Datasets:

Hendrik-a

/

arxiv

like 0

the quantity of dust in a spiral disk can be estimated using the dust s typical emission or the extinction of a known source . in this paper , we compare two techniques , one based on emission and one on absorption , applied on sections of fourteen disk galaxies . the two measurements reflect , respectively the average and apparent optical depth of a disk section . hence , they depend differently on the average number and optical depth of ism structures in the disk . the small scale geometry of the cold ism is critical for accurate models of the overall energy budget of spiral disks . ism geometry , relative contributions of different stellar populations and dust emissivity are all free parameters in galaxy spectral energy distribution ( sed ) models ; they are also sometimes degenerate , depending on wavelength coverage . our aim is to constrain typical ism geometry . the apparent optical depth measurement comes from the number of distant galaxies seen in hst images through the foreground disk , calibrated with the `` synthetic field method '' ( sfm ) . we discuss what can be learned from the sfm measurement alone regarding ism geometry . we measure the ir flux in images from the _ spitzer _ infrared nearby galaxy survey in the same section of the disk that was covered by hst . a physical model of the dust is fit to the sed to estimate the dust surface density , mean temperature , and brightness in these disk sections . the surface density is subsequently converted into the average optical depth estimate . the two measurements generally agree and the sed in order model finds a mostly cold dust ( t 25 k. ) . the ratios between the measured average and apparent optical depths of the disk sections imply optically thin ( ) clouds in these disks . optically thick disks , are likely to have more than a single cloud along the line - of - sight .

0707.4165

the dust content of a spiral galaxy disk can be mapped either by the characteristic dust emission in the far - infrared ( fir ) and sub - mm regimes , or by using the attenuation of known background sources . both techniques have seen recent significant improvements in accuracy and sensitivity , with complementary results shedding light on the dusty interstellar medium in spiral disks . the emission from the interstellar dust in the disks of spiral galaxies has been characterized with increasing accuracy by several infrared space missions ( iras , iso and , recently , _ spitzer _ ) , as well as the sub - mm observations of scuba on the jcmt . the improvements in spatial resolution and wavelength coverage have led to significant insight into the temperature components of the dust in spiral disks , and into the relation between dusty clouds and star - formation . the fir emission from spiral galaxies has revealed that the dust can be described by two dominant thermal components : warm ( 25 k @xmath0 t @xmath0 100 k ) and cold ( t @xmath0 25 k ) . both the warm and cold components can be found in spiral arms , and there is a smooth disk of cold dust between these arms . most of the dust mass in the spiral disk is cold ( see the review articles by * ? ? ? * ; * ? ? ? * ; * ? ? ? fir and sub - mm observations of galaxies find indications ( e.g. , * ? ? ? * ; * ? ? ? * ) or direct evidence of cold dust disks extending beyond the stellar disk @xcite . the studies of edge - on spirals by @xcite and @xcite quantified the radial profile as a scalelength of the dust that is 40% larger than that of the starlight . the contribution of the cold ism ( t @xmath0 25 k ) to the overall emission of spiral disks has been difficult to constrain because of the degeneracy between dust temperature and mass . hence , the cold ism s relation with hi and their relative distribution remain unknown . _ spitzer _ observations , mostly from the spitzer infrared nearby galaxy survey ( sings , * ? ? ? * ) , have already contributed greatly to the understanding of spiral disks . the relations between the tracers of cold dust ( 70 and 160 micron emission ) and star - formation , both obscured ( 24 micron ) and unobscured ( uv and h@xmath2 emission ) have already been studied with this multi - wavelength survey in several canonical galaxies and their substructure : the starburst m51 @xcite , the grand design spiral m81 @xcite , the rings of ngc 7331 @xcite and m31 @xcite , the superwind in m82 @xcite , and the dwarf ngc 55 @xcite . @xcite discuss the sed of all sings galaxies over all available wavelengths . @xcite find ample evidence for dust in all the sings galaxies , with the gas - to - dust ratio related to the metallicity . they find no evidence for very cold ( t @xmath0 10k ) dust , however . these studies find ample evidence of cold dust throughout the optical disks of spirals but , interestingly , also outside them in various places : on the edge of the optical disk @xcite , outside m82 s superwind @xcite and extending beyond the stellar disk @xcite . parallel to these investigations of dust emission has gone an observational effort to quantify the absorption by dust in spiral disks using known background sources . @xcite proposed using occulting galaxy pairs for this purpose . nearby occulting galaxy pairs were initially investigated with ground - based data , both images @xcite and spectra @xcite . subsequently , with the hubble space telescope ( hst ) , a more detailed picture of dust in these nearby disks emerged @xcite . the results of these studies are that extinction is gray when measured over disk sections greater than 100 pc , but resembles the galactic extinction law at smaller scales those that can only be resolved with hst . arms are found to be more opaque than the general disk , and some evidence suggests that the dust disk is a fractal , similar to the hi disk . @xcite investigated the use of the calibrated number of distant galaxies seen though the foreground disk in hst images . the calibrated counts of distant galaxies have been explored further in the previous papers in this series @xcite . both the occulting galaxy technique and counts of distant objects yield very similar opacities for disks and spiral arms @xcite . in recent years , models have been developed to explain the spectral energy distribution ( sed ) of edge - on spirals spanning wavelengths ranging from the uv to the fir . ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? + three scenarios have been proposed to explain the discrepancy between the apparent absorption in uv and optical wavelengths and the emission of dust in the fir and sub - mm regimes : * a young stellar population is embedded in the dense plane of the disk . this is proposed by @xcite and corroborated by @xcite . the embedded young stars pump the fir emission radiated by the dust plane . * a strongly clumped dusty medium . the clumping would lead to underestimate the dust mass from optical extinction in edge - on systems @xcite ; the dust mass would also be underestimated by a uv to fir sed @xcite . * a different emissivity of the cold dust grains higher than canonical in the fir and sub - mm . a change of emissivity has been proposed for denser ism regions @xcite or , alternatively , for the lower density regions of the disk @xcite . in all three of these scenarios , the clumpiness of the dusty ism is an important factor . while in some models the large scale structure of the dusty ism has been somewhat constrained ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , the small - scale geometry ( `` clumpiness '' ) of the cold ism remains unknown . therefore , an estimate of the prevalent dusty cloud size for spiral disks would provide a constraint for the sed models of spiral disks . given that sed and extinction techniques are sensitive to different characteristics of the dusty clouds in the spiral disk , a comparison between the optical depth derived from these two methods has the potential to figure out the structure of the dusty ism . here , we compare the @xmath3-band optical depths for a section of the spiral disk , one derived from a sed model of the _ spitzer _ fluxes ( `` average '' ) , and one determined from the number of distant galaxies found in an hst image ( `` apparent '' ) . the term `` average '' refers to what the optical depth proportional to the dust mass , uniformly distributed over the disk section . the term `` apparent '' means the effective optical depth of the disk section for a background uniform light source ( original definitions from * ? ? ? the term `` opacity '' is used throughout our previous papers for the apparent optical depth measured over a section of the disk for its whole height . this paper is structured as follows : [ sec : data ] discusses the _ spitzer _ and hst data used . in [ sec : analysis ] , the two different methods to derive optical depths are presented . we discuss the relation between dust geometry and galaxy counts in [ sec : csize ] . in [ sec : results ] , we present the derived optical depths ; [ sec : model ] presents a simple geometric model to interpret the results , and [ sec : concl ] lists our conclusions and future work . the data for this paper come from two archives , the hst archive and the fourth data release @xcite of the _ spitzer infrared nearby galaxy survey _ ( sings , * ? ? ? * ) . there is an overlap of fourteen galaxies between the sings sample and that of @xcite . two of these , ngc 3621 and ngc 5194 , have two wfpc2 exposures analyzed in @xcite . the hst / wfpc2 data reduction is described in @xcite . the reasoning behind the hst sample selection from the archive is explained in [ ssec : sfm ] . the overlap between hst and _ spitzer _ data is illustrated in figure [ fig : overlap ] , with the wfpc2 footprint projected on the 24 micron _ spitzer _ images . only the solid angle covered by the wf chips is used for further analysis ( the pc chip is excluded ) . the infrared array camera ( irac ) mosaic is made with the custom _ singsdither _ script , by m. regan , that combines the scan images into a single mosaic using the `` drizzle '' algorithm ( fourth data release notes , * ? ? ? the multiband imaging photometer for _ spitzer _ ( mips ) data products are calibrated , sky subtracted mosaics in all three bands , reduced as described in @xcite . the basic instrument parameters , pixel scale , and adopted psf fwhm for the seven main _ spitzer _ imaging modes are summarized in table [ table : data ] . two parallel estimates of the optical depth of disks are used in this paper : first , the apparent optical depth of the spiral disks is determined from the number of distant galaxies identified in the hst / wfpc2 images , calibrated with the `` synthetic field method '' ( sfm ) . secondly , the optical depth of the same section of the spiral disks is derived from the dust surface density , which is a result of the sed model fit to the _ spitzer _ fluxes using the model from @xcite . in principle , the number of distant galaxies seen through a spiral disk is a function of the dust extinction , as well as the crowding and confusion by the foreground disk . initial applications of the number of distant galaxies as an extinction tracer were on the magellanic clouds @xcite , but they lacked accuracy . the `` synthetic field method '' ( sfm ) was developed by @xcite to correct an extinction measurement based on the number of distant galaxies in an hst image , for the effects of crowding and confusion by the foreground spiral disk . the `` synthetic field method '' follows a series of steps . first , the number of distant galaxies in an hst science field is determined . second , a series of simulated ( `` synthetic '' ) fields are made . in each of these fields , a typical background ( e.g. , the hubble deep field ) is first dimmed by a gray screen and added to the science field . third , the added distant objects are identified in these synthetic fields . the fourth step is to measure the relationship between the number of these identified synthetic distant galaxies and background dimming . from this relation and the original number of actual distant galaxies found in the science field , an average opacity can be found . it is important to remake the synthetic fields for each science field because the crowding and confusion issues are unique in each case . an additional uncertainty in the resulting average extinction measurement is the cosmic variance in the intrinsic number of distant galaxies behind the foreground disk . the uncertainty due to cosmic variance can be estimated from the two - point correlation function of distant galaxies and folded into the poissonian error . the cosmic variance uncertainty is of the same order as the poisson statistical error for small numbers ( for a complete discussion on the uncertainties of the sfm , see * ? ? ? therefore , single - field sfm measurements remain uncertain , but a meaningful conclusion can be drawn from a combined set of science fields . we have applied this method successfully on archival wfpc2 data . @xcite present the average radial opacity profile of spiral disks and the effect of spiral arms . the spiral arms are more opaque and show a strong radial dependence , while the more transparent disk shows a flat profile . @xcite compare hi radial profiles to the opacity ones , and conclude that no good relation between disk opacity and hi surface density radial profiles can be found . however , @xcite find that the sub mm profile from @xcite generally agrees with their opacity measurements of m51 . @xcite compare the relation between surface brightness and disk opacity ; this relation is strong in the spiral arms , but weak in the rest of the disk . @xcite predicted , based on simulated data , that the optimum distance for the application of the sfm with current hst instruments is approximately that of virgo . the identification of background galaxies suffers in closer disks , as the stellar disk becomes more resolved , compounding confusion . this optimum distance , combined with the availability of deep hst / wfpc2 images from the cepheid extragalactic distance scale key project , resulted in the sample presented in @xcite . @xcite confirmed the results from @xcite using this sample , with foreground disks spanning distances between 3.5 and 35 mpc . a selection effect of the key project is that the majority of the hst science fields is concentrated on spiral arms and exclude the centers of the galaxies . here , we present average extinction values for the whole wfpc2 field of view , minimizing the uncertainties to the extent possible . in our initial papers , we did not apply an inclination correction to the optical depths because the correction depends strongly on the dust geometry ( see the discussion in * ? ? ? however , in this paper we assume a simple dust model in [ sec : model ] . the appropriate inclination correction @xmath4 has been applied to the points in figures [ fig : fit ] and [ fig : n_val ] , and in table [ table : a ] . the uncertainties in the tables and figures reflect the 1-sigma confidence levels produced by the combination of the poisson error and the cosmic variance of background galaxies . the average disk optical depth is derived from the _ spitzer _ observations . first , the surface brightnesses within the wf chips footprint are measured ( see figure [ fig : overlap ] ) . second , these are converted into a dust surface density using an sed model . third , this surface density is translated into an @xmath3-band optical depth . all the irac and mips data are convolved to the poorest resolution of the 160 micron observations ( see table [ table : data ] ) , and the pixel scale is set to 9 " . this is done with the _ gauss , wcsmap _ and _ geotran _ tasks , under iraf . subsequently , the overall flux is measured in the wfpc2 field of view ( figure [ fig : overlap ] ) . because the l - shaped aperture is a highly unusual one , the aperture correction remains uncertain , but not negligible , since the fwhm at 160 micron ( @xmath5 ) is of the order of the aperture diameter ( @xmath6 in an l - shape ) . published aperture corrections for the irac instruments @xcite overestimate the correction for extended objects @xcite . here we use the aperture corrections for extended sources from @xcite for irac fluxes , and from @xcite for the mips fluxes ( see also table [ table : data ] ) . table [ table : flux ] gives the average surface brightnesses for the seven _ channels in the field - of - view of the three wf chips of the wfpc2 array . the uncertainties are derived from the variance in the sky . generally , the surface brightnesses agree with the results presented by @xcite for the entire disks . the second step is to convert these surface brightnesses to a dust surface density . initially , we fitted only the mips fluxes with two blackbodies and derived surface densities from these @xcite . however , a more rigorous treatment of the ir fluxes can be done with a sed model , such as the one presented in @xcite . this model uses the physics of grain heating and reradiation , and a model distribution of grain sizes and types . the updated version from @xcite has been fit to the data , and the results are shown in figure [ fig : spec ] . the derived stellar and dust surface brightnesses , dust surface density , and mean temperature are presented in table [ table : model ] . dust surface densities are between 0.1 and @xmath7 , with mean temperatures between 14.6 and 17.8 k. the mean dust temperatures are obtained from the mean radiation scaling , @xmath8 , ) of the local interstellar radiation field to calculate the irradiation that the grains see . @xmath8 is the average of this scaling distribution . ] in the @xcite model ( @xmath9 in their equation 18 ) . the model uses a distribution of temperatures and grain sizes . therefore the mean temperature is an indication of the thermal equilibrium point of the bulk of the dust . most of the dust is cold ( @xmath10 ) . these results are an obvious improvement over a simple single - temperature fit , but the model parameters in table [ table : model ] are still not fully constrained ; disagreement between data and model at the pah peak at 8 micron could be an effect of metallicity or the presence of a bright hii region . the 70 and 160 @xmath11 m fluxes hint at colder or more dust in the disk ( figure [ fig : spec ] ) . there are three caveats to the fits : ( 1 ) a lack of sub mm data , ( 2 ) the single color isrf used and ( 3 ) averaging over different types of emission regions , i.e. , hii regions and the general disk . additional sub - mm data of comparable quality , needed to better constrain the model and especially the cold dust emission , will not be available until scuba2 starts operations on the jcmt and the launch of _ herschel_. at present , comparable - quality sub - mm maps are available only for ngc 5194 @xcite and ngc 7331 @xcite . @xcite discuss sed models with and without sub - mm data . the second caveat in the derivation of dust mass is the assumption of a constant color for the interstellar radiation field ( isrf ) illuminating the emitting dust . in reality , dust grains deeper in a dust structure will encounter radiation field that is not only dimmed but also reddened and , hence , will contribute less flux to the fir emission . locally , the isrf will also depend on the age of the nearby stellar population . due to the reddening of the isrf deeper in the cloud , dense clouds could contain more dust mass in their centers than inferred from just the fir emission . additional sub - mm observations will help resolve this uncertainty in the sed optical depth . @xcite discuss fits of the model to the total fluxes of the sings galaxies with and without additional sub - mm data . they find that the fir estimate underestimates of 1.5 times the dust mass more ( 5 out of 17 cases , notably ngc 3627 and 7331 of our sample ) than overestimates it ( only m51 of the 17 ) . the third caveat is that the model values are an average over many different types of ism regions , each with a different heating mechanism , dust structure and composition ( e.g. , photo - dissociation regions , cirrus , and star - forming regions in spiral arms ) . the @xcite model s assumptions hold better for some regions than for others , but we use the results as `` typical '' for these disks . the relative contribution of pah emission to the sed is a function of ism geometry as well as irradiation and composition ( e.g. , * ? ? ? * ; * ? ? ? together with a better - constrained fir / sub - mm sed one could constrain ism geometry solely from the relative contributions to the sed . @xcite discuss the application of the @xcite model to whole disks of the sings galaxies . the dust surface density is translated into an average optical depth with the absorption cross section per unit dust mass , @xmath12 , and grain albedo from @xcite for the johnson @xmath3-band ( 865.5 . ) : @xmath13 . these optical depth values are presented in table [ table : a ] . it continues to be difficult to constrain dusty cloud geometry from models of either extinction or emission . in this section , we review what can be learned , solely from the sfm measurements , about the geometry of the extincting medium . in @xcite , two indications that the dust disk is clumpy are identified : ( 1 ) the average color of the distant galaxies is independent of the disk opacity implied by their number , and ( 2 ) the measurement of disk opacity is independent of inclination . the lack of a relation between the average opacity of the disk and the average color of the detected galaxies can be explained by two scenarios : ( 1 ) some of the background galaxies are completely blocked by large clouds , and some are not . the color measurement is done on background objects that do not suffer from extinction and , hence , reddening . alternatively , ( 2 ) _ all _ background galaxies are dimmed by clouds smaller than the projected distant galaxies . consequently , some of the distant galaxies are dimmed enough to drop below the detection threshold . any detected galaxy s color is , however , measured from mostly unreddened flux . however , we note that a relation between the reddening and derived extinction from the distant galaxies is difficult to detect because ( 1 ) the spread in colors of distant galaxies is substantial and ( 2 ) color is measured from the detected galaxies automatically the least dimmed whereas opacity is measured form the _ missing _ galaxies . the inclination effect on the apparent optical depth from the number of distant galaxies is minimal ( see * ? ? ? * and figure 3 ) . assuming a thin layer of optically thick clouds , the apparent optical depth of the disk , measured from the number of distant galaxies , is dominated by the apparent filling factor of clouds . the projected filling factor does not change much with inclination : a flat cloud covering 40% of a certain disk section still covers 40% of the inclined section . only when the height of the cloud becomes important when the inclination is closer to edge - on , does the apparent filling factor change . this explanation for the lack of an inclination effect in the opacity profiles does not depend on the size of the clouds . it could be a single , large , cloud or many small ones in the plane of the disk . however , the optical depth values in @xcite are from different sections of the disks although generally centered on a spiral arm and the effect of small inclination differences could well have been masked by comparing different regions in the disks . in @xcite we did not apply an inclination correction because it depends on the assumed dust geometry . in this paper we do assume a dust geometry and hence make an inclination correction ( [ sec : model ] ) . the simulations in the sfm assume a gray screen , an uniform unclumped dust layer with opacity equal in the v and i bands . the sfm opacity measurements in this paper are based on such simulations . in @xcite , we ran a series of simulations on ngc 1365 to characterize the effect of average cloud cross - section on the number of distant galaxies observable through a disk . figure [ fig : sfmsize ] shows the effect of cloud size , expressed in pixels , on the simulated relation between average opacity and number of distant objects . in each simulation , we fix a cloud size and vary their number to increase disk opacity . an ensemble of unresolved clouds is effectively the gray screen . for clouds resolved with hst , more than 2 pixels in our data , after the drizzle reduction . ] , the relation between opacity ( cloud filling factor ) and number of distant objects becomes much shallower . the same number of distant galaxies observed would then imply a much _ higher _ opacity of the disk . because the sfm ( calibrated with a gray screen ) generally agrees well with measurements from overlapping galaxies @xcite , it seems unlikely that the disk s opacity is predominantly due to large , resolved clouds . a pixel of @xmath14 at the distance of ngc 1365 ( 18 mpc ) is 4 pc in linear size . it is therefore implied that the structure of the ism responsible for the disk opacity measured with the sfm varies in optical depth on scales of @xmath1510 pc or less . from the sfm measurements alone , the cloud geometry is impossible to determine . only when additional information is used e.g . , the general agreement with the occulting galaxy technique it favors small ( unresolved ) scales for the clouds . therefore , to constrain cloud geometry , information from two different techniques needs to be combined . table [ table : a ] presents the optical depth estimates from the synthetic field method and the sed model @xcite , for the wfpc2 field - of - view . the optical depths range between 0.1 and 3.5 magnitudes in the @xmath3-band . the measurements are for different parts of different spiral disks ( figure [ fig : overlap ] ) , explaining in part the range in values . the optical depth estimates presented here may appear high for the johnson @xmath3-band , compared with other extinction estimates ( e.g. , those from inclination effects or reddening ) , but these are ( 1 ) for the entire height of the disk , and ( 2 ) generally centered on a spiral arm . typical extinction values in the @xmath3-band are several tenths of a magnitude for a dust screen in front of the stellar spiral disk ( e.g. , * ? ? ? two of the derived sfm opacities are negative , possibly the effect of an overdensity of distant galaxies behind the target galaxy . the discrepancy in ngc5194 - 1 may be due to misidentification of background galaxies , as they are difficult to identify in this field . figures [ fig : fit ] and [ fig : n_val ] show the values of disk opacity by both methods , over the same section of the disk . both methods generally agree within the uncertainties of the measurements . the agreement is better than our initial estimate from a blackbody fit to the mips fluxes in @xcite . the general agreement and the mean temperature of the dust ( table [ table : model ] ) imply that most of a disk s opacity is due to the cold dust in the disk . the relation between the apparent and average @xmath3-band optical depth measurements the first from the number of distant galaxies in hst images and the second derived from the _ spitzer _ sed could reveal the nature of the prevalent structure in the ism . the _ a priori _ assumptions are that ( 1 ) all dust structure is transparent to the fir emission from which the dust surface density is estimated in the sed model , and ( 2 ) the _ entire _ volume of the cloud emits in the fir . we adopt _ model c _ from @xcite , in which a randomly distributed series of clumps covers the area . these authors define two optical depths : ( 1 ) the typical optical depth ( @xmath16 ) , i.e. , the average optical depth that is proportional to the dust mass ; and ( 2 ) the apparent optical depth ( @xmath17 ) , or the optical depth if a uniform layer would cover the area . our two measurements of optical depth sed and sfm correspond to these two optical depths , average and apparent . an optical depth based on the sed depends on the dust mass within the area , and hence corresponds to @xmath16 . the sfm optical depth is the apparent optical depth averaged over the area , and hence @xmath17 . these two optical depths need not be the same , and their relation is an indication of how clumped the medium is . let us assume a number of small dust structures with a height ( @xmath18 ) , an average grain cross - section ( @xmath19 ) , and a grain emissivity ( @xmath20 ) . the grain number density in the clouds is denoted by @xmath21 , and the average number of clouds in a line - of - sight is @xmath22 . we assume all clumps have the same optical depth @xmath23 : @xmath24 the average optical depth ( @xmath16 ) is then : @xmath25 and the apparent optical depth can be derived if one assumes a gaussian distribution of the number of clouds along the possible lines - of - sight and sum the contributions of all clouds ( see equations 15 and 17 - 19 in * ? ? ? * ) : @xmath26 with asymptotic values for both optically thin and thick clouds : @xmath27 @xmath28 we note that the inferred dust mass in the disk ( @xmath29 ) is proportional to the number of clouds ( @xmath22 ) and the cloud optical depth ( @xmath23 ) . the ratio of the apparent over the average optical depth is : @xmath30 leaving only the optical depth of the clouds ( @xmath23 ) , and hence the grain density ( @xmath21 ) and the cloud size ( @xmath18 ) as the variables . our fit to the relation between sfm and sed optical depth estimates only has @xmath23 as the variable ( figure [ fig : fit ] ) . we can now use equation [ eq : rat ] to derive @xmath23 from a fit to the optical depths from sed ( @xmath16 ) , and from the number of distant galaxies ( @xmath17 ) . we want to answer three questions . are the clouds in the disks typically optically thin or thick ? if all disks are equal , what is the implied cloud optical depth ? how many clouds lie typically along a given line - of - sight ? optically thin clouds ( @xmath31 ) result in a ratio of optical depths close to unity : @xmath32 optically thick clouds ( @xmath33 ) result in a ratio of : @xmath34 figures [ fig : fit ] and [ fig : n_val ] show how a majority of the data exhibit a ratio of order unity . the cold dust mass in the sed model could be better constrained with additional sub - mm information @xcite . however , the ratios are , within the errors , consistent with optically thin clouds in most of the disks . figure [ fig : fit ] shows the fit to the ratios of apparent to average optical depths , with @xmath23 as the single fit parameter , as per equation [ eq : rat ] . for simplicity , we assume here that all disks are made up of similar clouds , and that there is no difference between arm and disk regions . the negative sfm measurements are excluded from the fit . the best fit is for a cloud optical depth of @xmath350.4 with , on average , 2.6 clouds along the line - of - sight ( if the two negative points are included , the values change , respectively , to @xmath23 = 0.56 and n = 1.9 ) . the inferred value of @xmath23 is likely to be a mean between higher values in spiral arms and much lower values in the disk . optically thick disks have more clouds along the line of sight , while optically thin disks harbor a single cloud . reasonable values for the parameters in equation [ eq : tc ] are : @xmath36 , @xmath37 and @xmath38 with @xmath39 @xcite ( e.g. , * ? ? ? @xmath40 was fixed at 2 in the @xcite models . ] . the value of 0.4 for @xmath23 implies a cloud height , @xmath18 , of @xmath15 60 pc ! much larger clouds could be resolved in extinction maps of these disks based on stellar reddening . in the case of ngc 3627 , ngc 5194 , ngc 6946 and ngc 7331 , there is a clear spiral arm in the reddening map @xcite ; the other reddening maps are smooth or do not extend out to cover the whole wfpc2 pointing . a more likely scenario is that there is an inverse relation between the cloud density @xmath21 and scale @xmath18 . such a relation can be seen in giant molecular clouds ( gmc ) of our own galaxy ( e.g , * ? ? ? * ) . in this case , the single optically thin cloud can be replaced by smaller optically thick ones . we note that the largest gmcs are of the order of 60 pc . the value of 60 pc clouds appears in contradiction to the implied size in the sfm calibration ( @xmath1510 pc ) . however , the typical cloud size can be 60 pc and still the disk opacity can change over smaller scales , if several partially overlapping clouds are seen in projection . the inverse relation between scale and density would also help make the two scales compatible . our data are consistent with optically thin clouds ( @xmath1 ) and their average opacity value implies a typical cloud size that is unresolved with _ spitzer _ in our galaxies . the above fit to the relation between @xmath17 and @xmath16 indicates that , on average , more than one cloud is needed along the line - of - sight in most disks ( @xmath41 ) . however , the assumption was that @xmath23 had a single value for all the disks . in [ ssec : tc ] , we argued that the ratio between @xmath17 and @xmath16 implied the @xmath23 is optically thin . it logically follows that optically thick disks must have more than a single cloud along the line - of - sight . figure [ fig : n_val ] illustrates the effect of number of clouds along the line - of - sight ( @xmath22 ) on the relation between @xmath17 and @xmath16 , when @xmath23 is freely increased from 0 to @xmath42 ( the lines have been drawn according to equations [ eq : tm ] and [ eq : tap ] ) . we note that all disks are consistent with many clouds along the line - of - sight including the optically thin ones , also because they lie in the optically thin cloud ( @xmath43 , @xmath44 ) regime . there are many refinements to be made to the simple model presented here . some improvements for future comparisons between these two measurements of optical depth are : ( 1 ) a distribution of cloud sizes for both the sfm calibration , as well as in the model explaining the ratio between sfm and sed optical depths . a model distribution can be taken from observations of gmcs in our own galaxy and nearby ones @xcite . the cross - section distribution could be used in sfm measurements in the future . a cross - section distribution can only be applied , if the foreground disk is at a single , fixed distance ; only one counts though a single foreground galaxy are used . there are three face - on spirals with enough solid angle in hst imaging as well as additional _ spitzer _ data : m51 , m81 and m101 . ( 2 ) the sed model can be much better constrained with additional sub - mm observations , the opportunities for which will expand dramatically in the near future ( _ scuba2 _ on the jcmt and _ herschel _ satellite ) . ( 3 ) the effects of grand design spiral arms and galactic radius could be identified in a single disk ; the comparison sfm and sed is not made for different sections of the disks combined . ( 4 ) a future sed model can take into account the reddening of the interstellar radiation field , as it penetrates the ism . this would require a comprehensive treatment of the ism structure in addition to its temperature , composition and irradiation . to constrain models of the spiral disk s energy budget with typical values for the size of dusty clouds in the ism , we compare two techniques to extract the average and apparent optical depths of a section of spiral disk . from the comparison between sfm and sed results , we conclude the following : * the sfm s calibration alone implies projected cloud scales predominantly unresolved with hst ( of the order of 10 pc in ngc 1365 , see [ sec : csize ] and figure [ fig : sfmsize ] ) . * the dust responsible for the disk s opacity is predominantly cold ( t @xmath0 25 k , table [ table : model ] ) . * the average and apparent optical depths of these disk sections , measured from sed and sfm respectively , generally agree ( figure [ fig : fit ] and [ fig : n_val ] ) . this implies generally optically thin clouds ( @xmath45 , [ ssec : tc ] ) . * the fit to the ratio between apparent and average optical depth measurements , @xmath46 , indicates a cloud optical depth , @xmath23 , of 0.4 , more than a single cloud along the line - of - sight , and a cloud size of @xmath15 60 pc . if several partially overlapping clouds are seen in projection through the disk , the disk s opacity will change over smaller scales , consistent with conclusion 1 . * optically thick disks appear to have more than a single cloud along the line - of - sight ( figure [ fig : n_val ] ) and optically thin disks may have several clouds as well . future work using counts of distant galaxies through a foreground disk could be used to find cold dust structures at larger galactic radii , provided a sufficiently large solid angle has been imaged with hst / acs s superb resolution . notably , the acs data on m51 , m81 and m101 are very promising for such an analysis . _ spitzer _ observations of these nearby disks are also available , making a similar comparison between sed and apparent optical depth possible for portions of these disks . the typical cloud scale for spiral arms or disk sections or as a function of galactic radius could then be found . the scuba-2 instrument has recently been installed on the james clerk maxwell telescope . a project with scuba-2 to map the sings galaxies in two sub - mm bands will improve future sed modelling of these spiral disks significantly over the sed models presented here . this work is based in part on archival data obtained with the _ spitzer _ space telescope , which is operated by jpl , caltech , under a contract with nasa . this work is also based on observations with the nasa / esa hubble space telescope , obtained at the stsci , which is operated by the association of universities for research in astronomy ( aura ) , inc . , under nasa contract nas5 - 26555 . the authors would like to thank t. jarrett , for making his aperture corrections of extended sources available to us at an early stage , and erik rosolowsky , for useful discussions on local group cloud sizes . we would like to thank maarten baes for discussion of the motivation . we would also like to thank george bendo , erik hollenback and kristen keener for their comments on earlier drafts of this paper . , g. j. , dale , d. a. , draine , b. t. , engelbracht , c. w. , kennicutt , jr . , r. c. , calzetti , d. , gordon , k. d. , helou , g. , hollenbach , d. , li , a. , murphy , e. j. , prescott , m. k. m. , & smith , j .- d . t. 2006 , , d. , kennicutt , r. c. , bianchi , l. , thilker , d. a. , dale , d. a. , engelbracht , c. w. , leitherer , c. , meyer , m. j. , sosey , m. l. , mutchler , m. , regan , m. w. , thornley , m. d. , armus , l. , bendo , g. j. , boissier , s. , boselli , a. , draine , b. t. , gordon , k. d. , helou , g. , hollenbach , d. j. , kewley , l. , madore , b. f. , martin , d. c. , murphy , e. j. , rieke , g. h. , rieke , m. j. , roussel , h. , sheth , k. , smith , j. d. , walter , f. , white , b. a. , yi , s. , scoville , n. z. , polletta , m. , & lindler , d. 2005 , , 633 , 871 , d. a. , bendo , g. j. , engelbracht , c. w. , gordon , k. d. , regan , m. w. , armus , l. , cannon , j. m. , calzetti , d. , draine , b. t. , helou , g. , joseph , r. d. , kennicutt , r. c. , li , a. , murphy , e. j. , roussel , h. , walter , f. , hanson , h. m. , hollenbach , d. j. , jarrett , t. h. , kewley , l. j. , lamanna , c. a. , leitherer , c. , meyer , m. j. , rieke , g. h. , rieke , m. j. , sheth , k. , smith , j. d. t. , & thornley , m. d. 2005 , , 633 , 857 , d. a. , de paz , a. g. , gordon , k. d. , hanson , h. m. , armus , l. , bendo , g. j. , bianchi , l. , block , m. , boissier , s. , boselli , a. , buckalew , b. a. , buat , v. , burgarella , d. , calzetti , d. , cannon , j. m. , engelbracht , c. w. , helou , g. , hollenbach , d. j. , jarrett , t. h. , kennicutt , r. c. , leitherer , c. , li , a. , madore , b. f. , martin , d. c. , meyer , m. j. , murphy , e. j. , regan , m. w. , roussel , h. , smith , j. d. t. , sosey , m. l. , thilker , d. a. , & walter , f. 2007 , , 655 , 863 , b. t. , dale , d. a. , bendo , g. , gordon , k. d. , smith , j. d. t. , armus , l. , engelbracht , c. w. , helou , g. , kennicutt , jr . , r. c. , li , a. , roussel , h. , walter , f. , calzetti , d. , moustakas , j. , murphy , e. j. , rieke , g. h. , bot , c. , hollenbach , d. j. , sheth , k. , & teplitz , h. i. 2007 , , 663 , 866 , c. w. , gordon , k. d. , bendo , g. j. , prez - gonzlez , p. g. , misselt , k. a. , rieke , g. h. , young , e. t. , hines , d. c. , kelly , d. m. , stansberry , j. a. , papovich , c. , morrison , j. e. , egami , e. , su , k. y. l. , muzerolle , j. , dole , h. , alonso - herrero , a. , hinz , j. l. , smith , p. s. , latter , w. b. , noriega - crespo , a. , padgett , d. l. , rho , j. , frayer , d. t. , & wachter , s. 2004 , , 154 , 248 , c. w. , kundurthy , p. , gordon , k. d. , rieke , g. h. , kennicutt , r. c. , smith , j .- d . t. , regan , m. w. , makovoz , d. , sosey , m. , draine , b. t. , helou , g. , armus , l. , calzetti , d. , meyer , m. , bendo , g. j. , walter , f. , hollenbach , d. , cannon , j. m. , murphy , e. j. , dale , d. a. , buckalew , b. a. , & sheth , k. 2006 , , 642 , l127 , k. d. , bailin , j. , engelbracht , c. w. , rieke , g. h. , misselt , k. a. , latter , w. b. , young , e. t. , ashby , m. l. n. , barmby , p. , gibson , b. k. , hines , d. c. , hinz , j. , krause , o. , levine , d. a. , marleau , f. r. , noriega - crespo , a. , stolovy , s. , thilker , d. a. , & werner , m. w. 2006 , , 638 , l87 , k. d. , prez - gonzlez , p. g. , misselt , k. a. , murphy , e. j. , bendo , g. j. , walter , f. , thornley , m. d. , kennicutt , jr . , r. c. , rieke , g. h. , engelbracht , c. w. , smith , j .- d . t. , alonso - herrero , a. , appleton , p. n. , calzetti , d. , dale , d. a. , draine , b. t. , frayer , d. t. , helou , g. , hinz , j. l. , hines , d. c. , kelly , d. m. , morrison , j. e. , muzerolle , j. , regan , m. w. , stansberry , j. a. , stolovy , s. r. , storrie - lombardi , l. j. , su , k. y. l. , & young , e. t. 2004 , , 154 , 215 , k. d. , rieke , g. h. , engelbracht , c. w. , muzerolle , j. , stansberry , j. a. , misselt , k. a. , morrison , j. e. , cadien , j. , young , e. t. , dole , h. , kelly , d. m. , alonso - herrero , a. , egami , e. , su , k. y. l. , papovich , c. , smith , p. s. , hines , d. c. , rieke , m. j. , blaylock , m. , prez - gonzlez , p. g. , le floch , e. , hinz , j. l. , latter , w. b. , hesselroth , t. , frayer , d. t. , noriega - crespo , a. , masci , f. j. , padgett , d. l. , smylie , m. p. , & haegel , n. m. 2005 , , 117 , 503 , b. w. , meyer , m. , regan , m. , calzetti , d. , gordon , k. d. , smith , j. d. , dale , k. , engelbracht , c. w. , jarrett , t. , thornley , m. d. , bot , c. , buckalew , b. a. , kennicutt , r. c. , & gonzlez , r. a. 2007 , , _ submitted _ , j. l. , fazio , g. g. , allen , l. e. , ashby , m. l. n. , barmby , p. , deutsch , l. k. , huang , j. s. , marengo , m. , megeath , s. t. , melnick , g. j. , pahre , m. a. , patten , b. m. , smith , h. a. , wang , z. , willner , s. p. , hoffmann , w. f. , pipher , j. l. , forrest , w. j. , mcmurtry , c. w. , mccreight , c. r. , mckelvey , m. e. , mcmurray , jr . , r. e. , moseley , s. h. , arendt , r. g. , mentzell , j. e. , marx , c. t. , fixsen , d. j. , tollestrup , e. v. , eisenhardt , p. r. , stern , d. , gorjian , v. , bhattacharya , b. , carey , s. j. , glaccum , w. j. , lacy , m. d. , lowrance , p. j. , laine , s. j. , nelson , b. o. , reach , w. t. , stauffer , j. r. , surace , j. a. , wilson , g. , & wright , e. l. 2004 , in microwave and terahertz photonics . edited by stohr , andreas ; jager , dieter ; iezekiel , stavros . proceedings of the spie , volume 5487 , pp . 77 - 92 ( 2004 ) . , ed . j. c. mather , 7792 , r. c. , armus , l. , bendo , g. , calzetti , d. , dale , d. a. , draine , b. t. , engelbracht , c. w. , gordon , k. d. , grauer , a. d. , helou , g. , hollenbach , d. j. , jarrett , t. h. , kewley , l. j. , leitherer , c. , li , a. , malhotra , s. , regan , m. w. , rieke , g. h. , rieke , m. j. , roussel , h. , smith , j. t. , thornley , m. d. , & walter , f. 2003 , , 115 , 928 , p. g. , kennicutt , jr . , r. c. , gordon , k. d. , misselt , k. a. , gil de paz , a. , engelbracht , c. w. , rieke , g. h. , bendo , g. j. , bianchi , l. , boissier , s. , calzetti , d. , dale , d. a. , draine , b. t. , jarrett , t. h. , hollenbach , d. , & prescott , m. k. m. 2006 , , 648 , 987 , m. w. , thornley , m. d. , bendo , g. j. , draine , b. t. , li , a. , dale , d. a. , engelbracht , c. w. , kennicutt , r. c. , armus , l. , calzetti , d. , gordon , k. d. , helou , g. , hollenbach , d. j. , jarrett , t. h. , kewley , l. j. , leitherer , c. , malhotra , s. , meyer , m. , misselt , k. a. , morrison , j. e. , murphy , e. j. , muzerolle , j. , rieke , g. h. , rieke , m. j. , roussel , h. , smith , j .- d . t. , & walter , f. 2004 , , 154 , 204 -band , from sed ( @xmath16 ) and sfm ( @xmath17 ) . the best fit with equation [ eq : rat ] to the ratios of these values is also shown ( @xmath47 ) . the average value for number of clouds along the line - of - sight , @xmath22 , is then 2.6 . if the negative sfm values are included in the fit , the cloud optical depth rises to @xmath48 and the average @xmath22 becomes 1.9 . ] and @xmath17 , from equations [ eq : tm ] and [ eq : tap ] , for fixed values of the average number of clouds along the line - of - sight ( n = 0.5 , 1 , 2 , 3 , 4 ) . the cloud optical depth @xmath23 is left to vary proportionally to @xmath16 ( @xmath49 , equation [ eq : tm ] ) . hence maximum cloud optical depths are 6 , 3 , 1.5 , 1 and 0.75 respectively . if optically thin clouds are made up of optically thin clouds , it seems necessary that more than one cloud will lie along the line - of - sight . ] l l l l irac ( 3.6 ) & @xmath50 & @xmath51 & 0.9 + irac ( 4.5 ) & @xmath50 & @xmath51 & 0.9 + irac ( 5.8 ) & @xmath50 & @xmath51 & 0.7 + irac ( 8.0 ) & @xmath50 & @xmath51 & 0.75 + mips ( 24 ) & @xmath52 & @xmath53 & 1.16 + mips ( 70 ) & @xmath54 & @xmath55 & 1.2 + mips ( 160 ) & @xmath56 & @xmath5 & 1.4 + l l l l l l l l ngc 925 & 0.17 @xmath57 0.04 & 0.19 @xmath57 0.08 & 0.12 @xmath57 0.05 & 3.79 @xmath57 2.65 & 0.33 @xmath57 0.06 & 4.76 @xmath57 0.44 & 21.92 @xmath57 0.91 + ngc 2841 & 0.39 @xmath57 0.04 & 0.33 @xmath57 0.09 & 0.88 @xmath57 0.59 & 3.42 @xmath57 2.43 & 0.46 @xmath57 0.04 & 4.54 @xmath57 0.34 & 34.27 @xmath57 0.58 + ngc 3031 & 0.56 @xmath57 0.03 & 0.39 @xmath57 0.06 & 0.25 @xmath57 0.07 & 2.07 @xmath57 1.41 & 0.23 @xmath57 0.04 & 3.74 @xmath57 0.37 & 23.73 @xmath57 1.38 + ngc 3198 & 0.13 @xmath57 0.02 & 0.12 @xmath57 0.04 & 0.11 @xmath57 0.04 & 2.28 @xmath57 1.20 & 0.32 @xmath57 0.05 & 3.48 @xmath57 0.36 & 20.09 @xmath57 0.46 + ngc 3351 & 0.55 @xmath57 0.05 & 0.45 @xmath57 0.16 & 0.33 @xmath57 0.07 & 5.81 @xmath57 4.14 & 1.11 @xmath57 0.06 & 12.54 @xmath57 0.54 & 43.55 @xmath57 0.68 + ngc 3621 - 1 & 0.24 @xmath57 0.01 & 0.12 @xmath57 0.01 & 0.40 @xmath57 0.05 & 1.27 @xmath57 0.37 & 0.70 @xmath57 0.05 & 9.68 @xmath57 0.46 & 41.38 @xmath57 0.87 + ngc 3621 - 2 & 0.15 @xmath57 0.01 & 0.17 @xmath57 0.01 & 0.24 @xmath57 0.05 & 1.08 @xmath57 0.37 & 0.66 @xmath57 0.05 & 11.04 @xmath57 0.46 & 36.81 @xmath57 0.87 + ngc 3627 & 0.97 @xmath57 0.04 & 0.68 @xmath57 0.06 & 1.20 @xmath57 0.06 & 6.72 @xmath57 2.89 & 5.80 @xmath57 0.06 & 49.58 @xmath57 0.61 & 152.63 @xmath57 0.95 + ngc 4321 & 0.55 @xmath57 0.02 & 0.39 @xmath57 0.04 & 0.64 @xmath57 0.06 & 4.45 @xmath57 2.23 & 2.62 @xmath57 0.05 & 25.12 @xmath57 0.46 & 100.53 @xmath57 0.75 + ngc 4536 & 0.11 @xmath57 0.04 & 0.15 @xmath57 0.07 & 0.72 @xmath57 3.45 & 3.64 @xmath57 2.89 & 0.19 @xmath57 0.05 & 2.35 @xmath57 0.52 & 10.97 @xmath57 0.60 + ngc 4559 & 0.05 @xmath57 0.03 & 0.04 @xmath57 0.01 & 0.06 @xmath57 0.05 & 1.51 @xmath57 1.05 & 0.10 @xmath57 0.05 & 2.65 @xmath57 0.38 & 11.80 @xmath57 0.53 + ngc 4725 & 0.11 @xmath57 0.02 & 0.09 @xmath57 0.05 & 0.28 @xmath57 0.20 & 1.76 @xmath57 1.29 & 0.16 @xmath57 0.05 & 1.59 @xmath57 0.35 & 12.50 @xmath57 0.49 + ngc 5194 - 1 & 0.52 @xmath57 0.01 & 0.32 @xmath57 0.01 & 0.80 @xmath57 0.05 & 2.61 @xmath57 0.22 & 2.64 @xmath57 0.04 & 28.61 @xmath57 0.38 & 124.68 @xmath57 1.36 + ngc 5194 - 2 & 0.56 @xmath57 0.01 & 0.34 @xmath57 0.01 & 0.88 @xmath57 0.05 & 2.76 @xmath57 0.22 & 3.26 @xmath57 0.04 & 36.35 @xmath57 0.38 & 144.30 @xmath57 1.36 + ngc 6946 & 0.43 @xmath57 0.03 & 0.27 @xmath57 0.01 & 0.71 @xmath57 0.09 & 2.31 @xmath57 0.48 & 2.92 @xmath57 0.07 & 28.88 @xmath57 0.53 & 80.83 @xmath57 2.13 + ngc 7331 & 0.14 @xmath57 0.01 & 0.10 @xmath57 0.01 & 0.10 @xmath57 0.05 & 1.39 @xmath57 0.88 & 0.23 @xmath57 0.04 & 3.49 @xmath57 0.67 & 17.05 @xmath57 0.79 + ngc 0925 & 0.8 & 0.2 & 0.2 & 15.7 & 2.8 & 0.70 & 10 & 0.8 & 5.8 & u0.70_1e7_mw3.1_20 + ngc 2841 & 1.8 & 0.3 & 0.5 & 14.6 & 4.15 & 0.70 & 10 & 0.8 & 1.0 & u0.70_1e7_mw3.1_20 + ngc 3031 & 2.7 & 0.2 & 0.3 & 14.6 & 2.32 & 0.50 & 1.0 & 0.5 & 1.5 & u0.50_1e6_mw3.1_30 + ngc 3198 & 0.6 & 0.2 & 0.3 & 15.0 & 3.73 & 0.50 & 10 & 0.6 & 7.5 & u0.50_1e7_mw3.1_30 + ngc 3351 & 2.7 & 0.5 & 0.4 & 16.2 & 1.9 & 0.80 & 1.0 & 1.0 & 1.6 & u0.80_1e6_mw3.1_10 + ngc 3621 - 1 & 0.7 & 0.5 & 0.4 & 16.2 & 2.45 & 1.00 & 1.0 & 1.0 & 2.6 & u1.00_1e6_mw3.1_60 + ngc 3621 - 2 & 0.7 & 0.4 & 0.2 & 16.9 & 10.91 & 1.20 & 10 & 1.3 & 7.3 & u1.20_1e7_mw3.1_30 + ngc 3627 & 3.7 & 2.1 & 1.1 & 17.0 & 2.11 & 1.00 & 10 & 1.4 & 24.5 & u1.00_1e7_mw3.1_30 + ngc 4321 & 2.2 & 1.1 & 1.0 & 16.0 & 1.91 & 0.70 & 10 & 0.9 & 15.6 & u0.70_1e7_mw3.1_30 + ngc 4536 & 0.5 & 0.1 & 0.1 & 15.7 & 2.32 & 0.80 & 10 & 0.8 & 2.7 & u0.80_1e7_mw3.1_60 + ngc 4559 & 0.2 & 0.1 & 0.1 & 15.7 & 1.76 & 0.80 & 1.0 & 0.8 & 0.0 & u0.80_1e6_mw3.1_30 + ngc 4725 & 0.5 & 0.1 & 0.2 & 14.6 & 3.01 & 0.50 & 1.0 & 0.5 & 0.0 & u0.50_1e6_mw3.1_60 + ngc 5194 - 1 & 1.6 & 1.4 & 1.2 & 16.0 & 2.33 & 0.80 & 1.0 & 0.9 & 9.9 & u0.80_1e6_mw3.1_50 + ngc 5194 - 2 & 1.8 & 1.7 & 1.4 & 16.0 & 1.83 & 0.80 & 1.0 & 0.9 & 11.7 & u0.80_1e6_mw3.1_40 + ngc 6946 & 1.4 & 1.2 & 0.5 & 17.8 & 0.65 & 1.50 & 1.0 & 1.9 & 20.5 & u1.50_1e6_mw3.1_50 + ngc 7331 & 0.6 & 0.2 & 0.2 & 15.4 & 2.15 & 0.70 & 10 & 0.7 & 4.5 & u0.70_1e7_mw3.1_30 + l l l l l l l ngc 0925 & @xmath58 & -0.2 & 0.6 & 0.3 & -0.63 + ngc 2841 & @xmath59 & 0.4 & 1.4 & 0.7 & 0.53 + ngc 3031 & @xmath60 & 0.4 & 0.9 & 0.5 & 0.88 + ngc 3198 & @xmath61 & 0.3 & 0.7 & 0.4 & 0.97 + ngc 3351 & @xmath62 & 1.1 & 1.0 & 1.0 & 1.06 + ngc 3621 - 1 & @xmath63 & 1.1 & 1.0 & 0.5 & 2.05 + ngc 3621 - 2 & @xmath64 & 0.5 & 0.7 & 0.4 & 1.38 + ngc 3627 & @xmath65 & 1.0 & 3.1 & 1.7 & 0.61 + ngc 4321 & @xmath66 & 1.7 & 2.8 & 2.2 & 0.76 + ngc 4536 & @xmath67 & 0.4 & 0.3 & 0.1 & 2.58 + ngc 4559 & @xmath68 & 0.1 & 0.3 & 0.1 & 0.39 + ngc 4725 & @xmath61 & 0.5 & 0.5 & 0.4 & 1.38 + ngc 5194 - 1 & @xmath69 & -0.4 & 3.3 & 3.0 & -0.12 + ngc 5194 - 1 & @xmath70 & 1.2 & 3.9 & 3.5 & 0.34 + ngc 6946 & @xmath71 & 0.8 & 1.3 & 1.0 & 0.76 + ngc 7331 & @xmath72 & 0.2 & 0.5 & 0.3 & 0.62 +

the young smc cluster ngc 330 contains a number of blue stars that lie above the main - sequence turnoff found from our isochrone fitting and below the position of the blue supergiants . we used our own , new spectroscopy and published data on these stars to investigate their possible nature . problems in interpreting the evolutionary status of the blue giants have been found in several preceding studies . in theoretical h - r diagrams , these stars lie in the rapidly traversed post main - sequence gap , similar to the unexpected concentration found by fitzpatrick & garmany ( 1990 ) in the h - r diagram of the lmc . we argue that these stars probably are core h burning main - sequence stars that appear as blue stragglers resulting from binary evolution as described in the simulations of pols & marinus ( 1994 ) and effects of rapid rotation . many of the blue stragglers are be stars and likely rapid rotators . we suggest that there is evidence for the presence of blue stragglers also in ngc 1818 , ngc 2004 , and ngc 2100 . we point out that blue stragglers may be a general phenomenon in the cmds of young clusters in the magellanic clouds and discuss the implications for imf and age determinations . = 2.5 cm

astro-ph9512143

a few years ago , we started a project to determine ages , metallicities , and reddenings of distinct stellar populations employing a newly developed technique of simultaneous multi - colour isochrone fitting ( roberts 1996 , grebel et al . when applying this technique to the young cluster ngc 330 in the small magellanic cloud ( smc ) , we found a number of stars above the main - sequence turnoff , as indicated by our isochrone fit , but below the location of the blue supergiants ( sect . [ sect_photdata ] ) . when only photometric data are available , these bright blue stars would usually be considered main - sequence stars , which implies that either the cluster is much younger than our simultaneous multi - colour isochrone fit shows or forces the conclusion that there is a considerable age spread in star formation times . spectroscopy is needed in order to investigate the nature of these stars . we therefore obtained spectra to properly classify these stars ( sect . [ n330data ] ) . llllllccclll & & & & above & + arp & rob & fb & cjf & lmpbc & this study & v [ mag ] & b v [ mag ] & ms ? & & & + ii-13 & a1 & & & b0.5 iii / ve & b0 - 1 iii & @xmath0&@xmath1 & y & 145 & & @xmath2 + ii-14 & a2 & b5 i & b6 i & b4 ib & b5 i & @xmath3&@xmath4 & y & @xmath5 & @xmath6 & + ii-31 & a4 & & & & b2 - 3 iie & @xmath7&@xmath8 & y & 144 & & + ii-16 & a16&b9 i & a2 i & & & @xmath9&@xmath10 & y & @xmath11 & @xmath12 & + ii-33 & a25&a1 i & a0 i & & & @xmath13&@xmath14 & y & @xmath15 & @xmath16 & + ii-40 & a29&b9 i & & & & @xmath17&@xmath18 & y & @xmath19 & & + ii-35 & a47&b9 i & & & & @xmath20&@xmath21 & y & @xmath22 & & + i-243 & b2 & & a7 i & & & @xmath23&@xmath24 & y & & @xmath25 & + i-235 & b4 & & & b2 iiie & & @xmath26&@xmath27 & n & & & @xmath28 + i-234 & b5 & & & b2 iiie & & @xmath29&@xmath30 & n & & & + i-233 & b6 & & & b2 iiie & & @xmath31&@xmath32 & n & & & + i-232 & b7 & & & b2 iiie & & @xmath33&@xmath34 & n & & & + i-223 & b11 & & & a0 ii & & @xmath35&@xmath36 & n & & & + i-47 & b12 & & & b2 iiie & & @xmath37&@xmath38 & y & 154 & & + ii-28 & b13 & & & b2 iii / ive & & @xmath39&@xmath40 & n & 138 & & + i-222 & b16 & & a0 i & a0 ii & b9/a0 ib & @xmath41&@xmath42 & y & & & + i-204 & b18 & & & o9 iii / ve & & @xmath43&@xmath44 & n & & & + ii-46 & b21 & & b4e , n & b2.5 iie & & @xmath45&@xmath46 & y & & @xmath47 & + ii-45 & b22 & & & b2 iie & b3 ii & @xmath48&@xmath49 & y & @xmath50 & @xmath51 & + i - f & b24 & & & b2 iiie & b1 i ve & @xmath52&@xmath53 & y & & & + & b28 & & & b0.2 iiie & & @xmath54&@xmath55 & n & & & + ii-1 & b29&a0 i & a1 i & & & @xmath56&@xmath57 & y & @xmath58 & @xmath59 & + ii-3 & b30 & & b6 i & b2 ii & b1 ii & @xmath60&@xmath61 & y & @xmath62 & @xmath63 & + ii-4 & b33 & & & & b2 - 3 ve & @xmath64&@xmath65 & n & & & + & b35 & & & b2 iiie & & @xmath66&@xmath67 & y & & & + ii-8 & b37&b5 i & b5 i & b3 ib & & @xmath68&@xmath69 & y & @xmath70 & @xmath71 & + ii-9 & b38&a1 i & a5 i & & & @xmath72&@xmath73 & y & @xmath74 & @xmath75 & + i - c & & & & & b8 ii & @xmath76&@xmath77 & y & & & + i-201 & & & a1 i & & & @xmath78&@xmath79 & y & & @xmath80 & + i-206 & & & & & b9 ii & @xmath81&@xmath82 & y & & & + i-207 & & & & & b8 iii & @xmath83&@xmath84 & y & & & + i-211 & & & a0 i & & & @xmath85&@xmath86 & y & & @xmath87 & + i-218 & & & & & b1 v & @xmath88&@xmath89 & n & & & + i-219 & & & & & b9/a0 iii & @xmath90&@xmath91 & y & & & + i-220 & & & & & b3 iii - iv & @xmath92&@xmath93 & y & & & + in addition to our own spectral classifications we make use of previously published classifications ( sect . [ sect_specresults ] ) . these classifications and determinations of surface gravities and effective temperatures place these stars where one would not expect to find so many : in the post main - sequence gap of the hertzsprung - russell diagram ( h - r diagram ) . in the following , we define main - sequence stars as stars in the core h burning phase . when talking about `` ordinary '' main - sequence stars we are referring to stars undergoing the usual , undisturbed main - sequence evolution typical for single stars , to which also the geneva evolutionary tracks pertain . according to these models , an ordinary main - sequence star in the turnoff region of a 20-myr isochrone has a @xmath94 value of 3.7 to 3.8 . of course neither of these values nor a mere spectral classification alone can tell whether a star is core h burning or not . we shall refer to the area redward of the main sequence as post main - sequence gap or blue hertzsprung gap . this area lies between the terminal - age main sequence ( tams ) and the area where the blue loops of the supergiants occur . precise size and location of the gap depend on the stellar models used . according to stellar evolution theory , this region is traversed very quickly ( see , e.g. , langer & maeder 1995 for time scales ) making it highly unlikely to find a large number of stars there , or even any stars . in sect . [ sectbg_giants ] and [ sectbg_ms ] we discuss whether the blue stars above the main - sequence turnoff and below the blue supergiant locus are evolved stars or main - sequence stars . in sect . [ sect_exception ] we review indications that ngc 330 is not an exception among young magellanic cloud clusters in containing blue giants . the implications for the upper imf and age determinations are discussed in sect . [ sect_imf ] and [ sect_age ] . we obtained our photometric data with the 2.2 m mpia telescope at eso , la silla , chile , on 14 nov 1992 and on 08 oct 1993 . we used the eso faint object spectrograph and camera ( efosc ii ) with a 1024 @xmath95 1024 thomson chip ( eso # 19 ) . pixel scale was @xmath96 ( unbinned ) resulting in field of view of @xmath97 . we obtained short and long exposures in bessel ubvr , gunn i , and h@xmath98 with the 2.2 m s standard filter set . we observed several ubv(ri)@xmath99 standard fields from landolt ( 1992 ) on 08 oct 1993 resulting in a total of 47 standard star observations . the data were reduced using the standard procedures in daophot ii running under midas ( stetson 1992 ) . we then used own programs that correctly remove atmospheric extinction effects ( roberts & grebel 1995 ) and transform to the ubv(ri)@xmath99 standard system ( roberts 1996 ) . outside a radius of 22 pc around the cluster center the number of stars normalized by area is constant . we therefore consider stars outside this radius to be mainly field stars . in fig . [ fn330fjohnccmd ] , only stars within 22 pc around the cluster are plotted . our best - fitting isochrone is indicated by a solid line . for more details on the isochrone fitting procedure see roberts ( 1996 ) and grebel et al . ( 1994a ) . 1b in grebel et al . ( 1994b ) shows the full frame for ngc 330 and the surrounding field . in both diagrams , several stars are visible above the main - sequence turnoff and below the supergiant locus . the main - sequence turnoff as given by the isochrone is at @xmath100 . inevitably , the resulting magnitude difference between the loci of the blue supergiants and the main sequence is @xmath101 mag , which is expected from theory and is independent from the set of opacities and the amount of overshoot used ( stothers & chin 1992 ) . note that our isochrone fits the location of the blue and red supergiants . in past studies , the stars above the main - sequence turnoff defined by our isochrone have either been considered ordinary main - sequence stars whose tip indicates the main - sequence turnoff ( carney et al . 1985 ) or as main - sequence stars younger than the rest of the cluster , thus indicating an age spread . possible age spreads of 10 myr for geneva models ( bomans & grebel 1994 ) , 15 myr for semiconvective padua models , and as much as 38 myr for models with full overshoot ( chiosi et al . 1995 ) have been suggested . however , when fitting the position of the stars above our isochrone turnoff to younger isochrones there are no supergiants at the positions corresponding to the supergiant loci of these isochrones . this might be an effect of small - number statistics , though on the other hand it seems strange that the supposedly ongoing star formation would not have resulted in the evolution of additional , younger supergiants . obviously , additional data are needed . our spectroscopic data were obtained with a boller & chivens spectrograph attached to the eso 1.52 m telescope on 29 dec 1993 and on 28 july 1994 . the b&c spectrograph was equipped with a uv coated @xmath102 ford chip ( eso # 24 ) . the angular pixel scale was @xmath103/pixel . we used grating # 23 , covering a wavelength range from 390 to 780 nm in the first observing run and 370 to 730 nm in the second run . the slit width was 2@xmath104 . with 0.19 nm / pixel , we obtained a resolution of @xmath105 1000 in the blue . an order separation filter ( wg 360 ) with a cutoff at 360 nm was used to prevent second order overlap . the orientable slit allowed us to observe two target stars simultaneously . because of the wide slit we often obtained spectra of other stars as well in one exposure of course we did not get all the light from the other stars since some of them would be positioned on the edge of the slit . additional light loss occurred due to guiding errors ( an autoguider was not yet available at the 1.52 m telescope ) and refractory losses at higher airmasses . therefore no flux calibration was attempted , though we observed two spectrophotometric standards per night to determine the response curve of the instrument . the fwhm of the psf in spatial direction varied between @xmath106 and @xmath107 . in total we observed 20 stars located in the outer regions of the cluster ( not all of them blue ) . all spectra had exposure times of 30 min ( except for one exposure of 25 min ) . the raw data were reduced in the standard way , i.e. , bias subtracted and flatfielded by a normalized dome flat in order to remove pixel - to - pixel variations . the individual spectra were extracted , sky - subtracted , and wavelength - calibrated using the dispersion relation from the he ar comparison spectra . for the extinction correction the standard la silla extinction curve as incorporated into midas was used . an instrumental response curve was derived in the same way from the standard stars . then the target stars were response - corrected with this curve and rectified . the poor signal - to - noise ratio in the violet made it difficult to determine spectral types for the spectra from dec 93 . therefore the data from jul 94 were used whenever possible . for our spectral classifications we used digital versions of the ob star atlas of walborn & fitzpatrick ( 1990 ) and the spectral atlas by jacoby et al . fig . [ n330bfspec1 ] shows our spectra with line identifications and classifications . additional spectral classifications were performed by feast & black ( 1980 ) , carney et al . ( 1985 ) , and lennon et al . ( 1994 , 1995 ) . combining all these data , we find that the majority of blue stars with @xmath108 v @xmath109 mag belong to luminosity class iv , iii , and ii , while stars brighter than v @xmath110 mag are supergiants ( tab . [ tn330tbluefairies ] ) . we shall refer to the stars of luminosity class iv , iii , and ii in the following as _ blue giants_. the two spectroscopically classified main - sequence stars lie below the main - sequence turnoff indicated by our isochrones . thus the main - sequence turnoff determined from our isochrone fits agrees well with the position of stars identified spectroscopically as main - sequence stars , as well as with that of the blue and red supergiants . the supergiants are mostly intermediate- and late - type b supergiants and early a - type supergiants . the blue giants span the entire b - type range . the two b main - sequence stars in our sample are early b - type stars ( b1 v to b2 v ) , which is not surprising since only spectra of the brightest blue stars were taken . for a discussion of individual stars , see sect . [ sect_indiv ] . the majority of the stars show h@xmath98 emission . spectral types determined by different authors vary within a few subtypes , which may be due to different recording techniques ( e.g. , resolution , and ccd frames versus image tube spectra ) and analytical techniques . also , spectral classifications are difficult in the metal - poor smc due to the fewer / weaker metal lines than in luminous blue galactic stars ( walborn et al . 1995 , lennon et al.1995 ) . the spectra with the highest resolution and best s / n for b stars in ngc 330 are those obtained by lennon et al . ( 1994 ) . reitermann et al . ( 1990 ) analyzed high - resolution echelle spectra in the visual and low - resolution iue spectra of the b star b30 ( naming convention from robertson 1974 ) in ngc 330 . in their h - r diagram , b30 lies in the post main - sequence gap . reitermann et al . suggest that this star has not yet been a red supergiant . bessell ( 1991 ) determined surface gravities for b22 , b30 , and b37 in ngc 330 . in both studies , low surface gravities were found ( tab . [ tn330surf ] ) . reitermann et al . ( 1990 ) nicely illustrate in their fig . 2 the uncertainties in determinations of effective temperatures and surface gravities when comparing the results from different methods , namely those based on the `` reddening - free index '' q = ( @xmath111 ( ) , the equivalent width of balmer lines , and the siii / sii ionization equilibrium . often @xmath112 values are based on q alone and very susceptible to the effects of individual reddenings and colour excesses . transformation equations for determination of @xmath112 as given in massey et al.(1995 ) furthermore require advance knowledge of the luminosity class . however , since the analyses by different authors , which are in part based on different data and different methods , agree quite well within the errors , we believe that the location of the blue giants in the blue hertzsprung gap is not caused by erroneous transformations . cccccccccccc star id&&bessell & & & & & ba / sw + & @xmath94 & @xmath112 & @xmath94 & @xmath94 & @xmath112&@xmath94 & @xmath112 & @xmath94 & @xmath112 & be & variable + a01 & & & & & & 3.7 & 28,000 & 4.10 & 28,000 & y & + a02 & & & & & & 2.5 & 15,000 & 2.50 & 16,000 & n & + b04 & & & & & & 3.9 & 23,000 & 3.80 & 25,000 & y & + b05 & & & & & & 3.7 & 19,000 & 3.50 & 22,000 & y & ba964,@xmath113eri + b06 & & & & & & 3.7 & 22,000 & 3.40 & 22,000 & y & + b07 & & & & & & & & 3.80 & 24,000 & y & + b11 & & & & & & & & 3.25 & 12,000 & n & + b12 & & & & & & & & 3.50 & 23,000 & y & + b13 & & & & & & & & 3.60 & 22,000 & y & + b18 & & & & & & & & 4.50 & 32,000 & y & + b21 & & & & & & 3.4 & 22,000 & 2.80 & 20,000 & y & + b22 & & & 3.2 & 3.22 & 22,700 & 3.2 & 20,000 & 3.25 & 21,000 & y & + b24 & & & & & & & & 3.80 & 25,000 & y & hv1669 + b28 & & & & & & & & 4.50 & 30,000 & y & + b30 & 3.00 & 18,700 & 3.1 & 2.91 & 18,800 & 3.2 & 19,000 & 3.25 & 21,000 & n & + b32 & & & & & & 3.8 & 28,000 & & & ? & ba347,@xmath113eri + b35 & & & & & & & & 3.50 & 25,000 & y & + b37 & & & 2.8 & & & 2.7 & 17,000 & 2.60 & 18,000 & n & + caloi et al . ( 1993 ) analyzed iue spectra of many of the stars below the blue supergiant locus and used kurucz models to determine temperatures and surface gravities ( tab . [ tn330surf ] ) . though they could not determine spectral types , they found temperatures and surface gravities of these stars as well as their position in the post main - sequence gap of the h - r diagram incompatible with their being main - sequence stars . they pointed out problems in interpreting the evolutionary status of these stars , a finding that was confirmed by lennon et al . lennon et al . determined temperatures and surface gravities from fits of lte and non - lte model atmospheres and robertson s ( 1974 ) bv photometry . caloi et al . ( 1993 ) find the stars b4 , b5 , b6 , b21 , b22 , and b30 to lie on the subgiant branch in their theoretical h - r diagram . these are statistically too many stars for a phase traversed so quickly . they mention that b4 because of its low temperature resembles an evolved star , which would mean an age spread of 15 myr . masses of these apparently evolved stars seem to be of the order of 10 m@xmath114 , which would indicate that mass loss is a lot less efficient than usually assumed . in jttner s ( 1993 ) h - r diagram star b22 lies very close to the tams , while b30 lies in the post main - sequence gap . jttner derives a mass of @xmath105 10 m@xmath114 for b30 and @xmath105 15 m@xmath114 for b22 . lennon et al . ( 1994 ) suggest that some of the supergiants ( a2 , b37 ) and bright giants ( b21 , b22 , b30 ) with redder colours may be evolved post red supergiants , while a1 and b18 ( both bright , but bluer and higher @xmath94 ) could be blue stragglers . we are here using the notations introduced by robertson ( 1974 ) when referring to these stars . star b18 was re - classified as an o9 giant or main - sequence star ( lennon et al . , 1995 ) . in our h@xmath98 frames , this star is the brightest star at the center of a large , elliptical hii region of about @xmath115 in diameter , part of dems 87 ( davies et al . 1976 ) , located approximately @xmath116 northwest of ngc 330 . b18 may be ( one of ) the exciting star(s ) of the hii region , in which case it is probably not a member of ngc 330 . this suggestion is circumstantially supported also by the fact that b18 is outside the membership radius described in sect . [ sect_photdata ] . the comparatively low v magnitude ( for a late o star ) may be due to higher extinction in the hii region , which then would amount to a few tenths of a magnitude . it is possible that b18 is more heavily extincted than other stars in this region because this o star appears a little fainter than the brightest early b main - sequence stars . lennon et al . ( 1994 ) find the bright giants or supergiants b11 and b16 to be very red and suggest that these stars might be smc field stars . in our data the supergiant b16 is quite red as well and , in our photometry , shows h@xmath98 emission , while b11 does not have red colours and photometrically appears to be a main - sequence star . a number of stars classified as b giants and subgiants by lennon et al.(1995 ) or by us lie in the main - sequence turnoff region , partly even a little below ( tab . [ tn330tbluefairies ] ) . of these stars , b4 , b5 , b6 , b12 , b13 , and b28 were found to show h@xmath98 emission and are be stars ( grebel et al . 1992 , 1994b , lennon et al . 1994 , grebel 1995 , mazzali et al . we classify star b33 as a main - sequence be star below the main - sequence turnoff . the more distant ( and probable non - member ) star i-218 is a main - sequence star without balmer emission . judging from their @xmath94 values , b28 and b18 may also qualify as a main - sequence stars . many stars above the turnoff are be stars ( e.g. , a1 , a4 , a15 , a22 , a39 , b7 , b12 , b17 , b21 , b22 , b24 , b35 , tab . [ tn330tbluefairies ] , grebel 1995 ) , some of which were identified as @xmath113 eri variables or eclipsing binaries ( balona 1992 , sebo & wood 1994 , grebel 1995 ) . chiosi et al . ( 1995 ) plot the stars of caloi et al . ( 1993 ) in an h - r diagram together with main - sequence and blue - loop boundaries for different types of evolutionary models . they find that when using models with full overshoot , which considerably widens the main sequence and shifts the tams to the red , all stars are located within the permissible boundaries of the tams except for b21 , b22 , and b30 , which still lie in the post main - sequence gap . there are a number of stars above the main - sequence turnoff that do not show balmer emission , or did not show balmer emission at the times they were observed . some of these stars tend to have as blue colours as the main - sequence stars beneath them , while the be stars generally stand out through their red excess in the redder colour indices , as one would expect . we also note a decline in number density of stars above the main - sequence turnoff as determined spectroscopically and by isochrone fitting . this decline can be seen clearly in the and cmds ( fig . [ fn330fjohnccmd ] ) . we shall now discuss the possible evolutionary status and nature of those unexpected blue stars above the main sequence and below the area of the blue supergiants in ngc 330 . there are two possibilities : ( 1 ) evolved giants or ( 2 ) core h burning main - sequence stars . we have to consider that the blue giants in ngc 330 may be evolved stars . as mentioned already , to find any stars in the subgiant region of the h - r diagram at all is unlikely because evolution here proceeds rapidly ( e.g. , langer & maeder 1995 ) , and to find so many is statistically nearly impossible according to current evolutionary models . it can be excluded that all of these stars are subgiants . of course ngc 330 is heavily contaminated by the surrounding young field population . our blue giants might simply be older , evolved field stars that we see superimposed on the cluster . however , the number - density distribution of blue giants is concentrated toward the cluster , and several of the blue giants are radial velocity members of ngc 330 . we estimate that stars in the velocity range of 140 to 155 km s@xmath117 belong to ngc 330 ( tab . [ tn330tbluefairies ] ) following the measurements of feast & black ( 1980 ) , carney et al . ( 1985 ) , and lennon et al . ( 1996 ) , which make a1 , a4 , a16 , a29 , a47 , b4 , b12 , b22 , and b30 radial velocity members . potential additional members with radial velocities at the lower end of the distribution are a2 , a29 , and b13 . we therefore exclude that the blue giants are merely evolved field stars seen in superposition . radial velocity measurements for b37 and b38 by feast & black ( 1980 ) support their being members of ngc 330 , while carney et al.s ( 1985 ) results would make them field stars . the stars b2 , b21 , b29 , i-201 , and i-211 , all of which are supergiants except for b21 , are probably not members of ngc 330 ( which is further supported by their distances from the cluster ) . while the positions of the supergiants a2 and b37 in the h - r diagram ( caloi et al . 1993 , lennon et al . 1994 ) agree with lennon et al.s suggestion that they are he burning , evolved post - red supergiants , the blue loops of the geneva models do not reach far enough to the blue for the bright giants b21 , b22 , and b30 . missing opacity could account for missing blue loops for less massive stars ( stothers & chin 1994 ) . the atmospheric composition of post - red supergiants should have changed due to admixture of cno - cycle products ( walborn 1988 ) , so high - resolution spectroscopic analyses should be able to clarify their evolutionary status . the ledoux criterion plus semiconvection work well for low - metallicity stars ( stothers & chin 1994 ) , though adoption of the ledoux criterion has little effect on stars in their main - sequence phase ( stothers & chin 1992 ) . however , for metal - poor red supergiants the ledoux criterion will lead to normal c / n ratios because of shallow convection envelopes . the same holds for the blue supergiant phase . thus alteration of surface composition is not necessarily to be expected . fitzpatrick ( 1991 ) found a large spread in n equivalent widths among b - type supergiants in the lmc , and fitzpatrick & bohannan ( 1993 ) suggest that 10 to 20% of the stars in their sample are pre - red supergiants . for some galactic blue supergiants it has been suggested that they suffered partial mixing of cn - cycled gas near the main sequence ( gies & lambert 1992 , lennon 1994 , venn 1995 ) , i.e. , these stars , too , show cno abundances different from unevolved b main - sequence stars . high - resolution spectroscopy of blue supergiants in the magellanic clouds generally showed significant he and n enrichment , while c and o were depleted ( kudritzki et al . 1987 , lennon et al . 1991 , humphreys et al . 1991 , fitzpatrick & bohannan 1993 , walborn et al . 1995 ) , which has been attributed to these stars being post - red supergiants . this indicates that even in metal - poor environments such as in the magellanic clouds the surface composition changes in general when evolving through the supergiant phase . as caloi et al . ( 1993 ) point out , high - resolution studies of b30 in ngc 330 do not show he or n enrichment ( reitermann et al.1990 ) . the same was found for b22 ( jttner et al . 1993 , tab . [ tcno ] ) . it therefore does not seem very likely that these blue giants are post - red supergiants . in a high - resolution study of the narrow - line be stars a1 and b4 , lennon et al.(1996 ) find these stars highly enriched in nitrogen , while no helium enrichment and no carbon deficiency is observed as would be expected from supergiant evolution ( tab . [ tcno ] ) . when the schwarzschild criterion for convection is used , a main - sequence star can evolve after core h exhaustion to become a blue supergiant before having passed through the red - supergiant phase ( case - b evolution , chiosi & summa 1970 and discussion in fitzpatrick & garmany 1990 ) . a star of 15 m@xmath114 would be a case - b blue supergiant at an age of about 11 myr following the evolutionary calculations for case - b by brunish & truran ( 1982 ) . a comparison of brunish & truran s fig . 3 and the h - r diagram given by caloi et al . ( 1993 ) and lennon et al . ( 1994 ) shows that the location of the `` blue giants '' is compatible with the position of blue supergiants in brunish & truran s models . however , if we consider the blue giants to have evolved in the sense of brunish & truran , we face the problem of how to explain the presence and location of the `` normal '' supergiants , both blue and red . besides , we found ngc 330 to be about twice as old as the above - quoted model from fits to geneva isochrones , which of course are based on different model parameters such as a certain amount of convective core overshooting leading to higher ages , a different metallicity , and current opacity models . if we were to assume an age spread to explain the presence of such massive case - b b supergiants , we would again be confronted with the problem of explaining why within one cluster two different types of stellar evolution and convection could take place . it does not seem likely that the case - b scenario is at work in ngc 330 . the effects of rapid rotation may change the inferred spectral classification ( slettebak et al . 1980 ) . as collins ( 1987 ) points out , deviations from spherical symmetry caused by rapid rotation imply that there is no longer a global @xmath112 and @xmath94 . instead , one can only determine a specific @xmath112 and a specific @xmath94 that is valid for the observed inclination angle . both quantities vary across the stellar surface depending on co - latitude . from pole - on inclinations to equator - on , the spectroscopic gravity may differ by a factor of ten , and the photometrically derived effective temperature may vary by several thousand degrees ( see collins s ( 1987 ) fig . 2 , see also slettebak et al.1980 ) . this should be kept in mind when looking at the @xmath112 and @xmath94 values listed in tab . [ tn330surf ] . if rapid rotation plays a rle here , then these values do not necessarily imply that these stars are evolved . the effects of rotation will be discussed in detail in the next section . luminous core h burning main - sequence stars may spectroscopically appear as giants , as has long been known ( e.g. , massey et al . 1995 and references therein ) . however , ordinary core h burning stars would not be found in the blue hertzsprung gap of the h - r diagram . also the sudden decline in number density above the main - sequence turnoff indicates that the blue giants may be different from main - sequence stars . for the blue giants to be ordinary main - sequence stars , the main sequence needs to be a lot wider than given by the models of the geneva group so wide that it covers that part of the blue hertzsprung gap in which the blue giants are found . following the suggestion by tuchman & wheeler ( 1990 ) , chiosi et al.(1995 ) demonstrate that this can partially be achieved by full overshoot models , for which several but not all of the blue giants in ngc 330 lie close to the tams , while the blue hertzsprung gap and the blue supergiant loop are shifted further to the red . full overshoot is , though , an extreme hypothesis that does not fit well with modern mixing length theories ( e.g. , grossman et al . 1993 ) . both the padua group and the geneva group have meanwhile adopted a moderate amount of overshoot that seems to fit observations in the milky way and in the magellanic clouds best . thus , our blue giants do not seem to be ordinary main - sequence stars in the sense previously defined . bessell & wood ( 1993 ) pointed out that the location of the blue giants at lower effective temperatures could also be reproduced when using the evolutionary tracks of doom ( 1985 ) . these models use the roxburgh criterion for convection and overshoot , are more centrally concentrated , and reach higher luminosities and lower effective temperatures . the location of the tams is in agreement with the location of the blue giants . spin - up , proposed by bessell & wood as a possible cause for the be phenomenon due to rotationally induced mass loss , is fostered by these models . these models might be a valid explanation for what we observe . however , if with anticipation of the discussion in later sections we take into account that each cluster has a certain fraction of binary stars , in the models of doom these stars would have to appear shifted even further to the red , and as we shall see later , that does not agree with what is observed in galactic open clusters with binaries ( s. pols & marinus 1994 ) . models taking into account high rotational velocities may also be able to account for ( part of ) the blue giants . many of the blue giants are be stars , thus supposedly rapid rotators . we will discuss the implications of the presence of be stars below . rotational effects were studied by collins & smith ( 1985 , s. also maeder & peytremann 1970 , 1972 ) and collins et al . ( 1991 ) ) , who show that the amount of main - sequence widening depends , apart from rotational velocities , on whether stars rotate as rigid bodies or differentially . the stronger the differential rotation , the wider the main sequence . the main sequence of differentially - rotating stars broadens particularly for stars of later types ( a , f ) , while the effects are less pronounced in the steep part described by b - type stars . as collins et al . ( 1991 ) have shown , the observed brightness of a b star is ( among other things ) a function of its rotational velocity and its inclination to the line of sight . a pole - on rotator ( inclination angle @xmath118 ) is displaced upwards from the main sequence of `` non''-rotating stars and will be much ( up to a magnitude ) brighter than other b stars of the same spectral type but at different inclination angles . the larger the inclination angle ( maximum @xmath119 equator - on ; common definition for equator - on : @xmath120 , collins 1987 ) of a b star toward us , the redder the star becomes , and , though still brighter than a non - rotating star , less luminous . therefore , a star with a given , high rotational speed describes what collins et al . ( 1991 ) call the `` rotational displacement fan '' in a colour - magnitude diagram ( s. their fig . 1 ) , its actual colour and brightness depending on its inclination angle . as collins et al . explain , rotational distortion causes a pole - on rotator to present a larger surface area including the high - temperature polar regions free of limb darkening and gravity darkening . therefore such a star appears brighter than a non - rotating star seen pole - on . rapidly rotating stars seen equator - on exhibit very strong limb darkening in their polar regions and , through rotational distortion , have more extended , cooler equatorial regions than non - rotating stars . both effects make them fainter than non - rotating counterparts . the latter also causes them to appear redder . rapid rotation also influences the interior of the rotating star , in essence causing a lower surface temperature , and stars may appear even more shifted to redder colours than in models considering only rotational distortion - induced changes in the atmosphere . the positions of the be stars in ngc 330 , which mostly are redward of the main sequence in our cmds ( figs . [ fn330fjohnccmd ] ) , may to a large extent be caused by rotational displacement as just explained . stothers & chin ( 1992 ) suggest that either strong convective envelope overshoot or fast interior rotation may account for the h - r diagram of ngc 330 and point out the high fraction of presumably rapidly rotating be stars in ngc 330 . for most of the stars without balmer emission and vertically above the main sequence their height above the main sequence is in accordance with what one might expect for pole - on rotators . more about that in the next section . most of the blue giants have been identified as be stars ( grebel et al.1992 , 1994b , lennon et al . 1994 , grebel 1995 , tab . [ tn330surf ] and [ tn330tbluefairies ] ) . as already mentioned , not all of the blue giants currently show h@xmath98 emission , but the be phenomenon is known to be variable and episodic . most ( galactic ) be stars are rapid rotators ( slettebak 1988 ) . the effects of rotational distortion were just explained in the preceding section . be stars in addition often show an infrared excess due to free - free emission in their disks , which already becomes visible in red filters of the visual wavelength range ( fig . [ fn330fjohnccmd ] ) . the position of the majority of the blue giants in the h - r diagram can be explained by their being be stars and thus shifted to lower effective temperatures and lower surface gravities due to effects of rotation and ( infra-)red excess . since ngc 330 is unusually rich in be stars ( grebel et al . 1992 , 1994b ) , we will now discuss the hypothesis that all blue giants in ngc 330 are be stars , even though some do not show balmer emission at the moment . if all blue giants were be stars , this would imply that the stars vertically above the main sequence must be pole - on rotators . we assume that the rotational axes of these stars would have to be within about @xmath121 of the line of sight so that limb darkening and the small solid angle of the equatorial regions would make the stars significantly brighter and bluer ( and smaller angles would be significantly more effective ) . the solid angle within @xmath121 of _ both _ poles is about @xmath122 , so the probability of a be star s having such an orientation is roughly 1 in 8 . in our cmds of ngc 330 , we see nine blue giants that lie more or less vertically above the main sequence . if these are pole - on rotators , we expect approximately 63 non - pole - on be stars . rapid rotation may affect both b and be stars . furthermore , it has to be considered that pole - on rotators will not only be present above the main sequence , but also along the main sequence just as the be stars span a range of magnitudes . also , for our comparison of pole - on and non - pole - on rotators to be meaningful we must confine ourselves to stars only within the magnitude range that may give rise to the observed nine vertically displaced blue giants . thus we need to count b stars and be stars within one magnitude below the main - sequence turnoff ( and the be stars above it ) to establish a valid base of comparison . the number ratio is approximately 60 supposedly non - pole - on b and be stars versus nine possible pole - on rotators above the main sequence . ( at this point , we note again that these nine stars above the main sequence do not show h@xmath98 emission . ) considering the uncertainties of the numbers , incompleteness , and crowding , we can not rule out that those nine blue giants are pole - on rotators . statistically , the only possibility for all blue giants in ngc 330 with or without current h@xmath98 emission to be be stars is that _ all _ b stars in ngc 330 would have to undergo the be star phenomenon at some point of their evolution . if all bright b stars are rapid rotators this would suffice to explain the presence of the vertically displaced blue giants . as rapid rotators and/or be stars , the blue giants may well still be core h burning . it is not yet known if the b - type stars in ngc 330 are rapid rotators . lennon et al . ( 1994 ) and mazzali et al . ( 1995 ) determined rotational velocities for a large number of be stars in ngc 330 . however , these measurements are based on the h@xmath98 emission line , which originates in the circumstellar disk around the be stars and may not be representative for the rotational velocity of these stars . reitermann ( 1989 ) determined @xmath123 values for seven b type stars in ngc 330 . the @xmath123 values range from 45 km s@xmath117 to @xmath124 km s@xmath117 . since the inclination angles are not known , we do not know what the rotational velocities are . at the current time , it can not yet be verified whether or not b - type stars in ngc 330 are rapid rotators . based on fitting their models , stothers & chin ( 1992 ) suggest that stars in ngc 330 suffer fast interior rotation . balona ( 1992 ) estimates from the mean periods and magnitudes of the 19 @xmath113 eri variables that he found in ngc 330 that the mean equatorial rotational velocity of these stars may be 375 km s@xmath117 , a relatively high value as compared to galactic @xmath113 eri variables . fast rotation may facilitate the formation of be stars and short - periodic be variables such as @xmath113 eri stars . we point out in this context that ngc 1818 and ngc 2004 , two young clusters in the lmc that also seem to contain blue giants , have comparatively many fewer be stars ( grebel et al . 1994b , sect . [ sect_exception ] ) . those blue giants vertically above the main sequence could also be blue stragglers . by definition , blue stragglers are stars that appear to have `` straggled '' in their main - sequence evolution and , while still being core h burning main - sequence stars , lie above the main - sequence turnoff given by the age of the cluster in question ( stryker 1993 ) . blue stragglers have been found in all kinds of stellar aggregates , from old globular clusters down to young ob associations ( mermilliod 1992 ) . blue stragglers may be the result of delayed formation ( star formation over an extended period of time ) or of delayed evolution ( binaries with mass transfer , eggen & iben 1988 ) . in the first case , the blue stragglers are likely to be single stars . should the blue giants found vertically above the main sequence be the result of delayed formation , we would expect them to show spectral characteristics typical for earlier types of stars . in the case of ngc 330 , where the earliest stars on the main sequence found from isochrone - fitting are of type b0 to b1 , we would consequently expect the most massive younger stars to be o stars . this however is not what we observe . thus the delayed - evolution scenario appears to be more likely . in fact there is mounting evidence that blue stragglers result from binary interactions ( stryker 1993 ) . collier & jenkins ( 1984 ) present model calculations for case - b mass transfer , i.e. , mass transfer through roche - lobe overflow while the primary component is already a red giant . to quote , the `` two most notable features resulting from the inclusion of the binary population are the splitting of the main sequence , and the presence of a sequence of blue stars near the main sequence but _ above _ the cluster turn - off . '' ( collier & jenkins 1984 ) . this effect is visible in our cmds and even more pronounced in ngc 1818 ( will et al . 1995 ) . as collier & jenkins point out , for stars with primaries more massive than 2.5 m@xmath114 , case - a mass transfer is more prevalent than case b. in case - a evolution , the system is so close that roche - lobe overflow occurs already on the main sequence ( kippenhahn & weigert 1967 ) . following kippenhahn s & weigert s model calculations , the age of ngc 330 is compatible with its containing very massive secondaries ( mass @xmath125 11 m@xmath114 ) that would appear as blue stragglers . the binary system in this case looks like a semi - detached algol - type system , where the less massive component fills its roche lobe and appears as overluminous subgiant . on the other hand , de loore & vanbeveren ( 1994 ) argue that with decreasing metallicity , the importance of case - a mass transfer decreases as well . tuchman & wheeler ( 1990 ) present a scenario in which the hertzsprung gap can be filled if a main - sequence star gets enriched in he by a more massive red - supergiant companion through roche - lobe overflow . the main - sequence star becomes a peculiar blue giant while the more massive companion may become a supernova or a neutron star . as mentioned before , though , high - resolution studies of two bright giants in ngc 330 ( reitermann et al . 1990 , jttner et al . 1993 ) did not find evidence for he enrichment , nor for n enrichment ( compare with jttner et al.s results for field stars , tab . [ tcno ] ) . that would seem to exclude the possibility that we are seeing the effects of homogeneous evolution of massive stars leading to the formation of obn blue stragglers ( maeder 1987 , beech & mihalas 1989 ) . these two stars with their very low surface gravities may , however , be coalesced binaries or a similar result of massive close binary evolution . on the other hand , lennon et al.s ( 1996 ) finding of significant n enrichment in a1 and b4 without the accompanying c and o depletion ( compare to the results of rolleston et al . 1993 , who used the same method ; tab . [ tcno ] ) . the authors suggest binary mass transfer or rotationally induced mixing for these two stars with low v sin i values . they might be pole - on rotators , but a number of discrepancies are yet to be solved . for discussions , see luck & lambert ( 1992 ) and lennon et al . ( 1996 ) . cccccccccccc & sun & + element & milky way & & & & & lmc & smc + @xmath126(m ) & ll & rd & ll & rd & ll & bssm & jt & mbp & ga & + & 45 stars&4 stars&15 stars&5 stars&11 stars & 3 stars & 10 stars & 5 stars & & & + c & 8.15 & 8.04 & 7.97 & 7.73 & 7.65 & 7.7 & 8.0 & 7.7 & 8.6 & 7.90 & 7.16 + n & 8.47 & & 8.36 & & 7.72 & 6.9 & & & 8.0 & 7.07 & 6.55 + o & 8.61 & & 8.68 & & 8.21 & 7.9 & & & 8.9 & 8.13 & 8.37 + + & milky way & ngc2004 & & & + & dbfl & jswb & & jswb & & & + & & d12 & av304 & idk - d2 & av175 & 11 & 637 & b30 & b22 & a01 & b04 + & & & b0 v & b2 iii & b iii & b0 v & b0 v & b1 ii & b3iie & b0.5iii / ve & b2iiie + c & 8.2 & 8.00 & 6.9 & 7.1 & 7.68 & @xmath127 & 6.8 & 7.78 & 7.79 & 6.9 & 6.9 + n & 7.8 & 7.54 & @xmath127 & @xmath128 & 7.40 & @xmath129&@xmath130&6.92 & 6.97 & 7.7 & 7.7 + o & 8.9 & 8.14 & 8.2 & & 8.26 & 8.0 & 8.0 & 7.79 & 7.98 & 8.0 & 8.0 + pols ( 1994 ) presents detailed results for case - a and case - b mass transfer between massive close binaries . it is likely that these interactions result in one of the components becoming a ( comparatively low - mass ) he star , which then would appear blueward of the ordinary core h burning main sequence ( blue interloper ) . while we see some fainter stars displaced to the blue of the hydrogen main sequence , these stars may simply be there because of large photometric errors and scatter for fainter magnitudes . spectroscopy of these stars would be useful . studying the short - periodic variability of stars in ngc 330 , balona ( 1992 ) found three early - type eclipsing binaries , one of them probably a contact system . in a study of long - periodic variability of stars in the surroundings of ngc 330 , sebo & wood ( 1994 ) found one eclipsing binary in the region of the blue giants ( hv 1669 or b24 , a b1 i ve star , s. tab.[tn330surf ] ) and three other binaries , one of them a contact w uma - type star . of these , hv11348 was identified as be star ( grebel 1995 , suspected already by carney et al . both stars also lie above the main sequence . thus , there is observational proof for the presence of at least two binaries among the blue giants and be stars . pols & marinus ( 1994 ) performed monte - carlo simulations of binary - star evolution in young open clusters and compared their results to galactic open clusters . they consider four different types of mass transfer and assume a binary fraction dependent on the initial mass - ratio distribution ( overall 25 to 30 % binaries among b - type stars ) . for young open clusters , their models reproduce the observations very well . the binaries cause the main sequence to be widened , or even a second , binary main sequence to appear . blue stragglers form as result of mass exchange in close binaries , which may also lead to the formation of be stars . the blue stragglers can either appear as single stars with fully coalesced components , or have stripped main - sequence companions , he - star companions , or white dwarfs or neutron stars , depending on the closeness of the system and the initial masses and mass ratios of the components . the blue stragglers are rapid rotators . stars rejuvenated by accretion of matter often appear not only brighter but also bluer than the main sequence . the simulations of pols & marinus elegantly explain the features we see in the cmds and h - r diagrams of ngc 330 . combining the previously described effects of rapid rotation and multiplicity with various forms of mass exchange , they account for a widened main sequence , for blue giants that emerge as blue stragglers , for blue giants vertically above or even to the blue of the main sequence , and for ( some of the ) be stars among the blue stragglers and fainter main - sequence stars . additional observational tests can be carried out to check predictions of the models . apart from the spectroscopic survey for he stars suggested above , x - ray data of the ngc 330 region should be investigated for evidence of b or be stars with neutron - star or white - dwarf companions , which show a different spectral signature than x - ray flares originating in the disks of be stars ( cassinelli & cohen 1994 ) . the models of pols & marinus also predict that for many binaries radial - velocity variations are so small that they may escape detection , and in fact _ no _ indications for binarity were found in previous radial velocity studies ( feast & black 1980 , carney et al . 1985 ) . long - term photometric monitoring of ngc 330 similar to the four - year study of the surrounding field population carried out by sebo & wood would give further clues about the nature of various types of binaries in ngc 330 in addition to the short - term variability studied by balona ( 1992 ) . ngc 330 is exceptional in at least one respect : it contains the highest fraction of be stars of any young cluster investigated so far ( grebel et al . 1994b ) . however , ngc 330 is not unique in containing what we have named blue giants . the young lmc cluster ngc 2100 seems to contain several stars that could qualify as `` blue giants '' . bhm - vitense et al . ( 1985 ) analyzed iue spectra for a number of blue stars in ngc 2100 . the stars c13 and b20 ( robertson s numbers , robertson 1974 ) have quite low surface gravities though bhm - vitense et al . refer to b20 as a main - sequence star . from their placement in the h - r diagram the stars b20 , c13 , and c31 could be `` blue giants '' since all of them lie in the post main - sequence region that is traversed very quickly . admittedly , these are very few stars in comparison to ngc 330 . additional candidates should be investigated . analyzing iue spectra of blue stars in the young lmc cluster ngc 2004 , caloi & cassatella ( 1995 ) find about eight stars lying in the post main - sequence gap . in a cmd , these stars lie at the apparent upper end of the main sequence , just as in ngc 330 . indeed , if we fit an isochrone to cmds of ngc 2004 ( fig . 1 in grebel et al . 1994b ) we find the main - sequence turnoff at lower magnitudes in good agreement with where caloi & cassatella find the first probable main - sequence stars . our own photometry of the young lmc cluster ngc 1818 ( grebel et al . 1994b , grebel 1995 ) as well as the study by will et al . ( 1995 ) indicates that here as well there is a number of `` blue giants '' above the main - sequence turnoff and below the position of the blue supergiants . in fact , the cmds of will et al . show a forked `` upper main sequence '' with one group of stars shifted toward red colours , while the other group of stars looks like a slightly blueward extension of the main sequence . the h - r diagrams for the young field population of smc ( garmany & fitzpatrick 1989 ) and lmc ( fitzpatrick & garmany 1990 ) show a possibly related effect : the post main - sequence gap is not , as one would expect from stellar evolution theory , empty but instead well filled with stars . it is quite possible that this is at least partly the same phenomenon we observe in ngc 330 and several other young magellanic cloud clusters . we emphasize that in selecting their stars , fitzpatrick & garmany ( 1990 ) excluded all known emission - line objects , while in ngc 330 the majority of stars filling the hertzsprung gap shows balmer emission . possible sources for erroneous placement in the hertzsprung gap of the magellanic cloud field stars are discussed by tuchman & wheeler ( 1990 ) . it also has to be emphasized that all currently widely available evolutionary tracks and isochrones consider slow - rotating , single star evolution only . in galactic open clusters , widened main sequences , binary main sequences , and blue stragglers are well - known phenomena and common to clusters of all ages ( mermilliod 1992 ) . it is hardly surprising that the same effect should exist in the magellanic clouds . we suggest that blue giants or blue stragglers may be a common feature in young magellanic cloud clusters ( and possibly in clusters of all ages ) . this possibility should be carefully investigated since it has significant implications for the determination of ages and imfs in the magellanic clouds . the true nature of the stars at the apparent upper main sequence is of great significance for the determination of imfs . if the blue giants are all rapidly rotating single stars and single be stars , the upper imf will include stars of erroneously high masses since the inclination angle of a rapid rotator may shift it by up to one magnitude in m@xmath131 , which in turn implies a higher luminosity in the h - r diagram and a higher mass . if the blue giants are binaries it becomes almost impossible to estimate their mass from their position in the cmd / h - r diagram . their current position is determined by the past mass exchange episodes and various evolutionary effects . we usually do not have information about the current state of the companion , nor do we know the initial mass ratio ( see also discussion in clarke & pringle 1992 ) , which would be needed for the determination of the imf . unresolved binaries may easily result in giving too large a weight to the upper mass bins of the imf if the star consists of two ( or more ) components that initially had roughly the same mass . unrecognized neutron - star or white - dwarf companions may lead to an underestimate of the mass of the visible component . if instead the blue giants are evolved stars , the upper mass bins of the imf will be underestimated since the masses of the main - sequence progenitors are likely to have been higher . however , as discussed we think we can exclude that these stars are evolved . if the binary fraction is known , one can make assumptions about the frequency distribution of different mass ratios and the contribution of these components to the imf . detailed comparisons with the predictions of the models of pols & marinus ( 1994 ) may help to constrain the binary fraction and the types of mass transfer . however , since stars do not need to be binaries to become be stars and since ngc 330 is so rich in be stars , we may see the combined result of two different effects : the rotational distortion and red excess for single be stars , and the main - sequence widening and blue ( be ) stragglers due to binarity . at this point , we do not see how these effects can be disentangled . some of the be stars may indeed be single stars , since balona ( 1992 ) found a number of @xmath113 eri variables among them , which are not considered binaries . on the other hand , the majority of the galactic be stars ( slettebak 1988 ) appear to be be stars . thus in addition to observational uncertainties and crowding , the determination of the imf of ngc 330 is severely affected by the above described effects , probably even more so than other magellanic cloud clusters because of its high fraction of be stars . as massey et al . ( 1995 ) point out , spectral classifications in addition to photometry are a prerequisite for any imf determinations but as we have seen not even spectroscopy can give all the necessary information . we are still far from knowing `` typical '' imf slopes , let alone how the imf varies in different ( metallicity ) environments since especially binary fraction and binary mass ratios are unknown . synthetic cmds , e.g. , of pols & marinus may help with these problems , however , in the case of ngc 330 the code would need to be adjusted such that single be - star evolution and binary be - star evolution is taken into account . even then , reproducing what we see in ngc 330 would not necessarily lead to conclusive answers since there are many free parameters ( such as binary fraction , rotational velocities , initial mass ratios , be - star fraction and be lifetimes ) so that several combinations may reproduce the observed result . attempts to reproduce observed cmds of four young lmc clusters with synthetic cmds assuming binary fractions between 0 and 50% have been made by subramaniam & sagar ( 1995 ) . they note that in some cases , quite extreme binary fractions are required to reproduce the observations . subramaniam & sagar did not account for be stars in the synthetic cmds . the age calibration of magellanic cloud clusters is an important input parameter for population synthesis models . ages of magellanic clusters are in part based on integrated colours , a rather crude method , and in part on cmds obtained photographically or by ccd photometry . ages obtained from the cmds and isochrone fitting are being used to recalibrate ages based on integrated colours . integrated colours , in turn , provide an essential input base for population synthesis models . the properties of young clusters in particular are important for modeling active galaxies and starburst galaxies . for example , bica et al . ( 1990 ) investigated integrated spectra of lmc and smc clusters in the red and far - red ( 5601000 nm ) . they dated the red - supergiant phase to occur between 7 and 12 myrs . as we have shown , however , the much older cluster ngc 330 is heavily dominated ( 85% contribution in the @xmath132 band ) by red supergiants ( grebel 1995 , grebel et al . in prep . bica et al . ( 1990 ) based their age derivations on the integrated johnson ( ) @xmath133 and gunn ( u v)@xmath133 colours and the apparent main - sequence turnoff visible in cmds . young magellanic clouds clusters with a high be star content like ngc 330 may appear bluer in integrated light due to the uv excess of some of these bright stars , than one would expect for a cluster of the same age without be stars . furthermore , a contribution to the uv flux may come from he stars formed in binary interactions ( pols et al.1991 , pols 1994 , pols & marinus 1994 ) . thus these clusters may be dated younger than they really are . the age calibration of young magellanic cloud clusters will be off if blue stragglers are generally present , as they are in young galactic open clusters . if only photometric data with one colour index are used , unrecognized blue giants may place the turnoff region up to a magnitude or more too bright ( s.carney et al . 1985 , sect . [ sect_photdata ] ) , depending on the age of the cluster . obviously , this leads to a systematic underestimation of the cluster age . the exact age that is derived strongly depends on the parameters of the underlying stellar models ( see chiosi et al . 1995 ) . a first indicator for blue stragglers is the presence of stars above the main sequence and below the blue supergiant locus , if isochrones are fitted such that they include both the main - sequence band and the loci of blue and red supergiants . in fact , our simultaneous isochrone - fitting technique roberts 1996 ) in several colours does not accept the blue giants as part of the main sequence , and this is how we first identified them as anomalous , without spectroscopy . in absence of spectroscopy or multi - colour photometry , blue stragglers will not be recognized for what they are so that age spreads will be derived to fit their positions by isochrones as well ( sect . [ sect_photdata ] ) . the second implication of our findings on the age of ngc 330 is that an age spread , if present , is small and within the uncertainty of the age derived for ngc 330 from isochrone fitting that includes the supergiants . the stars in ngc 330 may be considered coeval within @xmath134 myr ( grebel 1995 , grebel et al . in prep . ) . if blue stragglers are generally present , large age spreads need no longer to be invoked to explain the morphology of a cmd . lack or presence of a considerable age spread in clusters will have important ramifications for theories of star formation . we believe that once spectroscopic data are available for a larger set of clusters , previously inferred age spreads will be much reduced . the discovery of blue giants in ngc 330 implies that this cluster is older than has been derived in several previous studies ( main - sequence turnoff is shifted to fainter magnitudes ) and that the previously inferred large age spread ( e.g. , chiosi et al . 1995 ) is much smaller . the location of the blue and red supergiants is in agreement with the main - sequence turnoff inferred from spectral classifications and isochrone fitting . we reject the possibility that these stars are subgiants evolving from the main sequence towards the red - supergiant phase , since evolution proceeds so rapidly that it is unlikely to catch stars in this phase . considering that high - resolution spectroscopic studies of blue giants in ngc 330 did not show enhancement of both he and n and the accompanying c and o deficiency , we argue that the blue giants probably are not evolved stars such as post - red supergiants . we argue that rotational displacement effects due to rapid rotation and properties of be stars , which is what most of the blue giants are , can account for their position in the h - r diagram . departures from spherical symmetry as discussed by collins ( 1987 ) prevent the measurement of global @xmath112 and @xmath94 values and may , depending on the inclination angle , make stars appear as giants . we find that the simulations of the effects of binaries on cmds of young clusters by pols & marinus ( 1994 ) describe the observed cmd and h - r diagram morphology very well . these models account for a strongly widened main sequence , the presence of be stars , and blue unevolved stars above the main sequence through binary interactions and mass transfer that lead to the formation of blue stragglers . we suggest that the blue giants are most likely a mixture of ( 1 ) rapidly rotating b / be stars of varying orientation and ( 2 ) blue stragglers formed by interactions in binary systems . both possibilities are compatible with these stars being core h burning . faint short - periodic early - type eclipsing variables and blue straggler binaries have been found in the surroundings of ngc 330 ( sect . [ sect_bs2 ] ) , and we suggest additional observational tests to quantify the binary presence in ngc 330 . isochrones in the observational plane considering a certain binary fraction and rotational displacement rather than representing low - rotation , single - star evolution would be most useful . we discuss evidence that ngc 330 is not an exception in containing `` blue giants '' , and that blue stragglers are possibly also present in ngc 1818 , ngc 2004 , and ngc 2100 . we argue that unrecognized blue stragglers and binaries can severely impair the determination of imfs . they can also lead to an underestimation of the ages of young magellanic clusters and to the derivation of erroneously large age spreads in star formation times , as happened in the case of ngc 330 . ekg gratefully acknowledges valuable discussions with jon & karen bjorkman , steve cranmer , katy garmany , reinhard hanuschik , and rens waters , who directed me to useful references that helped to uncover the probable nature of the blue giants in ngc 330 . we are grateful to danny lennon for making available his results prior to or during publication . we thank klaas de boer for a critical reading of the text and our referee , dr . v. caloi , for useful comments . ekg was supported through a graduate fellowship in the graduiertenkolleg `` the magellanic system , its structure , and its interaction with the milky way '' of the german research foundation ( dfg ) . wb was supported by the german research foundation ( dfg ) under grant yo 5/161 . this research has made use of the simbad database , operated at cds , strasbourg , france , and of the nasa astrophysics data system , operated at cfa , harvard , usa . grebel , e.k . , roberts , wm j. , van de rydt , f. 1994a , in `` the local group . comparative and global properties '' , 3rd ctio / eso workshop , eso conference & workshop proceedings no . 51 , eds . a. layden , r.c . smith , & j. storm , la serena , p. 148

we prepare a general framework for analyzing the dynamics of a cylindrical shell in the spacetime with cylindrical symmetry . based on the framework , we investigate a particular model of a cylindrical shell - collapse with rotational pressure , accompanying the radiation of gravitational waves and massless particles . the model has been introduced previously but has been awaiting for proper analysis . here the analysis is put forward : it is proved that , as far as the weak energy condition is satisfied outside the shell , the collapsing shell bounces back at some point irrespective of the initial conditions , and escapes from the singularity formation . the behavior after the bounce depends on the sign of , the shell pressure in the z - direction . when , the shell continues to expand without re - contraction . on the other hand , when ( i.e. it has a tension ) , the behavior after the bounce can be more complicated depending on the details of the model . however , even in this case , the shell never reaches the zero - radius configuration .

gr-qc0404080

investigations on how the spacetime structures would be during and after gravitational collapses are important topics in gravitational physics and spacetime physics . nonlinear nature of gravity is making our understanding of the phenomena very difficult , both conceptually and technically . steady development of numerical methods is contributing to our deeper understanding on the phenomena . however , it does not mean the analytical investigations are becoming less meaningful . on the contrary , the more the numerical results we get , the more variety of analytical investigations is needed to interpret them and to reach the whole understanding of the phenomena . in this respect , we have not yet obtained enough variety of mathematical models for gravitational collapses which can be studied , mainly by analytical methods . indeed , the cases of just spherical collapses with various matter content is already complicated enough , and investigations are still going on @xcite . it can be inferred from this situation that constructing and analyzing other types of models may not be an easy task . as the next step in this direction , cylindrical collapses have been investigated considerably . this class of collapsing models are important because elongated matter distribution can drastically alter the destination of the gravitational collapse . already in the newtonian gravity , the jeans scale plays a vital role and the elongated matter longer than its jeans scale tends to split into smaller scale objects . in the case of general relativity , there is the so - called hoop conjecture " @xcite ; black holes are formed when and only when the matter of mass @xmath3 is contained in a compact region whose circumference @xmath4 in any direction always satisfies @xmath5 . therefore it is of great importance to study the fate of the collapsing elongated matter , such as cylindrical collapses . the numerical investigation of the collapse of the non - rotating dust spheroid by shapiro and teukolsky @xcite suggested that the naked singularities could emerge when sufficiently elongated object collapses . this numerical result motivated further analytical investigations on the cylindrical collapses . here , in view of the cosmic - censorship hypothesis @xcite , investigations on more natural " conditions , rather than non - rotating cases , are of significance . in particular , the effect of rotation and the effect of matter pressure should be studied . then , an analytical model of collapsing cylindrical shell made of counter - rotating dust particles was investigated by apostolatos and thorne @xcite . ( here counter - rotating particles " indicates the situation in which the half of the particles are rotating in the opposite direction to the other half of the particles . ) it takes into account the rotation of particles through the rotational pressure , while the net angular - momentum of the shell is kept zero . by avoiding the frame - dragging effect in this manner , the analysis can be relatively simple . it turned out that even a small amount of rotation prevents the singularity formation as far as this particular model concerned . another analytical model of cylindrical gravitational collapses was presented and briefly studied by pereira and wang @xcite . this model has interesting features in that not only has it taken into account both the rotation and the pressure effect , but it also tries to take into account the radiation of gravitational waves and massless particles from the collapsing region . indeed , the numerical work suggests that the gravitational radiation becomes important after the bounce of the cylindrical shell @xcite . in spite of several interesting features , however , this model has not yet been properly investigated so far . since this model is what is investigated in this paper , let us briefly recall their work @xcite . firstly they prepare general formulas in the coordinate system @xmath6 for matching arbitrary two cylindrical spacetimes . here junction conditions @xcite play the central role . then they apply the formulas to look at a particular model . in this model , an interior flat spacetime and an exterior cylindrical spacetime are matched together at @xmath7 . the geometry of the exterior spacetime is so chosen that there is out - going flux which can be interpreted as gravitational waves and massless particles ( see sec.[sec:3 ] for the details of this model ) . then a dynamical equation for the shell is explicitly derived ( see eq.([eq : pw - dynamical1 ] ) in sec.[sec:4 ] ) . since this non - linear equation is complicated enough , a very special class of solutions is searched for , and three types of special solutions are claimed to be found . they are the solutions of ( i ) a collapsing shell which bounces back at some finite radius , followed by an eternal expansion afterwards , ( ii ) an eternally expanding shell , and ( iii ) a collapsing shell whose radius reaches zero , resulting in a line - like singularity formation @xcite . the solution ( iii ) looks in particular interesting , since it reminds us of the result of numerical calculations by shapiro and teukolsky @xcite , though the spacetime structures in these two cases are not the same . considering several interesting features of the model , it is quite regrettable that their investigation is a preliminary one in nature with the limitation to the very special class of solutions , and moreover it turns out that the derivation of the special solution faces with crucial problems , which could totally alter the main results ( see sec.[sec:4 ] for the detailed account on this point ) . thus it is meaningful to analyze this model in detail . with these situations in mind , the present paper aims mainly two goals . firstly , we set up a general framework for analyzing cylindrical shell collapses in the same spirit as pereira and wang @xcite , but paying more attention to the mutual relations of three kinds of natural frames of reference . they are ( 1 ) the ortho - normal frame @xmath8 associated with the cylindrical spacetime geometry , ( 2 ) the ortho - normal frame @xmath9 associated with a shell @xmath10 in question ( here @xmath11 is the normal unit 4-vector to the shell @xmath10 , while @xmath12 is the 4-vector tangent to the world - line of an observer located on the shell ) , and ( 3 ) the coordinate frame @xmath13 . computations based on the frame ( 1 ) can be carried out systematically , while the frame ( 2 ) is more suitable for interpreting the results . therefore the bridge between them would be helpful . the frame ( 3 ) has been also considered since it is often convenient for actual computations . the second purpose of this paper is to analyze the above - mentioned model , based on the framework we prepare , and investigate the dynamical behavior of the collapsing shell . we assume that the weak energy condition @xcite is satisfied outside the shell . it means , roughly speaking , that the collapsing matter behaves as normal source of gravity . it shall be proved that , _ the collapsing shell bounces back at some point irrespective of the initial conditions , and escapes from the singularity formation_. the behavior after the bounce depends on the sign of the shell pressure in the z - direction , @xmath0 . when @xmath1 , the shell continues accelerated expansion , and its velocity asymptotically tends to the light velocity . on the other hand , when @xmath14 ( i.e. tension ) , the behavior after the bounce could be more complicated depending on the details of the model . in either case , the shell never reaches the zero - radius configuration , contrary to the result ( the solution ( iii ) mentioned above ) of the preliminary analysis in refs . @xcite and @xcite . in sec.[sec:2 ] , the fundamental formulas on the cylindrical spacetime and the cylindrical shell in it are derived . we in particular pay attention to the extrinsic curvature of the shell , which plays the central role for studying the shell dynamics . the junction conditions , which is another essential part for the shell dynamics , are also investigated there . based on the framework so prepared , then , we investigate in detail a particular model of a collapsing cylinder in sec.[sec:3 ] . in particular , we prove a theorem that the shell never reaches the singularity in this model as far as the weak energy condition is satisfied outside the shell . section [ sec:4 ] is devoted to the detailed comparison between the present analysis and the one reported in refs . @xcite and @xcite , pinpointing why the latter had been led to the virtual singular solution . section [ sec:5 ] is devoted for summary . we consider a spacetime with cylindrical symmetry defined by a line - element , @xmath15 ( by a suitable coordinate transformation @xmath16 , one can actually assume @xmath17 , but we here keep the form eq.([eq : ds ] ) . ) looking at eq.([eq : ds ] ) , the theory of differential forms is efficiently applied @xcite , by introducing natural 1-forms , @xmath18 the local ortho - normal frame @xmath19 is then defined as the dual to the 1-forms @xmath20 . fundamental geometrical quantities , such as curvatures , w.r.t . ( with respect to ) the frame @xmath19 are derived and collected in appendix [ app : a ] for the convenience of later use and future applications . now let @xmath10 be a _ timelike _ 3-surface with cylindrical spatial symmetry embedded in a cylindrical spacetime described by the metric eq.([eq : ds ] ) . due to its symmetry , the surface @xmath10 is characterized by a single equation @xmath21 . let @xmath22 be a proper time of an observer located on the surface with @xmath23 . then , the metric eq.([eq : ds ] ) yields a relation between @xmath24 and @xmath25 , @xmath26 we can use @xmath27 as a coordinate system on @xmath10 , which defines ortho - normal vectors tangent to @xmath10 , @xmath28 and @xmath29 here @xmath30 " indicates the derivative w.r.t . @xmath22 and @xmath31 . a normal unit vector @xmath32 of the surface @xmath10 , which may also be written as @xmath11 or @xmath33 , and its dual one - form are given by , respectively , @xmath34 ( the suffix @xmath35 indicates the frame employed . ) we note that the set of vectors @xmath36 forms the local ortho - normal frame located on @xmath10 . at this stage , let us summarize notations employed throughout this paper . 1 . capital latin indices @xmath37 are reserved for the indices w.r.t . the local ortho - normal frame @xmath19 , and runs from @xmath38 to @xmath39 . einstein s summation convention is adopted throughout the paper . the first few greek letters @xmath40 , @xmath41 , @xmath42 , @xmath43 are reserved for the indices w.r.t . the ortho - normal frame @xmath44 , and take the values @xmath45 , @xmath22 , @xmath46 or @xmath47 ; the later greek letters @xmath48 are used for the indices w.r.t . the coordinate bases @xmath49 . since several frames appear in the paper , we indicate the frame explicitly when necessary . for instance , @xmath50 for a vector and @xmath51 for a 1-form . the symbols @xmath52 , @xmath53 and @xmath54 denote the partial derivatives of a function @xmath55 w.r.t . @xmath56 , @xmath57 and @xmath22 , respectively . in the same manner , @xmath58 indicate the directional derivative of @xmath55 w.r.t . the vector @xmath32 . we here summarize essential formulas frequently used below . we note the relations , @xmath59 therefore it follows @xmath60 for any function @xmath61 defined in the neighborhood of @xmath10 , it follows @xmath62 conversely , we also have @xmath63 each derivative in eqs.([eq : f - formula1 ] ) and ( [ eq : f - formula2 ] ) is understood to be estimated on @xmath10 . we now compute the extrinsic curvature of @xmath10 , which later plays the essential role in analyzing the shell dynamics . the extrinsic curvature is handled mainly in the _ ortho - normal _ frame @xmath65 in this paper . to remind us of it , we write @xmath66 etc . rather than @xmath67 etc . there is a standard formula which relates @xmath68 with the christoffel symbol @xmath69 and the normal vector @xmath32 , @xmath70 where in the last line , eq.([eq : gamma - rel ] ) in appendix [ app : b ] has been used to express the christoffel symbol @xmath71 in terms of @xmath72 since the latter , given in eq.([eq : gamma ] ) , is much easier to compute . let us first compute @xmath66 , @xmath73 the final result is @xmath74 we note that @xmath66 contains the second time - derivative of @xmath75 as is seen in eq.([eq : kttfinal ] ) , so that @xmath66 plays the most important role in the shell dynamics . in view of its essential significance , the detailed derivation of eq.([eq : kttfinal ] ) is shown in appendix [ app : c ] . it is by far easier to compute @xmath76 and @xmath77 . combining eq.([eq : kgeneral ] ) with eqs.([eq : e1 ] ) and ( [ eq : e3 ] ) , we get @xmath78 it is reasonable to compute every quantity in the ortho - normal frame @xmath80 as has been done just above . however , in order to compare the present results with the results in the preceding work ref.@xcite ( see sec.[sec:4 ] ) , it is also needed to represent main quantities in terms of the coordinate frame @xmath81 . after some computations ( see appendix [ app : c ] for more details ) , we finally reach the result , @xmath82 here the @xmath83-derivative in the last term on the r.h.s . ( right - hand side ) has been purposefully left untouched to make the formula look simpler ; the term will vanish whenever @xmath17 , as is the case for the model we analyze later . discontinuity in the extrinsic curvature across a 3-surface implies that the energy - momentum is accumulated on the shell ; this is analogous to the relation of the discontinuity in the electric field across a shell with the electric charge there . geometrical discontinuity across a 3-surface @xmath10 is described by a discontinuity - tensor induced on @xmath10 defined as @xmath84 - h_{\a \b}\left [ k \right ] \ \ , \label{sab}\ ] ] where @xmath85:=q_+ - q_-$ ] , the difference of @xmath86 across the surface ; @xmath87 is the einstein s gravitational constant . then the discontinuity is governed by the junction conditions @xcite , @xmath88\ \ . \label{eq : junction}\ ] ] here @xmath89 is the energy - momentum tensor in the neighborhood of @xmath10 ; @xmath90 and @xmath91 are given in eqs . ( [ eq : n - dual ] ) and ( [ eq : e1 ] ) , respectively , if one refers to the @xmath92-frame . let us now prepare more explicit expression for eq.([eq : junction ] ) . the metric induced on the surface @xmath10 is @xmath93 where @xmath94 , @xmath95 , and @xmath96 form a set of ortho - normal 1-forms on @xmath10 dual to the bases @xmath97 . thus the discontinuity - tensor in this frame becomes @xmath98 + \left [ k_{e_\phi e_\phi } \right ] \right ) = : \eps\ \ , \nonumber \\ s_{e_z e_z } & = & - \frac{1}{\k } \left(\left [ k_{e_\t e_\t } \right ] - \left [ k_{e_\phi e_\phi } \right ] \right ) = : p_z \ \ , \nonumber \\ s_{e_\phi e_\phi } & = & - \frac{1}{\k } \left(\left [ k_{e_\t e_\t } \right ] - \left [ k_{e_z e_z } \right ] \right ) = : p_\phi \ \ , \label{eq : e - p - p } \end{aligned}\ ] ] where @xmath99 , @xmath0 and @xmath100 are interpreted as the energy density of the shell , the pressure in the @xmath46-direction and pressure in the @xmath47-direction , respectively . it is easy to get @xmath101 ( we note @xmath102 forms the non - coordinate bases and , in particular , @xmath103 . ) thus @xmath104 in the @xmath105-coordinate . then ( ii ) in eq.([eq : junction ] ) reduces to @xmath106\ \ , \label{eq : junc-1}\ ] ] with @xmath107 and @xmath91 being given in eq.([eq : n - dual ] ) and eq.([eq : e1 ] ) , respectively . noting that @xmath108 we also get @xmath109 it is straightforward to get @xmath110 and @xmath111 in the @xmath112-coordinate , which is the same expression appears as in eq.([eq : dsab ] ) . therefore , ( ii ) in eq.([eq : junction ] ) reduces to eq.([eq : junc-1 ] ) in the @xmath112-frame also . having prepared all the necessary formulas in the preceding sections , let us now reanalyze the cylindrical collapsing - shell model considered by pereira and wang @xcite . the spacetime to be analyzed is constructed by matching two spacetimes @xmath113 described by the metrics @xmath114 at the timelike surface @xmath10 ( the suffix @xmath115 is for the interior geometry and @xmath116 is for the exterior geometry ) : the @xmath114 are given by @xmath117 where @xmath118 , a function of @xmath119 only : the surface @xmath10 is assumed to be described by @xmath120 . the metric @xmath121 has the form of @xmath122 in eq.([eq : ds ] ) with @xmath123 the @xmath4-energy ( cylindrical energy ) @xcite of this spacetime is given by @xmath124 so that we assume @xmath125 from now on . the interior spacetime is obviously flat , while the exterior geometry is given by , with the help of eq.([eq : ricci ] ) and noting that @xmath126 , @xmath127 , @xmath128 therefore @xmath129 and @xmath130 , or @xmath131 thus @xcite @xmath132 where @xmath133 is a null - vector . combined with the einstein equation , eq.([eq : einsteintensor ] ) implies @xmath134 when the weak energy condition for matter is imposed . since we are interested in the behavior of the collapsing shell in more or less natural " conditions in view of the cosmic censorship conjecture @xcite , it is reasonable to assume @xmath125 and @xmath134 from now on . it may be appropriate to make some remarks on the energy condition . without any condition imposed on matter , any strange spacetime would become possible . indeed , one can always claim that arbitrary spacetime is the solution of the einstein equation if the matter with energy - momentum tensor @xmath135 is prepared . therefore it is essential to assume some kind of energy condition to make any meaningful arguments @xcite . the weak energy condition is one of such appropriate conditions widely accepted , so we assume the condition here to make the definite arguments below . of course , there is still a room of considering the violation of the energy condition at the shorter scale , due to the casimir effect for instance . though such a possibility is certainly interesting to study , we do not consider it in this paper for definiteness and only investigate the purely classical situations . we now impose the junction conditions eq.([eq : junction ] ) for two geometries @xmath136 : .7 cm .4 cm we impose @xmath137 when @xmath138 , it implies @xmath139 where @xmath140 . by plugging the relation @xmath141 into the above relation and by solving the latter w.r.t . @xmath142 , we get an important formula @xmath143 taking the @xmath56-derivative of the above relation , one finds @xmath144 noting the relation , @xmath145 one also finds @xmath146 .7 cm .4 cm let us find the expressions for @xmath147 ( eqs.([eq : kzzfinal ] ) , ( [ eq : kppfinal ] ) and ( [ eq : kttfinal2 ] ) ) in this model , and then derive the formulas for @xmath148 , @xmath0 and @xmath100 ( eq.([eq : e - p - p ] ) ) . first of all , the factor @xmath149 becomes @xmath150 implying that @xmath151 then , noting the relation @xmath152 , eq.([eq : kttfinal2 ] ) with @xmath153 yields @xmath154 where eq.([eq : rhominus2dot ] ) has been used in the last line . noting also @xmath155 one finds @xmath156&= & -\frac{e^{-\g_+}}{\delta ( 1-{\dot{\rho}_+}^2)^{3/2 } } \left\ { ( \delta - 1)\ddot{\rho}_+ - \g_+ ' \dot{\rho}_+ ( 1 - \dot{\rho}_+)[(\delta + 1 ) \dot{\rho}_+ + 1 ] \right\ } \ \ , \\ \left [ k_{e_z e_z } \right ] & = & 0 \ \ , \\ \left [ k_{e_\phi e_\phi } \right ] & = & - e^{-\g_+ } \frac{\delta - 1 } { \rho_+ \sqrt{1-{\dot{\rho}_+}^2 } } \ \ .\end{aligned}\ ] ] equation ( [ eq : e - p - p ] ) then yields @xmath157 -\delta ( \delta -1 ) \frac{1-{\dot{\rho}_+}^2}{\rho_+ } \right\ } \ \ , \label{eq : pz - result } \\ \k p_\phi & = & \frac{e^{-\g_+}}{\delta ( 1-{\dot{\rho}_+}^2)^{3/2 } } \left\ { ( \delta - 1)\ddot{\rho}_+ -\g_+ ' \dot{\rho}_+ ( 1 - \dot{\rho}_+)[(\delta + 1 ) \dot{\rho}_+ + 1 ] \right\ } \ \ , \label{eq : pphi - result } \end{aligned}\ ] ] with @xmath158^{1/2 } $ ] . noting the relation @xmath159 , we see @xmath160 when @xmath161 , @xmath162 and @xmath163 , as it should be . setting @xmath164 eq.([eq : pz - result ] ) yields the dynamical equation of the shell , @xmath165 + \frac{\delta}{\delta-1}(1-\dot{\rho_+}^2 ) ^{3/2 } e^{\g_+ } p_0 \ \ . \label{eq : result1}\ ] ] to make the statements concise , we indicate the first , the second and the third terms on the r.h.s . of eq.([eq : result1 ] ) by the symbols , @xmath166 $ ] , @xmath167 $ ] and @xmath168 $ ] , respectively . the term @xmath168 $ ] indicates that when the shell has a positive pressure in the @xmath46-direction ( @xmath169 ) , it causes more acceleration in the expanding direction , while the tension of the shell ( @xmath170 ) works to reduce the acceleration . however the term @xmath168 $ ] does not change the whole dynamics so much : the term could become important only when @xmath171 , i.e. when @xmath172 . setting @xmath173 ( @xmath174 ) , the term behaves as @xmath175 , so that the contribution of the term is not significant . indeed , we shall show below a theorem that under the dynamical equation eq.([eq : result1 ] ) , the shell never reaches the singular configuration @xmath176 irrespective of the initial conditions nor the value of @xmath177 . first of all , we note the following obvious fact : * lemma 1 * + _ consider the case @xmath178 . once the shell satisfies the condition @xmath179 , it expands forever afterwards . _ + noting that @xmath180 and that the terms @xmath167 $ ] and @xmath168 $ ] are non - negative when @xmath179 , it follows that @xmath181 , so that the claim follows . in particular , when @xmath182 , eq.([eq : result1 ] ) becomes @xmath183 , so that the effective potential behaves like @xmath184 in the region where @xmath185 does not change so drastically . compared with the dust case @xcite , where @xmath186 ( @xmath187 , @xmath188 are constants ) , the tendency of expansion of the present model is obvious . next , let us investigate the contracting shell . to avoid any confusion , let us set @xmath189 , so that @xmath190 . then eq.([eq : result1 ] ) reads @xmath191 + \frac{\delta}{\delta -1}(1-v^2)^{3/2 } e^{\g_+ } p_0 \ \ , \label{eq : result1b}\ ] ] with @xmath192 ^{1/2}$ ] . the present model has been constructed by giving metrics eqs . ( [ eq : ds- ] ) and ( [ eq : ds+ ] ) , and there is no particular physical image regarding the matter content forming the shell . however , one can infer that the term @xmath166 $ ] comes from the angular momentum effect since it is a decreasing function of @xmath193 . on the other hand , the term @xmath167 $ ] may be identified with what corresponds to newtonian gravity under suitable conditions , since it is proportional to @xmath194 . let us pay attention to the last factor in the term @xmath167 $ ] , @xmath195 . the function @xmath196 is obviously monotonic , increasing , continuous function of @xmath197 , with @xmath198 and @xmath199 , so that @xmath196 has only one zero in @xmath200 . when the weak energy condition is satisfied , and the shell is not moving so fast ( @xmath201 ) , then @xmath202 , so that the term @xmath167 $ ] yields an attractive force , which is consistent with the attractive nature of newtonian gravity . a curious feature arises , however , when @xmath203 . then @xmath204 , and the same term now yields a repulsive force . due to this characteristic behavior of @xmath196 , the shell bounces back at some point , and never collapses to singularities as we shall see soon . before proving this statement , let us first get rough understanding on what happens when @xmath193 becomes small . the term @xmath166 $ ] is the repulsive force , produced by an infinite effective potential barrier near @xmath176 ( angular momentum effect ) . in case the shell would have ever reach @xmath176 , it should have been this repulsive effect becomes negligible compared to other terms , i.e. when @xmath205 approaches @xmath206 ( @xmath207 ) more rapidly than @xmath193 approaches to zero . however , in this case , the term @xmath167 $ ] causes a very strong repulsive effect since @xmath208 and @xmath171 , while the term @xmath168 $ ] becomes negligible . this behavior can be analyzed in more detail as follows . let us set @xmath209 ( @xmath174 ) . then @xmath210 , and one finds @xmath166 \sim \frac{\d}{\rho_+}$ ] , @xmath167\sim \d^{-1}$ ] , and @xmath168\sim \d^{1/2}$ ] , so that @xmath211 therefore , when @xmath212 , then the term @xmath167 $ ] causes a very strong repulsive effect . in summary , the shell never reaches @xmath176 due to the strong repulsive force caused by the term @xmath166 $ ] ( when @xmath197 is not so large ) , or by the term @xmath167 $ ] ( when @xmath203 ) . to make the statements below as concise as possible , let us first define the core - region " . consider the phase space @xmath213 . it is understood that suitable topology is endowed on @xmath69 . ) . the detailed topological property is not very important in the argument below , so that it suffices to imagine @xmath69 as @xmath214 . ] at each time @xmath56 , the region @xmath215 is defined such as ( i ) @xmath167+[[3 ] ] > 0 $ ] and @xmath216 on @xmath215 and ( ii ) @xmath215 is connected , and contains a neighborhood of @xmath217 . in other words , @xmath215 consists of phase points that are close enough " to the singularity - emerging point @xmath217 . * definition * + we call the above - mentioned region @xmath215 in @xmath69 _ the core - region _ ( at @xmath56 ) , for brevity . we also say that the shell enters ( or , leaves ) the core - region at @xmath218 " if at @xmath219 the shell s phase point is outside ( or , inside ) the core - region , and at @xmath218 , the phase point of the shell is on the edge of the core - region ; ( @xmath167+[[3 ] ] = 0 $ ] and @xmath220 ) . it is clear ( from the consideration of the order of magnitude of the terms @xmath166 $ ] , @xmath167 $ ] and @xmath168 $ ] when @xmath203 ) that , if the shell would have ever reached the singularity , the phase point of the shell should have reached the point @xmath217 _ through the core - region_. now we can show * lemma 2 * + _ under the dynamical equation eq.([eq : result1 ] ) , once the shell enters the core - region at @xmath218 , it bounces back without reaching zero - radius . indeed , @xmath221 is bounded from below as @xmath222 _ + noting that @xmath223 when the shell is in the core - region , it suffices to show the claim holds even for @xmath224 with arbitrary initial conditions @xmath225 and @xmath226 ( where @xmath227 and @xmath228 ) . however , its solution is given by @xmath229 , so that @xmath230 . thus the claim follows . * theorem 1 * + _ under the dynamical equation ( with or without @xmath177 ) eq.([eq : result1 ] ) , the shell never reaches @xmath231 irrespective of its initial conditions . _ + it suffices only to consider the contracting phase . then if the shell would have ever reached @xmath176 , the phase point of the shell should have reached the point @xmath217 through the core - region . however , _ lemma 2 _ indicates that this never happens . thus _ the line - like singularity never forms in this model irrespective of the value @xmath177 , as far as the weak energy condition is satisfied outside the shell_. the behavior in the expanding phase ( and/or after the bounce ) varies depending on the value @xmath177 . when @xmath232 , @xmath233 in the expanding phase , so that the shell continues to be accelerated and expand , and its velocity asymptotically approaches to the light velocity . on the other hand , the case @xmath234 , the term @xmath168 $ ] is negative , thus it might be possible that @xmath235 turns to negative when the shell velocity is not very large . therefore , the oscillating behavior can arise , which might be interesting to investigate further . ( however , even in this case , the shell never collapses to zero - radius due to _ theorem 1_. ) once the shell velocity becomes sufficiently large in the expanding phase , however , the shell continues to be accelerated and expands forever just as the case of @xmath232 . this is because , when the shell velocity becomes close to 1 , by setting @xmath236 ( @xmath237 , @xmath238 ) , the term @xmath167 $ ] behaves as @xmath239 and become dominant , while the terms @xmath166 $ ] and @xmath167 $ ] behave as @xmath240 . now let us illustrate what has been proved rigorously above by some numerical investigations . we set @xmath241 for simplicity . figure 1 shows a typical behavior of the contracting shell ( thinner curve ) along with @xmath100 ( thicker curve ) . as an example , we have set @xmath242 . initial conditions have been set @xmath243 and @xmath244 ( i.e. very close to the light velocity ) . the bouncing behavior discussed above is obviously seen in the figure . it is also seen that the turning point of the shell coincides with the peak of the rotational pressure . figure 2 shows the behavior of the same model with the initial conditions @xmath243 and @xmath245 ( i.e. relatively mild contraction ) . again , evolutions of @xmath221 ( thinner curve ) and @xmath100 ( thicker curve ) are indicated . due to the mild initial conditions , the shell soon bounces back . just as the previous case , the turning point of the shell coincides with the peak of the rotational pressure . noting that @xmath246 = 0 $ ] , eq.([eq : e - p - p ] ) with @xmath247 implies @xmath248 , so that @xmath249 this relation provides another confirmation of the validity of the condition @xmath250 ( see eq.([eq : g - delta - eps ] ) ) . finally let us note that the relation eq.([eq : result2 ] ) reveals a curious character of the model . we first consider the case @xmath251 . in the case of the counter - rotating dust cylinder , there is a relation @xmath252 , where @xmath253 is the velocity of the dust particle relative to an observer standing still " on the shell @xcite . comparing this with eq.([eq : result2 ] ) ( @xmath241 ) , the present model corresponds to the case of @xmath254 , so that the present model could be interpreted as a cylinder shell consisting of the counter - rotating massless particles , though the two models are not exactly the same . the situation in which massless particles are confined to a compact region arises naturally in superconducting objects , so that the present model should have some relevance . on the other hand , when @xmath2 , it behaves more or less like a normal matter and it is consistent with the attractive nature of the term @xmath168 $ ] . on the other hand , when @xmath255 , eq.([eq : result2 ] ) rather looks like a cosmic string . however , the cylinder tends to expand in the present case , contrary to the case of cosmic strings . it may not appropriate at present to pursue such comparisons too further . in view of our results above , we need to investigate whether the singularity - forming solution claimed in ref.@xcite and its correction ref.@xcite is a valid one . it turns out that the formal solution in question is not a relevant one and should be discarded . by comparing one by one our formulas with the corresponding ones shown in refs.@xcite and @xcite , it turns out that some formulas in ref.@xcite along with its correction ref.@xcite still need slight modifications ( see below and appendix [ app : d ] for more details ) . however , we first derive the above conclusion faithfully based on the formulas in refs.@xcite and @xcite . then we also show that the same conclusion follows based on our formulas , too , even with the @xmath177-term . firstly their functions @xmath256 , @xmath257 , @xmath258 , @xmath259 , @xmath260 , @xmath261 , @xmath262 and @xmath263 correspond to ours as @xmath264 in particular , let us note that the sign of the function @xmath188 is opposite to our function @xmath42 . thus we should understand that @xmath265 and @xmath266 . they correspond to the physical conditions @xmath267 and @xmath134 ( see arguments after eq.([eq : c - energy ] ) and eq.([eq : einsteintensor ] ) ) . the detailed comparison is shown in appendix [ app : d ] . here it suffices to pay attention only to their dynamical equation shown in ref.@xcite ( which is the correction of the original eq.(30 ) of ref.@xcite ) : @xmath268 where @xmath269 and its derivatives correspond to our @xmath221 and its @xmath56-derivatives ( see eq.([eq : notationmatch ] ) ) , and @xmath270 corresponds to our @xmath271 ; @xmath272 ^{1/2}$ ] . if the terms are rearranged to look as similar to eq.([eq : result1 ] ) ( with @xmath241 ) as possible , their formula eq.([eq : pw - dynamical1 ] ) reads @xmath273 ( compared with eq.([eq : result1 ] ) ( @xmath241 ) , the factors @xmath274 and @xmath275 of the second term should be replaced by @xmath276 and @xmath277 , respectively . ) based on eq.([eq : pw - dynamical1 ] ) , ref . @xcite ( just as ref . @xcite ) tries to search for special solutions for @xmath269 satisfying @xmath278 , \label{eq : pw - special2 } \end{aligned}\ ] ] where @xmath41 is assumed to be an arbitrary _ constant_. as is claimed in ref.@xcite , eq.([eq : pw - special1 ] ) itself is formally satisfied by @xmath279 or integrating once more , @xmath280 \big\}\ \ . \label{eq : pwsol}\end{aligned}\ ] ] here two integral constants @xmath281 and @xmath282 have appeared corresponding to the fact that the differential equation eq.([eq : pw - special1 ] ) is of the 2nd order . the constant @xmath281 indicates the time when @xmath276 vanishes , while @xmath282 ( along with @xmath281 ) determines the time @xmath283 when @xmath284 vanishes ( @xmath285 ) . these constants @xmath281 and @xmath282 are ( indirectly ) determined by the initial conditions for solving eq.([eq : pw - special1 ] ) . equation ( [ eq : pwsol ] ) would imply a line - like singularity formation , reaching @xmath286 at some finite time @xmath283 , so that its implication would be significant . we show below , however , that _ eq.([eq : pwsol ] ) is not the solution of the original differential equation eq.([eq : pw - dynamical2 ] ) _ , irrespective of the parameters @xmath281 and @xmath282 , i.e. independent of the initial conditions . to see this conclusion , let us first define @xmath287 to be @xmath206 times the r.h.s . of eq.([eq : pw - special2 ] ) , @xmath288 . \label{eq : b(t)}\ ] ] what we show below is that @xmath287 varies ( actually _ diverges _ ) along the collapsing solution eq.([eq : pwsol ] ) so that the starting assumption @xmath41=constant ( eq.([eq : pw - special2 ] ) ) is incompatible with the dynamical equation eq.([eq : pw - special1 ] ) . causing such self - inconsistency , then , eq.([eq : pwsol ] ) is not the solution of the original dynamical equation , eq.([eq : pw - dynamical2 ] ) . now we prove * theorem 2 * + _ the quantity @xmath289 diverges , and does not remain constant along the collapsing solution eq.([eq : pwsol ] ) . _ + let us look at the definition of @xmath287 ( eq.([eq : b(t ) ] ) ) . both of the terms in the brackets @xmath290 $ ] are positive , since @xmath291 , @xmath292 and @xmath293 . thus @xmath294 and accordingly , eq.([eq : pw - special2 ] ) implies that @xmath295 . we then get the estimation @xmath296 so that @xmath297 where @xmath283 is the shell - collapsing time , @xmath298 . + , the above estimation implies that @xmath299 as @xmath300 . + , the estimation further goes as @xmath301 where de lhopital s theorem has been applied in the last step . with the help of eq.([eq : pw - special1 ] ) , then , we get @xmath302 thus we get @xmath303 . the situation is the same for our correct equation , eq.([eq : result1 ] ) ( with @xmath169 , @xmath304 or @xmath305 ) , too . just as eqs.([eq : pw - special1 ] ) and ( [ eq : pw - special2 ] ) , we can ask whether the collapsing solution is allowed for the equation @xmath306\ \ , \label{eq : my - special2 } \end{aligned}\ ] ] with @xmath307 being a _ constant _ ; @xmath308 ^{1/2}$ ] . equation ( [ eq : my - special1 ] ) itself has a formal solution , which is just the same as eqs.([eq : pwsol0 ] ) and ( [ eq : pwsol ] ) . though more care is required , essentially the same argument as above follows and it turns out that the formal solution just like eq.([eq : pwsol ] ) is not the solution of the original dynamical equation eq.([eq : result1 ] ) . to show this fact , we first introduce some symbols . * definition * + \(a ) let @xmath309 be @xmath206 times the r.h.s . of eq.([eq : my - special2 ] ) ; @xmath310 \nonumber \\ & & \ + \frac{\delta}{\delta -1}(1-v^2)^{1/2 } e^{\g_+ } p_0 \big ] \ \ , \label{eq : tb(t)}\end{aligned}\ ] ] where @xmath311 ( @xmath312 ) . \(b ) let @xmath313 and @xmath314 be the second and the third terms in the brackets @xmath290 $ ] in eq.([eq : tb(t ) ] ) , respectively ; @xmath315 \ \ . \\ \te(t):&= & \frac{\delta}{\delta -1}(1-v^2)^{1/2 } e^{\g_+ } p_0 \ \ .\end{aligned}\ ] ] \(c ) let @xmath316 . the sign of @xmath196 determines the sign of @xmath317 . as is already discussed in the previous section ( after _ lemma 1 _ ) , the function @xmath196 is monotonic , increasing , continuous on @xmath318 $ ] , with @xmath198 and @xmath319 , so that @xmath196 has only one zero , say @xmath320 , in @xmath200 . when @xmath321 is looked as a function of @xmath197 , then , @xmath322 for @xmath323 $ ] and @xmath324 for @xmath325 . \(d ) let @xmath326 be the shell - collapsing time ( which is assumed to exist in the present context ) , i.e. @xmath326 is the time such that @xmath327 . ( it does not matter whether @xmath326 is finite or infinite in the argument below . ) we now prove a theorem ( _ theorem 3 _ below ) corresponding to _ theorem 2 _ step by step . we first note * lemma 3 * + _ @xmath328 as @xmath207 . _ + when @xmath197 is close to 1 , setting @xmath329 ( @xmath174 ) , then @xmath210 , and one finds @xmath330 and @xmath331 , so that @xmath332 . then the claim follows . now let us confirm * lemma 4 * + _ it follows that @xmath333 . _ + studying the solution of eq.([eq : my - special1 ] ) ( cf . eq.([eq : pwsol0 ] ) ) , we see @xmath334 ( i.e. @xmath335 ) . suppose @xmath336 . since @xmath196 is continuous on @xmath318 $ ] , then , it means @xmath337 , so that @xmath338 as well as @xmath339 as @xmath340 ( note @xmath293 ) . considering the limiting behavior of @xmath341 $ ] as @xmath340 , _ lemma 3 _ implies @xmath342 as @xmath340 . since @xmath307 ( @xmath343 ) is a constant by assumption , this should not happen . thus @xmath344 . * corollary of lemma 4 * + _ the quantities @xmath317 and @xmath345 are bounded , @xmath346 and @xmath347 . furthermore , @xmath348 and @xmath349 . _ + we note that @xmath313 and @xmath314 become unbounded only when @xmath207 and that @xmath313 and @xmath314 are continuous as functions of @xmath197 . now , the solution of eq.([eq : my - special1 ] ) ( cf . eq.([eq : pwsol0 ] ) ) implies @xmath334 . furthermore _ lemma 4 _ says @xmath333 . then the claim follows . then we can show * lemma 5 * + _ the constant parameter @xmath307 should be set negative , @xmath350 _ + we note that @xmath351 and @xmath352 due to _ corollary of lemma 4_. on the other hand , @xmath353 . then one can find appropriate @xmath354 and @xmath355 ( @xmath356 ) such that @xmath357 on @xmath358 . thus @xmath359 on @xmath360 . since @xmath307 ( @xmath361 ) is a constant by assumption , this means that @xmath307 should be set negative from the outset . finally we prove * theorem 3 * + _ the quantity @xmath362 diverges , and does not remain constant along the collapsing solution ( corresponding to eq.([eq : pwsol ] ) ) . _ + let us set @xmath363 $ ] . we note @xmath364 @xmath365 due to eq.([eq : tb(t ) ] ) . then it follows @xmath366 . noting @xmath367 ( _ lemma 5 _ ) , then , exactly the same argument as the proof of _ theorem 2 _ can be applied , and one concludes @xmath368 . on the other hand , it follows @xmath369 due to _ corollary of lemma 4 _ , then , it follows @xmath370 . we conclude that there is no relevant solution indicating the singularity formation in the present model . in this paper , we have studied the dynamics of a cylindrical shell in the spacetime cylindrical symmetry . we have started with constructing a general framework for analyzing a cylindrical spacetime and a shell in it , which might be useful for future investigations . based on the framework , we have investigated a cylindrical shell - collapse model which accompanies the out - going radiation of gravitational waves and massless particles . this model had been introduced by pereira and wang @xcite , but its proper analysis had been awaited . this model could be interpreted as a thin shell filled with radiation . in connection with the cosmic - censorship hypothesis , we are mostly interested in the collapse of normal " matter , so that the weak energy condition has been assumed outside the shell . it has been proved that , as far as the weak energy condition is satisfied outside the shell , the shell bounces due to the rotational - pressure effect . after the bounce , it continues to expand without re - contraction when the pressure of the shell in the z - direction @xmath0 satisfies @xmath371 , while in the case @xmath14 , the behavior after the bounce can be more complicated . however in either case , the shell never reaches the zero - radius configuration and it escapes from the line - like singularity formation . just as the case of a shell filled with counter - rotating dust particles considered by apocatopolas and thorne @xcite , the present case also shows a bouncing due to the effect of rotational pressure , while the eternal expansion after the bounce ( when @xmath1 ) is a unique feature of the present model . we have also performed numerical investigations which reveal explicit behaviors of the shell . the author would like to thank s. jhingan for valuable discussions at the beginning stage of this work . he also thanks h. sakuragi for valuable discussions on numerical investigations . important part of work has been completed during the author s stay at institute of cosmology , tufts university . he is grateful for its nice hospitality . in particular he would like to thank a. vilenkin for helpful suggestions on the interpretation of the results . this work has been supported by the japan ministry of education , culture , sports , science and technology with the grant # 14740162 . based on the metric eq.([eq : ds ] ) , it is convenient to introduce the 1-forms as eq.([eq : cartan ] ) . taking the exterior derivative of @xmath372 s , we get @xmath373 here @xmath374 " indicates the partial derivative w.r.t . @xmath56 , while @xmath375 " means the same w.r.t . @xmath57 . with the help of the first cartan structure - equation @xmath376 , we thus get @xmath377 on account of the relation , @xmath378 , we also get forms the _ non - coordinate bases _ and @xmath379 s are not tensors , so that there is no surprise in that @xmath380 @xmath381 . ] @xmath382 here @xmath30 " indicates the derivative w.r.t . the proper - time @xmath83 of an observer on the shell , and @xmath31 ; they are evaluated on the shell in consideration . taking the exterior - derivatives of @xmath383 s , we get @xmath384 thus , by the second cartan structure - equation @xmath385 , we find @xmath386 using the relation @xmath387 , we obtain the riemann curvature w.r.t . the frame @xmath388 , @xmath389 thus , the ricci curvature in the frame @xmath390 becomes @xmath391 finally , the scalar curvature becomes , independently of the frame , @xmath392 throughout this paper , we always consider a cylindrical spacetime defined by the metric eq.([eq : ds ] ) . the coordinates @xmath393 define the coordinate bases and their dual 1-forms , @xmath394 ( here the greek letters @xmath48 indicate the indices w.r.t . the coordinates @xmath395 . ) on the other hand , a cylindrical shell embedded in the spacetime naturally defines a local ortho - normal frame @xmath396 ( see sec.[sec:2 - 1 ] ) , @xmath397 where the symbol @xmath398 implies the normalized frame - vector . ( the first few greek letters @xmath40 , @xmath41 , @xmath42 , @xmath43 are used for the indices w.r.t . the above ortho - normal frame , and they take the values @xmath45 , @xmath22 , @xmath46 or @xmath47 . ) a set of 1-forms @xmath399 dual to @xmath396 is also introduced . as the third set of frames , the metric given in eq.([eq : ds ] ) defines a natural set of 1-forms @xmath400 , @xmath401 along with their dual ortho - normal bases @xmath388 . ( the capital latin letters @xmath402 , @xmath403 , @xmath4 , @xmath43 indicate the indices w.r.t . the frame determined by eq.([eq : frame3 ] ) , and they take the values @xmath404 . ) now we summarize the relations among @xmath405 , @xmath406 and @xmath407 , and those among @xmath408 , @xmath409 and @xmath410 . firstly , eq.([eq : ehat ] ) gives the relation between @xmath405 and @xmath406 , and those between @xmath408 and @xmath409 @xmath411 here @xmath91 and @xmath412 defined below are mutually inverse as matrices : @xmath413 secondly , eq.([eq : frame3 ] ) gives the relation between @xmath405 and @xmath407 , and those between @xmath408 and @xmath410 @xmath414 here @xmath415 and @xmath416 defined below are mutually inverse as matrices : @xmath417 finally , eqs.([eq : e1 ] ) and ( [ eq : e2 ] ) give the relation between @xmath407 and @xmath406 , and those between @xmath410 and @xmath409 @xmath418 here @xmath419 and @xmath420 given below are mutually inverse as matrices : @xmath421 now noting the relation @xcite , @xmath422 along with eq.([eq : e3 ] ) , we get @xmath423 equation ( [ eq : gamma - rel ] ) is the basis for calculating the extrinsic curvature . here derivations of the formula for the extrinsic curvatures ( eqs.([eq : kttfinal ] ) and ( [ eq : kttfinal2 ] ) ) are purposefully shown in detail . this is to ensure anyone interested in this topic reproducing all the results presented in this paper without any difficulty and to motivate him to go ahead the point where this paper ends . in order to derive eq.([eq : kttfinal ] ) , we note that the first term on the r.h.s . in eq.([eq : ktt ] ) becomes , with the help of eqs.([eq : gamma ] ) and ( [ eq : e3 ] ) , @xmath424 \right\ } \ \ . \label{eq : calc1}\end{aligned}\ ] ] similarly , with the help of eqs.([eq : e1 ] ) and ( [ eq : e3 ] ) , the second term on the r.h.s . in eq.([eq : ktt ] ) is calculated as @xmath425 the above expression can be further modified with the help of the following formulas that can be derived straightforwardly ; @xmath426 after some calculations we then get @xmath427 the last two terms together yield @xmath428 \\ & & \ \ \ \ \ + \frac{r}{2xt } { \rhodot}^{\ : 2 } \left [ xtr\ \del_n ( \ln r^2 ) - r^2 \rhodot\ ( \ln r^2)^\circ \right ] \\ & & = : ( v_1)+(v_2)+(v_3)+(v_4)\ \ , \end{aligned}\ ] ] where the formulas in eq.([eq : f - formula2 ] ) have been used . noting that @xmath429 , thus , the calculations go as @xmath430 yielding eq.([eq : kttfinal ] ) . now we derive eq.([eq : kttfinal2 ] ) . to start with , let us note the following formula derived straightforwardly from eq.([eq : rhodot ] ) : @xmath431 taking the @xmath56-derivative of the relation @xmath432 ( eq.([eq : rhodot ] ) ) , then , we obtain , @xmath433 we should return to the basic equation eq.([eq : ktt ] ) along with the calculations eqs.([eq : calc1 ] ) and ( [ eq : calc2 ] ) . noting the relations @xmath434 we finally reach eq.([eq : kttfinal2 ] ) . it is easily checked that their result is translated into our notation as @xmath439 where the @xmath22-derivative in the last term is for making the expression concise , and can be expressed in terms of @xmath440 and @xmath441 " if needed . comparing eq.([eq : k(pw2 ) ] ) with eq.([eq : kttfinal2 ] ) , we realize , @xmath442 here we have made use of eq.([eq : f - formula2 ] ) to reach the last line in eq.([eq : k(pw)comparison ] ) ; @xmath443 on the other hand , their expressions for the @xmath444-component and the @xmath445-component of the extrinsic curvature , @xmath446 and @xmath447 , ( see the second and the third formulas in eq.(10 ) of ref.@xcite ) are related to our @xmath76 and @xmath77 as @xmath448 it turns out that @xmath446 and @xmath447 are totally consistent with our @xmath76 and @xmath77 , taking into account that the former is based on the coordinate frame @xmath81 , while the latter is based on the ortho - normal frame @xmath449 . [ on the other hand , @xmath437 is the @xmath450-component ( and not @xmath451-component ) , and @xmath452 is a member of the ortho - normal frame @xmath449 . therefore @xmath437 is what should correspond to our @xmath66 ( up to sign ) without any further conversion factor . ] their expression eq.([eq : k(pw ) ] ) can be modified so as to coincide with our result @xmath66 ; @xmath453 comparing eq.([eq : k(pw)correct ] ) with eq.([eq : k(pw ) ] ) , we realize that the term @xmath454 in eq.([eq : k(pw ) ] ) should be removed and that the numerical factor 2 " in front of the factor @xmath455 should be corrected to 1 " . the expression for @xmath456 ( corresponding to our @xmath142 ) in ref.@xcite has been corrected in ref.@xcite , and the new expression coincides with our eq.([eq : dtdt ] ) . however , the expressions for @xmath457 and @xmath458 ( corresponding to @xmath459 and @xmath460 , respectively ) in eq.(28 ) of ref.@xcite are not mentioned in ref.@xcite ; they actually should be corrected as our eqs.([eq : dt2dt2 ] ) and ( [ eq : rhominus2dot ] ) , respectively . the expression for @xmath0 has been corrected in ref.@xcite as however , it still needs modification ( see our eq.([eq : pz - result ] ) ) , which is equivalent to replace the last two factors @xmath462 in eq.([eq : pw - pz ] ) with @xmath463 $ ] . the new expression of @xmath0 influences the dynamical equation for the shell , since the latter is obtained by setting @xmath247 ( they only consider the case @xmath247 ) . the rest is discussed in sec.[sec:4 ] . 99 for the basic accounts on the spherical collapses , see , e.g. p.s . joshi , _ global aspects in gravitation and cosmology _ ( clarendon , oxford , 1993 ) . k. s. thorne , in _ magic without magic : john archibald wheeler _ ( j. klauder ( ed . ) , freeman , san francisco , 1972 ) , p.1 . s. l. shapiro and s. a. teukolsky , phys . lett . * 66 * , 994 ( 1991 ) . r. penrose , riv . nuovo cimento * 1 * ( numero special ) , 252 ( 1969 ) . t. a. apostolatos and k. s. thorne , physical review d*46 * , 2435 ( 1992 ) . p. r. c. t. pereira and a. wang , physical review d*62 * , # 124001 ( 2000 ) . p. r. c. t. pereira and a. wang , physical review d*67 * , # 129902(e ) ( 2003 ) . t. piran , phys . lett . , 1085 ( 1978 ) . w. israel , il nuovo cimento * b44 * , 1 ( 1966 ) ; * b48 * , 463(e ) ( 1967 ) . see , for instance , r. m. wald , _ general relativity _ 9 - 2 . k. s. thorne , phys . rev . * 138 * , b251 ( 1965 ) . see , for instance , m. nakahara , _ geometry , topology and physics _ ( institute of physics publishing , bristol , 1990 ) , chapter 7 .

the substantial decrease in star formation density from to the present day is curious given the relatively constant neutral gas density over the same epoch . future radio astronomy facilities , including the ska and pathfinder telescopes , will provide pioneering measures of both the gas content of galaxies and star formation activity over cosmological timescales . here we investigate the commensalities between neutral atomic gas ( h ) and radio continuum observations , as well as the complementarity of the data products . we start with the proposed h and continuum surveys to be undertaken with the ska precursor telescope meerkat , and building on this , explore optimal combinations of survey area coverage and depth of proposed h and continuum surveys to be undertaken with the ska1-mid instrument . intelligent adjustment of these observational parameters results in a tiered strategy that minimises observation time while maximising the value of the dataset , both for h and continuum science goals . we also find great complementarity between the h and continuum datasets , with the spectral line h data providing redshift measurements for gas - rich , star - forming galaxies with stellar masses to , a factor of three lower in stellar mass than would be feasible to reach with large optical spectroscopic campaigns . [ firstpage ] surveys galaxies : general galaxies : evolution radio continuum : galaxies radio lines : galaxies

1605.04962

the dramatic decrease in the cosmic star formation density from @xmath0 to the present day has been well documented ( see @xcite , for example ) . conversely , the neutral gas content ( h ) of galaxies spanning the same redshift range is much less well known . at @xmath5 , large area surveys such as the arecibo legacy fast alfa survey ( alfalfa , @xcite ) and the h parkes all - sky survey ( hipass , @xcite , @xcite ) provide direct detections of the cold neutral gas in galaxies , from which the h density of the local universe can be derived . targeted studies , such as the blind ultra - deep h environmental survey ( budhies , @xcite ) , or the highz survey @xcite , have used many hundreds of hours of observing time to detect several galaxies in h at @xmath6 , but the detected objects are not representative of the general galaxy population , being either located in high density environments , or lying at the upper end of the galaxy mass distribution . an average measurement of the h content of galaxies not individually directly detected can be derived from spectral stacking , which has provided estimates of the cosmic hdensity to @xmath7 ( @xcite , and @xcite , for example ) , and can be extended to even higher redshifts . at high ( @xmath8 ) redshifts , the neutral gas content of galaxies and their environments is detected in absorption along the line - of - sight toward background quasars , resulting in damped ly@xmath9 systems ( dlas ) , lower column density lyman limit systems and the ly@xmath9 forest . the large numbers of these systems currently known ( several thousands , see @xcite ) gives an indirect measurement of the gas content of galaxies at redshifts beyond the reach of current radio telescopes . the lower redshift limit of @xmath10 is set by the ly@xmath9 absorption appearing in the ultra - violet , necessitating expensive space - based spectroscopy . as h provides the fuel for star formation , a connection between star formation activity and the neutral gas reservoirs available to feed the activity is expected . an intriguing detail is that while the level of star formation has dropped by more than an order of magnitude from @xmath0 to the local universe , the difference between the hdensity at @xmath11 measured from h in emission to @xmath8 measured from h in absorption has only changed by a factor of a few ( see @xcite , for example ) . the nearly constant h density since @xmath0 indicates replenishment of the gas reservoir , as the gas consumption timescales are relatively short in comparison . this discrepancy also points to at least one additional ingredient required to connect the reservoir of fuel to stellar buildup . molecular gas , as the intermediary between neutral gas and star formation , is an obvious line of investigation . alma , with its exceptional sensitivity , is revolutionizing the study of molecular gas , both at low and high redshift . observing h in emission spanning the redshift range @xmath12 is within the capabilities of the next generation of radio telescopes , and is one of the key science cases for the square kilometre array ( ska , @xcite ) . in the meantime , existing facilities , including the recently upgraded karl g. jansky very large array ( vla ) , and the aperture tile in focus ( apertif ) focal - plane array system planned for deployment on the westerbork synthesis radio telescope in 2016 @xcite , will allow measurement of the h content of galaxies to @xmath13 and @xmath14 , respectively . a deep h survey , the cosmos h large extragalactic survey ( chiles ) is indeed already underway with the vla ( see section [ sec : chiles ] ) . in preparation for the ska , mid - frequency pathfinder radio telescopes currently under construction including south africa s meerkat ( @xcite ) and the australian square kilometre array pathfinder ( askap , @xcite ) , will improve on the capabilities of existing facilities . thus , we will be able to begin exploring the gap in redshift , @xmath12 , corresponding to lookback times up to nearly 8 gyr , which has thus far remained almost entirely inaccessible with respect to the neutral gas content of galaxies . in practice , the amount of time available on a given facility is finite . performing observations _ commensally _ , i.e. multiple independent projects with distinct science goals all using the data from a single observation , greatly improves the efficiency of a telescope . clever adjustment of the observational parameters , including integration time , survey area , and frequency resolution , enables maximum scientific return for a given dataset . we focus here on the commensality between h and continuum observations , with the aim of optimising the scientific return for the hsurveys , by adjusting only the depths and areas of the observations . the ability to combine surveys with historically different observation modes , namely spectral line and radio continuum , has only recently become feasible due to expanded bandwidth of receivers and sufficient computing to handle the increased data rates . previously , observing in spectral line mode was necessarily restricted to a limited frequency range , corresponding to a very narrow volume slice , appropriate for targeted observations . performing purely commensal continuum observations in spectral line mode was not worth the increased computing overhead , because the probability of the small volume probed containing an h - rich galaxy was low . with the increased bandwidth available for spectral line observations , spanning many hundreds of mhz , the volume probed for h detections is vastly increased . we note that the different computing requirements of spectral line and continuum science , with respect to imaging , spatial and spectral resolution , among other things , will need to be considered . spatial and spectral resolution can be customized for each science case by implementing this flexibility within the data processing pipeline . fortunately , the h and continuum observations have similar initial calibration requirements , so the computing overhead for processing a set of observations to be suitable for both h and continuum science is minimal . satisfying the observational requirements of multiple science cases while staying within the boundary conditions of available observing time requires careful planning and preparation . we should also be mindful of existing and planned ancillary data efforts . survey area , depth , and particularly relevant for radio observations , spatial resolution , need to be considered . at the depths probed by next generation radio facilities , the continuum source density is high @xcite , requiring adequate spatial resolution , of the order of an arcsecond or less , to avoid source confusion . the h source density is much lower , and thus confusion is not as problematic . we also investigate the _ complementarity _ of the resulting data products . ultra - deep neutral hydrogen surveys are planned with both meerkat ( laduma , @xcite ) and askap ( dingo ) , spending several thousands of hours of observations of single fields to detect h in galaxies to higher redshifts and lower h masses than possible with existing telescopes . these same observations will be used by the radio continuum teams , who intend to create ultra - deep continuum maps using multi - frequency synthesis imaging . the converse is also true , with h spectral line data to be extracted from large area continuum observing programmes . thus , from the same observations , simultaneous h spectral line and broad - band continuum data can be obtained , probing both the neutral gas reservoirs and star formation activity for a large sample of galaxies spanning a cosmologically significant range in redshift . this dual - usage of a single dataset to explore two related , yet distinct , physical processes is unique to upcoming radio surveys . we focus only on surveys using the meerkat pathfinder and the ska phase 1 mid - frequency ( ska1-mid ) instrument . we have not included proposed apertif or askap surveys , as the final performance of the phased array feeds emerging technology has yet to be conclusively measured . also , while the single pointing field of view is @xmath1530 deg@xmath16 , the frequency range , and thus redshift range accessible for h , will not be as extensive as that of meerkat or ska1-mid . the current work has two goals . the first is to investigate commensal observations , focusing on a combination of surveys soon to begin , in section [ sec : cases ] . we then use what we learned from that example to suggest optimal commensality for upcoming ska surveys in section [ sec : skadesign ] , and focus on parametrization of the h mass function as a general science goal . the investigation here serves as input for the ongoing discussion regarding the definition of the surveys to be undertaken with next generation facilities . in the second half of the work , starting in section [ sec : deepsurveys ] , we focus on how h spectral line and radio continuum observations are complementary , and what information each provides with respect to the other . a case study for a pair of surveys currently underway is in section [ sec : chiles ] . a summary is in section [ sec : summary ] . concordance cosmology with @xmath17 km s@xmath18 mpc@xmath18 ( thus @xmath19/[100 km s@xmath18 mpc@xmath18]@xmath21 ) , @xmath22 , @xmath23 is assumed when computing masses and luminosities , unless noted otherwise . advances in computing power and engineering technology provide flexibility in observing techniques . in the past , observations were performed either in spectral line mode , with fine spectral resolution over a limited frequency range , or continuum mode , with coarser spectral resolution over a wider frequency range . upcoming facilities will have no such restriction , enabling simultaneous spectral line and continuum observations . while this is clearly efficient with respect to observing time , the science return of the observations can also be increased . the south african ska precursor facility , meerkat , is a 64-element array of 13.5 m dishes located in the karoo region of south africa . the smaller dish diameter of meerkat , with respect to the vla , results in a larger field of view ( fov ) , with the primary beam full width half maximum ( fwhm ) of nearly 1 degree at 1420mhz for the former compared to 0.5 degree for the latter . the majority of operational time has been allocated to several large programmes , with the remaining time available for smaller individual projects . looking at the distant universe with the meerkat array ( laduma ) , is the deep h survey to be undertaken with meerkat . laduma has been allocated 5000 hours of observing time for a single pointing , to be split between two frequency bands , covering @xmath24 for h ( 1670900 mhz ) in phase 1 operations , and @xmath25 ( 1015580 mhz ) in phase 2 . the meerkat international giga - hertz tiered extragalactic exploration ( mightee ; @xcite ) survey is the meerkat continuum key science survey , and follows a wedding - cake design . the exact parameters are yet to be finalised , but for our purposes here , we assume a medium - deep layer covering 35 deg@xmath16 , and a deep @xmath151 deg@xmath16 component coincident with the laduma pointing . the final area coverage will not change significantly from these values . the observations allocated to laduma will be made available to the mightee team for continuum science , while the data collected for mightee can be used for h science , as it will be collected in spectral line mode , with velocity resolution of 26khz , corresponding to 5.5 kms@xmath18 at @xmath11 for h , increasing to 7.8 kms@xmath18 at @xmath26 . thus , laduma and mightee are to be observed entirely commensally . the 5000 hours of integration for laduma is sufficient to directly detect the most h - massive systems at @xmath27 and beyond . this does not , however , directly imply that the continuum data will reach depths corresponding to such long integration times . the longest planned baseline of the meerkat array is 8 km , resulting in spatial resolution ranging from @xmath156 at 1420mhz to @xmath2815 at 600mhz . using the approximation listed in equation 27 in @xcite to compute the noiseless confusion distribution width @xmath29 , the continuum confusion noise is @xmath151@xmath30jy beam@xmath18 for these resolutions , which is reached in approximately 50 hours with meerkat , and can not be overcome without introducing longer baselines . a more significant contribution that laduma data can make to the mightee project is by providing redshifts for the h - rich , but relatively low stellar mass galaxies at @xmath31 , which will have significant levels of star formation , but are optically faint , and therefore difficult spectroscopic targets at visible wavelengths . while laduma provides a deep continuum pointing for mightee , mightee provides complementary wide - area h data with observations covering 35 deg@xmath16 , with tens of hours of integration per pointing . this is clearly not comparable to the thousands of hours devoted to laduma , but it is sufficient to detect the most massive h systems , which will be under - represented in the relatively small volume contained within a single pointing . these same gas - rich galaxies will also have significant star formation , and thus will be a prominent population within the mightee continuum dataset . we will then have simultaneous measures of the neutral gas content and star formation rates of galaxies extending beyond @xmath5 , including a wide range of environments , as the volume probed is large enough to include a number of massive clusters . the benefits of using the medium - deep continuum survey data to supplement h surveys is expanded and investigated in the following section . here we make qualitative predictions of what we can expect from planned surveys and how we can best combine the available information . in the absence of direct observations of the h content of galaxies at @xmath32 , we rely on information gathered from the low redshift universe , combined with simulations , to estimate the number of galaxies , and their distribution in mass and redshift , upcoming surveys will detect . we start by investigating the laduma mightee survey combination , focusing on the h aspect , as this is where the greatest complementarity is seen . the confusion issues for the continuum observations make a similar analysis much more complex . for the following , we use the information provided to the public for meerkat . the listed system equivalent flux density ( sefd ) values indicate meerkat s performance is expected to be better than that of the vla , possibly by as much as a factor of two in sensitivity . as listed in table [ tab : ska1 ] , for the low redshift ( phase 1 ) component of laduma , the integration time will be 1000 hours , which corresponds to a 5-@xmath33 h flux limit of 0.015 jy kms@xmath18 for meerkat over 250 kms@xmath18 . we choose this profile width as we expect the majority of detected objects to be massive systems , with profiles that can easily span several hundreds of kms@xmath18 . equivalently , for 24 hours per mightee pointing , the 5-@xmath33 h flux limit is 0.07 jy kms@xmath18 . we assume that the full 35 deg@xmath16 area will be composed of 35 individual pointings , instead of a single mosaic , which removes the complication of an increasing effective exposure time for a given patch of sky within a mosaic , as the primary beam increases in size going to lower frequency , and hence to higher redshift , and the individual pointings increasingly overlap . we note , however , that in practice there will be overlap between some of the pointings , thus the calculations here are conservative with respect to the mass detection limit . we use the parameters from the analytic fits to the @xmath34 relations derived from the simulations of @xcite to convert flux limits in h to numbers of galaxies detected as a function of redshift . the simulations are consistent with the observed @xmath11 h mass function and co(10 ) luminosity function , and allow for evolution of the h content of galaxies with redshift . the @xmath34 relations enable us to determine , for a given redshift shell and sky area , the total number of galaxies expected to be detected above a specified flux limit . appropriate relations for arbitrary integrated flux limits can be obtained by interpolating the function coefficients for the given set of flux limits . we note that the @xmath34 relations are based on the cosmological parameters of @xmath35 km s@xmath18 mpc@xmath18 with @xmath36 , @xmath37 and @xmath38 ( @xcite ) , which have been superceded by , for example , @xcite , but the effect on the results is negligible . we also make use of the h mass function ( himf ) derived at @xmath5 from alfalfa @xcite , which spans @xmath39log(@xmath40)@xmath41 . galaxies with @xmath40@xmath42@xmath3 are exeedingly rare , so we do not include them in our calculations . from the h mass function , we can estimate how the galaxies in each redshift shell are distributed in mass . for the single meerkat pointing of laduma and 35 deg@xmath16 of mightee , we can compute how many galaxies of each mass we expect to observe over the full survey redshift range . [ fig : lad_mig_zhist ] shows the result of this exercise , limited to @xmath31 . as seen , the continuum survey , mightee , detects nearly three times as many galaxies in h as the deep h survey , laduma , due to the larger area , and thus larger volume , surveyed . the difference is particularly notable at the high h - mass end , where very few massive h systems are detected by laduma . of mightee ( black solid line ) , both for @xmath43 . ] [ fig : lad_mig_mhi_z ] shows how the detected galaxies counted in fig . [ fig : lad_mig_zhist ] are distributed in the redshift@xmath40 plane . the plane is divided into cells of width @xmath44=0.01 and @xmath45log(@xmath40)=0.1 . the contours for both laduma and mightee are set at 1 , 10 and 25 galaxies per cell . complications due to spatial resolution determined by the size of the synthesized beam , how this size changes with redshift , and the resulting effect on column density sensitivity have been neglected for this exercise . the sensitivity is maximized when the size of the galaxy matches the size of the beam , and decreases for larger and smaller sources . therefore , the lower boundaries of detected galaxies are specifically for galaxies the same size as the beam , and extends to higher masses for galaxies of smaller or larger physical sizes . the two meerkat surveys are highly complementary , with mightee primarily detecting high h mass galaxies at low redshift , while laduma reaches further down the h mass function . this access to wide area will constrain the high mass end of the h mass function , which is populated by galaxies with low space density . the wide area is also essential for studies of galaxy environment and large scale structure , which are inaccessible to the narrow , deep observations . thus , including h data from the continuum survey greatly enhances the potential scientific output of the h studies . plane for laduma ( red contours , with ` l ' label ) and mightee ( greyscale contours , with ` m ' label ) . the parameter space is divided into cells of width 0.01 in redshift and 0.1 in log(@xmath40 ) . the contour levels for both surveys are the same , surrounding regions of parameter space containing at least one ( faint shading ) , 10 ( medium shading ) , and 25 ( dark shading ) galaxies per cell . due to the overlapping frequency coverage of the two meerkat receivers , the redshift range @xmath46 is observed for the full 5000 hours for laduma , and is correspondingly deeper . ] a tiered , or wedding - cake " , survey design , is often employed by large projects , as it provides dynamic range in depth , while also probing a large volume . the relative depths of the tiers must be set to provide optimal coverage of some parameter of interest . here we investigate surveys suggested to be undertaken with ska1-mid . the facility will be a combination of 133 15 m new ska dishes , in addition to the 64 13.5 m meerkat dishes , as outlined in the baseline design document version 2 . as the final telescope will be composed of two types of dishes , each with their own performance characteristics , accurately anticipating the sensitivity of ska1-mid is difficult . however , given the sefd values provided in the baseline design document for the ska dishes , and those provided in the public meerkat documentation , we conservatively estimate a factor of 5 increase in sensitivity of ska1-mid bands 1 and 2 , with respect to the vla , consistent with simulations by @xcite . we construct the survey tiers keeping in mind that the resulting data will be used commensally by both the h and continuum teams , and thus restrict ourselves to the range of parameters that are useful to both . we use the sensitivity to h mass as a function of redshift as a measure of the value of the survey tier combination , and optimise the tiers by rearranging their depth and area to maximise coverage of the redshift@xmath40 plane . the redshift and h mass distribution of galaxies expected from surveys displayed as in fig . [ fig : lad_mig_mhi_z ] can be used as a powerful diagnostic tool to explore combinations of depth and area which result in the most scientifically useful dataset . laduma and mightee are clearly well paired at @xmath31 , enhancing the usefulness of the data . the ska h science working group ( swg ) proposed a tiered observing strategy to be undertaken with ska1-mid to address many extragalactic h science cases , consisting of an ultra - deep , small area survey , a medium - wide , medium - deep survey , and a wide - area , relatively shallow survey . we refer to these three tiers as ultra - deep ( ud ) , medium - wide ( mw ) and wide ( w ) . their areas , integration times and approximate 5-@xmath33 sensitivities that can be expected are listed in table [ tab : ska1 ] as the `` fiducial '' case , as set by @xcite . a more detailed discussion within the h swg of the three tiers will come at a later date , when the technical specifications of the ska1-mid are better known , but the suggested fiducial case serves as a useful starting point . we have not included the `` strawman '' surveys outlined in the ska h galaxy evolution chapter ( @xcite ) , as they rely heavily on the ska - survey instrument , construction of which has been deferred . we show in fig . [ fig : ska1_proposed1_mhi_z ] the coverage in the redshift@xmath40 plane resulting from the three fiducial h survey tiers . increasing the integration time to 4000 hours per tier , to be consistent with the proposed surveys discussed below , improves the sensitivity by a factor of two , but the general features remain unchanged . a critical aspect missing from the h swg observing strategy is the possibility of observing commensally , in particular sharing observations with the ska continuum swg ( @xcite ) . while commensal observing necessarily involves some compromises to be made on some aspects , including field selection , the benefits are significant , including the increased available observing time . the continuum swg also proposed a three - tiered observing strategy , with areas comparable to those explored here , with a single pointing ud tier , a mw tier spanning 1030 deg@xmath16 , and a w tier of @xmath151000 deg@xmath16 . the suggested continuum rms values for the three tiers of 0.05 , 0.2 and 1@xmath30jy , respectively , can be compared with the estimated rms values computed for the fiducial and optimised surveys as listed in table [ tab : ska1 ] . we do not list continuum rms values for the requested or ideal cases , because the relevant receivers do not currently exist . we suggest a number of tiered survey strategies working within the framework of commensal extragalactic h and continuum observations . the three tiers under consideration are driven by different science cases . the ud tier is driven primarily by the redshift evolution of h science case , and has very little flexibility . in order to detect the faintest objects at the greatest distances , long integration times are required . the deepest integration is achieved by observing only a single pointing , the area of which is set by the fov of the telescope . at the other extreme , the w tier is of general use for many science cases . the size is restricted by the amount of area that can be covered with individual pointings long enough such that they are not dominated by calibration overheads . the mw tier is intermediate between the two , and requires advance planning to maximise the scientific return . the area and depth must be simultaneously adjusted such that a representative volume containing detectable galaxies is observed . for our purposes , we assume in the following that the continuum science will be the primary driver behind the mw tier , as it will provide the bulk of the continuum science return . the w tier is too shallow to explore a signficant redshift range , while the narrow ud tier will suffer from cosmic variance , which becomes a significant source of uncertainty for areas smaller than about 10 deg@xmath16 @xcite . this is consistent with the ska continuum swg specification that the mw tier should be 1030 deg@xmath16 , while the required sensitivity should be sufficient to sample the peak of star formation , at @xmath47 . continuum confusion is not an issue for ska1-mid as it is for meerkat and askap , as the array layout will enable sub - arcsecond spatial resolution . h science can employ a range of spatial resolutions , matching the resolution with the angular scale of the targets . we assume that the continuum and h data processing pipelines will be able to customize the spatial resolution of the respective maps for optimal sensitivity . .parameters for the two meerkat surveys , laduma ( phase 1 ) and mightee , and the four different combinations of survey tiers suggested for the ska phase 1 mid instrument . the confusion noise floor of @xmath151@xmath30jy for meerkat limits the continuum depth of laduma and mightee . the fiducial case is from the ska h science working group , whereas the optimised case alters the depth and areas of these three tiers . the requested case reflects the frequency coverage of the suggested revised band 1 and band 2 , while the ideal case is for a hypothetical , single - octave receiver system . for the optimized and ideal cases , the ud tier is two pointings covering the same redshift range , to alleviate cosmic variance . for the requested case , the two ud pointings are split between bands 1 and 2 . t@xmath48 is the effective integration time per pointing , and the h flux limits assume ska1-mid sensitivity 5 times that of the vla and profile widths of 250kms@xmath18 . continuum rms values for the ud tiers assume band 1 , and band 2 for mw and wide . the combination of lower sefd and wider available bandwidth for band 2 with respect to band 1 results in similar rms values for the ud and mw tiers for the optimised case . [ cols="<,^,^,^,^ " , ] plane for the fiducial ska1 ud ( red filled contours , with ` ud ' label ) , mw ( greyscale filled contours , with ` mw ' label ) , and w survey ( blue open contours ) . the parameter space is the same as in fig . [ fig : lad_mig_mhi_z ] , but extends to @xmath49 . the contour levels for all three surveys are the same , surrounding regions of parameter space containing at least 1 ( faint shading ) , 25 ( medium shading ) , and 100 ( dark shading ) galaxies per cell . the contours for the wide survey are unfilled to show the overlapping regions between w and mw . the vertical dashed lines indicate the lower edge of band 1 at 1050mhz , and the upper edge of band 2 at 950mhz . the black solid curve indicates the h mass sensitivity if the performance of band 1 is reduced by a factor of two . ] a critical aspect of the following investigation is the frequency coverage of the ska1-mid instrument . the current design for the mid - frequency receivers is for band 1 to cover 1050350mhz ( corresponding to @xmath50 for h observations ) , and band 2 1700950mhz ( @xmath51 ) , the extent of which are indicated as vertical dashed lines in fig . [ fig : ska1_proposed1_mhi_z ] . h surveys are particularly affected by the choice of frequency coverage , as it directly dictates the redshift range that can be surveyed in a given observation . this is highlighted by the discontinuity in coverage in fig . [ fig : ska1_proposed1_mhi_z ] at 1050mhz , or @xmath52 , indicating a mismatch in depth between the ud and mw components . constructing a receiver band such as the current planned band 1 is a challenging engineering task . we have thus marked on fig . [ fig : ska1_proposed1_mhi_z ] the equivalent h mass sensitivity if the band 1 performance is decreased by a factor of two . current measurements of the receiver indicate it performs slightly better than this at low redshift ( high frequencies ) , with decreasing sensitivity to high redshift ( lower frequencies ) . alternative possibilities for receiver bands are currently under discussion . one option that has been suggested is to replace the current wide band 1 with a single - octave band covering 875475mhz ( @xmath53 for h ) , and the low redshift band 2 with a second band covering 1470795mhz ( @xmath54 ) . this arrangement shifts the overlapping redshift range from @xmath55 to @xmath56 . this arrangement is listed as `` requested '' in table [ tab : ska1 ] . addtional options involve a single - octave band covering either 1420710mhz ( @xmath12 for h ) , or 1200600mhz ( @xmath57 ) . although the second range excludes observing h at @xmath11 , it is an attractive option for a number of reasons . first , with respect to survey commensality , detailed h observations of nearby galaxies have different requirements to the deep h surveys aiming for higher redshift , so it is sensible to perform the @xmath11 and @xmath58 surveys separately . in addition , there is known severe rfi at 11701300mhz , rendering extracting data from observations within this frequency range very problematic . continuum studies also stand to benefit , as the lower frequency limit of 600mhz results in a larger fov and enhances sensitivity due to the spectral shape of radio sources , albeit with poorer spatial resolution . including coverage at @xmath59 at the expense of @xmath60 also offers the opportunity to explore new parameter space , which is only accessible to the ska . the number of galaxies directly detected at such high redshifts will be modest , but we will gain the most new information with such observations . in the following , we investigate three cases for frequency coverage . the first assumes the frequency bands as they are currently defined , but optimising survey tier depth and area ( optimised ) . the second refers to the proposed shifting of the frequency coverage of bands 1 and 2 ( requested ) , and the third assumes a single band covering either 1420710mhz or 1200600mhz ( ideal ) . we restrict the total observing time for each survey tier to be nominally one year , which corresponds to approximately 4000 hours of usable integration time . for the optimised case , we first fix the ud component , which has very few options for variation . we choose to split the 4000 hours into two pointings instead of one , to alleviate cosmic variance . we are then free to adjust the area and exposure time per pointing of the mw and w components to address the science cases . as a general rule , increasing area improves coverage of the high mass region of redshift@xmath40 parameter space , whereas increasing integration time improves coverage at lower masses . [ fig : ska1_best1_mhi_z ] shows how the redshift@xmath40plane is populated for the optimised surveys listed in table [ tab : ska1 ] , after adjusting the three tiers for optimal coverage in mass and redshift . for this arrangement , in order to maintain high dynamic range in h mass at low redshifts , the mw tier is necessarily deep , with correspondingly smaller area to keep the total survey time within the 4000 hour budget . 10 deg@xmath16 is at the low end of the range of survey areas deemed to be acceptable for this tier by the ska continuum swg ( @xcite ) . we have increased the area of the wide tier from 400 to 1000 deg@xmath16 , which better samples the high - mass end of the h mass distribution . increasing the area further would require shorter exposure times per pointing , which we want to avoid . as h column density sensitivity also scales with exposure time , longer exposures have greater sensitivity to diffuse h , and therefore can be imaged at higher spatial resolution , than shorter exposures . short integration times over a larger area , with the accompanying poorer resolution , would not be a significant improvement over currently available technology , and would not efficiently exploit the full capabilities of the ska . [ fig : ska1_ecr_mhi_z ] shows a combination of survey tiers more appropriate for the requested frequency coverage . in order to span the full redshift range , the ud tier has been split in two , with one 2000h integration at low redshift ( band 2 ) , and one 2000h integration at high redshift ( band 1 ) . the redshift range @xmath56 is covered by both band 1 and band 2 , and thus receives 4000h of integration , and is subsequently deeper . the primary difference between the optimised and requested cases is in the mw tier , which no longer needs to be as deep . in fact , the coverage of the mw tier is achieved by using only 30 pointings of 50 hours each , using only 1500 of the nominally available 4000 hours . 30 deg@xmath16 is also well - matched to the planned survey area of the lsst deep fields , which will cover 36 deg@xmath16 in the first instance . the remaining time beyond 1500 hours could be used to more than double the area covered , if desired . fig . [ fig : ska1_best2_mhi_z ] shows the alternative arrangement for frequency coverage , listed as ideal in table [ tab : ska1 ] , using either the 1420710mhz or 1200600mhz single band . although we display the full redshift range of @xmath61 , note that only either the range @xmath12 or @xmath57 can be covered with a single observation . the mw and w tiers are unchanged from the requested case , but the ud tier is again composed of two , 2000h integrations , which span the full available redshift range . we emphasize here that , between the optimised and requested or ideal frequency setups , it is the mw tier that is the most effective for optimizing the coverage of the redshift@xmath40 plane . as the specifications of the mw tier are of critical importance to the continuum swg , the importance of full investigation of survey commensality becomes apparent . plane for an ska1 ud ( red filled contours , with ` ud ' label ) , mw ( greyscale filled contours , with ` mw ' label ) , and w survey ( blue open contours ) , corresponding to the optimised case in table [ tab : ska1 ] . the contour levels for all three surveys are the same , surrounding regions of parameter space containing at least 1 ( faint shading ) , 25 ( medium shading ) , and 100 ( dark shading ) galaxies per cell . the black solid curve indicates the h mass sensitivity of band 1 if it is reduced by a factor of two . the characteristic mass , which is @xmath62@xmath3 at @xmath11 , is well covered with this set of surveys . ] , but for the frequency band definition corresponding to the requested case . note that the ud component now must be built from two separate integrations , one with band 1 and one with band 2 , while the redshift range @xmath56 is observed twice . ] , but for the frequency band definition corresponding to the ideal in table [ tab : ska1 ] . note that either @xmath12 or @xmath57 can be observed at once , but not the full redshift range @xmath61 . the lack of the @xmath52 lower redshift cutoff for the ud component means the mw tier can be shallower and wider , and completed in only 1500 hours . the absence of the @xmath63 upper redshift cutoff enables the mw tier to continue contributing coverage of the high - mass galaxies up to @xmath0 . either band definition results in a more efficient set of surveys which covers more of the redshift@xmath40 plane than with the existing bands 1 and 2 . ] we have thus far not mentioned the nearby galaxies extragalactic hscience case , which aims to map the neutral gas distribution of a number of local , well - resolved galaxies in much greater detail than will be achieved with either the ultra - deep or wide - area surveys . these targeted observations will observe a small number of fields with long exposure times , similar to the parameters for the mw surveys discussed here . however , these mw observations can not be easily shared with the local galaxies communities , as the fields would be specifically chosen such that they do _ not _ have prominent nearby galaxies within them . although the spectral dimension allows us to see in h beyond the local galaxies , they are optically opaque , rendering the entire volume beyond the galaxies inaccessible to ancillary data . while there is no clear commensality between the deep extragalactic surveys discussed here and the resolved studies of nearby galaxies , the h and continuum teams targeting these galaxies will certainly exploit the high degree of commensality between their respective observations . with several options for survey design available to us , we need some way of quantifying the quality and usefulness of the resulting dataset . population studies require dynamic range in mass at a given redshift , whereas evolution studies require dynamic range in redshift at a given mass . h content as a function of environment requires sufficient survey volume to include a range of galaxy group and cluster masses . applications such as investigating the evolution of the h mass function with redshift would benefit from coverage of the redshift@xmath40 plane spanning the break of the relation , which is at log(@xmath40 ) = 9.96 @xmath3 at @xmath5 @xcite . if continuous coverage of h mass at a given redshift is a priority , then adjusting exposure times and survey areas can minimize gaps in the redshift@xmath40 plane . we have chosen as an illustrative target science goal the characterisation of the h mass function ( himf ) as a function of redshift . this fundamental relation can be integrated to give the cosmic h density , @xmath64 , as a function of redshift . knowledge of the himf not only provides information about the h content of galaxies , but it provides the contextual framework within which other hstudies are conducted , i.e. are the galaxies under investigation chosen from the rare , high - mass end of the distribution , or are they relatively common ? quantifying the parametrization of the himf at several redshifts allows us to move beyond the so - called `` strawman '' surveys , and set realistic guidelines for a combination of survey tiers that results in a coherent dataset which maximizes scientific return while minimizing duplicated effort . we use the schechter function parametrization of the himf as used by the alfalfa team in @xcite , given by : @xmath65 there are three parameters to determine : the faint - end slope @xmath9 , the characteristic mass @xmath66 , and the normalisation @xmath67 . the lack of observational data regarding the detailed distribution of h content of individual galaxies at @xmath32 provides little constraint on any evolution in the himf , so we do not include any evolution in the three parameters . the combination of observing h in emission at low redshift , and high redshift observations of damped ly@xmath9 absorption seen in the spectra of background quasars constrains the evolution of @xmath64 to be at most a factor of a few between @xmath68 , so we do not expect drastic evolution of the individual himf parameters over our redshift range of interest , @xmath61 . simulations from @xcite indicate that the high mass end of their himf steepens significantly , but only at high redshifts , @xmath69 . we construct himfs from the combinations of surveys introduced in section [ subsec : commensal ] and see how the different surveys provide complementary constraints . as an illustration , the advantage gained from combining surveys of different depth and area is displayed in fig . [ fig : lad_mig_himf_inset ] , which shows the himf resulting from the mightee and laduma surveys individually , and the result from combining the data . the uncertainties on the individual himfs from laduma and mightee are @xmath70 , where @xmath71 is the number of objects in the given mass bin of width @xmath45log(m ) = 0.1 . these are represented by the shaded areas in the main panel , and the final combined himf is shown in black with errorbars . the large volume covered by mightee is entirely responsible for constraining the high - mass end of the himf , as laduma does not contain any of these rare galaxies . at the low - mass end , the deep flux limit of laduma is required , and a small mass range around the break in the relation is covered by both surveys . when combined , the three parameters describing the himf are better constrained than considering each survey individually . . the two insets show the himf from mightee and laduma individually , and the main panel shows is the combination of the two . the x - axis in the main panel and the insets span the same range . the grey region is the himf from mightee including uncertainties , the red region is from laduma with the associated uncertainties , and the black line with errorbars is the combination of the two . the dotted grey line here and in the following figures is the input @xmath11 himf from alfalfa , for reference . ] [ fig : best1_best2_himf_z0203_inset ] compares the himf that can be constructed from the optimised surveys and either of the ideal options , at @xmath72 . for the top panel , the existing band 1 does not cover redshifts below @xmath52 , so the ud component is excluded . the break and the high - mass end of the himf are well constrained by the mw and w survey components , as seen by the small uncertainties . alternatively , the ideal cases in the bottom panel do incorporate the ud component , which has large uncertainties at high h masses , but extends the himf to lower masses , helping constrain the low - mass slope . . the ud component is not included in the top panel as band 1 does not extend to these low redshifts . the addition of the ud component in the bottom panel shows the himf can be probed to lower masses . the x - axis in the main panel and the insets span the same range . ] at higher redshift , @xmath73 , fig . [ fig : best1_best2_himf_z04_inset ] shows that the optimised and either of the ideal surveys both incorporate the ud component , but lose significant contribution from the w survey , which only detects galaxies in the final few h mass bins . the short vertical line in the top panel indicates the limiting h mass reached if the sensitivity of the current band 1 receiver is reduced by a factor of two . the bottom panel shows that the larger area of the ideal mw tier reduces the uncertainties at the high - mass end of the himf . if we allow the normalization of the himf to evolve with redshift such that by @xmath0 the integral is twice that at @xmath11 , we get the dashed curve shown in the bottom panel of fig . [ fig : best1_best2_himf_z04_inset ] . the uncertainties in the himf determination for both cases are sufficiently small to distinguish between the evolving and non - evolving cases . . the larger area of the mw component in ideal compared to the optimised case improves the constraints at the high - mass end . the short vertical line in the top panel marks the limiting h mass if the sensitivity of the current band 1 is reduced by a factor of two , and the dashed line in the bottom panel indicates the himf if allowed to evolve with redshift . the dotted line is the @xmath11 himf as in previous figures . the x - axis in the main panel and the insets span the same range . the benefits of observing commensally to minimize observational effort while maximizing science return are clear . provided sufficient advance effort with respect to survey design and science requirements is invested , it is possible to construct a coherent dataset from disparate surveys . with a major scientific effort such as the ska , the accessibility of previously unexplored parameter space is expected to transform our understanding of a number of key areas of astrophysics , so we must ensure we undertake surveys which will provide the data needed to significantly advance our current knowledge . here we explore the _ complementarity _ of the data products resulting from the commensal h and radio continuum observations , biases in the resulting dataset , limitations resulting from restricted sensitivity at different wavelengths , and how information can be combined to circumvent these limitations . with future observations , we will have simultaneous measures of both the h reservoirs of galaxies , along with star formation rate indicators from the radio continuum , over a cosmological redshift range . our current information , particularly h , is restricted to @xmath5 . while limited , the existing data can be used to provide the context within which we can expect the new observations to fit . to understand how h , optical and radio continuum surveys will fit together , we take advantage of the spatial overlap of two individual , large - area surveys . optical photometry and spectroscopy are provided by the sloan digital sky survey ( sdss , @xcite ) . the h data are from the arecibo legacy fast alfa survey ( alfalfa , @xcite ) @xmath9.40 source catalogue from @xcite , which covers @xmath152800 deg@xmath16 within the sdss imaging footprint , and contains 15041 secure extragalactic sources . alfalfa is flux - limited to @xmath150.47 jy kms@xmath18 , while the spectral range allows observations of galaxies to @xmath74 . from this catalogue we extract the h masses , @xmath40 , from the hprofiles , along with the galaxy redshifts provided within the alfalfa data release . the alfalfa team has crossmatched the @xmath9.40 catalogue to the sdss data release 7 ( dr7 , @xcite ) by hand to identify the most likely optical counterparts for the h detections . 11740 of the alfalfa galaxies have clear optical counterparts in the dr7 imaging , providing optical photometry , galaxy sizes , and for the majority of the objects , spectroscopic redshifts . we refer to this set of galaxies as the alfalfa sdss sample . at the intersection of alfalfa and sdss , we have , for a large number of galaxies , information regarding the neutral gas reservoir , the existing stellar properties , and a measure of the star formation rate from optical photometry . even though alfalfa is restricted to @xmath75 , the galaxies contained within the survey span a wide range of h masses , stellar masses , and star formation rates , and provide a useful low redshift frame of reference for extrapolating to the higher redshifts and greater depths which will be probed by upcoming surveys . as described in detail in @xcite , h selection results in a highly biased galaxy sample , preferentially including blue , star forming objects . constructing a galaxy sample flux - limited in the radio continuum also results in a biased census , composed of two populations of objects . the high luminosities are dominated by galaxies hosting active galactic nuclei ( agn ) , whereas at lower luminosities , the star - forming galaxies become more prevalent ( for example , @xcite , @xcite , @xcite , @xcite ) . continuum surveys with flux limits sensitive enough to probe below the agn regime will have significant numbers of star - forming galaxies , which are the same galaxies selected by deep h surveys . so although both h and continuum selection result in biased galaxy samples , they are biased toward the same galaxies . the galaxy mass function derived from the galaxy and mass assembly ( gama , @xcite ) shows that at the highest stellar masses , the galaxy counts are dominated by red galaxies @xcite , which tend to be h - poor and have little star formation activity . however , at stellar masses @xmath1@xmath76@xmath3 , the galaxy counts are dominated by blue galaxies , which is also where the bulk of star formation activity occurs , and is exactly the population of objects both h and continuum selection are biased toward . a multi - wavelength aspect , combining several individual surveys at different wavelengths , is becoming increasingly common with current and future surveys , including upcoming radio - based observations . these ancillary data , primarily optical imaging and spectroscopy , provide additional information for the galaxy population under examination . however , the optical observations impose their own flux limits on a given survey in addition to the h and continuum sensitivities imposed by the radio wavelength observing parameters . imaging large areas of sky to significant depths in order to obtain galaxy colours and morphological information is well within the capabilities of current optical instrumentation , as proven by , for example , the canada - france - hawaii telescope legacy survey ( cfhtls ) , the cosmic evolution survey ( cosmos , @xcite ) , ultravista ( @xcite ) , the vista deep extragalactic observations survey ( video , @xcite ) , and dark energy survey ( des ) and do not impose significant limitations on the galaxy sample to be observed . upcoming surveys such as the large synoptic survey telescope ( lsst , @xcite ) , euclid ( @xcite ) , and wfirst actively incorporate surveys at other wavelengths in their planning strategy . large area , flux - limited spectroscopic surveys , however , require extensive telescope time and effort . while photometric redshifts ( photo-@xmath77 ) , with accuracies of @xmath78 are sufficient for a number of scientific applications ( the evolution of luminosity functions , for example ) , the accuracy of spectroscopic redshifts ( @xmath79 ) are required for investigations of detailed galaxy environment and spectral stacking ( @xcite ) . the acquisition of sufficient spectroscopic redshifts to supplement upcoming h and continuum surveys is the topic of much discussion and effort ( see a good overview in @xcite ) , and here we investigate the effect of imposing an optical flux limit on a galaxy sample initially flux - limited in h as a function of redshift , and what other information we can use to alleviate the effects . scaling relations illustrate general trends between physical properties for an ensemble of galaxies . one such relation at @xmath5 is shown in fig . [ fig : a_mhi_mstellar ] , indicating how h mass , from the alfalfa catalogue , and stellar mass , from the mpa jhu catalogue ( @xcite , @xcite ) , are related for the h - selected alfalfa sdss galaxies . at low stellar masses , below @xmath1@xmath80@xmath3 , the galaxies are h - rich , lying above the diagonal line marking the 1-to-1 relation . at high stellar masses , the galaxies become increasingly h - poor . extensive discussion regarding the features of this relation can be found in @xcite . there are many galaxies not detected by alfalfa which would lie in the bottom right corner of fig . [ fig : a_mhi_mstellar ] , at high stellar mass and low hmass . these are the absent red sequence galaxies not included in an h - selected sample . while a small number of early - type galaxies are found to have significant masses of h @xcite , the contribution from these galaxies to the h mass census is negligible . as a function of @xmath1 for the alfalfa sdss galaxies , in units of @xmath3 . the diagonal black solid line marks the 1-to-1 relation . the vertical grey dotted lines indicate the stellar mass limits resulting from an optical flux limit of @xmath81 at various redshifts as described in section [ subsec : fluxlim ] , while the horizonal grey dashed lines indicate the h mass sensitivity of the chiles survey at @xmath82 , outlined in section [ sec : chiles ] . ] known scaling relations can also be useful for converting observable quantities , such as brightness , into quantities that are difficult to measure either directly or indirectly . for example , a galaxy s stellar content can be estimated using only it s brightness and colour by employing the known correlation between the optical luminosity and colours of a galaxy and its stellar mass ( @xcite ) . we use this relation to convert optical flux limits to approximate limiting stellar masses , as a function of redshift , to illustrate the effect these flux limits have on an h - selected galaxy sample . due to the biased nature of the alfalfa sdss galaxies , we derive a luminosity@xmath1 relation specifically for this dataset . as the h - selected galaxies are nearly all blue cloud galaxies , we remove the colour dependence of the usual luminosity@xmath1 relation by assuming all galaxies have the same representative colour . this approximation is sufficient for the illustration presented here . we find : @xmath83 where @xmath1 is in units of solar masses , and @xmath84 is the absolute @xmath85-band magnitude . we then use this relation to convert a given optical flux limit into an approximate redshift - dependent stellar mass limit , incorporating a @xmath86-correction appropriate for blue galaxies . the calculation is only an approximation , because it uses the relation derived from the @xmath11 population and assumes no evolution with redshift . shown on fig . [ fig : a_mhi_mstellar ] are vertical lines indicating the effective stellar mass limits resulting from imposing a flux limit of @xmath87 , appropriate for an optical spectroscopic campaign , on the @xmath11 h - selected galaxy ensemble . the distribution of galaxies in this figure is to provide the low redshift reference for the populated regions of parameter space , which will evolve with redshift in an unknown way , as we currently can not make such a figure at @xmath58 . the limits at increasing redshift then illustrate which parts of the parameter space are no longer accessible . at @xmath88 , all galaxies with @xmath1@xmath89@xmath3 are brighter than @xmath81 , but galaxies are rapidly lost with increasing redshift , leaving only objects with @xmath1@xmath90@xmath3 bright enough for spectroscopy at @xmath26 . note that these limits ignore the complicating factor of galaxy surface brightness , which works against observations of disk galaxies , particularly when viewed face - on . relaxing the optical flux limit to @xmath91 , more appropriate for optical photometry alone , shifts the effective stellar mass limits lower by an order of magnitude . this fainter flux limit is relevant for obtaining photometric redshifts with current facilities over moderately wide areas . also shown on fig . [ fig : a_mhi_mstellar ] are horizontal lines denoting the 5-@xmath33 h mass sensitivity of chiles as a function of redshift ( see section [ sec : chiles ] for details ) . the important detail to note is that for a given h mass , for example @xmath40@xmath92@xmath3 , the stellar masses range from @xmath93@xmath1@xmath94@xmath3 . even at @xmath95 , galaxies of @xmath1@xmath2@xmath3 will be detected in the most h - rich cases , a factor of more than three lower in stellar mass than is accessible to optical spectroscopy at the same redshift , unless the objects have very strong emission lines . thus , selection to a given h flux limit is effective at identifying galaxies with a wide range of stellar masses , whereas selection incorporating an optical flux limit is strongly biased toward the most massive stellar systems , particularly at increasing redshift . in fact , the h observations themselves are able to provide redshifts for these galaxies below the stellar masses accessible to optical spectroscopy . in order to explore the average galaxy population below the detection threshold of a given survey , statistical co - adding , or stacking , is often employed . stacking in the radio continuum has been successfully performed by a number of authors ( for example , @xcite , @xcite , @xcite , @xcite , or see @xcite for a good overview ) . the spatial coordinates of the galaxies are provided by deep optical or near - infrared imaging , and redshifts are required to convert fluxes into luminosities . with the commensal h and continuum observations , the galaxy population selected in h , rather than at optical wavelengths , can be used as the basis for a star formation census over the redshift range probed by the spectral observations . not only does the hselection provide a galaxy sample less biased by stellar mass , but the h detections themselves provide the redshifts of the galaxies , without the need for additional expensive spectroscopy , which quickly becomes unfeasible for large numbers of galaxies with @xmath1 @xmath96@xmath3 at @xmath32 . as described in section [ subsec : biases ] , observations in h and radio continuum tend to select the same galaxy population , so within a single observation , we gain information about both the neutral gas content and star formation activity in the same set of galaxies . we investigate a pair of complementary spectral line and radio continuum surveys currently underway with the vla . the cosmos h large extragalactic survey ( chiles ) project is a deep , 1000 hour observation of a single pointing . the instantaneous frequency coverage of 1450970 mhz allows h detections over the unprecedented redshift range @xmath97 , which will only be surpassed by ska - era facilities . two semesters of observations are already completed , and the feasibility is demonstrated by the pilot project , spanning @xmath98 @xcite . the same 1000 hours of observations are being shared with the chiles continuum polarization ( chiles con pol , @xcite ) team , who are exploiting the flexibility and power of the upgraded vla correlator to simultaneously observe in four 128 mhz continuum bands to create a deep continuum map of the same area of sky . the total observing time for chiles was chosen in order to detect a galaxy with @xmath40@xmath99@xmath3 at the highest accessible redshift . this corresponds to a 5-@xmath33 flux limit of 0.0295 jy kms@xmath18 , which translates directly to a redshift - dependent h mass limit via : @xmath100 where @xmath40 is in solar masses , the luminosity distance to the galaxy , @xmath101 , is in mpc , and the integral is the total flux in jy kms@xmath18 . the @xmath102 factor accounts for the difference between the observed and rest - frame profile width . assuming an hprofile width of 150 kms@xmath18 , the minimum detectable h mass at @xmath103 0.1 , 0.2 , 0.3 and 0.4 are marked as horizontal dashed lines on fig . [ fig : a_mhi_mstellar ] . at @xmath88 , galaxies with @xmath40@xmath2@xmath3 are easily detectable , thus including the bulk of the h mass . the stellar masses for these galaxies spans @xmath93@xmath1@xmath94@xmath3 . even at @xmath104 , the h mass limit is still below the break mass of the @xmath11 h mass function , at @xmath105@xmath3 @xcite . in reality , these mass limits have some variation for a given flux limit , as the resulting detection limit is dependent on the hi profile width , as determined by not only the mass of the galaxy , but it s inclination to the line of sight . for a given flux limit , galaxies seen face - on can be detected to lower mass limits than galaxies viewed edge - on , by about 0.3 dex . thus , a flux limit does nt result in a hard lower mass limit , but rather a lower mass limit with a decreasing number of galaxies to lower masses seen increasingly face - on . the limiting h masses marked on fig . [ fig : a_mhi_mstellar ] should then only be used for guidance . in a similar manner to h , the expected flux limit of 0.6@xmath30jy per beam in continuum for chiles con pol can be converted to an equivalent detectable star formation rate as a function of redshift . to convert the 1.4 ghz continuum flux , @xmath106 , into luminosity , we use : @xmath107 where the luminosity , @xmath108 , is in w hz@xmath18 sr@xmath18 and assuming a power - law spectral index @xmath109 and @xmath110 . then , using the relation to convert 1.4 ghz luminosity to a star formation rate , we use the relation from @xcite : @xmath111 even at @xmath26 , chiles con pol will be sensitive enough to detect star formation rates of @xmath15 1 @xmath3 year@xmath18 . thus , for the first time , we will have information about moderate levels of star formation , along with the h reservoirs , in galaxies over a range of redshifts . due to the computing capacities now available , this information can be collected with one set of observations . with the relatively small field of view of the vla , @xmath150.5 deg in diameter , and restricted redshift range of @xmath112 for h , the number of galaxies contained within the observed volume is manageably small , on the order of 300 galaxies . the chiles pointing is within the cosmos footprint , thus deep multi - wavelength imaging ( @xcite , @xcite ) , along with extensive spectroscopy , already exist ( @xcite , @xcite , @xcite , @xcite , @xcite , @xcite ) . these catalogues have been conveniently amalgamated , and for the zcosmos spectra , reprocessed , by @xcite . this indicates the amount of ancillary data required to support upcoming radio surveys . with the larger fields of view and wider frequency coverage of ska - era facilities , the ancillary data effort becomes even more substantial . despite h ultimately providing the raw material required for star formation , little is known about how galaxies acquire and consume this fuel over cosmological timescales . conversely , the star formation density in the universe has been traced with a number of techniques to the highest observable redshifts . with current and upcoming radio facilities , enabled by advances in engineering and computing , we will , for the first time , be able to trace both the h reservoir and the star formation activity in the same galaxies over a cosmologically significant timescale , when the sfr is dropping rapidly but the neutral gas content remains relatively stable . careful planning of commensal observations enables multiple science cases to benefit from the same data , and maximises the scientific return with no impact on telescope time . the meerkat programmes laduma and mightee are an excellent example of survey tiers well matched in terms of relative depth and area coverage , creating a comprehensive h dataset . conversely , the three h survey tiers suggested by the ska h science working group to be undertaken with the ska1-mid instrument require significant rearrangement in terms of area and depth to create a coherent dataset . the definition of the frequency bands is a critical aspect when designing the tiered survey , especially for h science . we find that alternative frequency band definitions , either the suggested alteration of bands 1 and 2 , or a new single - octave band covering 1200600mhz , require less extreme survey tiers , and less observing time , in order to optimally cover the redshift@xmath40 parameter space . in addition to h and continuum observations being highly commensal , the data products are also complementary . h selection results in a galaxy sample biased toward blue , star - forming galaxies , which are the same galaxies that dominate radio continuum selection at faint flux limits . h selection also results in a sample less biased in terms of stellar mass than the effective optical flux limit that results from the additional requirement of obtaining spectroscopic redshifts . in fact , data from the radio spectral line observations are able to provide redshifts for gas - rich galaxies without needing additional optical spectroscopy . commensal observing will become more common as new facilities with expanded capabilities come online , and large programmes compete for ever increasing time allocations . maximising the science return for many separate projects while minimising observational effort will be essential to achieve the ambitious goals set out by each team , whose individual survey plans would not be realised otherwise . the current investigation shows that with adjustments of survey parameters , including area and depth , a coherent strategy satisfying several communities is possible , provided advance effort is committed to the planning stages . the arecibo observatory is operated by sri international under a cooperative agreement with the national science foundation ( ast-1100968 ) , and in alliance with ana g. mndez - universidad metropolitana , and the universities space research association . we acknowledge the work of the entire alfalfa collaboration team in observing , flagging , and extracting the catalogue of galaxies used in this work . funding for the sdss and sdss - ii was provided by the alfred p. sloan foundation , the participating institutions , the national science foundation , the u.s . department of energy , the national aeronautics and space administration , the japanese monbukagakusho , the max planck society , and the higher education funding council for england . the sdss was managed by the astrophysical research consortium for the participating institutions . the sdss web site is http://www.sdss.org/. the sdss is managed by the astrophysical research consortium for the participating institutions . the participating institutions are the american museum of natural history , astrophysical institute potsdam , university of basel , university of cambridge , case western reserve university , university of chicago , drexel university , fermilab , the institute for advanced study , the japan participation group , johns hopkins university , the joint institute for nuclear astrophysics , the kavli institute for particle astrophysics and cosmology , the korean scientist group , the chinese academy of sciences ( lamost ) , los alamos national laboratory , the max - planck - institute for astronomy ( mpia ) , the max - planck - institute for astrophysics ( mpa ) , new mexico state university , ohio state university , university of pittsburgh , university of portsmouth , princeton university , the united states naval observatory , and the university of washington . this research has made use of nasa s astrophysics data system . this work has benefitted from useful discussions , particularly with ian heywood and chris hales . mjj acknowledges support from ska south africa and stfc . we thank the anonymous referee for helpful comments which improved this paper . blyth s. , et al . , 2015 , proceedings of advancing astrophysics with the square kilometre array ( aaska14 ) . 913 june , 2014 . giardini naxos , italy . online at http://pos.sissa.it/cgi-bin/reader/conf.cgi?confid=215,%20id.128 braun r. , bourke t. , green j. a. , keane e. , wagg j. , 2015 , proceedings of advancing astrophysics with the square kilometre array ( aaska14 ) . 913 june , 2014 . giardini naxos , italy . online at http://pos.sissa.it/cgi-bin/reader/conf.cgi?confid=215,%20id.174 jarvis m. , bacon d. , blake c. , brown m. , lindsay s. , raccanelli a. , santos m. , schwarz d. j. , 2015 , proceedings of advancing astrophysics with the square kilometre array ( aaska14 ) . 913 june , 2014 . giardini naxos , italy . online at http://pos.sissa.it/cgi-bin/reader/conf.cgi?confid=215,%20id.18 meyer m. , robotham a. , obreschkow d. , driver s. , staveley - smith l. , zwaan m. , 2015 , proceedings of advancing astrophysics with the square kilometre array ( aaska14 ) . 913 june , 2014 . giardini naxos , italy . online at http://pos.sissa.it/cgi-bin/reader/conf.cgi?confid=215,%20id.131 popping a. , meyer m. , staveley - smith l. , obreschkow d. , jozsa g. , pisano d. j. , 2015 , proceedings of advancing astrophysics with the square kilometre array ( aaska14 ) . 913 june , 2014 . giardini naxos , italy . online at http://pos.sissa.it/cgi-bin/reader/conf.cgi?confid=215,%20id.132 prandoni i. , seymour n. , 2015 , proceedings of advancing astrophysics with the square kilometre array ( aaska14 ) . 913 june , 2014 . giardini naxos , italy . online at http://pos.sissa.it/cgi-bin/reader/conf.cgi?confid=215,%20id.67 staveley - smith l. , oosterloo t. , 2015 , proceedings of advancing astrophysics with the square kilometre array ( aaska14 ) . 913 june , 2014 . giardini naxos , italy . online at http://pos.sissa.it/cgi-bin/reader/conf.cgi?confid=215,%20id.167 zwart j. , et al . , 2015 , proceedings of advancing astrophysics with the square kilometre array ( aaska14 ) . 913 june , 2014 . giardini naxos , italy . online at http://pos.sissa.it/cgi-bin/reader/conf.cgi?confid=215,%20id.172

we investigate new and unusual signals that arise in theories where dark matter is asymmetric and carries a net antibaryon number , as may occur when the dark matter abundance is linked to the baryon abundance . antibaryonic dark matter can cause _ induced nucleon decay _ by annihilating visible baryons through inelastic scattering . these processes lead to an effective nucleon lifetime of years in terrestrial nucleon decay experiments , if baryon number transfer between visible and dark sectors arises through new physics at the weak scale . the possibility of induced nucleon decay motivates a novel approach for direct detection of cosmic dark matter in nucleon decay experiments . monojet searches ( and related signatures ) at hadron colliders also provide a complementary probe of weak - scale dark - matter induced baryon number violation . finally , we discuss the effects of baryon - destroying dark matter on stellar systems and show that it can be consistent with existing observations . baryon destruction by asymmetric dark matter hooman davoudiasl , david e. morrissey , kris sigurdson , sean tulin + department of physics , brookhaven national laboratory , + upton , ny 11973 , usa + theory group , triumf , + 4004 wesbrook mall , vancouver , bc v6 t 2a3 , canada + department of physics and astronomy , university of british columbia , + vancouver , bc v6 t 1z1 , canada + email : hooman@bnl.gov , dmorri@triumf.ca , krs@physics.ubc.ca , tulin@triumf.ca

1106.4320

cosmological observations indicate that about 4.6% of the energy density of the universe consists of baryonic matter , while 23% is dark matter ( dm ) @xcite . neither of these results can be explained with our current understanding of elementary particles , the standard model ( sm ) . cosmology therefore requires new fundamental physics , and it is important to find ways to detect such new physics experimentally . in the majority of new - physics scenarios , the generation of baryons and dm occurs through unrelated mechanisms , offering no explanation for the similar magnitudes of their cosmological densities . the most thoroughly studied scenarios involve baryon production from cp - violating non - equilibrium processes during the electroweak phase transition , from decays of right - handed neutrinos , or from the coherent evolution of scalar fields @xcite , while the dm relic density is determined by thermal freeze out when a non - relativistic stable species falls out of equilibrium @xcite . in this context , there is no reason to expect similar cosmological densities of baryons and dm . this apparent coincidence may instead be a clue that both types of matter have a common origin . several models of asymmetric dark matter ( adm ) have been proposed along these lines where the dm density carries a net ( approximately ) conserved global charge shared by the sm @xcite , such as baryon number @xmath7 . these models generally fall into two classes depending on how the charge asymmetry is created : 1 . an initial charge asymmetry , generated in either the visible or dm sector , is partitioned between the two sectors by chemical equilibration through a transfer operator @xcite . these charges are separately `` frozen in '' once the transfer operator goes out of equilibrium . 2 . non - equilibrium dynamics generate equal and opposite charge asymmetries in the visible and dm sectors , without any net overall charge asymmetry @xcite . in order to avoid washout , transfer operators must always be out of equilibrium once the asymmetries are created . we term this process _ ( `` hyle '' = matter ) @xcite . subsequently , in both cases , dm particles and antiparticles are assumed to annihilate efficiently , leaving only a remnant asymmetric component determined by the charge density . in the present work , we investigate novel experimental signatures from hylogenesis scenarios where dm carries @xmath7 @xcite . here , the lowest - dimensional , gauge - invariant transfer operator is given by [ indop ] _ ~ u_r^i d_r^j d_r^k _ r + [ leff ] where @xmath8 label generation . in this case , dm has two components , a fermion / scalar pair @xmath9 with total baryon number @xmath10 . stability of both @xmath11 and @xmath12 requires @xmath13 . this operator is inactive cosmologically and does not wash out the baryon - dm asymmetry for @xmath14 provided the asymmetry is created at relatively low temperatures , below about a gev . our results may also be applicable to other adm scenarios in which the same transfer operator appears with one or both of @xmath15 making up the dm . in hylogenesis , the universe is net @xmath7-symmetric , and therefore the baryon asymmetry carried by dm is equal and opposite to that in visible baryons . specifically , in the hylogensis scenario of ref . @xcite , this has two important consequences : * the number densities of @xmath15 satisfy . therefore , cosmological observations imply @xmath16 . together with the dm stability requirement , we have @xmath17 gev . * there exist many scenarios in which adm is coupled to the sm via the `` neutron portal '' operator @xmath18 @xcite , generally falling into the first class of `` chemical equilibration '' scenarios . in this case , dm is very often baryonic ( @xmath19 ) , while for hylogenesis scenarios the dm must be antibaryonic ( @xmath20 ) . in equilibration adm scenarios . also , for other applications of the neutron portal operator to baryogenesis , see ref . @xcite . ] in ref . @xcite , we presented a specific realization for hylogenesis where the operator in eq . arises by integrating out heavy dirac fermion mediatiors @xmath21 . out - of - equilibrium decays of @xmath22 during reheating generate equal - and - opposite dark and visible baryon asymmetries . furthermore , in this scenario @xmath15 are charged under an additional hidden @xmath23 gauge symmetry , that couples to the sm via kinetic mixing with hypercharge , to facilitate annihilation of the symmetric dm densities . more details about this realization are given in appendix [ sec : appa ] . hylogenesis models with the operator of eq . have an interesting and unique signature : antibaryonic dm particles can annihilate visible baryonic matter , termed _ induced nucleon decay _ ( ind ) . ind is a novel and unusual dm signal . these events are @xmath24 and @xmath25 , where @xmath26 is a nucleon and @xmath27 is a meson . since the dm states are invisible , this mimics nucleon decay , with an effective lifetime dependent on the local dm density . this signal offers the new and exciting possibility of searching for dm in nucleon decay searches in deep underground detectors such as superkamiokande @xcite . in section [ sec : ind ] , we compute the rates and kinematics of ind , and discuss the implications of ind for existing nucleon decay searches . our main conclusions are : ( i ) due to different kinematics of ind , compared to standard nucleon decay , existing bounds do not apply over most of the region of parameter space of our dm model , and ( ii ) the effective nucleon lifetime can be around @xmath28 years , if the new physics scale @xmath29 in eq . is the weak scale . hadron colliders can probe ( anti)baryonic dm scenarios through direct production of the weak - scale mediators transfering @xmath7 between visible and dark sectors , discussed in section [ sec : coll ] . the operator of eq . can give rise to observable signatures in the form of monojets ( or jets and missing energy ) . within our hylogenesis model , if ind were detected in nucleon decay searches , monojet signals are inescapable and should be observed at the large hadron collider ( lhc ) . ind processes can potentially be relevant in stars . since the ind rate scales with the dm density , capture and accretion of dm in stars is important , possibly leading to modification of stellar evolution due to baryon destruction and energy injection . in section [ sec : star ] , we consider these effects in neutron stars , white dwarfs , and main - sequence stars . our conclusions are summarized in section [ sec : conc ] . in ref . @xcite . _ ] dark matter particles @xmath9 can annihilate nucleons @xmath30 , producing an energetic meson @xmath27 through inelastic scattering [ indprocess ] n | m , n ^m , shown in figure [ indfeyn ] . we restrict our attention to single meson final states , although multi - meson events are allowed and may have comparable rates . in general , both down - scattering and up - scattering can occur , defined as whether the heavier or lighter dm particle , respectively , is in the initial state ; however , the latter is kinematically forbidden if @xmath31 . since neither the initial dm nor final anti - dm particles are observed directly , these processes mimic standard nucleon decay events @xmath32 @xcite , with an undetected final state neutrino @xmath33 ( or antineutrino @xmath34 ) . the observable meson energy for each case in eq . is , respectively , e_m = , e_m = , neglecting the initial kinetic energy of the dm particles ( @xmath35 ) . although hylogenesis ( as a baryogenesis mechanism ) works for any quark flavor , the most interesting signatures arise if the ind operators involve @xmath36 quarks only . restricting our attention to the lightest mesons ( @xmath37 , @xmath38 , @xmath39 ) , the final state meson can have much more kinetic energy in ind than for standard nucleon decay ( snd ) , summarized in table . [ tab : indmom ] . for fixed masses @xmath40 and @xmath41 , the meson is either monochromatic or bichromatic ( up to fermi motion ) , depending on whether up - scattering is forbidden or allowed . the range of momenta @xmath42 corresponds to the allowed range @xmath43 gev , provided that @xmath15 are stable and account for the observed @xmath44 . ' '' '' ' '' '' & @xmath45 & @xmath46 [ up ] & @xmath46 [ down ] & @xmath47 bound ( @xmath48 yr ) + @xmath49 ' '' '' ' '' '' & 460 & @xmath50 & @xmath51 & @xmath52 @xcite , @xmath53 @xcite + @xmath54 ' '' '' & 340 & @xmath55 & @xmath56 & @xmath57 @xcite , @xmath58 @xcite + @xmath59 ' '' '' & 310 & @xmath60 & @xmath61 & @xmath62 @xcite + existing searches have been optimized for meson momenta @xmath63 mev , while for ind mesons are typically much more energetic , with momenta @xmath64 gev . here , we briefly summarize existing snd search strategies and how they might be adapted for ind searches . @xmath65 * : * the super - kamiokande experiment , a water erenkov detector , provides the strongest limits on these channels . for @xmath66 , they have three searches : ( _ i _ ) @xmath67 , giving three erenkov rings , ( _ ii _ ) @xmath68 with a prompt @xmath69 ( from @xmath70 ) , and ( _ iii _ ) a mono - energetic @xmath71 from @xmath72 , with no prompt @xmath69 . all three searches assume , as is the case for @xmath73 in snd , that the @xmath66 is emitted below erenkov threshold ( @xmath74 ) and comes to rest before decaying . for ind , we estimate that an @xmath75 fraction of @xmath66 s do come to rest before decaying . however , except for up - scattering events close to kinematic threshold , the @xmath66 from ind has @xmath76 , adding an extra ring to the event topology . furthermore , this additional radiation may make finding the prompt @xmath69 in search method ( _ ii _ ) more difficult . for @xmath77 , they have two searches : ( _ iv _ ) @xmath78 , giving four @xmath79-like rings , and ( _ v _ ) @xmath80 , giving two @xmath81-like rings . both searches assume @xmath82 mev , thereby excluding ind events ( again , except for up - scattering near kinematic threshold ) . one difficulty in search ( _ iv _ ) is identifying all four @xmath79-like rings . for ind , this may be more difficult as the rings would be more overlapping due to relativistic beaming . on the other hand , search ( _ v _ ) is promising for ind . in snd this mode suffers from a small efficiency that the @xmath83 are both above erenkov threshold . with greater energetics in ind , the efficiency may be much larger . @xmath84 * : * the best limit is provided by the soudan 2 experiment , an iron tracking calorimeter . nucleon decay event candidates were required to have a single @xmath85 track , with ionization consistent with mass @xmath86 or @xmath87 , initial momentum @xmath88 mev , and visible endpoint decays @xmath89 ) . their simulations showed that a @xmath85 originating from within an iron nucleus loses on average half its initial momentum . at higher @xmath90 , ind events may be more visible due to reduced background from atmospheric neutrinos . however , it is unknown to us what is the average momentum deposition in iron of the @xmath85 at much higher energy , and whether this can lead to fragmentation of the parent nucleus into exotic nuclear states . @xmath91 * : * the best limits on these modes come from the imb-3 experiment , a water erenkov detector . mev ( 20 events vs. 6.1 expected background ) @xcite . nearly all these events had between @xmath92 erenkov rings , large missing momenta ( @xmath93 mev ) , and large invariant masses ( @xmath94 mev ) . in ind , a large missing momentum would be expected , while a large invariant mass could arise through heavy meson or multi - meson final states . ] the @xmath95 channel may be more difficult at higher energies : due to decreased separation angle of the two photons ( from relativistic beaming ) , they can appear as a single electron - like track . the @xmath96 channel will have greater photon separation and may be more promising . we find that the ind rates into @xmath97 and @xmath39 final states are comparable and are sensitive to the same underlying ind operator ( shown below ) . an effective ind lifetime can be defined as the inverse scattering rate per target nucleon , @xmath98 , with local dm number density @xmath99 and ind scattering cross section @xmath100 . numerically , we have _ n^-1 ( 10 ^ 32 ) ^-1 ( ) ( ) . next , we compute @xmath100 using chiral perturbation theory . we perform an expansion in powers of @xmath101 , where @xmath102 mev is the pion decay constant , and truncate at leading order . since for ind we expect @xmath103 , our calculations should be regarded as order - of - magnitude estimates at best . our analysis closely follows snd rate computations in ref . @xcite . there are four effective interactions that are relevant for ind processes with single meson final states . these are given by @xmath104 , with operators ( given in two - component spinor notation ) @xmath105 \\ o_3 & = \frac{1}{\sqrt{2 } } \ , \epsilon_{\alpha\beta\gamma } \phi [ ( d^\alpha_r s_r^\beta)(u_r^\gamma \psi_r ) - ( s^\alpha_r u_r^\beta)(d_r^\gamma \psi_r ) ] \end{aligned}\ ] ] where @xmath106 are color indeces , and the coefficients @xmath107 have mass dimension @xmath108 . is relevant only for multi - kaon final states . a fifth operator @xmath109 $ ] vanishes by a fierz identity . ] the linear combinations have been chosen such that @xmath110 have strong isospin @xmath111 , respectively . here , it is useful to write @xmath112 where c ( ccc + & 0 & 0 + 0 & - & 0 + 0 & c_1 & - c_2 ) , o_ij _ _ jk ( q_rk^q_r^ ) ( q^_ir _ r ) , with @xmath113 . under @xmath114 chiral symmetry transformations , the right - handed quark fields transform in the @xmath115 representation , @xmath116 ( where @xmath117 ) , while the ind operator @xmath118 transforms in the @xmath119 representation , @xmath120 . if we treat @xmath121 as a spurion in the @xmath119 representation , @xmath122 is invariant under chiral transformations . ( the dm fields @xmath15 are chiral singlets . ) the ind interactions of baryon and ( pseudo - goldstone ) meson fields are determined by the chiral transformation properties of the spurion @xmath121 . following the conventions of ref . @xcite , the only invariant operator is [ eq : leff ] _ ind = [ c ^(b_r _ r ) ] , where @xmath123 . the meson and baryon fields are m = ( ccc + & ^+ & k^+ + ^- & - & k^0 + k^- & |k^0 & - ) , b = ( ccc + & ^+ & p + ^- & - & n + ^- & ^0 & - ^0 ) . is invariant since the quantity @xmath124 is in the @xmath119 representation @xcite . the unknown overall coefficient @xmath125 has been computed using lattice methods @xcite . the feynman diagrams for ind are given in fig . [ indfeyn ] . it is straight - forward to derive the feynman rules for the interactions of baryons , mesons , and dm by expanding the matrix expressions in eq . and working to linear order in @xmath126 . from these , we compute the matrix elements and cross sections for ind processes . we assume that only one coefficient @xmath107 is non - zero at a time . @xmath127 modes depend only on @xmath128 , while @xmath54 modes are governed by @xmath129 . for @xmath130 ( left ) and @xmath131 ( right ) as a function of fermion dm mass @xmath40 for @xmath132 . dotted ( dashed ) lines denote @xmath133 ( @xmath134 ) . solid lines denote total rates @xmath133 + @xmath134 . @xmath135 @xmath136/s corresponds to lifetime @xmath137 years.__,title="fig : " ] for @xmath130 ( left ) and @xmath131 ( right ) as a function of fermion dm mass @xmath40 for @xmath132 . dotted ( dashed ) lines denote @xmath133 ( @xmath134 ) . solid lines denote total rates @xmath133 + @xmath134 . @xmath135 @xmath136/s corresponds to lifetime @xmath137 years.__,title="fig : " ] our numerical results are shown in figs . [ ind1 ] and [ ind2 ] , for the five decay modes we consider . we plot the velocity - weighted cross sections @xmath100 for each mode , as a function of @xmath138 , for the allowed range @xmath139 ( as required by dm stability ) . in fig . [ ind1 ] , the solid lines show the total cross sections for the channels @xmath140 and @xmath141 ( left ) and @xmath131 ( right ) . all three modes arise from the same ind operator @xmath142 , and we have fixed @xmath143 . the individual cross sections for @xmath144 ( @xmath145 ) are shown by the dotted ( dashed ) curves . although the individual rates vanish where kinematically forbidden , the total rate is always non - vanishing . for @xmath146 , both up- and down - scattering rates are non - zero , and ind is bichromatic . otherwise , only down - scattering is allowed , and ind is monochromatic . for @xmath147 ( left ) and @xmath148 ( right ) as a function of fermion dm mass @xmath40 for @xmath149 . dotted ( dashed ) lines denote @xmath150 ( @xmath151 ) from operators @xmath152 . solid lines denote total rates @xmath153 + @xmath151 . grey regions show where existing nucleon decay bounds apply , described in text.__,title="fig : " ] for @xmath147 ( left ) and @xmath148 ( right ) as a function of fermion dm mass @xmath40 for @xmath149 . dotted ( dashed ) lines denote @xmath150 ( @xmath151 ) from operators @xmath152 . solid lines denote total rates @xmath153 + @xmath151 . grey regions show where existing nucleon decay bounds apply , described in text.__,title="fig : " ] in fig . [ ind2 ] , we show the strange ind channels @xmath147 ( left ) and @xmath148 ( right ) . here , there are two relevant operators @xmath152 . the solid lines denote the total ind cross sections induced by each @xmath154 independently , taking @xmath155 , for @xmath156 . the individual cross sections for @xmath144 ( @xmath145 ) are shown by the dotted ( dashed ) curves . again , the kaons can be monochromatic or bichromatic , depending on whether up - scattering is kinematically allowed . existing nucleon decay bounds do apply in select regions of parameter space for up - scattering close to threshold , where the meson momentum is reduced . we illustrate these regions in fig . [ ind2 ] , shown in grey , for the case of super - kamiokande . for @xmath147 , this region corresponds to @xmath157 : in this case , the @xmath66 is below erenkov threshold and the event topology is identical to @xmath158 . for @xmath148 , this region corresponds to a kinematic window @xmath82 mev , as in their @xmath159 search . we emphasize that super - kamiokande bounds constrain only the up - scattering ind rate , which can be suppressed compared to down - scattering in these parameter regions . the total ind rate depends sensitively on the unknown mass scale @xmath160 . the total ind cross sections and nucleon lifetimes for all channels are comparable , scaling as ( v)_ind 10 ^ -39 ^3/ ( ) ^-6 , _ n 10 ^ 32 ( ) ^6 ( ) . as we show in sec . [ sec : coll ] , the collider bound on this scale is @xmath161 gev . therefore , @xmath162 can in principle be as low as @xmath163 years . it is likely that such a short lifetime would be excluded from nucleon decay searches , but no dedicated ind search has yet been performed . we also note that future nucleon decay experiments are envisioned to have markedly better reach for @xmath164 . for example , a water erenkov detector with @xmath165 kton @xmath166 year of exposure can reach @xmath167 yr @xcite ; a similar capability is expected to be achieved with a liquid argon detector with @xmath168 kton @xmath166 year of exposure @xcite . for kaon final states , liquid argon technology is expected to provide improved efficiency due to better imaging capabilities @xcite . lastly , we note that the ind cross sections satisfy certain relations , as a consequence of strong isospin symmetry : ( v)_ind^p ^+ = 2 ( v)_ind^n ^0 , ( v)_ind^p k^+ = ( v)_ind^n k^0 . the latter relation holds only if @xmath54 modes are dominated by either @xmath169 or @xmath170 , as assumed in fig . if neither operator is negligible , then @xmath171 . in this case , both @xmath38 modes are complementary and can be used to disentangle the underlying ind operator structure . adm scenarios rely on transfer operators to connect global charge between the visible and dark sectors , and these operators can be probed at high - energy colliders . in the specific hylogenesis model of ref . @xcite , @xmath7 is mediated between the two sectors by heavy dirac fermions @xmath21 ( @xmath172 ) , through interactions of the form - _ a=1,2 ( u_r^i d_r^j)(x_a , l^ d^k_r ) + _ a ( x_a , l_l+x_a , r_r ) + [ eq : couplings ] where @xmath8 label generation , color indices are implicitly contracted antisymmetrically , and other fermion contractions are also possible . integrating out @xmath21 generates operators of the form of eq . . the @xmath21 particles can be produced at high - energy hadron colliders through the operator of eq . . with decays @xmath173 , this gives rise to events involving missing energy and one or more jets . in this section , we investigate the sensitivity of the tevatron and the lhc to such events and we derive a corresponding bound on the heavy mass scale @xmath27 suppressing the neutron portal operator . related studies in the context of wimp and other dark matter candidates can be found in refs . @xcite . to be concrete , we will focus on the lighter state @xmath174 , with the specific interaction - & & ( x^_l s_r)(u_r d_r ) + x+ h.c . , [ op1 ] we expect other flavour structures and fermion contractions to give qualitatively similar results . the operator of eq . can give rise to processes of the form q(p_1 ) q^(p_2 ) |q^(p_3 ) @xmath175 quarks . the corresponding summed and averaged squared matrix element can take two possible forms , depending on how the fermions are contracted . they are : = \ { cc ||^2 ||^2 ( p_1p_2 ) ; & s + & + . [ matrixel ] here , @xmath176 is the momentum carried by the intermediate @xmath177 state , and @xmath178 is the width of the @xmath177 state which we assume decays mainly into @xmath179 . the @xmath180-like form correponds to the case where both initial state fermions are contracted together in the underlying operator while the @xmath181-like form corresponds to a contraction between initial and final state quarks . at the tevatron and the lhc , we find that the full cross section derived from the matrix elements of eq . is frequently dominated by the pole in the intermediate @xmath177 propagator . this corresponds to the production of an on - shell @xmath177 state whose mass is not much smaller than the higher - dimensional operator scale @xmath27 , and corresponds to a large momentum transfer . in this case the higher - dimensional operator structures we are using become unreliable and the full dynamics of the unknown ultraviolet completion becomes relevant . without specifying the underlying theory , we can still parametrize the generic behaviour in a reasonable way as follows . for @xmath180-like contractions , whose structure would arise most naturally from a boson in the @xmath180-channel , we make the replacement , where @xmath182 is the parton - level mandelstam variable and @xmath183 is the decay width of the mediator . in the absence of an underlying theory , we parametrize this quantity as @xmath184 and consider the values @xmath185 . for @xmath181-like contractions , we make the replacement , where @xmath186 is the parton - level mandelstam variable . since @xmath187 is negative , we do not bother adding a width term . we apply these matrix elements to compute the leading - order ( lo ) monojet production cross sections at the tevatron by convolving with cteq6.1 m parton distribution functions @xcite and integrating over phase space . to match the most stringent tevatron monojet search bounds , we impose a cut of @xmath188 and @xmath189 on the outgoing jet . following ref . @xcite , we also apply a flat efficiency factor of @xmath190 to connect our parton - level cross section to the full hadronic jet reconstruction at the tevatron detectors . the cross sections computed in this way ( after applying cuts and the efficiency factor ) are shown in fig . [ tevatron ] for the parameter values @xmath191 and @xmath192 . we also set @xmath193 and @xmath194 . this figure shows a significant resonant enhancement when the intermediate state is narrow . a similar enhancement was seen in ref . the resonant enhancement only becomes fully operational at @xmath195 due to the cut on jet @xmath196 . the horizontal dashed line in fig . [ tevatron ] represents the current tevatron @xmath197 exclusion limit on the net monojet cross section of @xmath198 ( after cuts ) @xcite , based on the cdf analysis of refs . @xcite which uses the same set of jet cuts as applied to our signal estimates . this limit translates into a lower bound of @xmath199 for the operator of eq . , depending on the mass of the @xmath177 state and the width of the unspecified intermediate state . monojet signals can also be detected at the lhc . due to the expectation of a significant amount of associated qcd radiation , ref . @xcite investigated the reach of an inclusive search for a hard jet plus missing energy search at atlas with no veto on additional hard jets . to match this analysis , we compute the inclusive leading - order parton - level cross section at the lhc with @xmath200 subject to the cuts @xmath201 and @xmath202 on the outgoing jet . we also rescale the cross section by a conservative acceptance / efficiency factor of @xmath203 @xcite . the corresponding cross sections are shown in fig . [ lhc ] for several values of @xmath27 with @xmath191 and @xmath192 . we also set @xmath204 and @xmath194 . based on the background analysis of ref . @xcite , the study in ref . @xcite estimated a net sm background production rate after the applied cuts of @xmath205 . applying a simple @xmath206 measure on the detection significance , this leads to a sensitivity to monojet cross sections as small as @xmath207 ( @xmath208 ) with @xmath209 ( @xmath210 ) of data at @xmath211 . this is shown by the dotted lines in fig . the resulting lhc reach for the operator of eq . lies in the range @xmath212 . this is competitive with the sensitivity to this operator from searches for ind in existing nucleon decay experiments . additional quark operators beyond that given in eq . and considered above are expected to yield qualitatively similar collider rates and signatures in most cases . an interesting further possibility are quark operators involving top or bottom quarks . with a bottom quark in the final state , the monojet signal could be augmented with a @xmath213-tag . in the case of a top quark in the final state , the signal would be a hard single top quark with large missing energy . while the search reach in monobottom and monotop channels is likely greater than for the light quark channels , the operators involving third - generation quarks do not correlate directly with ind processes . the hylogenesis model in ref . @xcite also contains a hidden @xmath23 gauge symmetry that couples to @xmath15 , is spontaneously broken at the gev - scale , and couples to the sm only through kinetic mixing with hypercharge . direct pair production of dm particles via the corresponding @xmath214 vector boson with an associated jet can be another source of monojet signals @xcite . however , since this vector is relatively light , the existing bounds from the tevatron and the expected reach of the lhc are both much weaker than the limits from dm direct detection through elastic scattering mediated by the @xmath214 . dm can be captured in stars by elastic scattering with the nuclei they contain . once captured , ordinary self - annihilating dm will collect in the middle of the star and annihilate with other dm particles there , releasing energy @xcite . self - annihilation is not possible for hylogenic dark matter ( hdm ) due to its conserved global charge . instead , it can potentially destroy baryons within the star via ind processes such as @xmath215 . if it remains within the star , the anti - hdm @xmath216 reaction product can then annihilate with a @xmath12 particle already captured in the stellar core . the net result of this chain is the destruction of a baryon and the release of energy , both from the meson decay and the annihilation step . in this section we investigate the effects of ind processes on several varieties of stellar species including neutron stars , white dwarfs , and main - sequence stars . to be concrete , we will concentrate on the specific model of hylogenesis presented in ref . @xcite and described in appendix [ sec : appa ] with ind mediated by the operator of eq . . we will assume a fiducial spin - independent proton scattering cross section of @xmath217 ( and @xmath218 ) for both @xmath11 and @xmath12 induced by their coupling to the kinetically - mixed @xmath23 vector boson , , this value is consistent with existing direct detection searches @xcite . ] as well as an @xmath219-anti-@xmath219 annihilation cross section of @xmath220 . we shall consider two cases for the ind cross section : a large value of @xmath221 , and a small value of @xmath222 . the general behaviour for other ind rates will lie somewhere between these two extremes . relic @xmath11 and @xmath12 particles will be captured in stars by scattering with nuclei to energies below the local escape velocity . once captured , an hdm particle will undergo further scatterings , thermalize with the baryons in the star , and collect within the stellar core . this occurs quickly relative to the lifetimes of the stars we consider here for our large fiducial value of the proton scattering cross section . once they thermalize , the dm particles are largely confined to a core of radius @xcite r_i , th = ( ) ^1/2 , where @xmath223 , @xmath224 is the mean temperature and @xmath225 is the mean ( baryon ) density in the stellar core . the evolution of the total numbers of @xmath11 , @xmath12 , @xmath226 , and @xmath227 within a star is described the following system of equations : & = & c_- a_n_n_| - b_n _ [ neqy ] + & = & - a_n_n_| + _ |b_n _ [ neqyb ] + & = & c _ - a_n_n_^ - b_n _ [ neqf ] + & = & - a_n_n_^ + _ ^b_n _ [ neqfb ] here , the @xmath228 coefficients are the hdm capture rates , the @xmath229 coefficients describe hdm - anti - hdm annihilation , and the @xmath230 coefficients describe ind . a general expression for @xmath228 is given in ref . @xcite , while the @xmath231 and @xmath7 coefficients are given to a good approximation by a_i & & ( v)_i , ann / ( 4r_i , th^3/3 ) , + b_i & & ( v)_i , ind ( _ c / m_n ) , where @xmath232 is the mass of a nucleon . the @xmath233 terms appearing in eqs . and are the probabilities for the anti - hdm products of ind to be captured by the host star after they are created . in certain regimes additional processes can influence the evolution of the stellar populations of hdm and anti - hdm such as evaporation @xcite and direct annihilation to baryons ( _ e.g. _ @xmath234 ) . we will discuss these effects when they may be relevant . neutron stars are very dense objects supported by the fermi degeneracy pressure of their neutrons . despite their name , they also contain a significant mass fraction of protons and heavier nuclei , and the nuclear state of their cores is not fully understood @xcite . typical neutron star parameters are mass @xmath235 , with @xmath236 kg the solar mass , radius @xmath237 , core temperature @xmath238 , and core baryon density @xmath239 @xcite . the rate of capture of @xmath11 or @xmath12 upon a typical neutron star ( including general relativistic corrections ) is @xcite c_i 2.510 ^ 25 ^ -1 ( ) ( ) ( ) f , [ nscapture ] where @xmath240 is the local dm energy density , @xmath241 is the local dm velocity dispersion , and f = min\{1,(x_p_p+x_n_n)/(210 ^ -45 ^2 ) } , with @xmath242 and @xmath243 being the proton and neutron mass fractions . the factor @xmath244 accounts for the saturation of the cross section at the cross - sectional area of the star . this saturation sets in when the star becomes optically thick to dm when a dm particle impinging upon the star is likely to scatter multiple times with nucleons in the star . for the fiducial nucleon scattering cross section we are using and assuming a proton mass fraction of @xmath245 , we find that neutron stars are optically thick to both @xmath11 and @xmath12 . this leads to @xmath246 as well as @xmath247 . we also find r_i , th ( 140)()^1/2()^1/2 ( ) ^1/2 , implying @xmath248 and @xmath249 for the large ( small ) ind rate @xmath221 ( @xmath250 ) . consider first the case of a large ind rate . using the evolution equations of eqs . ( [ neqy][neqfb ] ) and assuming negligible initial hdm densities , we find that the numbers of hdm and anti - hdm particles within a typical neutron star reach a steady state . this behaviour is shown in fig . [ neutron ] for the fiducial cross sections given above and @xmath251 , @xmath252 . the steady - state populations in the limit of @xmath253 are approximated well by the analytic expressions n _ , n_| ( ) , n _ , n_^ ( ) , with @xmath254 . , we find : @xmath255 . ] a steady state is also attained when one of the ind rates vanishes due to kinematic suppression . for example , with @xmath256 and @xmath257 , we find n_= , n_|=0 , n_,^= + , where in the last equality the plus sign corresponds to @xmath12 and the minus sign to @xmath216 . the solution for @xmath258 and @xmath259 is identical but with @xmath12 and @xmath11 interchanged in the expressions above . in both cases , the steady - state particle populations are on the order of @xmath260 and @xmath261 ( when they are non - zero ) for the fiducial input values listed above . the time needed to reach this steady state from a negligible initial dark matter density in the star is on the order of _ ss ~max\{b^-1,}. where the first value corresponds to ind balancing capture and the second to annihilation balancing production of anti - hdm by ind . for the fiducial cross sections we are considering , we find that annihilation takes longer to balance , and leads to @xmath262 . this is ultra short relative to the lifetime of a typical neutron star . having reached a steady state , the main combined effect of the ind and annihilation processes is to inject energy into the host neutron star with rate @xmath263 . in this respect , hdm in the steady state limit has the same effect on neutron stars as ordinary self - annihilating dm . energy injection by dm can interfere with and halt the cooling of old neutron stars , and therefore the observation of a very cool , old neutron star in a region of large dm density could put significant constraints on a wide variety of dm scenarios @xcite . however , this effect is too small to be observed using existing observations , and appears to be challenging to probe in the near future @xcite . with hdm , baryons within the neutron star are also destroyed by ind , but the number is negligible compared to the total of @xmath264 within a typical neutron star over the lifetime of the universe unless the local dm density approaches an enormous value of @xmath265 . consider next the case of a vanishingly small ind cross section . the populations of of @xmath11 and @xmath12 hdm particles will now build up within the neutron star with rate given by eq . , corresponding to populations of about @xmath266 over the lifetime of the universe . we can compare this number to the populations required for dm particles to begin self - gravitating and to form a black hole . self - gravitation begins when @xcite n_i n_self(4r_i , th^3/3 ) 310 ^ 45()^5/2()^3/2 ( ) ^1/2 . [ nself ] this is much greater than the steady - state populations found above for the case of a large ind rate , and also greater than the numbers collected for a smaller ind rates for local hdm densities below @xmath267 . self - gravitating dm particles will form a black hole that can potentially destroy the host star unless they are stabilized in some way @xcite . for non - interacting fermions there is a degeneracy pressure that must be overcome . to do so , the total number of fermions must exceed n_i n_crit^f ( ) ^3 610 ^ 55()^3 , [ ncritf ] with @xmath268 the reduced planck mass . in the case of non - self - interacting bosons , there is still a zero - point pressure which can be overcome if the number of bosons exceeds n_i n_crit^b ( ) ^2 210 ^ 37()^2 . a bosonic black hole can also arise even before bulk self - gravitation sets in through the formation of a bose - einstein condensate @xcite . we argue that black holes are not likely to form in the specific theory of hylogenesis discussed in ref . @xcite and appendix [ sec : appa ] , in which @xmath12 and @xmath11 carry equal and opposite charges under a spontaneously broken @xmath23 gauge symmetry , until @xmath269 . at distances much smaller than the inverse vector mass @xmath270 , the effects of breaking the @xmath23 can be neglected and the @xmath214 vector boson mediates a repulsive force between particles with like - sign charges . this induces a pressure among @xmath12 particles that prevents them from collapsing into a black hole provided @xmath271 and @xmath272 ( where @xmath273 is the @xmath23 coupling of @xmath12 ) @xcite . instead , under these mild assumptions the formation of a black hole requires an approximately charge - neutral collection of hdm particles and therefore roughly equal numbers of @xmath11 and @xmath12 states . a necessary condition for this to occur is @xmath274 to overcome the fermion degeneracy pressure . therefore we do not expect this specific theory of hylogenesis to lead to the formation of hdm black holes within neutron stars unless the local hdm density around the host star exceeds @xmath275 . between the two extremes of the large and small ind rates considered above , no new obvious observational bounds arise . with a small but non - zero rate for ind , the @xmath11 and @xmath12 populations may grow large enough to form some amount of anti - hdm that will subsequently annihilate away . large hdm densities can also lead to direct annihilation of hdm to antibaryons , @xmath276 , and similarly for anti - hdm . this can be accounted for by adding terms of the form @xmath277 to the evolution equations for @xmath11 and @xmath12 , with @xmath278 and @xmath279 is the volume occupied by the hdm , whose evolution we must also keep track of . note that @xmath279 appears here rather than @xmath280 because this effect starts to compete with standard ind only after the hdm population begins to self - gravitate , when @xmath281 . we defer an analysis of the effects of self - gravitation and self - annihilation of hdm on the structure of neutron stars to a future work . we can perform a similar analysis for the effects of hdm on white dwarfs . a typical white dwarf consists primarily of carbon and oxygen and is supported by the degeneracy pressure of the electrons it contains . typical white dwarf parameters are @xmath282 , @xmath283 , @xmath284 , @xmath285 @xcite . we will approximate the internal structure of a white dwarf as consisting entirely of carbon with a uniform density @xcite . the capture rate of dm through elastic scattering with nuclei in a white dwarf is approximated well by the expression @xcite c_i ( ) ^1/2 ( ) ( ) _ eff , with @xmath241 being the local velocity dispersion , @xmath286 the escape velocity at the surface of the star , and the the effective cross section is defined to be _ eff min\{r^2 , ^si_p_k ( ) ^2 } , where the sum runs over all nuclear species @xmath287 in the star with mass fraction @xmath288 and atomic mass ( number ) @xmath289 ( @xmath290 ) , @xmath291 is the spin - independent nucleon scattering cross section off protons , @xmath292 and @xmath293 are the relative coupling strengths of the dm to protons and neutrons , and @xmath294 ( @xmath295 ) is the reduced mass of the dm - nucleus ( dm - proton ) system . the saturation of the effective cross section at @xmath296 corresponds to the star becoming optically thick . we find that a typical white dwarf is indeed optically thick for the fiducial scattering cross section of @xmath297 ( with @xmath298 , @xmath299 ) . this gives c_&=&c _ c + & & ( 610 ^ 27 ^ -1 ) ( ) ( ) ( ) ( ) ( ) , where @xmath300 km is the solar radius . once captured , hdm particles thermalize rapidly and collect within a thermal radius at the core of the star of size r_i , th ( 510 ^ 7)()^1/2 ( ) ^1/2 ( ) ^1/2 . for @xmath301 , we find @xmath302 and @xmath303 ( @xmath304 ) for the case of a large ( small ) ind rate @xmath221 ( @xmath250 ) in the case of a large ind rate , we find that the populations of @xmath11 , @xmath12 , @xmath226 , and @xmath305 all reach a steady state quickly relative to the lifetime of a typical white dwarf . in this state we find @xmath306 and @xmath307 for the fiducial parameter values listed above . the number of baryons destroyed by ind over the age of the universe is a small fraction of the total number in a white dwarf provided @xmath308 . the main effect of hdm capture , ind , and annihilation on white dwarf is therefore the injection of energy into the stellar interior with rate @xmath263 . this is equivalent to the effect of ordinary self - annihilating dm ( with slightly larger mass ) . heating of white dwarfs by dm capture and annihilation was studied recently in refs . these authors differ in their conclusions , with ref . @xcite finding an upper bound on the nuclear capture cross section of about @xmath309 and ref . @xcite finding essentially no bounds from current observations . the origin of the disagreement is the density of dm within globular clusters , from which ref . @xcite obtains their most stringent bounds , while ref . @xcite argues that they contain a much smaller dark matter abundance . the observation of a cool white dwarf within a dwarf spheroidal galaxy could help to resolve this question decisively @xcite . based on this uncertainty , we do not consider hdm models with nuclear scattering cross sections as large as @xmath310 to be ruled out . however , we do note that the model of hdm considered here can also accommodate much smaller elastic scattering cross sections . with a small ind rate , hdm collects and builds up within the stellar interior . as we argued above , we do not expect any significant effects on the star until at least @xmath311 given in eqs . ( [ nself],[ncritf ] ) . for this to occur over the lifetime of the universe , a local hdm density approaching @xmath312 is needed . we do not know of any observations of white dwarfs that approach these requirements . the capture rate of light ( @xmath313 ) dm in the sun is dominated by scattering with hydrogen and helium and is given to a good approximation by @xcite c_i & & ( 810 ^ 25 ^ -1 ) ( ) ( ) ( ) ( ) + & & , where the relative factor for helium relative to hydrogen comes from its slightly different distribution within the sun . in writing this expression we have also made use of the fact that the sun is optically thin for our fiducial nucleon elastic scattering cross section . a second consequence of the sun being optically thin is that nearly all the anti - hdm produced by ind escapes from the sun , since it is typically produced with velocities much larger than @xmath286 . thus we set @xmath314 in our evolution equations . the thermal radius in which the dm collects is r_i , th ( 510 ^ 9 ) ( ) ^1/2 ( ) ^1/2()^1/2 which yields @xmath315 for the large ( small ) ind rate . an additional effect that is important for the sun is evaporation , in which captured hdm particles are up - scattered above the escape velocity and leave the sun @xcite . fitting to the results of refs . @xcite , we take the evaporation rate to be e_i 10^[-3.5(m_i/)-4 ] ( ) ^-1 . this evaporation adds a dissipative @xmath316 term to each of eqs . ( [ neqy][neqfb ] ) . for the fiducial parameters we are considering ( and a large ind rate ) , evaporation becomes more important than ind for hdm masses below @xmath317 . we find that capture , evaporation , and ind reach a steady state on a timescale of @xmath318-@xmath319 billion years , much shorter than the age of the sun . the equilibrium populations are always less than @xmath320 , for both large and small ind rates . this is a very small fraction of the total mass of the sun , and is far below what is needed for hdm to self - gravitate or to contribute significantly to the properties of the solar interior @xcite . note as well that due to the low masses of hdm , the energies of any neutrinos produced by ind are well below the thresholds of dm searches for upward - going muons in neutrino telescopes such as ice cube @xcite . the effects of hdm on other main - sequence stars are also expected to be small , unless the star is immersed in a region of very large dm density @xcite . we have investigated signals of nucleon destruction that can arise in specific theories where dark matter is antibaryonic . antibaryonic dm is motivated by unified mechanisms for dm and baryon generation ( referred to as hylogenesis ) which address the cosmic coincidence between the energy densities of dark and visible matter in the universe . in the hylogenesis scenario considered here , the dm consists of an asymmetric density of fermions @xmath11 and scalars @xmath12 . these particles can scatter inelastically with nucleons via the reactions @xmath321 and @xmath322 , where @xmath30 is a nucleon and @xmath27 is a meson . these induced nucleon decay ( ind ) processes lead to distinctive signatures in nucleon decay searches , stellar evolution , and hadron colliders . ind is a novel signature of dm that can be searched for in terrestrial nucleon decay searches . the effective nucleon lifetime is expected to be @xmath323 years , if baryon transfer between dark and visible sectors is mediated by new physics at a scale @xmath324 . since the dm states are unobserved , ind events mimic standard nucleon decay with neutrino final states , but typically with greater final state meson energy @xmath325 . due to these different kinematics , existing searches do not directly apply in general . our study therefore motivates new searches in these experiments . we expect that the resulting sensitivities should be comparable to those in standard nucleon decay searches . the coupling of hylogenic dm to quarks responsible for ind can also give rise to observable signals in hadron colliders . in particular , such couplings can potentially lead to monojet signals at the tevatron and the lhc . existing monojet searches at the tevatron place a lower bound on the coupling of hylogenic dm to quarks , and this bound will be significantly improved at the lhc . we find that the coupling strengths that can be probed at the lhc are of the same size as those that can produce an observable effect in nucleon decay search experiments . this correlation may permit the characterization of hylogenic dm through a very diverse set of experimental probes . if hylogenic dm also scatters elastically with nucleons , it can be captured in stars . once captured , it will thermalize with the baryons in the star and sink to the stellar core . in compact stars , the local baryon density in the stellar core may be large enough for ind to occur at a significant rate , destroying both a nuclon and the hdm particle , and producing an anti - hdm particle and a meson . this process deposits energy in the star , an effect that may be probed by observing old and cool neutron stars or white dwarfs . however , current observations do not provide a definitive bound on or evidence for these processes . we thank brian batell , matthew buckley , robert caldwell , alejandra castro , jim cline , katie freese , patrick fox , michael graesser , ed kearns , jennifer kile , john ng , scott oser , maxim pospelov , michael - ramsey - musolf , jen raaf , adam ritz , pat scott , rishi sharma , hirohisa tanaka , and kathryn zurek for helpful discussions . we also thank jennifer kile for collaboration in the early stages of this work . dm and st would like to thank the weizmann institute of science for their hospitality . ks would like to thank perimeter institute for theoretical physics for their hospitality . the work of hd is supported by the united states department of energy under grant contract de - ac02 - 98ch10886 . the work of dm , ks , and st is supported by the national science and engineering research council of canada ( nserc ) . in this appendix we review some of the important features of the hylogenesis model presented in ref . the model consists of the sm together with a sm - neutral hidden sector containing two massive dirac fermions @xmath326 ( @xmath327 , with masses @xmath328 tev ) , a dirac fermion @xmath11 , and a complex scalar @xmath12 ( with masses @xmath329 gev ) . these fields couple through the `` neutron portal '' ( @xmath330 ) and a yukawa interaction : - ( x_a , l^d^k_r ) ( u_r^id_r^j ) + _ a ( x_a , l_l+x_a , r_r ) + [ eq : couplings ] where @xmath8 label flavors and the quark color indices are implicitly contracted antisymmetrically . many variations on these operators exist , corresponding to different combinations of quark flavors and spinor contractions . with this set of interactions one can define a conserved generalized global baryon number with charges @xmath331 . the proton , @xmath11 , and @xmath12 are stable due to their @xmath7 and gauge charges if their masses satisfy |m_- m_| < m_p + m_e , m_p - m_e < m_+ m _ . [ eq : massdiff ] @xmath11 and @xmath12 are the `` hidden antibaryons '' that comprise the dark matter . furthermore , there exists a physical cp - violating phase @xmath332 that can not be removed through phase redefinitions of the fields . we also introduce a hidden @xmath333 gauge symmetry under which @xmath11 and @xmath12 have opposite charges @xmath334 , while @xmath326 is neutral . we assume this symmetry is spontaneously broken at the gev scale , and has a kinetic mixing with sm hypercharge @xmath335 via the coupling @xmath336 , where @xmath337 and @xmath338 are the @xmath335 and @xmath23 field strength tensors . at energies well below the electroweak scale the effect of this mixing is primarily to generate a vector coupling of the massive @xmath214 gauge boson to sm particles with strength @xmath339 . the @xmath340-scale @xmath214 masses we consider here can be consistent with observations for @xmath341 @xcite . in the mechanism for hylogenesis presented in ref . @xcite , baryogenesis begins when a non - thermal , cp - symmetric population of @xmath22 and @xmath342 is produced in the early universe . these states decay through @xmath343 or @xmath344 ( and their conjugates ) . an asymmetry between the partial widths for @xmath345 and @xmath346 arises from interference between tree and loop diagrams and is characterized by = , where we have assumed that the total decay rate @xmath347 is dominated by @xmath348 over the three - quark mode , and that @xmath349 . for @xmath350 , @xmath22 decays generate a baryon asymmetry in the visible sector , and by cpt an equal and opposite baryon asymmetry in the hidden sector . these asymmetries can be `` frozen in '' by the weakness of the coupling between both sectors provided the temperature at which the @xmath22 are produced is not too high . for the asymmetry to be large enough to explain the observed value , @xmath351 can not be too much smaller than @xmath27 . once produced , @xmath11 and @xmath12 will thermalize by scattering with @xmath214 vectors present in the plasma . these interactions will also deplete the symmetric densities of @xmath11 and @xmath12 very efficiently through annihilation to pairs of @xmath214 vectors provided @xmath352 @xcite . only the asymmetries will remain . this is analogous to the annihilation of baryons with antibaryons . the cross section for @xmath353 is given by @xcite v= ( 1.610 ^ -25 ^ 3/)()^4()^2 . annihilation of @xmath354 is given by a similar expression . there is also a direct detection signal in our model due to the hidden @xmath214 mediating the elastic scattering of @xmath11 and @xmath12 off protons . the effective scattering cross section per nucleon for either @xmath11 or @xmath12 is spin - independent and given by _ 0^si = ( 510 ^ -39 ^ 2)()^2()^2 ( ) ^2 ( ) ^2 ( ) ^4 , where @xmath355 is the dm - nucleon reduced mass . for a dm mass of @xmath356 , this is slightly below the best current limits from cresst @xcite , cdms @xcite , and cogent @xcite . for reviews , see : a. riotto , m. trodden , ann . nucl . part . sci . * 49 * , 35 - 75 ( 1999 ) . [ hep - ph/9901362 ] ; m. dine , a. kusenko , rev . phys . * 76 * , 1 ( 2004 ) . [ hep - ph/0303065 ] ; s. davidson , e. nardi , y. nir , phys . rept . * 466 * , 105 - 177 ( 2008 ) . [ 0802.2962 [ hep - ph ] ] . s. nussinov , phys . lett . * b165 * , 55 ( 1985 ) ; s. m. barr , r. s. chivukula , e. farhi , phys . lett . * b241 * , 387 - 391 ( 1990 ) . s. m. barr , phys . * d44 * , 3062 - 3066 ( 1991 ) ; d. b. kaplan , phys . * 68 * , 741 - 743 ( 1992 ) ; d. e. kaplan , m. a. luty , k. m. zurek , phys . rev . * d79 * , 115016 ( 2009 ) . [ 0901.4117 [ hep - ph ] ] ; g. d. kribs , t. s. roy , j. terning , k. m. zurek , phys . rev . * d81 * , 095001 ( 2010 ) . [ 0909.2034 [ hep - ph ] ] ; t. cohen , d. j. phalen , a. pierce , k. m. zurek , phys . * d82 * , 056001 ( 2010 ) , [ 1005.1655 [ hep - ph ] ] . j. shelton , k. m. zurek , phys . rev . * d82 * , 123512 ( 2010 ) . [ 1008.1997 [ hep - ph ] ] ; n. haba , s. matsumoto , [ 1008.2487 [ hep - ph ] ] ; m. r. buckley , l. randall , [ 1009.0270 [ hep - ph ] ] ; e. j. chun , phys . rev . * d83 * , 053004 ( 2011 ) . [ 1009.0983 [ hep - ph ] ] ; j. mcdonald , [ 1009.3227 [ hep - ph ] ] ; m. blennow , b. dasgupta , e. fernandez - martinez , n. rius , jhep * 1103 * , 014 ( 2011 ) , [ 1009.3159 [ hep - ph ] ] ; l. j. hall , j. march - russell , s. m. west , [ 1010.0245 [ hep - ph ] ] ; b. dutta , j. kumar , phys . * b699 * , 364 - 367 ( 2011 ) . [ 1012.1341 [ hep - ph ] ] ; a. falkowski , j. t. ruderman , t. volansky , jhep * 1105 * , 106 ( 2011 ) , [ 1101.4936 [ hep - ph ] ] ; e. j. chun , jhep * 1103 * , 098 ( 2011 ) , [ 1102.3455 [ hep - ph ] ] ; z. kang , j. li , t. li , t. liu , j. yang , [ 1102.5644 [ hep - ph ] ] ; j. j. heckman , s. -j . rey , [ 1102.5346 [ hep - th ] ] ; d. e. kaplan , g. z. krnjaic , k. r. rehermann , c. m. wells , [ 1105.2073 [ hep - ph ] ] . m. t. frandsen , s. sarkar , k. schmidt - hoberg , [ 1103.4350 [ hep - ph ] ] ; a. hook , [ 1105.3728 [ hep - ph ] ] ; n. f. bell , k. petraki , i. m. shoemaker , r. r. volkas , [ 1105.3730 [ hep - ph ] ] ; c. cheung , k. m. zurek , [ 1105.4612 [ hep - ph ] ] . p. -h . gu , phys . * b657 * , 103 - 106 ( 2007 ) . [ 0706.1946 [ hep - ph ] ] ; p. -h . gu , u. sarkar , x. zhang , phys . * d80 * , 076003 ( 2009 ) . [ 0906.3103 [ hep - ph ] ] ; p. -h . gu , m. lindner , u. sarkar , x. zhang , [ 1009.2690 [ hep - ph ] ] . h. an , s. -l . chen , r. n. mohapatra , y. zhang , jhep * 1003 * , 124 ( 2010 ) . [ 0911.4463 [ hep - ph ] ] ; h. an , s. -l . chen , r. n. mohapatra , s. nussinov , y. zhang , phys . rev . * d82 * , 023533 ( 2010 ) . [ 1004.3296 [ hep - ph ] ] . s. dimopoulos , l. j. hall , phys . * b196 * , 135 ( 1987 ) ; j. m. cline , s. raby , phys . * d43 * , 1781 - 1787 ( 1991 ) ; s. d. thomas , phys . * b356 * , 256 - 263 ( 1995 ) . [ hep - ph/9506274 ] ; r. kitano , h. murayama , m. ratz , phys . * b669 * , 145 - 149 ( 2008 ) . [ 0807.4313 [ hep - ph ] ] ; r. allahverdi , b. dutta , k. sinha , phys . rev . * d83 * , 083502 ( 2011 ) . [ 1011.1286 [ hep - ph ] ] . j. goodman , m. ibe , a. rajaraman , w. shepherd , t. m. p. tait , h. -b . yu , phys . * b695 * , 185 - 188 ( 2011 ) . [ 1005.1286 [ hep - ph ] ] ; j. goodman , m. ibe , a. rajaraman , w. shepherd , t. m. p. tait , h. -b . yu , phys . rev . * d82 * , 116010 ( 2010 ) . [ 1008.1783 [ hep - ph ] ] . m. t. frandsen , s. sarkar , phys . lett . * 105 * , 011301 ( 2010 ) . [ 1003.4505 [ hep - ph ] ] ; d. t. cumberbatch , j. .a . guzik , j. silk , l. s. watson , s. m. west , phys . * d82 * , 103503 ( 2010 ) . [ 1005.5102 [ astro-ph.sr ] ] ; m. taoso , f. iocco , g. meynet , g. bertone , p. eggenberger , phys . rev . * d82 * , 083509 ( 2010 ) . [ 1005.5711 [ astro-ph.co ] ] ; j. casanellas , i. lopes , mon . not . astron . soc . * 410 * , 535 - 540 ( 2011 ) . [ 1008.0646 [ astro-ph.co ] ] . i. lopes , j. silk , astrophys . j. * 722 * , l95 ( 2010 ) . [ 1009.5122 [ astro-ph.sr ] ] . m. pospelov , phys . d * 80 * , 095002 ( 2009 ) . j. d. bjorken , r. essig , p. schuster and n. toro , phys . d * 80 * , 075018 ( 2009 ) [ 0906.0580 [ hep - ph ] ] . m. l. graesser , i. m. shoemaker , l. vecchi , [ 1103.2771 [ hep - ph ] ] ; h. iminniyaz , m. drees , x. chen , [ 1104.5548 [ hep - ph ] ] .

we show via triangulations that for there is exactly one class ( under unimodular equivalence ) of nondegenerate lattice simplices in with minimal volume and interior lattice points .

0804.2910

a _ @xmath3-polytope _ @xmath4 is a polytope of dimension @xmath3 . if its _ vertex set _ @xmath5 is a subset of @xmath6 , then @xmath4 is a _ lattice _ @xmath3-polytope . if in addition @xmath7 , then @xmath4 is a lattice _ @xmath3-simplex_. the convex hull of @xmath8 , denoted by @xmath9 , is a lattice polytope with at most @xmath10 vertices and dimension at most @xmath3 . this notation will be used loosely ; for convenience , we use @xmath11 to mean @xmath12 when it is clear @xmath4 is a polytope and @xmath13 is a point . as used in @xcite , we say that @xmath4 is _ clean _ if @xmath14 , where @xmath15 is the boundary of @xmath4 . if in addition @xmath16 , the _ interior _ of @xmath4 , contains @xmath2 lattice points , then @xmath4 is a clean _ @xmath2-point _ lattice polytope . if @xmath17 , then the polytope is _ empty_. we use @xmath18 to denote the collection of clean @xmath2-point lattice @xmath3-simplices . unless otherwise stated , all polytopes are taken to be convex @xmath3-polytopes . reznick proved in @xcite and @xcite that any lattice tetrahedron with at least one clean face is unimodularly equivalent to some @xmath19 , the lattice tetrahedron with vertex set @xmath20 where @xmath21 and @xmath22 . reznick also classified the set of clean 1-point tetrahedra , up to equivalence under unimodular transformations , using barycentric coordinates . very recently , bey , henk , and wills proved in @xcite that if @xmath4 is a lattice @xmath3-polytope , not necessarily clean , and @xmath4 has @xmath2 interior lattice points , then for @xmath23 , the volume of @xmath4 satisifies @xmath24 moreover , they showed that for @xmath25 , equality holds if and only if @xmath4 is unimodularly equivalent to the simplex @xmath26 , where @xmath27 and @xmath28 denotes the @xmath29-th unit point . this is not true for @xmath30 and @xmath31 . we will show that equality holds in ( [ volumeboundeq ] ) for all @xmath32 if and only if @xmath0 and @xmath4 is unimodularly equivalent to @xmath33 . we first prove that if @xmath34 and @xmath35 , then the interior points lie on a line passing through some vertex of @xmath36 . we then show that such simplices are unimodularly equivalent to @xmath37 , the @xmath3-simplex whose vertex set consists of the origin , the points @xmath28 @xmath38 , and the point @xmath39 , where @xmath40 for @xmath41 and @xmath42 . finally , we will show that @xmath43 and @xmath44 . the following definitions are taken from @xcite , @xcite , @xcite , and @xcite . a _ @xmath45-flat _ is a @xmath45-dimensional affine subspace of @xmath1 . _ points _ , _ lines _ , and _ planes _ are @xmath46-flats , @xmath47-flats , and @xmath48-flats , respectively . the _ affine hull _ of a set @xmath49 , denoted by @xmath50 , is the intersection of all flats containing @xmath51 . equivalently , @xmath50 is the smallest flat containing @xmath51 . we say @xmath51 is in _ general position _ if no @xmath52 points of @xmath51 lie in a @xmath45-flat , where @xmath53 . a hyperplane @xmath54 is a @xmath55-flat . if @xmath4 is a @xmath3-polytope , then @xmath56 is a _ supporting hyperplane _ of @xmath4 if @xmath4 lies entirely on one side of @xmath56 . a _ face _ of @xmath4 is an intersection @xmath57 , where @xmath56 is a supporting hyperplane . if we allow degenerate hyperplanes , then @xmath4 is a face of @xmath4 corresponding to @xmath58 ; @xmath59 is also a face of @xmath4 corresponding to a hyperplane that does not meet @xmath4 . @xmath45-face _ of @xmath4 is a @xmath45-dimensional face of @xmath4 . a @xmath55-face is a _ facet _ , a @xmath47-face is an _ edge _ , and a @xmath46-face is a _ vertex_. one property of @xmath3-polytopes is that any @xmath45-face of @xmath4 is contained in at least @xmath60 facets of @xmath4 . [ ifaceinjface ] ( * ? ? ? * section 3.1 ) if @xmath61 and if @xmath4 is a @xmath3-polytope , each @xmath29-face of @xmath4 is the intersection of the family of @xmath45-faces of @xmath4 containing it . there are at least @xmath62 such faces . we generally use capital letters to denote @xmath3-polytopes . in particular @xmath4 and @xmath63 are @xmath3-polytopes , and @xmath64 and @xmath36 are @xmath3-simplices . capital script letters will generally denote sets of @xmath3-simplices . in particular , @xmath51 is a set of points ( @xmath46-simplices ) . borrowing from @xcite , we let @xmath51 denote a set of @xmath10 distinct points in @xmath1 , where @xmath65 and @xmath66 . assume @xmath51 does not lie entirely in a hyperplane . let @xmath67 . a _ triangulation _ , @xmath68 , of @xmath51 ( or of @xmath4 with the dependence on @xmath51 understood ) is a set of nondegenerate @xmath3-simplices @xmath69 with the following properties . 1 . all vertices of each simplex are members of @xmath51 . the interiors of the simplices are pairwise disjoint . each facet of a simplex is either on the boundary of @xmath4 , or else is a common facet of exactly two simplices . [ partialcondition ] each simplex contains no points of @xmath51 other than its vertices . the union of @xmath69 is @xmath51 and the union of @xmath70 is @xmath4 . since each @xmath3-simplex has volume at least @xmath71 , one immediate consequence is @xmath72 to prove ( [ volumeboundeq ] ) , bey , henk , and wills showed @xmath4 with @xmath2 interior lattice points can be decomposed into at least @xmath73 nondegenerate @xmath3-subpolytopes . any @xmath3-polytope must contain a @xmath3-simplex as a subpolytope , so ( [ volbytriangulation ] ) still holds if @xmath68 is replaced by this decomposition . we will present a slight variation of their theorem and its proof by using triangulations . let @xmath4 be a lattice @xmath3-polytope . lattice triangulation _ @xmath68 of @xmath4 is a triangulation of some set @xmath74 such that @xmath75 . note that if @xmath76 and @xmath77 , we still have @xmath78 . on the other hand , the triangulation @xmath68 of @xmath51 and the triangulation @xmath79 of @xmath80 are necessarily different . fortunately , we can reconstruct the _ vertex set _ of @xmath68 , denoted by @xmath81 , by taking the union of all vertices of all @xmath82 . thus @xmath83 and @xmath84 . let @xmath4 be a lattice polytope and @xmath68 be a lattice triangulation @xmath4 . we say @xmath79 is a _ refinement _ of @xmath68 ( and write @xmath85 ) provided @xmath79 is a lattice triangulation of @xmath4 , @xmath86 , and for all @xmath87 there exists @xmath82 such that @xmath88 . we say that @xmath68 is a _ full lattice triangulation _ if @xmath89 . otherwise we say @xmath68 is a _ partial lattice triangulation_. naturally , triangulations of @xmath49 partition @xmath9 into simplices . on the other hand , there exist partitions of @xmath9 that satisify all but condition ( d ) in the definition of a triangulation . into triangles[triangulationexamples ] ] let @xmath90 and let @xmath91 . note that @xmath4 is a 1-point lattice polygon , and its interior lattice point is the origin . three possible partitions @xmath92 , @xmath93 , and @xmath94 of @xmath4 into triangles are shown in figure [ triangulationexamples ] . note that @xmath92 is a triangulation of @xmath95 and a full lattice triangulation of @xmath4 . however , @xmath92 is not a triangulation of @xmath51 . the middle partition @xmath93 is both a triangulation of @xmath51 and a partial lattice triangulation of @xmath4 ( viewed as a triangulation of @xmath96 ) . the partition @xmath94 on the right is not a triangulation since it fails condition ( d ) . the following theorem , the proof of which can be found in the appendix of @xcite , guarantees the existence of a triangulation of the vertex set of a polytope . [ nonewvert ] ( * ? ? ? * theorem 3.1 ) every convex polytope @xmath4 can be triangulated using no new vertices . that is , there exists a triangulation of @xmath5 . let @xmath4 be a clean , non - empty lattice @xmath3-polytope , and suppose @xmath97 . we construct a _ basic lattice triangulation _ @xmath98 of @xmath4 in the following manner . theorem [ nonewvert ] guarantees that each facet @xmath99 of @xmath4 , as a @xmath55-polytope , has a ( lattice ) triangulation @xmath100 of @xmath101 . let @xmath102 be the set of facets of @xmath4 and let @xmath103 be the set @xmath104 of @xmath55-simplices . finally , let @xmath105 . it is easy to check that @xmath98 is a lattice triangulation of @xmath4 . note that @xmath106 in particular , if @xmath34 , where @xmath107 , and @xmath108 are the facets of @xmath36 , then a basic lattice triangulation @xmath98 is simply the convex hull of the facets of @xmath36 with an interior lattice point @xmath109 of @xmath36 . moreover , @xmath98 is a refinement of the trivial triangulation @xmath110 . the main idea in proving the collinearity property of @xmath111 is to start with the trivial triangulation @xmath110 and obtain a sequence @xmath112 of refinements such that @xmath113 is a full triangulation of @xmath36 and @xmath114 , and then show that noncollinearity forces @xmath115 . the following lemma appears as an assertion in the proof of ( [ volumeboundeq ] ) in @xcite . we prove it here since it is crucial in computing @xmath116 . [ atleastdplusoneminusj ] suppose @xmath4 is a clean , non - empty , lattice @xmath3-polytope . let @xmath68 be a partial triangulation of @xmath4 . if @xmath64 is a @xmath45-face of some @xmath82 , and @xmath117 , then @xmath64 is contained in at least @xmath118 simplices in @xmath68 . let @xmath119 . by lemma [ ifaceinjface ] , @xmath64 is contained in at least @xmath60 facets of @xmath36 . moreover @xmath109 is a vertex of each facet of @xmath36 containing @xmath64 . it follows that these facets are not contained in facets of @xmath4 and are therefore shared by exactly two simplices in @xmath68 . on the other hand any two simplices in @xmath68 intersect in at most one common facet . thus @xmath64 is contained in at least @xmath60 other simplices in @xmath68 . [ refinementtheorem ] let @xmath4 be a clean , non - empty , lattice @xmath3-polytope . let @xmath68 be a partial triangulation of @xmath4 . for any @xmath120 there exists a refinement @xmath79 of @xmath68 such that @xmath121 and @xmath122 . since @xmath68 is a partial triangulation , there exists an interior lattice point @xmath109 of @xmath4 such that @xmath123 . moreover , @xmath109 must lie in the relative interior of some @xmath45-face ( @xmath124 ) , say @xmath64 , of some simplex in @xmath68 . note that @xmath64 is in fact a @xmath45-simplex . let @xmath125 and consider the basic triangulation @xmath98 of @xmath64 into @xmath45-simplices , where @xmath126 clearly @xmath98 is a refinement of @xmath64 into @xmath45-simplices . this refinement of @xmath64 induces a refinement of any @xmath3-simplex containing @xmath64 . more precisely , if @xmath82 contains @xmath64 , then the set @xmath127 is a lattice triangulation of @xmath36 . since @xmath128 and @xmath4 is clean , @xmath129 . thus @xmath130 . by lemma [ atleastdplusoneminusj ] , there are at least @xmath118 simplices in @xmath68 containing @xmath64 as a @xmath45-face . now consider @xmath68 with all such @xmath3-simplices in @xmath68 replaced with their respective induced triangulations and take this to be @xmath79 . by construction , @xmath109 is contained in a simplex @xmath87 if and only if @xmath131 . hence @xmath79 is a refinement of @xmath68 such that @xmath132 and @xmath133 finally , @xmath134 which implies @xmath135 . in the proof above , it is important to note that equality in ( [ atleastd ] ) holds if and only if @xmath136 or @xmath137 . equally important is that if @xmath138 , then @xmath79 is again a partial triangulation . [ simplextriangulation ] if @xmath34 and @xmath110 , then there exists a sequence @xmath139 of refinements of @xmath140 such that @xmath113 is a full triangulation of @xmath36 and @xmath141 . moreover , @xmath142 . for @xmath17 , then the sequence consists only of @xmath140 . if @xmath32 then @xmath110 is a partial triangulation of @xmath36 . let @xmath143 be an arbitrary enumeration of the interior lattice points of @xmath36 . for @xmath144 , we refine @xmath145 into @xmath146 , using theorem [ refinementtheorem ] with @xmath147 . after @xmath2 refinements , @xmath113 is a full triangulation of @xmath36 , and @xmath148 . lastly , ( [ volbytriangulation ] ) implies @xmath149 . [ nojsimplex ] suppose @xmath34 where @xmath150 . let @xmath143 be an arbitrary enumeration of @xmath111 , and let ( [ refinementsequence ] ) be the corresponding refinement sequence guaranteed by corollary [ simplextriangulation ] . if @xmath151 lies in a @xmath45-face of a simplex in @xmath152 , where @xmath153 and @xmath154 , then @xmath155 and @xmath156 . this follows immediately from theorem [ refinementtheorem ] and ( [ sizeofbasictriangulations ] ) , and ( [ atleastd ] ) . in the context of corollary [ simplextriangulation ] , ( [ atleastd ] ) implies that for each @xmath152 in ( [ refinementsequence ] ) , @xmath157 . since equality in ( [ atleastd ] ) holds if and only if @xmath137 or @xmath136 , @xmath158 if and only if for each @xmath29 , @xmath151 lies in the relative interior of an edge in @xmath152 , or in the relative interior of a simplex in @xmath152 . note that this must be true for any ordering of the @xmath159 s . finally , to prove ( [ volumeboundeq ] ) for a general non - empty lattice @xmath3-polytope @xmath4 , we start with a basic triangulation of @xmath4 and obtain a refinement sequence similar to that of corollary [ simplextriangulation ] . [ equivproof ] ( * ? ? ? * theorem 1.2 ) if @xmath4 is a lattice @xmath3-polytope with @xmath160 interior lattice points , then there exists a sequence @xmath161 of lattice triangulations of @xmath4 such that @xmath162 is a basic triangulation of @xmath4 , @xmath113 is a full triangulation of @xmath4 and @xmath114 . moreover , @xmath163 . if @xmath17 , take @xmath162 to be the triangulation guaranteed by theorem [ nonewvert ] . if @xmath107 , then consider an arbitrary enumeration @xmath164 of the the interior points of @xmath4 . let @xmath162 be the basic triangulation @xmath165 of @xmath36 . either @xmath25 and @xmath162 is a full triangulation , or we can apply theorem [ refinementtheorem ] , as in the proof of corollary [ simplextriangulation ] , to obtain a refinement sequence @xmath161 such that @xmath141 . finally , ( [ volbytriangulation ] ) implies @xmath166 . in general , we need not start with a basic triangulation of @xmath4 . it is easy to check that so long as @xmath68 is a partial triangulation such that @xmath167 then we can still refine @xmath68 into a full triangulation with at least @xmath73 simplices by applying theorem [ refinementtheorem ] . this is in fact equivalent to the inductive step in @xcite , with triangulations replaced by decompositions into @xmath3-subpolytopes . the proof of corollary [ equivproof ] is otherwise essentially the same as that in @xcite . in the context of corollary [ nojsimplex ] , if we can show @xmath136 implies @xmath156 as well , then any two interior lattice points of @xmath36 must be collinear with some vertex of @xmath36 . as a base case , we first consider the possible configurations of any @xmath169 , where @xmath0 . suppose @xmath36 has interior lattice points @xmath170 and @xmath171 , and let @xmath172 . the sets @xmath173 are all sets of @xmath168 points not contained in a hyperplane . many properties of such sets are discussed in @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , and @xcite . two relevant and well known results in the theory of convex bodies are carathodory s theorem @xcite and radon s theorem @xcite . if @xmath174 is not contained in a hyperplane , then every @xmath175 can be expressed as @xmath176 the coefficients @xmath177 in carathodory s theorem are the _ barycentric coordinates _ of @xmath178 relative to @xmath9 . if @xmath67 is a @xmath3-simplex , then the barycentric coordinates of @xmath178 relative to the simplex @xmath4 are the numbers @xmath179 satisfying @xmath180}_{1\times ( d+1 ) } \cdot \underbrace { \left [ \begin{array}{cc } v_1 & 1 \\ v_2 & 1 \\ \vdots & \vdots \\ v_{d+1 } & 1 \\ \end{array } \right]}_{(d+1)\times ( d+1 ) } = \underbrace{\left [ \begin{array}{cc } x & 1 \end{array } \right]}_{1\times ( d+1 ) } .\ ] ] for each @xmath29 , the sign of @xmath177 indicates the position of @xmath178 relative to the hyperplane @xmath181 containing the facet of @xmath4 opposite vertex @xmath182 . that is , @xmath183 when @xmath182 and @xmath178 are on the same side of @xmath181 , @xmath184 if @xmath182 and @xmath178 are on opposite sides of @xmath181 , and @xmath185 if @xmath178 lies in @xmath181 . the barycentric coordinates of @xmath178 relative to @xmath36 are all positive if and only if @xmath186 . thus any point @xmath187 ( cf . ( [ pisets ] ) ) , can be described in terms of its barycentric coordinates relative to the simplex @xmath188 . if @xmath189 is a set of @xmath190 points in @xmath1 , then there exist a partition @xmath191 of @xmath189 such that @xmath192 and @xmath193 . the partition in radon s theorem is called a _ radon partition _ of @xmath189 . a radon partition _ in _ @xmath189 is a radon partition of a subset of @xmath189 . let @xmath194 and @xmath195 be radon partitions of @xmath189 . then @xmath194 _ extends _ @xmath196 provided @xmath197 . in @xcite , hare and kennely introduced the notion of a _ primitive radon partition _ , a radon partition that is minimal with respect to extension . an immediate consequence is that if @xmath191 is a radon partition of @xmath189 , then there exists a primitive radon partition in @xmath189 such that @xmath191 extends it . breen proved in @xcite that @xmath198 is a primitive radon partition in @xmath189 if and only if @xmath199 is in general position in @xmath200 . recall that @xmath201 is in general position if no @xmath52 points in this union lie in a @xmath45-flat for all @xmath202 . peterson proved in @xcite that the radon partition of @xmath168 points in general position in @xmath1 is unique , and both breen and peterson showed that if @xmath199 is in general position in @xmath200 , then @xmath203 is a single point . lastly , proskuryakov proved in @xcite that if @xmath49 is a set of @xmath168 points in general position , then two points will lie in the same component of the ( unique ) radon partition of @xmath4 if and only if they are separated by the hyperplane through the remaining @xmath3 points . kosmak also proved this result in @xcite using affine varieties . these properties of radon partitions are equivalent to lawson s first and second theorems from @xcite . * theorem 1 ) let @xmath204 and suppose @xmath51 does not lie entirely in any hyperplane . there is a partition of @xmath51 into three sets @xmath205 , @xmath206 , and @xmath207 , and @xmath208 , satisifying @xmath209 the numbers @xmath177 are uniquely determined by the set @xmath51 . we set @xmath185 if @xmath210 . the sets @xmath205 and @xmath191 are also unique . since the sets @xmath206 and @xmath207 in lawson s first theorem form the unique primitive radon partition in @xmath51 , the point @xmath211 is in the relative interior of @xmath212 and @xmath213 by ( [ uniqueintersection ] ) , ( [ sumisone ] ) , and ( [ aiispositive ] ) . these sets also determine the possible triangulations of @xmath9 . * theorem 2 ) let @xmath214 , and let @xmath70 be the simplex with vertex set @xmath215 . there are at most two distinct triangulations of @xmath67 , namely @xmath216 where the sets @xmath206 and @xmath207 are as defined in lawson s first theorem . the set @xmath217 is a valid triangulation if and only if @xmath218 , where @xmath219 . [ lawsonscorollary ] ( * ? ? ? * corollary 1 ) in the context of building triangulations , an enumeration of all possible configurations of @xmath168 points in @xmath1 , not lying in any hyperplane , is given by all of the possible ways of assigning values to @xmath220 , @xmath221 , and @xmath222 satisfying @xmath223 consider the possible sets of five points in @xmath224 such that the convex hull is a nondegenerate polyhedron with two different triangulations . lawson s second theorem implies @xmath225 . inequalities ( [ aoneisatleastone ] ) , ( [ atwoisatleasttwo ] ) , ( [ azeroisatleastzero ] ) , and ( [ sumofaisdplusone ] ) imply either @xmath226 [ bitetraexample ] suppose @xmath227 is in general position . as shown on the left in figures [ bistellara ] and [ bistellarb ] , let @xmath91 . in the context of lawson s first theorem , @xmath228 , @xmath229 , and @xmath230 . the polyhedron @xmath4 is a `` bipyramid '' with triangulations @xmath231 and @xmath232 by lawson s second theorem . ( left ) and the two simplices of @xmath162 ( right)[bistellara ] ] ( left ) and and the three simplices of @xmath233 ( right ) [ bistellarb ] ] [ pyramidexample ] consider @xmath227 and @xmath67 shown on the left in figure [ pyramid ] . the polyhedron @xmath4 is a nondegenerate pyramid whose base is a planar quadrilateral with vertex set @xmath234 . in the context of lawson s first theorem , @xmath235 , @xmath236 and @xmath237 . lawson s second theorem implies @xmath4 has two triangulations @xmath238 and @xmath239 each triangulation is induced by the triangulation ( in dimension 2 ) of the base . ( left ) and both of its triangulations ( right ) [ pyramid ] ] in both examples [ bitetraexample ] and [ pyramidexample ] , @xmath240 , and @xmath241 can be viewed as two simplices joined at a common facet . moreover , the line segment formed by the vertices not in the common facet intersects that facet at a point . such a configuration is called a bipyramid . in general , a @xmath3-polytope @xmath4 is a _ @xmath3-bipyramid _ if @xmath4 is the convex hull of a line segment @xmath242 and a @xmath55-simplex @xmath64 such that the intersection @xmath243 is a single point contained in @xmath244 . a @xmath47-bipyramid is simply a line segment with a point in its interior ( or the convex hull of three collinear points ) . in example [ bitetraexample ] , @xmath4 itself is a 3-bipyramid , whereas in example [ pyramidexample ] @xmath4 contains a 2-bipyramid ( its base ) . the following result is a direct consequence of lawson s theorems . [ hasbipyramid ] let @xmath4 be a @xmath3-polytope with @xmath168 vertices . if there exists a subset @xmath245 such that @xmath9 is a @xmath45-bipyramid , where @xmath246 , then @xmath5 has exactly triangulations , one with cardinality 2 and another with cardinality @xmath45 . for @xmath150 , under what conditions will @xmath34 contain a bipyramid ? consider the following construction of possible bipyramids from a @xmath3-simplex . let @xmath64 be a @xmath3-simplex , and let @xmath247 . for any point @xmath178 exterior to @xmath64 , let @xmath248 be the barycentric coordinates of @xmath178 relative to @xmath64 . by the definition of a bipyramid , @xmath249 is a bipyramid if and only if exactly one @xmath177 is negative , and the remainder are positive . if we allow for , say , @xmath45 of the @xmath177 to be zero , then @xmath4 will contain a @xmath250-bipyramid . the geometric interpretation is as follows . let @xmath181 be the hyperplane containing the facet of @xmath64 opposite @xmath182 . then @xmath4 is a @xmath3-bipyramid if @xmath178 and @xmath64 are on opposite sides of exactly one @xmath181 , and @xmath178 is on the same side of the remaining hyperplanes as @xmath64 . if @xmath178 is contained in @xmath45 of these hyperplanes , then @xmath4 contains a @xmath250-bipyramid . [ musthavejbipyramid ] suppose @xmath34 , where @xmath150 . let @xmath170 and @xmath171 be any two interior lattice points of @xmath36 , and let @xmath251 be the basic triangulation of @xmath36 with respect to @xmath170 . if @xmath171 lies in the relative interior of a simplex in @xmath165 , say @xmath252 , then there exist numbers @xmath177 such that @xmath253 @xmath254 , @xmath255 , and @xmath256 otherwise . let @xmath257 and @xmath258 be the facet of @xmath36 opposite the vertex @xmath182 . we may assume without loss of generality that @xmath259 and @xmath260 ( i.e. @xmath261 ) . let @xmath242 be the line through @xmath262 and @xmath170 . * case 1 : * if @xmath171 lies on @xmath242 , then @xmath170 is necessarily between @xmath262 and @xmath171 . there exists @xmath263 such that @xmath264 choose @xmath265 , @xmath266 , and @xmath185 for @xmath267 . * case 2 : * suppose @xmath171 does not lie on @xmath242 . let @xmath268 . if @xmath269 , then the line segment @xmath270 , which contains @xmath170 , would be contained in a facet of @xmath36 and contradict the cleanliness of @xmath36 . thus @xmath271 and consequently @xmath178 can not be a lattice point . since @xmath271 if and only if there exists @xmath272 such that @xmath273 the set @xmath274 can be partitioned into @xmath275 according to lawson s first theorem . by lawson s second theorem , @xmath51 has exactly one ( non - lattice ) triangulation @xmath276 since @xmath277 and @xmath278 , @xmath171 must be in some ( non - lattice ) simplex in @xmath68 . without loss of generality , suppose @xmath279 then there exist @xmath280 such that @xmath281 for @xmath282 , @xmath283 since @xmath171 can not lie in any face of @xmath36 , @xmath284 . similarly , @xmath285 since @xmath171 is not in any face of @xmath286 . furthermore , the assumption that @xmath171 does not lie on @xmath242 implies one of the remaining @xmath280 ( @xmath287 must also be positive . since @xmath170 lies between @xmath262 and @xmath178 on @xmath242 , there exists @xmath288 such that @xmath289 it follows that @xmath290 let @xmath291 note that @xmath292\gamma_{d+1 } + \sum_{i=1}^{d } \gamma_i = 1,\ ] ] @xmath293 , and @xmath294 . the remaining @xmath177 are nonnegative , and at least one is positive since at least one of the @xmath280 is positive for @xmath295 . the geometric interpretation of theorem [ musthavejbipyramid ] is that for some simplex @xmath296 in @xmath165 , @xmath297 contains a @xmath45-bipyramid @xmath4 , and @xmath5 includes @xmath170 and @xmath171 . in terms of triangulations , @xmath298 has two triangulations , provided @xmath299 . [ thastwotriangulations ] suppose @xmath34 , where @xmath150 . if @xmath171 lies in the relative interior of a simplex in the basic triangulation @xmath165 of @xmath36 , then the set @xmath300 has two triangulations provided @xmath170 and @xmath171 are not collinear with any @xmath301 . let @xmath302 . without loss of generality , suppose @xmath278 . let @xmath303 be the basic triangulation of @xmath286 with respect to @xmath171 . one triangulation of @xmath304 is the refinement @xmath233 of @xmath162 guaranteed by theorem [ refinementtheorem ] ( with @xmath305 ) , where @xmath306 let @xmath307 be the facet of @xmath286 opposite @xmath182 for @xmath308 , and let @xmath309 be the facet of @xmath286 opposite @xmath170 . theorem [ musthavejbipyramid ] guarantees there exists @xmath310 such that @xmath311 contains a @xmath45-bipyramid . note that both @xmath312 and @xmath313 are simplices in @xmath314 . since @xmath170 and @xmath171 are not collinear with any @xmath301 , @xmath36 is clean , and no three vertices of @xmath36 are collinear , it follows that @xmath299 . corollary [ hasbipyramid ] implies @xmath315 has two triangulations , one of which is contained in @xmath233 . that is , @xmath316 thus @xmath304 has two triangulations . among all possible @xmath3-bipyramids that have two triangulations , 2-bipyramids are unique in that their triangulations have the same cardinality . the existence of lattice 2-bipyramids within lattice @xmath3-simplices has further implications . note that 2-bipyramids are simply convex planar quadrilaterals . [ noparallelogram ] suppose @xmath317 where @xmath150 . let @xmath170 and @xmath171 be any two of the interior points of @xmath36 . let @xmath318 where @xmath319 . if @xmath63 is a planar quadrilateral , then the opposing edges of @xmath63 can not be parallel . without loss of generality , suppose @xmath320 and @xmath321 are edges of @xmath63 . let @xmath322 be the hyperplane containing the facet of @xmath36 opposite @xmath323 , and let @xmath324 be the hyperplane parallel to @xmath322 and containing @xmath323 . since @xmath170 is an interior point of @xmath36 , it must lie between @xmath322 and @xmath324 . if @xmath325 , then @xmath171 must be opposite of @xmath170 relative to both @xmath322 and @xmath324 . this is impossible since @xmath171 would lie outside of @xmath36 as shown in the left figure above . if @xmath326 , edge @xmath327 of @xmath63 must be shorter than edge @xmath328 since @xmath327 is in the interior of @xmath36 . then @xmath329 , or @xmath330 , is a lattice point on the edge @xmath328 , as shown above on the right , which is also impossible as @xmath36 is clean . [ qhaslatticepoint ] suppose @xmath331 is a planar lattice quadrilateral . if the opposing edges of @xmath63 are not parallel , then interior of @xmath63 contains a lattice point @xmath332 for @xmath333 . moreover , @xmath109 lies in the interior of a triangle whose edge set is a subset of the edges and diagonals of @xmath63 . consider the pairs of adjacent edges of @xmath63 . without loss of generality , suppose @xmath334 and @xmath335 are edges of @xmath63 such that the triangle @xmath336 has minimal area . the lines containing each edge of @xmath337 partition the plane into 7 different regions as shown below ( left ) . since @xmath63 is a convex quadrilateral , @xmath338 can only lie in the three unshaded regions . without loss of generality , suppose @xmath338 lies in region @xmath339 as shown above . let @xmath322 be the line through @xmath334 and @xmath324 be the line through @xmath340 such that @xmath341 . similarly , let @xmath342 be the line through @xmath335 and @xmath343 be the line through @xmath344 such that @xmath345 . ( see right - side figure above . ) these four lines divide @xmath339 into four regions . since @xmath63 has no parallel edges , @xmath338 can not lie on any one of these four lines . if @xmath338 lies in region i , then @xmath346 is a triangle of smaller area than @xmath337 ( they both share the same leg @xmath335 , whereas the distance between @xmath338 and @xmath342 is shorter than the distance between @xmath344 and @xmath342 ) . similarly , @xmath338 can not lie in regions iii or iv . thus @xmath338 must lie in region ii . then @xmath347 is a lattice point contained in interior of @xmath348 . we can now prove the collinearity property of @xmath111 , where @xmath34 , @xmath0 , and @xmath349 . [ twocollinear ] suppose @xmath34 , where @xmath0 and @xmath349 . any two interior lattice points of @xmath36 are collinear with a vertex of @xmath36 . the claim is vacuously true for @xmath350 , so suppose @xmath150 . if @xmath351 then let @xmath170 and @xmath171 be the two interior lattice points of @xmath36 . otherwise let @xmath170 and @xmath171 be any two interior lattice points of @xmath36 . suppose @xmath170 and @xmath171 are not collinear with any @xmath301 . that is , suppose there does not exist a lattice 1-bipyramid consisting of @xmath170 , @xmath171 , and any @xmath301 . let @xmath302 , where @xmath165 is the basic triangulation of @xmath36 with respect to @xmath170 . by assumption , @xmath171 must lie in the relative interior of either a simplex in @xmath162 , or a @xmath45-face of a simplex in @xmath162 , where @xmath154 . * case 1 : * if @xmath171 lies in the relative interior of a @xmath45-face of a simplex in @xmath162 , where @xmath154 , then corollary [ nojsimplex ] implies @xmath156 , a contradiction . * case 2 : * if @xmath171 lies in the relative interior of a simplex in @xmath162 , then we may assume without loss of generality that @xmath352 . let @xmath303 be the basic triangulation of @xmath286 with respect to @xmath171 . the refinement @xmath233 of @xmath162 obtained by applying theorem [ refinementtheorem ] with @xmath305 is @xmath306 theorem [ musthavejbipyramid ] implies @xmath36 contains a lattice @xmath45-bipyramid @xmath4 , where @xmath353 and @xmath354 . if @xmath355 , then @xmath356 is convex planar lattice quadrilateral , where @xmath357 . lemmas [ noparallelogram ] and [ qhaslatticepoint ] imply @xmath4 contains a lattice point @xmath358 . this is a contradiction if @xmath351 . if @xmath31 , then the second part of lemma [ qhaslatticepoint ] implies @xmath359 lies in the relative interior of some triangle @xmath360 , where @xmath361 since @xmath0 , and each @xmath360 a 2-face of simplices in @xmath233 , @xmath156 by corollary [ nojsimplex ] . this contradicts the mimal volume property of @xmath36 . on the other hand , if @xmath362 , then @xmath304 has two possible triangulations @xmath233 and @xmath363 where @xmath364 subsequent applications of theorem [ refinementtheorem ] , starting with @xmath363 , and ( [ volbytriangulation ] ) imply @xmath365 this is again a contradiction . thus any two points in @xmath111 must be collinear with some @xmath301 . [ allcollinear ] suppose @xmath34 , where @xmath0 and @xmath349 . the points @xmath111 are collinear with some @xmath301 . the claim is vacuously true for @xmath366 . theorem [ twocollinear ] implies the claim also holds for @xmath351 . suppose @xmath367 . it suffices to show the claim holds for any three points in @xmath111 . let @xmath170 , @xmath171 , and @xmath359 be any three points in @xmath111 . theorem [ twocollinear ] implies there exists @xmath301 such that @xmath170 , @xmath171 , and @xmath13 are collinear similarly , there exists @xmath368 such that @xmath170 , @xmath359 , and @xmath323 are collinear . we need to show @xmath369 . suppose @xmath370 . let @xmath242 be the line through @xmath170 , @xmath171 and @xmath13 , and let @xmath371 be the line through @xmath170 , @xmath359 , and @xmath323 . without loss of generality , suppose @xmath171 is between @xmath170 and @xmath13 on @xmath242 . * case 1 : * if @xmath359 lies between @xmath170 and @xmath323 on @xmath371 , then @xmath372 is a convex planar quadrilateral contained within @xmath373 . lemmas [ noparallelogram ] and [ qhaslatticepoint ] imply @xmath16 contains a lattice point @xmath374 . since @xmath375 and @xmath0 , corollary [ nojsimplex ] implies @xmath376 . * case 2 : * if @xmath170 lies between @xmath359 and @xmath323 on @xmath371 , then @xmath377 . again , since @xmath0 , corollary [ nojsimplex ] implies @xmath378 . thus @xmath379 , and all points in @xmath380 and @xmath13 are collinear . [ evenlyspaced ] suppose @xmath34 and @xmath0 . let @xmath242 be the line through @xmath111 . the consecutive points on @xmath242 are evenly spaced . let @xmath381 and let @xmath382 be the lattice point closest to @xmath13 . since any line can be described as a linear combination of two points , any @xmath383 can be expressed as @xmath384 where @xmath385 . it suffices to show that for any @xmath386 , the @xmath387 in ( [ equationofline ] ) is an integer and @xmath388 . if @xmath389 , then @xmath390 , which is impossible . if @xmath391 then @xmath392 . let @xmath393 $ ] and @xmath394 denote the integer and fractional parts of @xmath387 , respectively . if @xmath395 , then @xmath396 ( w - v ) = v+\ { \alpha \}(w - v ) \in \z^d\ ] ] lies between @xmath13 and @xmath109 on @xmath242 , which contradicts our assumption that @xmath109 is closest to @xmath13 on @xmath242 . hence @xmath397 . finally , if @xmath398 , then @xmath36 has more than @xmath2 interior points , which is aslo impossible . [ uniquetriangulation ] if @xmath34 , @xmath349 , and @xmath399 , then @xmath400 has a unique triangulation . let @xmath401 . without loss of generality , suppose @xmath262 is collinear with @xmath111 . let @xmath170 be the point in @xmath111 closest to @xmath262 and @xmath402 let @xmath403 . for @xmath144 and @xmath404 , let @xmath405 it is easy to check that @xmath406 is a triangulation of @xmath400 . in fact , this triangulation corresponds to the refinement sequence in corollary [ simplextriangulation ] . let @xmath79 be another full triangulation of @xmath36 . consider any @xmath407 . since @xmath79 is a full triangulation , @xmath64 is necessarily empty and clean . these two conditions force @xmath408 * case 1 : * if @xmath409 , then there exist @xmath301 and @xmath410 such that @xmath411 since @xmath64 is empty , it follows that @xmath412 , @xmath413 , and @xmath414 . * case 2 : * if @xmath415 , then there exist @xmath357 and @xmath416 such that @xmath417 without loss of generality , suppose @xmath109 is closer to @xmath418 than @xmath419 is to @xmath418 . since @xmath64 is clean , @xmath420 for some @xmath421 . moreover , @xmath418 ( @xmath422 ) is a vertex of @xmath64 if and only if @xmath423 in ( [ ithlevel ] ) since @xmath64 must also be empty . thus @xmath424 for some appropriate @xmath45 , and @xmath425 . consequently , @xmath426 . a _ unimodular transformation _ @xmath427 is an affine map of the form @xmath428 where @xmath429 , @xmath430 is a @xmath431 matrix with entries in @xmath432 , and @xmath433 . translation _ is a unimodular transformation in which @xmath434 , where @xmath435 is the @xmath431 identity matrix . if @xmath433 , @xmath430 is invertible , then @xmath436 is unimodular , and @xmath437 is one - to - one . thus @xmath4 is a lattice @xmath3-polytope if and only if @xmath438 is a lattice @xmath3-polytope with the same number of vertices . using barycentric coordinates , it is easy to check that @xmath439 is an interior point of @xmath438 if and only if @xmath109 is an interior point of @xmath4 . similarly , @xmath439 is a boundary point of @xmath438 if and only if @xmath109 is a boundary point of @xmath4 ( cf . we say @xmath440 and @xmath441 are _ equivalent _ lattice @xmath3-polytopes ( and write @xmath442 ) if there exists a unimodular transformation @xmath437 such that @xmath443 . [ unitsimplex ] let @xmath36 be a @xmath3-simplex with @xmath444 . then @xmath36 is equivalent to @xmath445 , the @xmath3-simplex whose vertex set consists of the origin , @xmath28 for @xmath446 , and the point @xmath447 . let @xmath448 be an enumeration of the vertices of @xmath36 . if necessary , translate @xmath36 so that one of its vertices is the origin . without loss of generality , suppose @xmath449 . let @xmath450 note that @xmath4 is a parallelepiped containing @xmath36 , and @xmath451 , where @xmath452 is a @xmath431 matrix whose entry @xmath453 is the @xmath45-th coordinate of @xmath182 . since @xmath454 @xmath455 and @xmath452 is invertible . thus , there exists a unique , unimodular , @xmath431 matrix @xmath430 such that @xmath456 . thus @xmath457 defined by @xmath458 is a unimodular transformation such that and @xmath459 for @xmath308 and @xmath460 . let @xmath461 be the unimodular transformation given by @xmath462 where @xmath463.\ ] ] it is easy to check that @xmath464 for @xmath446 , @xmath465 , and @xmath466 . then the composition @xmath467 , where @xmath468 , is also a unimodular transformation , and @xmath469 . let @xmath470 denote the @xmath3-simplex whose vertex set consists of the origin , the unit points @xmath471 , and the point @xmath472 . what can we say about @xmath470 ? if @xmath473 , then we can apply @xmath427 , where @xmath474.\ ] ] note that @xmath437 is simpy a reflection in the @xmath3-th coordinate . by reflection , if necessary , we can take @xmath475 to be nonnegative . moreover , @xmath475 is determined by the volume of @xmath470 since @xmath476 = for the remaining @xmath477 , consider the integers @xmath478 such that such that @xmath479 and the transformation @xmath480 , where @xmath481.\ ] ] all the vertices of @xmath470 excluding @xmath482 are fixed under @xmath483 . on the other hand , @xmath483 sends @xmath484 to @xmath485 where @xmath486 and @xmath487 for @xmath446 . since @xmath470 is not contained in any hyperplane , @xmath488 and @xmath489 . thus the class of all @xmath490 is represented by @xmath491 where @xmath492 lastly , let @xmath493 and let @xmath494 be the barycentric coordinates of @xmath109 relative to @xmath495 , where @xmath496 the @xmath3-th coordinate of @xmath109 is @xmath497 , and for @xmath446 , the @xmath29-th coordinate is @xmath498 we can conclude that every point in @xmath499 is completely determined by its @xmath3-th coordinate . equivalently , no two interior lattice points can have the same @xmath3-th coordinate . [ hasgeneralform ] let @xmath0 and @xmath107 . if @xmath34 , and @xmath349 , then @xmath500 where @xmath501 . let @xmath502 , @xmath262 , @xmath503 , @xmath504 , and @xmath68 be as in the proof of corollary [ uniquetriangulation ] . for @xmath505 , lemma [ unitsimplex ] implies there exists a unimodular transformation @xmath437 such that @xmath460 , @xmath506 , and @xmath507 for @xmath508 . let @xmath509 we may assume @xmath510 by ( [ aiarepositive ] ) . since @xmath349 and unimodular transformations preserve volume , ( [ asubd ] ) implies @xmath511 . unimodular transformations also preserve interior lattice points , so @xmath512 , and @xmath513 is collinear with @xmath514 . corollary [ evenlyspaced ] implies @xmath515 . if @xmath516 are the barycentric coordinates of @xmath517 , where @xmath518 then ( [ ithcoordinate ] ) implies @xmath519 note that @xmath520 , and @xmath521 since @xmath522 @xmath523 for all @xmath29 , which implies @xmath524 . the final step is to show that @xmath40 . consider the interior point @xmath525 and let @xmath526 be the barycentric coordinates of @xmath527 . clearly @xmath528 . by ( [ ithcoordinate ] ) , we have @xmath529 since @xmath530 , it follows that @xmath531 using elementary algebraic manipulation , we obtain @xmath532 which implies @xmath533 finally , @xmath534 and ( [ remainingaj ] ) imply @xmath40 for @xmath404 . if @xmath34 , then @xmath349 if and only if @xmath535 . we first show that @xmath44 . using determinants , @xmath536 where @xmath537 = \underbrace{\left [ \begin{array}{c|ccccc } -1 \\ -1 & & { \mathbf{i}}_{d-1 } \\ -1 \\ \hline -k-1 & -k & \dotsm & -k \end{array } \right]}_{d\times d}.\ ] ] let @xmath538 be the matrix obtained by adding @xmath2 times the sum of the first @xmath539 rows of @xmath452 to row @xmath3 of @xmath452 . that is , @xmath540.\ ] ] it is easy to check that @xmath541 . if @xmath535 , then @xmath542 . on the other hand , if @xmath349 , then @xmath535 by theorem [ hasgeneralform ] provided @xmath33 is clean . since @xmath44 and any lattice triangulation of @xmath543 contains at least @xmath73 simplices , ( [ sizeofbasictriangulations ] ) and corollary [ simplextriangulation ] imply @xmath33 can not have any lattice points on its boundary except for its vertices . let @xmath427 be the translation @xmath544 and let @xmath461 be the transformation @xmath545 where @xmath546}_{d\times d}.\ ] ] that is , if @xmath547 is the entry in the row @xmath29 and column @xmath45 of @xmath430 , then @xmath548 consider the image of @xmath549 under the composition @xmath467 . for @xmath550 , the @xmath45-th coordinate of @xmath13 under @xmath467 is simply the dot product of @xmath551 with the @xmath45-th column of @xmath430 . it is easy to check that @xmath552 the composition @xmath467 is in fact a unimodular transformation which maps @xmath33 to @xmath470 , where the @xmath477 are as in theorem [ hasgeneralform ] . the collinearity property of the interior lattice points does not hold for all clean triangles in @xmath553 . pick s theorem states that if @xmath4 is a lattice polygon , then @xmath554 where @xmath555 and @xmath556 . for clean triangles , ( [ picks ] ) reduces to @xmath557 thus all clean lattice triangles satisfy the minimal volume condition . in @xcite , reznick proved that any @xmath558 is equivalent to some @xmath559 , where @xmath560 . using this representation for clean lattice triangles , reznick then showed that the number of equivalence classes of clean @xmath2-point lattice triangles increases with @xmath2 . thus there exist clean lattice triangles whose interior points are not collinear . consider the triangle @xmath561 in @xmath553 with vertex set @xmath562 , where @xmath563 and @xmath564 . it is easy to check that @xmath561 is clean . we first compute the number of interior points of @xmath561 . using determinants , @xmath565 pick s theorem implies @xmath566 , the interior points of @xmath561 are all on the line @xmath567 . however , for @xmath568 , this is not the case as they are covered by the lines @xmath569 ( @xmath570 ) . since these lines are parallel , they can not be contained within any one line . incidentally , this collection of counterexamples is related to the following summation identity . if @xmath571 and @xmath572 , then @xmath573 this is a simple counting argument using ( [ pointsindeltapq ] ) , the fact that the lines @xmath569 ( @xmath574 ) cover @xmath575 , and the fact that there are @xmath576 lattice points on the line @xmath569 and @xmath577 . the counterexample above can be generalized to a triangle with vertex set @xmath578 where @xmath579 , @xmath580 , and @xmath581 are positive integers and @xmath582 . the corresponding identity below is similar to ( [ countingidentityone ] ) . if @xmath583 and @xmath584 , then @xmath585 pick s theorem itself is a counterexample since only in @xmath553 is the volume of a lattice polytope @xmath4 an equality in terms of the number of lattice points on @xmath15 and in @xmath16 . tetrahedra @xcite have vertices @xmath586 where @xmath587 . these tetrahedra are clearly clean and empty , yet their volume increases with @xmath10 . reeve concluded that the volume of lattice of a polyhedron @xmath4 can not be expressed solely in terms of @xmath588 and @xmath589 . however , reeve was able to find and analogue to pick s theorem by considering sublattices . let @xmath590 denote the lattice consisting of all points in @xmath1 whose coordinates are multiples of @xmath591 . note that @xmath592 . for a lattice @xmath3-polytope @xmath4 , let @xmath593 reeve showed in @xcite that @xmath594,\ ] ] where @xmath595 and @xmath596 is the euler characteristic . reeve also conjectured a similar formula for @xmath597 . soon after , macdonald @xcite not only confirmed reeve s conjecture , but generalized reeve s formulas to @xmath598 . \end{split}\ ] ] in 1996 , kolodzieczyk @xcite showed that ( [ macdonaldvolume ] ) has the following alternate form in which only the numbers @xmath599 of interior lattice points in the sublattice @xmath600 are required . @xmath601.\ ] ] for a @xmath3-polytope @xmath4 all of whose vertices join @xmath3 edges , @xmath602 and ( [ kkequation ] ) assumes the form @xmath603 recently , kolodzieczyk showed in @xcite that pick - type formulas exist even when considering only the points in @xmath604 whose coordinates are odd multiples of @xmath591 . though no formula for the volume of a lattice polytope in terms of the numer of its boundary and interior lattice points exist , there does exist a pick - type inequality with sharp bounds on the volume of a polyhedron . any polyhedron in @xmath224 that does not intersect itself has an associated planar graph . we can use elementary graph theory and euler s formula to obtain a lower bound on the volume of lattice polyhedra . [ picksinequality ] let @xmath605 be a convex lattice polyhedron with @xmath606 lattice points on the boundary and @xmath107 interior lattice points . then @xmath607 by ( [ volbytriangulation ] ) , it suffices to show @xmath4 has a lattice triangulation @xmath68 satisfying @xmath608 any refinement sequence starting with any basic triangulation of @xmath4 will suffice . consider the graph @xmath609 whose vertex set @xmath610 is @xmath611 and whose edge set consists of the edges of @xmath15 . since @xmath4 is convex , @xmath609 is planar . ( just embed @xmath609 into @xmath612 and then map onto @xmath553 . ) thus every triangulation of @xmath609 has @xmath613 faces , implying @xmath15 can be triangulated into @xmath614 triangles . since @xmath107 , @xmath4 can be triangulated into @xmath614 subtetrahedra using any interior lattice point . the remaining interior points refine this triangulation , via theorem [ refinementtheorem ] , into a full triangulation having at least @xmath615 subtetrahedra . this bound is sharp ( consider any clean , non - empty lattice tetrahedron ) . a similar result is proved ( independently ) in ( * ? ? ? * lemma 3.5.5 ) by de loera , rambau , and santos leal . while their proof does not assume that @xmath4 is nonempty , it does assume that @xmath616 is in general position . they also showed that any polyhedron with @xmath10 vertices has a triangulation consisting of at most @xmath617 tetrahedra . what can we say about an upper bound for the volume of polyhedra ? hensley proved in @xcite that the volume of any lattice @xmath3-polytope is bounded above by a function in terms of @xmath3 and the number of interior lattice points . ziegler improved hensley s results and showed in @xcite that the volume of a @xmath2-point lattice @xmath3-polytope @xmath4 satisifies @xmath618^{d\cdot 2^{d+1}}.\ ] ] in @xmath224 , a few simple computations seem to indicate that the upper bound on the volume of clean tetrahedra is linear in @xmath2 . moreover , maximal - volume tetrahedra also appear to have collinear interior lattice points . unlike the minimal - volume tetrahedra , however , the line containing the interior points does not pass through any vertex . based on these computations , we make the following conjectures . [ firstconjecture ] if @xmath619 , then @xmath620 . [ secondconjecture ] if @xmath619 and @xmath621 , then @xmath622 . if @xmath623 , then @xmath624 is a set of collinear points that does not pass through any vertex of @xmath36 . these conjectures are true for @xmath25 , as proved independently by kasprzyk @xcite and reznick @xcite . it may be possible to gain some new insight on the relationship between lattice points of a lattice @xmath3-polytope and its volume using ehrhart theory . for example , it is well known that every convex lattice @xmath3-polytope is associated with an ehrhart polynomial of degree @xmath3 . moreover , the leading coefficient of the associated polynomial is precisely the volume of the @xmath3-polytope ( cf . @xcite , @xcite ) . i would like to thank my thesis advisor bruce reznick for having introduced me to lattice polytopes . i am also grateful to phil griffith , former director of graduate studies of the math department at the university of illinois in champaign - urbana , who was directly responsible for my opportunity to work with bruce reznick . i also would like to thank matthias beck and sinai robins , the authors of @xcite , for their enlightening course on lattice point enumeration in banff , canada .

we investigate massive star formation in turbulent , magnetized , parsec - scale clumps of molecular clouds including protostellar outflow feedback using three dimensional numerical simulations of effective resolution . the calculations are carried out using a block structured adaptive mesh refinement code that solves the ideal mhd equations including self - gravity and implements accreting sink particles . we find that , in the absence of regulation by magnetic fields and outflow feedback , massive stars form readily in a turbulent , moderately condensed clump of ( containing initial jeans masses ) , along with a cluster of hundreds of lower mass stars . the massive stars are fed at high rates by ( 1 ) transient dense filaments produced by large - scale turbulent compression at early times , and ( 2 ) by the clump - wide global collapse resulting from turbulence decay at late times . in both cases , the bulk of the massive star s mass is supplied from outside a 0.1 pc - sized core " that surrounds the star . in our simulation , the massive star is clump - fed rather than core - fed . the need for large - scale feeding makes the massive star formation prone to regulation by outflow feedback , which directly opposes the feeding processes . the outflows reduce the mass accretion rates onto the massive stars by breaking up the dense filaments that feed the massive star formation at early times , and by collectively slowing down the global collapse that fuel the massive star formation at late times . the latter is aided by a moderate magnetic field of strength in the observed range ( corresponding to a dimensionless clump mass - to - flux ratio a few ) ; the field allows the outflow momenta to be deposited more efficiently inside the clump . we conclude that the massive star formation in our simulated turbulent , magnetized , parsec - scale clump is outflow - regulated and clump - fed ( orcf for short ) . an important implication is that the formation of low - mass stars in a dense clump can affect the formation of massive stars in the same clump , through their outflow feedback on the clump dynamics . in a companion paper , we discuss the properties of the lower mass cluster members formed along with the massive stars , including their mass distribution and spatial clustering .

0908.4129

the most important factor that determines whether a massive star can form or not is its mass accretion rate ( larson & starrfield 1971 ; for recent reviews , see mckee & ostriker 2007 ; zinnecker & yorke 2007 ) . it follows from the fact that the outward radiative force on the accretion flow onto a forming massive star increases rapidly with the stellar mass . if the mass accretion rate is too low , there would be insufficient ram pressure to overcome the radiation pressure near the dust sublimation radius , and the infall would be choked . in the spherical model of wolfire & cassinelli ( 1987 ) , the required minimum accretion rate ranges from @xmath4 to @xmath5 for stars in the mass range @xmath6 , although the requirement can be relaxed somewhat if the accretion flow is flattened , by , e.g. , magnetic support ( nakano 1989 ) or rotation ( jijina & adams 1996 ) . in addition to overcoming the radiation pressure , a high accretion rate can also greatly modify the internal structure and appearance of a growing massive star ( e.g. , hosokawa & omukai 2009 ) . the associated high accretion luminosity may help heat up the region surrounding the protostar , perhaps suppressing the potential fragmentation that may adversely affect massive star formation at the early stages ( krumholz & mckee 2008 ) . a high mass accretion rate may not be difficult to achieve in principle , given that most , if not all , massive stars are thought to form in dense , massive , cluster - forming clumps ( mckee & ostriker 2007 ; zinnecker & yorke 2007 ) . for example , a parsec - sized dense clump of @xmath7 would have an average density of @xmath8 g @xmath9 , corresponding to a global free fall time @xmath10 myr . the ratio of the clump mass and free - fall time yields a characteristic accretion rate @xmath11 ( e.g. , cesaroni 2005 ) which , if channeled to a single object , would have a sufficiently large ram pressure to overcome the radiation pressure from even the most massive stars . in practice , the mass accretion rate onto any individual star will be reduced relative to the characteristic free - fall rate by several factors , such as the turbulence , magnetic fields , protostellar outflows , and radiative feedback . these factors are included in our simulations , except the last one . supersonic turbulence is observed ubiquitously in dense clumps of star cluster and massive star formation ( e.g. , garay 2005 ) . the turbulence is expected to reduce the mass accretion rate onto any individual star relative to @xmath12 in at least two ways . first , it resists the global collapse , thereby reducing the total amount of gas that collapses into stars per unit time . second , it fragments the clump material into many centers of collapse , creating a cluster of stars each accreting at a fraction of the total rate . these effects of the turbulence are difficult to quantify analytically ; numerical simulations are required ( mac low & klessen 2004 ) . for example , using sph simulation with sink particles , bonnell , bate & vine ( 2003 ) found that a dense turbulent clump of @xmath7 in mass and 0.5 pc in radius ( corresponding to @xmath13 ) produced a cluster of about 400 stars . in their simulation , the most massive star ( of 27 ) formed from an average accretion rate of order @xmath4 , much smaller than the global free - fall rate . in this paper , we will explore the effects of the turbulence on the stellar mass accretion further with a grid - based code , enzo , including adaptive mesh refinement ( amr ) and sink particles . the grid - based code also enables us to quantify , in addition , the effects of magnetic fields , which are generally difficult to treat in sph ( see , however , price & bate 2008 ) . there is ample observational evidence for magnetic fields in regions of massive star formation . in the nearest region of active massive star formation , omc1 , the well - ordered polarization vectors of submillimeter dust continuum leave little doubt that the ( slightly pinched ) magnetic field is dynamically important ( e.g. , vaillancourt et al . a cn zeeman measurement yields a line - of - sight field strength of 0.36 mg , corresponding to a magnetic energy density that is higher than the turbulent energy density and a mass - to - flux ratio ( before geometric correction ) that is 4.5 times higher than the critical value @xmath14 ( falgarone et al . the mass - to - flux ratio is close to the median value inferred by falgarone et al . for a sample of a dozen or so high - mass star forming regions where the cn zeeman measurements are made . their best estimate for the mean dimensionless mass - to - flux ratio ( in units of the critical value ) is @xmath15 after geometric corrections , although it can be uncertain by a factor of 2 in either direction . a goal of our calculations is to determine how a magnetic field of this magnitude affects the stellar mass accretion rate , using an mhd version of the enzo code ( wang & abel 2009 ) . the third element that we consider in this paper are protostellar outflows , which are routinely observed around forming stars of both low and high masses ( see the review by richer et al . their effects are expected to be particularly important in the dense clumps of active cluster formation , where many stars are formed close together in both space and time ( li & nakamura 2006 ; norman & silk 1980 ) . a case in point is the dense clump associated with the reflection nebula ngc 1333 in the perseus molecular cloud , which is sculpted by dozens of outflows detectable in co , optical forbidden lines and h@xmath16 emission ( sandell & knee 2001 ; walawender et al . 2008 ; maret et al . the outflows appear to inject enough momentum into the dense clump to replenish the turbulence dissipated in this region ( see , however , maury et al . 2009 for a discussion of ngc 2264 , where the current generation of active outflows appear incapable of supporting the cluster - forming clump ) . it is possible that a good fraction , perhaps the majority , of the cluster members are formed during the time when the clump turbulence is driven by protostellar outflows ( nakamura & li 2007 , matzner 2007 , carroll et al . 2008 , swift & welch 2008 ) . we will improve on the previous outflow feedback simulations of nakamura & li ( 2007 ) by including sink particles , making the outflow injection more continuous , and dramatically increasing the numerical resolution . these improvements enable us to evaluate the effects of the outflows on the accretion rates onto individual stars , especially massive stars , which are the focus of this paper . we will describe the properties of the lower mass cluster members formed in our simulations , including their mass distribution and spatial clustering , in a companion paper . the effects of radiative feedback , ignored in our current calculations , are discussed in [ limitation ] . the rest of the paper is organized as follows . in [ method ] , we describe the numerical method and simulation setup . these are followed by two result sections , focusing first on the global star formation history and clump dynamics ( [ global ] ) and then on the massive stars ( [ massivestars ] ) . we find that the massive stars in our simulations are formed through neither the collapse of pre - existing 0.1 pc - sized turbulent cores nor competitive accretion ; their formation is controlled to a large extent by the clump dynamics , which are found to be regulated strongly by the collective outflow feedback from all accreting stars and , to a lesser extent , by a moderate magnetic field . in [ discussion ] , we discuss this new scenario of outflow - regulated clump - fed " ( orcf for short ) massive star formation and contrast it with the existing scenarios . the last section contains a brief summary . our simulations are performed using the parallel amr code , enzo , with a newly added solver for unsplit conservative hydrodynamics and magnetohydrodynamics ( wang & abel 2009 ) . as usual , at the heart of the mhd solver lies the enforcement of the divergence - free condition @xmath17 . there are three classes of method for this purpose : constraint transport ( ct , evans & hawley 1988 ) , projection method and hyperbolic clean ( see , e.g. , tth 2000 for a review ) . we have evaluated all these methods and found that divergence cleaning is most suitable for the problem at hand , which involves rather extreme variation in physical quantities ( including the field strength ) due to both gravitational collapse and protostellar outflows . the fast and robust method of hyperbolic clean of dedner et al . ( 2002 ) is adopted . it enables us to follow the formation of hundreds of stars over several dynamical times . we monitor the value of @xmath18 to ensure that it is small in all of our mhd simulations . truelove et al . ( 1997 ) pointed out that in order to avoid artificial fragmentation , one need to use at least four cells to resolve the jeans length . for a given cell size at the finest level of refinement , the condition translates into a maximum density , which we will call the `` jeans density '' , above which artificial fragmentation may happen . since it is difficult in large - scale simulations such as ours to resolve the jeans length all the way to the stellar density , sink particles are used to handle the gravitational collapse beyond the jeans density . this approach has previously been successfully used in both sph ( e.g. , bate et al . 1995 ) and grid - based simulations of star formation ( e.g. , krumholz et al . 2004 ; offner et al . 2009 ) . our procedure for implementing the sink particles is as follows . when a cell violates the truelove criterion on the highest refinement level , a sink particle is inserted at the center of that cell . the initial mass of the sink particle is calculated such that after subtracting the sink mass , the cell will be at the jeans density . the initial velocity of the sink particle is calculated using momentum conservation . after creation , the sink particle accretes gas from its host cell according to a prescription inspired by the bondi - hoyle ( bh ) formula ( ruffert 1994 ) , @xmath19 where @xmath20 is the bondi radius , @xmath21 and @xmath22 are the gas density , velocity and sound speed of the ( uniform ) medium far from the point mass . even though equation ( [ bh ] ) is not strictly applicable to the highly turbulent medium under investigation , we will follow krumholz et al . ( 2004 ) and use it as a guide to construct a prescription for sink particle accretion . since the cloud material is assumed isothermal , it is natural to set @xmath23 to the isothermal sound speed @xmath24 . for @xmath25 , we adopt the simple prescription @xmath26 , where @xmath27 is the flow velocity in the cell and @xmath28 the velocity of the sink particle . we compute the bondi radius in equation ( [ bh ] ) using @xmath29 where @xmath30 is the sink mass . if @xmath20 is smaller than the cell length @xmath31 , the cell density @xmath32 is used for @xmath33 . otherwise an extrapolation assuming an @xmath34 density profile is used . in other words , we set @xmath35 $ ] , where @xmath36 au is the size of our finest cells . the amount of gas accreted during a single time step @xmath37 is then @xmath38 . the amount may or may not match the true value exactly at any given time , given the crudeness of the above prescription . this deficiency is corrected in a second step , through sink particle merging . sink particle merging is needed not only to ensure the correctness of the mass accretion algorithm , but also to save computation time when many sink particles are created . it is a tricky problem , because we want to eliminate , on the one hand , the artificial particles created by the mismatch between the true accretion rate and that given by equation ( [ bh ] ) , and to preserve , on the other hand , the legitimate stellar seeds . it requires sub - grid information that is not available in the simulation . at this early stage in the development of the sink particle technique , our design principle of the merging recipe is to make it as simple as possible and with as few parameters as possible . we have experimented with a number of recipes , and settled on one with two parameters : a merging mass @xmath39 and a merging distance @xmath40 . using @xmath39 we divide the sink particles into a `` small '' and a `` big '' group . at every time step , the merging is done in two steps . first , for each small particle , we search within the merging distance for its nearest big particle . if the search is successful , we merge this particle to the found big particle . second , for all small particles that remain after the first step , we group them use a friend - of - friend algorithm with the merging distance @xmath40 as the linking length ( davis et al . 1985 ) . in this work , we use @xmath41 and @xmath42 , which is about @xmath43 au . experimentation shows that increasing @xmath39 to @xmath44 or @xmath40 to @xmath45 does not change the star formation rate or the stellar mass distribution significantly . lowering the merging distance to @xmath46 has little effect on the total star formation rate , but can change the mass of a star by up to @xmath47 , probably because the flow pattern within a few cells of a sink particle is not well resolved . the flow so close to a forming star may be strongly affected by radiative heating ( offner et al . 2009 ; bate 2009 ) , which is not included in our simulations . for these reasons , we believe that the properties of circumstellar disks and binaries may not be reliably captured by our simulation ; they are not discussed further in the paper . with the inclusion of sink particle merging , the crude prescription for sink particle accretion based on the bh formula ( [ bh ] ) does not appear to pose any serious problem . we have experimented with changing the parameters in the formula and did not see any significant difference in either the stellar mass accretion rate or the final stellar mass distribution . as stressed by krumholz et al . ( 2004 ) , the reason is probably that , when the bh formula underestimates the true accretion rate , the sink particle creation routine would transform the un - accreted gas into small sink particles , which would merge quickly with the original particle and restore the correct rate . on the other hand , when the bh formula overestimates the true rate , which happens rarely and only when the sink particle is massive , the accretion flow is typically supersonic near the particle , and the over - accretion does not affect the flow further away . to avoid numerical instabilities , we restrict the maximum amount of the gas accreted in a single time - step to be less than @xmath48 of the cell mass . finally , the gravity of the sink particles is calculated using the standard particle - mesh method , and their positions and velocities are updated using the leapfrog method . protostellar outflows are a crucial element of our model . in large - scale global simulations such as ours , it is unfortunately impossible to follow their production from first principles . an observationally motivated prescription is used instead . we assume that the protostellar outflow momentum injection rate is proportional to the stellar mass accretion rate , with a mass dependent proportionality constant @xmath49 where @xmath50 km / s and @xmath51 . the mass dependence is chosen to reflect the fact that the outflows from high mass stars tend to be more powerful even after correcting for their larger masses ( richer et al . 2000 , d. shepherd , private communication ) . we represent the outflow by injecting a momentum @xmath52 into the surrounding gas every time the sink particle has gained a mass increment @xmath53 . ideally , one would like to make @xmath53 as small as possible , to render the injection continuous . however , we find continuous injection is too time consuming , because of the small courant timestep associated with the fast outflow speed and small grid size near the sink particles . furthermore , if @xmath53 is much less than the mass of an injecting cell , the injected momentum would have little dynamical effect on the cell . these considerations led us to adopt @xmath54 . we have experimented with @xmath55 , and it did not change the results significantly . to avoid too high an outflow speed when a surrounding cell happens to have a low density , we take @xmath56 out of the sink particle , and divide it evenly between the injecting cells . for simplicity , we inject the momentum in a bipolar jet , with the jet direction parallel ( and anti - parallel ) to the local magnetic field direction of the host cell when the jet is first injected ( the jet direction is kept fixed in time ) and a jet width of five finest cells . we consider a dense clump of molecular cloud that has an initial density profile with a uniform central region , @xmath57 where @xmath58 and @xmath59 is the length of the simulation box . the central density @xmath60 g @xmath9 and the box size is @xmath61 pc . this gives a cloud mass within a sphere of 1 pc in radius of @xmath62 . since a square box with @xmath61 pc is used , the total mass is somewhat larger , with @xmath63 . our choice of cloud density , mass and mass distribution is motivated by observations of infrared dark clouds ( e.g. , butler & tan 2009 ) , which are thought to be future sites of cluster and massive star formation ; they tend to be centrally condensed even before the onset of massive star formation . the equation of state is isothermal , with a sound speed @xmath64 km / s , which corresponds to a temperature of @xmath65 k for a mean molecular weight @xmath66 . a cloud with a density profile given by equation ( [ denprofile ] ) has a gravitational potential energy @xmath67 , which yields a virial parameter @xmath68 , where @xmath69 is the 3d rms velocity . in terms of the 3d mach number @xmath70 , we have @xmath71 the initial mach number is set to @xmath72 . following the standard treatment in supersonic turbulence simulations ( e.g. mac low et al . 1998 ) , we impose a velocity field of power spectrum @xmath73 with a minimum wave number 2 and maximum wave number 10 in units of inverse box length . in models with a magnetic field , the field is assumed to be initially uniform in the @xmath74 direction , with a strength given by @xmath75 where @xmath76 is the mass - to - flux ratio in units of the critical value . because the adopted density profile is centrally condensed , the central part of the clump within @xmath77 has a larger mass - to - flux ratio @xmath78 . thus the envelope is more magnetized than the central part . our adopted degree of magnetization is in the range inferred by falgarone et al . ( 2008 ) . for all simulations , the top grid has resolution @xmath79 and the maximum refinement level is 4 , which corresponds to a highest resolution of @xmath80 au . with this resolution and our sink particle algorithms , the risk of over producing low mass stars is reduced in our model even though we do not include radiative heating , which happens predominantly on the disk scales ( krumholz et al . 2007 , bate 2009 ) . we will not be confident about the mass spectrum at the lowest mass end . however , the formation of massive stars , the main focus of this paper , is adequately resolved . the boundary condition is periodic for all hydro and mhd quantities and gravity . in the model with protostellar outflows , to prevent the high speed outflows from re - entering the other side of the boundary and thus artificially increasing their effects , we reduce their speeds by a factor of @xmath81 when they reach the boundary . we consider four models of increasing complexities , starting with a base model ( base ) that is initially neither turbulent nor magnetized . we then add to this base model , one by one , an initial turbulence ( hd ) , magnetic fields ( mhd ) and outflows ( wind ) . the models are summarized in table 1 and discussed in turn below . movies for the hd , mhd and wind models are included in the electronic version of the paper . our centrally condensed clump described in [ setup ] contains many thermal jeans masses . in the absence of any support in addition to thermal pressure , the clump collapses dynamically towards the center , forming an object within 0.2 myr ( see fig . 1 ) , close to the free - fall time of the central plateau part of the initial clump configuration @xmath82 myr . there is a rapid initial increase in the mass accretion rate , characteristic of free - fall collapse of regions of flat density distribution . the accretion rate settles into a more or less constant value , as the material in the envelope part of the initial clump ( that has an @xmath83 density distribution ) starts to fall into the center . this nearly constant rate of mass accretion is reminiscent of the collapse of the singular isothermal sphere ( sis ; shu 1977 ) , although the rate @xmath84is @xmath85 times larger than the standard shu s value of @xmath86for 20 k gas . it is , however , close to the characteristic free - fall rate for the clump as a whole , @xmath87(also plotted on fig . 1 for comparison ) . the large characteristic accretion rate predisposes the clump to massive star formation , although this tendency can be weakened or even suppressed by various factors , including turbulence . turbulence reduces the total rate of mass accretion onto all stars as long as it provides a significant global support to the clump . it is therefore not surprising to find a smaller stellar mass accretion rate in the case with turbulence ( model hd ) compared to the base model ( base ) , especially at early times ( see fig . 1 ) . the rate increases as the turbulence decays , reaching values of order @xmath88 , not far below the characteristic free - fall rate @xmath89 , at @xmath90 myr , which is comparable to both the free - fall time at the average density of the clump and the crossing time for the initial turbulence . thereafter , the clump dynamics is dominated by global collapse , as in the non - turbulent model base . the global collapse is illustrated in the first panel of fig . 2 , which plots the mass distribution and velocity field at the end of the simulation ( @xmath91 ) on a slice through the most massive star . as expected , the dense materials are collected preferentially near the bottom of the gravitational potential well , where they are fed by the clump matter that falls towards the bottom . the total star formation rate is lowered further by magnetic fields and outflows . since the central plateau region of the clump has a mass - to - flux ratio that is significantly larger than the critical value , the magnetic field there is not able to reduce the initial star formation rate significantly . however , the magnetic field becomes more important at late times , after the initial turbulence has decayed substantially and the more strongly magnetized clump envelope begins to collapse . the ensuing global collapse is retarded by the magnetic field in the cross - field direction . the retardation reduces the total stellar accretion rate by a moderate factor of 2 - 3 . the global collapse is even less prominent when outflows are included ( the second panel of fig . 2 ) . the outflows have apparently supplied enough energy ( and momentum ) into the clump that the bulk of its material is prevented from either collapsing towards a global center or settling along field lines into a thin sheet ( although some degree of flattening in the mass distribution is still evident ) . the feedback has kept the total mass accretion rate at only @xmath92 of the characteristic free - fall rate ( shown in fig . 1 for comparison ) . the global support directly affects the properties of the massive stars that are formed in the clump , as we show next . in this section , we will concentrate on massive stars , defined somewhat arbitrarily as those sink particles with masses more than 10 . our concentration on these stars is motivated by the fact that their formation mechanism is still under debate . we postpone a discussion of the properties of the lower mass stars to another publication . massive stars are formed in all of the four simulations listed in table 1 . in the limit of no turbulence ( model base ) , all of the accreted mass goes to a single object at the center . at the end of the simulation ( @xmath93 ) , the object has grown to more than 400 , which is clearly unphysical . the addition of an initial turbulence in model hd enabled the formation of eight massive sink particles by @xmath94 . the time evolution of the positions of these particles are shown in fig . 3 . there are two features worth noting . first , only five lines are clearly distinguishable in the plot . this is because the first 4 of the 8 stars are formed close together both in space and in time ; they stay close together at all times , so that their trajectories are indistinguishable . for our main purpose of evaluating mass accretion rate , we will treat them as a single object . second , the remaining four massive stars are formed at different times and locations . two of them join the first object at later times to form a tight group , while the other two merge into another system . we refer to the merged systems as group a and b respectively , and the massive stars formed at different times in each group as different generations . for example , a1 is the first generation of massive stars to form in group a. we will refer to all massive stars in each generation as `` a massive object '' ( because they form at essentially the same time and same location , and partition of mass between the stars depends somewhat on the sink particle treatment ) . some properties of the massive objects are listed in tables 3 - 5 at three different times @xmath94 , @xmath95 and @xmath96 . there are drastic differences between the massive objects formed in different simulations . at a relatively early time @xmath94 ( @xmath97 myr , about one global free fall time ; table 2 ) , 5 massive objects ( 8 massive stars ) form in the hd model , with a total mass of 131 , which is 37% of the stellar mass . the presence of a magnetic field in model mhd reduces the number of massive objects ( stars ) to 3 ( 4 ) , and their total mass to 73.6 , although the mass fraction of the massive stars remains similar ( 45% ) . when outflows are turned on ( model wind ) , the formation of massive stars is suppressed up to this time , with no stars more massive than @xmath81 . massive stars do form at later times in model wind . at @xmath98 , there are two massive stars ( 16.9 and 10.5 each ) formed , with a total mass of 27.4 , which is much less than the total mass of massive stars formed at the same time in model mhd ( 183 ) . the difference re - enforces the notion that the outflows retard massive star formation . the massive stars do continue to grow with time , however . they attain @xmath99 and @xmath100 respectively near the end of the simulation at @xmath101 ( about two global free fall times ) . they are each joined by another massive star , forming two loose groups . together , the four massive stars contain 86.9 , accounting for 35% of all the stellar mass in the cluster . how do massive stars form in our simulations ? to address this question , we will concentrate on the most massive object formed in each simulation , starting with the simpler , hd model that has turbulence but neither magnetic fields nor outflow feedback . the most massive object , hd - a1 in table 2 , formally reached a mass of 61.1 at the end of the simulation ( @xmath94 or 0.63 myr ) . it is among the first objects to form , at a time @xmath102 myr . the average mass accretion rate is therefore @xmath103 . the instantaneous accretion rate for this object is shown in fig . [ massivestar ] . it quickly rises to a plateau of @xmath4 , before jumping at @xmath104 myr to a second plateau that is @xmath105 higher . about equal amounts of mass ( @xmath106 ) are accreted in the two plateau phases . during the first plateau phase , the stellar mass increased by an order of magnitude , while the mass accretion rate stayed more or less constant . this behavior is characteristic of core collapse ( such as the collapse of the singular isothermal sphere where the mass accretion rate is a constant , shu 1977 ) rather than the bondi - hoyle type accretion which depends sensitively on the stellar mass . the rapid increase in mass accretion rate around @xmath107 myr is not triggered by any sudden increase in stellar mass , which again points to an external control of the mass accretion rate ; it is caused by the onset of global collapse of the clump . to examine the nature of the accreting flow before the global collapse in more detail , we plot in fig . [ hdyslice ] the mass distribution , velocity field and gravitation potential at @xmath108 , when the eventual massive object a1 is still in an early stage of accretion . it is embedded in a dense filament , which is produced by converging flows in the turbulence ( see also banerjee , pudritz & anderson 2006 ) . here , the inclusion of self - gravity in the calculation is crucial . it allows the converged material to stay together , rather than re - bouncing and dispersing quickly ; in other words , the self - gravity softens the high - speed collision that creates the filament . the filament feeds the object at a high rate of @xmath4 , some 20 times the standard shu s rate for an singular isothermal sphere ( sis ) . as stressed by banerjee et al . ( 2006 ) , the high accretion rate is due to the rapid formation of non - equilibrium structures ( dense filaments ) that are fed continuously by converging flows ( and are thus very different from the equilibrium sis ) . their existence does not depend on the stellar objects embedded in them ; rather , the objects are produced by the runaway collapse of the densest parts of the filaments , often at the intersections with other filaments . in this sense , the formation of the most massive object in this early phase is not due to competitive accretion , since the gravity of the object is not the primary driver of the mass accumulation in the filaments ; there are few objects to compete against at this stage of star formation in any case . it may be viewed as a special version of core collapse , in which the `` core '' is an over - dense , highly out - of - equilibrium filament that existed before the object and that continues to grow out of the converging flow even as part of it disappears into the collapsed object . the removal of dense gas from the filament via star formation also promotes further mass accumulation in the filament , by reducing its internal pressure and thus the resistance to further compression . in this picture , the massive star formation is simply part of the rapid `` core '' formation process . the use of sink particles in our simulation enables us to go beyond the calculations of banerjee et al . ( which stopped before significant mass accretion onto the stellar object ) and follow the mass accretion for a long time , and to reveal a second phase of even more rapid accretion , driven by global collapse . the global collapse at the end of the pure hydro simulation ( t=3 ) was already shown in fig . it is further illustrated in fig . [ hdyslice ] , which shows the distribution of mass , velocity field and contours of gravitation potential in a sub - region centered on the most massive object , with the positions of stars superposed . it is clear that the global collapse has supplied the central region near the bottom of the gravitational potential well with plenty of dense material , which fuels both the enormous rate of total stellar mass accretion ( close to the characteristic free - fall rate , see fig . 1 ) and the high accretion rate of the most massive object , which is but one of many stars in the region . in this crowded region , competitive rather than core accretion is likely at work . the reason is that the stars in this region are not enveloped by permanent host cores , because the dense gas in the region is constantly drained onto stars and constantly replenished by the global collapse . the most massive object is located near the minimum of the global potential well and accretes the global collapse - fed material at a rate higher than any other object . nevertheless , its accretion rate is only @xmath109 of the total rate . in other words , the vast majority of the globally collapsing flow is diverted to stars other than the most massive object . its accretion rate is large ( @xmath110 ) during this late phase only because of an even larger global collapse rate . a common theme of the early and late phases of rapid accretion is that the dense gas that feeds the most massive object at a high rate is gathered by an agent external to the object . it is the converging flow set up by the initial turbulence in the early phase and the globally collapsing flow in the late phase . in this aspect , the formation mechanism is closer to core collapse than to competitive accretion . one may plausibly identify the dense filament in the early phase and the dense region at the bottom of the global gravitational potential well in the late phase as a mckee - tan core ( mckee & tan 2003 ) . however , the `` cores '' so identified are transient objects that are not in equilibrium . they are evolving constantly , with mass growing ( from converging or collapsing flow ) and depleting ( into one or more collapsed objects ) at the same time . the replenishment of dense gas in the `` core '' is the key to the long duration of high mass accretion rate for the most massive object , which is many times the free fall time of the small dense `` core '' and comparable to the global collapse time of the clump as a whole . in other words , there is no hint of any drop in the mass accretion rate after a finite reservoir of core material gets depleted over a ( short ) core free - fall time . indeed , the nearly constant or increasing accretion rate for the most massive object highlights the important issue of when and how the mass accretion is terminated ; there is an urgent need to tackle this issue if the final stellar mass is to be determined . we will return to a more detailed discussion of the nature of the massive star s high accretion rate in the absence of magnetic fields and outflow feedback in [ unregulated ] . the effect of the magnetic field in model mhd on the mass accretion rate onto the most massive object is relatively modest ( see fig . [ massivestar ] ) . as in the hd case , there are two plateau phases . the first starts somewhat later than that in the hd case , indicating that the initiation of star formation in general and massive star formation in particular is somewhat delayed , as expected , because of magnetic cushion of turbulent compression . once started , the accretion rate onto the most massive object is in fact slightly higher in the mhd case than in the hd case , indicating that the magnetic field does not significantly impede the initial phase of rapid mass accretion , which is enabled by the dense structures formed by turbulent compression . indeed , dense structure formation is aided by a strong magnetic field , which forces the turbulent flows to collide along the field lines . when viewed in 3d , the dense structure resembles a warped , fragmented disk , with transient spirals that come and go . the spirals indicate significant levels of rotation on relatively small scales , which may hinder the stellar mass accretion due to centrifugal barriers . the centrifugal barrier may be weakened or even removed by magnetic braking . evidence for the braking is shown in fig . [ mhdyslice ] , which shows a magnetic bubble driven from the central region , where active mass accretion is occurring . magnetic braking driven bubbles provide a form of energy feedback that should accompany any star formed in a magnetized cloud ( e.g. , draine 1980 ; tomisaka 1998 ; mellon & li 2008 ; hennebelle & fromang 2008 ) but is absent from purely hydrodynamic models . we expect this form of feedback to become more important with a higher spatial resolution , since the rotating material closer to a star would be able to wrap the field lines more quickly ( until the flux freezing under the ideal mhd assumption breaks down ) . indeed , the protostellar outflow feedback may be viewed as a special form of this magnetic bubble - driven feedback , since the outflow is thought to be driven by the field lines that are forced to rotate rapidly by the circumstellar disks ( knigl & pudritz 2000 ; shu et al . 2000 ) . despite the feedback through the magnetic bubbles , large - scale collapse does occur at late times . when enough mass has been accumulated near the bottom of the gravitational potential well , the self - gravity of the accumulated gas overwhelms the magnetic tension forces , leading to a rapid cross - field collapse and an elevated , second phase of rapid mass accretion . the mass accretion is facilitated by the magnetic braking , which enables the high angular momentum material to sink closer to the center than it would be in the absence of the braking . the collapse is not as global as in the hd case , however , as shown in fig . [ infall ] . at the end of the mhd simulation ( @xmath98 ) , the bulk of the clump material infalls towards the bottom of the potential well at a speed of order twice the sound speed or less , unlike in the hd case , where the infall is faster and more widespread . the collapse remains significantly retarded by the magnetic tension except deep in the potential well . the combined effect of outflows and magnetic fields on the formation of the most massive object is much greater than that of magnetic fields alone . from fig . [ massivestar ] , we find that , even though a massive star of more than 45 was eventually formed at the end of the wind simulation , it took nearly @xmath111 yrs to accrete the mass , with an average rate of only @xmath112 . the accretion rate at early times is even lower , with a value of @xmath113 during the initial third of the time ; it increases to @xmath114 during the second third . these rates are about a factor of 5 lower than those in the hd and mhd models at comparable times . the question is : why does the most massive object accrete at such a low rate ? the basic reason is of course the outflows in the wind model , which are much more powerful than the magnetic braking - driven bubbles that exist in the mhd case . the outflows modify both the mass distribution and velocity field , and thus the outcome of the gravitational dynamics . as a result , the most massive object forms at a somewhat different time and location . despite the additional physics in both mhd and wind models it is initially still embedded in a dense filament , as it is in the hd case . as discussed earlier , the filament is the key to the first , turbulence compression - induced phase of rapid mass accretion in the hd and mhd models . in the wind model , accretion from the filament is reduced in two ways . first , there are several stars formed in the filament . they all emit outflows which tend to break the filament into smaller segments . second , the outflows emerge more easily perpendicular to ( rather than along ) the filament . once they break out , they blow against the converging flow and slow down the mass accumulation that feeds the growth of the filament in the first place . as the outflows propagate to large distances , a fraction of their momentum is lost through the computation boundary . the remaining fraction is deposited in the lower density region that surrounds the denser region of active star formation near the bottom of the potential well . the momentum deposition allows the bulk of the clump material to be supported against rapid global collapse . the absence of a global collapse is illustrated in fig . [ windyslice ] , which shows a rather chaotic velocity field that involves both infall and outflow . it is further quantified in fig . [ infall ] , which shows that the mass - weighted infall speed toward the most massive object is subsonic over most of the volume , except within a radius of about 0.2 pc , where the speed becomes two to three times the sound speed ( still much lower than the local free - fall speed ) . the slower infall leads to a smaller amount of dense gas accumulating near the bottom of the potential well . the accumulated dense gas is also more fragmented because of interaction with outflows . both factors limit the mass accretion rate onto the most massive object , especially at late times . as in the hd and mhd cases , the most massive object is not formed out of a pre - existing dense core . the case against the pre - existing dense core picture is stronger in the wind case , because it takes longer ( nearly two global free - fall times ) to accumulate the final mass , whereas a pre - existing dense core should collapse and exhaust its mass in a local ( core ) free - fall time , which should be a very small fraction of the global free - fall time . the slow accretion rate and long accretion time are clear evidence that the formation of the most massive object is controlled by the global clump dynamics in our wind simulation . massive stars form quickly in our simulated parsec - scale clump in the absence of any magnetic field or outflow feedback ( see model hd ) . they are fueled by high mass accretion rates from either the dense filaments that are formed by turbulent compression or the clump - wide collapse due to the global turbulence dissipation . in both cases , the dense material that is depleted onto the massive star is constantly replenished . the replenishment is illustrated in the left panel of fig . [ starcore ] , where the mass of the most massive object is plotted as a function of time , along with the mass of the gas within a core " of 0.1 pc in diameter around the object ; the size was chosen to coincide with the fiducial value that mckee & tan ( 2003 ) used to define their turbulent cores . " it is clear that the core " so defined has an initial mass that is well below the final mass of the most massive object , and thus can not supply all of the mass of the object . as the object gains mass , the mass of the core " stays roughly constant or even increases ( rather than decreases ) , which demands that the core mass be replenished or fed from the surrounding clump . we are thus motivated to term this model the clump - fed massive star formation " model . our `` clump - fed massive star formation '' model ( cf model for short ) contains elements of the two widely discussed scenarios for massive star formation in the literature : the turbulent core model of mckee & tan ( 2003 ) and competitive accretion model of bonnell et al . it has in common with the turbulent core model in that the mass accretion rate onto the massive star is not primarily determined by the star itself , but rather by the properties of the pre - existing gas that produces the seed of the star in the first place and that continues to feed the stellar growth ; in other words , if the massive star were to be removed prematurely , another seed would be produced in its place and grow to a high mass . it is the properties of the fueling gas that determine the stellar mass accretion rate and thus the stellar mass . mckee & tan envisioned this gas to be a pre - existing massive turbulent core . it collapses to form the massive star , with a mass accretion rate that depends on the core structure . for example , if the turbulent core has a density profile of @xmath34 , the mass accretion rate would increase linearly with time @xmath115 ( mckee & tan 2003 ) . in our cf model , the pre - existing gas is the cluster - forming clump , which produces the transient dense material that feeds the massive star at a high rate through two mechanisms : ( 1 ) the collapse of the dense filaments produced by turbulent compression , as already emphasized by banerjee et al . ( 2006 ) , and ( 2 ) global collapse , driven by the turbulence decay in our simulations , but can in principle be caused by an external compression as well . the first mechanism depends on the detailed properties of the initial turbulence in the clump , which are not well constrained observationally . the second mechanism depends on the dissipation of ( supersonic ) turbulence that is inevitable , and should be more robust . our unregulated cf model has in common with the competitive accretion model in that a massive star can form even in the absence of a pre - existing turbulence - supported massive core and that the global gravitational potential and dynamics of the clump play an important , even the dominant , role in massive star formation . the latter is especially true at late times , when the global collapse feeds mass at a high rate to the compact region near the bottom of the potential well , where a large number of stars ( including massive stars ) are already present . in the absence of stellar feedback , competition for this clump - fed material in the crowded region is unavoidable . the most massive object tends to grow the fastest , mainly because it typically locates closest to the center of the potential , as emphasized by bonnell et al . although the numerical results of our grid - based amr hydro simulations are in broad agreement with those of sph simulations of bonnell et al . ( 2003 ) , we differ from their interpretation of the results regarding the formed stars , particularly as it relates to massive star formation . when a massive star is formed , the vast majority of the clump mass remains in the gas . the high accretion rates of the massive stars at late times of our hydro simulation derive directly from the global gas collapse . even if there were no stars near the center of the collapse ( or more likely the mass accretions onto individual stars are terminated by the stellar feedback ) , ( other ) massive stars can still form from scratch , as long as the collapse delivers mass to the center at a high enough rate . so we argue it is predominantly the structure and dynamics of the gaseous component that set the relevant physics in forming the massive stars rather than the properties of the stars made previously . a moderately strong magnetic field ( corresponding to the observationally inferred dimensionless mass - to - flux ratio of a few ) does not qualitatively change the clump - fed picture for massive star formation by itself . as discussed in [ mhd ] , the magnetic field has relatively little effect on the high mass accretion rate onto the most massive object induced by the initial turbulent compression and filament formation . indeed , it is conducive to filament formation by guiding the converging flows to collide along the field lines . the magnetic field does slow down the global collapse - induced rapid accretion by a modest factor of a few . the basic ingredients of the cf model remain , however , as can be seen from the middle panel of fig . [ starcore ] . specifically , the initial core mass is still much less the final mass of the most massive object , and most of the mass that goes into the object still needs to be replenished or fed from the surrounding clump . we conclude that the feeding processes for massive star formation are only weakly regulated by the magnetic field , perhaps because the field is only moderately strong ( i.e. , the clump is moderately supercritical ) , and it resists the feeding passively rather than actively ( unlike the outflows , see below ) . the magnetic field does change the dynamical coupling between different parts of the clump , an important aspect of cluster and massive star formation that we plan to return to in a future investigation . massive star formation through rapid accretion of the mass that is fed from outside the small region surrounding the forming star is particularly vulnerable to outflow feedback . this is because the feeding is directly opposed by the feedback . in the case of the rapid accretion fed by turbulent compression and filament formation , the filament is quickly chopped up into small segments by the outflows driven by the stars formed in it ( see the outflow movie in the electronic version of the paper ) . the outflows also slow down any further mass accumulation in the filament after star formation is initiated . as a result , this mode of turbulent compression - fed massive star formation is strongly regulated , perhaps even choked off completely , by outflow feedback . in the case of the rapid accretion fed by global collapse , the infall is countered by the outflows on all scales , especially on the global clump scale . the additional global support provided by the outflow feedback can reduce the total star formation rate by a large factor . this reduction affects the formation of all cluster members , especially the massive stars . this is because the massive stars receive a larger fraction of the collapse - fed material , and are thus more sensitive to the change of global clump dynamics . they also tend to complete their formation toward the end of cluster formation ( even though their seeds tend to be among the first objects to form ) , making them more prone to the accumulative influence of multiple generations of outflows that precede their eventual formation . since the outflows are believed to be driven by the release of gravitational binding energy from mass accretion in one form or another ( konigl & pudritz 2000 ; shu et al . 2000 ) , and most of the accreted mass goes to low - mass ( rather than high - mass ) stars for a salpeter - like imf , a large fraction if not the bulk of the outflow feedback should come from the low - mass stars . in this sense , the formation of low - mass stars in a dense clump can profoundly influence the formation of massive stars in the same clump , through their feedback on the clump dynamics . whether massive stars eventually form or not in a dense clump depends on the extent to which all the outflows in the clump collectively regulate the global collapse and slow down the star formation . if the total star formation rate is reduced , for example , below the fiducial minimum rate for massive star formation , @xmath116 , no massive stars would form at all . in the opposite extreme where the outflows are weak and the feedback is not strong enough to reverse the global collapse , stars ( especially massive stars ) would form quickly , as in our pure hydro simulation . the degree of outflow regulation will depend on the properties of the outflows , including their strengths and degrees of collimation , both of which are somewhat uncertain . what we have demonstrated explicitly through numerical simulation is that for well - collimated jets of reasonable strength the outflow feedback can prevent rapid global collapse , and keep the total star formation rate an order of magnitude below the characteristic free - fall rate . massive stars do eventually form in our simulation that includes outflow feedback . it demonstrates that outflows can strongly regulate massive star formation , but do not necessarily quench it completely , especially in dense massive clumps with a high characteristic free - fall rate @xmath12 . for such clumps , even when the bulk of the clump material is supported , a small fraction can still percolate down the global gravitational potential well , feeding the formation of massive stars near the center at a high enough rate . the fact that our outflow - regulated massive star formation remains clump - fed rather than core - fed is illustrated in the right panel of fig . [ starcore ] . as in the hd and mhd models that do not have outflow feedback , the mass of the 0.1 pc - sized `` core '' that surrounds the most massive object in the wind model is initially smaller than the final mass of the object , and does not decrease monotonically as the object accretes . again , the bulk of the accreted stellar material must come from outside the `` core , '' which is an essential feature of our new scenario of outflow - regulated clump - fed ( orcf ) massive star formation . in this picture , the parsec - scale cluster - forming clump , rather than a 0.1 pc - sized turbulent core , is the basic unit for massive star formation . as pointed out by bonnell et al . ( 2007 ) , a potential drawback of the turbulent core model is that massive , turbulence - supported , cores tend to fragment into many stars rather than collapse monolithically , although radiative feedback from the rapidly accreting massive stars can reduce the level of fragmentation ( krumholz & mckee 2008 ) . fragmentation is expected to be less of a problem in our clump - fed picture of massive star formation , since the material near the forming massive stars does not have to be supported by a strong turbulence ; it is typically in a state of rapid collapse that feeds the growing massive stars at a high rate , even though the clump as a whole may remain supported by ( possibly outflow - driven ) turbulence and , perhaps to a lesser extent , magnetic fields . if the parsec - scale dense clumps are indeed the basic units of massive star formation , how they form in the first place becomes an important problem that deserves close theoretical and observational attention . massive stars play a dominant role in galaxy formation and evolution . our orcf scenario suggests a rough criterion for their formation that can be used in global galaxy formation simulations that reach the scale of the cluster forming clumps but do not resolve their internal structure . the criterion is based on a threshold for mass accretion rate . a high mass accretion rate is needed not only to overcome the radiation pressure ( wolfire and cassinelli 1987 ) and quench the development of hii region ( walmsley 1995 ) , but also to satisfy observational constraints on the time scale of massive star formation . wood & churchwell estimated an age of @xmath117 years for uc hii regions , which probably represent a relatively late stage of massive star formation , when the bulk of mass accretion has completed and the mass accretion rate becomes too low to trap the hii region ( churchwell 2002 ) . the majority of the stellar mass may be accreted during the hot core ( and perhaps hypercompact hii ) phase , which lasts for a time of order @xmath118 years or less ( kurtz et al . the relatively short time scale is also consistent with the dynamical times estimated for massive molecular outflows ( which are presumably driven by rapid mass accretion during the main accretion phase , as in the case of low mass stars ; e.g. , bontemps et al . 1996 ) , which are typically of order @xmath118 years or less ( e.g. , zhang et al . if the time scale for massive star formation is indeed @xmath117 yrs or less , to form a star of 10 , a stellar mass accretion rate of @xmath119 or more would be needed . a minimum requirement for massive star formation in a clump is that the characteristic free - fall rate @xmath12 of the clump be greater than @xmath120 . the actual requirement will be more stringent . the accretion rate onto massive stars @xmath121 is related to the characteristic free - fall rate @xmath12 of the dense clump by two factors : @xmath122 , where @xmath123 is the actual rate of star formation normalized to the characteristic rate @xmath89 , and @xmath124 the fraction of the stellar mass accretion onto massive stars . we estimate the characteristic free - fall rate using the clump mass divided by the free - fall time at the average density @xmath125 ( where @xmath126 is the clump mass in units of @xmath127 and @xmath128 the clump radius in units of parsec ) , which yields @xmath129/yr for a @xmath130 clump of 1 pc in radius , consistent with the value obtained in our reference model of non - turbulent collapse . the most uncertain factor in the above equation is perhaps @xmath123 , the star formation rate compared to the characteristic free - fall rate . it depends on the extent to which the clump is supported globally . decaying turbulence can provide significant global support within a turbulence decay time , when @xmath123 can be significantly below unity . unless the star formation is terminated in one decay time , @xmath123 will eventually increase to a value not far from unity , as demonstrated in our hd model . a moderate magnetic field does not change the global support fundamentally , reducing @xmath123 by a factor of only a few ( see model mhd ) . together with the magnetic field , protostellar outflows can reduce @xmath123 by an order of magnitude , to values of order @xmath131 . the value of @xmath123 may range from about 0.1 ( before the virial turbulence decays away or after it is replenished ) to close to unity ( after the turbulence has decayed in the absence of magnetic support and turbulence replenishment ) . krumholz & tan ( 2007 ) argued that the star formation efficiency per free fall time is of order @xmath92 or less , implying @xmath132 . nakamura & li ( 2007 ) found a similarly low value for the well - studied nearby clump of active cluster formation ngc 1333 . based on these results , we choose @xmath133 as the fiducial value . a smaller @xmath123 would make it more difficult to form massive stars . the value for the remaining factor @xmath124 can be constrained both observationally and numerically . if the salpeter slope is universal for the upper part of the imf , then about 1/3 of the stellar mass must reside in stars more massive than 10 m@xmath134 ( assuming a lower cut - off at 0.3 @xmath135 ) . in this case , @xmath136 , which is similar to the ratios obtained in our simulations , according to tables 2 - 4 . to be conservative , we assume that all of the massive star accretion rate , @xmath137 , goes to a single star . if more than one massive star are fed at this rate , the requirement for massive star formation would be more stringent . taken together , the above considerations yield the following mass accretion rate for a massive star : @xmath138 which has to exceed @xmath139 for massive star formation to actually occur ( @xmath140 is the critical mass accretion rate for massive star formation in units of @xmath116/yr ) . this yields a rough condition on the ratio of the mass and radius of the clump for massive star formation : @xmath141 the significance of the above condition is that , if the fraction of the total stellar mass going into massive stars ( @xmath124 ) is constrained by observations , then whether a dense clump of a given mass and size ( and thus given @xmath12 ) can form massive stars or not boils down to the total star formation rate , @xmath142 . it depends on the clump dynamics , which is sensitive to the outflow feedback . the above condition is consistent with , for example , the sample of massive star forming clumps associated with water masers of plume et al . ( 1997 ) , which has a mean virial mass @xmath143 and mean radius @xmath144 . their ratio of 7.6 is comfortably above the fiducial value of @xmath145 in equation ( [ condition ] ) , indicating that massive stars can form , even when the total rate of star formation is an order of magnitude below the characteristic free - fall rate . a small value for the product @xmath146 ( @xmath147 ) may be the reason for the massive star formation to occur predominantly in those special regions of molecular clouds the massive dense clumps that are both massive and compact . the most severe limitation of the current work is perhaps the neglect of radiative feedback . radiative heating changes the clump fragmentation behavior , especially close to the forming stars ( krumholz et al . 2007 ; bate 2009 ) . this effect was mimicked to some extent by the relatively large sink particle merging distance adopted in our simulations ( with a length of 5 cells or @xmath7 au ) , which suppresses fragmentation on the small ( mostly disk ) scale . furthermore , if our orcf scenario is correct , the formation of massive stars may be more sensitive to the global clump dynamics ( which are less affected by the radiative heating ) than the gas properties close to the stars . nevertheless , we believe that the formation of massive stars will benefit from the suppression of fragmentation by radiative heating , especially near the bottom of the gravitational potential well of the clump , where the thermal jeans mass is formally smaller than the typical stellar mass . on small scales , radiative pressure may slow down the mass accretion onto individual massive stars somewhat , although the accretion may be enhanced to some extent by non - ideal mhd effects , such as ambipolar diffusion , which are not included in our ideal mhd calculations . another effect that we neglected was the hii region driven by the massive star s uv radiation . the expansion of hii regions provides a way to remove the clump gas , and perhaps terminate the cluster formation . it needs to be included in future simulations that aim to model the entire history of cluster formation . such studies may also need to include massive star winds , which are observed to have dramatic effects in some regions ( e.g. the carina nebula , see smith & brooks 2008 for a review ) . they are the main alternative to the hii regions as the means for terminating the cluster formation . a further limitation is the periodic boundary condition used in our simulation . it precludes any communication between the cluster - forming clump and its surrounding environment . if there is energy injection into the dense clump from the ambient medium , the externally supplied energy may aid the outflows in regulating the cluster and massive star formation . large - scale external compression may , on the other hand , lead to rapid clump collapse and massive star formation . in this case , the feeding of massive stars may extend beyond the parsec - scale clump , and the massive star formation may simply be part of the rapid clump formation . we have carried out amr - mhd simulations of massive star formation in dense , turbulent , parsec - scale clumps of cluster star formation including sink particles and outflow feedback . we find that , without regulation by magnetic fields and outflows , massive stars form quickly . they are fed at a high rate first by the converging flows in the initial turbulence and later by the global collapse induced by turbulence decay . a moderate magnetic field alone does not affect these feeding processes much . they are greatly modified , however , by a combination of protostellar outflows and magnetic fields . the outflows break up the turbulent compression - produced dense filaments that feed the massive stars at early times and stall the global collapse that fuel the massive star formation at later times . the outflow feedback is enhanced by a magnetic field , which links different parts of the clump together ; the coupling makes the deposition of the outflow momenta in the clump more efficient . the magnetically - aided outflow feedback can in principle reduce the total rate of star formation below the critical mass accretion rate for massive star formation and suppress the massive star formation completely . in practice , whether massive stars form in a dense clump or not depends on the properties of the clump ( particularly its mass and size ) and the degree of magneto - outflow regulation of its star formation ( see equation [ [ condition ] ] ) . for parsec - scale clumps of order @xmath7 , we have demonstrated explicitly through numerical simulations that the formation of massive stars is clump - fed and outflow - regulated . additional simulations and analysis are needed to determine whether this new scenario of outflow - regulated clump - fed massive star formation is applicable to more massive and/or more compact dense clumps . in a companion paper , we will explore the effects of the outflow feedback on the lower mass cluster members . walawender , j. , bally , j. , francesco , j. d. , jrgensen , j. , & getman , k. 2008 , handbook of star forming regions , volume i : the northern sky asp monograph publications , vol . 4 . edited by bo reipurth , p.346 llllll hd - a1 & 61.1 & 0.96 & 4 & ( 0.66 , 0.40 , 0.48 ) & + hd - a2 & 22.6 & 1.36 & 1 & ( 0.66 , 0.40 , 0.48 ) & + hd - a3 & 10.1 & 2.03 & 1 & ( 0.68 , 0.41 . 0.48 ) & + hd - b1 & 21.3 & 1.11 & 1 & ( 0.74 , 0.48 , 0.33 ) & + hd - b2 & 15.8 & 1.72 & 1 & ( 0.74 , 0.48 , 0.33 ) & + hd - all & 131 & & 8 & & 137 stars , 350 ; 37% massive + mhd - a1 & 12.3 & 1.05 & 1 & ( 0.41 , 0.34 , 0.51 ) & + mhd - a2 & 46.7 & 1.35 & 2 & ( 0.44 , 0.38 , 0.52 ) & + mhd - b1 & 14.6 & 2.30 & 1 & ( 0.51 , 0.65 , 0.36 ) & + mhd - all & 73.6 & & 4 & & 86 stars , 165 ; 45% massive + wind - all & 0 & & 0 & & 72 stars , 69.7 ; no massive stars + llllll mhd - a1 & 15.0 & 1.05 & 1 & ( 0.32 , 0.45 , 0.51 ) & + mhd - a2 & 110 & 1.35 & 2 & ( 0.44 , 0.40 , 0.51 ) & + mhd - a3 & 12.5 & 3.13 & 1 & ( 0.46 , 0.41 , 0.51 ) & + mhd - b1 & 45.8 & 2.30 & 1 & ( 0.46 , 0.65 , 0.39 ) & + mhd - all & 183 & & 5 & & 121 stars , 337 ; 54% massive + wind - a1 & 16.9 & 1.16 & 1 & ( 0.38 , 0.42 , 0.50 ) & + wind - b1 & 10.5 & 2.29 & 1 & ( 0.45 , 0.68 , 0.42 ) & + wind - all & 27.4 & & 2 & & 123 stars , 134 ; 20% massive + llllll wind - a1 & 46.4 & 1.16 & 1 & ( 0.47 , 0.36 , 0.51 ) & + wind - a2 & 11.8 & 1.34 & 1 & ( 0.43 , 0.33 , 0.47 ) & + wind - b1 & 16.7 & 2.29 & 1 & ( 0.49 , 0.67 , 0.44 ) & + wind - b2 & 12.0 & 2.31 & 1 & ( 0.51 , 0.67 , 0.43 ) & + wind - all & 86.9 & & 4 & & 183 stars , 250 ; 35% massive +

the simplest unified extension of the minimal supersymmetric standard model with bi - linear r parity violation naturally predicts a hierarchical neutrino mass spectrum , suitable to explain atmospheric and solar neutrino fluxes . we study whether the individual violation of the lepton numbers in the charged sector can lead to measurable rates for and . we find that some of the r parity violating terms that are compatible with the observed atmospheric neutrino oscillations could lead to rates for measurable in projected experiments . however , the obtained for those parameters is too high to be compatible with the solar neutrino data , excluding therefore the possibility of having measurable rates for in the model . hep - ph/0202054 + fisist/022002/cfif + * charged lepton flavor violation in supersymmetry with bilinear r parity violation * d. f. carvalho , m. e. gmez and j. c. romo + _ departamento de fsica , instituto superior tcnico _ + _ av . rovisco pais 1 , 1049 - 001 lisboa , portugal _ + ( 1,0)140 email addresses : dani@cfif.ist.utl.pt , mgomez@cfif.ist.utl.pt , jorge.romao@ist.utl.pt

hep-ph0202054

in the standard model ( sm ) , lepton number is exactly preserved in contradiction with the observed neutrino oscillations @xcite . several extension of the sm include patterns of neutrino masses and mixings which can provide a satisfactory explanation for these flavor oscillations . the consequences of the individual violation of the lepton numbers @xmath0 for charged lepton will be manifest in processes such as @xmath3 , @xmath7 , @xmath8 conversion in heavy nuclei , @xmath9 and @xmath10 @xcite . the experimental upper bound for these processes is quite restrictive , which imposes a significant constraint for the explanation of flavor in models beyond the sm . however , the mechanisms used to explain the origin of the tiny neutrino masses required to explain solar and atmospheric neutrino oscillations , typically imply that these processes may occur at small rates , motivating an increasing experimental interest in exploring further charged lepton flavor violating processes . the rates for charged lepton flavor violation ( lfv ) are extremely small in the sm with right handed neutrinos ( @xmath11 @xcite ) . in r parity conserving supersymmetric ( susy ) models , like the minimal supersymmetric standard model ( mssm ) , the presence of lfv processes is associated with vertices involving leptons and their superpartners @xcite . these processes are sensitive to the scalar mass matrices structure , a non diagonality of the latter in a basis in which fermions are diagonal , leads to a hard violation of flavor . the structure of the scalar mass matrices is very sensitive to the susy breaking , in particular in models where susy is softly broken , lfv imposes a severe constraint in the flavor dependence of the soft terms as they are generated in gut s and string inspired models @xcite . the inclusion of a `` see saw '' mechanism in the mssm provides a very attractive scenario to understand neutrino oscillations with very small neutrino masses , and at the same time gives rates for lfv processes accessible in projected experiments @xcite . the waiving of the r - parity symmetry in the mssm provides an alternative scenario to explain the generation of small neutrino masses . in this case the r - parity violating operators can be constrained by rare processes @xcite . the simplest extension of the mssm with bilinear r parity violation ( brpv ) @xcite ( allowing b conserving but l - violating interactions ) can explain neutrino masses and mixings which can account for the observed neutrino oscillations @xcite . the brpv model has been extensively discussed in the literature @xcite . it is motivated by the fact that it provides an effective truncation of models where r parity breaks _ spontaneously _ by singlet sneutrino vev s around the weak scale @xcite . moreover , they allow for the radiative breaking of r - parity , opening also new ways to unify gauge and yukawa couplings @xcite and with a potentially slightly lower prediction for @xmath12 @xcite . for recent papers on phenomenological implications of these models see ref . @xcite . as the parameters involved in the r parity violating operator are constrained in order to predict neutrino masses in the sub - ev range , we address in this paper the question of whether this operator will induce rates for charged lfv processes of experimental interest . some of them occur at tree level such as double @xmath13 decay @xcite and @xmath8 conversion in nuclei @xcite . one loop lfv decays as @xmath14 become interesting on this framework due to the experimental interest in improving the current limits @xcite : @xmath15 as we will show , the predictions for the last two processes are much lower than the above limits and will not constrain the brpv model . for @xmath5 the predictions are compatible with the current limit but could begin to constrain the model for the bounds that will be reached in current @xcite or planned experiments @xcite , if only the atmospheric neutrino data were taken in account . however the requirement that the one loop induced @xmath4 is in agreement with the solar neutrino data will imply that the predicted rates for @xmath5 will not be visible , even in those new experiments . this paper is organized as follows . in sections [ sec : themodel ] , [ sec : thepot ] and [ sec : massmatrices ] we describe the model , the scalar potential and the fermion mass matrices , respectively . in section [ sec : expressions ] we derive the expressions for the lfv processes . the results are presented in section [ sec : results ] and in section [ sec : conclusions ] we give our conclusions . the more technical questions regarding the mass matrices , couplings and the explicit formulas for the amplitudes are given in the appendices . using the conventions of refs . @xcite we introduce the model by specifying the superpotential , which includes brpv @xcite in three generations . it is given by @xmath16\ ] ] where the couplings @xmath17 , @xmath18 and @xmath19 are @xmath20 yukawa matrices and @xmath21 and @xmath22 are parameters with units of mass . the second bilinear term in eq . ( [ eq : wsuppot ] ) violates lepton number and therefore also breaks r parity . the inclusion of the r parity violating term , though small , can modify the predictions of the mssm . the most salient features are that neutrinos become massive and the lightest neutralino is no longer a dark matter candidate because it is allowed to decay . furthermore , we can observe that this model implies the mixing of the leptons with the usual charginos and neutralinos of the mssm . lepton yukawa couplings can be written as diagonal matrices without any loss of generality since it is possible to rotate the superfields @xmath23 in the superpotential , eq . ( [ eq : wsuppot ] ) , such that yukawa matrix @xmath24 becomes diagonal . conversely , in brpv models it is possible to apply a similar rotation to reduce the number @xmath25 parameters and provide a non - trivial structure to @xmath26 @xcite . supersymmetry breaking is parameterized with a set of soft supersymmetry breaking terms . in the mssm these are given by @xmath27\ ] ] where @xmath28 \ , .\nonumber\end{aligned}\ ] ] in addition to the mssm soft susy breaking terms in @xmath29 the brpv model contains the following extra term @xmath30 where the @xmath31 have units of mass . the electroweak symmetry is broken when the two higgs doublets @xmath32 and @xmath33 , and the neutral component of the slepton doublets @xmath34 acquire vacuum expectation values . we introduce the notation : @xmath35 where we shift the neutral fields with non zero vev s as @xmath36\ , , \quad h_{u}^{0}\equiv { 1\over { \sqrt{2}}}[\sigma ^{0}_{u } + v_{u}+i\varphi ^{0}_{u}]\ , , \quad \tilde{l}_{i}^{0}\equiv { 1\over { \sqrt{2}}}[\tilde{\nu } ^{r}_{i}+v_{i}+i\tilde{\nu } ^{i}_{i}]\ , .\ ] ] note that the @xmath37 boson acquires a mass @xmath38 , where @xmath39 . in addition to the above mssm parameters , our model contains nine new parameters , @xmath22 , @xmath40 and @xmath31 . the minimization of the scalar potential allows to relate some of these free parameters . the values of @xmath22 , @xmath40 are directly related with the neutrino masses and mixings as we will discuss below . the electroweak symmetry is broken when the neutral higgses and the neutral slepton fields acquire non zero vev s . these are calculated via the minimization of the effective potential or , in the diagrammatic method , via the tadpole equations . the full scalar potential at tree level is @xmath41 where @xmath42 is any one of the scalar components of the superfields in the superpotential in eq . ( [ eq : wsuppot ] ) , @xmath43 are the @xmath44-terms , and @xmath45 is given in eq . ( [ softbrpv ] ) . the tree level scalar potential contains the following linear terms @xmath46 where the different @xmath47 are the tadpoles at tree level , their explicit expressions can be found in ref . the five tree level tadpoles @xmath48 are equal to zero at the minimum of the tree level potential , therefore we can use them to express the parameters : @xmath49 in terms of : @xmath50 we have two possible solutions for @xmath51 : @xmath52 where @xmath53 , @xmath54 and @xmath55 where we have defined @xmath56 . as one can easily verify , the above relations lead to the mssm relation for @xmath57 in the limit of vanishing @xmath58 and @xmath59 . the uncertainty of the sign in the mssm is translated here into two possible values for @xmath51term . however for the values of @xmath58 and @xmath59 relevant to our work both solutions are close in module and of opposite sign . the values for @xmath60 and @xmath61 s can be expressed in terms of @xmath51 as : @xmath62\\ b_{i}&=&\frac{1}{v_{u}}\left[v_{d}\mu - \sum ^{3}_{j=1}\epsilon _ { j}v_{j}-\frac{1}{\epsilon _ { i}}(dv_{i}+ \frac{1}{2}\sum ^{3}_{j=1}v_{j}(m^{2}_{lij}+m^{2}_{lji}))\right]\end{aligned}\ ] ] the equivalent equations for the mssm equations are obtained by setting @xmath63 equal to zero . as we discussed in the previous section the presence of brpv terms in the superpotential , eq . ( [ eq : wsuppot ] ) , induces non zero vev s for the sneutrinos and enables the neutrinos to have a mass , with a value related to the size of @xmath64 and the susy parameters involved on the electroweak symmetry breaking . furthermore , the non conservation of the r parity allows the susy partners to mix with the sm particles . in this section we describe with detail the resulting neutralino neutrino and chargino charged lepton mass matrices , since they are the most directly related to our problem . the complete set of mass matrices for the brpv model can be found in ref . @xcite . the range of values of the @xmath65parameters is indirectly associated to the size of the neutrino masses predicted by the model . to explore this relation we describe next the mass mixings among neutralinos and neutrinos . in the basis @xmath66 the neutral fermion mass matrix @xmath67 is given by _ n= where , _ ^0 - 2pt=-4pt is the standard mssm neutralino mass matrix and m= characterizes the breaking of r - parity . the mass matrix @xmath67 is diagonalized by ^*m_n^-1=diag(m_^0_i , m__j ) , [ chi0massmat ] where @xmath68 for the neutralinos , and @xmath69 for the neutrinos . since @xmath70 the mass matrix @xmath67 is similar to the _ see saw _ mass matrices and takes approximately the form diag(@xmath71 ) , with : [ eq : meff ] m_eff = - m _ ^0 ^ -1 m^t = ( -2 mm ccc _ e^2 -1pt&-1pt _ e _ -1pt&-1pt _ e _ + _ e _ -1pt&-1pt _ ^2 -1pt&-1pt _ _ + _ e _ -1pt&-1pt _ _ -1pt&-1pt _ ^2 -3 mm ) . where the @xmath72 parameters in eq . ( [ eq : meff ] ) are defined as : _ i v_i + v_d _ i [ lambdai ] one of the neutrino species acquire a tree level non zero mass , given by : [ mnutree ] m__3 = tr(m_eff ) = ||^2 , where @xmath73 . the two other neutrinos can get masses at one loop as it is discussed in ref . @xcite . for our purposes it will be important to have an estimate of the values of @xmath74 . we will use the results of ref . @xcite where it was found that , to a very good approximation , @xmath75 and @xmath76 the explanation of the data on neutrino oscillations given in ref . @xcite requires the neutrino masses to be be on the sub - ev range in order to fit the data on atmospheric neutrino oscillations . in our examples we take a @xmath77 , which leads to values of the @xmath78 in the range of @xmath79 , for the values of the susy parameters that we will consider . considering that we take positive values for @xmath51 we should also take negative values for the product @xmath80 to avoid our analysis to be constrained to small values of @xmath58 . however , as we will see , for the values of the susy parameters that give the largest @xmath81 , the values of the @xmath82 have to be below 0.1 gev , if @xmath83 ( ev)@xmath84 , as required by the solar neutrino data . due to the r - parity violating terms in the superpotential , eq . ( [ eq : wsuppot ] ) , the charginos mix with the charged leptons , linking therefore the problem of the masses of the neutrinos with the problem of the charged lepton flavor violation . we describe in this subsection the chargino lepton mass matrix to explain how the flavor mixing on the charged lepton sector arises . in a basis where @xmath85 the corresponding charged fermion mass terms in the lagrangian are @xmath86 where the chargino charged lepton mass matrix @xmath87 is given in the appendix [ ap : a ] . as in the mssm , @xmath87 is diagonalized by two rotation matrices , and we include the physical charged leptons and charginos into a set of five charged fermions defined as : f_i^-=u_ij _ j^- ; f_i^+=v_ij _ j^+ such that ^ * ^-1 = where @xmath88 is the diagonal charged fermion mass matrix . in the previous expressions the @xmath89 are two component spinors . we construct the four component dirac spinors out of the two component spinors with the conventions , @xmath90 and @xmath91 to be the particles and not the anti particles . ] , _ i^-= ( ) the parameterization of the matrices @xmath92 , and @xmath93 given in appendix [ ap : a ] , that was introduced in ref . @xcite , provides a very accurate representation of the exact result . by comparing the numerical results with the analytical expressions shown in appendix a , we found discrepancies of less than the @xmath94 . to obtain this level of accuracy we had to introduce corrections in the definition of @xmath95 , @xmath96 , and @xmath97 with respect to the formulas of ref . these arise mainly from including the sub - matrix @xmath98 in our derivation ( see appendix a ) . although the size of the matrix elements of @xmath98 is smaller than the other components of @xmath87 , it must be taken into account in order to match the results of the smaller elements of @xmath93 and @xmath92 found in the exact diagonalization . our definition of @xmath95 and @xmath96 leads to the correct form for the lower right @xmath99 sub - matrices of @xmath93 and @xmath92 . we will make use of it to explain the details of our results . the inclusion of the matrix @xmath98 in the determination of @xmath97 allows to display the dependence of the matrix elements on the @xmath100-parameters , rather than a explicit dependence on the @xmath65 s as quoted in ref . @xcite . also , we must observe that the elements of @xmath101 exceed the ones of @xmath97 by several orders of magnitude . therefore we can anticipate that the couplings containing the matrix @xmath92 in appendix [ ap : c ] , will be suppressed with respect to the ones containing the elements of @xmath93 . the effective operators that generate the decays @xmath103 and the lepton anomalous magnetic moment can be written as : @xmath104 . ] the one loop contributions to @xmath105 in the model under consideration arise from the diagrams of figs . [ fig1][fig3 ] @xmath106 the partial contributions on the above expression correspond to the addition of the sets of diagrams represented in each figure : @xmath107 the superscripts in each contribution on the right denote the fermion and boson internal lines of the corresponding diagram . for the quarks - squarks diagrams we include the symbol @xmath108 to indicate whether the photon is attached to the fermion or the boson line . we follow the notation of @xcite indicating by @xmath109 the eigenstates of the charged scalar mass matrix , by @xmath110 and @xmath111 the eigenstates for the sneutrino higgs scalar mass matrices , cp even and cp odd , respectively . the contributions to @xmath112 arise from the diagrams in fig . the index @xmath113 corresponds to the eigenstates of the chargino lepton mass matrix , while the indices @xmath114 correspond to the lepton generation indices in the limit of the mssm with r parity conservation . these diagrams will become the sm contribution to @xmath115 , eq . ( [ leffg2 ] ) , in the limit @xmath116 . in the case of the sm it provides the main contribution to @xmath117 , no contribution for charged lfv processes when neutrinos are considered massless and a very suppressed contribution for the values of @xmath118 compatible with the experimental limits @xcite . the contributions to @xmath119 arise from the three diagrams of fig . [ fig2 ] , where the index @xmath120 refers to the eigenstates of scalar mass matrices . @xmath121 are the eigenstates of the @xmath122 charged higgs slepton mass matrix and @xmath123 represent the eigenstates of the @xmath124 neutral higgs sneutrino scalar and pseudoscalar mass matrices , respectively . in the limit where r parity is conserved these three diagrams will be combined in the two supersymmetric diagrams contributing to the @xmath125 in the mssm . in this this limit , these diagrams are flavor conserving when the soft - terms are universal as given by minimal supergravity version of the mssm . . ] @xmath126 arise from the four diagrams of fig . [ fig3 ] , where the indices @xmath127 refer to the eigenstates of @xmath128 squark mass matrices and indices @xmath129 are the quark generation indices . these diagrams are not present when r parity is conserved . the branching ratio for the rare lepton decays @xmath131 is given in the literature @xcite and we do not repeat the derivation here . the result is @xmath132 where the amplitudes @xmath133 and @xmath134 were defined in eq . ( [ alar ] ) . the complete expressions for the amplitudes corresponding to these processes in the brpv model are given in the appendix [ ap : c ] . in their derivation we have neglected the mass of the outgoing fermion . the expression for the muon anomalous magnetic moment can be obtained from the lagrangian given in eq . ( [ leffg2 ] ) . one obtains @xcite , @xmath135 the amplitudes @xmath136 can also be obtained from the formulas of appendix c by including the effect of @xmath137 in both external lines of the diagrams . to do that we just have to include a factor of 2 in the part of the amplitude containing the function @xmath138 : @xmath139 . ] the brpv model that we consider adds more free parameters to the ones already present on the mssm . however , if we consider the phenomenological constraints imposed on the mssm by the limits on the mass of the lightest neutral @xmath140even higgs @xmath141 , by the @xmath142 and by the value of the @xmath143 , as well as those derived from neutrino physics on the brpv parameters , we can narrow the space of parameters such that generic predictions for @xmath144 and @xmath145 can be made . we assume the parameter space of the mssm with universal soft - terms and gut unification , @xmath146 with the addition of the brpv parameters , @xmath147 @xmath148 is defined such that @xmath149 . the quantities @xmath150 ( @xmath151 being the gut gauge coupling constant ) and @xmath152 are evaluated consistently with the experimental values of @xmath153 , @xmath154 , and @xmath155 at @xmath156 . we integrate numerically the rge s for the brpv model , at two loops in the gauge and yukawa couplings and a one loop in the soft terms , from @xmath152 down to a common supersymmetric threshold @xmath157 . from this energy to @xmath156 , the rge s of the sm are used . as we explained before , the minimum conditions of the effective scalar potential allows us to express the values of @xmath158 in terms of @xmath159 . these are evaluated at the scale @xmath160 . the value of @xmath51 obtained at this scale is similar to the one obtained by minimizing the effective potential with the complete 1-loop mssm contributions @xcite . the 1-loop contributions arising from rp - violating terms for these parameters are comparatively much smaller . we fix the elements of the quark yukawa matrices at the gut scale , consistently with the experimental values of the quark masses and the absolute values of the ckm matrix elements . in the case of the charged leptons we have to make sure that the three lightest eigenvalues of the chargino charged lepton matrix are consistent with the experimental values of the charged lepton masses . the values of @xmath161 and @xmath162 are chosen in region of the parameter space favored by the considerations presented in ref . @xcite , so that we can compare our results with typical predictions for the @xmath1 in the mssm with a _ `` see saw '' _ mechanism , as discussed in ref . @xcite . obviously , since our model breaks r - parity , the lsp is not a dark matter candidate and therefore the cosmological preferred areas of ref . @xcite do not apply to our study . however , the restriction of considering points in the @xmath163 plane such that @xmath164 , is the most restrictive . the susy contribution to @xmath143 @xcite favors the sign of @xmath51 to be positive for the choice of susy parameters given below . we found that the upper bound of eq . ( [ eq : g2range ] ) ( see below ) on @xmath165 is less restrictive than the one imposed by @xmath164 . we analyze three sets of susy parameters , * @xmath166 , @xmath167 , @xmath168 , @xmath169 , @xmath170 . * @xmath171 , @xmath167 , @xmath172 , @xmath169 , @xmath170 . * @xmath171 , @xmath173 , @xmath172 , @xmath169 , @xmath170 . the six free brpv parameters @xmath64 reduce to three if we take into account the constraints imposed by the predictions for neutrino oscillations in this model , as given in ref . @xcite . by setting the atmospheric neutrino anomaly scale to the magnitude of the tree level non zero value of one of the neutrinos , eq . ( [ mnutree ] ) , we fix the value @xmath174 for each susy point , where the @xmath175 were defined in eq . ( [ lambdai ] ) . we then follow the discussion of ref . @xcite , where it was shown that the conditions @xmath176 satisfy both the atmospheric neutrino anomaly mixings and the chooz result @xcite . we then obtain a linear relationship between each couple @xmath64 . therefore we study the dependence of the process under consideration on the values of @xmath177 for a neutrino mass of @xmath77 , on the upper limit of the allowed range for the atmospheric neutrino anomaly . for comparison purposes we also present some results for @xmath178 on the middle of that range , anf for @xmath179 . we will assign random values to @xmath180 and @xmath181 in the range : @xmath182 however the requirement that @xmath183 will exclude values of @xmath184 . this will be explicitly shown in our results . the value of @xmath185 is kept fixed since our results are not altered when it varies on the above range . we perform a full numerical analysis with the exact diagonalization of matrices involved in the computation of branching ratios . the main contribution for @xmath1 , eq . ( [ eq : brm ] ) , comes from the amplitudes @xmath187 . the partial contributions from the various diagrams listed in eq . ( [ eq : pam2][eq : pam3 ] ) are displayed in fig . [ fig4 ] , for the set of parameters b ) . we have found that they are all independent of @xmath185 and that they display a linear behavior as a function of the product @xmath188 , when this product is larger than 0.1 gev except for the cancellation observed in @xmath189 . the values of the amplitudes arising from the diagrams of fig . [ fig1 ] depend on the @xmath65 s through the @xmath100 s which are kept fixed , and therefore remain constant . we also found the contributions of the diagrams of fig . [ fig3 ] to be of the same order of magnitude . as we can see , the sum of all of them , @xmath190 , is almost a linear function of @xmath188 . the amplitudes arising from the diagrams of fig . [ fig2 ] are the dominant ones . the one mediated by the neutralino ( @xmath191 ) is smaller than the dominant chargino exchange ( @xmath192 ) , except in the range where the cancellation takes place . corresponding to the choice of parameters b ) of section 6.1 for @xmath193 gev ( solid ) and @xmath194 gev ( dash ) . the values of @xmath195 change sign on each branch of the curve , the left one corresponds to positive values . , height=377 ] we can also observe in fig . [ fig4 ] that the cancellation which appear in the @xmath196 depends on the value of @xmath180 . the values that we show correspond to two different choices of @xmath180 ( note that if we allow @xmath180 to change randomly , as we have done with the other amplitudes , the values for @xmath196 would be a distribution of dots ) . the behavior of @xmath196 can not be attributed to an accidental cancellation between the scalar and pseudoscalar parts as one may naively expect , on the contrary both parts add constructively and almost vanish simultaneously . the behavior of that amplitude can be explained when we identify which are the particles running in the loops of fig . [ fig2 ] that are responsible for the main contributions : @xmath197 and @xmath198 . then we can obtain an accurate approximation for the amplitude by using the formulas given in the appendices [ ap : a ] , [ ap : b ] and [ ap : c ] . let s consider @xmath199 since the contribution of the corresponding pseudoscalar exchange is almost identical . we get from eq . ( [ eq : alcs ] ) for the dominant contributions , @xmath200\ ] ] using the definitions of appendix [ ap : b ] we find , @xmath201 where @xmath202 is the yukawa coupling of the muon . we can then write , @xmath203 where @xmath204 the quantities @xmath205 and @xmath206 are independent of the brpv parameters @xmath58 and can be evaluated given the susy parameters . the dependence of the amplitude @xmath207 on the @xmath58 comes from the matrix elements @xmath208 , @xmath209 and @xmath210 . using the appendix [ ap : a ] we can find approximate expressions for these matrix elements that display explicitly this dependence . we get , @xmath211 hence we can find the value of @xmath181 at which @xmath212 , @xmath213 the position of the cancellation changes , with the value of the susy parameters and also with value of @xmath214 , as we can see from eq . ( [ cancellation ] ) . this explains the qualitative changes we find in figs . [ fig5][fig8 ] . some of the the amplitudes contributing to the br(@xmath3 ) presented above have been previously discussed in refs . we agree with ref . @xcite in that the main contribution arises from @xmath195 except for the values of parameters affected by the cancellation mentioned above . however we find smaller values for @xmath215 than the ones quoted in @xcite . ( 115,100 ) ( 0,0 ) ) vs. @xmath216 for @xmath217 gev for case b.,title="fig:",height=377 ] the contribution of @xmath218 to the branching ratio is negligible compared with @xmath187 , due to the fact that the matrix @xmath93 is replaced by @xmath92 with suppressed mixings . this holds even for the element @xmath219 . as we can see in appendix [ ap : a ] this element is determined by @xmath96 which is obtained in a similar way as @xmath95 for the matrix u. however , we observe that the main contribution to @xmath219 is suppressed by a factor @xmath220 with respect to the corresponding one in @xmath210 . fig . [ fig5 ] shows the impact of the cancellation in @xmath195 on the predictions of br(@xmath221 ) for the choice of parameters b ) . as we can deduce from eq . ( [ cancellation ] ) the value of the @xmath181 at which the cancellation in @xmath195 takes place depend on the values of the susy parameters . this determines the shape of the curves of constant br in figs . [ fig6 ] , [ fig7 ] and [ fig8 ] . since the main contribution comes from the chargino mediated diagram of fig . [ fig2 ] , we can expect that the @xmath222 increases with @xmath223 and decreases as @xmath162 grows . the increase of @xmath161 produces the same qualitative effect as the increase of @xmath162 , however it has a lower impact on br than the changes in @xmath162 . our results can be compared with the predictions of a model based in the mssm with a _ `` see saw '' _ mechanism presented in ref . @xcite , where the results for @xmath224 are of order @xmath225 for case a ) and between @xmath226 and @xmath225 for b ) and c ) . if we observe our predictions for case b ) on the left graph of fig . [ fig6 ] , we can see that the model predicts ratios of @xmath227 for values of @xmath228 and @xmath229 ranging from @xmath230 to @xmath231 gev ( independently of the value of @xmath185 ) . values in the range of @xmath232 to @xmath230 gev would lead to rates of order @xmath233 , still interesting for the next generation experiments @xcite . similarly , a window of @xmath234 gev , is crossed only by the @xmath235 line in case c ) and by lines below this value for case a ) . such values of @xmath82 are however excluded if one takes in account the constraint coming from the solar neutrinos mass scale . this is shown in figs . [ fig6 ] and [ fig7 ] , where the dashed line gives the upper limit on the @xmath82 as obtained from eq . ( [ eq : mneu2 ] ) for the requirement that @xmath236 . in these figures we can appreciate that the parameters that enhance the ratios also make @xmath237 larger . from eq.([eq : mneu2 ] ) we can observe that @xmath237 increases with @xmath223 ( through the dependence on @xmath238 ) and decreases as the @xmath51-term increases ( i.e with @xmath162 and @xmath161 ) . furthermore , since we have choosen the @xmath100-parameters to be related , @xmath237 is proportional to a combination of @xmath239 with no accidental cancellation amoung them . therefore we can not find suitable values for @xmath223 , @xmath162 and @xmath161 , such that we can find an overlaping of areas with @xmath240 ev and br(@xmath241 ) @xmath242 . ( 160,69 ) ( 1,0 ) ) in @xmath243@xmath244 plane for case b ( left ) and c ( right ) . the dash lines correspond to @xmath245 ev.,title="fig:",width=302 ] ( 81,0 ) ) in @xmath243@xmath244 plane for case b ( left ) and c ( right ) . the dash lines correspond to @xmath245 ev.,title="fig:",width=302 ] in fig . [ fig8 ] we consider @xmath246 and @xmath247 both for case b ) . this decreases ( increases ) the values of the @xmath100 s by about a factor of @xmath248 ( @xmath249 ) , respectively . by looking at the parameterization of the matrix u in appendix a we can infer that these changes in the @xmath100 s have not a decisive impact on the @xmath250 rates . the reason for this is that the dominant contributions to @xmath251 are determined by the matrix @xmath101 and its elements that depend explicitly on the @xmath65 s are much larger than the ones containing @xmath100 s ( at least for values of @xmath252 s leading to relevant ratios ) . however the position of the cancellation on @xmath253 depends on @xmath214 as we can see in eq . ( [ cancellation ] ) and it therefore determines the changes in the figures . the changes on the @xmath100 s have only a direct impact on the smaller contributions , such as the ones arising from the diagrams of fig . [ fig1 ] and on the @xmath218 , which size is controlled by the elements of @xmath97 , which , as we have said , are several orders of magnitude below the main contribution coming from @xmath101 . ( 115,100 ) ( 0,0 ) for case a.,title="fig:",height=377 ] the predictions for @xmath254 that we obtain with this model are of the same order as those for @xmath224 , the reason being that we have assumed the @xmath100 s to be of the same order of magnitude , as it is required to explain neutrino oscillations . the results in this case are independent of @xmath255 . if we consider values for @xmath256 and @xmath257 in the same range as in fig . [ fig6 ] we obtain similar curves . this result contrasts with the lfv results on the framework of the r - parity conserving mssm , where @xmath81 is typically suppressed by several orders of magnitude with respect to @xmath254 . in this case the hierarchy of yukawas couplings makes a distinction between the two processes . we conclude therefore that the values of the parameters of the brpv model that successfully explain the solar and atmospheric neutrino data @xcite , predict rates for @xmath221 , @xmath258 and @xmath259 , that are well below the current limits and well as those of planned experiments . ( 160,69 ) ( 1,0 ) , case b for @xmath260 ev ( left ) and @xmath261 ev ( right).,title="fig:",width=302 ] ( 81,0 ) , case b for @xmath260 ev ( left ) and @xmath261 ev ( right).,title="fig:",width=302 ] the difference on the value of the muon anomalous magnetic moment found in the bnl e821 measurement @xcite with respect to the sm prediction , which originally was considered to be @xmath262 is now reduced to @xmath263 after a theoretical error has been corrected @xcite . when the @xmath264 range is considered , the allowed values for contributions beyond the sm become , @xmath265 several studies @xcite indicate that the mssm extension of the sm can account for this discrepancy . parity is broken the susy particles are allowed to enter in the sm diagrams ( fig . [ fig1 ] ) and conversely the sm particles run in the susy loops ( figs . [ fig2 ] and [ fig3 ] ) . the contribution due to the r - parity violating operators to @xmath266 is obtained by subtracting from the amplitudes arising from fig . [ fig1 ] ( @xmath267 , @xmath268 ) and fig . [ fig2 ] ( @xmath269 , @xmath270 ) the contribution of the sm and the mssm , respectively ( which are obtained in the limit of vanishing @xmath58 s ) . the contribution from fig . [ fig3 ] ( @xmath271 ) is not present in the mssm . all these contributions are found to be small when the r - parity violating terms are associated with neutrino masses of experimental interest . in table [ tab : table1 ] we show the different contributions to @xmath272 for the selected values of @xmath273 and @xmath223 discussed in the section [ sec : parameters ] . the main contributions to @xmath272 arise basically from the mssm components of the diagrams in fig . [ fig2 ] ( @xmath274 , @xmath275 ) . the contribution from brpv operators just adds a small percentage to the total values arising from physics beyond the sm . the values that we show correspond to the maximum value obtained in the conditions for the @xmath65parameters described in section [ sec : parameters ] , when we allow @xmath276 to range from @xmath277 to @xmath278 ( the result is almost independent of the value of @xmath185 ) . .contributions to @xmath279 from the graphs in fig . [ fig1][fig3 ] ( in units of @xmath280 ) . cases a c , refer to the choice of parameters given in section [ sec : parameters ] . see section [ sec : expressions ] for details on the notation . [ cols="^,^,^,^,^,^,^,^",options="header " , ] we studied the lfv in one loop induced rare processes @xmath281 in susy models with bi linear r parity violation . in this context , the r parity violating interactions can explain the neutrino masses and mixings without adding new fields to the particle content of the mssm which represent an appealing alternative to the _ see saw _ mechanism . in this work we addressed the question of whether neutrino masses in the sub ev range can be compatible with rates for charged lfv processes of experimental interest . we have performed an exhaustive study of the interactions and of the susy parameters involved in the processes @xmath282 and contributing to the muon anomalous magnetic moment . for the case of rare decays we find the diagram mediated by the higgs sneutrino scalars to be the dominant . contributions arising from diagrams including lepton squark and lepton gauge bosons vertices , possible in this model , are very suppressed in the range of neutrino masses considered in this work . regarding @xmath283 , the additional contributions introduced by the r parity violating interactions modify in a small percentage the value obtained in the mssm limit . as in a previous analysis @xcite we find the br(@xmath284 ) to be very sensitive to the product @xmath285 . however the presence of a cancellation in the main amplitude contributing to this process ( which we have analyzed in detail through an accurate parameterization of the matrices @xmath286 ) , makes our contour plots sensitive to the values of @xmath180 and @xmath181 in most of our examples . the rate increases with @xmath223 as it is the case in the mssm with _ see saw _ mechanism . however as the one loop induced neutrino masses will also grow with @xmath223 , the requirement that @xmath287 is compatible with the solar neutrino data excludes the region in parameter space where the br(@xmath5 ) could be of experimental interest . on the other hand the rates for @xmath288 found in our study are of the same order as the ones for @xmath5 , therefore also out of the experimental range . unlike the situation in these models , the rates for @xmath288 found in our study are of the same order as the ones for @xmath5 , therefore out of the experimental range . to conclude , we must say , that the obtained results for the @xmath284 show us that if the brpv model is the explanation for both the solar and atmospheric neutrino ocillations , the predicted lfv will not be testable at psi @xcite or at prism @xcite . the correlations of the brpv parameters with the neutralino decays , as proposed in ref . @xcite , will remain the main test of the model . * acknowledgments * this project was supported in part by the tmr network of the ec under contract hprn - ct-2000 - 00148 . d. f. c. and m. e. g. acknowledge support from the ` fundao para a cincia e tecnologia ' under contracts praxis xxi/ bd/9416/96 and sfrh / bpd/5711/2001 respectively . we are grateful to p. nogueira for his help with the figures presented in this work . the chargino charged lepton mass matrix , in the basis of of eq . ( [ eq : basisch ] ) , takes the form : @xmath289\ ] ] where @xmath290 is the charged leptons mass matrix and @xmath291 is the usual mssm chargino mass matrix , @xmath292\ ] ] the sub - matrix @xmath293 is @xmath294\ ] ] and @xmath295 can be written as @xmath296 , where @xmath297 is defined as : @xmath298\ ] ] as the r parity breaking parameters are small compared with the susy scale , it is possible to have an approximate diagonalization of @xmath299 . this will be very useful in understanding the numerical results as we can have approximate analytical formulas . this approximate diagonalization is obtained by using the following parameterization , introduced in refs . @xcite , for @xmath300and @xmath301 @xmath302 \left [ \begin{array}{cc } 1-\frac{1}{2}\omega ^{\dagger } _ { l}\omega _ { l } & \omega ^{\dagger } _ { l}\\ -\omega _ { l } & 1-\frac{1}{2}\omega _ { l}\omega ^{\dagger } _ { l } \end{array}\right ] \\ \vb{24 } v^{\dagger } & = & \left [ \begin{array}{cc } 1-\frac{1}{2}\omega ^{\dagger } _ { r}\omega _ { r } & -\omega ^{\dagger } _ { r}\\ \omega _ { r } & 1-\frac{1}{2}\omega _ { r}\omega ^{\dagger } _ { r } \end{array}\right ] \left [ \begin{array}{cc } u_{r}^{\dagger } & 0\\ 0 & v^{\dagger } _ { r } \end{array}\right ] \end{aligned}\ ] ] where @xmath303 are the mssm rotation matrices , @xmath304 the matrices @xmath305 and @xmath306 are to be determined from the unitarity of @xmath93 and @xmath92 , and from the defining condition @xmath307\ ] ] in the literature @xcite , the matrices @xmath305 and @xmath306 were obtained in the approximation @xmath308 . however we discovered that this approximation was not good enough to explain our numerical results . so we re - derived the expressions for these matrices without neglecting @xmath309 . we get the following expressions for @xmath305 , @xmath310\ ] ] @xmath311 \cdot ( m_{\chi^\pm}^{-1})^{t}\end{aligned}\ ] ] where summation over @xmath312 is implied in each matrix element . the expression for @xmath101 coincides with the one found in the literature but the expression for @xmath97 it is different . for @xmath306 we found that instead of the relation @xcite @xmath313 they should satisfy , @xmath314 for a general form of the matrix @xmath315 it will be difficult to have an explicit form for @xmath306 . however for the case , that we consider , where the matrix @xmath315 is diagonal , we can obtain an analytical approximate expression for these matrices , @xmath316\ ] ] where @xmath317 \nn \\ \vb{20 } & & + \frac{\epsilon_j\lambda_i } { \mu^2 v_d}\left [ -\frac{g^2 v_u v_d}{2\ , \det(m_{\chi^\pm } ) } - \frac{m_i^2}{m_i^2-m_j^2 } \right ] - \frac{\lambda_i\lambda_j } { \det(m_{\chi^\pm})\ , \mu^2}\ , \frac{g^2 v_u}{v_d}\ , \frac{m_i^2+m_j^2}{m_i^2-m_j^2 } \\ \vb{24 } \label{eq : etar } \eta_{ij}^r&= & \frac{\epsilon_i\epsilon_j } { \mu^2}\ , \frac{m_i m_j}{m_i^2-m_j^2 } - \frac{\epsilon_i\lambda_j } { \mu^2 v_d } \frac{m_i m_j}{m_i^2-m_j^2 } -\frac{\epsilon_j\lambda_i } { \mu^2 v_d } \frac{m_i m_j}{m_i^2-m_j^2}\nn \\ \vb{20 } & & -\frac{\lambda_i\lambda_j}{\det(m_{\chi^\pm})\ , \mu^2}\ , \frac{2\ , g^2 v_u}{v_d}\ , \frac{m_i m_j}{m_i^2-m_j^2 } \end{aligned}\ ] ] and @xmath318 are the charged lepton physical masses . putting everything together we can find analytical expressions for the matrix @xmath93 that will be useful in explaining our results . we get , @xmath319 for further reference we give the approximate expression for @xmath320 . we get @xmath321 where we have assumed that the parameters are in the ranges described in section [ sec : parameters ] . the relevant part of the lagrangian , using four component spinor notation is , @xmath322 \nn \\ & & \vb{20 } + \overline{\chi_{i}^{-}}(v^{ccs}_{liax}p_{l } + v^{ccs}_{riax}p_{r})\chi_{a}^{-}s^{0}_{x } + \overline{\chi_{i}^{-}}(v^{ccp}_{liax}p_{l } + v^{ccp}_{riax}p_{r})\chi _ { a}^{-}p^{0}_{x } \nn \\ & & \vb{20 } + \overline{\chi _ { i}^{-}}\gamma ^{\mu } ( v^{ccz}_{lia}p_{l } + v^{ccz}_{ria}p_{r})\chi _ { a}^{-}z_{\mu } ^{0 } \end{aligned}\ ] ] the definition of these couplings is given in the following sections . these definitions extend those of ref . @xcite which conventions we follow . \vb{22 } v^{cns}_{riax}&\hskip -4mm=\hskip -4 mm & \eta _ { i}\!\left[\vb{16 } gr^{s^{\pm}}_{x1}\left(\frac{1}{\sqrt{2}}n_{a2}u_{i2}-n_{a3}u_{i1}\right ) + gr^{s^{\pm}}_{x2+\alpha } \left(\frac{1}{\sqrt{2 } } u_{i2+\alpha } n_{a2 } - n_{a4+\alpha}u_{i1}\right ) \right . \nn \\ & & \left . \vb{16 } + \!\frac{g'}{\sqrt{2}}\left(r^{s^{\pm } } _ { x1}n_{a1}u_{i2 } \!+\ ! r^{s^{\pm}}_{x2+\alpha}u_{i2+\alpha } n_{a1}\right ) \!\!+\ ! \left(u_{i2}n_{a4+\alpha } \!-\ ! n_{a3}u_{i2+\alpha}\right ) h^{\alpha \beta } _ { e}r^{s^{\pm } } _ { x5+\beta } \right]\end{aligned}\ ] ] where the indices have the following ranges : @xmath324 , @xmath325 , @xmath326 and @xmath327 ( @xmath328 ) are the signs of the neutralinos ( respectively charginos ) as they are obtained from the numerical evaluation of the eigenvalues @xcite . @xmath329 \\ \vb{20 } v^{ccs}_{riax}&=&v^{ccs*}_{laix } \end{aligned}\ ] ] @xmath330 \\ \vb{20 } v^{ccp}_{riax}&=&v^{ccp*}_{laix } \end{aligned}\ ] ] @xmath332 \\ \vb{20 } v^{cnw}_{ria}&=&g\left [ \frac{1}{\sqrt{2}}n_{a4}v^{*}_{i2 } -n_{a2}v^{*}_{i1}\right]\end{aligned}\ ] ] @xmath334 \\ \vb{20 } v^{ccz}_{ria}&=&\frac{g}{\cos \theta_{w}}\left[v^{*}_{i1}v_{a1 } + \frac{1}{2}v^{*}_{i2}v_{a2 } -\sin ^{2}\theta _ { w } \delta _ { ia}\right ] \end{aligned}\ ] ] @xmath335 \\ \vb{20 } v^{cd\widetilde{u}}_{riax}&=&\eta _ { i}\left [ u_{i2}r^{\widetilde{u}}_{x\alpha } h^{\alpha \beta}_{d}r^{d}_{ra\beta } \right]\end{aligned}\ ] ] @xmath336 \\ \vb{20 } v^{cu\widetilde{d}}_{riax}&=&v_{i2}r^{\widetilde{d}}_{x\alpha } h^{\alpha \beta}_{u}r^{u*}_{ra\beta } \end{aligned}\ ] ] we collect here the various amplitudes corresponding to the diagrams of figs . [ fig1][fig3 ] . in these amplitudes the mass of the outgoing fermion was neglected . we give only the amplitudes @xmath218 because the @xmath187 can be obtained from these with the substitution rule @xmath337 99 y. fukuda et al . , super - kamiokande collaboration , _ phys . _ * b433 * ( 1998 ) 9 ; _ phys . lett . _ * b436 * ( 1998 ) 33 ; _ phys * 81 * ( 1998 ) 1562 . m. apollonio et al . , chooz collaboration , _ phys . * b420 * ( 1998 ) 397 . b. mukhopadhyaya and a. raychaudhuri _ phys . * d42 * ( 1990 ) 3215 ; a. belyaev _ et al . _ [ kaon physics working group collaboration ] , hep - ph/0107046 , hep - ph/0008276 . petcov , _ sov . * 25 * ( 1977 ) 340 ; t.p . cheng and l.f . li , _ phys . _ * d16 * ( 1977 ) 1425 ; b. lee and r. schrock , _ phys . rev . _ * d16 * ( 1977 ) 1444 ; j.d . bjorken et al . , _ phys . _ * d16 * ( 1977 ) 1474 ; s. m. bilenkii , s. t. petcov and b. pontecorvo , phys . b * 67 * ( 1977 ) 309 ; s. m. bilenkii and s. t. petcov , rev . * 59 * ( 1987 ) 671 [ erratum - ibid . * 61 * ( 1987 ) 169 ] . a. ilakovac and a. pilaftsis , nucl . b * 437 * ( 1995 ) 491 j. g. korner , a. pilaftsis and k. schilcher , phys . b * 300 * ( 1993 ) 381 r. barbieri and l.j . hall , _ phys . lett . _ * b338 * ( 1994 ) 212 ; r. barbieri et al . , phys . _ * b445 * ( 1995 ) 219 ; s. dimopoulos and d. sutter , _ nucl . phys . _ * b452 * ( 1995 ) 496 ; nima arkani - hamed , hsin - chia cheng and l.j . hall , _ phys . _ * d53 * ( 1996 ) 413 ; p. ciafaloni , a. romanino and a. strumia , _ nucl . phys . _ * b458 * ( 1996 ) 3 ; m.e . gmez and h. goldberg , rev . _ * d53 * ( 1996 ) 5244 b. de carlos , j.a . casas and j.m . moreno , _ phys . rev . _ * d53 * ( 1996 ) 6398 ; j. hisano , t. moroi , k. tobe and m. yamaguchi , _ phys . lett . _ * b391 * ( 1997 ) 341 , erratum - ibid . * b397 * ( 1997 ) 357 ; s.f . king and m. oliveira , _ phys . * d60 * ( 1999 ) 035003 . d. f. carvalho , m. e. gomez and s. khalil , jhep * 0107 * , 001 ( 2001 ) j. hisano , t. moroi , k. tobe and m. yamaguchi , _ phys . rev . _ * d53 * ( 1996 ) 2442 ; m. gmez , g. leontaris , s. lola and j. vergados , _ phys . rev . _ * d59 * ( 1999 ) 116009 j. ellis _ _ * c14 * ( 2000 ) 319 w. buchmuller , d. delepine and f. vissani , _ phys . _ * b459 * ( 1999 ) 171 ; w. buchmuller , d. delepine and l. t. handoko , _ nucl . phys . _ * b576 * ( 2000 ) 445 ; q. shafi and z. tavartkiladze _ phys . * b473 * ( 2000 ) 145 ; j. l. feng , y. nir and y. shadmi _ phys . rev . _ * d61 * ( 2000 ) 113005 . j. a. casas and a. ibarra , nucl . b * 618 * ( 2001 ) 171 ; s. lavignac , i. masina and c. a. savoy , phys . b * 520 * ( 2001 ) 269 ; a. kageyama , s. kaneko , n. shimoyama and m. tanimoto , hep - ph/0110283 . f. de campos , m.a . garca - jareo , a.s . joshipura , j. rosiek , and j. w. f. valle , ; t. banks , y. grossman , e. nardi , and y. nir , ; a. s. joshipura and m.nowakowski , ; h. p. nilles and n. polonsky , ; b. de carlos , p. l. white , ; s. roy and b. mukhopadhyaya , m. hirsch , w. porod , m. a. daz , j. w. f. valle and j. c. romo , proceedings of the corfu summer institute , corfu 2001 , ; m. hirsch , w. porod , m. a. daz , j. w. f. valle and j. c. romo , work in preparation . daz , j.c . romo and j. w. f. valle , nucl . * b524 * , 23 ( 1998 ) . for reviews see , e.g. j. w. f. valle , `` super - gravity unification with bilinear r - parity violation , '' proceedings of 6th international symposium on particles , strings and cosmology ( pascos 98 ) , boston , ma , 22 - 27 mar 1998 [ hep - ph/9808292 ] ed . nath ; m.a . daz , talk given at _ international europhysics conference on high energy physics _ , jerusalem , israel , 19 - 26 aug 1997 , a. masiero and j. w. f. valle , ; j.c . romo , c.a . santos , j. w. f. valle , ; j. c. romo , a. ioannisian and j. w. valle , phys . rev . * d55 * , 427 ( 1997 ) ; for a review see j.c . romo , proceedings of _ international workshop on physics beyond the standard model : from theory to experiment _ , valencia , 13 - 17 oct 1997 , ed . v. berezinsy , g. raffelt , j. w. f. valle , nuclear physics b , proceedings supplements , 2000,vol . 81 , issn 0920 - 5632 . for papers discussing r - parity violation phenomenology see , e.g. f. de campos , o. j. eboli , m. a. garcia - jareno and j. w. valle , nucl . * b546 * , 33 ( 1999 ) ; m. nowakowski and a. pilaftsis , nucl . * b461 * , 19 ( 1996 ) ; l. navarro , w. porod and j. w. valle , phys . * b459 * , 615 ( 1999 ) ; m. a. diaz , d. a. restrepo and j. w. valle , nucl . b * 583 * ( 2000 ) 182 ; b. allanach _ et al . _ , hep - ph/9906224 ; a. faessler , s. kovalenko and f. simkovic , phys . rev . * d58 * , 055004 ( 1998 ) ; s. davidson , m. losada and n. rius , nucl . b * 587 * ( 2000 ) 118 ; m. bisset , o. c. kong , c. macesanu and l. h. orr , phys . d * 62 * ( 2000 ) 035001 ; m. a. diaz , e. torrente - lujan and j. w. f. valle , nucl . * b551 * , 78 ( 1999 ) . the homepage of the prism project : http://www-prism.kek.jp/ ; y. kuno , _ lepton flavor violation experiments at kek / jaeri joint project of high intensity proton machine _ , in proceedings of workshop of `` lownu / noon 2000 '' , tokyo , december 4 - 8 , 2000 . for recent work on the topic see for example : u. chattopadhyay and p. nath , ; t. moroi , ; m. carena , m. giudice and c.e.m . wagner , ; e. gabrielli and u. sarid , ; t. ibrahim and p. nath , ; ; k.t . mahanthappa and s. oh , ; u. chattopadhyay , d. k. ghosh and s. roy , . m. knecht and a. nyffeler , hep - ph/0111058 . m. knecht , a. nyffeler , m. perrottet and e. de rafael , hep - ph/0111059 . m. hayakawa and t. kinoshita , hep - ph/0112102 . i. blokland , a. czarnecki and k. melnikov , hep - ph/0112117 . j. bijnens , e. pallante and j. prades , hep - ph/0112255 .

in the matrix approach to graph transformation we represent _ simple _ digraphs and rules with boolean matrices and vectors , and the rewriting is expressed using boolean operations only . in previous works , we developed analysis techniques enabling the study of the applicability of rule sequences , their independence , stated reachability and the minimal digraph able to fire a sequence . see for a comprehensive introduction . in , graph constraints and application conditions ( so - called _ restrictions _ ) have been studied in detail . in the present contribution we tackle the problem of translating post - conditions into pre - conditions and vice versa . moreover , we shall see that application conditions can be moved along productions inside a sequence ( _ restriction delocalization _ ) . as a practical - theoretical application we show how application conditions allow us to perform multidigraph rewriting ( as opposed to simple digraph rewriting ) using matrix graph grammars . * keywords : * matrix graph grammars , graph dynamics , graph transformation , restrictions , application conditions , preconditions , postconditions , graph constraints .

0912.2160

graph transformation @xcite is becoming increasingly popular in order to describe system behavior due to its graphical , declarative and formal nature . for example , it has been used to describe the operational semantics of domain specific visual languages ( dsvls , @xcite ) , taking the advantage that it is possible to use the concrete syntax of the dsvl in the rules which then become more intuitive to the designer . the main formalization of graph transformation is the so - called algebraic approach @xcite , which uses category theory in order to express the rewriting step . prominent examples of this approach are the double @xcite and single @xcite pushout ( dpo and spo ) which have developed interesting analysis techniques , for example to check sequential and parallel independence between pairs of rules @xcite or the calculation of critical pairs @xcite . frequently , graph transformation rules are equipped with _ application conditions _ ( acs ) @xcite , stating extra ( in addition to the left hand side ) positive and negative conditions that the host graph should satisfy for the rule to be applicable . the algebraic approach has proposed a kind of acs with predefined diagrams ( i.e. graphs and morphisms making the condition ) and quantifiers regarding the existence or not of matchings of the different graphs of the constraint in the host graph @xcite . most analysis techniques for plain rules ( without acs ) have to be adapted then for rules with acs ( see e.g. @xcite for critical pairs with negative acs ) . moreover , different adaptations may be needed for different kinds of acs . thus , a uniform approach to analyze rules with arbitrary acs would be very useful . in previous works @xcite we developed a framework ( matrix graph grammars , mggs ) for the transformation of simple digraphs . simple digraphs and their transformation rules can be represented using boolean matrices and vectors . thus , the rewriting can be expressed using boolean operators only . one important point is that , as a difference from other approaches , we explicitly represent the rule dynamics ( addition and deletion of elements ) instead of only the static parts ( rule pre and postconditions ) . this point of view enables new analysis techniques , such as for example checking independence of a sequence of arbitrary length and a permutation of it , or to obtain the smallest graph able to fire a sequence . on the theoretical side , our formalization of graph transformation introduces concepts from many branches of mathematics like boolean algebra , group theory , functional analysis , tensor algebra and logics @xcite . this wealth of available mathematical results opens the door to new analysis methods not developed so far , like sequential independence and explicit parallelism not limited to pairs of sequences , applicability , graph congruence and reachability . on the practical side , the implementations of our analysis techniques , being based on boolean algebra manipulations , are expected to have a good performance . in mggs we do not only consider the elements that must be present in order to apply a production ( left hand side , lhs , also known as _ certainty part _ ) but also those elements that potentially prevent its application ( also known as _ nihil _ or _ nihilation part _ ) . refer to @xcite in which , besides this , application conditions and graph constraints are studied for the mgg approach . the present contribution is a continuation of @xcite where a comparison with related work can also be found . we shall tackle pre and postconditions , their transformation , the sequential version of these results and multidigraph rewriting . * paper organization*. section [ sec : mggs ] gives an overview of matrix graph grammars . section [ sec : previous ] revises application conditions as studied in @xcite . postconditions and their equivalence to certain sequences are addressed in sec . [ sec : postconditions ] . section [ sec : movingconditions ] tackles the transformation of preconditions into postconditions . the converse , more natural from a practical point of view , is also addressed . the transformation of restrictions is generalized in sec . [ sec : delocalization ] in which _ delocalization _ how to move application conditions from one production to another inside the same sequence is also studied together with _ variable nodes_. as an application of restrictions to mggs , sec . [ sec : fromsimpledigraphstomultidigraphs ] shows how to make mgg deal with multidigraphs instead of just simple digraphs without major modifications to the theory . the paper ends in sec . [ sec : conclusions ] with some conclusions , further research remarks and acknowledgements . we work with * simple digraphs * which we represent as @xmath0 , where @xmath1 is a boolean matrix for edges ( the graph _ adjacency _ matrix ) and @xmath2 a boolean vector for vertices or nodes . in the corresponding position of the vector . ] the left of fig . [ fig : example_graph ] shows a graph representing a production system made up of a machine ( controlled by an operator ) which consumes and produces pieces through conveyors . self loops in operators and machines indicate that they are busy . well - formedness of graphs ( i.e. absence of dangling edges ) can be checked by verifying the identity @xmath3 , where @xmath4 is the boolean matrix product , @xmath5 is the transpose of the matrix @xmath1 , @xmath6 is the negation of the nodes vector @xmath2 , and @xmath7 is an operation ( a norm , actually ) that results in the * or * of all the components of the vector . we call this property * compatibility * ( refer to @xcite ) . note that @xmath8 results in a vector that contains a 1 in position @xmath9 when there is an outgoing edge from node @xmath9 to a non - existing node . a similar expression with the transpose of @xmath1 is used to check for incoming edges . a * type * is assigned to each node in @xmath10 by a function from the set of nodes @xmath11 to a set of types @xmath12 , @xmath13 . sets will be represented by @xmath14 . in fig . [ fig : example_graph ] types are represented as an extra column in the matrices , where the numbers before the colon distinguish elements of the same type . it is just a visual aid . for edges we use the types of their source and target nodes . a * typed simple digraph * is @xmath15 . from now on we shall assume typed graphs and shall drop the @xmath12 subindex . a * production * or grammar rule @xmath16 is a morphism of typed simple digraphs , which is defined as a mapping that transforms @xmath17 in @xmath18 with the restriction that the type of the image must be equal to the type of the source element .. ] more explicitly , @xmath19 being @xmath20 and @xmath21 partial injective mappings @xmath22 , @xmath23 such that @xmath24 and @xmath25 , where @xmath26 stands for domain , @xmath27 for edges and @xmath2 for vertices . a production @xmath16 is * statically represented * as @xmath28 . the matrices and vectors of these graphs are arranged so that the elements identified by morphism @xmath29 match ( this is called completion , see below ) . alternatively , a production adds and deletes nodes and edges , therefore they can be * dynamically represented * by encoding the rule s lhs together with matrices and vectors representing the addition and deletion of edges and nodes : for `` erase '' and @xmath30 for `` restock '' . ] @xmath31 , where @xmath32 contains the types of the new nodes , @xmath33 and @xmath34 are the deletion boolean matrix and vector , @xmath35 and @xmath36 are the addition boolean matrix and vector . they have a 1 in the position where the element is to be deleted or added , respectively . the output of rule @xmath29 is calculated by the boolean formula @xmath37 , which applies both to nodes and edges . is usually omitted in formulae , so @xmath38 with precedence of @xmath39 over @xmath40 . ] * example*.@xmath41figure [ fig : example_rule ] shows a rule and its associated matrices . the rule models the consumption of a piece ( pack ) by a machine ( mach ) input via the conveyor ( conv ) . there is an operator ( oper ) managing the machine . compatibility of the resulting graph must be ensured , thus the rule can not be applied if the machine is already busy , as it would end up with two self loops which is not allowed in a simple digraph . this restriction of simple digraphs can be useful in this kind of situations and acts like a built - in negative application condition . later we will see that the _ nihilation matrix _ takes care of this restriction . @xmath42 in order to operate with the matrix representation of graphs of different sizes , an operation called * completion * adds extra rows and columns with zeros to matrices and vectors , and rearranges rows and columns so that the identified edges and nodes of the two graphs match . for example , in fig . [ fig : example_rule ] , if we need to operate @xmath43 and @xmath44 , completion adds a fourth @xmath45-row and fourth @xmath45-column to @xmath44 . no further modification is needed because the rest of the elements have the right types and are placed properly . with the purpose of considering the elements in the host graph that disable a rule application , we extend the notation for rules with a new simple digraph @xmath46 , which specifies the two kinds of forbidden edges : those incident to nodes which are going to be erased and any edge added by the rule ( which can not be added twice , since we are dealing with simple digraphs ) . @xmath46 has non - zero elements in positions corresponding to newly added edges , and to non - deleted edges incident to deleted nodes . matrices are derived in the following order : @xmath47 . thus , a rule is _ statically _ determined by its lhs and rhs @xmath48 , from which it is possible to give a _ dynamic _ definition @xmath49 , with @xmath50 and @xmath51 , to end up with a full specification including its _ environmental _ behavior @xmath52 . no extra effort is needed from the grammar designer because @xmath46 can be automatically calculated : @xmath53 , with @xmath54 . denotes the tensor or kronecker product , which sums up the covariant and contravariant parts and multiplies every element of the first vector by the whole second vector . ] the evolution of the nihilation matrix ( what elements can not appear in the rhs ) call it @xmath55 is given by the inverse of the production : @xmath56 . see @xcite for more details . inspired by the dirac or bra - ket notation @xcite we split the static part ( initial state , @xmath17 ) from the dynamics ( element addition and deletion , @xmath29 ) : @xmath57 . operators ( those to the right side of the bra - ket ) can be moved to the _ bra _ ( left hand side ) by using their adjoints . * matching * is the operation of identifying the lhs of a rule inside a host graph . given a rule @xmath16 and a simple digraph @xmath58 , any total injective morphism @xmath59 is a match for @xmath29 in @xmath58 , thus it is one of the ways of _ completing _ @xmath17 in @xmath58 . besides , we shall consider the elements that must not be present . given the grammar rule @xmath16 and the graph @xmath60 , @xmath61 is called a * direct derivation * with @xmath62 and result @xmath63 if the following conditions are satisfied : 1 . there exist total injective morphisms @xmath64 and @xmath65 with @xmath66 , @xmath67 . the match @xmath68 induces a completion of @xmath17 in @xmath58 . matrices @xmath69 and @xmath30 are then completed in the same way to yield @xmath70 and @xmath71 . the output graph is calculated as @xmath72 . the negation when applied to graphs alone ( not specifying the nodes ) e.g. @xmath73 in the first condition above will be carried out just on edges . notice that in particular the first condition above guarantees that @xmath17 and @xmath46 will be applied to the same nodes in the host graph @xmath58 . in direct derivations _ dangling edges _ can occur because the nihilation matrix only considers edges incident to nodes appearing in the rule s lhs and not in the whole host graph . in mgg an operator @xmath74 takes care of dangling edges which are deleted by adding a preproduction ( known as @xmath75production ) before the original rule . refer to @xcite . thus , rule @xmath29 is transformed into the sequence @xmath76 , where @xmath77 deletes the dangling edges and @xmath29 remains unaltered . there are occasions in which two or more productions should be matched to the same nodes . this is achieved with the * marking operator * @xmath78 introduced in chap . 6 in @xcite . a grammar rule and its associated @xmath79-production is one example and we shall find more in future sections . in @xcite some analysis techniques for mggs have been developed which we shall skim through . one important feature of mgg is that sequences of rules can be analyzed independently to some extent of any host graph . a rule * sequence * is represented by @xmath80 where application is from right to left , i.e. @xmath81 is applied first . for its analysis , the sequence is completed by identifying the nodes across rules which are assumed to be mapped to the same node in the host graph . once the sequence is completed , sequence * coherence * @xcite allows us to know if , for the given identification , the sequence is potentially applicable , i.e. if no rule disturbs the application of those following it . the formula for coherence results in a matrix and a vector ( which can be interpreted as a graph ) with the problematic elements . if the sequence is coherent , both should be zero ; if not , they contain the problematic elements . a coherent sequence is * compatible * if its application produces a simple digraph . that is , no dangling edges are produced in intermediate steps . given a completed sequence , the * minimal initial digraph * ( mid ) is the smallest graph that permits the application of such sequence . conversely , the * negative initial digraph * ( nid ) contains all elements that should not be present in the host graph for the sequence to be applicable . in this way , the nid is a graph that should be found in @xmath82 for the sequence to be applicable ( i.e. none of its edges can be found in @xmath58 ) . see sec . 6 in @xcite or chaps . 5 and 6 in @xcite . other concepts we developed aim at checking * sequential independence * ( same result ) between a sequence and a permutation of it . * g - congruence * detects if two sequences , one permutation of the other , have the same mid and nid . it returns two matrices and two vectors , representing two graphs which are the differences between the mids and nids of each sequence , respectively . thus if zero , the sequences have the same mid and nid . two coherent and compatible completed sequences that are g - congruent are sequentially independent . see sec . 7 in @xcite or chap . 7 in @xcite in this section we shall brush up on application conditions ( acs ) as introduced for mgg in @xcite with non - fixed diagrams and quantifiers . for the quantification , a full - fledged monadic second order logic formula is used . one of the contributions in @xcite is that a rule with an ac can be transformed into ( sequences of ) plain rules by adding the positive information to the left hand side of the production and the negative to the nihilation matrix . a * diagram * @xmath83 is a set of simple digraphs @xmath84 and a set of partial injective morphisms @xmath85 with @xmath86 . the diagram @xmath83 is well defined if every cycle of morphisms commute . @xmath87 is a * graph constraint * where @xmath83 is a well defined diagram and @xmath88 a sentence with variables in @xmath89 and predicates @xmath90 and @xmath55 . see eqs . ( [ eq : p ] ) and ( [ eq : q ] ) . formulae are restricted to have no free variables except for the default second argument of predicates @xmath90 and @xmath55 , which is the host graph @xmath58 in which we evaluate the gc . gc formulae are made up of expressions about graph inclusions . the predicates @xmath90 and @xmath55 are given by : @xmath91 \\ \label{eq : q } q(x^1 , x^2 ) & = \exists e [ f(e , x^1 ) \wedge f(e , x^2)],\end{aligned}\ ] ] where predicate @xmath92 states that element @xmath93 ( a node or an edge ) is in graph @xmath94 . predicate @xmath95 means that graph @xmath96 is included in @xmath97 . predicate @xmath98 asserts that there is a partial morphism between @xmath96 and @xmath97 , which is defined on at least one edge ( @xmath69 ranges over all edges ) . the notation ( syntax ) will be simplified by making the host graph @xmath58 the default second argument for predicates @xmath90 and @xmath55 . besides , it will be assumed that by default total morphisms are demanded : unless otherwise stated predicate @xmath90 is assumed . we take the convention that negations in abbreviations apply to the predicate ( e.g. @xmath99 \equiv \exists a [ \overline{p}(a , g)]$ ] ) and not the negation of the graph s adjacency matrix . * example*.@xmath41the gc in fig . [ fig : diagexample ] is satisfied if for every @xmath100 in @xmath58 it is possible to find a related @xmath101 in @xmath58 , i.e. its associated formula is @xmath102 $ ] , equivalent by definition to @xmath103 $ ] . nodes and edges in @xmath100 and @xmath101 are related through morphism @xmath104 in which the image of the machine in @xmath100 is the machine in @xmath101 . to enhance readability , each graph in the diagram has been marked with the quantifier given in the formula . the gc in fig . [ fig : diagexample ] expresses that each machine should have an output conveyor . @xmath42 given the rule @xmath16 with nihilation matrix @xmath46 , an * application condition * ac ( over the free variable @xmath58 ) is a gc satisfying : 1 . @xmath105 such that @xmath106 and @xmath107 . @xmath108 such that @xmath109 is the only free variable . 3 . @xmath88 must demand the existence of @xmath17 in @xmath58 and the existence of @xmath46 in @xmath73 . for simplicity , we usually do not explicitly show the condition 3 in the formulae of acs , nor the nihilation matrix @xmath46 in the diagram which are existentially quantified before any other graph of the ac . notice that the rule s lhs and its nihilation matrix can be interpreted as the minimal ac a rule can have . for technical reasons addressed in sec . [ sec : movingconditions ] ( related to converting pre into postconditions ) we assume that morphisms @xmath110 in the diagram do not have codomain @xmath17 or @xmath46 . this is easily solved as we may always use their inverses due to @xmath110 s injectiveness . it is possible to embed arbitrary acs into rules by including the positive and negative conditions in @xmath17 and @xmath46 , respectively . intuitively : `` mgg + ac = mgg '' and `` mgg + gc = mgg '' . in @xcite two basic operations are introduced : * closure * @xmath111 that transforms universal into existential quantifiers , and * decomposition * @xmath112 that transforms partial morphisms into total morphisms . notice that a match is an existentially quantified total morphism . it is proved in @xcite that any ac can be embedded into its corresponding direct derivation . this is achieved by transforming the ac into some sequences of productions . there are four basic types of acs / gcs . let @xmath113 be a graph constraint with diagram @xmath114 and consider the associated production @xmath115 . the case @xmath116 $ ] is just the matching of @xmath117 in the host graph @xmath58 . it is equivalent to the sequence @xmath118 , where @xmath119 has @xmath117 as lhs and rhs , so it simply demands its existence in @xmath58 . we introduce the operator @xmath120 that replaces @xmath29 by @xmath121 and leaves the diagram and the formula unaltered . if the formula @xmath122 $ ] is considered , we can reduce it to a sequence of matchings via the closure operator @xmath123 whose result is : @xmath124,\end{aligned}\ ] ] with @xmath125 , @xmath126 , is an isomorphism@xmath127 . ] @xmath128 and @xmath129 . is a maximal non - empty partial morphism with @xmath130 . ] this is equivalent to the sequence @xmath131 . if the application condition has formula @xmath132 $ ] , we can proceed by defining the composition operator @xmath133 with action : @xmath134,\end{aligned}\ ] ] where @xmath135 contains a single edge of @xmath117 and @xmath136 is the number of edges of @xmath117 . this is equivalent to the set of sequences @xmath137 . less evident are formulas of the form @xmath138 = \forall a [ \overline{a } ] $ ] . fortunately , operators @xmath111 and @xmath112 commute when composed so we can get along with the operator @xmath139 . the image of @xmath140 on such acs are given by : @xmath141.\end{aligned}\ ] ] an ac is said to be * coherent * if it is not a contradiction ( false in all scenarios ) , * compatible * if , together with the rule s actions , produces a simple digraph , and * consistent * if @xmath142 host graph such that @xmath143 satisfies @xmath144 $ ] , written @xmath145 $ ] , if and only if @xmath146 $ ] , being @xmath147 is a total morphism@xmath148 . also , @xmath58 satisfies @xmath149 $ ] , written @xmath150 $ ] , if and only if @xmath151 $ ] . usually we shall abuse of the notation and write @xmath152 instead . for more details , please refer to @xcite . ] to which the production is applicable . as acs can be transformed into equivalent ( sets of ) sequences , it is proved in @xcite that coherence and compatibility of an ac is equivalent to coherence and compatibility of the associated ( set of ) sequence(s ) , respectively . also , an ac is consistent if and only if its equivalent ( set of ) sequence(s ) is applicable . besides , all results and analysis techniques developed for mgg can be applied to sequences with acs . some examples follow : * as a sequence is applicable if and only if it is coherent and compatible ( see sec 6.4 in @xcite ) then an ac is consistent if and only if it is coherent and compatible . * sequential independence allows us to delay or advance the constraints inside a sequence . as long as the productions do not modify the elements of the constraints , this is transformation of preconditions into postconditions . more on sec . [ sec : movingconditions ] . * initial digraph calculation solves the problem of finding a host graph that satisfies a given ac / gc . there are some limitations , though . for example it is necessary to limit the maximum number of nodes when dealing with universal quantifiers . this has no impact in some cases , for example when non - uniform mgg submodels are considered ( see _ nodeless mgg _ in @xcite ) . * _ graph congruence _ characterizes sequences with the same initial digraph . therefore it can be used to study when two gcs / acs are equivalent for all morphisms or for some of them . summarizing , there are two basic results in @xcite . first , it is always possible to embed an application condition into the lhs of the production or derivation . the left hand side @xmath17 of a production receives elements that must be found @xmath153 and @xmath46 those whose presence is forbidden @xmath154 . second , it is always possible to find a sequence or a set of sequences of plain productions whose behavior is equivalent to that of the production plus the application condition . in this section we shall introduce postconditions and state some basic facts about them analogous to those for preconditions . we shall enlarge the notation by appending a left arrow on top of the conditions to indicate that they are preconditions and an upper right arrow for postconditions . examples are @xmath155 for a precondition and @xmath156 for a postcondition . if it is clear from the context , arrows will be omitted . [ def : prepostcondition ] an application condition set on the lhs of a production is known as a _ precondition_. if it is set on the rhs then it is known as a _ postcondition_. operators @xmath157 and @xmath158 are defined similarly for postconditions . the following proposition establishes an equivalence between the basic formulae ( match , decomposition , closure and negative application condition ) and certain sequences of productions . [ prop : postconds ] let @xmath159 be a postcondition . then we can obtain a set of equivalent sequences to given basic formulae as follows : @xmath160 \quad \longmapsto \quad t_a\left(p\right ) = id_a ; p \\ \label{eq : closure } \textrm{(closure ) } \quad \mathfrak{f } & = \nexists a [ \overline{a } ] \quad \longmapsto \quad \widecheck{t}_a\left(p\right ) = id_{a^1 } ; \ldots ; id_{a^m } ; p \\ \label{eq : decomp } \textrm{(decomposition ) } \quad \mathfrak{f } & = \exists a [ \overline{a } ] \quad \longmapsto \quad \widehat{t}_a\left(p\right ) = \left\ { \overline{id}_{a_i } ; p \right\}_{i = 1 , \ldots , n } \\ \label{eq : postnac } \textrm{(nac ) } \quad \mathfrak{f } & = \nexists a [ a ] \quad \longmapsto \quad \widetilde{t}_a \left ( p \right ) = \left\ { \overline{id}_{a^1_{i_1 } } ; \ldots ; \overline{id}_{a^m_{i_m } } ; p \right\}_{i_j \in \{1 , \ldots , n\ } , j \in \{1 , \ldots , m\ } } \end{aligned}\ ] ] where @xmath93 is the number of potential matches of @xmath117 in the image of the host graph , @xmath136 is the number of edges in @xmath117 and @xmath161 asks for the existence of @xmath117 in the complement of the image of the host graph . _ + @xmath41 for the first case ( match ) , the ac states that an additional graph @xmath117 has to be found in the image of the host graph . this is easily achieved by applying @xmath119 to the image of @xmath17 , i.e. by considering @xmath162 . the elements in @xmath117 are related to those in @xmath18 according to the identifications in a morphism @xmath163 that has to be given in the diagram of the postcondition . in the four cases considered in the proposition we can move from composition to concatenation by means of the marking operator @xmath78 . recall that @xmath78 guarantees that the identifications in @xmath163 are preserved . the second case ( closure ) is very similar . we have to verify all potential appearances of @xmath117 in the image of the host graph because @xmath164 = \forall a [ a]$ ] . we proceed as in the first case but this time with a finite number of compositions : @xmath165 . for decomposition , @xmath117 is not found in the host graph if for some matching there is at least one missing edge . it is thus similar to matching but for a single edge . the way to proceed is to consider the set of sequences that appear in eq . ( [ eq : decomp ] ) . negative application conditions ( nacs ) are the composition of eqs . ( [ eq : closure ] ) and ( [ eq : decomp ] ) . @xmath42 one of the main points of the techniques available for preconditions is to analyze rules with acs by translating them into sequences of flat rules , and then analyzing the sequences of flat rules instead . [ th : reductionpost ] any well - defined postcondition can be reduced to the study of the corresponding set of sequences . _ + @xmath41the proof follows that of th . 4.1 in @xcite and is included here for completeness sake . let the depth of a graph for a fixed node @xmath166 be the maximum over the shortest path ( to avoid cycles ) starting in any node different from @xmath166 and ending in @xmath166 . the depth of a graph is the maximum depth for all its nodes . notice that the depth is @xmath167 if and only if @xmath168 in the diagram are unrelated . we shall apply induction on the depth of the ac . a diagram @xmath83 is a graph where nodes are digraphs @xmath135 and edges are morphisms @xmath169 . there are @xmath170 possibilities for depth @xmath167 in a ac made up of a single element @xmath117 , summarized in table [ tab : possibilitiessinglecase ] . ( 1 * ) & @xmath171 $ ] & ( 5 * ) & @xmath172 $ ] & ( 9 * ) & @xmath173 $ ] & ( 13 * ) & @xmath174 $ ] + ( 2 * ) & @xmath175 $ ] & ( 6 * ) & @xmath176 $ ] & ( 10 * ) & @xmath177 $ ] & ( 14 * ) & @xmath178 $ ] + ( 3 * ) & @xmath179 $ ] & ( 7 * ) & @xmath180 $ ] & ( 11 * ) & @xmath181 $ ] & ( 15 * ) & @xmath182 $ ] + ( 4 * ) & @xmath183 $ ] & ( 8 * ) & @xmath184 $ ] & ( 12 * ) & @xmath185 $ ] & ( 16 * ) & @xmath186 $ ] + elements in the same row for each pair of columns are related using equalities @xmath187 = \forall a[\overline{a}]$ ] and @xmath188 = \exists a[\overline{a}]$ ] , so it is possible to reduce the study to cases ( 1 * ) ( 4 * ) and ( 9 * ) identities @xmath189 and @xmath190 reduce ( 9 * ) ( 12 * ) to formulae ( 1 * ) ( 4 * ) : @xmath191 = \exists a\left[p(a , \overline{g})\right ] & , \ ; \exists a [ q(a ) ] = \exists a\left[\overline{p}(a , \overline{g})\right ] \nonumber \\ \nexists a [ q(a ) ] = \nexists a\left[\overline{p}(a , \overline{g})\right ] & , \ ; \nexists a [ \overline{q}(a ) ] = \nexists a\left[p(a , \overline{g})\right ] . \nonumber\end{aligned}\ ] ] proposition [ prop : postconds ] considers the four basic cases which correspond to ( 1 * ) ( 4 * ) in table [ tab : possibilitiessinglecase ] , showing that in fact they can all be reduced to matchings in the image of the host graph , i.e. to ( 1 * ) in table [ tab : possibilitiessinglecase ] , verifying the theorem . now we move on to the induction step which considers combinations of quantifiers . well - definedness guarantees independence with respect to the order in which elements @xmath135 in the postcondition are selected . when there is a universal quantifier @xmath192 , according to eq . ( [ eq : closure ] ) , elements of @xmath117 are replicated as many times as potential instances of @xmath117 can be found in the host graph . in order to continue the procedure we have to clone the rest of the diagram for each replica of @xmath117 , except those graphs which are existentially quantified before @xmath117 in the formula . that is , if we have a formula @xmath193 when performing the closure of @xmath117 , we have to replicate @xmath194 as many times as @xmath117 , but not @xmath195 . moreover @xmath195 has to be connected to each replica of @xmath117 , preserving the identifications of the morphism @xmath196 . more in detail : when closure is applied to @xmath117 , we iterate on all graphs @xmath197 in the diagram . there are three possibilities : * if @xmath197 is existentially quantified after @xmath117 @xmath198 then it is replicated as many times as @xmath117 . appropriate morphisms are created between each @xmath135 and @xmath199 if a morphism @xmath200 existed . the new morphisms identify elements in @xmath135 and @xmath199 according to @xmath163 . this permits finding different matches of @xmath197 for each @xmath135 , some of which can be equal . in the image of the host graph but only one of @xmath197 , then the three replicas of @xmath195 are matched to the same part of @xmath201 . ] * if @xmath197 is existentially quantified before @xmath117 @xmath202 then it is not replicated , but just connected to each replica of @xmath117 if necessary . this ensures that a unique @xmath197 has to be found for each @xmath135 . moreover , the replication of @xmath117 has to preserve the shape of the original diagram . that is , if there is a morphism @xmath203 then each @xmath204 has to preserve the identifications of @xmath163 ( this means that we take only those @xmath135 which preserve the structure of the diagram ) . * if @xmath197 is universally quantified ( no matter if it is quantified before or after @xmath117 ) , again it is replicated as many times as @xmath117 . afterwards , @xmath197 will itself need to be replicated due to its universality . the order in which these replications are performed is not relevant as @xmath205 . @xmath42 previous theorem and the corollaries that follow heavily depend on the host graph and its image ( through matching ) so analysis techniques developed so far in mgg which are independent of the host graphs can not be applied . the `` problem '' is the universal quantifier . we can consider the initial digraph and dispose to some extent of the host graph and its image . this is related to the fact ( sec . [ sec : movingconditions ] ) that it is possible to transform postconditions into equivalent preconditions . two applications of th . [ th : reductionpost ] are the following corollaries that characterize coherence , compatibility and consistency of postconditions . [ cor : equivpostac_seqs ] a postcondition is coherent if and only if its associated ( set of ) sequence(s ) is coherent . also , it is compatible if and only if its associated ( set of ) sequence(s ) is compatible and it is consistent if and only if its associated ( set of ) sequence(s ) is applicable . [ cor : postcondconscohcomp ] a postcondition is consistent if and only if it is coherent and compatible . * * example.**@xmath41let s consider the diagram in fig . [ fig : final_example_gc ] with formula @xmath206 $ ] . the postcondition states that if an operator is connected to a machine , such machine is busy . the formula has an implication so it is not possible to directly generate the set of sequences because the postcondition also holds when the left of the implication is false . the closure operator @xmath111 reduces the postcondition to existential quantifiers , which is represented to the right of the figure . the resulting modified formula would be @xmath207 $ ] . once the formula has existentials only , we manipulate it to get rid of implications . thus , we have @xmath208 = \exists a^0_1 a^0_2 a^1_1 a^1_2 [ ( \overline{a^0_1 } \wedge \overline{a^1_2 } ) \vee ( \overline{a^0_1 } \wedge a^1_2 ) \vee ( a^1_1 \wedge \overline{a^0_2 } ) \vee ( a^1_1 \wedge a^1_2)]$ ] . this leads to a set of four sequences : @xmath209 . thus , the graph @xmath201 and the production satisfy the postcondition if and only if some sequence in the set is applicable to @xmath201 . @xmath42 something left undefined is the order of productions @xmath210 and @xmath211 in the sequences . consistency does not depend on the ordering of productions as long as the first to be applied is production @xmath29 because productions @xmath212 ( and their negation ) are sequentially independent ( they do not add nor delete any edge or node ) . if they are not sequentially independent then there exists at least one inconsistency . this inconsistency can be detected using previous corollaries independently of the order of the productions . in this section we give two different proofs that it is possible to transform preconditions into equivalent postconditions and back again . the first proof ( sketched ) makes use of category theory while the second relies on the characterizations of coherence , g - congruence and compatibility . to ease exposition we shall focus on the certainty part only as the nihilation part would follow using the inverse of the production . we shall start with a case that can be addressed using equations , th . [ th : reductionpost ] and cor . [ cor : equivpostac_seqs ] : when the transformed postcondition for a given precondition does not change . the question of whether it is always possible to transform a precondition into a postcondition and back again in this restricted case would be equivalent to asking for sequential independence of the production @xmath29 and the identities @xmath212 or @xmath213 : @xmath214 where the sequence to the left of the equality corresponds to a precondition and the sequence to the right corresponds to its equivalent postcondition . in general the production may act on elements that appear in the diagram of the precondition , spoiling sequential independence . left and center of fig . [ fig : preposttrans ] in which the first basic ac ( match ) is considered suggest that the pre - to - post transformation is a categorical pushout is a pushout where @xmath29 , @xmath17 , @xmath215 , @xmath18 and @xmath155 are known and @xmath156 , @xmath216 and @xmath215 need to be calculated . ] in the category of simple digraphs and partial morphisms . theorem [ th : reductionpost ] proves that any postcondition can be reduced to the match case . besides , we can trivially consider total morphisms ( instead of partial ones ) by restricting the domain and the codomain of @xmath29 to the nodes in @xmath155 . for the post - to - pre transformation we can either use pullbacks or pushouts plus the inverse of the production involved . to see that precondition satisfaction is equivalent to postcondition satisfaction using category theory , we should check that the different pushouts can be constructed ( @xmath217 , etcetera ) and that @xmath218 and @xmath219 ( refer to fig . [ fig : preposttrans ] ) . although some topics remain untouched such as dangling edges , we shall not carry on with category theory . * example*.@xmath41let be given the precondition @xmath155 to the left of fig . [ fig : pre2postsimpleex ] with formula @xmath220 $ ] . to calculate its associated postcondition we can apply the production to @xmath155 and obtain @xmath156 , represented also to the left of the same figure . notice however that it is not possible to fing a match of @xmath17 in @xmath155 because of node @xmath221 . one possible solution is to consider @xmath222 and restrict the production to those common elements . this is done to the right of fig . [ fig : pre2postsimpleex ] @xmath42 [ th : prepostpre ] any consistent precondition is equivalent to some consistent postcondition and vice versa . _ proof _ + @xmath41for the post - to - pre transformation roles of @xmath29 and @xmath223 are interchanged so we shall address only the pre - to - post case . it is enough to study a single @xmath117 in the diagram as the same procedure applies mechanically ( th . 4.1 transforms any precondition into a sequence of productions ) . also , it suffices to state the result for @xmath119 because @xmath161 is similar but the evolution depends on @xmath223 . finally , we shall assume that @xmath29 and @xmath119 are not sequentially independent . recall that @xmath58-congruence guarantees sameness of the initial digraph , which is what the sequence demands on the host graph . therefore , all we have to do is to use @xmath58-congruence to check the differences in the two sequences : @xmath224 however , before that we need to guarantee coherence and compatibility of both sequences ( see the hypothesis of th . 4 in @xcite ) . coherence gives rise to the following equation : @xmath225 where @xmath69 and @xmath30 correspond to @xmath29 , @xmath226 to @xmath227 and @xmath228 to @xmath229 . fortunately , @xmath212 is a production that does nothing , so from the dynamical point of view any conflict should come from @xmath29 , i.e. @xmath230 which has been used in the implication of eq . ( [ eq:3 ] ) . by consistency we have that @xmath231 so eq . ( [ eq:3 ] ) will be fulfilled if the postcondition is the precondition but erasing the elements that the production deletes . a similar reasoning for the nihil part tells us that we should add to the postcondition all those elements added by the production . compatibility can only be ruined by dangling edges . in sec . 6.1 in @xcite dangling edges are deleted transforming the production via the opeartor @xmath74 . this is proved to be equivalent to defining a sequence by appending a so - called @xmath79-production . in essence the @xmath79-production just deletes any dangling edge , thus keeping compatibility . this very same procedure can be applied now : @xmath232 according to prop . 5 and th . 4 in @xcite , two compatible and coherent sequences are @xmath58-congruent if the following equation ( adapted to our case ) is fulfilled : @xmath233 we have that @xmath234 . also , @xmath235 because @xmath227 acts on the certainty part and @xmath236 ( see e.g. prop . 4.1.4 in @xcite ) . we are left with @xmath237 which is guaranteed by compatibility : once @xmath227 is transformed into @xmath238 in eq . ( [ eq:7 ] ) there can not be any potential dangling edge , except those to be deleted by @xmath29 in the last step . @xmath42 it is worth stressing the fact that the transformation between pre and postconditions preserve consistency of the application condition . we have seen in this section that @xmath29 not only acts on @xmath17 but on the whole precondition . we can therefore extend the notation : @xmath239 pre - to - post and post - to - pre transformations can affect the diagram and the formula . see the example below . there are two clear cases : * the application condition requires the graph to appear and the production deletes all its elements . * the application condition requires the graph not to appear and the production adds all its elements . for a given application condition ac it is not necessarily true that @xmath240 because some new elements may be added and some obsolete elements discarded . what we will get is an equivalent condition adapted to @xmath29 that holds whenever @xmath117 holds and fails to be true whenever @xmath117 is false . * example*.@xmath41 in fig . [ fig : pre2postex ] there is a very simple transformation of a precondition into a postcondition through morphism @xmath241 . the associated formula to the precondition @xmath155 that we shall consider is @xmath220 $ ] . the production deletes two arrows and adds a new one . the overall effect is reverting the direction of the edge between nodes @xmath167 and @xmath242 and deleting the self - loop in node @xmath167 . notice that @xmath68 can not match node @xmath242 to @xmath243 in @xmath58 because of the edge @xmath244 in the application condition . suppose we had a ( redundant ) graph @xmath195 made up of a single node @xmath167 with a self loop in the precondition and with formula @xmath245 $ ] . the formula in the postcondition would still be @xmath246 $ ] . the opposite transformation , from postcondition into precondition , can be obtained by reverting the arrow , i.e. through @xmath247 . more general schemes can be studied applying the same principles . let @xmath248 . if a pre - post - pre transformation is carried out , we will have @xmath249 because edge ( 2,1 ) would be added to @xmath155 . however , it is true that @xmath250 . note that in fact @xmath227 and @xmath29 are sequentially independent if we limit ourselves to edges , so it would be possible to simply move the precondition to a postcondition as it is . nonetheless , we have to consider nodes 1 and 2 as the common parts between @xmath17 and @xmath155 . this is the same kind of restriction as the one illustrated in fig . [ fig : pre2postsimpleex ] . @xmath42 if the pre - post - pre transformation is thought of as an operator @xmath251 acting on application conditions , then it fulfills @xmath252 where @xmath212 is the identity . the same would also be true for a post - pre - post transformation . a possible interpretation of eq . is that the definition of the application condition can vary from the _ natural _ one , according to the production under consideration . pre - post - pre or post - pre - post transformations adjust application conditions to the corresponding production . when defining diagrams some `` practical problems '' may turn up . for example , if the diagram @xmath253 is considered then there are two potential problems : 1 . the direction in the arrow @xmath254 is not the natural one . nevertheless , injectiveness allows us to safely revert the arrow , @xmath255 . even though we only formally state @xmath256 and @xmath257 , other morphisms naturally appear and need to be checked out , e.g. @xmath258 . new morphisms should be considered if they relate at least one element . graphs @xmath135 can be identified as the complete graph @xmath259 , in which nodes are graphs @xmath135 and morphisms are @xmath169 . whether this is a directed graph or not is a matter of taste ( morphisms are injective ) . ] in this section we touch on _ delocalization _ of graph constraints and application conditions as well as their equivalence . also , we shall pave the way to multidigraph rewriting to be studied in detail in sec . [ sec : fromsimpledigraphstomultidigraphs ] . let @xmath260 be a sequence of productions with their corresponding acs . we have seen in th . [ th : reductionpost ] that preconditions and postconditions are equivalent and in th . [ th : prepostpre ] that they can be transformed into sequences of productions . as a precondition in @xmath261 is the same as a postcondition in @xmath262 , we see that acs can be moved arbitrarily inside a sequence . similarly , constraints set on the intermediate states of a derivation can be moved among them . a graph constraint gc set in the initial state @xmath58 to which a production @xmath29 is going to be applied is equivalent to the precondition @xmath263.\ ] ] if the gc is set on the final state @xmath264 to which the production @xmath29 has been applied , there is an equivalent postcondition : @xmath265.\ ] ] in both cases the diagrams are given by the lhs or the rhs plus the diagram of the graph constraint . we call this property of application conditions and graph constraints _ delocalization_. we shall now address variable nodes which will be used to enhance mgg functionality to deal with multidigraphs . graph transformation with variables is studied in @xcite . we shall summarize the proposal in @xcite and propound an alternative way to close the section . if instead of nodes of fixed type variable , types are allowed we get a so called _ graph pattern_. a _ rule scheme _ is just a production in which graphs are graph patterns . a _ substitution function _ @xmath266 specifies how variable names taking place in a production are substituted . a rule scheme @xmath29 is instantiated via substitution functions producing a particular production . for example , for substitution function @xmath266 we get @xmath267 . the set of production instances for @xmath29 is defined as the set @xmath268 @xmath266 is a substitution@xmath269 . the _ kernel _ of a graph @xmath58 , @xmath270 , is defined as the graph resulting when all variable nodes are removed . it might be the case that @xmath271 . the basic idea is to reduce any rule scheme to a set of rule instances . note that it is not possible in general to generate @xmath272 because this set can be infinite . the way to proceed is not difficult : 1 . find a match for the kernel of @xmath17 . 2 . induce a substitution @xmath266 such that the match for the kernel becomes a full match @xmath273 . 3 . construct the instance @xmath274 and apply @xmath267 to get the direct derivation @xmath275 . as an alternative , we may extend the concept of type assignment . recall from sec . [ sec : mggs ] that types are assigned by a function from the set of nodes @xmath11 of a simple digraph @xmath58 to some fixed set @xmath12 of types , @xmath276 . instead , we shall define @xmath277 where @xmath278 is the power set of @xmath12 except for the empty set because we do not permit nodes without types . when two matrices are operated , the types of a fixed node will be the intersection of the nodes operated . for example , suppose that we * and * two matrices @xmath279 and that the ( set of ) nodes associated to the elements @xmath280 and @xmath281 are @xmath282 and @xmath283 , respectively . then , @xmath284 . the operation would not be allowed in case @xmath285 . * example*.@xmath41 let a type of nodes be represented by squares ( call them _ multinodes _ ) and the rest ( call them _ simple nodes _ ) by colored circles . the set of types @xmath12 is split into two : multinodes and simple nodes . let s consider the graph constraint @xmath286 , with @xmath287 the diagram depicted in fig . [ fig : simplifieddiagram2 ] made up of the graphs @xmath100 and @xmath101 , along with the formula @xmath288 $ ] . this graph constraint is `` edges must connect nodes and multinodes alternatively but no edge is allowed to be incident to two multinodes or to two simple nodes , including self - loops '' . in the graph @xmath100 of fig . [ fig : simplifieddiagram2 ] , @xmath289 and @xmath290 represent variable nodes while @xmath280 and @xmath281 in @xmath101 have a fixed type . we may think of @xmath280 and @xmath281 as variable nodes whose set of types has a single element . in this section we show how mgg can deal with multidigraphs ( directed graphs allowing multiple parallel edges ) just by considering variable nodes . at first sight this might seem a hard task as mgg heavily depends on adjacency matrices . adjacency matrices are well suited for simple digraphs but can not cope with parallel edges . this section can be thought of as a _ theoretical application _ of graph constraints and application conditions to matrix graph grammars . the idea is not difficult : a special kind of node ( call it _ multinode _ in contrast to _ simple node _ ) associated to every edge in the graph is introduced , i.e. edges in the multidigraph are substituted by multinodes in a simple digraph representation of the multidigraph . graphically , multinodes will be represented by a filled square while normal nodes will appear as colored circles . see the example by the end of sec . [ sec : delocalization ] operations previously specified on edges now act on multinodes : adding an edge is transformed into a multinode addition and edge deletion becomes multinode deletion . there are edges that link multinodes to their source and target simple nodes . some restrictions ( application conditions ) to be imposed on the actions that can be performed on multinodes exist , as well as on the shape or topology of permitted graphs ( graph constraints ) . not every possible graph with multinodes represents a multidigraph . * example*.@xmath41consider the simple production in fig . [ fig : simplemultigraph ] with two parallel edges between nodes @xmath167 and @xmath221 . as commented above , multinodes are represented by square nodes while normal nodes are left unchanged . when @xmath29 deletes an edge , @xmath291 deletes a multinode . adjacency matrices for @xmath291 are : @xmath292 \qquad r & = \left [ \begin{array}{cccccc } \vspace{-10pt } 0 & 0 & 0 & 1 & 1 & \vert\;\ ; 1_{\phantom{2 } } \\ \vspace{-10pt } 0 & 0 & 0 & 0 & 0 & \vert\;\ ; 2_{\phantom{2 } } \\ \vspace{-10pt } 0 & 0 & 0 & 0 & 0 & \vert\;\ ; 3_{\phantom{2 } } \\ \vspace{-10pt } 0 & 0 & 1 & 0 & 0 & \vert\;\ ; a_2 \\ \vspace{-10pt } 0 & 1 & 0 & 0 & 0 & \vert\;\ ; d_{\phantom{2 } } \\ \vspace{-10pt } \end{array } \right ] \\ k = \left [ \begin{array}{ccccccc } \vspace{-10pt } 0 & 0 & 0 & 0 & 0 & 0 & \vert\;\ ; 1_{\phantom{2 } } \\ \vspace{-10pt } 0 & 0 & 0 & 1 & 0 & 0 & \vert\;\ ; 2_{\phantom{2 } } \\ \vspace{-10pt } 0 & 0 & 0 & 1 & 0 & 0 & \vert\;\ ; 3_{\phantom{2 } } \\ \vspace{-10pt } 1 & 1 & 0 & 1 & 1 & 1 & \vert\;\ ; a_1 \\ \vspace{-10pt } 0 & 0 & 0 & 1 & 0 & 0 & \vert\;\ ; a_2 \\ \vspace{-10pt } 0 & 0 & 0 & 1 & 0 & 0 & \vert\;\ ; d_{\phantom{2 } } \\ \vspace{-10pt } \end{array } \right ] \qquad \;\;e & = \left [ \begin{array}{ccccccc } \vspace{-10pt } 0 & 0 & 0 & 1 & 0 & 0 & \vert\;\ ; 1_{\phantom{2 } } \\ \vspace{-10pt } 0 & 0 & 0 & 0 & 0 & 0 & \vert\;\ ; 2_{\phantom{2 } } \\ \vspace{-10pt } 0 & 0 & 0 & 0 & 0 & 0 & \vert\;\ ; 3_{\phantom{2 } } \\ \vspace{-10pt } 0 & 0 & 1 & 0 & 0 & 0 & \vert\;\ ; a_1 \\ \vspace{-10pt } 0 & 0 & 0 & 0 & 0 & 0 & \vert\;\ ; a_2 \\ \vspace{-10pt } 0 & 0 & 0 & 0 & 0 & 0 & \vert\;\ ; d_{\phantom{2 } } \\ \vspace{-10pt } \end{array } \right ] \nonumber\end{aligned}\ ] ] in a real situation , a development tool such as atom@xmath293 or agg and http://www.gratra.org/ for agg and some other tools . ] should take care of all these representation issues . a user would see what appears to the left of fig . [ fig : simplemultigraph ] and not what is depicted to the right of the same figure . @xmath42 some restrictions on what a production can do to a multidigraph are necessary in order to obtain a multidigraph again . think for example the case in which after applying some production we get a graph in which there is an isolated multinode ( which would stand for an edge with no source nor target nodes ) . all we have to do is to find the properties that define one edge and impose them on multinodes as graph constraints : 1 . a simple node ( resp . , multinode ) can not be directly connected to another simple node ( resp . , multinode ) . 2 . edges ( encoded as multinodes ) always have a simple node as source and a simple node as target . first condition above is addressed in the example of sec . [ sec : delocalization ] with graph constraint @xmath294 . see fig . [ fig : simplifieddiagram2 ] . the second condition can be encoded as another graph constraint @xmath295 . the diagram can be found in fig . [ fig : multidigraphgc2 ] and the formula is @xmath296 $ ] . [ th : multidigraph ] any multidigraph is isomorphic to some simple digraph @xmath58 together with the graph constraint @xmath297 . _ proof ( sketch ) _ + @xmath41a graph with multiple edges @xmath298 consists of disjoint finite sets @xmath2 of nodes and @xmath27 of edges and source and target functions @xmath299 and @xmath300 , respectively . function @xmath301 , @xmath302 , @xmath303 returns the node source @xmath304 for edge @xmath69 . we are considering multidigraphs because the pair function @xmath305 need not be injective , i.e. several different edges may have the same source and target nodes . we have digraphs because there is a distinction between source and target nodes . this is the standard definition found in any textbook . it is clear that any @xmath1 can be represented as a multidigraph @xmath58 satisfying @xmath306 . the converse also holds . to see it , just consider all possible combinations of two nodes and two multinodes and check that any problematic situation is ruled out by @xmath306 . induction finishes the proof.@xmath42 the multidigraph constraint @xmath306 must be fulfilled by any host graph . if there is a production @xmath16 involved , @xmath306 has to be transformed into an application condition over @xmath29 . in fact , the multidigraph constraint should be demanded both as precondition and postcondition . this is easily achieved by means of eqs . ( [ eq:8 ] ) and ( [ eq:9 ] ) . this section is closed analyzing what behavior we have for multidigraphs with respect to dangling edges . with the theory as developed so far , if a production specifies the deletion of a simple node then an @xmath79-production would delete any edge incident to this simple node , connecting it to any surrounding multinode . but restrictions imposed by mc do not allow this so any production with potential dangling edges can not be applied . in order to automatically delete any potential multiple dangling edge , @xmath79-productions need to be restated by defining them at a multidigraph level , i.e. @xmath79-productions have to delete any potential `` dangling multinode '' . a new type of productions ( @xmath307-productions ) are introduced to get rid of annoying edges that would dangle when multinodes are also deleted by @xmath79-productions . we will not develop the idea in detail and will limit to describe the concepts . the way to proceed is to define the appropriate operator @xmath308 and redefine the operator @xmath74 . a production @xmath16 between multidigraphs that deletes one simple node @xmath309 may give rise to one @xmath79-production that deletes one or more multinodes @xmath310 ( those `` incident '' to @xmath309 not deleted by the grammar rule ) . this @xmath79-production can in turn be applied only if any edge incident to the @xmath310 s has already been erased , hence possibly provoking the appearance of one @xmath307-production . this process is depicted in fig . [ fig : spomultidigraph ] where , in order to apply production @xmath29 , productions @xmath77 and @xmath311 need to be applied in advance @xmath312 eventually , one could simply compose the @xmath307-production with its @xmath79-production , renaming it to @xmath79-production and defining it as the way to deal with dangling edges in case of multiple edges , fully recovering the standard behavior in mgg . as commented above , a potential user of a development tool such as atom@xmath293 would still see things as in the simple digraph case , with no need to worry about @xmath307-productions . another theoretical use of application conditions and graph constraints is the encoding of turing machines and boolean circuits using matrix graph grammars ( see @xcite ) . however , they are not necessary for petri nets ( see chap . 10 in @xcite ) . in the present contribution we have introduced preconditions and postconditions for mggs , proving that there is an equivalent set of sequences of plain rules to any given postcondition . besides , coherence , compatibility and consistency of postconditions have been characterized in terms of already known concepts for sequences . we have also proved that it is always possible to transform any postcondition into an equivalent precondition and vice versa . moreover , we have seen that restrictions are delocalized if a sequence is under consideration . an alternative way to that in @xcite to tackle variable nodes has also been proposed . this allows us to extend mgg to cope with multidigraphs without major modifications to the theory . in @xcite there is an exhaustive comparison of the application conditions in mgg and other proposals . the main papers to the best of our knowledge that tackle this topic are @xcite ( with the definition of acs ) , @xcite where gcs and acs are extended with nesting and satisfiability , and also @xcite in which acs are generalized to arbitrary levels of nesting ( though restricted to trees ) . for future work , we shall generalize already studied concepts in mgg for multidigraphs such as coherence , compatibility , initial digraphs , graph congruence , reachability , etcetera . our main interest , however , will be focused on complexity theory and the application of mgg to the study of complexity classes , * p * and * np * in particular . @xcite follows this line of research . corradini , a. , montanari , u. , rossi , f. , ehrig , h. , heckel , r. , lwe , m. 1999 . _ algebraic approaches to graph transformation - part i : basic concepts and double pushout approach_. in @xcite , pp . : 163 - 246 ehrig , h. , heckel , r. , korff , m. , lwe , m. , ribeiro , l. , wagner , a. , corradini , a. 1999 . _ algebraic approaches to graph transformation - part ii : single pushout approach and comparison with double pushout approach . _ in @xcite , pp . : 247 - 312 . de lara , j. , vangheluwe , h. 2004 . _ defining visual notations and their manipulation through meta - modelling and graph transformation_. journal of visual languages and computing . special section on `` domain - specific modeling with visual languages '' , vol 15(3 - 4 ) , pp . : 309 - 330 . elsevier science . prez velasco , p. p. , de lara , j. 2006 . _ towards a new algebraic approach to graph transformation : long version . _ tech . rep . of the school of comp . autnoma madrid . ` http://www.ii.uam.es/\simjlara/investigacion/techrep_03_06.pdf ` . prez velasco , p. p. , de lara , j. 2007 . _ using matrix graph grammars for the analysis of behavioural specifications : sequential and parallel independence _ proc . prole07 , pp . : 11 - 26 . . notes theor . elsevier . prez velasco , p. _ matrix graph grammars : an algebraic approach to graph dynamics_. isbn 978 - 3639212556 . vdm verlag . also available as e - book at : ` http://www.mat2gra.info/ ` and ` arxiv:0801.1245v1 ` . prez velasco , p. p. , de lara , j. 2009 . _ a reformulation of matrix graph grammars with boolean complexes_. the electronic journal of combinatorics . vol 16(1 ) . available at : ` http://www.combinatorics.org/ ` .

rotation - induced time - odd components in the nuclear mean field are analyzed using the hartree - fock cranking approach with effective interactions siii , skm * , and skp . identical dynamical moments are obtained for pairs of superdeformed bandstb(2)dy(1 ) andgd(2)tb(1 ) . the corresponding relative alignments strongly depend on which time - odd mean - field terms are taken into account in the hartree - fock equations .

nucl-th9502030

a description of ground and excited states in terms of a mean field is a well established approach in nuclear structure . most of the nuclear phenomena can be considered as manifestations of the mean field properties , or the mean field can be used as a suitable framework and the first - order approximation . the mean field results from averaging the nucleon - nucleon interactions over states of individual nucleons . the averaging procedure , which must take into account the fermi statistics of nucleons , can be formalized in terms of a variational approach , and leads to the well - known hartree - fock ( hf ) selfconsistent equations @xcite . dynamic or time - dependent phenomena can be described by a corresponding time - dependent hf method . the nuclear state is then represented by a one - body density matrix , which evolves in time according to the hamilton equations , and represents a motion of a wave packet . such an approach has been applied to genuinely time - dependent problems , like nuclear reactions , but it is in fact best suited to describe stationary collective states . the nuclear rotation is an example of a collective motion for which linear combinations of stationary states of a given spin can be identified with a rotating wave packet . in this case , a transition from the time - dependent hf theory to a stationary problem can be achieved by introducing rotating intrinsic frame of reference . in this frame , the equations of motion are time - independent , however , the resulting density matrix is not invariant with respect to the time - reversal operator . as a consequence , the mean field obtained for such a density matrix also loses its time invariance , and acquires new terms which are odd with respect to the time reversal . properties of nuclear time - even mean fields are known rather well , because they are reflected in multiple static phenomena which can be studied experimentally @xcite . on the other hand , very little is known about properties of the time - odd mean fields . most studies were up to now carried out either within the adiabatic @xcite or semiclassical approximations @xcite . in particular , the adiabatic approximation to the nuclear translation , rotation , or quadrupole motion leads to the well - known thouless - valatin @xcite corrections to the mass , moment of inertia , or to the vibrational mass parameters , respectively . these corrections reflect the fact that a velocity - dependent mean field should be appropriately transformed to the intrinsic frame of reference @xcite corresponding to a given collective mode . in fact , simple estimations and numerical calculations @xcite show that the inertia obtained for effective forces without the effective - mass terms ( @xmath4=@xmath5 ) are very close to those for @xmath4@xmath6@xmath5 when the thouless - valatin corrections are consistently included . fast nuclear rotation is a phenomenon in which the collective motion should be described beyond the adiabatic limit . such a cranking model has been successfully used in explaining numerous high - spin effects in nuclei @xcite . in this approach , the properties of the rotating mean field explicitly depend on the angular velocity . however , most studies performed up to now were done in terms of phenomenological mean fields , which are not selfconsistently depending on the rotating states , and , therefore , do not incorporate time - odd terms . only in ref . @xcite an attempt has been made to include the effects of a zero - range interaction within the nilsson single - particle mean field . full selfconsistent cranking calculations @xcite are still rather scarce , and no explicit analysis of the time - odd mean - field components is available . on the other hand , the discovery of the superdeformation @xcite , and the resulting avalanche of very precise data on high - spin states , allow for an attempt to study these unknown aspects of the nuclear mean field . in particular , the phenomenon of identical bands ( see ref . @xcite for a recent review ) provides an extremely rich and puzzling information on properties of fast rotating states . in terms of the mean field approaches explicitly depending on the time - odd components , the calculated identical @xmath7-ray transitions have been obtained in two cases only , namely , ( i ) the hf cranking calculations with skyrme interaction gave yrast band of @xmath8hg identical to an excited band in @xmath8pb @xcite , and ( ii ) the identical bands in @xmath2dy and @xmath1 tb ( excited band ) were obtained in the relativistic mean - field ( rmf ) theory @xcite . in both studies , the authors invoked the time - odd components of the mean field as a crucial element of the obtained results . in the present study we aim at a detailed analysis of these time - odd terms in the context of the identical bands phenomenon . we focus our attention on two classic and experimentally well studied pairs of identical bands , namely , on those in @xmath2dy and @xmath1 tb ( excited band ) , and in @xmath1 tb and @xmath3gd ( excited band ) . we perform our calculations in terms of the hf cranking method with the skyrme interactions . it was not in our intentions to give here a full account of the time - reversal breaking description that the selconsistent hf cranking approach really offers and the experiment may test . such a field of studies seems to emerge now in relation with the fast progress in the available experimental information . in particular , our choice of the illustrative material purposely minimizes rather than maximizes a possible magnitude of the time - odd effects . this is so , because we mainly discuss the role of the @xmath9{1/2}$ ] nilsson orbital known to interact only weakly via the coriolis term in the hamiltonian . in all standard approaches , such as the woods - saxon ( ws ) , nilsson , hf , or rmf cranking models , one signature member of this orbital is to a good approximation given by a staight line as a function of the angular velocity . this type of dependence can be opposed to dramatic changes of other aligning or interacting orbitals which are much more sensitive to the rotating field and hence to its time - odd components . on the other hand , the specific behavior of the discussed orbital induces a regular behavior of various observables , and has a great advantage in the fact that the level repulsion or level crosings do not disturb the comparisons of primary interest here . in section [ sec3 ] we discuss the skyrme interaction and energy density with a particular emphasis on the relations between the time - even and time - odd components of the mean field . the hf cranking calculations are presented in section [ sec4 ] , where several variants of the skyrme functional are used and the influence of the time - odd terms on the rotational properties is analyzed , and conclusions are presented in section [ sec6 ] . the skyrme energy functional @xcite is a three - dimensional integral @xmath10 of the energy density @xmath11 which can be represented as a sum of the kinetic energy and of the potential - energy isoscalar and isovector terms @xmath12 where @xmath13 with @xmath14and the isospin index @xmath15 can have values 0 or 1 . coupling constants @xmath16 have superscripts corresponding to various density functions appearing in eq . ( [ eq109 ] ) . all of these coupling constants can , in principle , depend on particle densities , see refs . @xcite , but the most common choice @xcite restricts the density dependence to the @xmath17 and @xmath18 terms . in appendix [ appa ] we give coupling constants @xmath16 expressed through the traditional parameters of the skyrme interaction . apart from the density - dependent coupling constants , the isoscalar energy density , @xmath19 , depends on the isoscalar density functions and the isovector one , @xmath20 , depends on the isovector density functions . for the particle densities , @xmath21 , the isoscalar and isovector parts are defined in the usual way as a sum and a difference of the proton and neutron contributions , respectively , @xmath22 and analogous expressions are used to define other densities . altogether we have to consider six position - dependent density functions @xmath23 , @xmath24 , @xmath25 , @xmath26 , @xmath27 and @xmath28 with definitions given in ref . whenever the isospin indices are omitted , we understand that the symbols may refer either to the isoscalar ( @xmath15=0 ) or isovector ( @xmath15=1 ) parts . in the energy densities ( [ eq109 ] ) there appear two scalar time - even density functions , @xmath23 and @xmath24 , one vector time - odd , @xmath25 , and two pseudo - vector time - odd ones , @xmath26 and @xmath27 , and one pseudo - tensor time - even density function , @xmath28 . the vector time - even density @xmath29 which appears in eq . ( [ eq109 ] ) is given by the antisymmetric part of the pseudo - tensor density , i.e. , @xmath30=@xmath31 , and is not an independent quantity . terms in @xmath32 and @xmath33 are bilinear in time - even and time - odd densities , respectively . therefore , we denote them by the superscripts `` even '' and `` odd '' . this notation is only used to indicate the dependence on two different classes of densities . it should not be confused with the fact that both @xmath34 and @xmath33 are of course _ even _ with respect to the time reversal . to every term in @xmath32 there corresponds an analogous term in @xmath35 , as seen in eq . ( [ eq109 ] ) . by varying the energy density ( [ eq109 ] ) with respect to the six density functions @xmath23 , @xmath24 , @xmath25 , @xmath26 , @xmath27 and @xmath28 one obtains the mean fields . details of calculations are presented in ref . here we only repeat the final results conforming to the notation introduced above . time - even and time - odd mean fields are obtained by a variation of @xmath34 and @xmath35 , respectively , and they read @xmath36these fields are given by the six potential functions @xmath37 , @xmath38 , @xmath39 , @xmath40 , @xmath41 and @xmath42 which have tensor transformation properties respectively identical to the six density functions on which they depend through the following formulae @xmath43 the tensor gradient operators in eqs . ( [ eq209 ] ) and ( [ eq210e ] ) are defined @xcite as @xmath44=@xmath45 and @xmath46@xmath47@xmath48 . due to the density dependence of the coupling constants @xmath17 and @xmath49 one has to add to the isoscalar potential energy @xmath50 the rearrangement terms @xcite resulting from the variation of these coupling constants with respect to the isoscalar particle density , i.e. , @xmath51 these terms introduce an explicit dependence of the complete time - even isoscalar potential @xmath50+@xmath52 on the isovector density @xmath53 and on the time - odd densities @xmath54 . similar terms have to be also consistently taken into account whenever some other coupling constants are assumed to be density dependent . finally , the neutron and proton hamiltonians , @xmath55 and @xmath56 , are obtained by combining the kinetic energy with the isoscalar and isovector mean fields : @xmath57 we are now in a position to discuss different time - odd terms in the mean fields . it is clear that the time - odd mean fields , @xmath58 , in eq . ( [ eq209 ] ) directly result from the `` odd '' part , @xmath35 , of the energy density in ( [ eq109 ] ) , and depend on 10 time - odd coupling constants @xmath49 , @xmath59 , @xmath60 , @xmath61 , and @xmath62 for @xmath15=0 and @xmath15=1 . similarly , the time - even mean fields depend on 10 time - even coupling constants @xmath17 , @xmath63 , @xmath64 , @xmath65 , and @xmath66 . for the skyrme interaction , the time - odd coupling constants are linear combinations of the time - even ones ( see appendix [ appa ] ) , and therefore the time - odd mean fields are uniquely determined from the time - even mean fields . since the time - even fields are tested against numerous experimental observations which have a static character , they are much better known then the time - odd ones . in fact , the skyrme force parameters have been in the past almost uniquely fitted to the static properties only . a description of dynamic properties which do depend on the time - odd fields does not therefore ( for the skyrme force ) require new parameters to be introduced and fitted . on the other hand , one sometimes adopts a different point of view by considering the energy density to be a more fundamental construction than the skyrme interaction itself @xcite . in such a case , all 20 coupling constants of eq . ( [ eq109 ] ) should be treated and adjusted independently . however , in the next section we show that some relations between time - odd and time - even coupling constants have an origin in the local gauge invariance of the energy density . as noted in ref . @xcite , in the energy density derived from the skyrme interaction the kinetic density @xmath24 and the current density @xmath25 appear in the characteristic combination of @xmath67@xmath68@xmath69 . the same is true for two other pairs of densities appearing together in the combinations @xmath70@xmath71@xmath72@xmath68@xmath73 and @xmath74@xmath71@xmath75@xmath76@xmath77@xmath78@xmath79 . this gives the following relations between three pairs of time - even and time - odd coupling constants @xmath80 the skyrme functional has now the form @xmath81 in ref . @xcite this structure has been interpreted as a result of the galilean invariance of the skyrme interaction . however , it has a deeper origin in the fact that the skyrme force is locally gauge invariant . in order to illustrate the role of the locally gauge invariant , velocity dependent interactions let us consider an arbitrary finite - range and non - local , but velocity independent interaction given by @xmath82 to simplify the notation we disregard for a moment the spin and isospin variables . @xmath83 describes an interaction process where the particles 1 and 2 are located at @xmath84 and @xmath85 before the interaction , and at @xmath86 and @xmath87 after the interaction . when the system of particles is described by a one - body density matrix @xmath88 , its hf interaction energy reads @xmath89 for a local gauge transformation of the many - body hf wave function @xmath90 , @xmath91 where @xmath92 is an arbitrary real function of the position @xmath93 , one obtains the following gauge - transformed one - body density matrix @xmath94 the general interaction energy ( [ e202 ] ) is not invariant with respect to such a transformation . however , when the interaction is local @xcite , @xmath95 the corresponding interaction energy , @xmath96 becomes invariant with respect to the local gauge . the direct term is invariant because it depends only on the gauge - invariant local densities , which are denoted by the single argument , i.e. , @xmath97@xmath98@xmath99 . on the other hand , in the exchange term the gauge factors coming from two density matrices cancel one another . when the density functions defining the skyrme energy functional ( [ eq109 ] ) are calculated for the gauge - transformed density matrix ( [ eq203 ] ) one obtains the following relations : @xmath100 and the three characteristic combinations of density functions , which appear in the energy density of eq . ( [ eq109c ] ) , are then explicitly gauge invariant . transformation properties of @xmath24 and @xmath25 allow to interpret @xmath101 as a velocity field , @xmath102 which shows that the flow of matter obtained through the gauge transformation is irrotational , @xmath103@xmath78@xmath104=0 . the local gauge invariance of the skyrme interaction reflects the fact that its velocity dependence has been introduced only to simulate the finite range effects of the effective interaction . in this way the skyrme interaction conserves the local gauge invariance of a velocity - independent finite - range local interaction , such as the gogny force @xcite , for example . the galilean invariance is a special case of the local gauge invariance , for which the phase in eq . ( [ eq111 ] ) is given by @xmath105 where @xmath106 is a constant linear momentum of the boost transformation . since the interaction energy does not depend on @xmath106 , the total energy ( [ eq107 ] ) is only modified through the kinetic energy [ the first term in eq . ( [ eq108 ] ) ] . using the transformation property of @xmath24 , eq . ( [ eq112b ] ) , we have the energy increase under the boost transformation , @xmath107 equal to the translational energy of the boosted system . this result holds for an initially stationary solution ( i.e. , for vanishing currents , @xmath25=0 ) , however , due to the transformation property of @xmath25 , eq . ( [ eq112d ] ) , the boost transformations can be added to one another by adding the corresponding momenta @xmath106 . for the boost transformation , one obtains the obvious velocity field ( [ eq212 ] ) , i.e. , @xmath108 since the velocity field ( [ eq212 ] ) obtained through a gauge transformation is irrotational , it can not correctly describe physical rotations of nuclei . this is so , because the nuclei basically rotate as rigid bodies ( at least in the independent - particle approximation @xcite ) , and the velocity field of a rigid - body rotation , @xmath109=@xmath110 , has a non - zero curl , @xmath103@xmath78@xmath109=2@xmath111 , i.e. , is not irrotational . of course , interactions and nucleon - nucleon correlations ( pairing ) introduce an irrotational component in the velocity field ( moment of inertia decreases below the rigid - body value ) , but this field is never entirely irrotational . in an analogy to the boost transformation , one may try to induce the rotation of a many - fermion system by adding for all particles a constant value , @xmath112 , to their projections of the angular momentum on a fixed ( say @xmath113 ) axis . such a procedure can be realized in terms of the twirl transformation given by the gauge function @xmath114 its velocity field has the form @xmath115 where @xmath111 is the vector of angular velocity oriented along the @xmath113 axis , and @xmath116 is the distance to this axis . one can see that the velocity field ( [ eq213 ] ) is singular at the rotation axis . the energy increase resulting from the kinetic energy density ( [ eq112b ] ) is then infinite for any arbitrarily small @xmath112 . the irrotational velocity field ( [ eq213 ] ) is of course very much different from the rigid - rotation velocity field , @xmath109 , even though both contain the same factor @xmath110 . this illustrative example shows that the nuclear rotation can not be realized by democratically distributing the angular momentum among all particles . in the cranking approximation , at a given value of the angular velocity some particles receive larger contributions ( aligning states ) and some smaller contributions ( high-@xmath117 states ) . a precise distribution can not be found without actually solving the quantal cranking equations . this example also shows that the gauge - invariant interaction must contribute to the rotational energy , contrary to what happens in the case of the translational motion . in the present study we have performed the hartree - fock ( hf ) calculations of superdeformed rotational bands using the skyrme effective interactions . the calculations have been done using the numerical code hfodd which employs a three - dimensional cartesian deformed harmonic oscillator ( ho ) basis to describe the single - particle wave functions . the details concerning the hfodd code will be presented elsewhere @xcite ; here we only give a few of its basic parameters pertaining to the present application . the calculations have been performed using a fixed basis given by the ho frequencies @xmath118=11.200 and @xmath119=6.246mev in the directions perpendicular and parallel to the symmetry axis , respectively . these values have been obtained by standard prescriptions developed for diagonalizing the deformed ws hamiltonian @xcite in the ho basis , and correspond to the ws potential with deformations @xmath120=0.61 and @xmath121=0.10 . the basis has been restricted to a fixed number , @xmath38 , of basis states having the lowest single - particle ho energies @xmath122=@xmath123+@xmath124+@xmath125+@xmath126+@xmath127 . the actual calculations have been performed with @xmath38=306 . this corresponds to the maximal numbers of oscillator quanta equal to 8 and 15 in the perpendicular and parallel directions , respectively . the stability of results with respect to increasing the size of the ho basis has been tested by performing calculations with @xmath38=604 , which introduces basis states up to 11 and 20 quanta in these two directions . it has been found that the rotational characteristics of the studied nuclei are almost independent of such an increase . for example , numerical inaccuracies in the dynamical moment @xmath0 and in the total angular momentum @xmath128 can be estimated to be smaller than 0.2@xmath129mev@xmath130 and 0.1@xmath131 , respectively . inaccuracies of relative values between different angular frequencies @xmath132 , or between different nuclei , are smaller than these estimates , because the numerical errors then cancel out . in the present study we aim at investigating the role of different time - odd terms in the selfconsistent mean fields ( [ eq209 ] ) obtained for rotating superdeformed nuclei . as discussed in section [ sec3 ] , this can be done by considering different values of 10 time - odd coupling constants appearing in the skyrme energy density , eq . ( [ eq109 ] ) . for every given set of values of coupling constants we perform full selfconsistent calculations within the hf cranking method . in the present study , pairing correlations are not taken into account . below we separately discuss three cases corresponding to ( i ) the complete skyrme functional , ( ii ) the skyrme functional with certain time - odd terms omitted , but with the gauge invariance preserved , and ( iii ) with the gauge invariance violated . as discussed in section [ sec2 ] , the energy functionals corresponding to the standard skyrme interactions preserve the local gauge invariance . the 20 coupling constants appearing in eq . ( [ eq109 ] ) are then restricted by 6 conditions ( [ eq215a])-([eq215c ] ) , which leads to the energy density of eq . ( [ eq109c ] ) . in table [ tab1 ] , the remaining 14 coupling constants are listed for the siii @xcite , skm * @xcite , and skp @xcite skyrme interactions . the values of the density - dependent coupling constants @xmath17 and @xmath133 are given for the vacuum ( @xmath23=0 ) and for the nuclear - matter saturation density ( @xmath23=@xmath134 ) characterizing a given force . the time - even coupling constants corresponding to the siii , skm * , and skp interactions are rather similar . the main difference consists in different values of the isoscalar - effective - mass coupling constant @xmath135 which is equal to 0 for skp ( effective mass @xmath136=1 ) , and 44.4mev@xmath71@xmath137 and 34.7mev@xmath71@xmath137 for siii and skm * , respectively ( effective masses @xmath136=0.76 and 0.79 ) . apart from that , the absolute values of the @xmath138 coupling constants are larger for skp than for siii and skm * , which gives better symmetry - energy properties @xcite within the skp parametrization as compared to the other two forces . on the other hand , the skm * parameters have been adjusted so as to properly describe the surface energy at large deformation , and therefore this interaction was successfully applied in numerous studies of superdeformation . values of the time - odd coupling constants corresponding to the three skyrme interactions , table [ tab1 ] , differ much more than those of the time - even ones . this illustrates uncertainties in determining the time - odd components of the mean field . the differences partly result from the fact that skp has been adjusted to give attractive matrix elements in the pairing channel , while those given by siii and skm * are repulsive . one should bear in mind , however , that for the skyrme interaction the time - odd coupling constants are unique functions of the time - even ones ( see appendix [ appa ] ) and , in principle , one has no freedom for an independent readjustment . such a readjustment is possible only if we consider the hf theory based directly on the energy density functional and not on the skyrme interaction . in the previous applications of the skyrme force to nuclear rotation @xcite , in the energy density ( [ eq109c ] ) the terms @xmath59 and @xmath139 have been neglected in order to facilitate the calculations . the first of these terms gives purely time - odd contribution to the mean field , while the second one gives both time - even and time - odd contributions , because the gauge invariance implies that @xmath139=@xmath140 , eq . ( [ eq215b ] ) . in fact , the term @xmath140 has also been usually neglected in most skyrme parametrizations applied to problems where the time - reversal symmetry is concerved , cf . ref . @xcite . since the hfodd code is organized in a different way , omitting some terms would not provide any serious simplifications , and the code may in fact handle the complete skyrme functional . this gives us a possibility to test the importance of different terms for the rotational properties of nuclei . there exist several high - spin observables which , when calculated from the hf solutions ( wave functions ) behave differently depending on whether various time - odd terms are included or not . this offers , in principle , a possibility of both better readjustment of the interaction coupling constants and better understanding of the underlying mechanisms . some of the physical quantities , as e.g. alignments and @xmath0-moments of intruder orbitals , are well recognized as responding strongly to the coriolis and centrifugal interaction effects . the same quantities are expected to respond relatively strongly to the time - odd terms in the mean - field hamiltonian . similarly , various families of orbitals having large high-@xmath141 components are systematically responsible for such precisely measurable effects as band - crossings ( back- or up - bending effects ) and related phenomena . in this study we would like to illustrate the effects of the time - odd terms on yet another seemingly more subtle mechanism related to identical bands , leaving the aformentioned analysis of intruder orbitals for a later investigation . in the following sections we present results of the hf cranking calculations for the yrast superdeformed bands in dysprosium , @xmath2dy(1 ) , and terbium , @xmath1tb(1 ) , and for the first excited bands in the corresponding isotones , @xmath1tb(2 ) and @xmath3gd(2 ) . according to the standard notation , the numbers in parentheses refer to numbers attributed in experimental studies in connection with relative intensities of gamma transitions . the experimental data are taken from refs . @xcite where the most recent and precise results are given . the pairs of superdeformed bands , @xmath1tb(2)@xmath2dy(1 ) and @xmath3gd(2)@xmath1tb(1 ) , are identical with a very high precision @xcite . for each pair , the @xmath7-ray transition energies @xmath142 are identical up to 2kev . the identity of the bands can be characterized in two ways , ( i ) by their relative alignments , and ( ii ) by their relative dynamical moments . the relative alignment @xmath143 is defined as a difference of spins in two bands at fixed angular velocity @xmath132 . similarly , the relative dynamical moment @xmath144 is the difference of @xmath0 at the same value of @xmath132 . in calculations , the latter is a derivative of the former with respect to the angular velocity . in the present study we have fixed the yrast proton configurations of @xmath2dy and @xmath1 tb to ( 16,16,17,17 ) and ( 15,16,17,17 ) , respectively . the numbers in parentheses denote the numbers of lowest states occupied in the parity - signature blocks ordered as ( + + , + @xmath68,@xmath68+,@xmath68@xmath68 ) , where the signs denote the parity quantum number and the sign of the ( imaginary ) signature @xcite quantum number , @xmath145=@xmath146 . one somtimes uses a notation based on the signature index @xmath147 which equals to @xmath681/2(+1/2 ) for @xmath145=+@xmath148(@xmath149 ) . for all bands considered in the present study , the neutron configurations are fixed at ( 22,22,21,21 ) . the yrast configuration in @xmath1 tb corresponds to a hole in the 16-th orbital in the + + block , 16@xmath150 , which in the standard notation is described as the @xmath151{3/2}$ ] ( @xmath145=@xmath152 ) or @xmath153 nilsson intruder orbital . since in the experiment the angular velocity is associated with half of the transition energy @xmath142 , and the above pairs of bands have identical transition energies , the relative alignments must be close to a half - integer value . this is simply a consequence of the fact that the spins in the odd and even nuclei are half - integer and integer , respectively . departures from half - integer relative alignments can only be caused by differences in the @xmath7-ray transition energies , which are very small for the two pairs of bands studied here . since the values of spins have not yet been measured , the relative alignments are known up to an additive integer value and a theoretical input is necessary if one wishes to put forward one value or another . already at a very early stage of the identical band studies , it has been suggested @xcite that the first excited bands @xmath1tb(2 ) and @xmath3gd(2 ) correspond to the @xmath9{1/2}$ ] ( @xmath145=@xmath152 ) holes ( signature index @xmath147=@xmath681/2 ) in the yrast states of the respective cores ( see also ref . @xcite ) . in the present study we have followed this interpretation and we have constructed the excited bands by creating a hole in the 17-th orbital of the @xmath68 + block , 17@xmath154 . in all the ws and hf cranking calculations , the single - particle routhian @xmath9{1/2}$ ] ( @xmath145=@xmath152 ) increases with rotational frequency with a constant slope of about + 0.5@xmath131 , i.e. , it has the single - particle alignment ( the average value of the projection of angular momentum on the rotation axis ) close to @xmath680.5@xmath131 . therefore , a hole created in this orbital must lead to relative alignment of about + 0.5@xmath131 . in this paper we adopt this half - integer value for fixing the unknown integer additive constant , @xmath155=+0.5@xmath131 , required to extract relative alignment from experimental data . within this choice , the experimental average relative alignment for the @xmath1tb(2)@xmath2dy(1 ) pair of bands equals to + 0.564(18)@xmath131 . the error given here is the average error resulting from experimental errors of transition energies . the experimental relative alignment for the @xmath3gd(2)@xmath1tb(1 ) pair increases slightly with the angular velocity and has the average value of + 0.479(14)@xmath131 . in fact , experimentally , only the average departures from the half - integer constant @xmath155 , @xmath156=@xmath155 + 0.064(18)@xmath131 and @xmath156=@xmath155@xmath680.021(14)@xmath131 , are established . in figs . [ fig05 ] and [ fig06 ] we show results of calculations for the @xmath1tb(2)@xmath2dy(1 ) and @xmath3gd(2)@xmath1tb(1 ) pairs of bands , respectively . the complete skyrme energy - density functionals of the siii , skm * , and skp forces have been used . parts ( a ) of these figures ( bottom ) present the dynamical moments @xmath0 as functions of the angular velocity @xmath132 . for @xmath2dy(1 ) , fig . [ fig05 ] , all forces overestimate the experimental values of @xmath0 by about 5 - 10% . all three forces give very similar results , within 2@xmath129mev@xmath130 . a similarity of results obtained for different forces is also visible in table [ tab6 ] , where we give the values of proton quadrupole moments calculated at @xmath132=0.5mev/@xmath131 . the siii interaction gives values larger by only 0.30.4b as compared to those given by skm * , while the skp force leads to intermediate results . a similarity of results for changes of proton quadrupole moments induced by creating holes in proton orbitals is even more pronounced . a polarization by the @xmath9{1/2}$ ] ( @xmath145=@xmath152 ) hole gives , for different forces , the changes between 0.12b and 0.15b . that induced by the @xmath151{3/2}$ ] ( @xmath145=@xmath152 ) hole gives values between @xmath680.94b and @xmath681.05b . the hf proton quadrupole moment of @xmath2dy(1 ) agrees very well with the result obtained within the nilsson single - particle potential @xcite . in fig . [ fig06](a ) we show the dynamical moments @xmath0 calculated for the @xmath1tb(1 ) band . due to the hole in the @xmath151{3/2}$ ] ( @xmath145=@xmath152 ) intruder orbital , @xmath0 decreases here with @xmath132 much faster than that of the @xmath2dy(1 ) band . however , at @xmath132@xmath1570.45mev/@xmath131 one obtains values of @xmath0@xmath15794@xmath129mev@xmath130 which are almost identical for all the forces and for both bands . this contradicts simple perturbative estimates . indeed , a hole in the @xmath2dy(1 ) core should in principle cause a decrease of the moment of inertia due to the smaller mass and deformation of @xmath1tb(1 ) , and also due to the fact that the @xmath151{3/2}$ ] ( @xmath145=@xmath152 ) routhian has negative second derivative with respect to @xmath132 and , therefore , the hole in this orbital should bring negative perturbative contribution to @xmath0 . nevertheless , at @xmath132@xmath1570.45mev/@xmath131 the @xmath0 values calculated for @xmath2dy(1 ) and @xmath1tb(1 ) are the same . this illustrates the fact that the polarization effects obtained by selfconsistent calculations do not necessarily follow perturbative estimates . in @xmath1tb(1 ) , the siii and skm * forces give a better agreement with data than skp . this can be attributed to a different effective masses , @xmath136=1 for skp and 0.760.79 for siii and skm * , which leads to different time - odd components in the mean field ( see the next section ) . then , the interaction between the intruder orbital @xmath151{3/2}$ ] ( @xmath145=@xmath152 ) and the time - odd mean field is modified , and gives a more significant departure from experiment . in parts ( b ) of figs . [ fig05 ] and [ fig06 ] we present the relative dynamical moments @xmath144 calculated for the pairs of bands @xmath1tb(2)@xmath2dy(1 ) and @xmath3gd(2)@xmath1tb(1 ) , respectively . one should note that the scale in ( b ) is enlarged five times with respect to that in ( a ) . at @xmath132@xmath1580.3mev/@xmath131 , both pairs of bands have dynamical moments identical up to about 1@xmath129mev@xmath130 , with lighter isotones having slightly larger values . in the scale of ( a ) this would lead to curves identical up to the size of the data marks . this result confirms the observation @xcite that the identity of dynamical moments can in fact be obtained in self - consistent theories , and does not necessarily follow semiclassical @xcite or perturbative estimates . the experimental values of @xmath144 are not shown in parts ( b ) , because they would be scattered between @xmath1591.5@xmath129mev@xmath130 with errors between 1 and 2.5@xmath129mev@xmath130 , i.e. , in the scale of ( b ) they would cover the whole presented region of @xmath144 . the calculated relative alignments @xmath143 , shown in parts ( c ) of figs . [ fig05 ] and [ fig06 ] , do not reproduce experimental results with a sufficient precision . for the @xmath1tb(2)@xmath2dy(1 ) pair [ fig . [ fig05](c ) ] , in the region of @xmath132 where the data are available , one obtains a gradual increase of @xmath143 by almost 0.5@xmath131 . for the siii and skp forces the value obtained at @xmath132=0.3mev/@xmath131 is correct , but for skm * the whole curve is additionally shifted up by about 0.5@xmath131 . for the @xmath3gd(2)@xmath1tb(1 ) calculations this increase becomes smaller ( siii and skp ) and similar to the small increase seen in the experimental data . however , the values of @xmath143 are still slightly ( siii and skp ) or significantly ( skm * ) too large . values of relative alignments can be translated into the differences @xmath160 of @xmath7-transition energies between the two bands . using a linear local expansion of spin as function of the angular frequency we obtain that @xmath161 where @xmath162 is the difference of frequencies at spins of physical states in two nuclei . hence the departures of calculated relative alignments @xmath143 from @xmath155=0.5@xmath131 correspond to the values of @xmath163 between 0 and 10kev , while the measured values are between 0 and 2kev . in the following two sections we present calculations obtained with modified skyrme functionals . in this way we try to analyze the influence of the time - odd terms on the relative alignments discussed here . as discussed above , the gauge - invariance conditions ( [ eq215a])-([eq215c ] ) restrict values of 6 time - odd coupling constants and leave a freedom to modify the values of @xmath49 and @xmath59 . if one decides to depart from the complete skyrme - interaction functional , in which the time - odd coupling constants are uniquely defined by the time - even ones ( appendix [ appa ] ) , one may , in principle , use arbitrary values of @xmath49 and @xmath59 . however , independent variations of these coupling constants have never been considered in the literature , and their effects are , up to now , unknown . therefore , in the present study we restrict our analysis to the functionals in which one or more of the skm * coupling constants , table [ tab1 ] , are assumed to be equal to zero . moreover , in order to further restrict the number of possible variants , we only consider simultaneous modifications of the isoscalar and isovector coupling constants of a given species . in the frame of gauge - invariant functionals , this leaves us a possibility of putting @xmath49=0 and/or @xmath59=0 . another modified gauge - invariant functional can be obtained by removing the term @xmath70@xmath71@xmath72@xmath68@xmath73 , i.e. , by putting @xmath164=@xmath165=0 , in accordance with eq . ( [ eq215b ] ) . this leads to a modification of time - odd _ and _ time - even mean fields . however , the term @xmath73 was anyhow neglected in most parametrizations o the skyrme forces used for time - even studies , and in particular in skm*. therefore , below we discuss four possibilities corresponding to ( i ) the complete functional , ( ii ) @xmath60=@xmath65=0 , ( iii ) @xmath59=@xmath164=@xmath65=0 , and then ( iv ) @xmath166=@xmath59=@xmath60=@xmath65=0 . the results are presented in figs . [ fig07 ] and [ fig08 ] for the pairs of bands @xmath1tb(2)@xmath2dy(1 ) and @xmath3gd(2)@xmath1tb(1 ) , respectively . we have studied several other possibilities , but principal conclusions can be drawn from these four cases . for @xmath2dy(1 ) , the dynamical moments @xmath0 , fig . [ fig07](a ) , are ( at high spin ) sensitive only to the @xmath60=@xmath65 coupling constants . the influence of @xmath59 and @xmath49 is visible only at low spins , and moreover , their effects have opposite signs and partially cancel one another . removing from the functional the term @xmath70@xmath71@xmath72@xmath68@xmath73 decreases the dynamical moments , and , at the high - spin - end of the band , brings the calculated value down to the experimental result . at lower spins one obtains a smaller than previously overestimation of the data ( by about 5% ) . a very interesting results is obtained for the relative dynamical moments @xmath144 , fig . [ fig07](b ) . for any studied combination of the time - odd mean fields taken into account , the dynamical moments in both nuclei are almost identical . even if the _ values _ of @xmath0 depend on whether the term @xmath70@xmath71@xmath72@xmath68@xmath73 is taken into account or not , the _ differences _ of @xmath0 do not depend on it at all . this is a rather general observation , valid also for other cases of modified functionals discussed below . it means that the identity of dynamical moments is fairly independent of at least some of the details of effective interactions , and can probably be attributed to rather simple geometric effects . the relative alignments @xmath143 , fig . [ fig07](c ) , very strongly depend on the presence of the @xmath70@xmath71@xmath72@xmath68@xmath73 term and are almost independent of the @xmath54 and @xmath70@xmath71@xmath167 terms . as soon as @xmath164=@xmath65=0 , the value of @xmath143 at @xmath132=0.3mev/@xmath131 comes down to the experimental result . on the other hand , the increase of @xmath143 as function of the angular frequency is still too large , and does not give the experimental identity of the @xmath7-transition energies , eq . ( [ eq216 ] ) . in fig . [ fig08 ] , for the pairs of bands @xmath3gd(2)@xmath1tb(1 ) we present a similar analysis of the role of different time - odd terms in the skyrme functional . in this case , the influence of the @xmath70@xmath71@xmath72@xmath68@xmath73 term on @xmath0 is weaker than for the @xmath1tb(2)@xmath2dy(1 ) pair , while its removal leads to a good agreement with experiment , with only a slightly too slow decrease of @xmath0 as function of the angular velocity . at the same time the relative alignments become much closer to experimental data , but an overestimation by a few tenth of @xmath131 persists at all angular frequencies . the influence of the other two time - odd terms considered here , @xmath54 and @xmath70@xmath71@xmath167 , is much weaker , although a non - negligible influence on relative alignments can be seen . we conclude this section by noting that the importance of the gauge - invariant time - odd terms for the rotational properties is _ not _ simply related to the order of the given term . in fact , the zero - order term @xmath54 , which depends on the density of spin , has a very small influence on the results . at the same time , two second - order terms , @xmath70@xmath71@xmath167 and @xmath70@xmath71@xmath72@xmath68@xmath73 , which in the previous studies have been simultaneously neglected , have a small and rather large influence , respectively . on the other hand , the latter term , which in principle can not be neglected because of its magnitude , leads to larger deviations from experimental data , as compared with calculations which do disregard it . of course , the present analysis is not sufficient for a more accurate derivation of the magnitude of coupling constants from experiment . this can only be done by a simultaneous consideration of many different available data , and moreover , should also involve a careful readjustment of properties of the time - even components of the mean fields . in this section we present results of calculations for skyrme functionals with coupling constants which do not obey the gauge - invariance conditions ( [ eq215a])-([eq215c ] ) . in order to simplify the discussion , the gauge - invariant time - odd terms @xmath54 and @xmath70@xmath71@xmath167 are disregarded , i.e. , @xmath49=@xmath59=0 . similarly as in the previous section , we only consider simultaneous modifications of the isoscalar and isovector coupling constants . this leaves us with a possibility of modifying the time - odd coupling constants , @xmath61 , @xmath164 , and @xmath62 , while leaving unchanged their time - even gauge - invariance partners , @xmath64 , @xmath65 , and @xmath66 . first we have checked the separate role of the time - even and time - odd parts in the term @xmath70@xmath71@xmath72@xmath68@xmath73 which has been discussed in the previous section . it turned out that the time - even part , @xmath73 , influences the results for rotational properties in a negligible way . therefore , for all practical purposes the results discussed previously should be attributed to the time - odd part , @xmath70@xmath71@xmath72 . this allows us to concentrate here on the two other time - odd terms , @xmath69 and @xmath70@xmath71@xmath168@xmath78@xmath169 . it is worth recalling at this point that in the gauge - invariant energy functional ( [ eq109c ] ) the @xmath69 term is associated with the effective - mass term @xmath67 , and the term @xmath70@xmath71@xmath168@xmath78@xmath169 comes together with the spin - orbit term @xmath21@xmath168@xmath71@xmath170 . therefore , the first one is absent in any theory which has the effective mass @xmath4 equal to the free mass @xmath5 . this is the case for all studies based on phenomenological single - particle potentials , like the nilsson or ws ones . on the other hand , the spin - orbit term is crucially important for a correct ordering of single - particle shells and is always taken into account in realistic calculations . however , the studies based on phenomenological potentials always disregard its time - odd gauge partner , and therefore should be considered as gauge - violating approaches , similar to the selfconsistent calculations considered in this section . in figs . [ fig09 ] and [ fig10 ] we present the results obtained for the same pairs of superdeformed bands as considered above , namely , for @xmath1tb(2)@xmath2dy(1 ) and @xmath3gd(2)@xmath1tb(1 ) , respectively . open symbols refer to the functionals with @xmath171=0 and squares to those with @xmath62=0 . consequently , the full circles repeat here the same results as those shown by full circles in figs . [ fig07 ] and [ fig08 ] . on the other hand , open squares correspond to the skyrme functionals with _ all _ time - odd terms neglected . as expected from the semiclassical analysis @xcite , omitting the @xmath69 term leads to much smaller values of @xmath0 . the obtained decrease is of the order of 10 - 13% , depending on the value of the angular frequency , i.e. , it is much smaller than the decrease of about 40% obtained from non - selfconsistent estimates @xcite . with a very high precision , this decrease is identical for both nuclei in pairs differing by a hole in the @xmath9{1/2}$ ] ( @xmath145=@xmath152 ) orbital . indeed , even if the values of @xmath0 change by as much as 13@xmath129mev@xmath130 , the relative dynamical moments , figs . [ fig09](b ) and [ fig10](b ) , do not change at all , and are always well below 1@xmath129mev@xmath130 . in addition , the @xmath69 term has only a minor influence on the values of relative alignments @xmath143 . the presence of the second term studied here , @xmath70@xmath71@xmath168@xmath78@xmath169 , has very little effect on the values and relative values of @xmath0 . however , its impact on the values of relative alignments is very large . removing this term from the functional , decreases the values of @xmath143 by about 0.5@xmath131 . for the @xmath1tb(2)@xmath2dy(1 ) pair of bands , this leads to relative alignment which has the experimental value at the high - spin end of the band , fig . [ fig09](c ) . at the low - spin end the value of @xmath143 is now underestimated by about 0.3@xmath131 . according to eq . ( [ eq216 ] ) , this gives the differences of the @xmath7-ray energies increasing from about @xmath686kev to 0 , i.e. , the identity of @xmath142 is still not reproduced with a sufficient precision . the sensitivity of results to the @xmath62 coupling constant suggests , however , that a fit of this kind of term to experimental data may , in principle , improve the agreement . as mentioned previously , such a fit should take into account many different bands and can not be based on a rather restricted set of examples discussed here . for the @xmath3gd(2)@xmath1tb(1 ) pair of bands , the removal of the @xmath70@xmath71@xmath168@xmath78@xmath169 term , while keeping the term @xmath69 unchanged , leads to a very good description of the values of @xmath0 , and of the relative alignments @xmath143 simultaneously , fig . [ fig10 ] . in this case , the relative alignments agree with the experiment to better than 0.1@xmath131 , which corresponds to the identity of the @xmath7-ray energies to better than 1kev . in the present study we have applied the hartree - fock cranking method with the skyrme interactions to describe rotational states in selected superdeformed nuclei . we have analyzed in detail properties of two pairs of bands , namely , @xmath1tb(2)@xmath2dy(1 ) and @xmath3gd(2)@xmath1tb(1 ) . experimentally , these bands are pair - wise identical , i.e. , the corresponding @xmath7-transition energies are identical within 2kev each . in agreement with the previous interpretations , we have fixed the configurations of the @xmath1tb(1 ) band as a hole - structure created in the @xmath151{3/2}$ ] ( @xmath145=@xmath152 ) or @xmath153 nilsson intruder orbital with respect to the magic superdeformed @xmath172dy@xmath173 core . similarly , the @xmath1tb(2 ) and @xmath3gd(2 ) bands have been constructed as the @xmath9{1/2}$ ] ( @xmath145=@xmath152 ) hole configurations in the corresponding @xmath2dy(1 ) and @xmath1tb(1 ) cores . the dynamical moments @xmath0 calculated for the @xmath1tb(2)@xmath2dy(1 ) pair of bands and for the @xmath3gd(2)@xmath1tb(1 ) pair of bands are identical within 2@xmath129mev@xmath130 . this result correctly reproduces the identity of these two pairs of bands obtained in experiment . it confirms that the internal structure of the @xmath9{1/2}$ ] ( @xmath145=@xmath152 ) orbital is responsible for the occurrence of these particular identical bands . the obtained identity of @xmath0 does not depend on the version of the skyrme force used , neither it depends on including or disregarding various time - odd components in the mean field of the rotating nucleus . the fact that the two bands calculated in @xmath1 tb , i.e. , @xmath1tb(1 ) and @xmath1tb(2 ) , are _ not _ identical ( similarly as in experiment ) , indicates a crucial role of the single - particle structure of the involved orbitals , and shows that in the hf cranking theory the phenomenon of identical bands is not a generic , built - in - by - assumption result . the relative alignments of the studied pairs of bands are more difficult to reproduce than the simple identity of the dynamical moments . first of all , the results do depend on the version of skyrme interaction and on the time - odd components included in the mean field . when the complete skyrme functionals are used , i.e. , when all time - odd terms are taken into account , the calculated relative alignments do not reproduce the experimental data . the disagreement obtained for the skm * interaction is particularly large , while the siii and skp interactions also give too large relative alignments . within the skm * parametrization , good agreement with data is obtained when all time - odd terms are disregarded except of the one which involves the density of current , @xmath174 , and is related to the effective - mass term by the gauge transformation . this specific result does not imply that the time - odd terms present in the formalism should be generally eliminated from its applications . by studying cases with some time - odd terms removed we only aimed at gaining information which may be useful when improving the parametrisation of effective interactions . as seen from the results obtained in the present study , a readjustment of parameters of effective forces in the time - odd channel seems to be necessary for a detailed description of the identical @xmath7-ray transitions in superdeformed nuclei . however , this should be done by taking into account many available high - spin data simultaneously , and also should involve a readjustment of the time - even part ( the spin - orbit interaction , in particular ) which is responsible for the ordering of the single - particle energies and routhians . the work along these lines is now in progress . we would like to express our thanks to the _ institut du dveloppement et de ressources en informatique scientifique _ ( idris ) of cnrs , france , which provided us with the computing facilities under project no . this research was supported in part by the polish committee for scientific research under contract no . 2 p03b 034 08 . in its standard form , the skyrme interaction ( see e.g. ref . @xcite ) depends on 10 parameters , @xmath175 , @xmath176 , @xmath177 , @xmath178 , @xmath179 , @xmath180 , @xmath181 , @xmath182 , @xmath147 , and @xmath183 . for @xmath178=@xmath180=0 , the energy density which corresponds to the skyrme interaction has been derived in ref . @xcite . in terms of the isovector and isoscalar coupling constants , and for the complete interaction , this energy density is given in eqs . ( [ eq108])([eq109 ] ) . the zero - order coupling constants , @xmath17 and @xmath184 , correspond to the velocity - independent terms of the interaction , and can be expressed by the @xmath175 , @xmath176 , @xmath181 , and @xmath182 parameters , as it shows table [ tab2 ] . the zero - order time - odd coupling constants @xmath49 ( for @xmath15=0 and 1 ) are linear combinations of the zero - order time - even coupling constants @xmath17 ( table [ tab4 ] ) . similarly , the second - order coupling constants , @xmath63 , @xmath64 , @xmath59 , and @xmath164 , correspond to the velocity - dependent terms of the interaction given by the parameters @xmath177 , @xmath178 , @xmath179 , and @xmath180 , and are presented in table [ tab3 ] . the second - order time - odd coupling constants , @xmath59 and @xmath60 , are linear combinations of the second - order time - even coupling constants , @xmath63 and @xmath64 ( table [ tab5 ] ) . another four second - order coupling constants , @xmath65 and @xmath61 , which also depend on the same @xmath177 , @xmath178 , @xmath179 , and @xmath180 parameters , are given by the gauge - invariance conditions ( [ eq215a ] ) and ( [ eq215b ] ) . finally , the second - order coupling constants @xmath66 are given by the spin - orbit term of the skyrme interaction and depend on the parameter @xmath183 , @xmath185=@xmath68@xmath186 and @xmath187=@xmath68@xmath188 , while the other two , @xmath62 , follow from the gauge - invariance condition ( [ eq215c ] ) . an extension of the skyrme energy density , which introduces the coupling constants @xmath189 and @xmath187 independent of one another , is discussed in ref . @xcite p. ring and p. schuck , the nuclear many - body problem ( springer , berlin , 1980 ) . s. berg , h. flocard and w. nazarewicz , ann . nucl . part . sci . * 40 * , 439 ( 1990 ) ; and references therein . m. baranger and m. vnroni , ann . of phys . * 114 * , 123 ( 1978 ) . b. grammaticos and a. voros , ann . of phys . * 123 * , 359 ( 1979 ) ; * 129 * , 153 ( 1980 ) . d.j . thouless and j.g . valatin , nucl . phys . * a31 * , 211 ( 1962 ) . a. bohr and b.r . mottelson , nuclear structure , vol . 2 ( w.a . benjamin , new york , 1975 ) . migdal , nucl . phys . * 13 * , 655 ( 1959 ) . giannoni and p. quentin , phys . rev . * c21 * , 2076 ( 1980 ) . j. dobaczewski and j. skalski , nucl a369 * , 123 ( 1981 ) . z. szymaski , fast nuclear rotation ( clerandon , oxford , 1983 ) . k. neergrd , v.v . pashkevich , and s. frauendorf , nucl . phys . * a262 * , 61 ( 1976 ) . p. bonche , h. flocard and p .- h . heenen , nucl . phys . * a467 * , 115 ( 1987 ) . p. bonche , h. flocard and p .- h . heenen , nucl . phys . * a523 * , 300 ( 1991 ) . b.q . chen , p .- h . heenen , p. bonche , m.s . weiss and h. flocard , phys . rev . * c46 * , 1582 ( 1992 ) . b. gall , p. bonche , j. dobaczewski , h. flocard , and p .- h . heenen , z. phys . * a348 * , 183 ( 1994 ) . h. flocard , p. bonche , p .- h . heenen , and r. mehrem , proceedings of the antibes conference , 1994 . j. terasaki , p .- h . heenen , p. bonche , j. dobaczewski , and h. flocard , to be published . heenen , p. bonche , and h. flocard , submitted to nucl . bonche , h. flocard , and p .- h . heenen , to be published . w. koepf and p. ring , nucl a493 * , 61 ( 1989 ) ; * a511 * , 279 ( 1990 ) . j. knig and p. ring , phys . 71 * , 3079 ( 1993 ) . egido and l.m . robledo , phys . lett . * 70 * , 2876 ( 1993 ) . janssens and t.l . khoo , ann . nucl . part . sci . * 41 * , 321 ( 1991 ) ; and references therein . c. baktash , b. haas , and w. nazarewicz , to be published in ann . rev . nucl . part . sci . ; and references therein . d. vautherin and d.m . brink , phys . * c5 * , 626 ( 1972 ) . engel , d.m . brink , k. goeke , s.j . krieger , and d. vautherin , nucl . a249 * , 215 ( 1775 ) . s. krewald , v. klemt , j. speth , and a. faessler , nucl . phys . * a281 * , 166 ( 1977 ) . pethick , d.g . ravenhall , and c.p . lorentz , urbana preprint p-94 - 08 - 068 . m. beiner , h. flocard , n. van giai and p. quentin , nucl . a238 * , 29 ( 1975 ) . negele and d. vautherin , phys . rev . * c5 * , 1472 ( 1972 ) . j. dobaczewski and j. dudek , to be submitted to computer physics communications . j. dudek , z. szymaski , and t.r . werner , phys . rev . * c23 * , 920 ( 1981 ) . s. wiok , j. dudek , w. nazarewicz , j. skalski , and t. werner , comp . . commun . * 46 * , 379 ( 1987 ) . j. bartel , p. quentin , m. brack , c. guet , and h.b . hkansson , nucl . phys . * a386 * , 79 ( 1982 ) . j. dobaczewski , h. flocard , and j. treiner , nucl . * a422 * , 103 ( 1984 ) . j. dobaczewski , w. nazarewicz , and t.r . werner , physica scripta , in press . et al . _ , lett . * 71 * , 1800 ( 1993 ) . et al . _ , b335 * , 313 ( 1994 ) . b. kharraja _ * b341 * , 268 ( 1995 ) . et al . _ , phys . lett . * 64 * , 1650 ( 1990 ) . w. nazarewicz , p.j . twin , p. fallon , and j.d . garrett , phys . lett . * 64 * , 1654 ( 1990 ) . i. ragnarsson , nucl . phys . * a557 * , 167c ( 1993 ) . k. bencheikh , p. quentin , and j. bartel , nucl . * a571 * , 518 ( 1994 ) . reinhard and h. flocard , orsay preprint 1994 . .coupling constants in the gauge - invariant energy density ( [ eq109c ] ) obtained for the siii , skp , and skm * skyrme interactions . zero - order coupling constants @xmath17 and @xmath49 ar given in mev@xmath71@xmath190 and the remaining ( second - order ) ones are given in mev@xmath71@xmath137 . [ cols= " < , > , > , > , > , > , > " , ]

we consider solutions to the linear wave equation on a non - extremal maximally extended schwarzschild - de sitter spacetime arising from arbitrary smooth initial data prescribed on an arbitrary cauchy hypersurface . ( in particular , no symmetry is assumed on initial data , and the support of the solutions may contain the sphere of bifurcation of the black / white hole horizons and the cosmological horizons . ) we prove that in the region bounded by a set of black / white hole horizons and cosmological horizons , solutions converge pointwise to a constant faster than any given polynomial rate , where the decay is measured with respect to natural future - directed advanced and retarded time coordinates . we also give such uniform decay bounds for the energy associated to the killing field as well as for the energy measured by local observers crossing the event horizon . the results in particular include decay rates along the horizons themselves . finally , we discuss the relation of these results to previous heuristic analysis of price and brady et al .

0709.2766

the introduction of a positive cosmological constant in the einstein equations of general relativity gives rise to a wide variety of new interesting solution spacetimes , in particular , spacetimes containing both `` black hole '' and `` cosmological '' regions . as in the case of black - hole spacetimes with vanishing cosmological constant , the stability of these spacetimes as solutions to the einstein equations is a fundamental open problem of gravitational physics . yet even the simplest questions concerning the behaviour of linear waves on such spacetime backgrounds today remain unanswered . in this paper , we initiate in the above context the mathematical study of decay for solutions to the linear wave equation . the simplest family of black - hole spacetimes with positive cosmological constant is the so - called _ schwarzschild - de sitter _ family . if the cosmological constant @xmath2 is considered fixed , this is a @xmath3-parameter family of solutions @xmath4 to the _ einstein vacuum equations _ @xmath5 with parameter @xmath6 , called the _ mass_. we shall consider only the non - extremal black - hole case , corresponding to parameter values @xmath7 as with the schwarzschild family , the first manifestation of the schwarzschild - de sitter family of solutions was an expression for the metric in local coordinates , in this case first published in 1918 by kottler @xcite , and independently by weyl @xcite , in the form @xmath8 here @xmath9 denotes the standard metric on the unit @xmath10-sphere . the global structure of maximal spherically symmetric vacuum extensions of such metrics was only understood much later @xcite based on the methods of formal penrose diagrams introduced by b. carter . in fact , maximally extended spherically symmetric vacuum spacetimes @xmath4 with various different topologies can be constructed , all of which equally well merit the name `` schwarzschild - de sitter with parameter @xmath6 and cosmological constant @xmath11 '' . such solutions @xmath4 all share the property that the universal cover @xmath12 of the @xmath10-dimensional lorentzian quotient @xmath13 consists of an infinite chain of regions as depicted in the penrose diagram below : @xmath14 the results of this paper do not depend on the topology , and for definiteness , one may assume in what follows that the name `` schwarzschild de - sitter '' and the notation @xmath4 refer to the spacetime with quotient precisely the universal cover depicted above . it is then the wave equation @xmath15 on this background @xmath4 whose mathematical study we wish to initiate here . there is already a rich body of heuristic work on this problem in the physics literature . ( see section [ discussion ] below for a discussion . ) the motivation for the study of @xmath16 in the present context is multifold . in particular , as in the case of vanishing cosmological constant , studied in our previous @xcite , we believe that proving bounds on decay rates for solutions to @xmath16 is a first step to a mathematical understanding of non - linear stability problems for spacetimes containing black holes , that is to say , to the problem of stability in the context of the dynamics of @xmath17 . for a more detailed discussion , we refer the reader to the introductory remarks of @xcite . we are interested in solutions of @xmath16 arising from suitably regular initial data prescribed on a cauchy surface @xmath18 of @xmath19 . for future applications to non - linear stability problems , it is crucial that all assumptions have a natural geometric interpretation independent of special coordinate systems . moreover , our primary concern in this paper is the region @xmath20 bounded by a set of black / white hole horizons @xmath21 and cosmological horizons @xmath22 : @xmath23 as depicted below , @xmath24 , etc . ) , and penrose diagrams . see @xcite . ] @xmath25 by causality , the global behaviour of @xmath1 in @xmath20 can be understood independently of the behaviour near @xmath26 and @xmath27 . the behaviour in say @xmath28 is completely determined by the behaviour of appropriate initial data on @xmath29 . is not necessarily @xmath30 . ] we review briefly in the next paragraph the solvability and domain of dependence property for the initial value problem for @xmath16 . let @xmath31 be a smooth cauchy surface and let @xmath32 denote the future - directed unit normal of @xmath18 . for @xmath33 , let @xmath34 be an @xmath35 function and @xmath36 an @xmath37 function . then there exists a unique global solution @xmath38 of @xmath0 such that for all smooth spacelike hypersurfaces @xmath39 with future directed unit normal @xmath40 , @xmath41 , @xmath42 , and @xmath43 , @xmath44 . moreover , if @xmath45 is closed and @xmath46 , @xmath47 are two such solutions corresponding to data @xmath48 , @xmath49 such that @xmath50 , @xmath51 , then @xmath52 on @xmath53 . in particular , setting @xmath54 , we obtain that @xmath52 on @xmath55 . since our results will be quantitative , we need to introduce relevant norms on the compact manifold with boundary @xmath29 . let @xmath56 denote the riemannian @xmath57 norm on @xmath29 . this induces a norm on sections of the tangent bundle , a norm we will denote also by @xmath56 . if @xmath58 , @xmath59 , then let us denote by @xmath1 the unique solution of @xmath0 corresponding to initial data @xmath60 . let us define now for all real @xmath61 the quantity @xmath62 where @xmath63 denotes the projection of @xmath1 to the @xmath64th eigenspace of @xmath65 , i.e. the @xmath64th spherical harmonic of @xmath1 . the function @xmath66 is discussed in section [ cordsec ] . if @xmath18 itself is spherically symmetric , then we may replace @xmath63 , @xmath67 be @xmath68 and @xmath69 , and the above expression is a sum of integrals on initial data . for general @xmath18 , a sufficient condition for the finiteness of @xmath70 is that @xmath71 , @xmath72 . in the case @xmath73 an integer , we can characterize @xmath74 geometrically as follows . let @xmath75 , @xmath76 denote a basis of killing fields generating the lie algebra @xmath77 associated to the spherical symmetry of @xmath4 . we call @xmath75 _ angular momentum operators_. it easily follows that @xmath78 again , if @xmath18 itself is spherically symmetric , we may replace @xmath1 with @xmath34 in the first term , and remove the @xmath79 from the second , replacing @xmath1 with @xmath80 . the main result of this paper is contained in the following [ gendata ] let @xmath4 denote the schwarzschild - de sitter spacetime with parameter @xmath6 and cosmological constant @xmath11 satisfying @xmath81 and let @xmath18 be a cauchy surface for @xmath19 . let @xmath82 denote a region as defined in @xmath83 and let @xmath61 . then , there exist constants @xmath84 depending only on @xmath85 , @xmath6 , @xmath11 , and the geometry of @xmath86 such that for all solutions @xmath1 of the wave equation @xmath0 on @xmath19 such that @xmath87 is finite , where @xmath88 , @xmath89 , and for all achronal hypersurfaces @xmath90 , the bound @xmath91 holds , where @xmath92 and @xmath93 denote fixed eddington - finkelstein advanced and retarded coordinates . although these coordinates are only defined in @xmath94 , the statements @xmath95 , @xmath96 can be interpreted in all of @xmath20 in view of conventions @xmath97@xmath98 . ] , @xmath99 , @xmath100 , etc , @xmath101 denotes the killing field coinciding in the interior of @xmath20 with @xmath102 , @xmath103 denotes the standard energy - momentum tensor , and @xmath104 is the future - directed unit normal wherever @xmath105 is spacelike , in which case the integral is taken with measure the induced volume form . and the measure of integration for general achronal @xmath105 can be derived by a limiting procedure . ] in addition , @xmath95 holds if @xmath101 is replaced by the vector field @xmath106 defined in section [ smesec ] . if @xmath107 , then the pointwise bound latexmath:[\[\label{ptwise1 } holds in @xmath55 , where @xmath109 is a constant satisfying @xmath110 * in particular , the theorem applies to arbitrary smooth initial data @xmath111 , @xmath112 , where @xmath85 can be taken arbitrarily large . * there are no unphysical assumptions regarding vanishing of @xmath1 at the _ sphere of bifurcation _ of the horizons , i.e. at the sets @xmath113 and @xmath114 . the decay rates @xmath96 , @xmath115 are uniform , i.e. they hold up to and including the horizons , setting @xmath116 or @xmath117 . in particular , @xmath105 in @xmath115 can be taken ( as depicted below ) @xmath118 to contain subsets of @xmath119 and/or @xmath120 . the statement of theorem [ gendata ] should be compared with the results of our previous @xcite concerning the wave equation on a schwarzschild exterior . recall that in the region @xmath121 , the schwarzschild metric is given by the expression @xmath122 for @xmath123 , @xmath124 . the penrose diagramme of the closure of this region in the maximally extended schwarzschild spacetime is given below : @xmath125 in @xcite , an analogue of @xmath95 is proven for all @xmath126 and spherically symmetric initial data . modulo an @xmath127 , this result is expected to be sharp , as it not expected to be true for @xmath128 , in view of heuristic arguments due to price @xcite . in @xcite , an analogue of @xmath95 is proven for @xmath129 for arbitrary , not necessarily spherically symmetric , initial data . in view of the fact that solutions of the wave equation vanish on @xmath130 , the results of @xcite allow one to obtain the uniform pointwise decay rate @xmath131 . as a uniform decay bound in @xmath93 , this decay rate is in fact sharp . the loss of angular derivatives in the result of theorem [ gendata ] can be more precisely quantified by decomposing @xmath1 into spherical harmonics . each spherical harmonic @xmath132 decays at least exponentially , but the bound on the exponential rate obtained here _ decreases _ inverse quadratically in the spherical harmonic number . we have the following [ second ] let @xmath4 , @xmath18 , @xmath20 be as in theorem [ gendata ] . then there exists a constant @xmath133 depending only on @xmath6 and @xmath11 , and @xmath134 depending only on @xmath6 , @xmath11 , and the geometry of @xmath29 , such that for all @xmath132 solutions of the wave equation on @xmath19 with spherical harmonic number @xmath64 with @xmath135 and all achronal hypersurfaces @xmath136 , the bound @xmath137 holds for all @xmath138 , and , again as before , also with @xmath101 above replaced with @xmath106 defined in section [ smesec ] . in addition , the pointwise bounds @xmath139 for @xmath140 , and @xmath141 for @xmath142 , hold in @xmath55 , where @xmath109 is a constant satisfying @xmath143 the above theorem can easily be seen to imply theorem [ gendata ] . in this paper , we insist on a framework of proof that in principle may have relevance to the non - linear stability problem , that is to say , the problem of the dynamics of @xmath17 starting from initial data close to those induced on a cauchy hypersurface @xmath18 of schwarzschild - de sitter . this leads us to try to exploit compatible currents . in this section , we will describe this general approach , and the natural relation of the currents we will define with various geometric and analytical aspects of the problem at hand . for quasilinear hyperbolic systems ( like @xmath17 ) in @xmath144 dimensions , all known techniques for studying the global dynamics are based on @xmath57 estimates . in the lagrangian case , the origin of such estimates can be understood geometrically in terms of _ compatible currents _ ( see christodoulou @xcite ) . these are @xmath3-forms @xmath145 such that at each point @xmath146 , both @xmath145 and the divergence @xmath147 depend only on the @xmath3-jet of @xmath1 . an important class of these are the currents @xmath148 obtained by contracting the energy momentum tensor @xmath149 with an arbitrary vector field @xmath150 . see section [ ccsec ] for a discussion in the context of the linear wave equations @xmath0 studied here . all estimates in this paper are obtained by exploiting the integral identities @xmath151 corresponding to compatible currents of the form @xmath148 and straightforward modifications thereof @xmath152 , where the vector fields @xmath153 are directly related to the geometry of the problem , and the region @xmath154 is suitably chosen . the timelike hypersurface @xmath156 is known as the _ photon sphere_. this has the ominous property of being spanned by null geodesics . if additional regularity is not imposed , then it is clear by a geometric optics approximation that solutions of the wave equation can concentrate their energy along such geodesics for arbitrary long times , and one can thus not achieve a quantitative bound for the rate of decay in terms of initial energy alone . in particular , @xmath95 can not hold for @xmath157 if @xmath158 is replaced by @xmath159 . it is truly remarkable that this obstruction arising from geometrical optics is captured , and quantified , by a current @xmath145 associated to a vector field @xmath153 of the form @xmath160 for a well - chosen function @xmath161 . the story is not entirely straightforward , however . the desired current is in fact not precisely of the form @xmath148 , but a modification thereof , to be denoted @xmath162 , which is associated in a well defined way to a collection of vector fields @xmath163 . the current is defined by summing over currents @xmath164 which act on individual spherical harmonics @xmath63 . the current @xmath165 yields a nonnegative @xmath166 , modulo an error term supported near the horizons . in a first approximation , we may pretend that in fact @xmath167 , but degenerates ( in regular coordinates ) near the horizon . the identity @xmath168 can then be used as an estimate for its left hand side , in view of the fact that its right hand side will in fact be bounded by the flux of @xmath169 , for the killing field @xmath170 , which is conserved . the role of the photon sphere will be exemplified by the degeneration at @xmath156 of the quantity controlled by this spacetime integral . in order to obtain decay results from the above , one would have to gain information about the quantity estimating the boundary terms namely @xmath171 , from the control of spacetime integral . the difficulty for this is that the spacetime integral estimates one obtains degenerate at the photon sphere @xmath156 and at the horizons . this does not allow one to control @xmath172 there . the problem at the photon sphere is cured by applying the estimate also to angular derivatives . it is here that the argument `` loses '' an angular derivative . it is this loss that leads to the form of decay proven in @xmath95 . the problem on the horizon , on the other hand , turns out to be illusionary . the horizon is in fact a very favourable place for estimating the solution ! for this , we will need to consider the `` local observer '' vector fields @xmath173 , @xmath174 , to be described in the next section . the heuristic mechanism ensuring decay near the horizons has been understood for many years , and is known as the _ red shift effect_. this is typically described in the language of geometric optics . if two observers @xmath177 and @xmath178 cross the event horizon at advanced times @xmath179 , and @xmath177 sends a signal to @xmath178 at a certain frequency , as he ( @xmath177 ) measures it , then the frequency at which @xmath178 receives it is exponentially damped in the quantity @xmath180 . it turns out that this exponential damping property can be captured by the integral identities @xmath168 corresponding to the currents @xmath181 and @xmath182 associated to vector fields @xmath173 , @xmath183 , defined in section [ ysec ] . these vector fields are supported near the horizons @xmath119 , @xmath120 , respectively . the estimates @xmath168 corresponding to the currents @xmath181 , @xmath182 fulfill the double role of ( a ) correcting for the error region where @xmath184 by dominating this term near the horizon by @xmath185 and ( b ) controlling the spacetime integrated energy measured by local observers near the horizon . the choice of @xmath173 , @xmath183 is delicate , because there is an `` error region '' where @xmath186 , which must be controlled with the help of the currents of the previous section . the use of the currents corresponding to @xmath187 , @xmath173 and @xmath183 are thus strongly coupled . to see the above arguments in context , the reader may wish to compare with our previous @xcite , where versions of the currents @xmath155 , @xmath175 are also employed . the relation of our arguments with the physical mechanisms at play are in fact much clearer in the present paper , than in @xcite . this is due on the one hand to the absence here of the morawetz - type vector field ( denoted @xmath188 in @xcite ) , and , on the other hand , to the relative simplicity here in the construction of the current @xmath155 . we give here some comments on these points . the morawetz vector field employed in @xcite is a highly unnatural quantity at the horizon from the geometric point of view . on the other hand , in view of its weights , it somewhat magically captures a polynomial ( as opposed to the proper exponential ) version of the red shift . the pointwise decay rates achieved via @xmath188 at the horizon are worse than the decay rates away from the horizon , but sufficient if one is only interested in the behaviour of the solution away from the horizon . ( see also @xcite . ) in our @xcite , uniform decay rates up to the horizon were indeed obtained with the help of @xmath175 . but these estimates could be obtained _ a posteriori_. from the point of view of the non - linear stability problem , this decoupling appears to be an exceptional feature . it is in this sense that the scheme proposed in the present paper is perhaps more naturally connected to the geometry of general black holes . the second point to be made here concerns the construction of @xmath155 . in @xcite , positivity of the analogue of what we denote here @xmath166 relied on an unmotivated recentring and rescaling of the derivatives of the functions @xmath189 which obscured perhaps the fundamental connection with the photon sphere . here , this connection appears much more clear . of course , this is at the expense of having to bound @xmath190 from @xmath191 and @xmath192 . this should in no way be thought of as a disadvantage . the red - shift effect has a lot to offer . it should be used and not obscured . as noted above , the study of the asymptotic behaviour of solutions to @xmath0 on both schwarzschild and schwarzschild - de sitter backgrounds has a long tradition in the physics literature . in the schwarzschild case , the pioneering heuristic study is due to price @xcite . see also @xcite . for the schwarzschild - de sitter case , there is numerical work of brady et al @xcite , the subsequent @xcite , and references therein . the above studies are based entirely upon decomposition of @xmath1 into spherical harmonics . the results of these heuristics or numerics are typically presented in terms of the asymptotic behaviour of the tail : @xmath193 for schwarzschild , where @xmath194 is fixed in the first formula , @xmath195 in the second , and @xmath196 in the third , and @xmath197 for schwarzschild - de sitter and @xmath140 , where @xmath198 is fixed in the first formula , and @xmath195 in the second , and @xmath199 in the third . at first glance , statements @xmath200 may appear stronger than what is actually proven in theorem [ second ] . as quantitative statements of decay , however , statements @xmath201 and @xmath200 are in fact much weaker than what has now been mathematically proven , here and in @xcite . for , rewriting , in particular , the first formula of @xmath200 as @xmath202 depends on , indeed , if there is any bound on @xmath203 provided by some norm of initial data , and if so , what is the behaviour as @xmath204 . this does not concern a mathematical pathology , but is intimately connected with the physical effect caused by the photon sphere . indeed , a geometric optics approximation shows easily that if @xmath205 is to hold and if @xmath203 is to depend , say , on the initial energy of the spherical harmonic , then @xmath206 as @xmath207 . _ it is the rate of this divergence that would then determine the decay rate ( if any ) for @xmath1 . _ if one is interested in quantitative statements of decay , a statement like @xmath200 provides no more information than the statement @xmath208 it is worth noting that the above statement at the level of individual spherical harmonics , together with the ( uniform ) boundedness was shown for schwarzschild in fundamental work of kay and wald @xcite . our @xcite gives an alternative proof not relying on the discrete symmetries of the maximal development . for schwarzschild - de sitter , the statement @xmath209 of course follows from theorem [ gendata ] for any @xmath107 . we have not found another statement of this in the literature . ] result latexmath:[\[\label{ubd } can indeed be used to show , for fixed @xmath66 , the statement @xmath211 for the total @xmath1 . this can be termed the statement of ( uniform ) decay without a rate . thus it is truly only @xmath212 , and not the results of @xcite or theorem [ gendata ] , that can be said to be suggested by heuristic and numerical studies . results like @xmath212 or even just @xmath209 are sometimes referred to as `` linear stability '' in the physics literature . is termed `` linear stability '' . such a result does not even imply @xmath209 . it is in fact entirely consistent with the statement @xmath213 ! ] one should keep in mind , however , that were @xmath212 the sharp decay result , it would in fact suggest _ instability _ for schwarzschild or schwarzschild - de sitter once one passes to the next order in perturbation theory . at very least , it would exclude all known techniques for proving non - linear stability for supercritical non - linear wave equations like @xmath17 . * it is only quantitative uniform decay bounds with decay rate sufficiently fast , such that moreover the bound depends only on a suitable norm of initial data , which indeed can the thought of as suggestive of non - linear stability . * one should thus be careful in associating the heuristic and numerical tradition exemplified in @xcite with the conjecture that black holes are stable . while the final version of this manuscript was being prepared , an interesting preprint @xcite appeared addressing a special case of the problem under consideration here with the methods of time - independent scattering theory . the special case where @xmath1 is not supported at @xmath214 and @xmath114 is considered and quantitative exponential decay bounds are proven for @xmath215 in the coordinate @xmath216 , where one must restrict to @xmath217 $ ] , for @xmath218 , @xmath219 . the bounds lose only an @xmath127 of an angular derivative , but depend on @xmath220 , @xmath221 , and the initial support of @xmath1 in an unspecified way . the work @xcite depends in an essential way on a previous detailed analysis of s barreto and zworski @xcite concerning resonances of an associated elliptic problem . for the special case of the data considered in @xcite , given that result , then the estimates of the present paper , in particular , those provided by the currents @xmath175 , @xmath176 , can be applied _ a posteriori _ to obtain uniform ( i.e. holding up to the horizons ) exponential decay bounds . the authors thank the massachusetts institute of technology for hospitality in the spring of 2006 where this research began . is supported by a clay research scholarship . i.r . is supported in part by nsf grant dms-0702270 . we refer the reader to the references @xcite for detailed discussions of the geometry of schwarzschild - de sitter . we recall that so - called schwarzschild coordinates @xmath222 map @xmath94 onto @xmath223 , in which the metric takes the form @xmath122 . let the choice of the @xmath216 coordinate be fixed . here @xmath224 denote the two positive roots of the equation @xmath225 the function @xmath66 can be given a geometric interpretation @xmath226 where here @xmath227 is the natural projection . thus @xmath66 can be defined as a smooth function on all of @xmath19 . it is known as the _ area - radius function_. this function also clearly descends to @xmath228 . the @xmath222 coordinates degenerate along the horizons @xmath229 and @xmath230 , on which @xmath231 , @xmath232 , respectively . it is immediate from the explicit form of the metric that the vector field @xmath233 is killing in @xmath94 . this extends to a globally defined killing field @xmath170 on @xmath4 , which is null along @xmath229 and @xmath234 , and vanishes along @xmath235 and @xmath236 . we now proceed to define two related coordinate systems on @xmath94 . let us denote the unique negative root of @xmath238 as @xmath239 , and let us set @xmath240 and similarly @xmath241 , @xmath242 . we now set @xmath243 where @xmath244 is a constant we may choose arbitrarily . for convenience , let us choose @xmath244 so that @xmath245 when @xmath156 , the so - called _ photon sphere_. we call the coordinates @xmath237 so - defined _ regge - wheeler coordinates_. from regge - wheeler coordinates @xmath237 , we can define now _ retarded and advanced eddington - finkelstein coordinates _ @xmath92 and @xmath93 , respectively , by @xmath247 and @xmath248 these coordinates turn out to be null : setting @xmath249 the metric takes the form @xmath250 we shall move freely between the two coordinate systems @xmath237 and @xmath246 in this paper . note that in either , region @xmath94 is covered by @xmath251 . by appropriately rescaling @xmath92 and @xmath93 to have finite range , one can construct coordinates which are in fact regular on @xmath252 and @xmath253 . by a slight abuse of language , one can parametrize the future and past horizons in our present @xmath246 coordinate systems as @xmath254 @xmath255},\ ] ] @xmath256 @xmath257 under these conventions , the statements of theorem [ gendata ] can be applied up to the boundary of @xmath20 . finally , we collect various formulas for future reference : @xmath258 @xmath259 @xmath260 @xmath261 @xmath262 @xmath263 @xmath264 @xmath265 @xmath266 here @xmath267 denotes the induced covariant derivative on the group orbit spheres . as discussed in the introduction , the results of this paper will rely on @xmath57-based estimates . such estimates arise naturally in view of the lagrangian structure of the wave equation . we review briefly here . let @xmath1 be a solution of @xmath0 . in general coordinates , the _ energy - momentum tensor _ @xmath268 for @xmath1 is defined by the expression @xmath269 the tensor @xmath268 is symmetric and divergence - free , i.e. we have @xmath270 for the null coordinate system @xmath271 we have defined , where @xmath272 , @xmath273 denote coordinates on @xmath274 , we compute the components @xmath275 @xmath276 @xmath277 here the notation @xmath278 . note moreover that @xmath279 . let @xmath280 denote an arbitrary vector field . let @xmath281 denote the deformation tensor of @xmath153 , i.e. , @xmath282 in local coordinates we have the following expression : @xmath283 set @xmath284 the relations @xmath285 and @xmath286 give @xmath287 and the divergence theorem applied to an arbitrary region @xmath154 gives @xmath168 . identity @xmath168 is particularly useful when the vector field @xmath153 is killing , for instance the vector field @xmath170 defined previously . for then , @xmath288 and one obtains a _ conservation law _ for the boundary integrals . moreover , when @xmath289 corresponds to two homologous timelike hypersurfaces , the integrands @xmath290 on the right hand side of @xmath291 when properly oriented are positive semi - definite in the derivatives of @xmath1 . were the vector field @xmath170 timelike in all of @xmath20 , then by applying @xmath168 to @xmath1 , @xmath292 , etc . , one could show the uniform boundedness of all derivatives of @xmath1 . since @xmath170 becomes null on @xmath293 , the integrand does not control all quantities on the horizon . it is for this reason that even proving uniform boundedness for solutions of @xmath0 on @xmath20 is non - trivial . ( see @xcite . ) for @xmath1 a solution to @xmath0 , the @xmath3-form @xmath294 defined above has the property that both it and its divergence @xmath295 depend only on the @xmath3-jet of @xmath1 . following christodoulou @xcite , we shall call one - forms @xmath145 and their divergences @xmath296 with the aforementioned property ( thought of as form - valued and scalar valued maps on the bundle of @xmath3-jets , respectively ) _ in the course of this proof we shall require special values @xmath297 , @xmath298 , @xmath299 , satisfying @xmath300 eventually , specific choices of these constants will be made , and these choices will depend only on @xmath6 , @xmath11 . constants @xmath301 , @xmath302 are in fact only constrained by lemma [ vanlem ] . constants @xmath303 , @xmath304 are constrained by the necessity of satisfying proposition [ polukrisimo ] of section [ choicesec ] . to choose @xmath220 , @xmath221 , on the other hand , is more subtle , as we will have to keep track of a certain competition of constants as @xmath220 , @xmath221 vary . we adopt , thus , the convention described in the next section . in all formulas that follow in this paper , constants which can be chosen independently of @xmath307 , @xmath308 shall be denoted by @xmath305 . constants @xmath305 will thus depend on @xmath6 , @xmath11 , @xmath297 , @xmath298 , for @xmath309 , and , after @xmath297 , @xmath298 have been chosen , will depend only on @xmath6 , @xmath11 . constants which depend on @xmath6 , @xmath11 , @xmath307 and @xmath308 and tend to @xmath310 as @xmath311 , @xmath312 will be denoted by @xmath127 . finally , all other constants depending on @xmath6 , @xmath11 , @xmath307 and @xmath308 , will be denoted by @xmath306 . * constants denoted by @xmath306 in principle diverge as @xmath313 , @xmath312 . * we will also use the convention @xmath314 whenever @xmath315 with a constant @xmath305 understood as above . in view of our above conventions , note finally the obvious algebra of constants : @xmath316 , @xmath317 , @xmath318 , @xmath319 , etc . associated to these special values of @xmath66 , we will define a number of cutoff functions . it is convenient to introduce also the notation @xmath320 let @xmath322 be a nonnegative smooth cutoff function which is equal to @xmath3 in @xmath323 $ ] and @xmath310 outside @xmath324 $ ] . for @xmath325 , define @xmath326 clearly @xmath321 has the property that @xmath327 in @xmath328 $ ] and @xmath329 in @xmath330\cup(\frac12 r_{i-1}^*,\infty)$ ] . moreover , @xmath331 as @xmath332 , @xmath333 . now let @xmath336 , @xmath337 be nonnegative smooth cutoff functions such that @xmath338 is @xmath3 in @xmath339 $ ] and @xmath310 in @xmath340 , and @xmath341 is @xmath3 in @xmath342 and @xmath310 in @xmath343 $ ] . now for @xmath344 define @xmath345 clearly , @xmath334 has the property that @xmath346 in @xmath347 $ ] , and @xmath310 in @xmath348 . similarly @xmath335 has the property that @xmath349 in @xmath350 , and @xmath310 in @xmath351 $ ] . let @xmath303 , @xmath304 be the special values announced in section [ specialval ] . for all @xmath216 , define @xmath354 \cup \{(t - r_1^*)/2\}\times[(t+r^*_1)/2,\infty ] \cup[(t - r^*_1)/2,\infty]\times\{(t+r^*_1)/2\}\ ] ] and for any @xmath355 , define @xmath356 the diagram below may be helpful : @xmath357 we shall repeatedly apply the identity @xmath168 in the region @xmath353 . we record below its explicit form : @xmath358 here , @xmath359 whenever @xmath246 belongs to the space - like portion of @xmath352 , and the measure of integration , call it @xmath360 , is there understood to be the induced volume form . let us moreover define @xmath361 and @xmath362 whenever @xmath246 belongs to the @xmath363 and @xmath364 portions of @xmath352 respectively . with this choice , the measure of integration @xmath360 in the respective null segments is understood to be given by @xmath365 all integrals over @xmath352 that appear in the sections that follow are understood to be with respect to the measure @xmath360 defined above . the integral over @xmath353 is to be understood to be with respect to the spacetime volume form . see section [ useful ] . in this section we will give a geometric statement of the main estimates . recall the cutoffs functions @xmath321 , @xmath369 , @xmath370 , from section [ cutsec ] . [ vfsec ] define the vector fields @xmath371 for convenience , let us denote @xmath372 by @xmath368 . the above vector fields will provide the fundamental directions in which the energy momentum tensor @xmath149 is to be contracted and/or @xmath1 is to be differentiated in the definition of the fundamental quantities appearing in the main estimates . see section [ thequansec ] below . the coordinate - dependent definitions given above notwithstanding , the important features of these vector fields can be understood geometrically . all three are invariant with respect to the action of @xmath373 , the one - parameter group of differentiable maps @xmath374 generated by the killing field @xmath170 , @xmath366 is future - directed timelike on @xmath352 , @xmath367 is future - directed timelike everywhere except @xmath156 , where it vanishes quadratically , and @xmath368 is supported away from the horizon and orthogonal to @xmath170 . let @xmath149 be the energy - momentum tensor defined in section [ ccsec ] . define the quantities @xmath378 the quantity @xmath379 is equivalent to the spacetime integral of the density @xmath380 defined by @xmath381 in the sense of the formula @xmath382 understood with the conventions of section [ depcon ] . we will make use of this equivalence often in what follows . recall also that the spacetime integral on the right hand side of @xmath383 is to be understood with respect to the volume form . see section [ useful ] . note that the quantity @xmath384 has integrand positive definite in @xmath385 . it is in fact precisely the flux through @xmath352 of the current @xmath386 . the quantity @xmath387 differs from @xmath384 in that control of the angular and @xmath216-derivatives degenerates quadratically at @xmath156 . similarly , the integrand of @xmath388 also degenerates at @xmath156 . this hypersurface @xmath156 is the so - called photon sphere discussed already in the introduction . the main estimates of this paper are contained in the following [ me ] there exists a constant @xmath305 depending only on @xmath6 , @xmath11 such that for all @xmath355 and all sufficiently regular solutions @xmath1 of @xmath0 in @xmath353 we have @xmath389 @xmath390 @xmath391 more generally than @xmath392 , if @xmath393 is achronal then @xmath394 these estimates will be used in section [ proofgendata ] to prove theorems [ gendata ] and [ second ] . as described in the introduction , the proof of theorem [ me ] will be accomplished in section [ proofme ] with the help of so called energy currents @xmath145 associated to the vector fields @xmath395 , @xmath173 , @xmath183 and @xmath396 . we turn in the next sections to the definition of these currents . were it @xmath375 on the right hand side of @xmath397 , or alternatively , were @xmath377 defined as the time - integral of @xmath398 , then inequality @xmath397 would immediately lead to exponential decay in @xmath216 for @xmath399 ( cf . lemma [ calculus ] ) . the appearance of @xmath400 on the right hand side of @xmath401 signifies that the estimates `` lose '' an angular derivative . at the level of any fixed spherical harmonic @xmath63 , estimates @xmath397 and @xmath401 lead immediately to exponential decay for @xmath402 and @xmath403 . the nature of the loss of angular derivative in @xmath401 means that for the total @xmath404 and @xmath384 , one can only obtain polynomial decay in @xmath216 , where the bound on the decay rate exponent is linear in the angular derivatives lossed . exponential decay for @xmath1 would be retrieved if the `` loss in angular derivatives '' in estimate @xmath401 were logarithmic . see the dependence in @xmath64 in lemma [ calculus ] . we define in this section a family of currents , all loosely based on vector fields parallel to @xmath406 . the role of these currents in capturing the role of the `` photon sphere '' has already been discussed in the introduction . let @xmath161 be a function of @xmath409 and consider a vector field @xmath410 define the currents @xmath411 and the divergences @xmath412 note the identities @xmath413 @xmath414 we compute @xmath415 @xmath416 @xmath417 @xmath418 @xmath419 the expression @xmath420 is not a compatible current in the sense of section [ ccsec ] , since @xmath421 depends on the @xmath10-jet of @xmath1 , but it can be treated as such when restricted to eigenfunctions of @xmath65 . the relation of the photon sphere to currents based on vector fields @xmath153 of the form @xmath160 is most clear upon examining the modified current @xmath422 and noting the coefficient of @xmath423 in @xmath424 vanishes precisely at @xmath156 in view of @xmath425 . this indicates that if one is to have say @xmath426 , the function @xmath161 must change sign at @xmath156 , and the control of the angular derivatives must degenerate at least quadratically . the task of choosing a suitable @xmath161 is simplified by passing to the further modified current @xmath427 which effectively `` borrows '' positivity from the @xmath428 term . finally , one can take advantage of the further flexibility provided by chosing separately @xmath153 for each spherical harmonic , where now , after passing to the current @xmath420 , the @xmath310th order terms are united with the angular derivative terms . in section [ tvfx ] , we shall construct a current @xmath429 we do not in fact ensure that @xmath167 everywhere , but rather , only in the region @xmath430 . the region near the horizon will be handled with the help of the spacetime integral terms controlled by the currents @xmath431 , @xmath432 to be discussed in the next section . once an initial positive definite spacetime integral ( albeit modulo an error ) is constructed , other quantities can be controlled with the help of auxiliary currents . these are defined in section [ auxcursec ] . it is there where we also use the current template @xmath433 . define @xmath436 and set @xmath437 . let @xmath438 denote the @xmath310th spherical harmonic of a solution @xmath0 of the wave equation . consider @xmath439 , @xmath440 as defined above . we compute @xmath441 note that @xmath442 consider for now the expression @xmath421 for a general vector field @xmath153 of the form , applied to a spherical harmonic @xmath63 with spherical harmonic number @xmath140 . we have @xmath443 define @xmath444 the following lemma is proven in appendix [ vanapp ] [ vanlem ] for all @xmath140 , there exists a unique zero @xmath445 of the function @xmath446 in @xmath447 $ ] , and there exist constants @xmath448 , @xmath449 , depending only on @xmath6 and @xmath11 , such that @xmath450 moreover , @xmath451 . _ let @xmath452 , @xmath453 now be fixed , chosen according to the above lemma . _ were the middle term on the right hand side of @xmath454 absent , then we would have @xmath455 for any @xmath161 such that @xmath456 , @xmath457 . the middle term of @xmath454 vanishes if @xmath458 , but , with the requirement that @xmath457 , in this case the function @xmath161 can not be bounded . by suitably cutting off a function linear in @xmath409 , we can ensure that @xmath455 in @xmath430 . let @xmath459 be the cutoff of section [ cutsecchi ] . define @xmath460 and the vector field @xmath395 by @xmath461 we have @xmath462 in the region @xmath463 . moreover , in this region we have in fact , @xmath464 ( recall our conventions for constants @xmath305 from section [ depcon ] . ) finally , for @xmath466 define the `` total '' currents @xmath467 and their divergences @xmath468 @xmath469 besides obtaining nonnegativity for @xmath470 in the region @xmath430 , we need to understand the error in the region @xmath471 and @xmath472 . it turns out that this error can be controlled by in section [ depcon ] . ] @xmath473 , where @xmath474 is a slightly stronger quantity than the energy density @xmath380 . define the quantity @xmath475 here @xmath476 , @xmath477 , @xmath478 are the cut - off functions defined in section [ cutsec ] . note that @xmath479 compare with @xmath480 where @xmath380 is the density of the main quantity @xmath481 defined in . the inequality replacing @xmath482 which holds globally is given by [ controllem ] the inequality @xmath483 holds on all spheres of symmetry . it suffices to consider the regions @xmath471 and @xmath472 . relation implies @xmath484 in these two regions , while @xmath485 the function @xmath486 is supported in the region @xmath487\cup [ r_0^ * , 2r_0^*]$ ] and obeys the pointwise bound latexmath:[\[\label{byourapp } constants denoted @xmath305 . note that @xmath489 in the support of @xmath490 . therefore , @xmath491 holds on all spheres of symmetry . we obtain @xmath492 with @xmath493 we will also need several `` auxiliary '' currents . let us first define , however , certain auxiliary positive definite quantities . the auxiliary currents @xmath188 will be seen to bound these quantities when integrated on spheres of symmetry in the region @xmath430 . define @xmath494 here @xmath321 are the cut - off functions defined in section [ cutsec ] . note that @xmath495 @xmath496 for @xmath497 , and that @xmath498 for all spheres of symmetry . the currents to be described in what follows are motivated by the problem of bounding the positive definite quanitites whose latin superscripts they share . define @xmath500 where @xmath501 is as defined in section [ cutsecchi ] . note that @xmath502 on @xmath503 $ ] , @xmath504 , and @xmath505 for @xmath506 , @xmath507 . set @xmath508 and define @xmath509 @xmath510 the current @xmath511 will allow us to bound the spacetime integrals of the quantities @xmath512 and @xmath513 . see lemma [ q1lem ] . at this point , we can see from and that for each @xmath64 , the pointwise bound @xmath514 holds in @xmath430 . define @xmath517 set @xmath518 and define as before the currents @xmath519 , @xmath520 and @xmath521 , @xmath522 . the current @xmath523 will allow us to bound in addition to the previous the spacetime integral of the quantity @xmath524 . see lemma [ q3lem ] . at this point , we can deduce from , and the bound @xmath525 on each sphere of symmetry with @xmath526 $ ] . the current @xmath520 , in conjunction also with @xmath527 and @xmath528 to be defined below , will be used to help bound the quantity @xmath384 in proposition [ metaaux ] . for now , note that from , , and , the estimate @xmath529 holds on each sphere of symmetry with @xmath530 $ ] . note that the left hand side of _ a priori _ does not have a sign , in view of the term containing @xmath531 . after defining @xmath527 and @xmath528 below , we shall be able to improve with . define @xmath533 set @xmath534 , and define as before @xmath535 , @xmath536 . we have that @xmath537 let us then finally define @xmath539 set @xmath540 and define as before the currents @xmath541 and @xmath542 . the currents @xmath527 and @xmath543 will allow us to bound the spacetime integral of the quantity @xmath544 . see lemma [ q4lem ] . for now , it follows from , and that on each sphere of symmetry with @xmath526 $ ] , we have @xmath545 the currents @xmath520 , @xmath527 and @xmath528 together allow us to estimate , on each sphere of symmetry for @xmath546 $ ] , the quantity @xmath547 again , the left hand side of @xmath548 does not have a sign . in this section , we define the currents @xmath181 and @xmath176 , associated to two vector fields @xmath173 and @xmath183 , supported near the black hole and cosmological horizons , respectively . the role of these currents in capturing the `` red - shift '' effect has been discussed in the introduction . let @xmath549 be a small number , and define functions @xmath550 , @xmath551 as follows . recall the cutoff functions @xmath369 , @xmath476 from section [ cutsecchi ] , and define @xmath552 @xmath553 we have @xmath554 , @xmath555 . to see the latter , note that @xmath556 since @xmath557 are three non - degenerate roots of @xmath558 , which is non - negative on @xmath447 $ ] . the above expression approaches @xmath310 as @xmath6 and @xmath11 tend to the extremal values . similarly , define functions @xmath559 , @xmath560 as follows . recall the cutoff @xmath370 from section [ cutsecchi ] , and define @xmath561 @xmath562 again , note that @xmath563 , @xmath564 . set @xmath173 to be the vector field @xmath565 and @xmath183 to be @xmath566 the application of cutoff @xmath476 is in fact not essential for the arguments of the paper . the cutoff simply ensures that @xmath173 and @xmath183 have smooth extensions beyond @xmath119 and @xmath120 . define the currents @xmath567 @xmath568 and set @xmath569 @xmath570 we have @xmath571 and similarly @xmath572 in this section we exhibit the `` red - shift '' property of the currents @xmath431 , @xmath432 . this is essentially contained in proposition [ polukrisimo ] below . in the context of proving this proposition , we will choose the constants @xmath303 , @xmath304 . note that the polynomial powers in the definitions @xmath573 @xmath574 would be unnecessary had they not been present in the definition @xmath575 . ( their presence in @xmath575 is in turn necessitated by our application of @xmath576 . ) a slightly unpleasant feature of these polynomial decaying expressions is that , if left bare , they would lead to vector fields @xmath173 , @xmath183 which fail to be @xmath577 at @xmath119 , @xmath120 , respectively . this would not in fact pose a problem for the analysis here . we prefer , however , to introduce a cutoff @xmath476 in @xmath573@xmath578 to emphasize the geometric nature of all objects involved in the proof . [ polukrisimo ] for @xmath579 sufficiently small and @xmath580 sufficiently large , depending only on @xmath6 , @xmath11 we have @xmath581 in @xmath582 and @xmath583 , respectively . we give the proof only for @xmath191 . the proof for @xmath584 is similar . recall the functions @xmath585 , @xmath586 from section [ ysec ] . let us denote by @xmath587 in the region @xmath588 , we have @xmath589 as a consequence , for all @xmath590 , @xmath591 @xmath592 recall that @xmath593 and note in addition that @xmath594 in the region @xmath582 . recall now the expression @xmath595 for @xmath191 . in the region @xmath582 , we compute @xmath596 note that @xmath597 it follows that @xmath303 can be chosen sufficiently close to @xmath598 , where the choice depends only on @xmath6 , @xmath11 , such that , in the region @xmath582 , we have @xmath599 in deriving , we have used the cauchy - schwarz inequality to bound the @xmath600 term . ( the fact that the constants work out is ensured by the limiting inequality @xmath601 . it is here that the presence of the constant factor @xmath602 in the definition of @xmath586 is paramount . ) it now follows from that , in the region @xmath582 , @xmath603 as desired . note finally , that , in conformance with our conventions of section [ depcon ] , the constant @xmath305 above does not depend on @xmath220 , @xmath221 , despite the appearance of the cutoff @xmath476 . _ henceforth , let @xmath303 , @xmath304 be chosen so that the conclusion of the above proposition holds . _ the special values @xmath303 , @xmath304 , @xmath301 , @xmath302 now being fixed , constants denoted @xmath305 now depend only on @xmath6 , @xmath11 . let @xmath605\to[0,1]$ ] be a cutoff function such that @xmath606 define @xmath607 let @xmath608 be the heaviside step - function and let @xmath396 to be the vector field @xmath609 and define as before the currents @xmath604 and @xmath610 . despite the appearance of the heaviside function , @xmath396 is a @xmath611 vector field . it is of the form @xmath612 , where @xmath613 is a spacetime cut - off function adapted to the @xmath611 foliation @xmath352 , constant on each leaf @xmath352 . we have @xmath614 let us define the auxiliary quantities @xmath615 where @xmath616 is as defined in section [ apdpq ] , with @xmath617 , and @xmath474 is as defined in section [ errcontrol ] . for the boundary terms , note first that on the support of @xmath618 , @xmath619 on the other hand , in the expression , each term gives a nonnegative contribution to @xmath620 , and we have on @xmath352 @xmath621 as a consequence , @xmath622 and @xmath623 in this section we address questions of size of the boundary terms generated by the currents @xmath625 from section [ x ] and @xmath626 from section [ ysec ] . [ bxs ] for @xmath627 , @xmath628 , @xmath519 , @xmath520 , @xmath527 , @xmath528 we have @xmath629 @xmath630 we shall here only prove the proposition in a representative case of the current @xmath631 defined in section [ aux ] . recall that @xmath632 since on the space - like part of @xmath352 , we have @xmath633 , it follows that there @xmath634 since in addition @xmath635 . on the other hand , since @xmath636 and @xmath637 on the null segments @xmath93=const and @xmath92=const of @xmath352 , respectively , @xmath638 @xmath639 since the space - like portion of @xmath352 corresponds to @xmath66-values @xmath640 we have that @xmath641 therefore , @xmath642 on the other hand , @xmath643 we compare now with the expression for @xmath644 . start with the current @xmath645 since @xmath646 , we infer that on @xmath352 @xmath647 for @xmath648 , consider now @xmath649 recall that the functions @xmath650 on the interval @xmath651 $ ] and vanish for @xmath652 and @xmath653 . on the space - like portion of @xmath352 , i.e. , for @xmath654 $ ] , we have @xmath655 while on the null segments of @xmath352 , we have @xmath656 for @xmath363 , and @xmath657 for @xmath364 . therefore , @xmath658 summing over @xmath64 and using the identity @xmath659 we obtain @xmath660 as desired . on the other hand , @xmath661 on the horizon @xmath662 , we have @xmath663 , and thus @xmath664 , @xmath665 and @xmath666 as a consequence , since @xmath667 , @xmath668 as desired . similar arguments give the inequality on the horizon @xmath669 , as well as the inequalities for the other currents of the statement of the proposition . [ posvan ] for @xmath670 , we have @xmath671 moreover @xmath672 @xmath673 @xmath674 @xmath675 once again , we shall here consider only the current @xmath676 . the considerations for @xmath432 are practically identical . by the construction in section [ ysec ] , the support of @xmath676 is contained in the region @xmath677 . this immediately implies that @xmath678 . moreover @xmath173 is a bounded future - directed time - like vector field in the region @xmath679 , which implies that there we have @xmath680 restricted to the support of @xmath173 , the remaining part of @xmath352 is contained in a null segment @xmath363 . thus , we have @xmath681 the functions @xmath682 are non - negative and in the region @xmath582 are given by @xmath683 which implies that @xmath684 comparing this to the expression for @xmath685 , given in , we see that @xmath686 on the null portion @xmath363 of @xmath352 . since on this portion we have @xmath687 , we finally obtain the desired inequality @xmath688 on the other hand , on @xmath662 we have @xmath689 and @xmath690 since @xmath691 and @xmath692 on @xmath662 . thus @xmath693 in this section we exploit the divergence theorem for compatible currents to relate various integral quantities . from proposition [ bxs ] , proposition [ posvan ] and identity @xmath291 , the following two propositions follow immediately : [ stb ] for @xmath694 , @xmath628 , @xmath519 , @xmath520 , @xmath527 , @xmath528 and @xmath296 , we have @xmath695 [ oldiys ] we have @xmath696 @xmath697 applying @xmath291 with the energy current @xmath698 gives us the following proposition [ evaveo ] @xmath699 the statement follows immediately from @xmath288 and @xmath700 , @xmath701 . applying @xmath291 with the energy current @xmath702 , we obtain [ achronal ] let @xmath703 be achronal . then @xmath704 in particular , @xmath705 the vector field @xmath706 is timelike and one sees easily @xmath707 while certainly @xmath708 . we have an alternative bound on @xmath709 as follows [ privavafora ] @xmath710 recall that by the construction given in section [ pragmaux ] , the vector field @xmath396 is future timelike , @xmath711 and @xmath712 . the result now follows from the statement @xmath713 and the divergence theorem . in this section we establish the first key part of theorem [ me ] , that is to say , statement @xmath397 . we begin with the following [ megalnprot ] @xmath714 the proposition follows from the three lemmas below : [ q1lem ] @xmath715 [ q3lem ] @xmath716 [ q4lem ] @xmath717 the statements of the above three lemmas follow directly from , , and , combined with the observation that , since the supports of the currents @xmath718 , @xmath719 and @xmath720 , as well as of the quantities @xmath721 ) , are contained in the region @xmath722 , since @xmath723 is nonnegative , and since @xmath166 is nonnegative in @xmath724 , it suffices to apply lemma [ controllem ] . proposition [ megalnprot ] now follows from proposition [ stb ] and proposition [ evaveo ] . [ kyboundf ] @xmath725 this follows immediately by adding @xmath726 , ( @xmath727 of proposition [ oldiys ] , in view also of proposition [ posvan ] . [ eukoloprop ] @xmath728 the statement follows from the inequality @xmath729 which is a direct consequence of , and proposition [ polukrisimo ] . it now immediately follows from proposition [ megalnprot ] , proposition [ kyboundf ] and proposition [ eukoloprop ] and our conventions regarding constants @xmath305 , @xmath306 and @xmath127 that the following holds : [ teliknepilogn ] if @xmath307 is chosen sufficiently small , and @xmath308 is chosen sufficiently large , depending only on @xmath6 , @xmath11 , then @xmath730 _ henceforth , let @xmath220 , @xmath221 be so chosen so that the conclusion of the above proposition holds . in particular , in what follows we shall need only make use of constants @xmath305 depending only on @xmath6 , @xmath11 . _ finally , since by , @xmath731 the statement @xmath397 of theorem [ me ] follows immediately . bound @xmath401 of theorem [ me ] is contained in the following [ metaaux ] for all @xmath732 @xmath733 by proposition [ privavafora ] @xmath734 the current @xmath735 was defined in section [ pragmaux ] . one easily sees @xmath736 comparing this with , we infer that the statement of the proposition would follow from the estimate @xmath737 from we have that @xmath738 it thus follows from that @xmath739 the desired @xmath740 now follows from propositions [ stb ] and [ evaveo ] . the final statements of theorem [ me ] follows from [ xwrisaux ] let @xmath393 be achronal . then @xmath742 in particular , @xmath743 by proposition [ achronal ] @xmath704 recall that the current @xmath744 ( respectively , @xmath745 ) is positive for @xmath582 ( respectively , @xmath583 ) and vanishes for @xmath746 ( respectively , @xmath747 ) . moreover , comparing and easily implies the bound @xmath748 for @xmath749 , and @xmath750 . the result now follows immediately . [ prwtodecay ] there exist constants @xmath305 , @xmath133 , depending only on @xmath6 , @xmath11 , such that for all @xmath63 solutions of @xmath751 in @xmath752 with spherical harmonic number @xmath64 , then @xmath753 for all @xmath216 and all @xmath754 . this follows immediately from estimates @xmath397 and @xmath401 , in view of the following lemma , proved in appendix [ calcap ] , applied to the functions @xmath755 and @xmath756 : [ calculus ] let @xmath757 , @xmath758 be positive constants and let @xmath759 , @xmath760 be nonnegative continuous functions satisfying @xmath761 for all @xmath762 , and @xmath763 . then there exists a constants @xmath133 depending only on @xmath757 , and a universal constant @xmath305 such that @xmath764 for all @xmath216 and for all @xmath754 . [ apauto2 ] there exist constants @xmath305 , @xmath133 , depending only on @xmath6 , @xmath11 , such that for all @xmath63 solutions of @xmath751 in @xmath752 with spherical harmonic number @xmath64 and for all achronal @xmath765 , @xmath766 in particular @xmath767 this follows immediately from propositions [ prwtodecay ] and the inequality @xmath768 of theorem [ me ] . the energy decay statements of theorem [ second ] now follow from the following let @xmath18 be a cauchy surface for @xmath19 , and let @xmath34 , @xmath80 , @xmath769 be as in the statement of theorem [ gendata ] or theorem [ second ] . then , there exist constants @xmath305 , @xmath770 depending only on @xmath6 , @xmath11 and the geometry of @xmath29 , such that @xmath771 and such that for all solutions @xmath1 to the wave equation @xmath0 on @xmath772 , the estimate @xmath773 holds . this is completely standard . we give a sketch to emphasize here too the role of compatible currents based on vector fields ! extend say @xmath774 from @xmath775 to an arbitrary future - timelike vector field @xmath366 in @xmath776 , and consider an arbitrary spacelike foliation @xmath777 of this region by manifolds with boundary , such that @xmath778 , @xmath779 , and @xmath780 . consider the analogue of @xmath291 in @xmath781 . set @xmath782 in view of the fact that @xmath366 is future timelike , we obtain @xmath783 on the other hand , one easily sees that there exists a @xmath305 depending on the geometry of our chosen foliation of the compact set @xmath784 such that for all @xmath785 $ ] , @xmath786 it now follow that @xmath787 . the result now follows by noting that @xmath788 the pointwise decay statements follow easily . consider the function @xmath789 recall that @xmath790 we may thus rewrite the function as @xmath791 for @xmath140 , @xmath177 is concave in @xmath66 : @xmath792 the value of @xmath177 at @xmath793 is given by @xmath794 where we have used @xmath795 . the same formula evidently holds at @xmath231 with @xmath598 replacing @xmath796 everywhere . since @xmath797 , @xmath798 and @xmath799 , we have @xmath800 from @xmath801 , we obtain similarly @xmath802 by concavity the function @xmath803 then has exactly one zero on the interval @xmath804 , and thus , so does @xmath805 . the second statement of the lemma is clear from the final , which in turn follows immediately from the form of the function @xmath806 . by replacing @xmath757 with @xmath807 , we may assume in what follows that @xmath808 . let @xmath809 be a sequence with @xmath810 . we can choose @xmath811\ ] ] such that by pigeonhole principle @xmath812 assumptions of the lemma then imply that @xmath813 @xmath814 therefore , @xmath815 thus if @xmath816 @xmath817 then @xmath818 @xmath819 now we have that @xmath820 , @xmath821 indeed hold for @xmath822 , where @xmath823 , and @xmath824 is defined so as to satisfy @xmath825 , with @xmath826 . this proves that for the sequence @xmath827 , defined above , @xmath828 using the assumptions of the lemma again we immediately obtain the desired statement . p. blue and a. soffer _ errata for `` global existence and scattering for the nonlinear schrodinger equation on schwarzschild manifolds '' , `` semilinear wave equations on the schwarzschild manifold i : local decay estimates '' , and `` the wave equation on the schwarzschild metric ii : local decay for the spin 2 regge wheeler equation '' _ , gr - qc/0608073 , 6 pages

we present near - infrared ( 0.8 - 1.8 ) spectra of 105 bright ( 10 ) stars observed with the low resolution spectrometer on the rocket - borne cosmic infrared background experiment ( ciber ) . as our observations are performed above the earth s atmosphere , our spectra are free from telluric contamination , which makes them a unique resource for near - infrared spectral calibration . two - micron all sky survey ( 2mass ) photometry information is used to identify cross - matched stars after reduction and extraction of the spectra . we identify the spectral types of the observed stars by comparing them with spectral templates from the infrared telescope facility ( irtf ) library . all the observed spectra are consistent with late f to m stellar spectral types , and we identify various infrared absorption lines .

1701.03609

precise ground - based measurements of stellar spectra are challenging in the near - infrared ( ir ) because of the contaminating effects of telluric lines from species like water , oxygen , and hydroxyl in the earth s atmosphere . telluric correction using standard stars is generally used to overcome this problem , but these corrections are problematic in wavelength regions marked by strong line contamination , such as from water and hydroxyl . in contrast , space - based spectroscopy in the near - ir does not require telluric correction , so can provide new insights into stellar atmospheres ( e.g. @xcite ) , especially near @xmath3 where starlight is not reprocessed by dust in the circumstellar environment @xcite . in particular , near - ir spectra can be used to study the age and mass of very young stars @xcite , and the physical properties of very cool stars @xcite . of particular interest in the study of the atmospheres of cool stars is water . according to early models of stellar photospheres @xcite , h@xmath4oexisted only in later than m6 type stars , and until recently observations have supported this . in 1963 , the balloon - borne telescope stratoscope ii observed h@xmath4o in two early m2-m4 giant stars @xcite at 1.4 and @xmath5 . several decades later , @xcite measured h@xmath4o absorption in an m2.5 giant star using the infrared space observatory ( @xcite ) , and @xcite observed water at 1.4 , 1.9 , 2.7 , and @xmath6 for 67 stars with the infrared telescope in space ( @xcite ) . surprisingly , @xcite discovered water features in late k - type stars . these results required a new stellar photosphere model to explain the existence of h@xmath4o features in hotter than m6 type stars @xcite . the low resolution spectrometer ( lrs ; @xcite ) on the cosmic infrared background experiment ( ciber ; @xcite ) observed the diffuse infrared background from 0.7 to 2.0 @xmath0 during four flights above the earth atmosphere . the lrs was designed to observe the near - ir background @xcite , and as a result finds excess extragalactic background light above all known foregrounds @xcite . furthermore , we precisely measure astrophysical components contributing to the diffuse sky brightness ( see @xcite for a review ) . for example , @xcite observed a component of the zodiacal light absorbed by silicates in a broad band near @xmath7 nm . by correlating the lrs with a 100 @xmath0 dust map @xcite , @xcite measured smooth diffuse galactic light spectrum from the optical band to the near - ir and constrained the size distribution of interstellar dust , which was dominated by small particles ( half - mass radius @xmath80.06 @xmath0 ) . the lrs also observed many bright galactic stars , enabling us to study their near - ir seds . in this paper , we present flux - calibrated near - ir spectra of 105 stars from @xmath9 with spectral resolution @xmath10 over the range . the paper is organized as follows . in section [ s : instrument ] , the observations and instrumentation are introduced . we describe the data reduction , calibration , astrometry , and extraction of the stellar spectra in section [ s : datareduction ] . in section [ s : stars ] , the spectral typing and features are discussed . finally , a summary and discussion are given in section [ s : summary ] . the lrs is one of the four optical instruments of the ciber payload @xcite ; the others are a narrowband spectrometer ( @xcite ) and two wide - field imagers @xcite . the lrs @xcite is a prism - dispersed spectrometer with five rectangular @xmath11 slits imaging a 5.8 @xmath12 @xmath13 5.8 @xmath12 field of view . the detector has @xmath14 pixels at a pixel scale of @xmath15 . ciber has flown four times ( 2009 february , 2010 july , 2012 march , and 2013 june ) with apogees and total exposure times of over 325 km and @xmath8 240 s , respectively , in the first three flights and of 550 km and 335 s in the final , non - recovered flight . due to spurious signal contamination from thermal emission from the shock - heated rocket skin , we do not use the first flight data in this work @xcite . eleven target fields were observed during the three subsequent flights , as listed in table [ tbl1 ] . details of the field selection are described in @xcite . during the observations , the detector array is read nondestructively at @xmath16hz frame@xmath17 . each field is observed for many tens or hundreds of frames , and an image for each field is obtained by computing the slope of the accumulated values for each pixel @xcite . figure [ flight_image ] shows an example image of the north ecliptic pole ( nep ) region obtained during the second flight . more than 20 bright stars ( @xmath18 @xmath2 11 ) are observed . the stellar spectra are characterized by a small amount of field distortion as well as an arc - shaped variation in constant - wavelength lines along the slit direction . the latter is known as a `` smile '' and is a known feature of prism spectrometers @xcite . details of the treatment of these distortions are described in section [ ss : sbr ] and [ ss : starsel ] . in this section , we describe how we perform background subtraction , calibration , photometric estimation , astrometric registration , and spectral extraction from the lrs - observed images . we measure the relative pixel response ( flat field ) in the laboratory before each flight @xcite . the second- and the third - flight data are normally corrected with these laboratory flats . however , for the fourth flight from the laboratory calibrations do not extend to the longest wavelengths ( @xmath19 ) because the slit mask shifted its position with respect to the detector during the flight . we therefore use the second - flight flat field to correct the relative response for the fourth - flight data , as this measurement covers @xmath20 . to apply this flat field , we need to assume that the intrinsic relative pixel response does not vary significantly over the flights . to check the validity of this assumption , we subtract the second flat image to the fourth flat image for overlapped pixels and calculate the pixel response difference . we find that only 0.3 % of pixels with response measured in both are different by @xmath21 , where @xmath22 is the standard deviation of the pixel response . finally , we mask 0.06 % of the array detectors to remove those pixels with known responsivity pathologies and those prone to transient electronic events @xcite . for each flight , the absolute brightness and wavelength irradiance calibrations have been measured in the laboratory in collaboration with the national institute of standards and technology . the details of these calibrations can be found in @xcite . the total photometric uncertainty of the lrs brightness calibration is estimated to be @xmath23% @xcite . the raw image contains not only spectrally dispersed images of stars but also the combined emission from zodiacal light @xmath24 , diffuse galactic light @xmath25 , the extragalactic background @xmath26 , and instrumental effects @xmath27 @xcite . the measured signal @xmath28 can be expressed as @xmath29 where we have decomposed the intensity from stars into a resolved component @xmath30 and an unresolved component arising from the integrated light of stars below the sensitivity of the lrs @xmath31 . it is important to subtract the sum of all components except @xmath30 from the measured brightness to isolate the emission from detected stars . at this point in the processing , we have corrected for multiplicative terms affecting @xmath28 . dark current , which is the detector photocurrent measured in the absence of incident flux , is an additional contribution to @xmath27 . the stability of the dark current in the lrs has been shown to be 0.7 nw m@xmath32 sr@xmath17 over each flight , which is a negligible variation from the typical dark current ( i.e. , 20 nw m@xmath32 sr@xmath17 ; @xcite ) . as a result , we subtract the dark current as part of the background estimate formed below . the relative brightnesses of the remaining background components are wavelength - dependent , so an estimate for their mean must be computed along constant - wavelength regions , corresponding to the vertical columns in figure [ flight_image ] . furthermore , because of the lrs s large spatial psf , star images can extend over several pixels in the imaging direction and even overlap one another . this complicates background estimation in pixels containing star images and reduces the number of pixels available to estimate the emission from the background components . to estimate the background in those pixels containing star images , we compute the average value of pixels with no star images along each column , as summarized in figure [ flow_chart ] . we remove bright pixels that may contain star images , as described in @xcite . the spectral smile effect shown in figure [ flight_image ] introduces spectral curvature along a column . we estimate it causes an error of magnitude @xmath33 , which is small compared to the spectral width of a pixel . approximately half of the rows remain after this clipping process ; the fraction ranges from 45 % to 62 % depending on the stellar density in each field . this procedure removes all stars with @xmath34 , and has a decreasing completeness above this magnitude @xcite . to generate an interpolated background map , each candidate star pixel is replaced by the average of nearby pixels calculated along the imaging direction from the @xmath35 pixels on either side of the star image . we again do not explicitly account for the spectral smile . this interpolated background image is subtracted from the measured image , resulting in an image containing only bright stellar emission . the emission from faint stars and bright stars that inefficiently illuminate a grating slit that contributes to @xmath36 is naturally removed in this process . the bright lines dispersed in the spectral direction in the background - subtracted images are candidate star spectra . to calculate the spectrum of candidate sources , we simply isolate individual lines of emission and map the pixel values onto the wavelength using the ground calibration . however , this procedure is complicated both by the extended spatial psf of the lrs and by source confusion . to account for the size of the lrs spatial psf ( fwhm @xmath81.2 pixels ) as well as optical distortion from the prism that spreads the star images slightly into the imaging direction , we sum five rows of pixels in the imaging direction for each candidate star . since the background emission has already been accounted for , this sum converges to the total flux as the number of summed rows is increased . by summing five rows , we capture @xmath37% of a candidate star s flux . the wavelengths of the spectral bins are calculated from the corresponding wavelength calibration map in the same way . from these spectra , we can compute synthetic magnitudes in the @xmath38- and @xmath39-bands , which facilitate comparison to two - micron all - sky survey ( 2mass ) measurements . we first convert surface brightness in nw m@xmath32 sr@xmath17 to flux in nw m@xmath32 hz@xmath17 , and then integrate the monochromatic intensity over the 2mass band , applying the filter transmissivity of the @xmath38- and @xmath39-bands @xcite . to determine the appropriate zero magnitude , we integrate the @xmath38- and @xmath39-band intensity of vega s spectrum @xcite with the same filter response . the @xmath38- and @xmath39-band magnitudes of each source are then calculated , allowing both flux and color comparisons between our data and the 2mass catalog . candidate star spectra may be comprised of the blended emission from two or more stars , and these must be rejected from the catalog . such blends fall into one of two categories : ( _ i _ ) stars that are visually separate but are close enough to share flux in a 5 pixel - wide photometric aperture , and ( _ ii _ ) stars that are close enough that their images overlap so as to be indistinguishable . we isolate instances of case ( _ i _ ) by comparing the fluxes calculated by summing both three and five rows along the imaging direction for each source . if the magnitude or @xmath40 color difference between the two apertures is larger than the statistical uncertainty ( described in section [ ss : errors ] ) , we remove those spectra from the catalog . to find instances of case ( _ ii _ ) , we use the 2mass star catalog registered to our images using the procedure described in section [ ss : astrometry ] . candidate sources that do not meet the criteria presented below are rejected . to ensure the catalog spectra are for isolated stars rather than for indistinguishable blends , we impose the following requirements on candidate star spectra : ( _ i _ ) each candidate must have @xmath41 ; ( _ ii _ ) the @xmath38-band magnitude difference between the lrs candidate and the matched 2mass counterpart must be @xmath42 ; ( _ iii _ ) the @xmath40 color difference between the lrs candidate star and the matched 2mass counterpart must be @xmath43 ; and ( _ iv _ ) among the candidate 2mass counterparts within the 500@xmath44 ( @xmath45 pixel ) radius of a given lrs star , the second - brightest 2mass star must be fainter than the brightest one by more than 2 mag at the j band . criterion ( _ i _ ) excludes faint stars that may be strongly affected by residual backgrounds , slit mask apodization , or source confusion . the second and third criteria mitigate mismatching by placing requirements on the magnitude and color of each star . in particular , the @xmath46 color of a source does not depend on the slit apodization or the position in image space ( see figure [ color_comp ] ) , so any significant change in @xmath46 color as the photometric aperture is varied suggests that more than a single star could be contributing to the measured brightness . finally , it is possible that two stars with similar @xmath46 colors lie close to each other , so the last criterion is applied to remove stars for which equal - brightness blending is an issue . approximately one in three candidate stars fails criterion ( _ iv _ ) . the number of candidate stars rejected at each criterion is described in table [ tbl3 ] . in addition , three of lrs candidate stars are identified as variables in the simbad database . we also identify two stars as binary and multiple - star systems as well as four high proper motion stars . through these stringent selection requirements , we conservatively include only the spectra of bright , isolated stars in our catalog . finally , 105 star spectra survive all the cuts , and the corresponding stars are selected as catalog members . we match the synthesized lrs @xmath38 , @xmath39 , and @xmath46 information with the 2mass point source catalog @xcite to compute an astrometric solution for the lrs pointing in each sky image . this is performed in a stepwise fashion by using initial estimates for the lrs s pointing to solve for image registration on a fine scale . as a rough guess at the lrs pointing , we use information provided by the rocket s attitude control system ( acs ) , which controls the pointing of the telescopes @xcite . this provides an estimated pointing solution that is accurate within 15 @xmath47 of the requested coordinates . however , since the acs and the lrs are not explicitly aligned to one another , a finer astrometric registration is required to capture the pointing of the lrs to single - pixel accuracy . to build a finer astrometric solution , we simulate images of each field in the 2mass _ j_-band using the positional information from the acs , spatially convolved to the lrs psf size . next , we apodize these simulated 2mass images with the lrs slit mask , compute the slit - masked magnitudes of three reference stars , and calculate the @xmath48 statistic using @xmath49 where index _ i _ represents each reference star and subscripts _ p _ and _ q _ index the horizontal and vertical positions of the slit mask , respectively . @xmath50 and @xmath51 are the fluxes in the lrs and 2mass @xmath38-band , and @xmath52 is the statistical error of the lrs star ( see section [ ss : errors ] ) . the minimum @xmath48 gives the most likely astrometric position of the slit mask . since , on average , there are around five bright stars with @xmath53 per field , spurious solutions are exceedingly unlikely , and all fields give a unique solution . using this astrometric solution , we can assign coordinates to the rest of the detected lrs stars . we estimate that the overall astrometric error is 120@xmath44 by computing the mean distance between the lrs and 2mass coordinates of all matched stars . the error corresponds to 1.5 times the pixel scale . we check the validity of the astrometric solutions by comparing the colors and fluxes between the lrs and matched 2mass stars . in figures [ color_comp ] and [ flux_comp ] , we show the comparison of the @xmath46 colors and fluxes of the cross - matched stars in each field . here , we multiply the lrs fluxes at the j- and h - band by 2.22 and 2.17 , respectively , to correct for the slit apodization . the derivation of correction factors is described in section [ s : summary ] . on the whole , they match well within the error range . even following careful selection , the star spectra are subject to various kinds of uncertainties and errors , including statistical uncertainties , errors in the relative pixel response , absolute calibration errors , wavelength calibration errors , and background subtraction errors . statistical uncertainties in the spectra can be estimated directly from the flight data . we calculate the @xmath54 slope error from the line fit ( see section [ s : instrument ] ) as we generate the flight images ; this error constitutes the estimate for the statistical photometric uncertainty for each pixel . in this statistical error , we include contributions from the statistical error in the background estimate and the relative pixel response . the error in the background signal estimate is formed by computing the standard deviation of the @xmath5510 pixels along the constant-@xmath56 direction for each pixel to match the background estimate region . this procedure captures the local structure in the background image , which is a reasonable measure of the variation we might expect over a photometric aperture . neighboring pixels in the wavelength direction have extremely covariant error estimates in this formulation , which are acceptable since the flux measurements are also covariant in this direction . a statistical error from the relative pixel response correction is applied by multiplying 3% of the relative response by the measured flux in each field @xcite . to compute the total statistical error , each constituent error is summed in quadrature for each pixel . several instrumental systematic errors are present in these measurements , including those from wavelength calibration , absolute calibration , and relative response correction . in this work , we do not explicitly account for errors in the wavelength calibration , as the variation is @xmath55 1 nm over 10 constant - wavelength pixels , which is @xmath57 . in all flights , @xmath2 3 % absolute calibration error is applied @xcite . for the longest - wavelength regions ( @xmath56 @xmath58 1.6 @xmath0 ) of the fourth - flight data that are not measured even in the second - flight flat , we could not perform flat correction . instead , we apply a systematic error amounting to 5.3 % of the measured sky brightness . the error is estimated from pixels in the short - wavelength regions ( @xmath56 @xmath2 1.4 @xmath0 ) of the fourth - flight flat . we calculate deviations from unity for those pixels and take a mean of 5.3 % . the linear sum of systematic errors is then combined with statistical error in quadrature . the 105 stellar spectra that result from this processing can be used to test spectral type determination algorithms and study near - ir features that are invisible from the ground . despite the relatively low spectral resolution of our stellar spectra , we identify several molecular bands , particularly for the late - type stars . we present the @xmath59band - normalized lrs spectra for each of the catalog stars in figure [ sp_fig1 ] . general information for each spectrum is summarized in table [ tbl2 ] with the corresponding star i d . all spectra are publicly available in electronic form.mgkim/ ] the spectra are presented without the application of interstellar extinction corrections , since extinction correction assumes both a color index and the integrated galactic extinction along the line of sight . therefore , without knowing the stars distances , it is difficult to make progress . for ciber fields , typical extinction ranges from 0.005 to 0.036 mag at the j - band if we assume extinction coefficients r(j ) with 0.72 @xcite the star spectral types are determined by fitting known spectral templates to the measured lrs spectra . we use the infrared telescope facility ( irtf ) and pickles @xcite templates for the sed fitting . the spex instrument installed on the irtf observed stars using a medium - resolution spectrograph ( _ r _ @xmath60 2000 ) . the template library contains spectra for 210 cool stars ( f to m type ) with wavelength coverage from 0.8 to 2.5 @xmath0 @xcite . the pickles library is a synthetic spectral library that combines spectral data from various observations to achieve wavelength coverage from the uv ( 0.115 @xmath0 ) to the near - ir ( 2.5 @xmath0 ) . it contains 131 spectral templates for all star types ( i.e. , o to m type ) with a uniform sampling interval of 5 @xmath61 . to perform the sed fit , we degrade the template spectra to the lrs spectral resolution using a mean box - car smoothing kernel corresponding to the slit function of the lrs . both the measured and template spectra are normalized to the @xmath38-band flux . we calculate the flux differences between the lrs and template spectra using @xmath62 where @xmath63 and @xmath64 are the fluxes of the observed and template spectra at wavelength @xmath56 normalized at @xmath38-band and @xmath65 is the statistical error of the observed spectrum . the best - fitting spectral type is determined by finding the minimum @xmath48 . no early - type ( i.e. , o , b , a ) stars are found in our sample ; all stars have characteristics consistent with those of late - type stars ( f and later ) . because the irtf library has about twice the spectral type resolution of the pickles library , we provide the spectral type determined from the irtf template in table [ tbl2 ] . since the irtf library does not include a continuous set of spectral templates , we observe discrepancies between the lrs and best - fit irtf templates , even though the @xmath46 colors are consistent between 2mass and the lrs within the uncertainties . the pickles and irtf fits are consistent within the uncertainty in the classification ( @xmath8 0.42 spectral subtypes ) . a color - color diagram for the star sample is shown in figure [ ccd ] . although the color - color diagram does not allow us to clearly discriminate between spectral types , qualitatively earlier - type stars are located in the bluer region , while later - type stars are located in the redder region , consistent with expectations . lrs stars well follow the color - color distributions of typical 2mass stars in lrs fields , as indicated by the gray dots . to estimate the error in our spectral type determination , we compare our identifications with the simbad database @xcite , where 63 of the 105 stars have prior spectral type determinations . figure [ type_difference ] shows the spectral types determined from the irtf fit versus those from the simbad database . the 1@xmath22 error of type difference is estimated to be 0.59 spectral subtypes , which is comparable with those in other published works @xcite . the error can be explained with two factors : ( _ i _ ) the low spectral resolution of the lrs and ( _ ii _ ) the sed template libraries , which do not represent all star types . five stars are observed twice in different flights ( ba2_5 and bb4_6 , n2_6 and n3_5 , ba2_1 and ba3_4 , bb2_1 and bb3_1 , and bb2_4 and bb3_4 ; see figure [ dup_star ] ) , enabling us to investigate the interflight stability of the spectra . for ba2_5 and bb4_6 , the spectral type is known to be f8 , while our procedure yields f7v and f1ii from the second- and fourth - flight data , respectively . for n2_6 and n3_5 , the known type is k5 while we determine m0.5v for both flights . for ba2_1 and ba3_4 , the known type is f5 while we determine f7iii and f2iii - iv in the second and third flights . for bb2_1 and bb3_1 , the fitted types are g8iiife5 and k4v for a k1 type star , and the type of bb2_4 and bb3_4 are not known but are fitted to f9v for both flights . the determined spectra are consistent within an acceptable error window , though the longer - wavelength data exhibit large differences , which can be attributed to calibration error . we present the spectra of each star from both flights in table [ tbl2 ] . this duplication results in our reporting of 110 spectra in the catalog , even through only 105 individual stars are observed . we determined the spectral type of 105 stars as well as the associated typing error ( 0.59 spectral subtypes ) assessed by comparing the type against a set of 63 previously determined spectral types . representative examples of the measured spectra for different spectral types are shown in figure [ type_example ] . molecular absorption lines are evident in these spectra , including the caii triplet and various cn bands . since we observed stars above the earth s atmosphere , observations of the h@xmath4o molecular band are possible . however , they are not able to distinguish between cn and h@xmath4o at 1.4 @xmath0 since both have the same bandhead and appear in late - type stars @xcite . for example , the spectral features of m2-m4 ( super)giant stars observed by stratoscope ii , previously identified as cn , were identified as h@xmath4o @xcite . several subsequent observations show clear evidence that water features exist even in k type stars , requiring modifications of present stellar photosphere models @xcite . in our spectral catalog , most k and m type stars exhibit a broad absorption band around 1.4 @xmath0 . although it is not possible to identify specific molecular bands with our data , we can not exclude the presence of h@xmath4o in the spectra of these stars . future mid - ir measurements at @xmath66 would help disentangle the source of the spectral features by removing the spectral degeneracies between cn and h@xmath4o @xcite . as these spectra are free from telluric contamination and the lrs is calibrated against absolute irradiance standards @xcite , _ in principle _ these measurements could be used as near - ir spectral standards . however , our lack of knowledge of the instrument response function ( irf ) on the spectral plane complicates the use of these measurements for the absolute photometric calibration of stars . specifically , the lrs s irf depends on the end - to - end optical properties of the instrument . because we use a slit mask at the focus of an optical coupler @xcite , the full irf knowledge of the focusing element of the optical coupler is difficult to disentangle from other effects . as a result , we would need to know the precise irf to assign an absolute error estimate to an absolute calibration of the star images . this response function was not characterized during ground testing . nevertheless , we consider it instructive to check the validity of photometric results whether or not the estimated magnitudes of the lrs stars are reasonable compared to previous measurements . we perform an empirical simulation as follows . for each lrs star , we generate a point source image with the flux of the 2mass counterpart convolved to the lrs psf . instrumental noise and source confusion from faint stars ( @xmath34 ) based on the 2mass stars around a target star are also added . we measure the photometric flux of the simulated star image in the same way as for the lrs stars as described in this paper . an aperture correction is applied to the lrs stars , since stars that are clipped by the slit mask will appear to have a reduced flux measurement . figures [ model_j ] and [ model_h ] show the ratios of the band - synthesized flux of each lrs star to the flux of the corresponding 2mass star with statistical errors . the range explained by our simulations is illustrated as a color - shaded area . the lrs stars fall within the expected flux range . also , the flux ratios of the stars between flights well agree , validating the stability of the photometric calibrations for the three ciber flights . the large scatter at faint stars is caused by background noise , including adjacent faint stars and the instrument . the statistical j- and h - band flux errors are 3.89 % and 4.51 % , with systematic errors of 2.98 % and 3.82 % . we conclude that the achievable uncertainties on the absolute photometric amplitudes of these spectra are not competitive with other measurements ( e.g. the existing 2mass j and h - band flux errors are 1.57 % and 2.36 % , respectively ) . the slit mask apodization correction ultimately limits the accuracy of our absolute calibration measurement and can lead to subtle biases . however , by connecting them with precise spectral measurements , we can improve the accuracy of lrs stellar spectra . the european space agency s gaia @xcite mission is a scanning all - sky survey that uses a blue photometer ( 0.33@xmath0 @xmath2 @xmath56 @xmath2 0.68@xmath0 ) and a red ( 0.64@xmath0 @xmath2 @xmath56 @xmath2 1.05@xmath0 ) one to cover 0.33@xmath0 to 1.05@xmath0 with spectral resolution similar to that of the lrs . because the gaia photometers spectrally overlap with the lrs , we expect to eventually be able to unambiguously correct for the slit mask apodization and achieve an absolute flux calibration with less than 2 % accuracy over the full range @xmath67 m for our 105 stars . in addition , the data reduction procedure described here may be a useful guide for the gaia analysis . since gaia uses a prism - based photometer source detection , the data will show a nonlinear spatial variation of constant - wavelength bands and flux losses by a finite window size , as in our measurements . the background estimation will also require careful treatment with precise estimation of the end - to - end gaia psf . this work was supported by nasa apra research grants nnx07ai54 g , nng05wc18 g , nnx07ag43 g , nnx07aj24 g , and nnx10ae12 g . initial support was provided by an award to j.b . from the jet propulsion laboratory s director s research and development fund . japanese participation in ciber was supported by kakenhi ( 2034 , 18204018 , 19540250 , 21340047 , 21111004 , and 26800112 ) from japan society for the promotion of science ( jsps ) and the ministry of education , culture , sports , science , and technology . korean participation in ciber was supported by the pioneer project from the korea astronomy and space science institute . m.g.k . acknowledges support from the global phd fellowship program through the nrf , funded by the ministry of education ( 2011 - 0007760 ) . h.m.l . and m.g.l were supported by nrf grant 2012r1a4a1028713 . m.z . and p.k . acknowledge support from nasa postdoctoral program fellowships , and a.c . acknowledges support from nsf career awards ast-0645427 and nsf ast-1313319 . we thank the dedicated efforts of the sounding rocket staff at the nasa wallops flight facility and white sands missile range and also thank dr . allan smith , dr . keith lykke , and dr . steven brown ( nist ) for the laboratory calibration of the lrs . this publication makes use of data products from the 2mass , which is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by the nasa and the nsf . this research has made use of the simbad database , operated at cds , strasbourg , france , and the spex library . arai , t. , matsuura , s. , bock , j. , et al . 2015 , , 806 , 69 bock , j. , battle , j. , cooray , a. , et al . 2006 , new a rev . , 50 , 215 bock , j. , sullivan , i. , arai , t. , et al . 2013 , , 207 , 32 bohlin , r. c. , & gilliland , r. l. , 2004 , , 127 , 3508 ( bg ) cohen , m. , wheaton , wm . a. , & megeath , s. t. 2003 , , 126 , 1090 cushing , m. c. 2005 , , 623 , 1115 fisher , j. , baumback , m. m. , bowles , j. h. , grossmann , j. m. , & antoniades , j. a. 1998 , proc spie , 3438 , 23 garnett , j. , d.p . , & forrest , w.j . , 1993 , spie , 1946 , 395 g gliese , w. 1971 , veroeffentlichungen des astronomischen rechen - instituts heidelberg , 24 , 1 hauser , m.g . & dwek , e. , 2001 , ara&a , 39 , 249 houk , n. , & swift , c. 1999 , university of michigan catalogue of two - dimensional spectral types for the hd stars , vol . 5 ( ann arbor : univ . michigan ) jaschek , m. 1978 , cds inf . 15 , 121 jaschek , c. , & jaschek , m. 1973 , in iau symp . 50 , spectral classification and multicolour photometry , ed . c. fehrenbach & b. e. westerlund ( dordrecht : reidel ) , 43 jordi , c. , gebran , m. , carrasco , j. m. , et al . 2010 , a&a , 523 , a48 joyce , r. r. , hinkle , k. h. , wallace , l. , dulick , m. , & lambert , d. l. , 1998 , , 116 , 2520 kessler , m. f. , steinz , j. a. , & anderegg , m. e. , et al . 1996 , a&a , 315 , l27 korngut , p. m. , renbarger , t. , arai , t. , et al . 2013 , , 207 , 34 lee , d. h. , kim , m. g. , tsumura , k. , et al . 2010 , journal of astronomy and space sciences , 27 , 401 leinert , ch . , bowyer , s. , haikala , l. k. , et al . 1998 , a&as , 127 , 1l madau , p. & pozzetti , l. , 2000 , mnras , 312 , l9-l15 matsumoto , t. , matsuura , s. , murakami , h. , et al . 2005 , , 626 , 31 matsuura , m. , yamamura , i. , murakami , h. , freund , m. m. , & tanaka , m. 1999 , a&a , 348 , 579 matsuura , s. , arai , t. , bock , j. , et al . 2016 , apj , submitted meyer , m. r. , edwards , s. , hinkle , k. h. , & strom , s. e. , 1998 , , 508 , 397 murakami , h. , freund , m. m. , ganga , k. , et al . 1996 , , 48 , l41 peterson , d. e. , megeath , s. t. , luhman , k. l. , et al . 2008 , , 685 , 313 perryman , m. a. c. , de boer , k. s. , gilmore , g. , et al . 2001 , a&a , 369 , 339 pickles , a. j. 1998 , , 110 , 863 rayner , j. t. 2009 , , 185 , 289 roeser , s. & bastian , u. , 1988 , a&a , 74 , 449 russell , h. n. 1934 , , 79 , 317 schlegel , d. j. 1998 , , 500 , 525 skrutskie , m. f. , cutri , r. m. , stiening , r. , et al . 2006 , , 131 , 1163s sorahana , s. , yamamura , i. 2014 , , 793 , 47 tsumura , k. , battle , j. , bock , j. , et al . 2010 , , 719 , 394 tsumura , k. , arai , t. , battle , j. , et al . 2013 , , 207 , 33 tsuji , t. , ohnaka , k. , aoki , w. , & yamamura , i. 1997 , a&a , 320 , l1 tsuji , t. 2000 , , 538 , 801 tsuji , t. 2001 , a&a , 376 , l1 tsuji , t. 2015 , pasj , 67 , 26 t wenger , m. , ochsenbein , f. , egret , d. , et al . 2000 , a&as , 143 , 9 wing , r. f. , & spinrad , h. 1970 , , 159 , 973 woolf , n. j. , schwarzschild , m. , & rose , w. k. 1964 , , 140 , 833 yuan , h. b. , liu , x. w. , & xiang , m. s. 2013 , , 430 , 2188 zemcov , m. , arai , t. , battle , j. , et al . 2013 , , 207 , 31 ccc elat10 - 2 & 15:07:60.0 & -2:00:00 + elat30 - 2 & 14:44:00 & 20:00:00 + elat30 - 3 & 15:48:00 & 9:30:00 + elat10 - 4 & 12:44:00 & 8:00:00 + elat30 - 4 & 12:52:00 & 27:00:00 + nep & 18:00:00 & 66:20:23.987 + swire & 16:11:00 & 55:00:00 + bootesa & 14:33:54.719 & 34:53:2.396 + bootesb & 14:29:17.761 & 34:53:2.396 + lockman & 10:45:12.0 & 58:00:00 + dgl & 16:47:60.0 & 69:00:00 cccccccccccccc & elat10 & e102_1 & tyc5000 - 614 - 1 & 15:06:50.134 & -00:02:47.746 & 9.020 & 8.283 & 8.283 & 7.608 & k2 & k3iii & 0.720 & ... + & elat10 & e102_2 & ... & 14:59:05.568 & -01:08:23.294 & 9.095 & 8.279 & 8.350 & 7.484 & ... & m0iiib & 4.582 & ... + & elat10 & e102_3 & hd131553 & 14:54:20.898 & -01:52:19.938 & 9.576 & 9.241 & 8.673 & 8.472 & f0v & g0ib - ii & 0.522 & ... + & elat10 & e102_4 & hd134456 & 15:09:58.320 & -00:52:47.269 & 7.872 & 7.754 & 6.982 & 6.854 & f2iii & f2iii - iv & 0.076 & ... + & elat10 & e102_5 & tyc5001 - 847 - 1 & 15:14:43.328 & -01:31:43.763 & 9.940 & 9.633 & 9.226 & 8.898 & ... & f8ib & 0.416 & ... + & elat10 & e102_6 & bd-01 - 3038 & 15:14:15.481 & -01:37:09.268 & 8.273 & 7.633 & 7.477 & 6.862 & k0 & m0.5v & 0.462 & ... + & elat10 & e102_7 & hd133213 & 15:03:28.468 & -03:10:05.732 & 8.802 & 8.751 & 8.066 & 8.030 & a2iii & f5ii - iii & 0.086 & ... + & elat30 & e302_1 & bd+22 - 2745 & 14:46:03.405 & 22:04:37.528 & 8.065 & 7.499 & 7.158 & 6.664 & g5 & k7v & 0.304 & ... + & elat30 & e302_2 & hd127666 & 14:32:02.149 & 22:04:47.600 & 8.645 & 8.396 & 7.866 & 7.676 & g5 & f8v & 0.045 & ... + & elat30 & e302_3 & hd131132 & 14:51:16.019 & 18:38:59.284 & 6.648 & 6.111 & 5.803 & 5.334 & k0 & g8iiife5 & 0.260 & ... + & elat30 & e302_4 & bd+19 - 2867 & 14:49:56.793 & 18:37:29.741 & 10.875 & 10.668 & 10.195 & 9.928 & g5 & g1ii - iiife-1ch0.5 & 0.705 & ... + & elat30 & e302_5 & bd+19 - 2857 & 14:45:32.922 & 18:40:20.255 & 7.342 & 6.643 & 6.466 & 5.815 & k2 & m0v & 0.234 & ... + & elat30 & e302_6 & tyc1481 - 620 - 1 & 14:46:48.921 & 17:30:12.359 & 10.208 & 9.644 & 9.620 & 9.138 & ... & k4v & 0.551 & ... + & elat30 & e302_7 & bd+18 - 2928 & 14:45:45.544 & 17:30:17.950 & 6.555 & 5.752 & 5.752 & 5.050 & m0 & k3iiife-0.5 & 1.488 & ... + & nep & n2_1 & bd+68 - 954 & 17:43:43.944 & 68:24:26.593 & 10.067 & 9.742 & 9.394 & 9.168 & f5 & f0ii & 0.064 & ... + & nep & n2_2 & ... & 17:38:56.867 & 66:22:12.587 & 10.726 & 10.216 & 10.440 & 9.937 & ... & g5iiia & 0.240 & ... + & nep & n2_3 & bd+67 - 1039a & 17:52:45.953 & 67:00:12.935 & 8.925 & 8.587 & 8.571 & 8.130 & ... & f8ib & 0.045 & ... + & nep & n2_4 & tyc4208 - 116 - 1 & 17:49:23.407 & 65:28:22.606 & 7.646 & 6.837 & 6.840 & 6.047 & ... & k4iii & 0.807 & ... + & nep & n2_5 & bd+67 - 1067 & 18:20:50.229 & 67:55:01.776 & 8.199 & 7.694 & 7.430 & 6.939 & k0 & k3v & 0.119 & ... + & nep & n2_6 & hd166779 & 18:07:35.504 & 63:54:12.298 & 6.544 & 5.874 & 5.706 & 5.078 & k5 & m0.5v & 0.221 & ... + & swire & s2_1 & hd144245 & 16:01:58.920 & 56:36:03.496 & 6.921 & 6.238 & 6.173 & 5.505 & k5 & k3iii & 0.201 & ... + & swire & s2_2 & hd144082 & 16:01:09.819 & 56:26:23.172 & 7.929 & 7.644 & 7.135 & 6.944 & f5 & g1vfe-0.5 & 0.051 & ... + & swire & s2_3 & hd147733 & 16:20:51.242 & 54:23:10.320 & 8.172 & 8.125 & 7.414 & 7.351 & a3 & f8iv & 0.059 & ... + & swire & s2_4 & hd234317 & 16:32:27.630 & 54:20:14.320 & 8.713 & 8.283 & 7.999 & 7.564 & g5 & k1v & 0.081 & ... + & swire & s2_5 & hd146736 & 16:15:15.896 & 52:01:48.338 & 8.929 & 8.618 & 8.140 & 7.884 & g5 & f9iiia & 0.060 & ... + & bootesa & ba2_1 & hd126878 & 14:27:13.534 & 34:43:19.996 & 8.631 & 8.385 & 7.783 & 7.640 & f5 & f7iii & 0.046 & ... + & bootesa & ba2_2 & tyc2557 - 719 - 1 & 14:41:46.727 & 33:34:23.452 & 10.800 & 10.557 & 10.045 & 9.783 & ... & f2iii - iv & 0.331 & ... + & bootesa & ba2_3 & tyc2556 - 652 - 1 & 14:33:46.073 & 33:34:53.886 & 10.341 & 9.620 & 9.352 & 8.717 & k9v & m1.5v & 1.125 & high - proper - motion + & bootesa & ba2_4 & bd+34 - 2527 & 14:25:57.827 & 33:34:32.984 & 9.846 & 9.426 & 9.250 & 8.973 & g5iii & g5v & 0.120 & ... + & bootesa & ba2_5 & hd126210 & 14:23:24.060 & 33:34:19.099 & 8.480 & 8.274 & 7.653 & 7.492 & f8 & f7v & 0.039 & ... + & bootesa & ba2_6 & bd+34 - 2522 & 14:21:54.490 & 33:34:35.580 & 7.311 & 6.514 & 6.307 & 5.545 & k5 & k3iiife-0.5 & 0.584 & ... + & bootesa & ba2_7 & ... & 14:41:50.085 & 32:24:33.790 & 10.848 & 10.330 & 10.178 & 9.587 & ... & m2v & 1.521 & ... + & bootesa & ba2_8 & tyc2553 - 127 - 1 & 14:29:10.917 & 32:27:40.871 & 10.252 & 9.490 & 9.130 & 8.483 & ... & k2iii & 1.255 & ... + & bootesb & bb2_1 & tyc2560 - 1157 - 1 & 14:38:39.909 & 35:31:13.224 & 9.347 & 8.799 & 8.611 & 8.100 & k1 & g8iiife5 & 0.143 & ... + & bootesb & bb2_2 & bd+36 - 2489 & 14:24:52.634 & 35:32:12.714 & 9.026 & 8.530 & 8.773 & 8.484 & g5 & g7iv & 0.107 & ... + & bootesb & bb2_3 & bd+32 - 2490 & 14:34:03.366 & 32:06:02.588 & 9.640 & 9.089 & 8.835 & 8.414 & k0 & g8iiife1 & 0.127 & ... + & bootesb & bb2_4 & bd+31 - 2630 & 14:33:01.264 & 30:56:33.554 & 10.240 & 9.793 & 9.504 & 9.246 & ... & f9v & 0.336 & ... + & bootesb & bb2_5 & tyc2553 - 961 - 1 & 14:24:21.497 & 30:58:03.684 & 10.323 & 9.713 & 9.351 & 8.864 & ... & g8iiife1 & 0.580 & ... + & elat30 & e303_1 & bd+11 - 2874 & 15:52:08.230 & 10:52:28.103 & 7.882 & 7.169 & 6.692 & 6.012 & k5v & m0.5v & 0.330 & spectroscopic binary + & elat30 & e303_2 & hd141631 & 15:49:47.057 & 10:48:24.520 & 8.251 & 7.922 & 7.555 & 7.096 & k2 & g4o - ia & 0.206 & ... + & elat30 & e303_3 & tyc947 - 300 - 1 & 15:50:53.577 & 09:41:15.828 & 10.379 & 9.841 & 9.861 & 9.310 & ... & k1iiife-0.5 & 0.595 & ... + & elat30 & e303_4 & hd141531 & 15:49:16.496 & 09:36:42.408 & 7.718 & 7.052 & 6.971 & 6.337 & k & m1v & 0.089 & ... + & nep & n3_1 & hd164781 & 17:57:03.647 & 68:49:19.744 & 8.948 & 8.601 & 7.733 & 7.423 & k0 & g8v & 0.076 & ... + & nep & n3_2 & tyc4428 - 1122 - 1 & 17:54:46.231 & 68:06:42.016 & 9.753 & 9.250 & 9.009 & 8.353 & ... & k1iiibcn1.5ca1 & 0.629 & ... + & nep & n3_3 & bd+67 - 1050 & 18:06:45.898 & 67:50:40.686 & 8.273 & 7.722 & 7.485 & 6.976 & k2 & k1iiibcn1.5ca1 & 0.134 & ... + & nep & n3_4 & bd+65 - 1248 & 18:12:21.398 & 65:36:17.381 & 7.214 & 6.492 & 6.359 & 5.635 & k5 & k5iii & 0.919 & ... + & nep & n3_5 & hd166779 & 18:07:35.504 & 63:54:12.298 & 6.711 & 6.077 & 5.706 & 5.078 & k5 & m0.5v & 0.455 & ... + & nep & n3_6 & tyc4226 - 812 - 1 & 18:25:26.020 & 66:00:38.783 & 9.655 & 9.417 & 8.924 & 8.714 & ... & f8ia & 0.293 & ... + & swire & s3_1 & bd+55 - 1802 & 16:01:45.359 & 54:48:40.882 & 10.325 & 10.033 & 9.570 & 9.330 & g0 & g2iv & 0.392 & ... + & swire & s3_2 & tyc3870 - 1085 - 1 & 15:54:21.929 & 53:36:47.786 & 10.417 & 10.198 & 9.554 & 9.300 & ... & g2ii - iii & 0.871 & ... + & swire & s3_3 & tyc3870 - 366 - 1 & 15:53:29.099 & 53:28:36.008 & 8.669 & 8.062 & 7.928 & 7.281 & ... & m1v & 0.285 & ... + & swire & s3_4 & tyc3877 - 704 - 1 & 16:10:22.667 & 54:28:38.784 & 9.017 & 8.472 & 8.258 & 7.715 & ... & k1iiibcn1.5ca1 & 0.239 & ... + & swire & s3_5 & tyc3877 - 1592 - 1 & 16:01:43.031 & 53:06:25.855 & 10.233 & 9.746 & 9.566 & 9.077 & ... & g9iii & 0.136 & ... + & swire & s3_6 & tyc3878 - 216 - 1 & 16:25:31.829 & 53:25:25.453 & 9.065 & 8.709 & 8.364 & 8.020 & ... & g1iiich1 & 0.214 & ... + & lockman & l3_1 & v*dm - uma & 10:55:43.521 & 60:28:09.613 & 7.975 & 7.476 & 7.194 & 6.621 & k0iii & g2ib & 0.233 & ... + & lockman & l3_2 & hd94880 & 10:58:21.518 & 59:16:53.422 & 7.787 & 7.482 & 6.900 & 6.629 & g0 & g0ib - ii & 0.115 & ... + & lockman & l3_3 & hd92320 & 10:40:56.905 & 59:20:33.065 & 7.947 & 7.662 & 7.148 & 6.852 & g0 & f2-f5ib & 0.109 & high - proper - motion + & lockman & l3_4 & hd237955 & 10:57:44.114 & 58:10:01.103 & 9.799 & 9.619 & 8.705 & 8.508 & g0 & f5iii & 0.038 & ... + & lockman & l3_5 & tyc3827 - 847 - 1 & 11:01:59.570 & 56:58:11.510 & 9.498 & 9.094 & 8.816 & 8.279 & ... & m2v & 0.479 & ... + & lockman & l3_6 & hd237961 & 11:00:12.007 & 56:59:49.481 & 9.267 & 9.049 & 8.495 & 8.271 & g0 & g1vfe-0.5 & 0.304 & ... + & bootesa & ba3_1 & bd+362491 & 14:26:05.241 & 35:50:00.776 & 8.897 & 8.498 & 8.095 & 7.676 & k0 & g3ii & 0.515 & ... + & bootesa & ba3_2 & hd128368 & 14:35:32.053 & 34:41:11.540 & 7.436 & 6.789 & 6.530 & 5.942 & k0 & m0.5v & 0.215 & ... + & bootesa & ba3_3 & bd+35 - 2576 & 14:32:31.567 & 34:42:09.493 & 9.291 & 8.834 & 9.058 & 8.737 & k0 & f5ib - g1ib & 0.143 & ... + & bootesa & ba3_4 & hd126878 & 14:27:13.534 & 34:43:19.996 & 9.190 & 9.091 & 7.783 & 7.640 & f5 & f2iii - iv & 0.060 & ... + & bootesb & bb3_1 & tyc2560 - 1157 - 1 & 14:38:39.909 & 35:31:13.224 & 9.416 & 8.918 & 8.611 & 8.100 & k1 & k4v & 0.124 & ... + & bootesb & bb3_2 & bd+32 - 2503 & 14:41:07.455 & 32:04:45.095 & 9.628 & 9.449 & 8.853 & 8.624 & ... & f8ib & 0.198 & ... + & bootesb & bb3_3 & bd+32 - 2456 & 14:18:52.718 & 32:06:31.003 & 9.191 & 8.531 & 7.992 & 7.444 & k2iii & k0.5iiicn1 & 0.534 & ... + & bootesb & bb3_4 & bd+31 - 2630 & 14:33:01.264 & 30:56:33.554 & 10.170 & 9.940 & 9.504 & 9.246 & ... & f9v & 0.438 & ... + & elat10 & e104_1 & hd111645 & 12:50:42.449 & 08:52:30.238 & 8.908 & 8.691 & 8.124 & 7.920 & f8 & f7iii & 0.041 & ... + & elat10 & e104_2 & bd+11 - 2491 & 12:46:07.870 & 11:09:25.744 & 10.229 & 9.992 & 9.486 & 9.201 & f8 & f2-f5ib & 0.162 & ... + & elat10 & e104_3 & ... & 12:41:28.720 & 10:52:57.907 & 10.959 & 10.368 & 10.599 & 10.096 & ... & k5v & 0.702 & ... + & elat10 & e104_4 & hd110777 & 12:44:20.102 & 06:51:16.916 & 8.442 & 8.212 & 7.663 & 7.418 & g0 & f8ia & 0.148 & ... + & elat10 & e104_5 & bd+10 - 2440 & 12:33:51.920 & 09:31:54.156 & 8.139 & 7.372 & 6.662 & 5.860 & ... & k3ii - iii & 1.012 & ... + & elat10 & e104_6 & hd109824 & 12:37:48.044 & 04:59:07.195 & 6.860 & 6.296 & 6.092 & 5.542 & k0 & k0.5iib & 0.570 & ... + & elat30 & e304_1 & ... & 13:02:54.144 & 26:23:27.762 & 8.966 & 8.441 & 8.267 & 7.756 & ... & k1iiibcn1.5ca1 & 0.478 & ... + & elat30 & e304_2 & bd+27 - 2207 & 13:02:50.671 & 26:50:00.402 & 10.924 & 10.630 & 10.141 & 9.899 & f8 & f8ib & 0.262 & ... + & elat30 & e304_3 & tyc1995 - 264 - 1 & 13:02:50.439 & 27:29:22.283 & 10.212 & 10.004 & 9.586 & 9.251 & ... & g1vfe-0.5 & 0.121 & ... + & elat30 & e304_4 & bd+27 - 2197 & 12:57:45.577 & 27:01:51.600 & 10.562 & 10.374 & 9.873 & 9.672 & f5 & f2ib & 0.098 & ... + & elat30 & e304_5 & tyc1995 - 1123 - 1 & 12:57:25.736 & 28:18:25.992 & 9.837 & 9.006 & 8.997 & 8.229 & ... & m1.5v & 0.608 & ... + & elat30 & e304_6 & lp322 - 154 & 12:57:04.818 & 29:30:36.860 & 10.454 & 9.808 & 9.740 & 9.096 & k5v & m0.5v & 1.460 & high - proper - motion + & elat30 & e304_7 & tyc2532 - 820 - 1 & 12:56:45.236 & 30:44:22.556 & 10.678 & 10.006 & 9.838 & 9.324 & k1v & m1v & 0.344 & ... + & nep & n4_1 & bd+68 - 951 & 17:38:51.760 & 68:13:16.536 & 9.137 & 8.449 & 7.942 & 7.438 & k0 & k1.5iiife-0.5 & 0.273 & multiple - star + & nep & n4_2 & hd161500 & 17:41:10.318 & 65:13:10.301 & 7.442 & 6.860 & 6.633 & 6.119 & k2 & k1iiibcn1.5ca1 & 0.312 & ... + & nep & n4_3 & g227 - 20 & 17:52:11.850 & 64:46:08.720 & 9.077 & 8.391 & 8.249 & 7.615 & m0.5v & m1.5v & 0.449 & high - proper - motion + & nep & n4_4 & tyc4208 - 1599 - 1 & 17:52:05.421 & 64:37:15.827 & 10.278 & 9.725 & 9.929 & 9.259 & ... & m2v & 0.486 & ... + & nep & n4_5 & bd+64 - 1227a & 17:52:17.178 & 64:14:16.411 & 8.816 & 8.500 & 8.400 & 8.125 & ... & f8ib & 0.046 & ... + & nep & n4_6 & tyc4213 - 161 - 1 & 18:03:24.923 & 67:12:41.681 & 10.171 & 9.868 & 9.327 & 9.115 & ... & f7iii & 0.109 & ... + & nep & n4_7 & bd+66 - 1074 & 18:03:15.008 & 66:20:29.069 & 7.609 & 6.866 & 6.739 & 6.046 & k5 & k3ii - iii & 1.262 & ... + & nep & n4_8 & hd170592 & 18:25:24.759 & 65:45:34.470 & 7.474 & 7.143 & 6.722 & 6.409 & k0 & g5v & 0.148 & ... + & swire & s4_1 & tyc3870 - 1026 - 1 & 15:55:16.319 & 54:45:12.510 & 10.127 & 9.564 & 9.332 & 8.829 & ... & k3v & 0.261 & ... + & swire & s4_2 & tyc3496 - 1361 - 1 & 15:56:04.610 & 52:13:29.543 & 8.240 & 7.566 & 7.519 & 6.825 & ... & k3iii & 0.421 & ... + & swire & s4_3 & tyc3880 - 1133 - 1 & 16:03:15.627 & 56:02:35.210 & 8.711 & 7.821 & 7.791 & 6.995 & ... & m2.5iiiba0.5 & 2.347 & ... + & swire & s4_4 & tyc3877 - 484 - 1 & 16:03:12.065 & 54:44:27.658 & 9.047 & 8.361 & 7.846 & 7.288 & ... & k2iiife-1 & 0.147 & ... + & swire & s4_5 & hd234308 & 16:26:05.554 & 52:18:08.266 & 8.652 & 8.101 & 7.932 & 7.407 & k0 & k1iiife-0.5 & 0.237 & ... + & dgl & d4_1 & tyc4419 - 1623 - 1 & 16:14:22.875 & 69:55:54.455 & 10.093 & 9.624 & 9.419 & 8.810 & ... & m2v & 0.373 & ... + & dgl & d4_2 & tyc4419 - 1631 - 1 & 16:18:10.929 & 69:16:36.761 & 9.923 & 9.466 & 9.229 & 8.916 & ... & k1v & 0.124 & ... + & dgl & d4_3 & bd+67 - 943 & 16:29:52.210 & 66:47:45.154 & 9.390 & 9.120 & 8.606 & 8.417 & f8 & f8ia & 0.110 & ... + & dgl & d4_4 & tyc4196 - 2280 - 1 & 16:34:34.354 & 65:36:05.818 & 10.424 & 9.946 & 9.783 & 9.339 & ... & g4v & 0.232 & ... + & dgl & d4_5 & hd151286 & 16:40:37.776 & 70:34:14.772 & 7.110 & 6.668 & 6.237 & 5.794 & ... & g3ii & 0.070 & ... + & dgl & d4_6 & bd+69 - 873 & 16:47:31.365 & 68:51:02.603 & 8.338 & 7.820 & 7.495 & 7.010 & k0 & g7.5iiia & 0.111 & ... + & dgl & d4_7 & hd154273 & 16:58:40.137 & 69:38:05.431 & 7.022 & 6.508 & 6.197 & 5.746 & k0 & g7.5iiia & 0.106 & ... + & dgl & d4_8 & tyc4424 - 1380 - 1 & 17:08:33.058 & 71:00:28.044 & 9.242 & 8.911 & 9.008 & 8.727 & ... & g2iv & 0.109 & ... + & dgl & d4_9 & tyc4421 - 2278 - 1 & 17:16:54.688 & 67:38:26.279 & 8.993 & 8.460 & 8.269 & 7.792 & ... & k1iiife-0.5 & 0.174 & ... + & bootesb & bb4_1 & tyc2557 - 870 - 1 & 14:40:08.540 & 34:40:29.669 & 10.107 & 9.545 & 9.249 & 8.768 & ... & m2v & 0.331 & ... + & bootesb & bb4_2 & hd128094 & 14:34:10.846 & 30:59:10.356 & 7.857 & 7.240 & 6.963 & 6.405 & k0 & k2iii & 0.226 & ... + & bootesb & bb4_3 & tyc2559 - 388 - 1 & 14:34:47.808 & 35:34:09.419 & 9.761 & 9.346 & 9.011 & 8.550 & g8v & g6iii & 0.184 & ... + & bootesb & bb4_4 & tyc2553 - 947 - 1 & 14:28:52.868 & 31:30:30.316 & 8.505 & 7.763 & 7.642 & 6.917 & ... & k2iii & 0.170 & ... + & bootesb & bb4_5 & v*kt - boo & 14:29:02.513 & 33:50:38.929 & 8.699 & 8.271 & 7.846 & 7.465 & g & g0ib - ii & 0.074 & ... + & bootesb & bb4_6 & hd126210 & 14:23:24.060 & 33:34:19.099 & 8.764 & 8.749 & 7.653 & 7.492 & f8 & f1ii & 0.194 & ... + & bootesb & bb4_7 & tyc2549 - 413 - 1 & 14:23:23.452 & 34:33:24.854 & 9.399 & 8.885 & 8.510 & 7.947 & ... & k1iiibcn1.5ca1 & 0.269 & ... +

we report on the x - ray spectral analysis of the black hole candidate xte j1752223 in the 20092010 outburst , utilizing data obtained with the maxi / gas slit camera ( gsc ) , the swift / xrt , and suzaku , which work complementarily . as already reported by maxi monitored the source continuously throughout the entire outburst for about eight months . all the maxi / gsc energy spectra in the high / soft state lasting for 2 months are well represented by a multi - color disk plus power - law model . the innermost disk temperature changed from.7 kev to.4 kev and the disk flux decreased by an order of magnitude . nevertheless , the innermost radius is constant at km , where is the source distance in units of 3.5 kpc and the inclination . the multi - color disk parameters obtained with the maxi / gsc are consistent with those with the swift / xrt and suzaku . the suzaku data also suggests a possibility that the disk emission is slightly comptonized , which could account for broad iron - k features reported previously . assuming that the obtained innermost radius represents the innermost stable circular orbit for a non - rotating black hole , we estimate the mass of the black hole to be 5.51.28 , where the correction for the stress - free inner boundary condition and color hardening factor of 1.7 are taken into account . if the inclination is less than 49 as suggested from the radio monitoring of transient jets and the soft - to - hard transition in 2010 april occurred at 14% of eddignton luminosity , the fitting of the suzaku spectra with a relativistic accretion - disk model derives constraints on the mass and the distance to be 3.155 and 2.322 kpc , respectively . this confirms that the compact object in xte j1752223 is a black hole .

1109.1378

galactic black hole candidates ( bhcs ) show various states characterized by their spectral shapes , temporal properties , and luminosities . they mostly take two major states , the `` low / hard state '' and the `` high / soft state '' , which are referred to as `` the hard x - ray state '' and `` the thermal dominant state '' , respectively , in more recent classification ( see @xcite and references therein ) since the state of a bhc is not always determined by an x - ray luminosity alone . in the low / hard state , the x - ray energy spectra are dominated by a power - law component with a photon index of @xmath01.7 and a high energy cutoff at @xmath0100 kev that shows strong short - time variability @xcite . they can be explained by thermal comptonization of soft photons from the accretion disk by hot plasmas with a temperature of @xmath010@xmath8 k ( e.g. , @xcite , @xcite , @xcite ) . in a typical outburst of a transient bhc , it generally exhibits a spectral transition from the low / hard to the high / soft state through the intermediate ( or very high ) state when the luminosity reaches @xmath010% of the eddingtion limit . the x - ray spectra in the high / soft state are characterized by a ultra - soft component which is considered to originate from an optically - thick and geometrically - thin accretion disk ( so called `` standard disk '' ; @xcite ) . this emission can be successfully described by the multi - color disk ( mcd ) model ( @xcite , @xcite ) with the innermost temperature of @xmath01 kev . an important signature in the high / soft state is that the innermost radius ( @xmath9 ) of the accretion disk is kept constant , independent of the flux or innermost disk temperature ( e.g. , @xcite ) . hence , it is believed that the @xmath9 reflects the innermost stable circular orbit ( isco ) , which is determined through general relativity by the mass and angular momentum of the black hole ; the radius of isco for a non - spinning black hole is 6@xmath10 ( @xmath11 is the gravitational radius , where @xmath12 , @xmath13 , @xmath14 are the gravitational constant , black hole mass , and light velocity , respectively ) . by assuming that the obtained innermost @xmath9 to the isco , the black hole mass can be estimated from the x - ray spectrum alone for a given spin parameter . xte j1752223 was first discovered with the proportional counter array ( pca ) onboard rossi x - ray timing explorer ( rxte ) on 2009 october 23 ( mjd 55127 ) during a galactic bulge scan observation ( @xcite ) . the gas slit camera ( gsc ) onboard monitor of all - sky x - ray image ( maxi ; @xcite ) detected the source simultaneously @xcite . as described in ( 2010 ; hereafter paper i ) , maxi continuously monitored the source during the entire outburst until 2010 june . the maxi light curves and hardness - intensity diagram revealed that the source initially stayed in the low / hard states with two stable flux levels for about three months , and then moved into the high / soft state ( paper i ) . a radio flare was detected simultaneously with the spectral transition ( @xcite , @xcite ) , and the proper motion of the jet ejecta was later found with vlbi observations ( @xcite ) . @xcite identified it as an approaching jet ejected coincidently with the x - ray state transition . from the observed flux of the approaching jet and its upper limit of the receding one , they constrained the jet speed and the inclination angle from the line - of - sight to be @xmath15 and @xmath16 , respectively , by assuming that the axes of the twin jets are aligned . @xcite estimated the black hole mass and distance to be 9.6@xmath40.9 @xmath5 and 3.5@xmath40.4 kpc , respectively , using the spectral - timing correlation technique ( @xcite ) . however , the compact object mass has not been estimated via optical mass - function technique . to firmly establish the nature of xte j1752223 , however , it is quite important to constrain the black hole mass by an independent approach . in this paper , we present the results from a detailed spectral analysis of xte j1752223 utilizing the maxi / gsc data , together with those of swift / xrt and suzaku , mainly focusing on the spectra in the high / soft state . we then discuss the constraints on the black hole mass and distance based on our results . in the appendix , we present the current status of spectral calibration of the maxi / gsc using the crab nebula , which is fully reflected in this paper and in a similar work for the black hole candidate gx 3394 reported by @xcite . the spectral fitting was carried out on xspec version 12.6 . errors are quoted at statistical 90% confidence limits for a single parameter throughout the paper . as the first astronomical mission on the international space station ( iss ) , maxi has been operating since 2009 august . the payload carries two kinds of x - ray all - sky monitors ; the gas - slit camera ( gsc ; @xcite ) and the solid - state slit camera ( ssc ; @xcite , @xcite ) . the gsc consists of xe - gas proportional counters for the x - ray detector and slat collimators with a slit to localize the direction of the incident x - rays . the counters employ resistive - carbon wires for detector anodes to determine the x - ray position along the anode wire . twelve identical units ( refereed to as gsc_0 , ... , gsc_9 , gsc_a and gsc_b ) are assembled so that they instantaneously cover two rectangular field of views ( fovs ) of 3@xmath17160 deg@xmath18 with an equal area . after the operation started in 2009 august , two counters , gsc_6 and gsc_9 were stopped on 2009 september 8 and 14 , respectively , for high - voltage breakdown . other two counters on gsc_a and gsc_b were stopped temporarily on september 23 , because diagnostic data indicated that they may also break down rather soon . consequently , the outburst of xte j1752223 from 2009 october to 2010 june was covered by eight gsc camera units out of the twelve . the gsc fov is limited by slats collimators . the visibility for a target at a given sky position changes according to the iss orbital motion ; hereafter we call the visibility time as `` transit '' . each transit lasts for 40 - 150 s , and the effective area of each gsc counter change due to the triangular - shaped collimator transmission function , with a peak value of 4 - 5 cm@xmath18 . the maxi / gsc scanned the direction of xte j1752223 2041 times in total from the discovery on 2009 october 23 ( mjd 55127 ) to 2010 june 3 ( mjd 55350 ) . the total exposure times effective area becomes 534.8 cm@xmath18 ksec . for the data analysis , we used the maxi specific analysis tools , which were developed by the maxi team . we analyzed the gsc event data version 0.3b , which include the data taken by counters operated at the nominal high voltage ( = 1650 v ) but excluding those of anode # 1 and # 2 whose energy responses have not been enough calibrated yet . we discarded events taken while the gsc fovs were interfered by the solar panels and other iss payloads . the events detected at the anode - end area were also screened since the background is higher therein . these event were cut with a condition that the photon incident angle ( @xmath20 ; see @xcite for the definition ) is higher than 36@xmath7 . as shown in figure [ fig : contour_region ] we carefully extracted events for source and background from circular regions with a 1.5@xmath7 radius , excluding regions within 1.7@xmath7 of nearby sources , gx 51 , gx 9 + 1 and sax j1748.92021 in ngc 6440 . we used only those data when the source and background regions were both fully scanned in a transit . we performed spectral analysis of the data taken between mjd=55200 and mjd=55293 , during which both the `` hard - to - soft '' and `` soft - to - hard '' transitions took place ( paper i ) . the net effective exposure was 194.5 cm@xmath18 ksec for 752 transits . we divided the whole dataset into groups with typical lengths of a few days so that the spectrum in each group has enough photon statistics , except for the epoch around the `` hard - to - soft '' state transition when the spectrum changed rapidly . eventually , data were separated into 49 groups whose exposures times effective area were 0.2@xmath012 cm@xmath18 ksec . the energy response matrix for each group was calculated by the response builder ( @xcite ) . the validity of the energy response was verified using the crab nebula ( see appendix ) . ( 80mm,80mm)figure1.eps ( 80mm,80mm)figure2.eps figure [ fig : lightcurves ] shows the maxi / gsc light curves of xtej1752223 during the 20092010 outburst in the 24 kev , 410 kev , and 1020 kev bands , together with the hardness ratio between the 410 kev and 24 kev bands . they are updated from those presented in paper i , after applying the latest calibration and the same data screening as used for the spectral analysis ( section [ subsec - spec ] and appendix ) . the maxi / gsc first detected xte j1752223 at 15:05 ( ut ) on 2009 october 23 ( mjd 55127 ) when the 220 kev x - ray intensity was 30 mcrab . it monitored the source almost uniformly except for the time when the fov was close to the sun ( mjd 5515455159 and mjd 5517255188 ) and when poles of the scanning axis was close to the source ( around mjd 55300 ) . the light curves reveal that the source stayed in the initial low / hard state for the extraordinary long period of about 3 months , and took two flux - plateau phases meantime ( paper i ) . the `` hard - to - soft '' state transition occurred on mjd 55218 and the following behavior agrees well with those of the typical bhc outbursts ( @xcite ) . the x - ray intensity was peaked at 420 mcrab on 2010 january 22 ( mjd 55218 ) , then decayed exponentially with an e - folding time of 34 days through the high / soft state . the `` soft - to - hard '' transition started on 2010 march 30 ( mjd 55285 ) , and then the source returned to the low / hard state on 2010 april 6 ( mjd 55292 ) . on 2010 june 28 ( mjd 55375 ) , the source flux fell below the maxi / gsc detection sensitivity per day ( 20 mcrab ) . thus , the total duration of the outburst became about eight months . for the spectral analysis , we employed the standard model for bhcs in the high / soft state , a multi - color disk ( mcd ; _ diskbb _ in xspec ) model plus a power - law representing the hard tail . the _ wabs _ ( @xcite ) model with solar abundances by @xcite was applied for the interstellar absorption . the hydrogen column density ( @xmath21 ) toward xte j1752223 was fixed at 0.6@xmath1710@xmath22 @xmath23 , based on the swift and suzaku results as described below . the model has four parameters : the innermost temperature @xmath24 and innermost radius @xmath9=@xmath25 @xmath26 , where @xmath27 is the normalization of the _ diskbb _ model . the distance to the source of @xmath28 kpc and inclination angle @xmath3=0 are assumed . ] for the mcd component , with the photon index @xmath29 and normalization at 1 kev for the power - law component . we first applied the model to all the data . when the mcd component was found to be not necessary ( i.e. , the 90% confidence range of the mcd normalization contains zero ) , only the power - law model was applied with its @xmath29 set free . when the mcd component was required , we fixed @xmath30 for all the spectra because it is often difficult to determine from individual spectrum due to poor statistics and the limited energy coverage of the maxi / gsc . this photon index corresponds to a typical value obtained when the power - law component is relatively strong ( @xmath3150% of the total flux in the 220 kev band ) . although @xcite suggest from the rxte / pca data that the power - law slope is variable between @xmath32 by using a spectral model different from ours , we confirmed that varying @xmath29 value within this range only changes the best - fit mcd normalization by @xmath33 , and hence does not affect our conclusions . ( 170mm,170mm)figure3.eps ( 170mm,170mm)figure4.eps rcccccccr dataid & mjd & exposure & @xmath29 & power - law & @xmath24 & @xmath9 & disk & @xmath34/dof + & start end & ( cm@xmath18 ksec ) & & flux@xmath35 & ( kev ) & ( km ) & flux@xmath35 & + 1 & 55200.0355200.66 & 3.42 & 1.71@xmath40.09 & 11.6@xmath40.7 & & & & 1.33(59 ) + 2 & 55200.7355201.30 & 3.41 & 1.67@xmath40.09 & 12.1@xmath40.7 & & & & 1.36(61 ) + 3 & 55201.4255202.00 & 3.42 & 1.73@xmath40.09 & 12.4@xmath40.7 & & & & 0.92(64 ) + 4 & 55202.0655202.76 & 3.41 & 1.77@xmath40.09 & 11.8@xmath40.7 & & & & 0.79(60 ) + 5 & 55202.8255203.71 & 3.41 & 1.63@xmath40.09 & 13.0@xmath40.7 & & & & 0.81(65 ) + 6 & 55203.7755204.72 & 3.40 & 1.75@xmath40.09 & 12.1@xmath40.7 & & & & 0.93(62 ) + 7 & 55205.4255206.56 & 2.37 & 1.70@xmath40.11 & 12.6@xmath40.8 & & & & 1.10(47 ) + 8 & 55207.3855208.73 & 2.01 & 1.76@xmath36 & 11.6@xmath40.9 & & & & 0.57(36 ) + 9 & 55209.4155211.84 & 2.01 & 1.67@xmath37 & 12.2@xmath40.9 & & & & 1.13(39 ) + 10 & 55214.8855214.94 & 0.24 & 1.51@xmath38 & 15.4@xmath39 & & & & 0.67(4 ) + 11 & 55215.6455215.83 & 0.83 & 2.09@xmath40.19 & 13.0@xmath41.3 & & & & 0.55(18 ) + 12 & 55215.8955216.15 & 0.83 & 2.00@xmath40 & 12.3@xmath41 & & & & 0.87(17 ) + 13 & 55216.6555216.91 & 1.06 & 2.20(fix ) & 10.7@xmath42 & 0.71@xmath43 & 13.3@xmath44 & 2.2@xmath40.6 & 1.09(24 ) + 14 & 55216.9755217.29 & 1.28 & 2.20(fix ) & 8.5@xmath45 & 0.79@xmath46 & 13.5@xmath47 & 4.2@xmath40.6 & 0.94(29 ) + 15 & 55217.4255218.37 & 3.39 & 2.20(fix ) & 8.0@xmath48 & 0.67@xmath40.07 & 23.2@xmath49 & 4.9@xmath40.4 & 0.86(51 ) + 16 & 55218.6255219.38 & 2.93 & 2.20(fix ) & 5.5@xmath41.0 & 0.67@xmath50 & 31.8@xmath51 & 9.2@xmath40.5 & 0.73(42 ) + 17 & 55219.5755220.39 & 3.22 & 2.20(fix ) & 4.7@xmath40.9 & 0.66@xmath40.04 & 33.1@xmath52 & 9.6@xmath53 & 0.77(43 ) + 18 & 55220.5855221.35 & 3.06 & 2.20(fix ) & 4.3@xmath40.9 & 0.66@xmath40.04 & 32.0@xmath54 & 8.7@xmath40.4 & 0.63(40 ) + 19 & 55221.5455222.30 & 2.27 & 2.20(fix ) & 3.0@xmath55 & 0.61@xmath40.04 & 40.1@xmath56 & 8.8@xmath40.5 & 1.11(33 ) + 20 & 55224.8455225.85 & 3.03 & 2.20(fix ) & 2.6@xmath40.8 & 0.65@xmath40.04 & 32.5@xmath57 & 8.0@xmath40.4 & 1.15(35 ) + 21 & 55226.3055227.31 & 3.31 & 2.20(fix ) & 3.6@xmath40.7 & 0.61@xmath40.04 & 36.9@xmath58 & 7.4@xmath40.4 & 0.63(39 ) + 22 & 55227.3755228.39 & 3.53 & 2.20(fix ) & 1.4@xmath40.7 & 0.62@xmath40.03 & 38.0@xmath59 & 8.2@xmath40.4 & 1.13(33 ) + 23 & 55228.4555229.34 & 3.33 & 2.20(fix ) & 1.5@xmath40.7 & 0.60@xmath40.04 & 38.2@xmath60 & 7.1@xmath40.4 & 1.19(30 ) + 24 & 55229.4155230.36 & 3.82 & 2.20(fix ) & 1.8@xmath40.6 & 0.55@xmath40.03 & 51.7@xmath61 & 7.6@xmath40.3 & 1.25(32 ) + 25 & 55230.5555231.31 & 3.40 & 2.20(fix ) & 2.7@xmath40.7 & 0.57@xmath40.04 & 44.3@xmath62 & 6.7@xmath40.4 & 1.29(34 ) + 26 & 55231.6355232.90 & 4.30 & 2.20(fix ) & 2.9@xmath63 & 0.56@xmath64 & 48.0@xmath65 & 6.8@xmath40.3 & 0.93(38 ) + 27 & 55237.1555237.62 & 2.66 & 2.20(fix ) & 1.5@xmath40.7 & 0.60@xmath40.05 & 36.0@xmath66 & 6.0@xmath40.4 & 0.79(26 ) + 28 & 55237.8755238.45 & 2.84 & 2.20(fix ) & 1.7@xmath40.6 & 0.58@xmath40.05 & 39.3@xmath67 & 5.8@xmath40.4 & 0.96(27 ) + 29 & 55238.8255239.14 & 1.56 & 2.20(fix ) & 0.5@xmath68 & 0.61@xmath40.06 & 34.3@xmath69 & 5.9@xmath40.5 & 1.32(18 ) + 30 & 55239.8455240.99 & 1.60 & 2.20(fix ) & 1.8@xmath40.9 & 0.58@xmath70 & 37.5@xmath71 & 5.6@xmath40.5 & 1.16(20 ) + 31 & 55247.0655247.89 & 3.22 & 2.20(fix ) & 0.7@xmath40.7 & 0.55@xmath40.05 & 40.2@xmath72 & 4.7@xmath40.3 & 0.73(23 ) + 32 & 55247.9555248.84 & 3.24 & 2.20(fix ) & 1.1@xmath73 & 0.55@xmath74 & 42.1@xmath75 & 4.6@xmath40.3 & 0.61(24 ) + 33 & 55248.9155249.86 & 3.17 & 2.20(fix ) & 1.0@xmath40.6 & 0.57@xmath40.05 & 36.0@xmath76 & 4.2@xmath40.3 & 0.92(24 ) + 34 & 55249.9255250.94 & 3.32 & 2.20(fix ) & 0.5@xmath40.5 & 0.55@xmath74 & 39.1@xmath77 & 4.2@xmath40.3 & 0.96(21 ) + 35 & 55251.0055252.02 & 3.22 & 2.20(fix ) & 0.7@xmath40.6 & 0.56@xmath40.05 & 36.7@xmath78 & 4.2@xmath40.3 & 0.41(23 ) + 36 & 55252.0855254.24 & 5.51 & 2.20(fix ) & 0.9@xmath40.4 & 0.53@xmath64 & 41.6@xmath79 & 3.7@xmath40.2 & 0.84(28 ) + 37 & 55257.0955259.12 & 3.83 & 2.20(fix ) & 1.0@xmath40.5 & 0.54@xmath40.05 & 36.9@xmath80 & 3.1@xmath40.2 & 0.66(23 ) + 38 & 55259.5255261.16 & 6.53 & 2.20(fix ) & 0.8@xmath40.4 & 0.52@xmath40.04 & 41.4@xmath81 & 3.1@xmath40.2 & 0.53(26 ) + 39 & 55261.4955263.19 & 6.82 & 2.20(fix ) & 0.3@xmath40.3 & 0.55@xmath64 & 34.7@xmath82 & 3.3@xmath40.2 & 0.99(27 ) + 40 & 55263.4455265.22 & 8.85 & 2.20(fix ) & 0.7@xmath40.3 & 0.51@xmath40.03 & 40.4@xmath83 & 2.9@xmath40.1 & 0.54(29 ) + 41 & 55265.3555266.94 & 8.52 & 2.20(fix ) & 0.4@xmath40.3 & 0.50@xmath64 & 42.4@xmath84 & 2.8@xmath40.1 & 1.11(25 ) + 42 & 55267.0055268.66 & 8.37 & 2.20(fix ) & 0.6@xmath40.3 & 0.50@xmath64 & 42.4@xmath85 & 2.6@xmath40.1 & 0.94(27 ) + 43 & 55268.7255272.02 & 16.93 & 2.20(fix ) & 0.6@xmath40.2 & 0.49@xmath40.03 & 43.7@xmath86 & 2.3@xmath40.1 & 0.94(36 ) + 44 & 55272.2855276.41 & 12.51 & 2.20(fix ) & 0.5@xmath40.2 & 0.46@xmath40.03 & 51.3@xmath87 & 2.0@xmath40.1 & 0.72(29 ) + 45 & 55277.2355281.63 & 3.98 & 2.20(fix ) & 0.8@xmath40.5 & 0.47@xmath88 & 39.0@xmath89 & 1.4@xmath40.2 & 0.94(14 ) + 46 & 55285.4455286.97 & 2.34 & 2.20(fix ) & 2.1@xmath90 & 0.31@xmath91 & 118.7@xmath92 & 0.5@xmath40.2 & 0.46(9 ) + 47 & 55287.4155288.05 & 2.42 & 2.63@xmath93 & 2.0@xmath40.4 & & & & 0.39(9 ) + 48 & 55288.4355291.16 & 8.98 & 2.19@xmath94 & 2.0@xmath40.2 & & & & 0.58(32 ) + 49 & 55291.3555292.49 & 3.41 & 1.60@xmath95 & 2.7@xmath53 & & & & 0.61(14 ) + the mcd plus power - law model or single power - law model gave good fits for all the maxi / gsc data . table [ tab : parametertab ] summarizes the obtained spectral parameters , while figure [ fig : time_vs_parameters ] plots the evolution of the model parameters . figure [ fig : nufnuspec ] shows typical response - unfolded @xmath96 spectra , together with their best - fit models . the spectra during mjd 5520055214 ( dataid = 110 ) required no mcd component . the photon indices and fluxes was almost constant meanwhile at @xmath971.7 and 1.2@xmath1710@xmath98 erg @xmath23 s@xmath99 , respectively . on mjd 55215.6455216.15 ( dataid = 1112 ) , the spectrum dramatically softened to @xmath1002.0 . the mcd component then appeared after mjd 55216.65 ( dataid = 13 ) and lasted until mjd 55286 ( dataid = 46 ) , although the power - law flux was dominant mjd 55216.6555218.37 and on mjd 55286.97 in the 2 - 20 kev band . the mcd flux reached to a peak of @xmath01@xmath1710@xmath98erg s@xmath99 @xmath23 on 2010 january 23 ( mjd 55219 ) , and then decreased . the innermost temperature @xmath24 gradually decreased from @xmath00.7 kev to @xmath00.4 kev toward mjd 55286.97 . by contrast , the innermost radius @xmath9 was almost constant at @xmath041 km ( for @xmath28 kpc and @xmath101 ) then , except for the epoch before the state transition on mjd 55218.37 ( dataid = 1315 ) . based on these results , we identify that the source was likely in the low / hard state before mjd 55214 , the intermediate state over mjd 5521555218 , the high / soft state over mjd 5521955282 , and then came back to the low / hard state after mjd 55292 through the intermediate state between mjd 55285 and 55292 . for later discussions , we calculate the weighted average of @xmath9 in the high / soft state to be 41.0@xmath42.1 km , using the observations on mjd 5523055282 when the contribution of the power - law component was sufficiently small . as @xcite pointed out the initial variation of @xmath9 observed on mjd 5521655218 in the intermediate state may be caused by the ignoring of comptonized photons in the calculation of @xmath9 . in these epochs , the fraction of the power - law component was @xmath3150% of the total x - ray flux in the 2 - 20 kev band , and hence the compton up - scattering of the disk blackbody emission can significantly reduce the normalization of the direct mcd component ( @xcite ) . we find it difficult , however , to obtain a reliable estimate of the innermost radius corrected for this effect from the maxi / gsc data alone , which strongly couples with the photon index of the compton scattered component . we leave detailed investigation of the spectra in the intermediate state for future work , and concentrate on those in the high / soft state in the following analysis . from 2009 october 25 to 2010 july 29 , the swift / xrt ( @xcite ) carried out 67 pointing observations of xte j1752223 in the windowed timing ( wt ) mode . due to the sun angle constraint , the observations were interrupted between 2009 november 4 ( mjd 55139 ) and 2010 february 3 ( mjd 55230 ) . we analyzed data of 17 observations taken in the high / soft state after 2010 february 4 ( mjd 55231 ) . all the xrt spectra and light curves were produced by the web interface ( @xcite ) supplied by the uk swift science data centre at the university of leicester . we used the xrt response matrix file for the wt mode version 12 . all the 17 observations with an exposure time of a few ksec each were analyzed independently . first results of the swift / xrt spectral analysis were already reported by @xcite . they employed the same mcd plus power - law model as we used in the maxi / gsc analyses , but derived a different result that the innermost radius changed significantly with time . the difference is considered to come from the fact that they left the power - law index free and obtained a wide range of @xmath29 between @xmath00 and @xmath03 , even though this quantity is not constrained when the mcd component is strong . hence we reanalyzed the xrt spectra by fixing @xmath29 at 2.2 , the same value as adopted for the maxi / gsc spectra . the @xmath21 was left free because the swift / xrt is sensitive down to @xmath00.3 kev . resultant parameters are summarized in table [ tab : parametertab_xrt ] , which are also plotted on figure [ fig : time_vs_parameters ] . the values of @xmath102 were thus constant , with at most 5% scatter around weighted mean of @xmath102=41.4 @xmath40.3 km . ccccccccc obsid & mjd & exposure & n@xmath103 & power - law & @xmath24 & @xmath9 & disk & @xmath34 /dof + & start & ( sec ) & ( @xmath1710@xmath22 @xmath23 ) & flux@xmath35 & ( kev ) & ( km ) & flux@xmath35 & + 00031532010 & 55231.6 & 1374.3 & 0.56@xmath40.01 & 2.34@xmath40.23 & 0.60@xmath40.01 & 38.4@xmath104 & 5.66@xmath40.20 & 1.68(224 ) + 00031532011 & 55233.7 & 1595.0 & 0.60@xmath40.01 & 2.40@xmath40.19 & 0.59@xmath40.01 & 38.0@xmath41.0 & 5.39@xmath40.14 & 1.29(238 ) + 00031532012 & 55238.9 & 2360.7 & 0.62@xmath40.01 & 1.26@xmath40.11 & 0.58@xmath40.01 & 44.3@xmath105 & 6.39@xmath40.11 & 1.50(244 ) + 00031532014 & 55242.0 & 1421.4 & 0.56@xmath40.01 & 2.41@xmath40.19 & 0.57@xmath40.01 & 42.2@xmath41.1 & 5.21@xmath40.15 & 1.64(234 ) + 00031532015 & 55244.0 & 1123.8 & 0.55@xmath40.01 & 1.07@xmath40.17 & 0.56@xmath40.01 & 40.9@xmath106 & 4.31@xmath107 & 1.22(191 ) + 00031532016 & 55246.2 & 1411.8 & 0.59@xmath40.01 & 2.80@xmath40.18 & 0.57@xmath40.01 & 41.7@xmath41.1 & 4.79@xmath36 & 1.56(241 ) + 00031532018 & 55248.0 & 835.6 & 0.63@xmath40.01 & 1.29@xmath40.16 & 0.55@xmath40.01 & 45.3@xmath41.4 & 5.01@xmath108 & 0.95(202 ) + 00031532019 & 55250.1 & 1144.3 & 0.61@xmath40.01 & 0.47@xmath40.12 & 0.54@xmath40.01 & 41.7@xmath41.4 & 3.96@xmath109 & 1.22(177 ) + 00031532020 & 55251.3 & 3026.5 & 0.61@xmath40.01 & 0.87@xmath40.07 & 0.55@xmath40.01 & 41.9@xmath40.7 & 4.07@xmath40.07 & 1.62(238 ) + 00031532021 & 55252.6 & 1524.0 & 0.57@xmath40.01 & 0.65@xmath40.09 & 0.53@xmath40.01 & 43.4@xmath41.1 & 3.80@xmath40.06 & 1.50(193 ) + 00031640001 & 55257.6 & 890.9 & 0.60@xmath40.01 & 0.69@xmath40.11 & 0.52@xmath40.01 & 42.5@xmath110 & 3.05@xmath40.08 & 1.22(174 ) + 00031640002 & 55260.7 & 909.9 & 0.61@xmath40.01 & 0.25@xmath40.05 & 0.52@xmath40.01 & 36.9@xmath104 & 2.23@xmath88 & 1.34(180 ) + 00031640003 & 55266.8 & 1209.8 & 0.61@xmath40.01 & 0.17@xmath40.06 & 0.50@xmath40.01 & 43.4@xmath111 & 2.38@xmath40.09 & 1.25(161 ) + 00031640004 & 55269.8 & 1184.1 & 0.67@xmath40.01 & 0.32@xmath40.04 & 0.50@xmath40.01 & 39.3@xmath112 & 2.00@xmath40.05 & 1.61(197 ) + 00031640005 & 55271.8 & 1214.6 & 0.66@xmath40.01 & 0.33@xmath40.04 & 0.48@xmath40.01 & 43.3@xmath41.3 & 1.75@xmath40.06 & 1.41(181 ) + 00031640006 & 55276.9 & 977.8 & 0.57@xmath40.01 & 0.34@xmath40.04 & 0.48@xmath40.01 & 41.5@xmath113 & 1.67@xmath40.04 & 1.34(177 ) + 00031640007 & 55279.8 & 128.3 & 0.54@xmath114 & @xmath40.12 & 0.47@xmath40.01 & 35.8@xmath115 & 0.99@xmath116 & 1.00(108 ) + we also observed xte j1752223 in the high / soft state with suzaku ( @xcite ) as a target - of - opportunity ( too ) from 2010 february 24 ( mjd 55251 ) 04:58:00 to february 25 04:27:24 ( obsid : 904005010 ) . suzaku carries 4 x - ray telescopes ( xrt , @xcite ) , each with a focal - plane x - ray ccd imager called xis ( x - ray imaging spectrometer ; @xcite ) covering the 0.212 kev energy band . since xis2 has not been available , the two remaining front - illuminated ( fi ) ccds ( xis0 and xis3 ) were operated for this observation together with the back - illuminated ( bi ) ccd ( xis1 ) . the three xis cameras were operated with burst and window options . the burst time , window size , and editing mode were `` 0.3 sec , 1/4 window , and 2@xmath172 '' for xis0 and xis3 , and `` 0.1 sec , normal window , and 3@xmath173 '' for xis1 , respectively . the hard x - ray detector ( hxd ; @xcite ; @xcite ) , covering the 1070 kev energy band with si pin photo - diodes ( hxd - pin ) , and the 50600 kev range with gso scintillation counters ( hxd - gso ) , was operated in the normal mode . the source was observed at the hxd nominal position . the data reduction and analyses were performed using suzaku ftools in heasoft version 6.10 provided by the nasa / gsfc suzaku gof . all the xis and hxd data were reprocessed by the suzaku pipeline processing version 2.4.12.27 , with the latest calibration database ( caldbver ) hxd20090902 , xis20090925 , and xrt20080709 . the image degradation due to `` thermal wobbling '' of the satellite pointing was corrected by using the attitude file updated with aeattcor2 . we screened the xis data under the standard criteria , and discarded events during time intervals when the telemetry of the xis was saturated . resultant net exposure was 10.6 ksec and 449 sec for fis and bi , respectively . the bi exposure was short , because 0.1 sec burst time with normal window make the exposure 0.1 sec in 8 sec period and the exposure is one twelfth of each fi s exposure . furthermore , the fraction of out - of - time events ( @xcite ) is more significant in the bi data , which would introduce systematic uncertainties in the spectral analysis . hence , we did not use the xis1 data in this paper . the average 0.510 kev count rate was @xmath0400 cts s@xmath99 and @xmath0700 cts s@xmath99 with fis and bi , respectively . we extracted the xis0 and 3 events from a rectangle region of @xmath17 tracing the 1/4 window area centered on the image peak , and the background from another region avoiding contaminating point sources . the source was so bright that the xis suffered from photon pileup at the image center . using the software of aepileupcheckup ( @xcite in preparation ) , which automatically detects the extent of pileup utilizing radial surface brightness profiles and other sets of information , we excluded a circular within of the image peak for xis0 and 3 , to extract the events whose pileup fraction is below 1% . we combined the spectra and response files of the two fi cameras to increase the photon statistics in the spectra . to account for possible uncertainties in the instrument calibration ( e.g. @xcite ) , systematic errors of 1% was assigned to each energy bin of the xis spectra . we obtained the cleaned hxd events with the standard criteria . after the dead - time correction , the net exposures of pin and gso were 21.2 and 27.0 ksec , respectively . the dead - time fraction was 18.2 and 7.5 % for pin and gso , respectively . the cosmic x - ray background ( cxb ) was modeled assuming an exponentially cutoff power - law model ( @xcite ) . the non - x - ray background ( nxb ) model is provided by the hxd team ( @xcite ) . we used the model with method=``lcfitdt(bgd_d ) '' , and the version of methodv=``2.0ver0804 '' and `` 2.4ver0912 - 64 '' for the data of pin and gso , respectively . after subtracting the modeled nxb and cxb , the source signals were significantly detected by pin up to 50 kev , above the systematic uncertainty on the nxb model ( @xmath03% ) . since the gso signals were comparable to the systematic errors of the nxb model there in ( @xmath01% ) , we consider the source detection with gso insignificant . as shown in figure [ suzaku_specfit](a ) , we used an energy range of the 1.210 kev for the xis spectrum and 1550 kev for the hxd - pin spectrum , where the calibrations of the energy responses were well established . the energy bands of 1.62.0 kev and 2.22.4 kev were excluded to avoid large systematic uncertainties of the effective area near the silicon k edge ( 1.74 kev and 1.84 for k@xmath117 and k@xmath118 , respectively ) and gold m edge ( 2.29 kev ) , respectively . in the combined fit to the xis and hxd - pin spectra , we employed a cross normalization factor of 1.18 between xis and pin ( suzaku - memo-2008 - 06 ) . we firstly fitted the spectra with the mcd plus power - law model modified by interstellar absorption ( _ wabs _ model ) , the same one as adopted to fit the maxi / gsc and swift / xrt spectra in the previous sections . the residuals in units of @xmath119 are shown in the figure [ suzaku_specfit](b ) . the fit was found to be unacceptable ( @xmath120 ) , leaving an emission - line feature at @xmath976.5 kev , and significant broad edge - like residuals above @xmath07 kev . these features may be explained away by several different approaches ; e.g. , by invoking a broad fe - k line concept @xcite or applying so - called smeared - edge model ( e.g. , @xcite ; @xcite ) to account for the edge structure . the latter approach with _ wabs*smedge*(diskbb+powerlaw+gaus ) _ model gave an acceptable fit ( @xmath121 ; fig [ suzaku_specfit](c ) ) with a maximum optical depth of @xmath122 for fixed edge energy and width of 7.11 kev and 10 kev , respectively . a narrow emission line was found at @xmath123=6.59@xmath124 kev with a 1@xmath125 line width of 10 ev ( fixed ) and an equivalent width of @xmath126=63@xmath127 ev . these approaches are empirical and may be degenerate , however . as an alternative possibility , we resorted to the idea of @xcite that the disk emission is broader than the simplest mcd model , which could lead to artificial detection of broad iron - k features . assuming that weak comptonization of the mcd emission took place , we replaced the _ diskbb _ model with a _ compps _ model ( @xcite ) , which computes compton scattering by hot corona using exact numerical solution radiative transfer equation . we chose `` slab '' ( geom=1 ) geometry and a mcd spectrum as the seed photons . we left the optical depth of the corona @xmath128 as a free parameter but fixed the electron temperature at @xmath129 kev and covering fraction at @xmath130 to avoid strong coupling between these parameters ; the choice of @xmath131 did not affect our results on the disk parameters over the statistical errors . the presence of a fluorescence iron - k line at @xmath976.6 kev indicates that a reflection continuum must be included in the model as well . hence , we adopted _ pexriv _ model ( @xcite ) for the power - law component , which was considered to be the main source of irradiation onto the disk in the high / soft state . the model thus became _ wabs*(compps+pexriv+gaus ) _ in the xspec terminology . for the _ pexriv _ component , we assumed no high energy cutoff , and fixed the solid angle of the reflector at @xmath132 , inclination at @xmath133 , and ionization parameter at @xmath134 erg cm s@xmath99 , which consistently explain both iron - k line intensity and center energy . the fit was found to be acceptable with @xmath135 ; as shown in the figure [ suzaku_specfit](d ) , the broad edge - like residuals mostly disappeared . the model gave an absorbed 220 kev flux of ( 3.76@xmath40.02)@xmath1710@xmath136 erg s@xmath99@xmath23 . we obtained @xmath137 and @xmath138 in the _ pexriv _ component , a reasonable value in the high / soft state ( e.g. , @xcite ) . more detailed analysis , more advanced disk emission modeling , and relativistic fe - k lines , is beyond the scope of this paper , and will be reported elsewhere . the best - fit parameters of these models are listed in table [ suzaku_bestfit ] . we note that by replacing the _ diskbb _ model with the _ compps _ model , the best - fit value of @xmath9 increased by 13% . this is because a spectrum hardened by comptonization was assumed in the _ compps _ model , an intrinsic temperature @xmath24 became lower , leading to an increase in @xmath9 . while the difference should be regarded as a systematic modeling uncertainty , we confirm that it does not affect our conclusion . the photon index of the hard tail ( @xmath138 ) is smaller than the result obtained from the same suzaku data by @xcite ( @xmath139 ) , who did not include the 15 - 20 kev and 45 - 50 kev bands in fitting the hxd / pin spectrum . ( 75mm,50mm)figure5.eps c c c c c c + & _ diskbb _ / _ compps_@xmath140 & _ powerlaw _ / _ pexriv_@xmath141 & _ gaus_@xmath142 & _ smedge_@xmath143 & + @xmath144 & @xmath24 ( kev ) & @xmath29 & @xmath123 ( kev ) & @xmath145 & @xmath146 / d.o.f . + & @xmath147 ( km ) & norm@xmath148 & @xmath126 ( ev ) & + & @xmath149 & & & + + 0.62@xmath150 & 0.53@xmath151 & 2.29@xmath40.06 & ... & ... & 230.1/136 + ... & 42.9@xmath152 & 0.16@xmath40.03 & ... & ... & ... + ... & ... & ... & ... & ... & ... + + 0.65@xmath150 & 0.526@xmath153 & 2.22@xmath40.05 & 6.59@xmath124 & @xmath122 & 141.80/133 + ... & 44.0@xmath154 & 0.16@xmath40.02 & 63@xmath127 & ... & ... + ... & ... & ... & ... & ... & ... + + 0.67@xmath150 & 0.507@xmath155 & 2.13@xmath151 & 6.60@xmath124 & ... & 131.4/133 + ... & 47.2@xmath156 & 0.05@xmath157 & 98@xmath158 & ... & ... + ... & 0.41@xmath159 & ... & ... & ... & ... + we have analyzed the maxi / gsc , the swift / xrt , and suzaku data of xte j1752223 in the high / soft state from mjd 55218.62 to 55281.63 . the overall continuum spectra were reproduced by the mcd plus power - law model with an interstellar absorption . the innermost temperature @xmath24 decreased from @xmath00.7 kev to @xmath00.4 kev , while the innermost radius @xmath9 remained constant . by assuming @xmath28 kpc and @xmath101 , the values of @xmath9 derived from the maxi / gsc , swift / xrt , and suzaku data are 41.0@xmath42.1 km , 41.4@xmath40.3 km , and 42.9@xmath152 , respectively . the results from the three instruments are mutually consistent with each other within the statistical errors . for the following discussion , we employ 41.0@xmath42.1 km for @xmath9 as determined from the maxi / gsc , although we also discuss the case when the suzaku result with the _ compps _ model ( @xmath9=47.2@xmath156 km ) is adopted to take into account the possible systematic uncertainty . since the value of @xmath102 thus estimated is significantly larger than those found in luminous low - mass x - ray binaries ( 10 km ; @xcite ) , the black hole interpretation of xtej1752223 ( paper i ; @xcite ) is considerably reinforced . furthermore , its constancy allows us to identify it with the isco in the high / soft state . the stable @xmath9 at 41.0 km is supposed to reach the isco in the high / soft state . we note that this @xmath160 is an `` apparent '' innermost radius , and the `` realistic '' innermost radius ( @xmath161 ) should be estimated as @xmath161= @xmath162 where the spectral hardening factor@xmath163 is 1.7 ( @xcite ) and correction factor for the boundary condition @xmath164 is 0.412 ( @xcite ) . the value of @xmath163 has been confirmed in recent work on disk models ( e.g. , @xcite ; @xcite ) . when the central object is assumed to be a non - spinning black hole ( i.e. schwarzschild black hole ) , @xmath161 should equal to @xmath165 . then , the black - hole mass is estimated as @xmath166 using the inclination angle @xmath16749@xmath7 obtained from radio observations ( @xcite ) , the black hole mass and distance are constrained in the shadowed area shown in figure [ fig : distancemass ] . if we assume the distance to be 3.5 kpc , the mass will be 5.27.1 @xmath5 , and if @xmath168=10 kpc , 15.020.4 @xmath5 . to validate our method to estimate the black hole mass by employing the mcd model which is a simplified approximation of a true disk spectrum , we also performed a spectral fit to the suzaku spectra using the _ kerrbb _ model ( @xcite ) . this model calculates the x - ray spectrum of a relativistic accretion disk around a rotating black hole by taking into account the innermost boundary condition and the effects of self - irradiation of the disk . here we fixed the distance at @xmath168=3.5 kpc , spin parameter at @xmath169=0 , and color correction factor at @xmath163=1.7 for consistency with the previous discussion . the fit with the _ wabs*(kerrbb+pexriv ) _ model was not acceptable ( @xmath170 ) , leaving similar residuals seen in figure[suzaku_specfit](b ) . thus , we employed the _ simpl _ model ( @xcite ) with a steep photon index fixed at 7.4 to approximately represent weak comptonization of the disk emission ; the model became _ wabs*(simpl*kerrbb+pexriv+gaus ) _ , which was found to give an acceptable fit ( @xmath171 ) . the black hole mass derived for @xmath3=0@xmath7 and @xmath17249@xmath7 is @xmath173 m@xmath174 and 8.43@xmath175 m@xmath174 , respectively , which differs by @xmath020% from the mass presented in equation ( 1 ) . from past observations of bhcs and neutron stars @xcite pointed out that the state transition from the high / soft state to the low / hard state occur at 1 - 4% ( centered at 2% ) of the eddington luminosity . = 1.5@xmath1710@xmath176 @xmath177erg s@xmath99 for the solar abundances ] judging from the change of the photon index , xtej1752223 went back to the low / hard state around mjd 55292 ( dataid 49 ) . the bolometric flux at the transition to be @xmath178=(6.68@xmath40.83)@xmath1710@xmath136 erg @xmath23 s@xmath99 assuming a cutoff power - law is continuum with @xmath179 and a cutoff energy of 200 kev . using the relation 0.01@xmath180/@xmath1810.04 and @xmath182 , we can further constrain the distance and black hole mass of xte j1752223 as illustrated with the hatched region in figure [ fig : distancemass ] . thus we can obtain @xmath168=2.518 kpc and @xmath13=3.736 @xmath5 . when we instead use the suzaku results , @xmath168=3.020 kpc and @xmath13=5.344 @xmath5 ( based on the _ wabs*(compps+pexriv+gaus ) _ model ) or @xmath168=2.322 kpc and @xmath13=3.155 @xmath5 ( based on the _ wabs*(simpl*kerrbb+pexriv+gaus ) _ ) are derived . since we obtain @xmath1833@xmath5 , our method of estimating @xmath13 assuming a black hole is self consistent . the conclusion that the compact object in xte j1752223 is a black hole is robust against the assumptions that the rotational axis of the accretion disk and jet axis is exactly aligned and that the black hole is non - spinning , since the extreme conditions @xmath101 and @xmath169=0 yield the smallest mass estimate . the values of @xmath13=9.6@xmath40.9 @xmath5 and @xmath168=3.5@xmath40.4 kpc estimated by @xcite are outside the hatched region . if the black hole in xte j1752223 is spinning and/or the disk has a larger inclination than 49@xmath184 , then the discrepancy could be solved . using data obtained by the maxi / gsc , the swift / xrt , and suzaku , we have performed x - ray spectral analysis of the black hole candidate xtej1752223 in the high / soft state . as commonly seen in bhcs , the innermost radius remained constant in this state during the continuous observation with the maxi / gsc . the results using the mcd plus power - law model were consistent between the three observatories . detailed spectral modeling using suzaku data suggests a possibility that the mcd emission may be slightly comptonized , which could explain apparently broad iron - k features . assuming that the innermost radius of the disk corresponds to the isco and the black hole is non - spinning , we estimated the mass of the central object as 5.51@xmath40.28 @xmath5 @xmath185 with corrections for the stress - free inner boundary condition and color hardening factor of 1.7 . furthermore , using the observational results that the inclination angle _ i _ is smaller than 49@xmath7 from radio observations and the `` soft - to - hard '' transition occurs at 14 % eddington luminosity , the distance and black hole mass of xtej1752223 were constrained as 3.155 @xmath5 and 2.322 kpc , respectively , based on an analysis of the suzaku spectra with a relativistic accretion disk model . thus we can conclude that xte j1752223 is likely a black hole . + this research has made use of maxi data provided by riken , jaxa and the maxi team . suzaku data is provided by a collaborative mission between the space agencies of japan ( jaxa ) and the usa ( nasa ) . we also thank the swift team for their observation . this work made use of data supplied by the uk swift science data centre at the university of leicester . this research was partially supported by the ministry of education , culture , sports , science and technology ( mext ) , grant - in - aid no.19047001 , 20041008 , 20540230 , 20244015 , 20540237 , 21340043 , 21740140 , 22740120 . one of the authors ( s. n. ) is grateful to a grant from the hayakawa satio fund awarded by the astronomical society of japan . anders , e. , & ebihara , m. , 1982 , geochimica et cosmochimica acta 46 , 2363 brocksopp , c. , et al . , 2010 , the astronomer s telegram # 2400 boldt , e. , et al . , , 1987 , 146 , 215 burrows , d. n. , et al . 2005 . , 120 , 165 curran , p. a. , et al.,2011 , , 410 , 1 , 541 done , c. , & davis , s. w. , 2008 , , 683 , 389 dotani , t. , et al . , 1997 , , 485 , l87 ebisawa , k. , et al . , 1994 , , 46 , 375 evans , p. a. , et al . , 2009 , mnras , 397 , 1177 fukazawa , y . , et al . , 2009 , , 61 , s17 grove , j. e. , and strickman , m. s. , 1998 , , 502 , l45 kirsch , m. g. , et al . 2005 , , 5898 , 22 koyama , k. , et al . , 2007 , , 59 , s23 kolehmainen , m. , et al . , 2011 , , 416 , 311 kokubun , m. , et al . , 2007 , , 59 , s53 kubota , a. , et al . , 1998 , , 50 , 667 kubota , a. , et al . , 2010 , , 714 , 860 li , l .- x . , zimmerman , e. r. , narayan , r. , & mcclintock , j. e. , 2005 , , 157 , 335 mcclintock , j. e. , remillard , r. e. , at al . , 2009 , , 698 , 1398 maccarone , t j. , 2003 , , 409 , 697 morrison , r. , & mccammon , d. , 1983 , , 270 , 11 magdziarz , p. , & zdziarski , a. a. , 1995 , , 273 , 837 makishima , k. , et al . , 1986 , , 308 , 635 makishima , k , et al . , 2008 , , 60 , 585 markwardt , c. b. , et al . , 2009 , the astronomer s telegram # 2258 . matsuoka , m. , et al . , 2009 , , 61 , 999 mihara , t. , nakajima , m. , sugizaki , m. , et al . , 2011 , , in print , arxiv:1103.4224 miller - jones , j. c. a. , et al . , 2011 , , 415 , 306 mitsuda , k. , et al . , 1984 , , 36 , 741 mitsuda , k. , et al . , 2007 , , 59 s1 morii , m. , sugimori , k. , & kawai , n. , 2011 , physica e : low - dimensional systems and nanostructures , 43 , 692 mu@xmath186oz - darias , t. , et al . , 2010 , , 404 , l94 nakahira , s. , et al . , 2009 , the astronomer s telegram # 2259 nakahira , s. , et al . 2010 , , 62 , l28 negoro , h. , et al . , 2010 , the astronomer s telegram # 2396 poutanen , j. & svensson , r. , 1996 , , 470 , 249 reis , r. c , et al . , 2010 , , 410 , 2497 remillard , r. e. & mcclintock , j. e. , 2006 , , 44 , 49 serlemitsos , p. j. , et al , 2007 , , 59 , s9 shakura , n. i. , & sunyaev , r. a. 1973 , , 24 , 337 shaposhnikov , n . , and titarchuk , l. , , 2009 , 699 , 453 shaposhnikov , n , et al . , 2010 , , 723 , 1817 shidatsu , m. , et al . , 2011a , , in print , arxiv:1105.3586 shidatsu , m. , et al . , 2011b , , submitted in this issue shimura , t. , & takahara , f. , 1995 , , 445 , 780 steiner , j. f. , narayan , r. , mcclintock , j. e. , ebisawa , k. , 2009 , , 121 , 1279 sugizaki , m. , mihara , t. , et al . , 2011 , , in print , arxiv:1102.0891 takahashi , h. , et al . , 2008 , , 60 , s69 takahashi , t. , et al . , 2007 , , 59 , s35 tanaka , y. , & shibazaki , n. , 1996 , , 34 , 607 tomida , h. , et al . , 2011 , , in print . toor , a. , & seward , f.d . , 1974 , , 79 , 995 tsunemi , h. , et al . , 2010 , , 62 , 1371 wilson - hodge , c. a. , et al . 2011 , , 727 , l40 yamada , s. , et al . 2011 in preparation yang , j. , brocksopp , c et al , 2010 , , l64 zdziarski , a. a. , & gierli@xmath187ski , m. , 2004 , prog . we here show the status of the gsc energy response calibration with crab - nebula data to confirm the validity of the spectral analysis . we used event data with the process version 0.3b , which is screened with the operation high voltage of a nominal 1650 v and the anode number of # 0 , # 3 , # 4 , and # 5 whose positional response had been well established . the event selection is same as that employed in the first performance verification in @xcite . we here screened the event data with more severe conditions to verify the response calibration with a better accuracy . we selected events taken only during such a good scan transit that the source incident angle @xmath20 is @xmath188 and the area for both the source and the background are completely covered . the calibration of the energy response was performed using the screened background and crab nebula data in the following steps ; the energy - pha relations , which depend on the detector position along the anode wires , are corrected by a gain factor for each output amplifier . the parameter was calibrated using fluorescent lines in the background spectrum from ti ( 4.51 kev ) and cu ( 8.04 kev ) as well as the calibration source @xmath189fe ( 5.895 kev ) . we verified that the gain factor obtained with a 0.1 % precision successfully reproduce gain position dependence within the discrepancy of 1% in rms throughout the whole detector area . the threshold and the resolution of the lower - discriminator ( ld ) of each amplifier were then calibrated using the crab nebula spectra . the ld parameters were tuned so that the results of the spectral fit to an absorbed power - law model agrees with those of the canonical values , photon index @xmath190 and hydrogen absorption column density @xmath191 @xmath23 ( i.e. @xcite , @xcite ) . figure [ fig : crabana ] shows the best - fit parameters to gsc crab nebula spectra . the data of multiple scan transits whose total exposure of @xmath313.2 cm@xmath18 ksec were grouped so that each spectrum has enough photon statistics to constrain parameters of a power - law model . table [ table : crabresult ] summarizes the results . the derived parameters agree with the canonical values within the statistical errors . lccc parameter & canonical & best fit & rms@xmath192 + @xmath21 ( @xmath193 @xmath23 ) & 0.35 & 0.39 & 0.25 + @xmath29 & 2.10 & 2.11 & 0.05 + @xmath194 & & 2.11 & 0.03 + flux@xmath195 @xmath196 & 3.0@xmath410% & 3.08 & 0.06 +

the possible occurence of highly deformed configurations in theca di - nuclear system formed in thesi +c reaction is investigated by analyzing the spectra of emitted light charged particles . both inclusive and exclusive measurements of the heavy fragments ( a 10 ) and their associated light charged particles ( protons and particles ) have been made at the ires strasbourg vivitron tandem facility at bombarding energies of(si ) = 112 mev and 180 mev by using the icare charged particle multidetector array . the energy spectra , velocity distributions , in - plane and out - of - plane angular correlations of light charged particles are compared to statistical - model calculations using a consistent set of parameters with spin - dependent level densities . this spin dependence approach suggests the onset of large nuclear deformation inca at high spin . this conclusion might be connected with the recent observation of superdeformed bands in theca nucleus . the analysis of particles in coincidence withs fragments suggests a surprisingly strongbe cluster emission of a binary nature . pacs number(s ) : 25.70.gh , 25.70.jj , 25.70.mn , 24.60.dr

nucl-ex0201021

the formation and binary decay of light nuclear systems in the a@xmath8 @xmath9 60 mass region produced by low - energy ( e@xmath10 @xmath9 7 mev / nucleon ) heavy - ion reactions has been extensively studied both from the experimental and the theoretical points of view @xcite . in most of the reactions studied the binary breakup of the compound nucleus ( cn ) is seen as either a fusion - fission ( ff ) @xcite or a deep - inelastic ( di ) orbiting @xcite process . the large - angles orbiting yields are found to be particularly strong in the @xmath1si + @xmath2c reaction @xcite , as illustrated by fig . 1 which summarizes some of the experimental results that have been collected for this system ; i.e. , orbiting cross sections @xcite and total evaporation residue ( er ) cross sections @xcite . since many of the conjectured features for orbiting yields are similar to those expected for the ff mechanism , it is difficult to fully discount ff as a possible explanation for the large energy - damped @xmath1si + @xmath2c yields @xcite . however , ff calculations @xcite significantly underpredict the cross sections measured in the carbon channel by almost a factor of 3 , thus suggesting an alternative mechanism ( see fig . 1 ) . ff , di orbiting , and even molecular - resonance behavior may all be active @xcite in the large - angle yields of the @xmath1si + @xmath2c reaction @xcite . the back - angle elastic scatterings of @xmath1si ions from @xmath2c displays structured excitation functions and oscillatory angular distributions in agreement with the relatively weak absorption of this system @xcite . moreover , the resonant gross structure @xcite is fragmented into very striking intermediate width resonant structure @xcite . superdeformed ( sd ) rotational bands have been found in various mass regions ( a = 60 , 80 , 130 , 150 and 190 ) and , very recently , sd bands have also been discovered in the n = z nuclei @xmath11ar @xcite and @xmath0ca @xcite . these new results make the a@xmath8 @xmath12 40 mass region of particular interest since quasimolecular resonances have also been observed in both the @xmath11ar and @xmath0ca dinuclear systems @xcite . although there is no experimental evidence to link the sd bands with the higher lying rotational bands formed by known quasimolecular resonances , both phenomena are believed to originate from highly deformed configurations of these systems . since the detection of light charged particles ( lcp ) is relatively simple , the analysis of their spectral shapes is another good tool for exploring nuclear deformation and other properties of hot rotating nuclei at high angular momenta . experimental evidence for angular - momentum - dependent spectral shapes has already been extensively discussed in the literature @xcite and , in particular , the @xmath13 mg + @xmath14o reaction @xcite , which reaches the @xmath0ca cn , has been studied in detail . strong deformation effects were also deduced from angular correlation data for the fusion reaction @xmath1si(@xmath2c,2@xmath4)@xmath6s@xmath15 at e@xmath10 = 70 mev @xcite . we decided to investigate the @xmath0ca nucleus produced through the @xmath1si + @xmath2c reaction at beam energies of e@xmath10 = 112 mev and 180 mev . as can be observed in fig . 1 , at the lowest incident energy of the present work the orbiting process is dominant whereas at e@xmath10 = 180 mev a large part of the o and n fully - damped yields may also result from a ff mechanism . it is interesting to note that the lower energy , e@xmath16 = 33.6 mev , corresponds to a well known quasimolecular resonance ( see fig . 6 of ref . @xcite ) . in this article we will focus on the lcp s found in coincidence with heavy fragments . data have also been collected for the @xmath1si + @xmath1si reaction ( leading to the n = z nucleus @xmath17ni ) in the same experimental conditions at two bombarding energies @xcite . the present paper is organized in the following way . ii describes the experimental procedures and the data analysis . iii presents the inclusive and the exclusive @xmath1si + @xmath2c data ( part of the experimental results presented here in detail have already been briefly reported elsewhere @xcite ) . the data are analyzed using the hauser - feshbach evaporation code cacarizo @xcite using a consistent set of parameters which has been found to succesfully reproduce @xmath13 mg + @xmath14o reaction results @xcite . the full statistical - model calculations , using monte carlo techniques to account for the experimental acceptance when comparing to the experimental exclusive data , are presented in sec . the unexpected strong cluster emission of @xmath7be as due to a binary decay process is also discussed in this section . we end with a conclusion in sec . v. the experiments were performed at the vivitron tandem facility of the ires strasbourg laboratory using 112 mev and 180 mev @xmath1si beams which were incident on @xmath2c targets ( 160 and 180 @xmath18g/@xmath19 thick , respectively ) mounted in the icare scattering chamber @xcite . the effective thicknesses of the @xmath2c targets were accurately determined using rutherford back scattering ( rbs ) techniques with @xmath20h and @xmath21he beams provided by the strasbourg 4 mv van de graaff accelerator @xcite . the carbon buildup corrections were found to be less than 2@xmath22 of the total of c atoms in the targets . both the heavy fragments ( a @xmath3 10 ) and their associated lcp s ( protons and @xmath4 particles ) were detected in coincidence using the icare charged - particle multidetector array @xcite which consists of nearly 40 telescopes . inclusive data have also been collected for heavy fragments and lcp s and presented in sec . iii.a . for the measurement at e@xmath10(@xmath1si ) = 112 mev , the heavy fragments consisting of er as well as quasi - elastic , deep - inelastic , and fusion - fision fragments , were detected in 8 gas - silicon hybrid telescopes ( ic ) , each composed of a 4.8 cm thick ionization chamber , with a thin mylar entrance window , followed by a 500 @xmath18 m thick si(sb ) detector . the ic s were located at @xmath23 = @xmath2415@xmath25 , -20@xmath25 , @xmath2425@xmath25 , -30@xmath25 , -35@xmath25 , and -40@xmath25 in two distinct reaction planes ( for each plane , the positive and negative angles are defined in a consistent manner as for the lcp detectors described below ) . the in - plane detection of coincident lcp s was done using 4 three - element telescopes ( tl3 ) ( 40 @xmath18 m si , 300 @xmath18 m si , and 2 cm csi(tl ) ) placed at forward angles ( @xmath23 = + 15@xmath25 , + 25 @xmath25 , + 35@xmath25 , and + 45@xmath25 ) , 16 two - element telescopes ( tl2 ) ( 40 @xmath18 m si , 2 cm csi(tl ) ) placed at forward and backward angles ( + 40@xmath25 @xmath9 @xmath23 @xmath9 + 115@xmath25 ) and , finally , two other ic telescopes located at the most backward angles @xmath23 = + 130@xmath25 and + 150@xmath25 . the csi(tl ) scintillators were coupled to photodiode readouts . the ic s were filled with isobutane at a pressure of 30 torr for the backward angle telescopes and of 60 torr for the forward angle detectors , thus allowing for the simultaneous measurement of both light and heavy fragments . for the measurement at e@xmath10(@xmath1si ) = 180 mev , three distinct reaction planes were defined . two for in - plane correlations and a third one , perpendicular to the lcp detection plane , for out - of - plane correlation measurements . the heavy fragments were detected in 10 ic s located at @xmath23 = @xmath2410@xmath25 , @xmath2415@xmath25 , @xmath2420@xmath25 , and @xmath2425@xmath25 ( @xmath26 = 0@xmath25 ) for the in - plane coincidences , and at @xmath23 = + 10@xmath25 and + 20@xmath25 ( @xmath26 = 90@xmath25 ) for the out - of - plane coincidences . both the in - plane and out - of - plane coincident lcp s were detected using 3 tl3 s placed at forward angles ( 2 at @xmath23 = + 30@xmath25 and one at @xmath23 = + 35@xmath25 ) and 24 tl2 s placed at forward and backward angles ( + 40@xmath25 @xmath9 @xmath23 @xmath9 + 95@xmath25 ) . the ic s were filled with isobutane at a pressure of 60 torr . the acceptance of each telescope was defined by thick aluminium collimators . the distances of these telescopes from the target ranged from 10.0 to 30.0 cm , and the solid angles varied from 1.0 msr at the most forward angles to 5.0 msr at the backward angles , according to the expected counting rates . the energy calibrations of the different telescopes of the icare multidetector array were done using radioactive @xmath27th and @xmath28am @xmath4-particle sources in the 5 - 9 mev energy range , a precision pulser , and elastic scatterings of 112 mev and 180 mev @xmath1si from @xmath29au , @xmath1si , and @xmath2c targets in a standard manner . in addition , the @xmath2c(@xmath14o,@xmath4)@xmath13mg@xmath30 reaction at e@xmath10 = 53 mev @xcite was used to provide known energies of @xmath4 particles feeding the @xmath13 mg excited states , thus allowing for calibration of the backward angle detectors . the proton calibration was achieved using scattered protons from formvar targets bombarded in reverse kinematics reactions with both @xmath1si and @xmath14o beams . on an event - by - event basis , corrections were applied for energy loss of heavy fragments ( a @xmath3 10 ) in the targets and in the entrance window mylar foils of the ic s and thin al - mylar foils of the si diodes , and for the pulse height defect in the si detectors . the ic energy thresholds and energy resolution for heavy fragments are better than 1.5 mev / nucleon and 0.7@xmath22 , respectively , as shown in fig . 2 for the c and o exit fragments for the @xmath1si(112 mev ) + @xmath2c reaction at @xmath23 = 15@xmath31 . the total energy resolution of 8.78 mev @xmath4 particles from thorium sources has been found to be better than 2.2@xmath22 for both the three - element and two - element light - ion csi(tl ) telescopes . absolute cross sections of inclusive measurements could be obtained within 10 - 12@xmath22 error bars as due to 3 - 5@xmath22 uncertainties in the target thickness and to 8 - 10@xmath22 uncertainties in the electronic deadtime corrections . more details on the experimental setup of icare and on the analysis procedures can be found in refs . @xcite and references therein . the fragments with z = 15 - 17 have typical inclusive energy spectra of er s which are not displayed here , but their lcp exclusive data are discussed in sec . we rather focuss on inclusive data of the binary fragments with z = 5 - 14 . the c and o fragment energy spectra measured at @xmath23 = 15@xmath31 for the @xmath1si + @xmath2c reaction at e@xmath10 = 112 mev are displayed in fig . as expected they are very similar to the spectra measured by shapira _ et al . _ @xcite for the same reaction at e@xmath10 = 115 mev . the peak identifications are given for the ground states as well as for the single and mutual excited states for the inelastic and @xmath4-transfer exit - channels , respectively . the measured yields for different binary - channel products ( z = 5 to 11 ) were converted to center - of - mass cross sections by integrating their energy spectra . the resulting cross sections are shown in fig . 3 and fig . 4 for the two bombarding energies e@xmath10 = 112 mev and 180 mev , respectively . 3 the present data ( open symbols ) are in fairly good agreement with the previously measured c , n , and o angular distributions of shapira _ et al . _ @xcite ( full symbols ) obtained at e@xmath10 = 115 mev . all angular distributions of figs . 3 and 4 vary as 1/sin@xmath32 , providing strong evidence that the damped reaction component ( of a fusion - fission or di orbiting origin ) has a long lifetime , thus for these light fragment ( z = 5 to 11 ) a fast di mechanism seems unlikely to contribute significantly . on the other hand such a fast di mechanism does appear to be significant for the heavier fragments ( z = 12 to 14 ) . by integrating their 1/sin@xmath32 angular distributions , fully - damped cross sections for the binary - channel products can be obtained . the results of the experimental cross sections are given in table [ table1 ] and compared with the previously measured excitation functions @xcite in fig . the agreement between both sets of data is good . however , at the highest bombarding energy e@xmath10 = 180 mev ( e@xmath16 = 54 mev ) a rather small systematic disagreement can be observed . this 10@xmath22 discrepancy is attributed to uncertainties in the target thickness and electronic deadtime corrections . typical inclusive energy spectra of @xmath4 particles are shown in fig . 5 at the indicated angles for the @xmath1si(180 mev ) + @xmath2c reaction . the solid points ( with error bars visible when greater than the size of the points ) are the experimental data whereas the solid and dashed lines are statistical model calculations discussed in sec . the @xmath4-particle energy spectra have a maxwellian shape with an exponential fall - off at high energy which reflects a relatively high effective temperature ( t@xmath33 @xmath12 [ 8e@xmath34/a@xmath35@xmath36 = 3.67 mev ) of the decaying nucleus . the spectral shape and high - energy slopes are also found to be essentially independent of angle in the c.m . system . these observations suggest a statistical deexcitation process arising from a thermalized source such as the @xmath0ca cn . similar @xmath4-particle spectra have been measured for the lowest bombarding energy . these are also consistent with a previous study at e@xmath10 = 150 mev @xcite . the velocity spectra of protons and @xmath4 particles , presented next , are strongly supportive of a statistical deexcitation process . the velocity contour maps of the lcp galilean - invariant differential cross sections ( d@xmath37/d@xmath38de)p@xmath39c@xmath39 as a function of the lcp velocity are known to provide an overall picture of the reaction pattern . from this pattern the velocity of the emission source can be determined in order to better characterize the nature of the reaction mechanism . 6 shows such typical plots of invariant cross - section in the ( v@xmath40,v@xmath41 ) plane for @xmath4 particles ( left side ) and protons ( right side ) , respectively , measured in singles mode . the symbols v@xmath40 and v@xmath41 denote laboratory velocity components parallel and perpendicular to the beam , respectively . for the sake of clarity the velocity cutoffs arising from the detector low - energy thresholds are indicated for each of the telescopes . the present data at e@xmath10 = 112 and 180 mev have the same trends as the plots measured at e@xmath10 = 150 mev ( see fig . 1 of ref . @xcite ) . the dashed circular arcs , centered on the center - of - mass v@xmath16 and defined to visualize the maxima of particle velocity spectra , describe the data trends rather well . they have radii very close to the coulomb velocities of @xmath4 particles and protons in the decay of @xmath0ca@xmath30 @xmath42 @xmath11ar + @xmath21he , and of @xmath0ca@xmath30 @xmath42 @xmath43k + @xmath20h , respectively . the apparent worsening of the agreement between the experimental and calculated @xmath4-particle spectra ( dashed circular arcs ) at larger angle results from the relatively large low - energy thresholds of the most backward - angle telescopes . despite these artifacts , the spectra can be understood by assuming a sequential evaporative process and successive emission sources starting with the thermally equilibrated @xmath0ca@xmath30 cn and ending with the final source characterised by a complete freeze - out of the residual nucleus . it is clear from this figure that the galilean - invariant cross section contours fall on the dashed circular arcs centered at v@xmath16 , as expected for a complete fusion - evaporation ( cf ) mechanism followed by isotropic evaporation . the exclusive lcp events are consistent with a cf mechanism ( as shown in the previous analysis of the inclusive lcp data ) as they are mainly in coincidence with er s with z = 15 - 17 . figs . 7 and 8 display the exclusive energy spectra of @xmath4 particles detected at the indicated angles ( from @xmath44 = + 40@xmath31 to @xmath44 = + 65@xmath31 ( + 95@xmath31 ) ) , in coincidence with individual er s ( z = 15 and 16 ) identified in the ic detector located at @xmath45 = -15@xmath31 ( and -10@xmath25 ) in the @xmath1si + @xmath2c reaction at e@xmath10 = 112 mev and 180 mev , respectively . the data taken with the ic s located at more backward angles ( larger than -15@xmath31 ) are not considered in the following analysis since the statistics for fusion - evaporation events are too low . the experimental data are given by the solid points , with error bars visible when greater than the size of the points . the spectral shapes of @xmath4 particles in coincidence with p ( z=15 ) er s , shown in fig . 8 , are very similar to the inclusive energy spectra of fig . on the other hand the energy spectra of @xmath4 particles in coincidence with s ( z=16 ) er s are more complicated as they show other sub - structures which are superimposed on the statistical " maxwellian shape . in fig . 7 for z = 15 the high - energy components showing up at the large angles may arise from @xmath4 , 3p evaporation cascades . 9 presents the corresponding @xmath1si + @xmath2c exclusive energy spectra of protons emitted in coincidence with individual er s ( z = 15 and 16 ) at e@xmath10 = 180 mev . their spectral shapes are also maxwellian with the typical exponential fall - off at high energy , characteristic of a statistical cn decay process . the in - plane angular correlations of @xmath4 particles and protons ( measured with -115@xmath31 @xmath9 @xmath44 @xmath9 + 115@xmath31 ) in coincidence with er s ( 14 @xmath9 z @xmath9 17 ) , produced in the @xmath1si + @xmath2c reaction , are shown in figs . 10 and 11 at e@xmath10 = 112 mev and 180 mev , respectively . the angular correlations are peaked strongly on the opposite side of the beam direction from the er detectors which were located at @xmath45 = -15@xmath31 or @xmath45 = -10@xmath31 for the two energies , respectively . the observed peaking of the lcp yields on the opposite side of the beam from the ic is the result of momentum conservation . the angular correlations of both the protons and @xmath4 particles show the same behavior for all er s . the solid lines shown in the figures are the results of statistical - model predictions described in the next section . 12 displays the out - of - plane angular correlations of @xmath4 particles ( circles ) and protons ( triagles ) in coincidence with individual er s detected at @xmath23 = 10@xmath25 , produced in the @xmath1si(180 mev ) + @xmath2c reaction . the angular distributions have a behaviour following a typical exp(-a@xmath46(@xmath47 ) ) shape @xcite , with possibly two components visible for @xmath4 particles in coincidence with z = 14 , 15 and 16 , plus a broadening of the angular correlations at backward angles . this broadening effect may result from @xmath4 particles being able to be emitted at the beginning or at the end of the decay chain , where the angular momentum becomes smaller towards the end of the chain . as the protons can not remove as much angular momentum as do the @xmath4 particles the broadening effect is less significant in the proton angular correlation . the solid lines shown in the figure are the results of statistical - model predictions for cf and equilibrium decay using the monte carlo evaporation code cacarizo @xcite , as discussed in the next section . the evaporation of light particles from a highly excited cn is a well known decay process up to very high excitation energies and spins @xcite . the interpretation of lcp data requires a careful treatment of the light particle emission properties in the statistical - model description . most of the available statistical - model computer codes @xcite are based on the hauser - feshbach formalism and are able to follow the cn decay by a cascade of evaporated lcp s and neutrons . in particular , a detailed analysis of the exclusive data can be undertaken by the use of monte carlo versions @xcite of some of these statistical - model codes in which the filtering of the events can reproduce the experimental conditions . the statistical - model analysis of the present data has been performed using the hauser - feshbach evaporation code cacarizo @xcite . cacarizo is a monte carlo version of the statistical - model code cascade @xcite , which has evolved with many modifications and extensions @xcite from the original code . in this program the effective experimental geometry of the icare detectors is properly taken into account . it is assumed that a single cn is created with a well defined excitation energy and angular momentum distribution , and the de - excitation chain is followed step by step and recorded as an event file . the generated events are then analyzed using a subsequent filtering code analysis @xcite in which the locations and the solid angles of all the icare telescopes are explicitly specified . this program allows the determination of the different types of events of interest . such events can be sorted ( singles events , coincidence events , etc . ) and the corresponding particle spectra and angular distributions can be created . the cn angular momentum distributions needed as the primary input for the calculations were specified using the er critical angular momentum l@xmath48 and the diffuseness parameter @xmath49l . they were taken from er cross section data compiled for the @xmath1si + @xmath2c fusion process by vineyard _ et al . _ @xcite , without including fission competition . a fixed value of @xmath49l = 1@xmath50 ( optimized at low energy by a previous statistical - model analysis of this reaction @xcite ) was assumed for the calculations . it has been checked that the calculated spectra are not sensitive to slight changes in the critical angular momentum or to explicit inclusion of the fission competition . the parameter sets used for the calculations are summarised in table ii . the other standard ingredients for statistical - model calculations are the formulations of the nuclear level densities and of the barrier transmission probabilities . the transmission coefficients were derived from optical model ( om ) calculations using potential parameters of light particle induced reactions deduced by wilmore and hodgson @xcite , perey and perey @xcite , and huizenga @xcite for the neutrons , protons and @xmath4 particles , respectively . for spin regions where the standard rotating liquid drop model ( rldm ) @xcite as well as the finite - range liquid drop model ( frldm ) @xcite still predict essentially spherical shapes , these sets of transmission coefficients have been found adequate in the considered mass region . however , in recent years it has been observed that when the angular momentum is increased to values for which frldm predicts significant deformations , statistical - model calculations using such standard parameters can not always predict satisfactorily the shape of the evaporated @xmath4-particle energy spectra @xcite . the calculated average energies of the @xmath4 particles are found to be much higher than the corresponding experimental results . several attempts have been made to explain this anomaly either by changing the emission barrier or by using spin - dependent level densities . adjusting the emission barriers and corresponding transmission probabilities affects the lower - energy part of the calculated evaporation spectra . on the other hand the high - energy part of the spectra depends crucially on the available phase space obtained from the level densities at high spin . in hot rotating nuclei formed in heavy - ion reactions , the energy level density at higher angular momentum is spin dependent . the level density , @xmath51 , for a given angular momentum @xmath52 and energy @xmath53 is given by the well known fermi gas expression with equidistant single - particle levels and a constant level density parameter @xmath54 : @xmath55^{1/2 } ) \label{lev}\ ] ] where t is the nuclear " temperature and @xmath49 is the pairing correction @xcite . the quantity e@xmath56 = @xmath57j(j+1 ) is the rotational energy , with @xmath58 being the effective moment of inertia , where @xmath59 at high excitation energy and high angular momentum is considered to be the rigid body moment of inertia and @xmath60 and @xmath61 are the deformability parameters " @xcite . the level density parameter is constant and is set equal to @xmath54 = a/8 mev@xmath39 , a value which is in agreement with previous works @xcite . in principle , the value of @xmath54 may be affected by dynamical deformation : rotation induces rearrangement of the single - particle level scheme and the altered nuclear surface area @xcite affects the macroscopic energy of the system . the @xmath54 parameter becomes more important when the nuclear deformation increases @xcite . however , in the present work we assume a constant value and rather introduce deformation effects through the deformability parameters . a constant value of @xmath54 = a/8 is in agreement with various authors @xcite , as well as with theoretical studies by shlomo and natowitz @xcite , by tke and swiatecki @xcite , and with experimental results obtained very recently in the a@xmath8 = 60 mass region @xcite . rather than to adjust the spin - dependence of the moment of inertia as done here , with effective emission barriers being unchanged , an alternative method to mock up nuclear deformation is to vary the radius paramater r@xmath62 ( from rldm @xcite parameters proposed by myers and swiatecki @xcite ) of the rigid - body moment of inertia @xmath59 , as discussed recently for very light nuclear systems such as @xmath63p @xcite . this will affect the transmission coefficients by lowering the effective emission barrier @xcite . however , a crude increase of the radius parameter ( up to 25@xmath22 ) in the om transmission coefficients has been shown @xcite to enhance the intensity of the high - energy @xmath4 particles in the spectra . although for hot heavy nuclei at high excitation energy a lowering of the @xmath4-particle emission barriers has to be taken into account @xcite , this artifice has been extensively criticized in the past for lighter systems @xcite . in fact this would result in a poorer reproduction of the present @xmath1si + @xmath2c data . therefore no attempt was made to modify the transmission coefficients since it has been shown that the effective barrier heights are fairly insensitive to the nuclear deformation @xcite . on the other hand , by changing the deformability parameters @xmath60 and @xmath61 one can simulate the spin - dependent level density @xcite associated with a larger nuclear deformation , and thus reproduce the experimental data in a much better way . in the present analysis , following the procedure proposed by huizenga _ et al . _ @xcite , we empirically modify the phase space open to statistical decay by lowering the yrast line with adjustment of the deformability parameters so as to fit the available experimental data @xcite . we take into account the fact that the deformation should be attenuated during the subsequent emission processes : i.e. , there is a readjustment of shape of the nascent final nucleus and a change of collective to intrinsic excitation during the particle - evaporation process . a similar analysis was suggested earlier by blann and komoto @xcite , but with the assumption that the deformation is a frozen degree of freedom through the decay chain . dynamical effects related to the shape relaxation during the de - excitation process have been incorporated into statistical - model codes @xcite . for the cacarizo calculations done here , it is assumed that memory of formation details are lost after each step , with only the conserved quantities such as total energy and spin preserved during the decay sequence . the cacarizo calculations have been performed using two sets of input parameters : the first one with standard liquid drop parameters ( parameter * set a * ) , consistent with the deformation of rldm @xcite and of frldm with finite - range corrections of sierk @xcite , and the second one with larger values for the deformability parameters @xcite ( parameter * set b * ) which are listed in table [ table3 ] . the dashed lines in fig . 5 show the predictions of cacarizo for @xmath1si + @xmath2c at e@xmath10 = 180 mev using the parameter * set a * consistent with frldm deformation @xcite . it is clear that the average energies of the measured @xmath4-particle inclusive spectra are lower than those predicted by these statistical - model calculations . the solid lines of fig . 5 show the predictions of cacarizo using the increased values of the spin deformation parameters ( see parameter * set b * given in table ii ) , and the agreement is considerably improved . the exclusive energy spectra of the @xmath4 particles in coincidence with individual er s ( z = 15 and z = 16 ) are displayed in figs . 7 and 8 for the two bombarding energies e@xmath10 = 112 mev and 180 mev , respectively . it can be observed that the spectra in coincidence with the p residues are well reproduced by using the deformation effects @xcite . the solid lines in figs . 7 and 8 show the predictions of cacarizo using the parameter * set b * with @xmath60 = 2.5 x 10@xmath64 and @xmath61 = 5.0 x 10@xmath65 chosen to reproduce the data consistently at the two bombarding energies . on the other hand , by using the standard liquid drop deformability parameter * set a * with no extra deformation ( i.e. with small values of @xmath66 and @xmath61 ) , the calculated average energies from the exclusive @xmath4-particles spectra ( not shown in figs . 7 and 8) are , as found for the inclusive data , lower than those predicted @xcite by the statistical model . in this case the cacarizo parameters are similar to the standard parameters used in a previous study of the 130 mev @xmath14o + @xmath13 mg reaction @xcite , with the use of the angular momentum dependent level densities . the exclusive energy spectra of @xmath4 particles measured in coincidence with individual s er s , which are shown in figs . 7 and 8 for e@xmath10 = 112 mev and 180 mev , respectively , are quite interesting . a large difference can be noticed when comparing the energy spectra associated with s residues and those associated with p er s @xcite . the latter are reasonably well reproduced by the cacarizo calculations , whereas the model does not predict the shape of the spectra obtained in coincidence with s residues at backward angles ( @xmath67 at e@xmath10 = 112 mev and @xmath68 at e@xmath10 = 180 mev ) . additional non - evaporative components appear to be significant in this case . this is consistent with the discrepancies also observed at backward angles in the in - plane angular correlations of fig . 10 . in the following section we will discuss the hypothesis that the discrepancy occurs because a strong cluster transfer reaction mechanism leads to significant @xmath7be production , with the @xmath7be subsequently breaking up into two @xmath4 particles . as shown in fig . 9 cacarizo calculations are also capable to reproduce the shape of exclusive proton spectra for the @xmath1si(180 mev ) + @xmath2c reaction . compared to the @xmath4 particles , it may be mentioned that the energy spectra of the protons do not shift as significantly as the spin - dependent parametrization of the moment of inertia is introduced . the statistical - model results using the two different parameter sets reproduce equally well the experimental velocity spectra and angular correlations . the statistical - model calculations displayed for protons in fig . 9 have been performed with parameter * set b * ( solid lines ) including the deformation effects ( calculations with parameter * set a * are not displayed ) . in order to better determine the magnitude of the influence of deformation effects in the cn and the residual nuclei which are suggested by our choice of statistical - model approach , we have proposed a very simple procedure @xcite . the effective moment of inertia is expressed as @xmath69 = @xmath70mr@xmath71 = @xmath72m(b@xmath71+a@xmath71 ) with the volume conservation condition : v = @xmath73abc , where b and a are the major and minor axis , and c is the rotational axis of the cn . in the case of an oblate shape a = b and @xmath69 = @xmath70ma@xmath71 and v = @xmath73a@xmath71c . the axis ratio is equal to @xmath74 = a / c = ( 1+@xmath60j@xmath71+@xmath61j@xmath21)@xmath75 . in the case of a prolate shape a = c and @xmath59 = @xmath72m(b@xmath71+a@xmath71 ) and v = @xmath73a@xmath71b . we obtain the equation : 1+(3-@xmath76)x+3x@xmath71+x@xmath77 = 0 with x = @xmath78 = @xmath79 and @xmath76 = 8(1+@xmath60j@xmath71+@xmath61j@xmath21)@xmath77 . the quadrupole deformation parameter @xmath80 is equal to @xmath80 = @xmath81 . + the effects of the yrast line lowering ( increase of the level density ) due to the nuclear deformation and the variation of the deformation parameter @xmath80 can be quantitatively discussed using the values in table [ table3 ] for several reactions . the values of the minor to major axis ratio b / a and of the deformation parameter @xmath80 have been extracted from the fitted deformability parameters by assuming either a symmetric prolate shape or a symmetric oblate shape , respectively , with sharp surfaces @xcite . in our analysis we have chosen to follow the procedure proposed by huizenga _ _ @xcite and successfully used for the @xmath1si + @xmath1si data @xcite . no attempt to modify the parametrization of om transmission coefficients has been undertaken since it has been shown that the effective barrier heights are fairly insensitive to the nuclear deformation @xcite . it is interesting to note that the deformation found necessary to reproduce the @xmath1si + @xmath2c reaction results is smaller than the deformation introduced by the deformability parameter used by kildir _ et al . _ @xcite , who also change the transmission coefficients . the in - plane angular correlations of the @xmath4 particles ( circles ) and the protons ( triangles ) in coincidence with individual er s at both energies e@xmath10 = 112 mev and 180 mev are shown in figs . 10 and 11 , respectively . the solid lines shown in the figures are the results of statistical - model predictions for cf and equilibrium decay using the evaporation code cacarizo . it can be observed in fig . 10 that for e@xmath10 = 112 mev the experimental angular correlations are well reproduced by the evaporation calculations for the data at the opposite side from the er detector , and this is true for correlations with both s and p er s . however the calculations fail to predict the experimental data at the same side as the er detector . the question of these large yields measured at negative angles remains open @xcite . similarly the cacarizo calculations reproduce in fig . 11 the in - plane angular correlations of @xmath4 particles ( circles ) and protons in coincidence with all er s , at e@xmath10 = 180 mev , for the data on the opposite side from the er detector . they are also able to describe the in - plane angular correlations of protons in coincidence with individual z = 14 and z = 15 on both sides of the beam . however , the excess of yields observed at backward angles ( @xmath23 = + 50@xmath25 to + 90@xmath25 ) for @xmath4 particles in coincidence with s may indicate the occurence of a non - evaporative process , possibly of a binary nature . the solid lines shown in fig . 12 for the out - of - plane angular correlations are the results of cacarizo statistical - model predictions . once again it can be observed that the statistical - model calculations are able to reproduce the proton coincidences well , but they fail to describe the @xmath4-cl coincidences and the large yields found in coincidence with p and si er s in the most forward direction . the reason why the experimental anisotropy factor is not well reproduced by the calculations is not well understood . the angular momentum dependence has been tested by performing calculations with two different angular - momentum windows : 10 - 20@xmath50 and 20 - 30@xmath50 . whereas for protons the anisotropy is almost constant with the l - window , for the @xmath4 particles the anisotropy is strongly depending of the chosen l - window . nevertheless the flat behavior shown around 0@xmath25 is present for the two particle species . the problem of this discrepancy in the out - of - plane angular correlations may need to use a more complete formulation for the angular distribution of the lp which is treated in a semi - classical way . the main conclusion that can be drawn from the above analysis indicate that the shape of the @xmath0ca nuclear system is expected to be elongated by rather significant deformation effects during the evaporative processes of @xmath4 particles . this strong deformation is compatible with the previous analysis of @xmath1si(@xmath2c,2@xmath4)@xmath6s@xmath15 angular correlation data @xcite . the extent to which these effects can be reasonably well quantified is dependent on the degree of complexity of the experimental device and , in particular , on the power of the coincidence trigger . it is of particular interest to note that the value of @xmath80 @xmath12 0.5 found for the quadrupole deformation parameter of @xmath1si + @xmath2c ( see table iii ) might be connected with the recent observation of sd bands in the doubly - magic @xmath0ca nucleus by standard @xmath76-ray spectroscopy methods @xcite . correlating large deformations in the hot cn with the presence of sd bands in @xmath0ca is obviously not straightforward , since the deformation deduced from the lcp data is the average , while sd bands are one of the possible configurations . we made the same discussion with the possible comparison between lcp results for the @xmath1si + @xmath1si reaction @xcite and @xmath76-ray data displaying very deformed bands in the doubly magic @xmath17ni nucleus @xcite . it has been shown from the analysis of figs . 7 and 8 with cacarizo that additional non - statistical components appear to be significant at both bomabarding energies e@xmath10 = 112 mev and 180 mev . however no evidence was found for additional processes at the lower bombarding energies e@xmath10 = 70 mev @xcite and 87 mev @xcite . to better understand the origin of these components , @xmath4 particle energies are plotted in fig . 13 against the energies of the s residues detected at @xmath82 = -10@xmath25 for the @xmath1si(180 mev ) + @xmath2c reaction for a number of @xmath4-particle emission angles . with increasing @xmath4-particle angles an increase of the energy of s residues and a decrease of the @xmath4 energy is observed which is consistent with kinematics . at @xmath83 = + 40@xmath25 , + 45@xmath25 , and + 50@xmath25 the bulk of events in fig . 13 are of a statistical origin , and consistent with cacarizo calculations , as demonstrated in fig . 14 ( for @xmath83 = + 40@xmath25 ) . another statistical - model code pace 2 @xcite gives similar predictions . the calculations suggest that these @xmath4 particles result from a cascade of a single @xmath4 , two protons , and x neutrons rather than a 2-@xmath4,xn evaporation process . for larger angles , the two branches , corresponding to the contours labelled 1 and 2 , although lying outside the `` statistical evaporation region '' , still correspond to an evaporation process as shown by the cacarizo calculations displayed in fig . 14 for @xmath83 = + 40@xmath25 and + 70@xmath25 . these two branches 1 and 2 correspond to a 2-@xmath4 fusion - evaporation channel with both the @xmath4 particles emitted respectively at backward and forward angles in the center of mass . however , at more backward angles other contributions , corresponding to the contours labelled 3 and 4 , appear more and more significantly as shown for instance in fig . 14 for + 70@xmath25 . the corresponding `` folding angles '' are compatible with the two - body kinematics required for the @xmath6s + @xmath7be binary exit - channel . in contrast , the energy correlations for the @xmath4 particles in coincidence with cl and p residues ( not shown ) do not exhibit any of these two - body branches 3 and 4 and , thus , the `` statistical evaporation region '' is consistent with the cacarizo predictions , for all the measured angles . although in principle the identification of the @xmath7be cluster requires the coincident detection and mass identification of both decaying @xmath4 particles @xcite , a kinematic reconstruction of this decay process is still possible from the present coincident data to determine the excitation energy of the @xmath7be nucleus by assuming a two - body @xmath6s + @xmath7be@xmath30 process . on the left side of fig . 15 the excitation energy of @xmath7be is presented for the contributions labelled 2 , 3 , and 4 ( fig . 14 ) at the indicated @xmath83 angles . from @xmath83 = 70@xmath25 to @xmath83 = 85@xmath25 the strongest peak appears with a very narrow width . this large component , which corresponds to the contribution of the contour 4 visible in fig . 14 , is centered at the energy of the ground state of @xmath7be ( the relative energy of the two @xmath4 particles of the @xmath7be breaking up in flight is 92 kev ) and displayed as the squared part of the left pannel of fig . 15 . from @xmath83 = 55@xmath25 to @xmath83 = 95 @xmath25 the main bulk of the yields from contour 3 is centered at around e@xmath30 = 3.1 mev with an experimental width of approximately 1.5 mev , which values correspond well to the known energy ( e@xmath30 = 3.04 mev ) and width ( @xmath84 = 1500 kev ) of the first 2@xmath85 excited level of @xmath7be @xcite . the short - lived @xmath7be 4@xmath85 excited level at e@xmath30 = 11.4 mev @xcite is not clearly observed due to its very broad width ( @xmath84 = 3.7 mev ) and the significant @xmath4-statistical background arising from the contribution of the contour 2 . for the same reasons it is hazardous to assign the bumps around 15 mev to the known 2@xmath85 doublet @xcite at e@xmath30 = 16.6 and 16.9 mev . the right pannels of the fig . 15 display the reconstructed excitation energy spectra of the s binary fragments measured at @xmath82 = -10@xmath25 in coincidence with @xmath4 particles detected at the indicated @xmath83 angles by gating either on the groud state ( g.s . ) contour 4 ( upper panel ) or the 2@xmath85 state contour 3 ( lower panel ) . we have performed fusion - fission calculations ( not shown ) , using the extended hauser - feshbach method @xcite . they fail to reproduce both the excitation energies of the @xmath6s fragments , and the yields from the contributions 3 and 4 @xcite . these contributions might result from a faster binary process governed by the @xmath4-transfer reaction mechanism @xmath1si + @xmath2c @xmath42 @xmath6s@xmath30 + @xmath7be , as proposed by morgenstern _ et al . _ this conclusion is in agreement with previous inclusive results published in ref . @xcite . in the cluster - transfer picture @xcite the reaction is characterised by a `` q - value '' window centered at the so - called `` q - optimum '' , which value can be estimated semi - classically by q@xmath86 = ( z@xmath87z@xmath88/z@xmath89z@xmath90 - 1)e@xmath91 , where the indices 1,2 and 3,4 indicate the entrance ( i ) and exit channel , respectively . the corresponding excitation energy e@xmath30 = q@xmath92 - q@xmath86 , where q@xmath92 is the ground - state q - value of the reaction . in this case the expected excitation energy in the @xmath6s nuclei is equal to 12.9 mev . the right hand side of fig . 15 represents the calculated excitation energy of @xmath6s in coincidence with the g.s . , and with the first 2@xmath93 ( e@xmath94 = 3.04 mev ) excited state of @xmath95be , respectively . the dashed lines correspond to e@xmath30 = 12.9 mev , the energy expected for @xmath4-transfer reaction mechanisms . in both cases the excitation energies of @xmath6s are consistent with these values . in the same way we can also have a @xmath7be - transfer reaction mechanism @xcite @xmath1si + @xmath2c @xmath42 @xmath11ar@xmath30 + @xmath4 . in this case the @xmath11ar@xmath30 ejectile has enough excitation energy to emit either one proton or one @xmath4 particle . this type of `` transferlike '' reaction can explain the disagreement observed in fig . 10 between data and cacarizo calculation for the in - plane angular correlation between @xmath4 particles and cl residues . the possible occurence of highly deformed configurations in the @xmath0ca dinuclear system has been investigated by using the icare charged - particle multidetector array at the vivitron tandem facility of the ires strasbourg . the properties of the emitted lcp s in the @xmath1si + @xmath2c reaction , have been analysed at two bombarding energies e@xmath10 = 112 mev and 180 mev , and compared with a statistical model that was adopted to calculate evaporation spectra and angular distributions for deformed nuclei . a monte carlo technique has been employed in the framework of the well documented hauser - feshbach code cacarizo . the measured observables such as velocity distributions , energy spectra , in - plane and out - of - plane angular correlations are all reasonably well described by the monte carlo calculations when they include spin - dependent level densities , and indicate that effects due to differences in nuclear shapes are large enough to be observed . the magnitude of the adjustements in the yrast line suggests significant deformation effects at high spin for the @xmath0ca dinuclear system comparable to recent @xmath76-ray spectroscopy data for the @xmath0ca nucleus at much lower spins @xcite . the extent to which the resonant behaviour @xcite is responsible to the observed nuclear deformation is still an open question . a non - statistical binary component is found in the @xmath4-particle energy spectra measured in coincidence with s residues that is attributed to the cluster decay of unbound @xmath96be nuclei , produced through the @xmath4-cluster - transfer reaction @xmath1si + @xmath2c @xmath42 @xmath6s + @xmath7be @xcite . this new type of `` transferlike '' mechanism , which does not appear at lower bombarding energies @xcite below threshold , allows to populate some n = z nuclei with a well defined excitation energy . therefore , sophisticated particle-@xmath76 experiments ( see refs . @xcite for instance ) using sc euroball iv and/or sc gammasphere should be performed in the very near future in order to well define and understand what are the best types of reaction which can populate significantly the superdeformed bands discovered and/or predicted in this mass region @xcite . this paper is based upon the ph.d . thesis of m. rousseau , universit louis pasteur , strasbourg , 2000 . the authors would like to thank the staff of the vivitron for providing us with good stable beams , m.a . saettel for preparing the targets , and j. devin and c. fuchs for their excellent support in carrying out these experiments . particular appreciation to m.a . saettel for preparing the targets , and to j.p . stockert and a. pape for assistance during their rbs measurements . we also whish to thank n. rowley and w. catford for useful discussions and for a careful reading of the manuscript . one of us ( m.r . ) would like to acknowledge the conseil rgional dalsace for the financial support of his ph.d . thesis work . parts of this work has also been done in collaboration with c.e . during his summer stay at the ires with a janus grant of in2p3 . this work was sponsored by the french cnrs / in2p3 , and partially by the cnrs / nsf and cnrs / cnpq collaboration programs . d. shapira , r. novotny , y. c. chan , k. a. erb , j. l. c. ford jr . , j. c. peng , and j. d. moses , phys . lett . * 114b , 111(1982 ) ; d. shapira , d. schull , j. l .c . ford , jr . , b. shivakumar , r. l. parks , r. a. cecil , and s. t. thornton , phys . * 53 , 1634 ( 1984 ) . * * b. fornal , f. gramegna , g. prete , r. burch , g. derasmo , e. m. fiore , l. fiore , a. pantaleo , v. paticchio , g. viesti , p. blasi , n. gelli , f. lucarelli , m. anghinolfi , p. corvisiero , m. taiuti , a. zucchiatti , p. f. bortignon , j. ruiz , g. nebbia , m. gonin , and j. b. natowitz , + phys . b * 255 * , 325 ( 1991 ) . c. bhattacharya , m. rousseau , c. beck , v. rauch , r.m . freeman , d. mahboub , r. nouicer , p. papka , o. stezowski , a. hachem , e. martin , a.k . dummer , s.j . sanders , and a. szanto de toledo , phys . rev . c * 65 * , 014611 ( 2002 ) . m. rousseau _ et al . _ , in proceedings of the 7@xmath97 international conference on clustering aspects of nuclear structure and dynamics , eds . m. korolija , z. basrak and r. caplar ( world scientific publishing co. , singapore , 2000 ) , p.189 . et al . _ , in proceedings of the 9@xmath97 international conference on nuclear reactions , varenna , italy , 2000 , edited by e. gadioli [ ricerca scientifica ed educatione permanente , suppl . * 115 , ( 2000 ) p. 407 ] . * m. rousseau _ et al . _ , in proceedings of the xxxix international winter meeting on nuclear physics , bormio , italy , 2001 , edited by i. iori [ ricerca scientifica ed educatione permanente , suppl . * 117 , ( 2001 ) p. 370 . ] . * g. viesti , v. rizzi , d. fabris , m. lunardon , g. nebbia , m. cinausero , b , e. fioretto , g. prete , a. brondi , g. la rana , r. moro , e. vardaci , m. aiche , m. m. aleonard , g. barreau , d. boivin , j. n. scheurer , j. f. chemin , k. hagel , j. b. natowitz , r. wada , s. courtin , f. haas , n. rowley , b. m. nyako , j. gal and j. molnar , phys . lett . * 521b , 165 ( 2001 ) . * r. nouicer , c. beck , r .m . freeman , f. haas , n. aissaoui , t. bellot , g. de france , d. disdier , g. duchne , a. elanique , a. hachem , f. hoellinger , d. mahboub , v. rauch , s. j. sanders , a. dummer , f. w. prosser , a. szanto de toledo , sl . cavallaro , e. uegaki , and y. abe , phys . c * 60 , 41303 ( 1999 ) . * c. beck , r. nouicer , d. disdier , d. duchne , g. de france , r. m. freeman , f. haas , a. hachem , d. mahboub , v. rauch , m. rousseau , s. j. sanders , and a. szanto de toledo , phys . rev . c * 63 , 014607 ( 2001 ) . * figure 1 : experimental c ( solid squares ) , n ( solid triangles ) , and o ( solid circles ) cross sections measured in the @xmath1si + @xmath2c reaction @xcite as compared to the calculations ( dotted curves ) performed with the equilibrium model of orbiting @xcite . the solid curves are the predictions of the transition - state model @xcite . the open squares , triangles and circles are the present data of the c , n , and o fully - damped yields with error bars smaller than the size of the symbols . the full diamonds correspond to er cross sections quoted in refs . @xcite . figure 3 : angular distributions ( open symbols ) of heavy fragments ( c , n , o , and ne ) measured in the @xmath1si + @xmath2c reaction at e@xmath10 = 112 mev . the full symbols are the c , n and o data taken from ref . @xcite at e@xmath10 = 115 mev . the ne angular distribution has been scaled down by a factor 10 for the sake of clarity . the curves correspond to 1/sin@xmath32 functions . figure 4 : angular distributions of heavy fragments ( z = 5 to 11 ) measured in the @xmath1si + @xmath2c reaction at e@xmath10 = 180 mev . the n and na angular distributions have been scaled down by a factor 10 for the sake of clarity . the curves correspond to 1/sin@xmath32 functions . figure 5 : inclusive energy spectra of @xmath4 particles measured in the @xmath1si + @xmath2c reaction at e@xmath10 = 180 mev between @xmath44 = 30@xmath25 and 55@xmath25 . the experimental data are shown by the solid points with error bars visible when greater than the size of the points . the solid and dashed lines are statistical - model calculations discussed in the text . figure 6 : two - dimensional scatter plots of galilean - invariant cross sections ( d@xmath37/d@xmath38de)p@xmath39c@xmath39 of inclusive @xmath4 particles ( left side ) and protons ( right side ) measured in the ( v@xmath41,v@xmath40 ) plane for the @xmath1si + @xmath2c reaction at e@xmath10 = 112 mev ( up ) and 180 mev ( down ) . the experimental detector thresholds are drawn along the laboratory angles of each telescope . the circular arcs are centered on the velocity of the center of mass . figure 7 : exclusive energy spectra of @xmath4 particles emitted at the angles + 40@xmath98 + 65@xmath31 , in coincidence with individual p and s er s detected at -15@xmath25 in the @xmath1si + @xmath2c reaction at e@xmath10 = 112 mev . the experimental data are given by the solid points with error bars visible when greater than the size of the points . the solid lines are statistical - model calculations discussed in the text . figure 8 : exclusive energy spectra of @xmath4 particles emitted at the angles + 40@xmath98 + 95@xmath31 , in coincidence with individual p and s er s detected at -10@xmath25 in the @xmath1si + @xmath2c reaction at e@xmath10 = 180 mev . the experimental data are given by the solid points with error bars visible when greater than the size of the points . the solid lines are statistical - model calculations discussed in the text . figure 9 : exclusive energy spectra of protons emitted at the angles + 40@xmath99 + 70@xmath31 , in coincidence with individual p and s er s detected at -10@xmath25 , at the indicated laboratory angles , in the @xmath1si(180 mev ) + @xmath2c reaction . the solid lines are statistical - model calculations discussed in the text . figure 10 : in - plane angular correlations of coincident @xmath4 particles ( circles ) and protons ( triangles ) measured in the @xmath1si + @xmath2c reaction at e@xmath10 = 112 mev . the proton correlations have been multiplied by a factor 10@xmath100 for the sake of clarity . the arrow indicates the position of the ic detector at @xmath23= -15@xmath25 . on the abscissa , the positive angle refer to the opposite side of the beam from the direction of the er detected in ic . the solid lines correspond to statistical - model calculations discussed in the text . figure 11 : in - plane angular correlations of coincident @xmath4 particles ( circles ) and protons ( triangles ) measured in the @xmath1si + @xmath2c reaction at e@xmath10 = 180 mev . the proton correlations have been multiplied by a factor 10@xmath101 for the sake of clarity . the arrow indicates the position of the ic detector at @xmath23= -10@xmath25 . on the abscissa , the positive angle refer to the opposite side of the beam from the direction of the er detected in ic . the solid lines correspond to statistical - model calculations discussed in the text . figure 12 : out - of - plane angular correlations of coincident @xmath4 particles ( circles ) and protons ( triangles ) measured in the @xmath1si + @xmath2c reaction at e@xmath10 = 180 mev . the proton correlations have been multiplied by a factor 10@xmath101 for the sake of clarity . the er s are detected at @xmath23 = -10@xmath25 . the solid lines correspond to statistical - model calculations discussed in the text . figure 13 : energy - correlation plots between coincident @xmath4 particles and s er s produced in the @xmath1si + @xmath2c reaction at e@xmath10 = 180 mev . the heavy fragment is detected at @xmath82 = -10@xmath25 and the @xmath4-particle angle settings are given in the figure . the dashed lines correspond to different contours with their associated labellings discussed in the text . figure 14 : experimental ( left side ) and calculated ( right side ) energy - correlation plots between coincident @xmath4 particles and s er s produced in the @xmath1si + @xmath2c reaction at e@xmath10 = 180 mev . the s is identified at @xmath82 = -10@xmath25 and the @xmath4 particles are detected at @xmath83 = + 40@xmath25 , and + 70@xmath25 , respectively . cacarizo calculations are discussed in the text . figure 15 : excitation - energy spectra calculated for the @xmath1si + @xmath2c reaction at e@xmath10 = 180 mev for @xmath7be ( left side ) and @xmath6s ( right side ) in coincidence with the g.s . ( above ) and first excited level ( below ) of @xmath7be . the solid line corresponds to the energy of the first excited state of @xmath7be ( 3.08 mev ) . the dashed lines correspond to an excitation energy in @xmath6s expected for an @xmath4-transfer process .

the study of stellar populations in galaxies is entering a new era with the availability of large and high quality databases of both observed galactic spectra and state - of - the - art evolutionary synthesis models . in this paper we investigate the power of spectral synthesis as a mean to estimate physical properties of galaxies . spectral synthesis is nothing more than the decomposition of an observed spectrum in terms of a superposition of a base of simple stellar populations of various ages and metallicities , producing as output the star - formation and chemical histories of a galaxy , its extinction and velocity dispersion . our implementation of this method uses the bruzual & charlot ( 2003 ) models and observed spectra in the 36508000 range . the reliability of this approach is studied by three different means : ( 1 ) simulations , ( 2 ) comparison with previous work based on a different technique , and ( 3 ) analysis of the consistency of results obtained for a sample of galaxies from the sloan digital sky survey ( sdss ) . we find that spectral synthesis provides reliable physical parameters as long as one does not attempt a very detailed description of the star - formation and chemical histories . robust and physically interesting parameters are obtained by combining the ( individually uncertain ) strengths of each simple stellar population in the base . in particular , we show that besides providing excellent fits to observed galaxy spectra , this method is able to recover useful information on the distributions of stellar ages and , more importantly , stellar metallicities . stellar masses , velocity dispersion and extinction are also found to be accurately retrieved for realistic signal - to - noise ratios . we apply this synthesis method to a volume limited sample of 50362 galaxies from the sdss data release 2 , producing a catalog of stellar population properties . emission lines are also studied , their measurement being performed after subtracting the computed starlight spectrum from the observed one . a comparison with recent estimates of both observed and physical properties of these galaxies obtained by other groups shows good qualitative and quantitative agreement , despite substantial differences in the method of analysis . the confidence in the method is further strengthened by several empirical and astrophysically reasonable correlations between synthesis results and independent quantities . for instance , we report the existence of strong correlations between stellar and nebular metallicites , stellar and nebular extinctions , mean stellar age and equivalent width of h and 4000 break , and between stellar mass and velocity dispersion . galaxies : stellar content - galaxies : statistics - galaxies : fundamental parameters - galaxies : evolution

astro-ph0412481

galaxy spectra encode information on the age and metallicity distributions of the constituent stars , which in turn reflect the star - formation and chemical histories of the galaxies . retrieving this information from observational data in a reliable way is crucial for a deeper understanding of galaxy formation and evolution . the mapping of observed onto physical properties of galaxies has been a major topic of research for over a generation of astronomers since the pioneering works of morgan ( 1956 ) , wood ( 1966 ) and faber ( 1972 ) on the one hand , and tinsley ( 1968 ) and spinrad & taylor ( 1972 ) on the other . the first group of authors introduced the so - called empirical population synthesis methods , which aim at reproducing a set of observations of a given galaxy by means of a linear combination of simpler systems of known characteristics , like individual stars or chemically homogeneous and coeval groups of stars . bica ( 1988 ) , pelat ( 1997 ) , cid fernandes ( 2001 ) , moultaka ( 2004 ) are examples of modern studies following this approach . the second group pioneered the so - called evolutionary population synthesis methods , which compare galaxy data with models that follow the time evolution of an entire stellar system by combining libraries of evolutionary tracks and stellar spectra with prescriptions for the initial mass function ( imf ) , star formation and chemical histories . this approach has enjoyed more widespread use in recent years , e.g. , arimoto & yoshii ( 1987 ) , guiderdoni & rocca - volmerange ( 1987 ) , bressan ( 1994 ) , fioc & rocca - volmerange ( 1997 ) , vazdekis ( 1999 ) , bruzual & charlot ( 2003 , hereafter bc03 ) , le borgne ( 2004 ) , among many others ( see cardiel 2003 for a large set of references ) . in short , empirical synthesis relies on nature for its basic ingredients , whereas evolutionary synthesis relies mostly on models . however , since models are made and calibrated to mimic nature , this difference is gradually vanishing as models improve . besides different methodologies , there are also differences in what type of data is actually modeled . colours ( e.g. , wood 1966 ) , absorption line equivalent widths or spectral indices ( e.g. , worthey 1994 ; kauffmann 2003 , hereafter k03 ) and emission features , both stellar ( leitherer , robert & heckman 1995 ; schaerer & vacca 1998 ) and nebular ( mas - hesse & kunth 1991 ; kewley 2001 ) , have all been used in stellar population synthesis . more recently , the full spectral information has been incorporated in the modeling process , both including ( charlot & longhetti 2001 ) , and excluding emission lines ( vazdekis & arimoto 1999 ; reichardt , jimenez & heavens 2001 ; cid fernandes 2004a , hereafter cf04 ) . recovering the stellar content of a galaxy from its observed integrated spectrum is not an easy task , as can be deduced from the amount of work devoted to this topic over the past @xmath1 half century . the situation is however much more favorable nowadays . huge observational and theoretical efforts in the past few years have produced large sets of high quality spectra of stars ( e.g. , prugniel & soubiran 2001 ; le borgne 2003 ; bertone 2004 ; gonzlez delgado 2004 ) . these libraries are being implemented in a new generation of evolutionary synthesis models , allowing the prediction of galaxy spectra with an unprecedented level of detail ( vazdekis 1999 ; bc03 , le borgne 2004 ) . at the same time , galaxy spectra are now more abundant than ever ( loveday 1996 ; york 2000 ) . the sloan digital sky survey ( sdss ) , in particular , is providing a homogeneous data base of hundreds of thousands of galaxy spectra in the 38009200 range , with a resolution of @xmath2 ( york 2000 ; stoughton 2002 ; abazajian 2003 , 2004 ) . this enormous amount of high quality data will undoubtedly be at the heart of tremendous progress in our understanding of galaxy constitution , formation and evolution . indeed , significant steps in this direction have recently been made ( k03 ; brinchmann 2004 , tremonti 2004 ; heavens 2004 ; panter , heavens & jimenez 2004 ) . in order to take advantage of the recent progress in evolutionary synthesis to analyze data sets such as the sdss , a methodology must be set up to go from the observed spectra to physical properties of galaxies . in this paper we discuss one possible method to achieve this goal . the method is based on fitting an observed spectrum with a linear combination of simple theoretical stellar populations ( coeval and chemically homogeneous ) computed with evolutionary synthesis models at the same spectral resolution as that of the sdss . our goal here is to demonstrate that , besides providing excellent starlight templates to aid emission line studies , spectral synthesis recovers reliable stellar population properties out of galaxy spectra of realistic quality . we show that this simple method provides robust information on the stellar age ( @xmath3 ) and stellar metallicities ( @xmath4 ) distributions , as well as on the extinction , velocity dispersion and stellar mass . the ability to recover information on @xmath4 is particularly welcome , given that stellar metallicities are notoriously more difficult to assess than other properties . in order to reach this goal we follow : ( 1 ) _ a priori _ arguments , based on simulations ; ( 2 ) comparisons with independent work based on a different method , and ( 3 ) an _ a posteriori _ empirical analysis of the consistency of results obtained for a large sample of sdss galaxies . other papers in this series on the semi - empirical analysis of galaxies ( seagal ) will explore various astrophysical implications of the results obtained with this method . this paper is organized as follows . section [ sec : synthesis ] presents an overview of our synthesis method and simulations designed to test it and evaluate the uncertainties involved . the discussion is focused on how to use the synthesis to derive robust estimators of physically interesting stellar population properties . section [ sec : syntesis_sdss ] defines a volume limited sample of sdss galaxies and presents the results of the synthesis of their spectra , along with measurements of emission lines . in section [ sec : comparisons ] we compare our results to those recently published by brinchmann ( 2004 ) . stellar population and emission line properties are used in section [ sec : correlations ] to investigate whether the synthesis produces astrophysically plausible results . finally , section [ sec : conclusions ] summarizes our main findings . our synthesis code , which we call starlight , was first discussed in cf04 ( see also cid fernandes 2004b and garcia - rissman 2005 for different applications of the same code ) . starlight mixes computational techniques originally developed for empirical population synthesis with ingredients from evolutionary synthesis models . briefly , we fit an observed spectrum @xmath5 with a combination of @xmath6 simple stellar populations ( ssp ) from the evolutionary synthesis models of bc03 . extinction is modeled as due to foreground dust , and parametrized by the v - band extinction @xmath7 . the galactic extinction law of cardelli , clayton & mathis ( 1989 ) with @xmath8 is adopted . line of sight stellar motions are modeled by a gaussian distribution @xmath9 centered at velocity @xmath10 and with dispersion @xmath11 . with these assumptions the model spectrum is given by @xmath12 \otimes g(v_\star,\sigma_\star)\ ] ] where @xmath13 is the spectrum of the @xmath14 ssp normalized at @xmath15 , @xmath16 is the reddening term , @xmath17 is the synthetic flux at the normalization wavelength , @xmath18 is the _ population vector _ and @xmath19 denotes the convolution operator . each component @xmath20 ( @xmath21 ) represents the fractional contribution of the ssp with age @xmath22 and metallicity . ] @xmath23 to the model flux at @xmath15 . the base components can be equivalently expressed as a mass fractions vector @xmath24 . in this work we adopt a base with @xmath25 ssps , encompassing 15 ages between @xmath26 and @xmath27 yr and 3 metallicities : @xmath28 , 1 and 2.5 @xmath29 . their spectra , shown in fig . [ fig : base ] , were computed with the stelib library ( le borgne 2003 ) , padova 1994 tracks , and chabrier ( 2003 ) imf ( see bc03 for details ) . the fit is carried out with a simulated annealing plus metropolis scheme which searches for the minimum @xmath30 ^ 2 $ ] , where @xmath31 is the error in @xmath5 . regions around emission lines , bad pixels or sky residuals are masked out by setting @xmath32 . pixels which deviate by more than 3 times the rms between @xmath5 and an initial estimate of @xmath33 are also given zero weight . the minimization consists of a series of @xmath34 likelihood - guided metropolis explorations of the parameter space . from each iteration to the next we increase the @xmath35 weights geometrically ( which corresponds to a decrease in the `` temperature '' in statistical mechanics terms ; e.g. , mackay 2003 ) . the step - size in each parameter is concomitantly decreased and the number of steps is increased . this scheme gradually focuses on the most likely region in parameter space , avoiding ( through the logic of the metropolis algorithm ) trapping onto local minima . after completion , the whole fit is fine tuned repeating the full loop excluding all @xmath36 components . an important difference ( from the computational point of view ) with respect to the code in cf04 is that we now perform a series expansion of the @xmath37 extinction factor , which allows a much faster computation of @xmath38 . naturally , there are a number of technical parameters in this complex fitting algorithm , which we have optimized by means of extensive simulations ( see cf04 ) . at any rate , results reported here are robust with respect to variations in these technical parameters . figs . [ fig : examples_fits_1 ] and [ fig : examples_fits_2 ] illustrate the spectral fits obtained for two galaxies drawn from the sdss database . the top - left panel shows the observed spectrum ( thin line ) and the model ( thick ) , as well as the error spectrum ( dashed ) . the bottom - left panel shows the @xmath39 residual spectrum , while the panels in the right summarize the derived star - formation history encoded in the age - binned population vector . these examples , along with those in k03 and cf04 , demonstrate that this simple method is capable of reproducing real galaxy spectra to an excellent degree of accuracy . an important application of the synthesis is to measure emission lines from the residual spectrum , as done by k03 ( see also section [ sec : emissionlines ] ) . another , of course , is to infer stellar population properties from the fit parameters . this was not the approach followed by k03 and their subsequent papers , who prefer to derive stellar population properties mainly from the @xmath40 versus h@xmath41 diagram . our central goal here is to investigate whether spectral synthesis can also recover reliable stellar population properties . in the remainder of this section we address this issue by means of simulations . the existence of multiple solutions is an old known problem in stellar population synthesis . this multiplicity arises from a combination of three factors : ( 1 ) algebraic degeneracy ( number of unknowns larger than number of observables ) , ( 2 ) intrinsic degeneracies of stellar populations and ( 3 ) measurement uncertainties . unlike in methods which synthesize only a handful of spectral indices , algebraic degeneracy is not a problem for the method outlined above , as the number of @xmath5 points in any decently sampled spectrum far exceeds the number of parameters even for large bases . similarly , by modeling the whole spectrum one should be able to alleviate degeneracies in spectral indices associated with different stellar populations ( jimenez 2004 ) . measurement errors , however , are still a problem . a corollary of these complications is that even superb spectral fits as those shown in figs . [ fig : examples_fits_1 ] and [ fig : examples_fits_2 ] do not guarantee that the resulting parameter estimates are trustworthy . this brief discussion illustrates the need to assess the degree to which one can trust the parameters involved in the fit before using them to infer stellar populations properties . as posed above , the spectral fits involve @xmath42 parameters : @xmath43 of the @xmath18 components ( one degree of freedom is removed by the normalization constraint ) , @xmath17 , @xmath7 and the two kinematical parameters , @xmath10 and @xmath11 . the reliability of parameter estimation is best studied by means of simulations which feed the code with spectra generated with known parameters , add noise , and then examine the correspondence between input and output values . cf04 performed this kind of simulation for a @xmath44 base and spectra in the 35005200 interval . their analysis concentrated on how well the method recovers @xmath45 for varying degrees of noise . the main results of that study are : ( 1 ) in the absence of noise , the method recovers all components of @xmath18 to a high degree of accuracy . ( 2 ) in the presence of noise , however , the individual output @xmath20 fractions may deviate drastically from the input values . essentially , what happens is that noise washes away the differences between spectrally similar components , so it becomes impossible to distinguish them , and the code splits @xmath20 among neighboring components in spectral space . this is a common problem in population synthesis ( e.g. , panter 2004 ; tremonti 2003 ; cid fernandes 2001 ) . once we have identified the origin of the problem , the remedy to fix it is evident : bin - over spectrally similar components . in other words , instead of attempting a fine graded description of stellar populations mixtures in terms of many ages and metallicities , it is much better to `` marginalize over the details '' and work with a coarser but more robust description based on combined @xmath20 fractions . as shown by cf04 , condensed versions of the population vector which project its @xmath6 components onto just a few physically interesting axes , yield very reliable results . our approach and objectives are conceptually similar to those of the moped code of heavens , jimenez & lahav ( 2000 ; see also reichardt 2001 , panter 2004 ) . operationally , however , the two codes differ . moped compresses data on input by degrading the sdss spectral resolution to 20 and performing weighted linear combinations of @xmath5 , reducing its dimension to one datum per output parameter while at the same time minimizing the loss of information . starlight , on the other hand , works with the spectrum at its full resolution and an overdimensioned set of parameters ( i.e. , large @xmath6 ) , which we compress on the description of the output . the disadvantage of our approach is computational . indeed , it is described as a `` slow '' and `` brute force '' method by panter ( 2004 ) . this , however , is not a severe problem given the abundance of cpus nowadays ( starlight takes about 4 minutes per galaxy on a 2 ghz linux - workstation ) . furthermore , there is plenty of room for improvement in the efficiency of the algorithm , for instance , by implementing a smarter exploration of the parameter space ( slosar & hobson 2003 ) , to the point that computational constraints could soon become a minor concern . an advantage of starlight is that it also measures kinematical parameters ( mainly @xmath11 ) and provides a high resolution template spectrum . we have carried out new simulations designed to test the method and investigate which combinations of the parameters provide robust results . these simulations differ from those in cf04 in three main aspects : ( 1 ) the spectral range is now 36508000 ; ( 2 ) the new base is larger ; ( 3 ) a more realistic error spectrum was used . a further difference is that we use a larger number of iterations , partly because we have more base elements than cf04 and partly because the code is now over 200 times faster . at each step in the annealing scheme , the number of steps per parameter is set to @xmath46 , where @xmath47 is the length of the allowed range for the parameter ( e.g. , @xmath48 for @xmath20 fractions , which range from 0 to 1 ) , and @xmath49 is the step - size ( see cf04 for a more detailed description of these technical aspects ) . this insures that each parameter can in principle random - walk twice its whole allowed range , which is an adequate criterion for metropolis sampling ( e.g. , mackay 2003 ) . furthermore , there are 12 stages in the annealing schedule . overall , over @xmath50 combinations of parameters are sampled for each galaxy . several sets of simulations were performed . given our interest in modeling sdss galaxies , here we focus on simulations tailored to match the characteristics of this data set . test galaxies were built from the average @xmath18 , @xmath7 and @xmath11 within 65 boxes in the mean stellar age versus mean stellar metallicity plane obtained for the sample described in section [ sec : syntesis_sdss ] . each spectrum covers the 36508000 range in steps of 1 , the same sampling of the bc03 models . we generate 20 perturbed versions of each synthetic spectrum for each of 5 levels of noise : @xmath51 , 10 , 15 , 20 and 30 , where @xmath52 is the signal - to - noise ratio per in the region around @xmath53 . unlike in cf04 , who used a flat error spectrum , the error at each @xmath54 was assumed to follow a gaussian distribution with amplitude obtained by scaling a normalized mean sdss error spectrum to yield the desired @xmath52 at @xmath15 . this error spectrum decreases by a factor of @xmath55 from 3650 to 6200 and then increases again by @xmath56 towards 8000 . finally , as for the actual data fits , we mask points around @xmath573726,3729 , @xmath58 , h@xmath49 , h@xmath59 , h@xmath60 , h@xmath61 , @xmath574959,5007 , @xmath62 , nad@xmath63 , @xmath546300 , @xmath54@xmath546548,6583 , h@xmath0 , @xmath576717,6731 . these new simulations confirm that the individual components of @xmath18 are very uncertain , so we skip a detailed comparison between @xmath45 and @xmath64 and jump straight to results based on more robust descriptions of the synthesis output . a coarse but robust description of the star - formation history of a galaxy may be obtained by binning @xmath18 onto `` young '' ( @xmath65 yr ) , `` intermediate - age '' ( @xmath66 yr ) , and `` old '' ( @xmath67 yr ) components ( @xmath68 , @xmath69 and @xmath70 respectively ) . these age - ranges were defined on the basis of the simulations , by seeking which combinations of @xmath20 s produce smaller input @xmath71 output residuals . table [ tab : parameter_uncertainties ] and figure [ fig : input_x_ouput_1 ] show that these 3 components are very well recovered by the method , with uncertainties smaller than @xmath72 , @xmath73 , and @xmath74 for @xmath75 . lrrrrr + parameter & + & 5 & 10 & 15 & 20 & 30 + @xmath68 & 0.35 @xmath76 7.06 & 0.21 @xmath76 4.50 & 0.62 @xmath76 4.04 & 0.56 @xmath76 3.04 & 0.64 @xmath76 2.63 + @xmath69 & 0.76 @xmath76 13.94 & -0.06 @xmath76 9.00 & 0.01 @xmath76 7.88 & -0.22 @xmath76 6.26 & -0.02 @xmath76 5.07 + @xmath70 & -1.12 @xmath76 13.02 & -0.15 @xmath76 8.83 & -0.63 @xmath76 7.61 & -0.34 @xmath76 6.19 & -0.62 @xmath76 5.05 + @xmath77 & 0.02 @xmath76 1.66 & 0.11 @xmath76 1.41 & 0.16 @xmath76 1.18 & 0.18 @xmath76 1.05 & 0.19 @xmath76 0.80 + @xmath78 & 1.11 @xmath76 10.26 & 0.99 @xmath76 7.57 & 0.93 @xmath76 6.10 & 1.00 @xmath76 5.03 & 1.17 @xmath76 4.22 + @xmath79 & -1.13 @xmath76 10.54 & -1.10 @xmath76 8.17 & -1.09 @xmath76 6.54 & -1.18 @xmath76 5.59 & -1.36 @xmath76 4.63 + @xmath7 & 0.01 @xmath76 0.09 & 0.00 @xmath76 0.05 & 0.01 @xmath76 0.03 & 0.01 @xmath76 0.03 & 0.00 @xmath76 0.02 + @xmath10 & 0.24 @xmath76 17.73 & -0.11 @xmath76 8.55 & 0.07 @xmath76 5.92 & 0.00 @xmath76 4.23 & 0.05 @xmath76 2.81 + @xmath11 & -3.12 @xmath76 24.32 & -1.40 @xmath76 12.36 & -1.01 @xmath76 7.71 & -0.75 @xmath76 5.73 & -0.57 @xmath76 3.78 + @xmath80 & 0.01 @xmath76 0.11 & -0.01 @xmath76 0.08 & -0.01 @xmath76 0.06 & -0.01 @xmath76 0.05 & -0.02 @xmath76 0.04 + @xmath81 & -0.01 @xmath76 0.14 & 0.01 @xmath76 0.08 & 0.00 @xmath76 0.06 & 0.01 @xmath76 0.05 & 0.01 @xmath76 0.04 + @xmath82 & -0.03 @xmath76 0.20 & -0.04 @xmath76 0.14 & -0.03 @xmath76 0.11 & -0.04 @xmath76 0.10 & -0.04 @xmath76 0.08 + @xmath83 & -0.01 @xmath76 0.15 & -0.01 @xmath76 0.09 & 0.00 @xmath76 0.08 & 0.00 @xmath76 0.06 & 0.00 @xmath76 0.05 + @xmath84 & -0.03 @xmath76 0.18 & -0.02 @xmath76 0.13 & -0.03 @xmath76 0.11 & -0.02 @xmath76 0.09 & -0.03 @xmath76 0.08 + @xmath85 & 0.03 @xmath76 0.16 & 0.00 @xmath76 0.10 & 0.00 @xmath76 0.08 & -0.01 @xmath76 0.07 & -0.01 @xmath76 0.06 + @xmath86 & -0.04 @xmath76 0.09 & -0.02 @xmath76 0.06 & -0.01 @xmath76 0.05 & -0.01 @xmath76 0.04 & 0.00 @xmath76 0.04 + @xmath87 & -0.01 @xmath76 0.21 & 0.00 @xmath76 0.14 & 0.02 @xmath76 0.11 & 0.02 @xmath76 0.10 & 0.02 @xmath76 0.09 + @xmath88 & -0.11 @xmath76 0.26 & -0.05 @xmath76 0.19 & -0.03 @xmath76 0.16 & -0.02 @xmath76 0.14 & -0.02 @xmath76 0.12 + @xmath89 & 0.95 @xmath76 0.02 & 0.95 @xmath76 0.02 & 0.95 @xmath76 0.02 & 0.95 @xmath76 0.02 & 0.95 @xmath76 0.02 + @xmath90 & 11.64 @xmath76 4.05 & 5.37 @xmath76 1.19 & 3.54 @xmath76 0.78 & 2.65 @xmath76 0.58 & 1.76 @xmath76 0.39 + @xmath91 & 0.78 @xmath76 0.27 & 0.68 @xmath76 0.20 & 0.64 @xmath76 0.21 & 0.59 @xmath76 0.22 & 0.51 @xmath76 0.23 + fig . [ fig : input_x_ouput_1 ] also shows the input versus output values of @xmath7 , @xmath11 and the stellar mass @xmath92 . the latter is not an explicit input parameter of the models , but may be computed from @xmath24 and the @xmath93 ratio of the different populations in the base . the uncertainties in these parameters are @xmath94 mag , @xmath95 dex and @xmath96 km@xmath97s@xmath98 for @xmath75 . ( table [ tab : parameter_uncertainties ] ) . if one had to choose a single parameter to characterize the stellar population mixture of a galaxy , the option would certainly be its mean age . we define two versions of mean stellar age ( the logarithm of the age , actually ) , one weighted by light @xmath99 and another weighted by stellar mass @xmath100 note that , by construction , both definitions are limited to the 1 myr13 gyr range spanned by the base . the mass weighted mean age is in principle more physical , but , because of the non - constant @xmath101 of stars , it has a much less direct relation with the observed spectrum than @xmath81 . [ fig : input_x_ouput_mean_age_and_z ] shows the input against output @xmath81 for simulations with @xmath102 and 20 . the plots show that the mean age is a very robust quantity . the rms difference between input and output @xmath81 values is @xmath103 dex for @xmath104 , and @xmath105 dex for @xmath82 ( table [ tab : parameter_uncertainties ] ) . although the uncertainties of @xmath81 and @xmath82 are comparable in absolute terms , the latter index spans a smaller dynamical range ( because of the large @xmath101 ratio of old populations ) , so in practice @xmath81 is the more useful of the two indices . given that the mean stellar age is so well recovered by the method one might attempt more detailed descriptions of the star - formation history involving , say , higher moments of the age distribution . for instance , @xmath106^{1/2}\ ] ] measures the flux - weighted standard deviation of the log age distribution , and might be useful to distinguish galaxies dominated by a single population from those which had continuous or bursty star - formation histories . the uncertainty in this index is of order 0.1 dex . given an option of what to chose as a second parameter to describe a mixed stellar population , the choice would likely be its typical metallicity . analogously to what we did for ages , we define both light and mass - weighted mean stellar metallicities : @xmath107 and @xmath108 both of which are bounded by the 0.22.5 @xmath29 base limits . fig . [ fig : input_x_ouput_mean_age_and_z ] and table [ tab : parameter_uncertainties ] show that the rms of @xmath109 is of order 0.1 dex . in absolute terms this is comparable to @xmath110 , but note that @xmath111 covers a much narrower dynamical range than @xmath112 , so that in practice mean stellar metallicities are more sensitive to errors than mean ages . this is not surprising , given that age is the main driver of variance among ssp spectra , metallicity having a `` second - order '' effect ( e.g. , schmidt 1991 ; ronen , aragon - salamanca & lahav 1999 ) . this is the reason why studies of the stellar populations of galaxies have a much harder time estimating metallicities than ages , to the point that one is often forced to bin - over the @xmath113 information and deal only with age - related estimates such as @xmath112 ( e.g. , cid fernandes 2001 ; cid fernandes , leo & rodrigues lacerda 2003 ; k03 ) . notwithstanding these notes , it is clear that uncertainties of @xmath114 dex in @xmath111 are actually good news , since they do allow us to recover useful information on an important but hard to measure property . this new tracer of stellar metallicity is best applicable to large samples of galaxies such as the sdss . the statistics of samples help reducing uncertainties associated with @xmath111 estimates for single objects and allows one to investigate correlations between @xmath111 and other galaxy properties ( sodr , in preparation ; section [ sec : stellar_x_nebular_metallicities ] ) . our method tends to underestimate @xmath111 for metal - rich systems and vice - versa . this bias is due to the infamous age - metallicity degeneracy . in order to verify to which degree our synthesis is affected by this well known problem ( e.g. , renzini & buzzoni 1986 ; worthey 1994 ; bressan , chiosi & tantalo 1996 ) we have examined the correlation between the output minus input residuals in @xmath115 and @xmath83 . the age-@xmath113 degeneracy acts in the sense of confusing old , metal - poor systems with young , metal - rich ones and vice - versa , which should produce anti - correlated residuals . these are indeed seen in figure [ fig : age_z_degeneracy ] , where small symbols represent all 1300 individual simulations and filled circles correspond to the mean residuals obtained for each set of 20 perturbations of the 65 test galaxies . this anti - correlation is also present with the mass - weighted mean stellar metallicity @xmath116 , but is not as strong as for @xmath117 . on the other hand , the uncertainty in @xmath116 is always larger than for @xmath117 ( table [ tab : parameter_uncertainties ] ) . the age-@xmath113 degeneracy is thus present in our method , introducing systematic biases in our @xmath111 and @xmath112 estimates at the level of up to @xmath1180.2 dex . none of the results reported in this paper rely on this level of precision . we have also carried out a separate set of simulations for 100 test galaxies generated by random combinations of @xmath18 , @xmath7 , @xmath11 . these simulations differ from the ones presented above in that the test galaxies are not restricted to match the range of properties inferred from the application of the synthesis to sdss galaxies ( section [ sec : syntesis_sdss ] ) . given that these random galaxies span a broader ( but less representative of our sample galaxies ) region of the parameter space , one might expect to find larger uncertainties in the derived properties . this was indeed confirmed in the numerical experiments , although the effect is small . the uncertainties in the @xmath119 binned fractions increase by just a couple of percentage points with respect to those in table [ tab : parameter_uncertainties ] . similarly , the increase in @xmath120 is of just 0.02 mag , while @xmath121 and @xmath122 increase by @xmath123 and 0.03 dex respectively . the only properties whose uncertainties are substantially larger that those reported in table [ tab : parameter_uncertainties ] are the stellar mass and velocity dispersion . for instance , while the sdss - based simulations for @xmath102 yield @xmath124 dex and @xmath125 km@xmath97s@xmath98 , with random galaxies these errors increase to 0.15 dex and 24 km@xmath97s@xmath98 respectively . the reason for this apparent discrepancy is due to the fact that the set of random test - galaxies contains a larger proportion of systems dominated by very young stellar populations than we find for sdss galaxies . an error @xmath126 in the light fraction associated to these populations must be compensated by errors in the older components , which carry most of the mass , even when @xmath68 is large . as illustrated by the size of the error - bars in figure [ fig : input_x_ouput_1 ] , @xmath126 increases as @xmath68 increases , so the errors in the mass fractions components increase too , leading to larger dispersion in @xmath92 . furthermore , galaxies with large @xmath68 have few absorption features to constrain the kinematical broadening of the spectrum , which explains the larger dispersion in our @xmath11 estimates . to prove this point , we have re - evaluated the uncertainties in @xmath92 and @xmath11 excluding test galaxies with @xmath127% , which corresponds to systems which formed @xmath128% of their stellar mass over the past @xmath129 yr . for this subset of the simulations , which comprise 72 galaxies ( each one split onto 20 different spectra corresponding to independent monte carlo realizations of the noise ) , the uncertainties in mass and velocity dispersion decrease to @xmath130 dex and @xmath131 km@xmath97s@xmath98 ( for @xmath132 ) , only slightly larger than those reported in section [ sec : mass_and_vd ] . uncertainties in other properties also decrease to values very similar to those listed in table [ tab : parameter_uncertainties ] . we thus conclude that the parameter uncertainties studied in section [ sec : robustdecription ] and summarized in table [ tab : parameter_uncertainties ] are only moderately affected by the design of the simulations , and represent fair estimates of the limitations of our synthesis method . finally , we have carried out simulations using the @xmath133 bc03 ssps to generate further test galaxies . galaxies with such low metallicity are not expected to be present in significant numbers in the sample described in section [ sec : sample_definition ] , given that it excludes low luminosity systems like hii galaxies and dwarf ellipticals ( which are also the least metallic ones by virtue of the mass - metallicity relation ) . still , it is interesting to investigate what would happen in this case . when synthesized with our 0.2@xmath134 base ( fig . [ fig : base ] ) , these extremely metal poor galaxies are modeled predominantly with the @xmath135 components , as intuitively expected . moreover , the mismatch in metallicity introduces non - negligible biases in other properties , like masses , mean ages and extinction ( @xmath92 , for instance , is systematically underestimated by 0.3 dex ) . similar problems should be encountered when modeling systems with @xmath136 . these results serve as a reminder that our base spans a wide but finite range in stellar metallicity , and that extrapolating these limits has an impact on the derived physical properties . while there is no straightforward _ a priori _ diagnostic of which galaxies violate these limits , in general , one should be suspicious of objects with mean @xmath4 too close to the base limits . summarizing this theoretical study , we have performed simulations designed to evaluate the accuracy of our spectral synthesis method . the simulations mimic in as much as possible the wavelength range , spectral resolution , error spectrum , and @xmath52 of the actual sdss data studied below . several physically motivated combinations of the synthesis parameters were investigated to establish their precision at different @xmath52 . table [ tab : parameter_uncertainties ] summarizes the uncertainties in these quantities and a few additional ones not explicitly mentioned above . in what follows we focus on 5 parameters : stellar mass , velocity dispersion , extinction , mean stellar ages and metallicities , all of which were found to be adequately recovered . this exercise demonstrates that we are capable of producing reliable estimates of several parameters of astrophysical interest , at least in principle . we must nevertheless emphasize that this conclusion relies entirely on models and on an admittedly simplistic view of galaxies . when applying the synthesis to real galaxy spectra , a series of other effects come into play . for instance , the extinction law appropriate for each galaxy likely differs from the one used here : in our galaxy , the ratio of total to selective extinction of stellar sources , @xmath137 , is known to depend on the line of sight ( cardelli 1989 ; patriarchi 2001 ) ; in addition , the wavelength dependence of the attenuation of light from an extended source such as a galaxy includes the effects of scattering back into the light beam , and depends on the relative distribution of stars and dust ( witt 1992 ; gordon 1997 ) . also , one might expect that each population of stars is affected by a distinct extinction ( panuzzo 2003 ; charlot & fall 2000 ) . similarly , while in the evolutionary tracks adopted here the metal abundances are scaled from the solar values , non - solar abundance mixtures are known to occur in stellar systems ( e.g. , trager 2000a , b ) , not to mention uncertainties in the ssp models and the always present issue of the imf . in short , evidence against these simplifying hypothese abound . accounting for all these effects in a consistent way is not currently feasible . we mention these caveats not to dismiss simple models , but to highlight that all parameter uncertainties discussed above are applicable within the scope of the model . hence , while the simulations lend confidence to the synthesis method , one might remain skeptical of its actual power . the next sections further address the reliability of the synthesis , this time from a more empirical perspective . in this section we apply our synthesis method to a large sample of sdss galaxies to estimate their stellar population properties . we also present measurements of emission line properties , obtained from the observed minus synthetic spectra . the information provided by the synthesis of so many galaxies allows one to address a long menu of astrophysical issues related to galaxy formation and evolution . before venturing in the exploration of such issues , however , it is important to validate the results of the synthesis by as many means as possible . hence , the goal of the study presented below is not so much to explore the physics of galaxies but to provide an empirical test of our synthesis method . the results reported in this section are used in sections [ sec : comparisons ] and [ sec : correlations ] with this purpose . the spectroscopic data used in this work were taken from the sdss . this survey provides spectra of objects in a large wavelength range ( 38009200 ) with mean spectral resolution @xmath138 , taken with 3 arcsec diameter fibers . the most relevant characteristic of this survey for our study is the enormous amount of good quality , homogeneously obtained spectra . the data analyzed here were extracted from the sdss main galaxy sample available in the data release 2 ( dr2 ; abazajian 2004 ) . this flux - limited sample consists of galaxies with reddening - corrected petrosian @xmath139-band magnitudes @xmath140 , and petrosian @xmath139-band half - light surface brightnesses @xmath141 mag arcsec@xmath142 ( strauss 2002 ) . from the main sample , we first selected spectra with a redshift confidence @xmath143 . following the conclusions of zaritsky , zabludoff , & willick ( 1995 ) , we have imposed a redshift limit of @xmath144 ( trying to avoid aperture effects and biases ; see e.g. gmez 2003 ) and selected a volume limited sample up to @xmath145 , corresponding to a @xmath139-band absolute magnitude limit of @xmath146 . the absolute magnitudes used here are k - corrected with the help of the code provided by blanton ( 2003 ; ` kcorrect v3_2 ` ) and assuming the following cosmological parameters : @xmath147 = 70 km s@xmath98 mpc@xmath98 , @xmath148 and @xmath149 . we also restricted our sample to objects for which the observed spectra show a @xmath52 ratio in @xmath150 , @xmath139 and @xmath151 bands greater than 5 . these restrictions leave us with a volume limited sample containing @xmath152 galaxies , which leads to a completeness level of @xmath1 98.5 per cent . all 50362 spectra were brought to the rest - frame ( using the redshifts in the sdss database ) , sampled from 3650 to 8000 in steps of 1 , corrected for galactic extinction using the maps given by schlegel , finkbeiner & davis ( 1998 ) and the extinction law of cardelli ( 1989 , with @xmath153 , and normalized by the median flux in the 40104060 region . the @xmath52 ratio in this spectral window spans the 530 range , with median value of 14 . besides the masks around the lines listed in section [ sec : simulations ] , we exclude points with sdss flag @xmath154 , which signals bad pixels , sky residuals and other artifacts . after this pre - processing , the spectra are fed into the starlight code described in section [ sec : synthesismethod ] . on average , the synthesis is performed with @xmath155 points , after discounting the ones which are clipped by our @xmath156 sigma threshold ( typically 40 points ) and the masked ones . the spectral fits are generally very good , as illustrated in figs.[fig : examples_fits_1 ] and [ fig : examples_fits_2 ] . the mean value of @xmath89 is 0.78 . in fact , this is somewhat too good , since from the simulations we expect @xmath157 . this is a minor difference , which could be fixed decreasing the errors in @xmath5 by @xmath158% . we further quantify the quality of the fits by @xmath90 , the mean value of @xmath159 over all non - masked points . from the simulations we expect this alternative figure of merit to be of order of 0.6 times the noise - to - signal ratio at @xmath15 ( table [ tab : parameter_uncertainties ] ) . this is exactly the mean value of @xmath160 in the actual fits , again indicating acceptable fits . the total stellar masses of the galaxies were obtained from the stellar masses derived from the spectral synthesis ( which correspond to the light entering the fibers ) by dividing them by @xmath161 , where @xmath162 is the fraction of the total galaxy luminosity in the @xmath163-band outside the fiber . this approach , which neglects stellar population and extinction gradients , leads to an increase of typically 0.5 in @xmath80 . we did not apply any correction to the velocity dispersion estimated by the code given that the spectral resolution of the bc03 models and the data are very similar . we point out that we did not constrain the extinction @xmath7 to be positive . there are several reasons for this choice : ( a ) some objects may be excessively dereddened by galactic extinction ; ( b ) some objects may indeed require bluer ssp spectra than those in the base ; ( c ) the observed light may contain a scattered component , which would induce a bluening of the spectra not taken into account by the adopted pure extinction law ; ( d ) constraining @xmath7 to have only positive values produces an artificial concentration of solutions at @xmath164 , an unpleasant feature in the @xmath7 distribution . interestingly , most of the objects for which we derive negative @xmath7 ( typically @xmath165 to @xmath166 mag ) are early - type galaxies . these galaxies are dominated by old populations , and expected to contain little dust . this is consistent with the result of k03 , who find negative extinction primarily in galaxies with a large @xmath40 . the distribution of @xmath7 for these objects , which can be selected on the basis of spectral or morphological properties , is strongly peaked around @xmath167 , so that objects with @xmath168 can be considered as consistent with having zero extinction . in any case , none of the results reported in this paper is significantly affected by this choice . besides providing estimates of stellar population properties , the synthesis models allow the measurement of emission lines from the `` pure - emission '' , starlight subtracted spectra @xmath169 . we have measured the lines of @xmath573726,3729 , @xmath544363 , h@xmath61 , @xmath574959,5007 , @xmath546300 , @xmath546548 , h@xmath0 , @xmath546584 and @xmath576717,6731 . each line was treated as a gaussian with three parameters : width , offset ( with respect to the rest - frame central wavelength ) , and flux . lines from the same ion were assumed to have the same width and offset . we have further imposed @xmath170/@xmath171 and @xmath172/@xmath173 flux ratio constraints . finally , we consider a line to have significant emission if its fit presents a @xmath52 ratio greater than 3 . in some of the following analysis , galaxies with emission lines are classified according to their position in the /h@xmath61 versus /h@xmath0 diagram proposed by baldwin , phillips & terlevich ( 1981 ) to distinguish normal star - forming galaxies from galaxies containing active galactic nuclei ( agn ) . we define as normal star - forming galaxies those galaxies that appear in this diagram and are below the curve defined by k03 ( see also brinchmann 2004 ) . objects above this curve are transition objects and galaxies containing agn . a point that deserves mention here is the bias that may be introduced in the analysis due to the use of small fibers to measure the galaxy spectra . this effect , known as aperture bias , may introduce a redshift dependence in the measured galaxy spectra , since the fraction of galaxy light received by a fiber increases with increasing distance . zaritsky ( 1995 ) , based on an analysis of spectra from the las campanas redshift survey , find that a lower limit on redshifts of @xmath174 minimizes the aperture bias . this effect has also been discussed by k03 , gmez ( 2003 ) and tremonti ( 2004 ) for sdss spectra . for instance , these authors found that the galaxy @xmath163-band @xmath101 ratio and the gas - phase oxygen abundance decrease by @xmath1 0.1 dex over the redshift range of the survey , indicating that these quantities are moderately affected by aperture bias . the redshift range considered in our study is smaller than that of the entire survey , implying that the aperture effects are expected to be even smaller in our sample . in order to verify whether our results are significantly affected by this bias , we investigated the behaviour of some of the quantities resulting from our analysis ( @xmath92 , @xmath111 , @xmath112 and @xmath7 ) as a function of redshift for galaxies with luminosities above and below @xmath175 , the median luminosity of our sample . we divided the sample in several redshift bins containing the same number of objects , and computed the median value and the quartiles of each quantity in each bin . [ fig : aperturebias ] shows as solid lines the median values of each distribution , as well as their respective quartiles . none of the quantities seems to be strongly affected by this bias . the largest correlation with @xmath163 appears for light - weighted ages , corresponding to a change along our redshift distribution of about @xmath176 and @xmath177 dex for faint and bright galaxies , respectively . this is a plausible result since the light of nearby objects seen by the fiber aperture is dominated by the older stellar populations of their bulges . for the other quantities the variations are well below 0.1 dex . the sdss database has been explored by several groups , using different approaches and techniques . the mpa / jhu group has recently publicly released catalogues of derived physical properties for 211894 sdss galaxies , including 33589 narrow - line agn ( k03 , see also brinchmann 2004 ) . these catalogues are based on the k03 method to infer the star formation histories , dust attenuation and stellar masses of galaxies from the simultaneous analysis of the 4000 break strength , @xmath40 , and the balmer line absorption index h@xmath41 . these two indices are used to constrain the mean stellar ages of galaxies and the fractional stellar mass formed in bursts over the past few gyr , and a comparison with broad - band photometry then allows to estimate the extinction and stellar masses . the mpa / jhu catalogues provide very useful benchmarks for similar studies . in this section we compare the values of some of the parameters from these catalogues with our own estimates . catalogues of galaxy properties obtained with our synthesis method will be made available in due course . besides comparing directly measurements of physical quantities , our aim is also to highlight the differences that may appear due to the use of different methods and procedures . -band determined by the mpa / jhu group with that estimated though the method described in this paper . the solid line shows a @xmath178 linear fit to the data , while the dash ones is the identity line ( @xmath179 ) . ] the mpa / jhu group estimates the @xmath163-band stellar extinction @xmath180 through the difference between model and measured colours , assuming an attenuation curve proportional to @xmath181 . in our case , the extinction @xmath7 is derived directly from the spectral fitting , carried out with the milky way extinction law ( cardelli 1989 , c.f . section [ sec : synthesismethod ] ) , for which @xmath182 assuming @xmath183 . [ fig : comparisons_av ] shows that these two independent estimates are very strongly and linearly correlated , with a spearman rank correlation coefficient @xmath184 . however , the values of @xmath180 reported by the mpa / jhu group ( column 17 of their stellar mass catalogue ) are systematically larger than our values : @xmath185 in the median . this discrepancy is only apparent . the galactic extinction law is substantially harder than @xmath181 . one thus expects to need less extinction when modeling a given galaxy with the former law than with the latter . this was confirmed by starlight fits to a sub - set of sdss galaxies using a @xmath181-law , which yield a value of @xmath7 typically 1.77 times larger than those obtained with the cardelli ( 1989 ) law . since the @xmath186 conversion factors are 0.7030 and 0.4849 for the @xmath181 and cardelli laws , one finds that the @xmath180 values obtained for the two laws should differ by a factor of @xmath187 . this is very close to the empirically derived factor of 2.51 ( fig . [ fig : comparisons_av ] ) . we thus conclude that there are no substantial differences between the mpa / jhu and our estimates of the stellar extinction other than those implied by the differences in the reddening laws adopted in the two studies . [ fig : comparisons_masses]a compares our results for the total stellar masses to the mpa / jhu extinction - corrected stellar masses ( column 9 of their stellar mass catalogue ) . the two estimates of @xmath92 correlate very well , with @xmath188 . the quantitative agreement is also good , with a median difference of just 0.1 dex . this small offset can not be attributed to the different imfs employed in the two studies ( chabrier 2003 here and kroupa 2001 in k03 ) , since , as illustrated in figure 4 of bc03 , these two imfs yield practically identical @xmath101 ratios . instead , this offset seems to be due to a subtle technicality . whereas we adopt the @xmath101 ratio of the best @xmath38 model , the mpa / jhu group derives @xmath101 comparing the observed values of the @xmath40 and @xmath189 indices with a library of 32000 models . each model is then weighted by its likelihood , and a probability distribution for @xmath101 is computed . the mpa / jhu mass is the median of this distribution , which is not necessarily the same as the best-@xmath38 value . in fact , the stellar mass catalogue of brinchmann ( 2004 ) lists in its column 8 the best @xmath38 masses , which are systematically larger than the median ones by @xmath114 dex , identical to the median difference identified above . part of the scatter in fig . [ fig : comparisons_masses]a can be attributed to the different extinction laws . extinction contributes @xmath190 to the estimated @xmath80 . as discussed above , using a @xmath181 law in the synthesis yields values of @xmath7 which are 1.77 times larger than using the cardelli ( 1989 ) law . furthermore , the mass - to - light ratios obtained with the two laws are very similar . from this we expect that @xmath191 , in good agreement with the observed relation ( fig . [ fig : comparisons_masses]b ) . correcting for this effect by adding @xmath192 to our mass estimates indeed produces a better correlation , with @xmath193 , as illustrated in fig . [ fig : comparisons_masses]c . overall , we conclude that the two mass estimates agree to within 0.4 dex . this level of agreement is similar to that recently found by drory , bender & hopp ( 2004 ) in their comparison between the mpa / jhu masses and estimates based on sdss plus 2mass photometry . it is quite remarkable that , despite the substantial differences between our approaches and the underlying assumptions , the estimated stellar masses are so similar over a wide range of masses . on the other hand , this may not be so surprising given that the mpa / jhu group estimates of physical properties are ultimately based on a comparison of observed indices with an extensive library of galaxy spectra constructed out of the bc03 evolutionary synthesis models ( k03 ) . since this library spans a wide of metallicities and star - formation histories , the agreement between our estimates and those of the mpa / jhu group may be simply indicating that the @xmath25 ssps from bc03 used in our spectral synthesis span a similar parameter - space to that covered by the k03 models . linear fit to the data , while the dash ones is the identity line ( @xmath179 ) . ] we use the sub - sample of galaxies with active nuclei to compare our measurements of absorption line broadening due to galaxy velocity dispersion and/or rotation , @xmath11 , with those of the mpa / jhu group , since they list this quantity only in their agn catalogue ( in column 16 ) . the comparison is displayed in fig . [ fig : comparisons_vd_and_zneb]a . the spearman correlation - coefficient in this case is @xmath194 and the median of the difference between the two estimates is just 9 km@xmath97s@xmath98 , indicating an excellent agreement between both studies . brinchmann ( 2004 ) also provide , in their emission line catalogue , emission lines fluxes and equivalent widths which can be compared with our own measurements . in both studies the emission lines are measured after subtracting from the observed spectrum a model spectrum representing the stellar emission . in our case this is done with our synthesized spectra . the mpa / jhu group adopts a similar approach ( see tremonti 2004 for a brief description ) by fitting the observed continuum with bc03 models . they adopt , however a single metallicity model and a different extinction law . we have compared the fluxes and equivalent widths of h@xmath0 , @xmath172 , @xmath195 , h@xmath61 and @xmath170 as measured by our code and that obtained by the mpa / jhu group . we do not find any significant difference between these values ; the largest discrepancy ( @xmath1 5 per cent ) was found for the equivalent widths of h@xmath0 and , probably due to different estimates of the continuum level and the associated underlying stellar absorption . our emission line measurements are also in good agreement with those in stasiska ( 2004 ) , who fit the balmer lines with emission and absorption components , instead of subtracting a starlight model . in fig . [ fig : comparisons_vd_and_zneb]b we plot our estimates of the nebular oxygen abundance against those obtained by the mpa / jhu group ( catalogue of gas phase metallicities , median values ; see tremonti 2004 ) . in order not to introduce any bias due to the use of different indicators of the oxygen abundance , we have estimated o / h using the calibration of the ( @xmath543726,3729 + @xmath544959,5007)/ h@xmath61 ratio given by tremonti ( 2004 ) in their equation ( 1 ) . this calibration is based on simultaneous fits of the most prominent emission lines with a model designed for the interpretation of integrated galaxy spectra ( charlot & longhetti 2001 ) . the oxygen abundances estimated in such a way differ by just @xmath196 dex , as shown in fig . [ fig : comparisons_vd_and_zneb]b . overall , we conclude that our spectral synthesis method yields estimates of physical parameters in good agreement with those obtained by the mpa / jhu group , considering the important differences in approach and underlying assumptions . yet another way to assess the validity of physical properties derived through a spectral synthesis analysis is to investigate whether this method yields astrophysically reasonable results . in this section we follow this empirical line of reasoning by comparing some results obtained from our synthesis of sdss galaxies ( which excludes emission lines ) with those obtained from a direct analysis of the emission lines . our aim is to demonstrate that our synthesis results do indeed make sense . a more detailed discussion of most of the points below will be presented in other papers of this series . our spectral synthesis approach yields estimates of the mean metallicity of the stars in a galaxy , @xmath111 . the analysis of emission lines , on the other hand , gives estimates of the present - day abundances in the warm interstellar medium . although stellar and nebular metallicities are not expected to be equal , it is reasonable to expect that they should roughly scale with each other . fig . [ fig : zneb_zst ] shows the correlation between mass - weighted stellar metallicities and the nebular oxygen abundance ( computed as in section [ sec : comparison_emissionlines ] ) , both in solar units ( allende prieto , lambert & apslund 2001 ) . ] , for our sample of normal star - forming galaxies . a correlation is clearly seen , although with large scatter ( @xmath197 ) . galaxies with large stellar metallicities also have large nebular oxygen abundances ; galaxies with low stellar abundances tend to have smaller abundances . the observed scatter is qualitatively expected due to variations in enrichment histories among galaxies . a robust linear fitting gives the following relation : @xmath198 with a dispersion of 0.08 dex . notice that in this expression both stellar and nebular metallicities are normalized to solar units ( @xmath199 and @xmath200 respectively ) . nebular and stellar metallicities are estimated through completely different and independent methods , so the correlation depicted in fig . [ fig : zneb_zst ] provides an _ a posteriori _ empirical validation for the stellar metallicity derived by the spectral synthesis . the possibility to estimate stellar metallicities for so many galaxies is one of the major virtues of spectral synthesis , as it opens an important window to study the chemical evolution of galaxies and of the universe as a whole ( panter 2004 ; sodr in prep . ) . the stellar extinction in the v - band is one of the products of our starlight code . a more traditional and completely independent method to evaluate the extinction consists of comparing the observed h@xmath0/h@xmath61 balmer decrement to the theoretical value . the intrinsic value of @xmath201 is not very sensitive to the physical conditions of the gas , ranging from 3.03 for a gas temperature of 5000 k to 2.74 at 20000 k ( osterbrock 1989 ) . adopting a value of 2.86 for this ratio , the `` balmer extinction '' ( stasiska 2004 ) is given by @xmath202\ ] ] where the 6.31 coefficient comes from assuming the cardelli ( 1989 ) extinction curve . [ fig : av_chb ] presents a comparison between the stellar @xmath7 and @xmath203 . these two extinctions are determined in completely independent ways , and yet , our results show that they are closely linked , with @xmath204 . a linear bisector fitting yields @xmath205 note that the angular coefficient in this relation indicates that nebular photons are roughly twice as extincted as the starlight . this `` differential extinction '' is in very good qualitative and quantitative agreement with empirical studies ( fanelli 1988 ; calzetti , kinney & storchi - bergmann 1994 ; gordon 1997 ; mas - hesse & kunth 1999 ) . we shall explore the implications of this result for the intrinsic colours of star - forming galaxies in another paper of this series ( stasiska , in prep . ) . ) and stellar ( @xmath7 ) extinctions for our sample of normal star - forming galaxies . the solid line is a robust fit for the relation . ] versus the light - weighted mean stellar age for normal emission line galaxies in our sample . ( b ) relation between the d4000 index and the light - weighted mean stellar age . the solid lines are robust fits for the relations . ] the equivalent width ( ew ) of h@xmath0 is related to the ratio of present to past star formation rate of a galaxy ( e.g. , kennicutt 1998 ) . it is thus expected to be smaller for older galaxies . fig . [ fig : age_ewhalpha_d4000]a shows the relation between ew(h@xmath0 ) and the mean light - weighted stellar age obtained by our spectral synthesis . the anti - correlation , which has @xmath206 , is evident . from this plot we can derive an empirical relation which can be used to estimate @xmath81 ( for @xmath53 ) of star - forming galaxies through the measurement of ew(h@xmath0 ) : @xmath207 for @xmath3 in yr and ew(h@xmath0 ) in . it is worth stressing that these two quantities are obtained independently , since the spectral synthesis does not include emission lines . another quantity that is considered a good age indicator , even for galaxies without emission lines , is the 4000 break , d4000 . we measured this index following bruzual ( 1983 ) , who define d4000 as the ratio between the average value of @xmath208 in the 40504250 and 37503950 bands . the relation between @xmath81 and d4000 is shown in fig . [ fig : age_ewhalpha_d4000]b . note that the concentration of points at the high age end reflects the upper age limit of the base adopted here , 13 gyr ( c.f . section [ sec : simulations ] ) . the correlation is very strong ( @xmath209 ) , showing that indeed d4000 can be used to estimate empirically mean light - weighted galaxy ages , despite its metallicity dependence for very old stellar populations ( older than 1 gyr , as shown by k03 ) . from the tight relation between @xmath81 and d4000 we derive the following empirical relation @xmath210 where @xmath3 is in yr . this robust fit reproduces @xmath115 to within an rms dispersion of 0.15 dex . [ fig : mass - sigma ] shows the relation between stellar mass and @xmath11 , obtained from the synthesis . the relation is quite good , with @xmath211 . the solid line displayed in the figure is @xmath212 for @xmath92 in @xmath213 and @xmath11 in km@xmath97s@xmath98 , obtained with a bisector fitting . the figure also shows as a dashed line a fit assuming @xmath214 , expected from the virial theorem under the ( unrealistic ) assumption of constant mass surface density . in both cases we have excluded from the fit galaxies with @xmath215 km@xmath97s@xmath98 , which corresponds to less than half the spectral resolution of both data and models . this is another relation that is expected _ a priori _ if we have in mind the faber - jackson relation for ellipticals and the tully - fisher relation for spirals . for early - type galaxies , @xmath11 is a measure of the central velocity dispersion , which is directly linked to the gravitational potential depth , and , through the virial theorem , to galactic mass . for late - type systems , @xmath11 has contributions of isotropic motions in the bulges , as well as of the rotation of the disks , and is also expected to relate with galactic mass . another aspect that it is interesting to point out in fig . [ fig : mass - sigma ] is that the dispersion in the @xmath92@xmath11 relation decreases as we go from low - luminosity , rotation - dominated systems , for which the values of @xmath11 depend on galaxy inclination and bulge - to - disk ratio , to high - luminosity , mostly early - type systems , which obey a much more regular ( and steeper ) relation between @xmath11 and @xmath92 . this relation , between a quantity that is not directly linked to the synthesis , @xmath11 , and another one that is a product of our synthesis , @xmath92 , is yet another indication that the results of our starlight code do make sense . . we have developed and tested a method to fit galaxy spectra with a combination of spectra of individual simple stellar populations generated with state - of - the art evolutionary synthesis models . the main goal of this investigation was to examine the reliability of physical properties derived in this way . this goal was pursued by three different means : simulations , comparison with independent studies , and analysis of empirical results . our main results can be summarized as follows : 1 . simulations tailored to match the characteristics of sdss spectra show that the individual ssp strengths , encoded in the population vector @xmath18 , are subjected to large uncertainties , but robust results can be obtained by compressing @xmath18 into coarser but useful indices . in particular , physically motivated indices such as mean stellar ages and metallicities are found to be well recovered by spectral synthesis even for relatively noisy spectra . stellar masses , velocity dispersion and extinction are also found to be accurately retrieved . we have applied our starlight code to a volume limited sample of over 50000 galaxies from the sdss data release 2 . the spectral fits are generally very good , and allow accurate measurements of emission lines from the starlight subtracted spectrum . catalogues of physical and emission line properties derived for this sample will be made available in due course . we also report that work is underway to produce a flexible , user friendly and publicly available version of starlight . we have compared our results to those obtained by the mpa / jhu group ( k03 ; brinchmann 2004 ) with a different method to characterize the stellar populations of sdss galaxies . the stellar extinctions and masses derived in these two studies are very strongly correlated . furthermore , differences in the values of @xmath7 and @xmath92 are found to be mostly due to the differences in the model ingredients ( extinction law ) . our estimates of stellar velocity dispersions and emission line properties are also in good agreement with those of the mpa / jhu group . the confidence in the method is further strengthened by several empirical correlations between synthesis results and independent quantities . we find strong correlations between stellar and nebular metallicites , stellar and nebular extinctions , mean stellar age and the equivalent width of h@xmath0 , mean stellar age and the 4000 break , stellar mass and velocity dispersion . these are all astrophysically reasonable results , which reinforce the conclusion that spectral synthesis is capable of producing reliable estimates of physical properties of galaxies . overall , these results validate spectral synthesis as a powerful tool to study the history of galaxies . other papers in this series will take advantage of this tool to address issues regarding aspects of galaxy formation and evolution . we thank the organizers of the guillermo haro workshop 2004 at the instituto nacional de astronomia , ptica y electrnica ( puebla , mexico ) for having allowed us to work in a very pleasant and stimulating environment and we thank the participants for many useful discussions . we are also in debt with the anonymous referee for her / his comments and helpful suggestions . partial support from cnpq , fapesp and the france - brazil pics program are also acknowledged . last but not least , we wish to thank g. bruzual , s. charlot , and the sdss team for their dedication to projects which made the present work possible . the sloan digital sky survey is a joint project of the university of chicago , fermilab , the institute for advanced study , the japan participation group , the johns hopkins university , the los alamos national laboratory , the max - planck - institute for astronomy ( mpia ) , the max - planck - institute for astrophysics ( mpa ) , new mexico state university , princeton university , the united states naval observatory , and the university of washington . funding for the project has been provided by the alfred p. sloan foundation , the participating institutions , the national aeronautics and space administration , the national science foundation , the u.s . department of energy , the japanese monbukagakusho , and the max planck society .

the temperature distribution of field li - rich red giants suggests the presence of a population of li - rich red clump ( rc ) stars . one proposed explanation for this population is that all stars with masses near 2 experience a short - lived phase of li - richness at the onset of core he - burning . many of these stars have lowc/c , a signature of deep mixing that is presumably associated with the li regeneration . to test this purported mechanism of li enrichment , we measured abundances in 38 rc stars and 6 red giant branch ( rgb ) stars in four open clusters selected to have rc masses near 2 . we find six li - rich stars ((li).50 dex ) of which only two may be rc stars . none of the rc stars have li exceeding the levels observed in the rgb stars , but given the brevity of the suggested li - rich phase and the modest sample size , it is probable that stars with larger li - enrichments were missed simply by chance . however , we find very few stars in our sample with lowc/c . such lowc/c , seen in many field li - rich stars , should persist even after lithium has returned to normal low levels . thus , if li synthesis during the he flash occurs , it is a rare , but potentially long - lived occurrence rather than a short - lived phase for all stars . we estimate a conservative upper limit of the fraction of stars going through a li - rich phase to be , based on stars that have lowc/c for their observed(li ) .

1608.01678

a small fraction of red giant stars have lithium abundances ( @xmath3(li ) ) exceeding the predictions of standard evolution models even though the majority of red giants ( rgs ) exhibit @xmath3(li ) orders of magnitude below standard model predictions @xcite . the low @xmath3(li ) demonstrates that non - convective mixing processes contribute significantly to li depletion , which makes the li - rich stars appear even more unusual . the most li - rich rgs have abundances exceeding the meteoritic abundances ( @xcite , @xcite , ) and require a nucleosynthetic origin for the li . @xcite described a pathway for li nucleosynthesis through the reactions @xmath6he(@xmath7,@xmath8)@xmath9be and @xmath9be(@xmath10,@xmath11)@xmath9li . this process requires high temperatures ( @xmath12 k ) for the first reaction to occur and fast mixing ( such as convection ) to transport the by - products to a cool region of the star ( @xmath13 k ) for lithium to be long - lived . these conditions are met at the base of the convection zones in luminous asymptotic giant branch ( agb ) stars . however , understanding li - rich stars(li ) meets or exceeds the commonly used threshold of 1.5 dex . ] found along the red giant branch ( rgb ) presents a problem because these stars convection zones are too cool to synthesize @xmath9be . thus , the @xmath6he reaction must occur below the convection zone , and a fast non - convective mixing mechanism is required to connect the convection zone to the deeper layers of the star in order to explain the enriched surface abundances . through a series of observations and advances in evolution modeling ( e.g. , , @xcite , @xcite ) , it has become largely accepted that red giants evolving through the luminosity bump stage of evolution likely experience a short - lived phase of enriched li . a number of deep mixing mechanisms for this stage of evolution have been proposed , including thermohaline mixing , magnetic buoyancy @xcite , and the hybrid magneto - thermohaline mixing @xcite . however , as the number of the relatively rare li - rich stars continues to grow , it has become clear that they are not restricted to just the luminosity bump and agb , but instead are found along the rgb . a new short - lived phase of li - richness was hypothesized by ( * ? ? ? * hereafter k11 ) , who suggested that li might be regenerated during the he - flash . this mechanism would account for the population of li - rich red giants with temperatures too warm to be luminosity bump stars and that fall in a narrow luminosity range that coincides with the red clump ( rc ) . this he - flash phenomenon would be relevant for stars in a narrow mass range of @xmath141.52.25 @xmath0 , the upper limit being defined by the maximum mass that experiences a he flash , and the lower limit by the stars that maintain a relatively large reservoir of @xmath6he . for the upper bound , showed that the transition occurs closer to 1.81.9 @xmath0 . at least one li - rich star is known to be in this mass range and is confirmed to be a he - burning star via asteroseismology @xcite . as a test of the hypothesis suggested by , we observed rc stars in four open clusters with ages and metallicities that place their rc stars in the mass range specified by . in three of the four clusters , we also observed rgs at other evolutionary stages to constrain the pre - he flash abundances . in addition to @xmath3(li ) , we also measure @xmath1c/@xmath2c . many of the li - rich field stars in have @xmath1c/@xmath2c that is much lower than predicted from standard evolution models , presumably a consequence of the same mixing that brought the synthesized li into the convection zone . although newly synthesized li can be destroyed if exposed to high enough temperatures , the altered @xmath1c/@xmath2c will persist as evidence of the past deep mixing . the selection and observation of the stars in this study are described in sections [ sec : clusters ] and [ sec : obs ] , respectively . in section [ sec : methods ] , we outline the measurement of li abundances and @xmath1c/@xmath2c . in section [ sec : discuss ] , we present the significance of our results , and we give our conclusions in section [ sec : end ] . the four clusters in this study ( collinder 110 , ngc 2204 , ngc 2506 , and ngc 6583 ) were selected so that the rc stars fell within the mass range of 1.52.25 @xmath0 , based on the predictions in . table [ tab : clusters ] lists the adopted cluster parameters ( age , dereddened distance modulus , reddening , and metallicity ) that were used to select the four clusters for study . we used @xcite isochrones to estimate the initial masses of the stars currently at the rc stage in each cluster , and our sample spans 1.62.2 @xmath0 . because of uncertainties in the cluster parameters , some of the true masses may fall outside of the narrow mass range we wish to probe . llllll collinder 110 & 1.7 & 11.82 & 0.38 & + 0.03 & 1.9 + ngc 2204 & 2.0 & 13.06 & 0.08 & @xmath15 & 1.6 + ngc 2506 & 1.99 & 12.53 & 0.04 & @xmath16 & 1.6 + ngc 6583 & 1.0 & 11.55 & 0.51 & + 0.37 & 2.2 + individual stars for each cluster were selected using color magnitude diagrams ( cmds ) , as illustrated in figure [ fig : cmds ] . the photometric data were downloaded from the webda database , and the photometric sources for each cluster can be found in tables [ tab : oc_starlist1][tab : oc_starlist4 ] . the black dots represent all of the stars in the respective photometric sources , and the large symbols represent stars observed for this study . candidate rc stars are identified by circles , and rgs at other evolutionary stages are represented by squares . ( the meaning for the different symbol colors / shading will be described later . ) cluster isochrones @xcite adopting ages from table [ tab : clusters ] and metallicities derived in section [ sec : spec_params ] are also plotted . the age of ngc 6583 is such that the red giants are at the upper edge of the critical mass range defined by and higher than the upper bound suggested by , which accounts for the lack of the luminosity bump feature in the isochrone . all of the observations for this project were carried out with the magellan inamori kyocera echelle ( mike ) spectrograph on the clay 6.5 m telescope at las campanas observatory . depending on the night s seeing conditions and other projects being observed , we used either the 0.5 slit or the 0.7 slit , which yields resolving powers of @xmath1444,000 and 31,000 , respectively . on each night , we also observed 23 radial velocity ( rv ) standard stars . we observed a th ar spectrum with each pointing toward a field star or cluster . additional th ar spectra were taken approximately every 30 minutes during extended pointings toward a given cluster . to reduce the data , we used the carnegie python pipeline @xcite . the pipeline first performs standard ccd processing tasks , such as overscan subtraction , bad pixel masking , flat fielding , and sky subtraction on the stellar spectra and calibration spectra . it extracts both a sky spectrum and stellar spectrum from each target star observation and wavelength calibrates these spectra with the th ar lamp spectra . all of the echelle orders from the red arm of the spectrograph , spanning 48009400 , were combined to a one - dimensional format . after measuring the rv ( section [ sec : rv_helio ] ) , the spectra were velocity shifted to the stellar rest frame . tables [ tab : oc_starlist1][tab : oc_starlist4 ] list the red giants observed in each cluster . nine stars were observed in collinder 110 ( table [ tab : oc_starlist1 ] ) using the 0.5 slit , and all of these stars are rc candidates . nineteen stars were observed in ngc 2204 ( table [ tab : oc_starlist2 ] ) using the 0.7 slit , of which sixteen are rc candidate stars , two are lower rgb stars , and one is near the rgb tip . one of the stars , mmu 4119 , is a suspected spectroscopic binary @xcite . in ngc 2506 ( table [ tab : oc_starlist3 ] , ) only five stars were observed , of which four are rc candidates . the fifth star is near the luminosity bump of the cluster . all stars were observed with the 0.5 slit . finally , eleven stars were observed in ngc 6583 ( table [ tab : oc_starlist4 ] ) . both slit sizes were utilized because seeing conditions changed dramatically over the course of the night . nine of the stars are rc candidates , one star is at the base of the rgb , and one star is on either the upper rgb or agb . lllrrrrrrc 1134 & 06:38:45.0 & 02:04:23.5 & 13.70 & 1.36 & 2013 jan 26 & 3600 & 113 & 0.5 & rc + 2119 & 06:38:43.3 & 02:02:18.2 & 13.52 & 1.26 & 2013 jan 27 & 4200 & 92 & 0.5 & rc + 2129 & 06:38:41.1 & 02:01:05.3 & 13.66 & 1.34 & 2013 jan 26 & 3240 & 118 & 0.5 & rc + 2223 & 06:39:03.5 & 01:59:19.8 & 13.48 & 1.28 & 2013 jan 27 & 2820 & 92 & 0.5 & rc + 3122 & 06:38:34.7 & 02:01:41.0 & 13.46 & 1.38 & 2014 mar 19 & 4050 & 105 & 0.5 & rc / rgb + 3144 & 06:38:30.3 & 02:03:03.0 & 13.49 & 1.31 & 2014 mar 20 & 2000 & 78 & 0.5 & rc + 3244 & 06:38:16.0 & 02:02:24.3 & 13.74 & 1.34 & 2013 jan 26 & 2640 & 104 & 0.5 & rc + 4260 & 06:38:32.3 & 02:07:24.7 & 13.87 & 1.37 & 2013 jan 27 & 2400 & 77 & 0.5 & rc + 5125 & 06:38:40.2 & 02:01:38.5 & 13.82 & 1.36 & 2013 jan 27 & 3200 & 83 & 0.5 & rc + lllrrrrrrc 1124 & 06:15:29.0 & @xmath1718:39:09.8 & 13.84 & 1.00 & 2014 jan 06 & 1800 & 94 & 0.7 & rc + 1212 & 06:15:20.1 & @xmath1718:37:57.8 & 13.88 & 0.98 & 2014 jan 07 & 1800 & 121 & 0.7 & rc + 1330 & 06:15:26.7 & @xmath1718:33:25.1 & 13.76 & 1.04 & 2014 jan 07 & 1400 & 113 & 0.7 & rc + 2212 & 06:15:49.7 & @xmath1718:37:39.4 & 12.82 & 1.24 & 2014 jan 06 & 800 & 122 & 0.7 & agb + 2229 & 06:15:36.9 & @xmath1718:36:08.9 & 13.83 & 1.01 & 2014 jan 07 & 1400 & 106 & 0.7 & rc + 2311 & 06:16:02.1 & @xmath1718:38:46.5 & 13.64 & 1.09 & 2014 jan 07 & 1200 & 109 & 0.7 & rc + 2330 & 06:15:34.4 & @xmath1718:35:03.1 & 14.48 & 1.00 & 2014 jan 07 & 2600 & 114 & 0.7 & rgb + 3205 & 06:15:46.0 & @xmath1718:40:44.0 & 13.91 & 0.98 & 2014 jan 07 & 1400 & 109 & 0.7 & rc + 3215 & 06:15:45.3 & @xmath1718:43:34.9 & 13.75 & 1.00 & 2014 jan 06 & 1400 & 114 & 0.7 & rc + 3321 & 06:15:43.3 & @xmath1718:46:20.1 & 13.83 & 0.99 & 2014 jan 06 & 2000 & 116 & 0.7 & rc + 4115 & 06:15:27.9 & @xmath1718:40:34.0 & 14.15 & 1.02 & 2014 jan 07 & 1680 & 108 & 0.7 & rgb + 4116 & 06:15:27.5 & @xmath1718:40:13.9 & 13.93 & 1.06 & 2014 jan 06 & 1600 & 117 & 0.7 & rc / rgb + 4119 & 06:15:27.2 & @xmath1718:40:44.3 & 13.69 & 1.00 & 2014 jan 07 & 1200 & 104 & 0.7 & rc / rgb + 4211 & 06:15:13.6 & @xmath1718:41:49.3 & 13.68 & 0.98 & 2014 jan 06 & 1200 & 103 & 0.7 & rc + 4223 & 06:15:32.9 & @xmath1718:43:11.9 & 13.89 & 1.03 & 2014 jan 06 & 1840 & 113 & 0.7 & rc + 4303 & 06:15:03.9 & @xmath1718:41:07.6 & 13.90 & 0.96 & 2014 jan 06 & 1400 & 100 & 0.7 & rc + 5352 & 06:15:50.8 & @xmath1718:34:19.4 & 13.91 & 1.02 & 2014 jan 06 & 1800 & 102 & 0.7 & rc + 5980 & 06:15:33.7 & @xmath1718:42:12.1 & 13.89 & 0.98 & 2014 jan 06 & 1800 & 112 & 0.7 & rc + 6330 & 06:15:26.1 & @xmath1718:31:48.8 & 13.91 & 0.98 & 2014 jan 06 & 1800 & 114 & 0.7 & rc + lllrrrrrrc 2380 & 08:00:09.2 & @xmath1710:49:09.0 & 13.14 & 0.98 & 2013 jan 27 & 1680 & 68 & 0.5 & rc + 3265 & 07:59:50.8 & @xmath1710:46:40.0 & 13.15 & 1.07 & 2013 jan 27 & 1320 & 52 & 0.5 & rgb / rc ? + 4138 & 08:00:01.4 & @xmath1710:45:39.3 & 13.31 & 0.91 & 2013 jan 27 & 2220 & 78 & 0.5 & rc + 4205 & 07:59:51.3 & @xmath1710:46:17.4 & 13.24 & 0.96 & 2013 jan 26 & 1680 & 96 & 0.5 & rc + 4240 & 07:59:52.6 & @xmath1710:44:50.0 & 13.13 & 0.93 & 2013 jan 26 & 1620 & 94 & 0.5 & rc + lllrrrrrrc 10 & 18:15:51.1 & @xmath1722:07:15.2 & 12.94 & 1.84 & 2014 jun 19 & 840 & 112 & 0.5 & agb / rgb + 33 & 18:15:52.6 & @xmath1722:09:53.2 & 14.02 & 1.60 & 2014 jun 19 & 2280 & 109 & 0.5 & rc + 34 & 18:15:43.9 & @xmath1722:09:00.9 & 13.96 & 1.59 & 2014 jun 19 & 1920 & 100 & 0.5 & rc + 38 & 18:15:48.2 & @xmath1722:09:53.2 & 14.05 & 1.59 & 2014 jun 19 & 1600 & 129 & 0.7 & rc + 39 & 18:15:56.6 & @xmath1722:07:32.4 & 14.07 & 1.62 & 2014 jun 19 & 1800 & 119 & 0.7 & rc + 42 & 18:15:50.2 & @xmath1722:09:55.5 & 14.12 & 1.70 & 2014 jun 19 & 900 & 70 & 0.7 & rc + 46 & 18:15:51.2 & @xmath1722:07:26.6 & 14.10 & 1.58 & 2014 jun 19 & 2400 & 96 & 0.5 & rc + 50 & 18:15:48.3 & @xmath1722:09:33.5 & 14.15 & 1.58 & 2014 jun 19 & 1800 & 106 & 0.7 & rc / rgb ? + 62 & 18:15:51.2 & @xmath1722:08:28.1 & 14.27 & 1.65 & 2014 jun 19 & 2000 & 109 & 0.7 & rc + 72 & 18:15:54.3 & @xmath1722:08:05.7 & 14.36 & 1.63 & 2014 jun 19 & 2200 & 98 & 0.7 & rc / rgb ? + 92 & 18:15:51.7 & @xmath1722:08:30.6 & 14.55 & 1.60 & 2014 jun 19 & 3800 & 107 & 0.5 & rgb + heliocentric radial velocities ( @xmath18 ) were measured for each star by cross - correlating the 1d spectra with the high - resolution arcturus atlas spectrum @xcite with the iraf procedure _ fxcor_. the spectra were also cross - correlated with the @xcite telluric spectrum in the wavelength ranges of 64756675 , 69807060 , and 78508125 to measure slit centering errors , since the seeing disk was at times smaller than the slit width . we found rv offsets in the telluric line centers as high as 2.4 km@xmath19s@xmath20 . the final @xmath18 measurements reported here are @xmath21 , where @xmath22 is the relative rv of the object spectra with respect to the arcturus spectrum ( which is in the stellar rest frame ) , @xmath23 is the heliocentric rv correction , and @xmath24 is the relative rvs of the telluric lines in the observed spectra with respect to the atlas telluric spectrum . using the corrections from the telluric line cross - correlation better reproduced the @xmath18 of the rv standard stars that we observed . the average cluster velocities were calculated using an iterative sigma - clipping algorithm , clipping at @xmath25 . the results are given in table [ tab : oc_rv ] , along with literature measurements . two stars in ngc 2204 were clipped : 2330 with @xmath26 km@xmath19s@xmath20 and 4119 with @xmath27 km@xmath19s@xmath20 . the latter is a suspected spectroscopic binary star that is still a cluster member @xcite . we maintain star 2330 in our analysis , but caution that it is a potential non - member based on its rv . if it is a member , it is on the lower rgb . individual @xmath18 are listed in the final column of table [ tab : stell_param ] . the uncertainty in the @xmath18 is dominated by systematic errors . the random errors are all @xmath28 km@xmath19s@xmath20 with an average 0.04 km@xmath19s@xmath20 . comparing the @xmath18 of five rv standard stars ( hd 107328 , hd 171391 , hd 66141 , hd 26162 , and hd 203638 ) measured on multiple nights to the accepted values in `` the astronomical almanac '' @xcite , we estimate a systematic uncertainty of 0.20.3 km@xmath19s@xmath20 . lrrr|rrrr collinder 110 & @xmath29 & @xmath30 & 1 & @xmath31 & 0.04 & + 0.03 & 1 + & & @xmath32 & 2 & & & & + ngc 2204 & @xmath33 & @xmath34 & 3 & @xmath35 & 0.04 & @xmath15 & 3 + & & @xmath36 & 4 & & & @xmath37 & 7 + ngc 2506 & @xmath38 & @xmath39 & 4 , 5 & @xmath40 & 0.09 & @xmath16 & 4 + & & & & & & @xmath41 & 7 + & & & & & & @xmath42 & 8 , 9 + ngc 6583 & @xmath43 & @xmath44 & 6 & @xmath45 & 0.05 & @xmath46 & 6 + although it is common in open cluster studies to use photometric temperatures and isochrones to constrain surface gravities , we opted to do a full spectroscopic derivation of the stellar parameters . this provides a more homogenous determination of stellar @xmath47 and @xmath48 across our sample . we used the iron line list of @xcite , which was compiled for deriving stellar parameters of red giant stars . equivalent widths ( eqws ) were all measured by hand using iraf s _ splot _ routine , using gaussian profiles and de - blending when necessary . the high - resolution arcturus atlas @xcite was used as a template to identify blended lines . we used the 2014 version of the stellar line analysis program , moog @xcite , and marcs spherical atmosphere models @xcite to compute abundances . to solve for the stellar @xmath47 , @xmath48 , [ fe / h ] , and @xmath49 , we used the standard requirements that there should be no trend of the output @xmath3 ( ) with either the excitation potential or strength of the iron lines and that @xmath3 ( ) should equal @xmath3 ( ) . the 2014 version of moog adopts @xmath3(fe)@xmath50 , while the marcs models adopt @xmath3(fe)@xmath51 . however , the difference in the atmosphere structure of two models that differ in [ m / h ] by 0.05 dex is too small to affect our abundances . the @xcite line list yields a slightly higher solar abundance of @xmath3(fe)@xmath52 , which we adopt to translate our stellar @xmath3(fe ) to [ fe / h ] . in table [ tab : stell_param ] , we give the results of our stellar parameter analysis and the uncertainties in each parameter . the uncertainties in @xmath49 are the variations that arise when the slope of @xmath3 ( ) vs. eqw/@xmath53 is varied within the 1@xmath54 uncertainty of the fitted slope . similarly , the uncertainty in @xmath47 uses the uncertainty in the slope of @xmath3 ( ) vs. excitation potential . it also includes the contribution from the @xmath49 uncertainty . the uncertainties in [ fe / h ] and @xmath48 are the standard deviations in the and lines , respectively . these dominate over the uncertainty arising from sensitivities to the other stellar parameters . lrrrrrrrrrrrrr collinder 110 & 1134 & 4960 . & & 2.80 & 0.09 & @xmath55 & 0.10 & 1.47 & 0.06 & 1.0 & 0.0 & 4.99 & + 37.2 + collinder 110 & 2119 & 4970 . & 64 . & 2.80 & 0.14 & @xmath56 & 0.10 & 1.14 & 0.06 & 5.9 & 0.2 & 5.23 & + 35.6 + collinder 110 & 2129 & 4940 . & 59 . & 2.70 & 0.09 & @xmath57 & 0.10 & 1.35 & 0.05 & 1.1 & 0.0 & 4.98 & + 38.0 + collinder 110 & 2223 & 4990 . & 69 . & 2.80 & 0.07 & @xmath58 & 0.12 & 1.43 & 0.06 & 1.0 & 0.0 & 5.34 & + 36.2 + collinder 110 & 3122 & 4800 . & 80 . & 2.65 & 0.10 & @xmath59 & 0.13 & 1.33 & 0.07 & 1.0 & 0.0 & 4.81 & + 39.2 + collinder 110 & 3144 & 4870 . & 72 . & 2.55 & 0.12 & @xmath58 & 0.12 & 1.37 & 0.06 & 1.4 & 0.0 & 4.97 & + 37.2 + collinder 110 & 3244 & 4960 . & 62 . & 2.85 & 0.09 & @xmath60 & 0.10 & 1.39 & 0.06 & 1.0 & 0.0 & 4.95 & + 38.0 + collinder 110 & 4260 & 4980 . & 64 . & 2.75 & 0.13 & @xmath59 & 0.11 & 1.19 & 0.06 & 1.0 & 0.0 & 4.85 & + 36.7 + collinder 110 & 5125 & 5130 . & 72 . & 3.05 & 0.11 & + 0.00 & 0.12 & 1.45 & 0.07 & 1.0 & 0.0 & 5.39 & + 34.6 + ngc 2204 & 1124 & 5030 . & 56 . & 2.65 & 0.13 & @xmath42 & 0.10 & 1.44 & 0.06 & 2.7 & 0.0 & 5.43 & + 91.2 + ngc 2204 & 1212 & 5080 . & 55 . & 2.75 & 0.08 & @xmath35 & 0.10 & 1.28 & 0.05 & 2.5 & 0.0 & 5.54 & + 88.1 + ngc 2204 & 1330 & 5120 . & 71 . & 2.85 & 0.10 & @xmath61 & 0.12 & 1.43 & 0.07 & 1.0 & 0.0 & 5.83 & + 88.9 + ngc 2204 & 2212 & 4570 . & 2.00 & 0.10 & @xmath62 & 0.10 & 1.55 & 0.05 & 1.9 & 0.0 & 5.21 & + 91.7 + ngc 2204 & 2229 & 5060 . & 52 . & 2.70 & 0.14 & @xmath61 & 0.09 & 1.35 & 0.05 & 1.5 & 0.0 & 5.54 & + 91.3 + ngc 2204 & 2311 & 4950 . & 61 . & 2.55 & 0.13 & @xmath63 & 0.11 & 1.46 & 0.06 & 1.5 & 0.0 & 5.40 & + 91.8 + ngc 2204 & 2330 & 5080 . & 59 . & 3.15 & 0.10 & @xmath64 & 0.10 & 1.23 & 0.06 & 1.0 & 0.0 & 4.83 & + 96.5 + ngc 2204 & 3205 & 5070 . & 72 . & 2.85 & 0.07 & @xmath35 & 0.12 & 1.38 & 0.07 & 1.0 & 0.0 & 5.47 & + 93.3 + ngc 2204 & 3215 & 5010 . & 2.75 & 0.06 & @xmath64 & 0.10 & 1.42 & 0.06 & 1.0 & 0.0 & 5.45 & + 91.5 + ngc 2204 & 3321 & 5100 . & 58 . & 2.80 & 0.11 & @xmath65 & 0.10 & 1.32 & 0.06 & 2.3 & 0.0 & 5.67 & + 91.3 + ngc 2204 & 4115 & 4880 . & 63 . & 2.70 & 0.09 & @xmath35 & 0.10 & 1.28 & 0.06 & 2.2 & 0.0 & 4.62 & + 91.5 + ngc 2204 & 4116 & 4860 . & 2.65 & 0.11 & @xmath63 & 0.11 & 1.35 & 0.06 & 1.0 & 0.0 & 4.80 & + 92.9 + ngc 2204 & 4119 & 4820 . & 71 . & 2.65 & 0.14 & @xmath42 & 0.11 & 1.27 & 0.06 & 1.0 & 0.0 & 4.96 & + 76.8 + ngc 2204 & 4211 & 5030 . & 2.70 & 0.10 & @xmath66 & 0.11 & 1.41 & 0.06 & 1.0 & 0.0 & 5.63 & + 90.7 + ngc 2204 & 4223 & 5260 . & 101 . & 3.06 & 0.11 & @xmath65 & 0.17 & 1.62 & 0.10 & 1.3 & 0.0 & 6.15 & + 91.0 + ngc 2204 & 4303 & 5010 . & 71 . & 2.60 & 0.10 & @xmath63 & 0.12 & 1.47 & 0.07 & 1.0 & 0.0 & 5.27 & + 91.3 + ngc 2204 & 5352 & 5010 . & 68 . & 2.80 & 0.09 & @xmath65 & 0.11 & 1.36 & 0.06 & 1.0 & 0.0 & 5.25 & + 90.6 + ngc 2204 & 5980 & 5050 . & 63 . & 2.70 & 0.10 & @xmath15 & 0.11 & 1.46 & 0.06 & 1.8 & 0.0 & 5.43 & + 91.0 + ngc 2204 & 6330 & 5000 . & 63 . & 2.70 & 0.06 & @xmath40 & 0.11 & 1.42 & 0.06 & 1.0 & 0.0 & 5.22 & + 90.9 + ngc 2506 & 2380 & 4980 . & 65 . & 2.55 & 0.09 & @xmath37 & 0.11 & 1.45 & 0.06 & 1.0 & 0.0 & 5.31 & + 82.8 + ngc 2506 & 3265 & 5250 . & 3.05 & 0.15 & @xmath59 & 0.17 & 1.52 & 0.10 & 1.0 & 0.0 & 6.24 & + 80.3 + ngc 2506 & 4138 & 5030 . & 64 . & 2.70 & 0.09 & @xmath67 & 0.11 & 1.45 & 0.06 & 1.0 & 0.0 & 5.27 & + 82.9 + ngc 2506 & 4205 & 5050 . & 61 . & 2.80 & 0.08 & @xmath40 & 0.10 & 1.41 & 0.06 & 1.0 & 0.0 & 5.42 & + 83.7 + ngc 2506 & 4240 & 5000 . & 66 . & 2.70 & 0.09 & @xmath62 & 0.11 & 1.43 & 0.06 & 1.0 & 0.0 & 5.38 & + 82.4 + ngc 6583 & 10 & 4350 . & 103 . & 2.10 & 0.28 & + 0.05 & 0.15 & 1.61 & 0.09 & 2.3 & 0.6 & 4.26 & @xmath68 + ngc 6583 & 33 & 4870 . & 81 . & 2.80 & 0.12 & + 0.19 & 0.13 & 1.60 & 0.07 & 1.5 & 0.0 & 4.54 & @xmath69 + ngc 6583 & 34 & 4860 . & 73 . & 2.65 & 0.11 & + 0.16 & 0.12 & 1.60 & 0.06 & 1.9 & 0.0 & 4.58 & @xmath70 + ngc 6583 & 38 & 4900 . & 75 . & 2.75 & 0.16 & + 0.18 & 0.12 & 1.54 & 0.07 & 2.1 & 0.0 & 4.59 & @xmath71 + ngc 6583 & 39 & 4830 . & 76 . & 2.50 & 0.12 & + 0.15 & 0.12 & 1.64 & 0.06 & 5.0 & 0.0 & 4.38 & @xmath72 + ngc 6583 & 42 & 4840 . & 90 . & 2.60 & 0.23 & + 0.18 & 0.14 & 1.68 & 0.08 & 3.3 & 0.0 & 4.36 & @xmath73 + ngc 6583 & 46 & 4910 . & 77 . & 2.80 & 0.13 & + 0.17 & 0.12 & 1.56 & 0.07 & 1.5 & 0.0 & 4.56 & @xmath73 + ngc 6583 & 50 & 4970 . & 85 . & 3.00 & 0.16 & + 0.27 & 0.13 & 1.64 & 0.08 & 2.5 & 0.0 & 4.67 & @xmath74 + ngc 6583 & 62 & 4910 . & 84 . & 2.80 & 0.15 & + 0.20 & 0.13 & 1.61 & 0.07 & 2.0 & 0.0 & 4.39 & @xmath75 + ngc 6583 & 72 & 4940 . & 92 . & 2.95 & 0.16 & + 0.16 & 0.14 & 1.55 & 0.08 & 5.0 & 0.2 & 4.37 & @xmath70 + ngc 6583 & 92 & 4980 . & 71 . & 2.95 & 0.14 & + 0.19 & 0.11 & 1.42 & 0.06 & 1.1 & 0.0 & 4.28 & @xmath76 + in figure [ fig : cmds2 ] , we find overall good agreement between our derived @xmath47 and @xmath48 and the values predicted from isochrones . however , the classification of objects as either candidate rc stars ( circles ) or candidate non - rc stars ( squares ) from the photometry ( figure [ fig : cmds ] ) do not necessarily persist when plotted in this phase space . our rc stars tend to scatter along a diagonal line in @xmath47@xmath48 but at a given temperature , the clump and first ascent stars appear to cleanly separate in @xmath48 , as is especially apparent in ngc 2204 . in that cluster , two of the rc candidate stars appear to be first ascent stars just below the luminosity bump . the same is true of one of the rc candidates in collinder 110 . these stars have revised classifications of ` rgb ' in tables [ tab : oc_starlist1 ] and [ tab : oc_starlist2 ] . all of the clump candidates in ngc 2506 appear to be bona fide rc stars . the picture in ngc 6583 is still murky . the two hottest rc candidates may be first ascent stars , and their revised classifications in table [ tab : oc_starlist4 ] is ` rgb ? . ' in the discussion that follows , ` candidate rc ' stars refers to the stars that have initial classifications of ` rc ; ' their revised classification may be different . comparing figures [ fig : cmds ] and [ fig : cmds2 ] , we were surprised to find that the single non - rc star that we observed in ngc 2506 has the hottest spectroscopic temperature ( 5250 k ) , even though it has the reddest @xmath77 of the stars we observed . our spectroscopic measurements place ngc 2506 3265 somewhat off of the isochrone , but we note that this star has the lowest signal - to - noise ratio ( s / n ) in our entire sample ( only 68 ) . using the color temperature relations of @xcite , the @xmath78 color yields @xmath79 k. to test the source of this discrepancy , we computed color temperatures from two independent photometric sources in the literature . optical @xmath80 photometry from @xcite yields a hotter temperature of @xmath81 k , while 2mass @xmath82 ( using @xcite color transformation ) yields an even higher @xmath83 k. this latter value is most consistent with the spectroscopic temperature but is still 200 k cooler . all of the literature photometry of ngc 2506 3265 is consistent with the rc and bump magnitude of the isochrones . the optical colors correspond to the luminosity bump , while the infrared color prefers the rc . thus , this star s evolutionary stage remains ambiguous . it is also possible that the star is a non - member considering that both its @xmath18 and [ fe / h ] appear to be deviant , although neither parameter is sufficiently deviant compared to the errors to confidently confirm its non - membership . comparing our @xmath18 measurement ( @xmath84 km@xmath19s@xmath20 ) to the literature suggests that the star may in fact be a binary . @xcite finds @xmath85 while @xcite find @xmath86 km@xmath19s@xmath20 . + in table [ tab : oc_rv ] and figure [ fig : compfeh ] , we compare our average cluster metallicities to those in the literature . most of the literature sources are those listed in table [ tab : clusters ] . for the two most metal - poor clusters , we get excellent agreement with at least one literature source , although other literature sources disagree outside the uncertainties . for the more metal - rich clusters , we find lower metallicities than the literature studies do . to test whether the cause of this discrepancy arises from our stellar parameter derivation , we also compute the metallicities using the same eqw measurements but using @xmath47 derived from the dereddened colors and magnitudes of the stars and @xmath48 derived from the stellar masses estimated from the isochrones . the bottom panel of figure [ fig : compfeh ] shows that comparison . the photometric cluster metallicities are in good agreement with the spectroscopic ones but are systematically lower . another possible cause of the discrepancy may be in the eqw measurements . the presence of many , weak unresolved metal lines could lead to systematic differences in the continuum placement . fortuitously , there are two stars each in collinder 110 and ngc 6583 with literature eqw abundances . @xcite measured abundances of collinder 110 stars 2129 and 3144 , and there are 29 iron lines overlapping our studies . similarly , @xcite measured abundances of the ngc 6583 stars 46 and 62 with 22 lines overlapping our studies . figure [ fig : compeqw ] shows that our measurements agree quite well , and there is no systematic underestimation of eqws in this study . the mean and standard deviation of @xmath87eqw are similar in size to that attained by @xcite , when they compared their results to earlier studies . the projected rotational velocities ( @xmath88 ) were measured by fitting the profiles of up to six relatively isolated fe lines in our line list ( 5307.36 , 5638.26 , 5855.08 , 6151.62 , 6165.36 , and 6750.15 ) . in addition to the rotation , there are two other major sources of broadening that must be accounted for : the line spread function ( or instrumental broadening ) and the macroturbulent broadening ( @xmath89 ) . the former can be measured from the th ar lines appearing in the same echelle order as the fe lines . discriminating uniquely between @xmath88 from @xmath89 is difficult ( if not impossible ) at the spectral resolution and s / n of our data . instead , we estimate @xmath89 using previously defined relationships between @xmath89 and other standard stellar parameters . we tested two different prescriptions : the @xmath89(@xmath47 , lc ) relationships of , where ` lc ' refers to the luminosity class , and the @xmath89(@xmath90 @xmath47 , @xmath91 ) relationships of @xcite . the former prescription tended to give larger @xmath89 by @xmath140.52.5 km@xmath19s@xmath20 compared to the latter . we adopted the @xcite prescription since we have fairly accurate luminosities for our stars . we then generated synthetic spectra of varying @xmath88 for each of the six spectra lines , and used a @xmath92 minimization to select the best @xmath88 . the results of the broadening analysis ( @xmath88 , @xmath93 , and @xmath89 ) are given in columns 1113 of table [ tab : stell_param ] . the majority of the rgs in this study are very slow rotators , where the instrumental and macroturbulent broadening fully account for the line profile , and in some cases over accounts for it . upper limits are provided for these stars , which are indicated with @xmath94 km@xmath19s@xmath20 . only three stars have @xmath88 that are not upper limits . the star with the largest rotation , collinder 110 2119 , has @xmath88@xmath95 km@xmath19s@xmath20 . its @xmath88 was also measured by @xcite via cross - correlation using high - resolution , low s / n data and was found to be @xmath96 km@xmath19s@xmath20 , consistent with this result . the remaining collinder 110 stars overlapping that study were all upper limits consistent with what is found here . @xmath3(li ) were measured by fitting spectra synthesized with moog to the observed data between 6706.7 and 6708.4 using the line list from @xcite . the spectra were fit by hand , making small adjustments to the overall continuum level , velocity scale , and broadening to get a good match between the observed and synthetic spectra . the resonance lines are blended with a line and cn lines at the resolution of our spectra . we made adjustments to the abundances of these elements first before adjusting @xmath3(li ) . our synthesis fits assume local thermodynamic equilibrium ( lte ) , and we correct for nlte effects by interpolating the grid of corrections to each stars observed parameters . the results of the synthesis fitting are given in table [ tab : stell_abun ] . the eqw of the li feature corresponding to the synthesis measurement or limit is also provided in the table . for some spectra , the `` best fit '' was still not a great match to the data , and we therefore provide a reliability parameter to flag the cases where the fits were more uncertain . as an example , ngc 2506 is the most metal - poor cluster in the sample and , because of observing conditions , the average s / n achieved for these stars is lower than all of the other clusters . therefore , the spectra had both the weakest features to fit and the largest noise . the stars with a quality parameter of ` a ' are good fits , whereas ` b ' indicates more uncertain fits . the @xmath1c/@xmath2c ratio was measured by fitting the spectral region between 8001 and 8006 , containing four @xmath1cn features ( one blended with an line ) and one @xmath2cn feature . we used the line list from @xcite to generate synthetic spectra . a linear correction to the continuum is fit , and the observed spectrum is cross - correlated with the synthetic spectrum to correct for small mismatches between the velocity scales . we keep the ratio of c / n fixed at 1.5 but allow the total abundance of the two elements to vary to fit the @xmath1cn features . then we vary the @xmath1c/@xmath2c to find the best fit to the @xmath2cn features . telluric features are abundant in this part of the spectrum . in most cases the telluric lines are weak , and we account for their presence by adding in their contribution using the @xcite atlas telluric spectrum , broadened to our instrumental resolution and scaled to fit the depth of the features . there is one strong telluric feature near 8007.5 that caused significant problems . ngc 2204 was observed at an unfortunate rv that placed the @xmath2cn lines at nearly the same wavelength as this strong feature . for ngc 2506 , the blue wing of this strong telluric feature is blended with the @xmath2cn line . for the stars in these two clusters , we corrected the spectra using the telluric standard stars that were observed on each night . these standards were observed at airmasses that bracketed the airmass of the cluster stars . we divided each of the cluster stellar spectra with the hot star spectrum that was closest in airmass , adjusting the velocity and overall scaling of the hot star spectrum to get the cleanest division . in seven cases , the @xmath2cn feature was not recoverable . for the ngc 2204 stars where the feature was recovered , it should be noted that the telluric feature was always stronger that the underlying stellar feature , and uncertainties in the telluric division may be large . therefore , all of the @xmath1c/@xmath2c measurements for ngc 2204 should be used with caution . the results of the @xmath1c/@xmath2c analysis are given in table [ tab : stell_abun ] , which includes a quality parameter . stars with measurements have a quality parameter ranging from ` a ' through ` d , ' which corresponds to decreasing quality of the fits . those with quality of ` x ' indicates that @xmath1c/@xmath2c was not measurable . the two lowest quality measurements ( ` d ' and ` x ' ) apply only to the stars that had the hot star correction to the telluric features . lrrrrrrrrrr collinder 110 & 1134 & @xmath97 & @xmath98 & @xmath99 & 0.15 & a & 17 & 5 & a + collinder 110 & 2119 & @xmath100 & @xmath101 & @xmath102 & 0.16 & a & 22 & 5 & b + collinder 110 & 2129 & @xmath103 & @xmath104 & @xmath105 & 0.00 & a & 12 & 5 & a + collinder 110 & 2223 & @xmath106 & @xmath107 & @xmath104 & 0.00 & a & 15 & 5 & c + collinder 110 & 3122 & @xmath108 & @xmath109 & @xmath110 & 0.22 & a & 18 & 5 & b + collinder 110 & 3144 & @xmath111 & @xmath101 & @xmath112 & 0.00 & b & 10 & 5 & c + collinder 110 & 3244 & @xmath113 & @xmath114 & @xmath115 & 0.15 & a & 16 & 5 & b + collinder 110 & 4260 & @xmath116 & @xmath117 & @xmath118 & 0.00 & a & 17 & 5 & b + collinder 110 & 5125 & @xmath119 & @xmath120 & @xmath121 & 0.00 & a & 20 & 5 & b + ngc 2204 & 1124 & @xmath122 & @xmath123 & @xmath124 & 0.00 & b & & & x + ngc 2204 & 1212 & @xmath125 & @xmath126 & @xmath127 & 0.13 & a & & & x + ngc 2204 & 1330 & @xmath128 & @xmath129 & @xmath130 & 0.00 & a & 16 & 5 & d + ngc 2204 & 2212 & @xmath131 & @xmath132 & @xmath133 & 0.00 & a & 12 & 5 & d + ngc 2204 & 2229 & @xmath103 & @xmath118 & @xmath98 & 0.00 & b & 12 & 5 & d + ngc 2204 & 2311 & @xmath134 & @xmath135 & @xmath136 & 0.00 & a & 12 & 5 & c + ngc 2204 & 2330 & @xmath137 & @xmath138 & @xmath139 & 0.14 & a & 12 & 5 & c + ngc 2204 & 3205 & @xmath140 & @xmath141 & @xmath142 & 0.17 & a & & & x + ngc 2204 & 3215 & @xmath143 & @xmath126 & @xmath144 & 0.15 & b & 8 & 5 & d + ngc 2204 & 3321 & @xmath145 & @xmath144 & @xmath146 & 0.13 & a & 9 & 5 & d + ngc 2204 & 4115 & @xmath147 & @xmath148 & @xmath149 & 0.16 & a & 27 & 5 & c + ngc 2204 & 4116 & @xmath150 & @xmath151 & @xmath138 & 0.16 & a & & & x + ngc 2204 & 4119 & @xmath152 & @xmath153 & @xmath154 & 0.19 & a & 16 & 5 & c + ngc 2204 & 4211 & @xmath155 & @xmath156 & @xmath157 & 0.16 & a & & & x + ngc 2204 & 4223 & @xmath158 & @xmath159 & @xmath160 & 0.00 & a & 15 & 5 & d + ngc 2204 & 4303 & @xmath161 & @xmath162 & @xmath163 & 0.00 & b & 10 & 5 & d + ngc 2204 & 5352 & @xmath164 & @xmath165 & @xmath166 & 0.17 & b & & & x + ngc 2204 & 5980 & @xmath167 & @xmath168 & @xmath169 & 0.00 & b & 18 & 0 & d + ngc 2204 & 6330 & @xmath170 & @xmath98 & @xmath171 & 0.15 & b & 13 & 5 & d + ngc 2506 & 2380 & @xmath155 & @xmath172 & @xmath126 & 0.00 & b & & & x + ngc 2506 & 3265 & @xmath173 & @xmath165 & @xmath174 & 0.00 & b & 21 & 5 & b + ngc 2506 & 4138 & @xmath175 & @xmath176 & @xmath177 & 0.15 & b & 10 & 0 & c + ngc 2506 & 4205 & @xmath178 & @xmath179 & @xmath157 & 0.14 & b & 10 & 5 & d + ngc 2506 & 4240 & @xmath180 & @xmath181 & @xmath182 & 0.00 & b & 9 & 5 & c + ngc 6583 & 10 & @xmath183 & @xmath184 & @xmath185 & 0.33 & a & 24 & 5 & a + ngc 6583 & 33 & @xmath186 & @xmath112 & @xmath187 & 0.22 & a & 21 & 5 & a + ngc 6583 & 34 & @xmath188 & @xmath146 & @xmath189 & 0.20 & b & 23 & 5 & a + ngc 6583 & 38 & @xmath190 & @xmath191 & @xmath192 & 0.20 & a & 22 & 5 & a + ngc 6583 & 39 & @xmath193 & @xmath31 & @xmath194 & 0.00 & a & 21 & 5 & a + ngc 6583 & 42 & @xmath195 & @xmath112 & @xmath189 & 0.00 & b & 19 & 5 & a + ngc 6583 & 46 & @xmath196 & @xmath176 & @xmath197 & 0.20 & a & 23 & 5 & b + ngc 6583 & 50 & @xmath198 & @xmath199 & @xmath151 & 0.22 & a & 22 & 5 & a + ngc 6583 & 62 & @xmath200 & @xmath201 & @xmath202 & 0.22 & a & 23 & 5 & a + ngc 6583 & 72 & @xmath203 & @xmath149 & @xmath204 & 0.24 & a & 31 & 5 & c + ngc 6583 & 92 & @xmath205 & @xmath206 & @xmath139 & 0.18 & a & 27 & 5 & a + in figure [ fig : li_histo ] , we show the distribution of @xmath3(li)@xmath207 for the four clusters in this study , with histograms for both the full observed sample and the subset of rc candidate stars . the non - rc stars tend to fall at either extremes of the distributions , but are more commonly at the high extreme . for three of the clusters , the majority of the rc candidates fall into a single distribution that peaks around @xmath208 dex and is composed of mostly upper limit measurements . two of the those clusters ( collinder 110 and ngc 2204 ) have rc candidates in a second group or tail that extends into the li - rich regime . in contrast , the fourth cluster ( ngc 6583 , both the most - metal rich and youngest cluster in this study ) shows a peak in the distribution at much higher @xmath3(li ) with very few upper limit measurements . only a single rc star has significantly lower @xmath3(li ) . ( li ) for all stars observed in each cluster ( shaded gray histograms ) and the subset of candidate rc stars ( open blue histograms ) . the hatched regions denote stars with only upper limits on @xmath3(li ) in the full sample ( diagonal lines ) and the rc sample ( horizontal lines ) . [ fig : li_histo],scaledwidth=50.0% ] these bimodal - like lithium distributions are similar to three clusters with @xmath3(li ) dichotomies reported previously in the literature for ngc 752 , ngc 3680 , and ic 4651 ( @xcite , , ) . all three clusters have similar ages ( 1.51.9 gyr ) and are older than two of our clusters ( ngc 6583 and collinder 110 ) but younger than the other two ( ngc 2204 and ngc 2506 ) . the dichotomy of @xmath3(li ) in these three literature clusters has been interpreted as discriminating between first ascent rgs and rc stars in each cluster , with the higher @xmath3(li ) corresponding to the former , and lower @xmath3(li ) to the latter . this interpretation is also borne out in collinder 110 and ngc 2204 , where it is clear from figure [ fig : cmds2 ] that the highest li stars ( filled symbols ) are the ones that appear to be first ascent rgb stars and not rc stars . however , because rc should be more common than first ascent rgs , one would expect all of the clusters to have more low @xmath3(li ) stars than high @xmath3(li ) stars . this is not the case for ngc 3680 ( or ngc 6583 in this work ) , which led to suggest that the high @xmath3(li ) stars in ngc 3680 were in fact the rc stars . an expanded sample of giants in ngc 3680 by @xcite confirmed this li dichotomy . ngc 3680 is the oldest and most metal - poor ( [ fe / h]@xmath209 dex ) of those three literature clusters , but ngc 6583 is the youngest and most metal - rich of the clusters studied here . since we have previously estimated evolutionary stages for our cluster stars , we can explore the li distribution of each cluster in more detail . in figure [ fig : li_results ] , we show @xmath3(li ) as a function of @xmath210 magnitude and @xmath1c/@xmath2c . in general , the rgb stars ( identified as squares ) show the largest abundances , while the candidate rc stars ( triangles ) show depleted abundances . this is true even in ngc 6583 ; the rc stars have depleted @xmath3(li ) compared to the first ascent rgb stars , but the depletion in this cluster is less than that in the other three clusters . the apparent dichotomous @xmath3(li ) distribution in this cluster is in fact three different levels of @xmath3(li ) : rgb stars with the least li dilution , rc stars with moderate li dilution , and one rc star with significant dilution . our four cluster sample suggests that if li - replenishment is commonly occurring at the he flash , the subsequent destruction of li reduces the surface abundance below the rgb levels . a few outliers are worth noting . first , in ngc 6583 the most luminous star has @xmath3(li ) comparable to the other rgb star , whereas the most luminous star in ngc 2204 has severely depleted levels . this implies that the ngc 2204 luminous star is on the agb , not the rgb . in all of the clusters except ngc 2506 , there are one or more candidate rc stars with @xmath3(li ) more similar to the rgb star levels . rc stars are plotted as red circles in figures [ fig : cmds ] ( which was used to determine their rc candidacy ) and [ fig : cmds2 ] . we find that all but one of the candidate rc stars with high li are closer matches to the first ascent rgb in figure [ fig : cmds2 ] and were reclassified based on those positions . the exception is ngc 6583 72 , whose classification is ambiguous in figure [ fig : cmds2 ] . the second panel in figure [ fig : li_results ] shows @xmath3(li ) as a function of @xmath1c/@xmath2c . most of the stars fall along a trend of decreasing @xmath1c/@xmath2c with decreasing @xmath3(li ) . this behavior is expected since both quantities are altered by mixing in stars . the stars along the linear trend follow the same distribution as the open cluster stars studied by @xcite . there are again some exceptions . there is a grouping of eleven stars with low @xmath1c/@xmath2c but @xmath3(li)@xmath211 dex . seven of these have limits in one of the parameters such that they could fall along the trend . four stars , on the other hand , are still outliers . all of the stars are in the clusters that had the difficult telluric removal , so it is possible that the low @xmath1c/@xmath2c measurements are spurious . however , our confidence in the validity of the measurements are due to the facts that ( 1 ) the @xmath2cn feature is stronger ( more easily detectable ) for low @xmath1c/@xmath2c , and ( 2 ) owing to the blend of lines creating the @xmath2cn feature , the width of the line would be difficult to mimic with residuals from the telluric removal . this sample of stars was selected specifically to test the hypothesis that li is generated at the he flash . in the sample , all of the li - rich stars that have @xmath1c/@xmath2c available have @xmath3(li)@xmath212 dex . this minimum @xmath3(li ) in the sample is influenced by two factors . first , the newly discovered rgs in were pre - screened for the presence of the li line in low resolution spectra , leading to a temperature - dependent minimum @xmath3(li ) . second , the lowest @xmath3(li ) stars in both the new rg and literature rg samples in do not have @xmath1c/@xmath2c measurements . none of our stars have li abundances at this level , implying a low occurrence rate for li - rich stars . using the binomial statistics in @xcite , we can estimate the @xmath213 upper bound for the occurrence rate of rc clump stars more li - rich than @xmath214 dex to be @xmath215% . this result is the same if we use all 38 photometrically identified rc candidates or if we reduce the sample size to 35 to account for the ones that may be rgb stars . the lack of a li - rich star in our sample is still consistent with the field giant study , which estimated the incidence to be around 1% . if we factor @xmath1c/@xmath2c into our consideration , we can perform a more stringent test of the fraction of stars that go through a li - enriched phase early in the core he - burning stage . in figure [ fig : cratio_hist ] , we show the @xmath1c/@xmath2c distribution of the subset of the sample that overlaps our rc sample in parameter space . we remove stars from the paper that have @xmath216 and @xmath47@xmath217 k. the first cut removes four luminous rgs in that are outside of the mass range and are near the second dredge - up phase on the early - agb ( see ) . the second cut removes cooler rgs that are more likely to be at the luminosity bump . overlaid in figure [ fig : cratio_hist ] is the @xmath1c/@xmath2c distribution of our sample of 38 rc candidates . the distributions are very different . the majority of the li - rich stars have @xmath1c/@xmath2c@xmath218 . such a low @xmath1c/@xmath2c is a signature of deep mixing that dredges up material from the vicinity of the h - burning shell , where @xmath1c/@xmath2c is 3.5 @xcite . this implies that the mechanism responsible for the enhanced li requires ( or is accompanied by ) very deep mixing . because @xmath1c/@xmath2c of the stellar atmosphere can only be lowered by mixing , this signature of low @xmath1c/@xmath2c will persist even when the lithium levels return to normal low values . therefore , the @xmath1c/@xmath2c distribution of our rc sample is inconsistent with the hypothesis that most / all stars of 1.52.2 @xmath0 experience a brief li - rich stage . instead , the li - rich stage must only be experienced by a small fraction of stars . we can identify the subset of stars in our sample that likely went through a li - rich phase as the outlier group of low @xmath1c/@xmath2c stars we identified in figure [ fig : li_results ] . if the four stars that are clear outliers ( i.e. , no limit on either abundance measurement ) are the only true outliers , then only @xmath219 of the rc stars went through a li - rich phase . the fraction is larger if we include all 11 stars in the low @xmath1c/@xmath2c outlier group , which results in an occurrence rate of @xmath220 . the rc sample for these two calculations only use the rc candidates without strong evidence for reclassification ( i.e. , the ` reclassified ' rc candidates in collinder 110 and ngc 2204 are removed ) and only considers stars with @xmath1c/@xmath2c measurements . these constraints yield 29 rc stars . c/@xmath2c in the 38 rc candidates in this work ( open histogram ) and for the li - rich stars with @xmath221 and @xmath47@xmath222 k in ( filled histogram ) . the temperature and luminosity limits were imposed on the sample to restrict it to stars most similar to the rc stars in this work . [ fig : cratio_hist],scaledwidth=50.0% ] our interpretations may also be affected by the unknown rotational histories of the individual stars . the main sequence progenitors of the stars in the mass range we are studying have a large distribution of rotational velocities , with a typical value of 150 km@xmath19s@xmath20 . despite this large variation on the ms , @xcite showed that the red giant descendants of these stars are still almost uniformly slow rotators , a result that is reproduced here ( section [ sec : rot ] ) . this fact makes it difficult to discern which stars were faster or slower ms rotators . the initial ms rotation affects the depth of mixing and thus the observed surface @xmath3(li ) of rgs . furthermore , rotation extends the main sequence lifetime . in an effort to explain the extended ms turn - offs of clusters in the magellanic clouds , @xcite explored the differences of isochrone morphology of models including large initial rotation compared to non - rotating models . they found that the isochrones made with fast rotating models were nearly indistinguishable from the slow rotating models . the implication to this work is that some of the rgs may be both more massive than we expect and may have experienced extensive rotational mixing . these stars would be difficult to distinguish from the stars that conform to the non - rotating model assumptions . among intermediate age milky way open clusters , some show extended rc morphologies that are inconsistent with simple single stellar population models . some of this extension is due to the fact that the ages of these clusters are such that the rc stars span the mass range that delineates the transition from quiescent to he flash core burning , resulting in a primary and secondary rc @xcite . however , this can not explain all of the spread in the msto , nor can it be explained with an age spread . the authors note that variations in the mass loss rate on the rgb or in the core overshoot efficiencies on the ms ( which affects the mass transition for the different types of he core burning onset ) could be responsible . if these factors are at play in the clusters of this work , they may also influence the variation of @xmath3(li ) observed in our clusters . we have measured @xmath3(li ) and @xmath1c/@xmath2c in a sample of rgs in four southern open clusters . most of the stars are rc stars having masses between 1.6 and 2.2 @xmath0 , which were selected to test the hypothesis that stars in this mass range synthesize li during the he flash and spend a short fraction of their rc lifetimes as li - rich stars . we find seven stars with @xmath3(li ) near the threshold of li - richness at 1.5 dex . three of these stars are rgb stars , and all of the four that were identified as candidate rc stars on cmds are or could be consistent with the rgb on spectroscopic @xmath47@xmath223 diagrams . given the modest sample size , the absence of li - rich stars in the rc constrains the occurrence ( or fractional lifetime ) to @xmath215% , which is compatible with the 1% lifetime quoted by . however , the majority of the li - rich rc field giants in have low @xmath1c/@xmath2c , the result of deep mixing which dredges - up material with @xmath1c/@xmath2c near 3.5 . even if the freshly synthesized li is destroyed during the rc lifetime , the evidence of the past deep mixing ( i.e. , low @xmath1c/@xmath2c ) should remain . the rc sample in this study has a @xmath1c/@xmath2c distribution that is much higher , demonstrating that these stars experienced less mixing . we therefore conclude that if a li - enrichment episode occurs at the he flash , it must only affect a fraction of the stars evolving through that phase . from a set of outliers in the @xmath3(li)@xmath1c/@xmath2c distribution , we estimate that fraction to be @xmath219 . accounting for stars with uncertain evolutionary stage and limits in their abundances , we can set a conservative upper limit for the incidence to be @xmath5 . we are grateful to s. h. lee and h. b. ann for providing us their optical photometry of the ngc 2506 red giants in this study . we also thank the referee for valuable feedback that improved this manuscript . jkc acknowledges support by an appointment to the nasa postdoctoral program at the goddard space flight center , administered by universities space research association through a contract with nasa . this paper includes data gathered with the 6.5 m magellan telescopes located at las campanas observatory , chile and made use of the webda database , operated at the department of theoretical physics and astrophysics of the masaryk university .

if the lepton mixing angle is not maximal , there arises a problem of ambiguity in determining due to the existence of two degenerate solutions , one in the first and the other in the second octant . we discuss an experimental strategy for resolving the octant degeneracy by combining reactor measurement of with accelerator disappearance and appearance experiments . the robustness of the degeneracy and the difficulty in lifting it only by accelerator experiments with conventional ( and ) beam are demonstrated by analytical and numerical treatments . our method offers a way to overcome the difficulty and can resolve the degeneracy between solutions and if at 95% cl by assuming the t2k phase ii experiment and a reactor measurement with an exposure of 10 gwktyr . the dependence of the resolving power of the octant degeneracy on the systematic errors of reactor experiments is also examined . 1.4 cm

hep-ph0601258

the atmospheric neutrino observation by super - kamiokande ( sk ) , which discovered neutrino oscillation , established that the mixing angle @xmath0 is large , which may be even close to the maximum @xcite . then , it became clear that the solar mixing angle @xmath9 is also large , though not maximal , as indicated by accumulation of the solar and the reactor data @xcite . curiously enough , the third angle @xmath1 is known to be small @xcite . comprehending the structure of lepton mixing described by the maki - nakagawa - sakata ( mns ) matrix @xcite which is composed of a nearly maximal , a large , and a small angle should reveal the key to understand the underlying physics of lepton flavor mixing . toward the explanation of the coexistence of nearly maximal and small mixing angles some symmetries @xcite which are motivated by more phenomenological @xmath10 exchange symmetries @xcite are discussed . interestingly , most of them share an attractive feature that @xmath11 and @xmath12 in the symmetry limit , suggesting a possible common origin of the two small quantities in the lepton flavor mixing , @xmath1 and @xmath13 , the parameter which measures deviation from the maximal mixing . the quark - lepton complementarity @xcite , if extends to the 2 - 3 sector , might be relevant for maximal or nearly maximal @xmath0 . if these small quantities take non - vanishing values it should give us a further hint for deciding if the symmetries are the right explanation for the two small quantities , and if so , for identifying the correct symmetry . in this paper , we describe an experimental strategy for determining @xmath0 following @xcite , where it was proposed that the @xmath0 octant degeneracy can be lifted by combining reactor measurement of @xmath1 with accelerator appearance and disappearance measurement of certain combinations of @xmath0 and @xmath1 . see ref . @xcite for an earlier suggestion of this possibility . we thoroughly examine this method by taking a concrete ( and probably the best thinkable ) setting for accelerator experiments , phase ii of the t2k experiment @xcite , with hyper - kamiokande as a detector @xcite and a 4 mw neutrino beam from the j - parc facility . for knowledges of reactor measurement , we are benefited by the outcome of the world - wide effort @xcite . we carry out a detailed quantitative analysis to obtain the region of @xmath1 and @xmath0 in which the @xmath0 octant degeneracy can be resolved by our method . for a related analysis , see @xcite . the @xmath0 octant degeneracy is a part of a larger structure called the parameter degeneracy . it may be understood as the intrinsic degeneracy @xcite duplicated by the unknown sign of @xmath14 @xcite and @xmath0 octant ambiguity @xcite , which lead to the total eight - fold degeneracy @xcite . ( see @xcite for exposition of analytic structure of the degeneracies . ) then , one has to address a problem ; is it possible to resolve only the @xmath0 degeneracy , leaving the other two types of degeneracies untouched ? we will answer in the positive this question in the context of conventional @xmath2 beam experiments . in short , in modest baseline distances for which perturbative treatment of the matter effect is valid , the @xmath0 degeneracy is decoupled from the other degeneracies and can be resolved independently in the presence of them . the current bounds on these small quantities in the lepton flavor mixing are rather mild , @xmath15 at 90% cl @xcite , which is nothing but a translation of the bound @xmath16 . at present , there is neither indication of deviation from the maximal @xmath0 , nor preference of the particular octant ( apart from very slight preference of the 2nd octant ) in the analysis by the sk group with their current data set @xcite . on the other hand , the bound on @xmath1 is much stronger if one uses the same variable , @xmath17 at 90% ( 3@xmath18 ) cl for 1 degree of freedom , as obtained by the global analysis @xcite with use of all the data including the chooz , the atmospheric , and the k2k @xcite one . as discussed in @xcite ( see also @xcite ) , one of the major difficulties for accurate measurement of @xmath0 with accelerator neutrinos is the @xmath0 degeneracy . hence , we expect that our method will help to improve the situation . in sec . [ how ] , we explain how the @xmath0 octant degeneracy can be resolved by our method . in sec . [ robust ] , we discuss how robust is the @xmath0 octant degeneracy by indicating the difficulties in resolving it only by accelerator measurement . in sec . [ analysis_method ] , we fully explain the statistical procedure of our analysis . in sec . [ analysis_results ] , we present the results of our analysis . in sec . [ conclusion ] , we give concluding remarks . in appendix [ number ] , we describe some details of how the referred numbers of events are computed in our paper . for completeness of the presentation , we start by reviewing the method for solving @xmath0 octant degeneracy described in @xcite . at the end of this section , we will try to elucidate the difference between this work and the reference @xcite . we invite the readers to look at fig . [ 23fig1 ] , and first focus on the upper four panels , the case where the input value of @xmath0 is in the first octant . ( the upper four and the lower four panels of fig . [ 23fig1 ] are for input values of @xmath19 and 0.542 , respectively . ) [ 23fig1]a describes the constraints imposed by each accelerator experiment , @xmath2 ( and @xmath4 ) disappearance and @xmath3 ( and @xmath20 ) appearance measurement . although these contours come from our full analysis whose details will be explained in secs . [ analysis_method ] and [ analysis_results ] , the main features of fig . [ 23fig1 ] can be understood by the vacuum oscillation approximation , and essentially it is all that we need . -1.6 cm octant degeneracy can be resolved for the case where the true value of @xmath21 = 0.458 ( 0.542 ) , corresponding to @xmath22 . the other input mixing parameters are given as @xmath23 ev@xmath24 , @xmath25 = 0.1 and @xmath26 , @xmath27 ev@xmath24 , @xmath28 = 0.31 ( the input values of @xmath25 and @xmath21 are indicated by the symbol of star in the plot ) . ( a ) the regions enclosed by the solid and the dashed curves are allowed regions only by the results of appearance and disappearance accelerator measurement , respectively . ( b ) the regions that remain allowed when results of appearance and disappearance measurement are combined . ( c ) the regions allowed by reactor measurement . ( d ) the regions allowed after combining the results of appearance and disappearance accelerator experiments with the reactor measurement . the exposures for accelerator are assumed to be 2 ( 6 ) years of neutrino ( anti - neutrino ) running with 4 mw beam power with hyper - kamionande whose fiducial volume is 0.54 mt , whereas for the reactor we assume an exposure of 10 gw@xmath8kt@xmath8yr . the case of optimistic systematic error is taken . ( see secs . [ analysis_method ] and appendix [ number ] for details . ) , scaledwidth=96.0% ] -1.6 cm the @xmath2 disappearance and @xmath3 appearance probabilities in one @xmath29 dominance approximation @xcite , which may be justified by @xmath30 , are given by @xmath31 where @xmath32 denotes the neutrino energy and @xmath33 is the baseline distance . we use the standard notation of the mns matrix including the symbol @xmath34 for @xmath35 @xcite . @xmath36 is defined as @xmath37 by using the neutrino masses @xmath38 ( @xmath39 1 - 3 ) . in this approximation @xmath40 . accelerator disappearance measurement is expected to determine both @xmath14 and @xmath41 with high accuracies . it is obvious that , if @xmath42 , one has two - fold solutions of @xmath0 for a given value of @xmath41 ; @xmath43 $ ] . this is a simple illustration of how @xmath0 octant degeneracy arises . the almost horizontal two lines in fig . [ 23fig1]a are nothing but these two solutions of @xmath44 for a given value of @xmath22 . on the other hand , accelerator appearance measurement determines a particular combination of two angles , @xmath45 , as seen in ( [ pvac_mue ] ) for a given value of @xmath14 . the latter quantity is expected to be well determined by disappearance measurement . the curved strip in fig . [ 23fig1]a , which takes the approximate form @xmath46 , is the outcome of the @xmath3 and @xmath20 appearance measurement . in fig . [ 23fig1]b we combine the disappearance and appearance measurement and we end up with a pair of degenerate solutions of @xmath1 and @xmath0 . now , we combine the information from reactor experiments . in a very good approximation what they measure is equal to the vacuum oscillation probability @xmath47 which is given again in the one @xmath29 dominance approximation as @xmath48 given the value of @xmath14 by disappearance measurement , the reactor experiments determine @xmath1 independent of other mixing parameters . this is illustrated in fig . [ 23fig1]c . when we combine the accelerator disappearance and appearance experiments as well as reactor measurement , one of the two allowed solutions in fig . [ 23fig1]b disappears as shown in fig . [ 23fig1]d . thus , the @xmath0 octant degeneracy can be lifted . though using the same method as proposed in @xcite , our analysis goes beyond that given in the reference in a number of ways . we have taken into account errors in accelerator disappearance and appearance as well as background so that the event number distributions roughly reproduce those obtained by the experimental group @xcite . our treatment of the reactor measurement of @xmath1 is elaborated to include uncorrelated and correlated errors in order to treat the so called phase ii type high statistics measurement . by using the highest sensitivity reactor and accelerator experiments , we aim at revealing the ultimate sensitivity for resolution of the @xmath0 octant degeneracy achievable by this method . differences between this paper and @xcite exist not only in the method for the analysis but also in some features of the results . for the particular set of parameters used in fig . [ 23fig1 ] , the resolving power of the @xmath0 degeneracy is greater for the case with true values of @xmath0 in the first octant than for the case in the second octant . it is in disagreement with the naive expectation in @xcite based on the difference between @xmath0 of a true and a fake solutions . see sec . [ analysis_results ] for more details . we illustrate how difficult is to resolve the octant degeneracy by accelerator experiments with conventional neutrino beam . apparently , this fact has been recognized by people in the community , but to our knowledge , a coherent discussion of why it is so has never been given . therefore , we try to fill the gap . as a byproduct , the argument illuminates the nature of the @xmath0 degeneracy and may indicate a unique feature of our approach . we note that , strictly speaking , our following discussion in this section is valid under circumstances that the matter effect can be treated perturbatively , and hence for baseline shorter than @xmath49 km . ( see below . ) robustness of the degeneracy is obvious from fig . [ 23fig1]a and fig . [ 23fig1]b , but we want to elaborate the discussion to bring our understanding to a little deeper level . the reasons for the robustness are mainly twofold : * the difference in energy spectra predicted by degenerate solutions in either the appearance or disappearance channels is too small to distinguish between the first- and the second - octant solutions . * the matter effects in the disappearance channels are not strong enough to lift the octant degeneracy , unless one goes to very long baseline . to explain the first point , we introduce the quantity @xmath50 , @xmath51 the difference between probabilities with parameters of two degenerate solutions , @xmath52 and @xmath53 ( i=1st , 2nd ) , determined at a particular value of energy in this exercise . in fig . [ app - disapp ] , plotted are @xmath54 ( left upper panel ) , @xmath55 ( right upper panel ) , and their antineutrino counterparts in the lower two panels . as one can see in fig . [ app - disapp ] , @xmath56 s are less than @xmath57% in most of the energy region . in this exercise , the cp phase @xmath58 ( i=1st , 2nd ) is kept equal and we did not try to adjust it to make @xmath56 smaller . we have also examined three other values of delta , @xmath59 , in addition to the @xmath60 case in fig . [ app - disapp ] , and reached the same conclusion . the smallness of @xmath54 and @xmath55 discourages the possibility of resolution of the @xmath0 octant degeneracy by using spectrum informations . this is in sharp contrast to the case of intrinsic degeneracy of @xmath1-@xmath61 , for which the spectrum analysis is proved to be powerful in the presently discussed original t2k setting provided that @xmath1 is not too small @xcite . -1.5 cm 1.0 cm degenerate solutions are shown to have almost the same energy spectrum in kamioka with baseline of 295 km from j - parc . in the upper ( lower ) panels , we show the difference of the probabilities of two degenerate solutions for neutrino ( anti - neutrino ) channels . the solid and the dashed lines are for normal ( @xmath62 ) and the inverted ( @xmath63 ) mass hierarchies , respectively . , title="fig:",scaledwidth=100.0% ] -3.5 cm we present an analytic framework to understand better the reasons for the robustness of the @xmath0 octant degeneracy . we restrict ourselves to baselines where the earth matter effect can be treated as perturbation in the appearance channel . for a consistent treatment we start from the expression of the disappearance probability @xmath64 in the one @xmath29 dominant vacuum oscillation approximation but with leading order @xmath65 correction ) is small because of the suppression by a factor of @xmath66 at around the first oscillation maximum . ] @xmath67 \sin^2 \left(\frac{\delta m^2_{31}l}{4e}\right ) . \label{pvac_mumu2}\end{aligned}\ ] ] then , a disappearance measurement determines @xmath44 to first order in @xmath65 as @xmath68 where @xmath69 is the solution obtained by ignoring @xmath65 . by reading off the coefficient of sine squared term in ( [ pvac_mumu2 ] ) , one can determine @xmath41 and @xmath44 is given by @xmath70 $ ] as noted before . we note that the linear dependence of @xmath71 on @xmath65 is clearly seen in fig . [ 23fig1]a . for the appearance channel , we use the @xmath3 ( @xmath72 ) appearance probability with first - order matter effect @xcite @xmath73 \nonumber \\ & = & \sin^2{2\theta_{13 } } s^2_{23 } \left [ \sin^2 \left(\frac{\delta m^2_{31 } l}{4 e}\right ) -\frac { 1}{2 } s^2_{12 } \left(\frac{\delta m^2_{21 } l}{2 e}\right ) \sin \left(\frac{\delta m^2_{31 } l}{2 e}\right ) \right . \nonumber \\ & & \hspace*{22 mm } { } \pm \left . \left(\frac { 4 ea(x)}{\delta m^2_{31}}\right ) \sin^2 { \left(\frac{\delta m^2_{31 } l}{4 e}\right ) } \mp \frac{a(x)l}{2 } \sin \left(\frac{\delta m^2_{31 } l}{2 e}\right ) \right ] \nonumber \\ & & + 2j_{r } \left(\frac{\delta m^2_{21 } l}{2 e } \right ) \left [ \cos{\delta } \sin \left(\frac{\delta m^2_{31 } l}{2 e}\right ) \mp 2 \sin{\delta } \sin^2 \left(\frac{\delta m^2_{31 } l}{4 e}\right ) \right ] . \label{pmue}\end{aligned}\ ] ] in ( [ pmue ] ) , @xmath74 @xcite where @xmath75 is the fermi constant , @xmath76 denotes the electron number density at the point @xmath77 in the earth , @xmath78 @xmath79 denotes the reduced jarlskog factor , and the upper and the lower sign @xmath80 refer to the neutrino and anti - neutrino channels , respectively . we make an approximation of ignoring terms of order @xmath81 . note that keeping only the leading order in this quantity is reasonable because @xmath82 , @xmath83 , and @xmath84 for @xmath85 . then , the two degenerate solutions obey an approximate relationship @xmath86 or , @xmath87 ignoring higher order terms in @xmath88 . we can neglect the leading order correction in @xmath65 to @xmath44 in these relations because it gives @xmath89 terms . with our machinery well oiled we can understand better the behavior of @xmath54 and @xmath55 given in fig . [ app - disapp ] , in particular their small values , @xmath91 . we also note that even if we switch off the matter effect , there is no visible change in @xmath54 and @xmath92 in fig . [ app - disapp ] . now , these features can be understood in our analytic framework . noting that @xmath93 in leading order in @xmath94 , the difference between probabilities with the first and the second octant solutions can be given by @xmath95 . \label{deltapmue}\end{aligned}\ ] ] the size of @xmath56 is about @xmath96 for @xmath97 , which is roughly consistent with the results given in fig . [ app - disapp ] . the remarkable feature of ( [ deltapmue ] ) is that the leading - order matter effect terms drops out completely , because they depend upon @xmath0 and @xmath1 only though the invariant combination ( [ 1st_order ] ) . therefore , we suspect that our statement about robustness of the @xmath0 octant degeneracy may apply not only to the t2k experiment but also to experiments with longer baseline , including the no@xmath98a project @xcite . does the disappearance channel @xmath99 help ? using the approximate relationship ( [ 1st_order ] ) and the first order ( in @xmath65 ) corrected formula for @xmath44 mentioned after ( [ pvac_mumu2 ] ) , we obtain the expression of @xmath55 to first order in matter effect as a sum of the vacuum and the matter effect contributions , @xmath100 . the vacuum term @xmath101 \left(\frac{\delta m^2_{21}l}{2e}\right ) \sin \left(\frac{\delta m^2_{31}l}{2e}\right ) \label{deltapmumu_vac } \end{aligned}\ ] ] is small , @xmath91 , because of the suppression by either one of @xmath65 or @xmath78 , and @xmath102 . the matter term is given by @xcite @xmath103 where @xmath104 , @xmath105 and the function @xmath106 is defined as @xmath107 . since the term flips sign under the transformation @xmath108 , it may be used to discriminate if @xmath109 or @xmath110 @xcite . unfortunately , the size of ( [ deltapmumu_matter ] ) and hence its contribution to @xmath55 is small , @xmath111 note that @xmath106 is a monotonically increasing function of @xmath77 in @xmath112 , and @xmath113 . thus , the difference @xmath114 is of order @xmath115 at @xmath116 km ( less than 1% even at @xmath117 km ) with @xmath1 at around the chooz limit @xcite . it should be noticed that @xmath56 flips sign not only under @xmath118 but also under @xmath119 . therefore , measurement of the sign of @xmath56 determines the combined sign , @xmath120 . hence , the use of this effect to resolve the @xmath0 degeneracy requires prior knowledge of the neutrino mass hierarchy , i.e. , the sign of @xmath14 . thus , @xmath56 is small , typically 0.1% level in the t2k setting , both in the appearance and the disappearance channels . this feature explains well the behavior shown in fig . [ app - disapp ] . under the approximation of ignoring @xmath56 in the appearance and the disappearance channels , there is no way to resolve the @xmath0 degeneracy by spectrum analysis . but , on the other hand , it implies that the octant degeneracy decouples from the other types of degeneracies , @xmath14-sign and the intrinsic ones . the desirable feature prevails after the @xmath0 degeneracy is resolved because it is executed by combining the reactor measurement of @xmath1 in our method , which is free from any degeneracies . we can not lift the @xmath14-sign degeneracy by the method we explore in this paper , but we can resolve the @xmath0 octant degeneracy independent of the sign of @xmath14 . we will explicitly verify this point in sec . [ analysis_results ] by performing the analysis under assumptions of the right and the wrong mass hierarchies . a simple remark on the intrinsic degeneracy of @xmath1-@xmath61 ; the coupling between the @xmath0 and the intrinsic degeneracy can be avoided by doing measurement at the oscillation maximum , or more precisely the `` thinnest ellipse '' limit @xcite . or , if necessary , it can be resolved relatively easily ( compared to the @xmath14-sign degeneracy ) by doing the spectrum analysis , as demonstrated in the t2k setting in @xcite . in this section , we summarize the statistical method and the procedure of our analysis . we ask the readers to refer to appenix [ number ] for any details of how the numbers of events are computed . we use the full three - flavor oscillation formulas in our analysis . they are obtained by numerically solving the neutrino evolution equation with the constant electron number density . for concreteness , we consider for accelerator experiment the phase ii of the t2k project @xcite where the beam power is upgraded to 4 mw and the far detector will be hyper - kamiokande with 0.54 mt fiducial volume . we consider the 2.5 degree off axis @xmath121 beam which has a peak at around 0.65 gev . we assume the exposures of 2 and 6 years of neutrino and anti - neutrino running , respectively @xcite . for the reactor experiment , we consider the exposure of 10 gw@xmath8kt@xmath8yr . the @xmath122 ( @xmath123 ) appearance mode is important to determine precisely the value of @xmath124 , provided that @xmath125 is well determined , which is possible by the disappearance mode . as discussed in ref . @xcite , if the experiment is done at or close to the oscillation maximum , we can determine well the quantity @xmath124 , even if we do not know the value of the cp phase @xmath61 . since information on the energy dependence of this channel can not be important in resolving @xmath0 degeneracy ( as we saw in the previous section ) , we consider only the total number of events and define @xmath126 for the appearance channel as follows , @xmath127 where @xmath128 and @xmath129 are the number of events to be observed and the theoretically expected one , respectively , for given values of the oscillation parameters . we note that background events come mainly from neutral current interactions as well as the events induced by @xmath130 ( @xmath131 ) which is inevitably contained in the initial flux . we assume optimistic systematic errors , @xmath132 % for the t2k ii experiment . 0.5 cm .number of events in the appearance mode with and without oscillation . we assume 2 ( 6 ) years of neutrino ( anti - neutrino ) running with the t2k phase ii set up . for the case with oscillation , we set @xmath23 ev@xmath24 , @xmath133 , @xmath134 , and @xmath135 , with the matter density @xmath136 g/@xmath137 and the electron fraction ( number of electron per nucleon ) being @xmath138=0.5 . numbers in parentheses correspond to the case where the matter effect is switched off . [ cols="^,^,^,^",options="header " , ] [ tab - events ] 0.5 cm in table i , we show the expected number of events for the t2k phase ii for the 2 ( 6 ) years of exposure for neutrino ( anti - neutrino ) running with and without oscillation effect . for the case with oscillation , we assumed the oscillation parameters @xmath23 ev@xmath24 , @xmath133 , @xmath134 , and @xmath135 . the choice is to compare the number of events to the one quoted in ref . @xcite for t2k i after properly scaling the fiducial volume , exposure time and the beam power . we have confirmed that our results agree reasonably well with the numbers quoted in @xcite . -2.5 cm ev@xmath24 , @xmath139 , @xmath140 , @xmath27 ev@xmath24 , @xmath28 = 0.31 , @xmath141 . , scaledwidth=80.0% ] -2.5 cm although we do not use the spectrum informations in our analysis , we present in fig . [ app_spectrum ] some examples of the energy distribution of the @xmath3 appearance events for completeness . for this plot , we set the oscillation parameter @xmath142 ev@xmath24 , @xmath139 , @xmath143 , @xmath27 ev@xmath24 , @xmath28 = 0.31 , @xmath141 . throughout our analysis , we fix the solar neutrino mixing parameters @xmath144 and @xmath28 to these values . one can observe in the figure that background is quite small in size and has similar shape as the signal events . notice also that the modulation of energy spectrum by neutrino oscillation is rather modest . the @xmath145 ( @xmath146 ) disappearance mode is important to determine @xmath147 as well as @xmath14 accurately . in fig . [ event_distribution ] , we show the expected event number distribution as a function of the reconstructed neutrino energy in the absence ( left panel ) and in the presence ( right panel ) of oscillation with the mixing parameters , @xmath139 , @xmath148 ev@xmath24 , @xmath135 , and @xmath141 . unlike the case of the appearance channel , the energy distribution is significantly modified by the oscillation effect . -1.5 cm ev@xmath24 , @xmath139 , @xmath140 , @xmath27 ev@xmath24 , @xmath28 = 0.31 and @xmath141 . the histogram by the solid lines indicate the sum of the signal and background events whereas the ones by the dashed line indicate only the background events . , scaledwidth=80.0% ] -3.8 cm for the disappearance mode , we consider 36 bins with 50 mev width from 0.2 gev to 2.0 gev in terms of the reconstructed neutrino energy . the @xmath126 function is defined as follows , @xmath149 ^ 2 } { n_i^{\text{obs}}+n_{i,\text{bg}}^{\text{obs } } } + \left(\frac{\alpha_{\text{sig}}}{\sigma_{\text{sig}}}\right)^2 + \left(\frac{\alpha_{\text{bg}}}{\sigma_{\text{bg}}}\right)^2 , \label{chi2_dis}\ ] ] where @xmath150 and @xmath151 are , the number of signal events to be observed and the theoretically expected one , respectively for the @xmath152-th bin , and @xmath153 and @xmath154 are the corresponding background event numbers . we assume the optimistic values for the systematic errors , @xmath155% for the t2k phase ii . in fig.[fig : sensitivity_angle ] , we show the expected sensitivity for @xmath147 assuming the pure 2 flavor oscillation ( @xmath12 ) as a function of the true value of @xmath125 , which compares reasonably well with that of @xcite . we have checked that this simple @xmath126 can reproduce the expected t2k sensitivity based on more realistic calculations obtained by monte carlo simulation given in ref . @xcite . 1.0 cm as a function of the true value of @xmath156 obtained with the @xmath126 defined in ( [ chi2_dis ] ) assuming the pure 2 flavor oscillation . , scaledwidth=80.0% ] -7.0 cm for the reactor experiment , we use the same @xmath126 function used in our previous works @xcite , which is defined as , @xmath158 + \frac { \alpha^2 } { \sigma_{\rm db}^2 } , \label{chi2_reac}\end{aligned}\ ] ] where @xmath159 represents the theoretical number of events at a near ( @xmath160 ) or a far ( @xmath161 ) detector within the @xmath152-th bin whose width is 0.425 mev . we set the distance to the near and the far detectors to be 300 m and 1500 m , respectively . @xmath162 are number of events to be observed . we consider four types of systematic error : @xmath163 , @xmath164 , @xmath165 and @xmath166 . the subscript d ( d ) represents the fact that the error is correlated ( uncorrelated ) between detectors . the subscript b ( b ) represents the fact that the error is correlated ( uncorrelated ) among bins . the @xmath167 s are parameters to be varied freely in order to take into account these systematic errors . in this work , we consider the high precision reactor experiments whose sensitivity can go beyond the ones currently expected by experiments such as double - chooz @xcite and kaska @xcite . namely , we assume sensitivities below @xmath168 , the one expected to be achievable by phase - ii type experiments such as the braidwood @xcite , the daya bay @xcite , and the angra @xcite projects . computation of the event number is done in the same way as in ref . we ignore the possible contribution from geo - neutrinos . it is demonstrated recently by kamland that its flux is consistent with the one expected by geo - chemical earth models @xcite . in this case , the effect of geo - neutrinos is negligibly small at the baseline of @xmath1691 km , as one can easily guess by extrapolation of the situation in a reactor @xmath9 experiment with baseline of @xmath16960 km @xcite . to characterize reactor measurement , we use gw@xmath8kt@xmath8yr , the `` total exposure '' unit , which is defined as the product of the net values of the reactor thermal power ( in gw ) , detector fiducial volume ( in kton ) and running time ( in year ) . we consider the reactor measurement for 10 gw@xmath8kt@xmath8yr . the total number of events at the far detector is @xmath170 . we consider two different sets of systematic errors : a relatively conservative choice and an optimistic one . for the conservative choice , we adopt similar values of the systematic errors used in ref . @xcite , @xmath171 % and @xmath172 % , and for the optimistic one , we set @xmath173 % and @xmath174 0.2% and @xmath175 % . the latter extremely small errors may be difficult to reach , but they are used to estimate the upper limit of resolving power of the @xmath0 degeneracy by the present method . the sensitivity limit of @xmath1 ( assuming no depletion ) at @xmath176 ev@xmath24 is @xmath177 ( @xmath178 ) at 1@xmath18 ( 3@xmath18 ) cl for the relatively conservative errors , and @xmath179 ( @xmath180 ) at 1@xmath18 ( 3@xmath18 ) cl for the optimistic ones . for the combined analysis , we simply sum all the @xmath126 functions defined in eqs . ( [ chi2_app ] ) , ( [ chi2_dis ] ) and ( [ chi2_reac ] ) , @xmath181 the allowed region in the @xmath182 plane is determined by the usual condition , @xmath183 2.3 , 6.18 and 11.83 for 1 , 2 and 3 @xmath18 cl for two degrees of freedom . we will establish the parameter regions where we can resolve the @xmath0 octant degeneracy for 1 degree of freedom by imposing the condition @xmath184 2.71 , 4 and 6.63 for 90 , 95 and 99% cl , respectively , where @xmath185 and @xmath186 are , respectively , the true and the false value of @xmath0 . in this section , we show our results based on our analysis with the combined @xmath126 of all channels . to remind the readers , our analysis is based on the t2k ii experiment of 2 ( 6 ) years running of neutrino ( anti - neutrino ) modes with 4mw beam power with the hyper - kamiokande detector whose fiducial volume is 0.54 mt @xcite . for the reactor experiment , the exposure of 10 gw@xmath8kt@xmath8yr is assumed . we assume throughout this section that the mass hierarchy ( determined by nature ) is normal type ( @xmath187 ) unless otherwise stated . even if we repeat the same procedure with the true mass hierarchy of inverted type ( @xmath63 ) , the region in which the degeneracy is solved is remarkably similar . therefore , we decided to concentrate on the normal hierarchy case . of course , we examine the stability of our results by assuming the wrong hierarchy in the analysis . -1.0 cm plane determined by the combined analysis for various different input parameters . for all plots , the input values of @xmath188ev@xmath24 and the cp phase @xmath61 = 0 . the input value of @xmath1 for the upper ( lower ) four panels is @xmath189 ( 0.04 ) . the input values of @xmath0 are ( a ) @xmath190 , ( b ) @xmath191 , ( c ) @xmath192 and ( d ) @xmath193 . the conservative values of the systematic errors are taken for reactor measurement ; @xmath171 % and @xmath172 % . , scaledwidth=100.0% ] -1.5 cm -1.0 cm but for smaller systematic errors for the reactor experiment , @xmath194 % , @xmath195 % and @xmath196 % . , scaledwidth=100.0% ] -1.5 cm in fig . [ allowed_region1 ] , we show the allowed regions in the @xmath197 plane at 1 @xmath18 ( light blue curve ) , 2 @xmath18 ( black curve ) , and 3 @xmath18 ( red curve ) cl for 2 degree of freedom , for 8 different sets of input parameters ; @xmath198 = ( 0.08 , 0.4 ) , ( 0.08 , 0.6 ) , ( 0.08 , 0.45 ) , ( 0.08 , 0.55 ) , ( 0.04 , 0.4 ) , ( 0.04 , 0.6 ) and ( 0.04 , 0.45 ) , ( 0.04 , 0.55 ) , which are indicated by the symbol of a star . they are determined by the combined @xmath126 analysis using all the channels , @xmath122 ( @xmath123 ) appearance mode , by @xmath145 ( @xmath199 ) disappearance mode and by @xmath200 disappearance mode . the relatively conservative values of the systematic errors are taken for the reactor measurement ; @xmath171 % and @xmath201 % . in our @xmath126 analysis , for given values of input parameters , we vary not only @xmath0 and @xmath1 but also @xmath61 and @xmath14 , as these parameters should be determined by the fit . note , however , that the range of @xmath14 and @xmath0 are restricted to the ones constrained by atmospheric neutrino experiments @xcite . for a larger value of @xmath202 , the octant degeneracy is resolved for @xmath203 ( see fig . [ allowed_region1]a1 and b1 ) , whereas for @xmath204 , the degeneracy can be resolved only at 1 @xmath18 cl but not at 2 @xmath18 cl or higher ( see fig . [ allowed_region1]c1 and d1 ) . for a smaller value of @xmath205 , for @xmath203 , the octant degeneracy is resolved only at 1 @xmath18 cl ( see fig . [ allowed_region1]a2 and b2 ) and for @xmath204 , the degeneracy can not be resolved even at 1 @xmath18 cl ( see fig . [ allowed_region1]c2 and d2 ) . in fig . [ allowed_region2 ] , we show the same quantities but for the small systematic errors for reactor experiments , @xmath194 % , @xmath206 % , the very optimistic ones . in this case , the octant degeneracy is completely resolved for a large @xmath1 , @xmath202 both for @xmath203 and 0.99 ( see fig . [ allowed_region2]a1 , b1 , c1 , and d1 ) . for a small value of @xmath1 , @xmath205 , however , we have a mixed result ; for @xmath203 , the @xmath0 degeneracy is resolved both for @xmath203 and 0.99 apart from a tiny 3@xmath18 cl region in fig . [ allowed_region2]a2 . for @xmath204 , the degeneracy is not lifted , except in fig . [ allowed_region2]d2 where the clone solution disappears at 1@xmath18 cl . -2.5 cm space where the @xmath0 octant degeneracy can be resolved at 90% ( thin green ) and 99% ( thick red ) cl . the solid ( dashed ) curve is for the case of taking the normal ( inverted ) hierarchy to perform the fit , assuming the normal hierarchy as input . conservative systematic errors , as indicated in the figure , are considered here . , scaledwidth=100.0% ] -4.5 cm -2.5 cm but with optimistic systematic errors . , scaledwidth=100.0% ] -4.5 cm in fig . [ region_resolved1 ] and fig . [ region_resolved2 ] , we show the region bounded by the solid and the dashed curves in which we can resolve the @xmath0 octant degeneracy in the @xmath207-@xmath21 plane at 90% ( thin green curves ) and 99% ( thick red curves ) cl for 1 degree of freedom for the conservative and the optimistic values of the reactor systematic errors , respectively . namely , the region inside the bands correspond to the parameters which satisfies @xmath208 2.71 and 6.63 where @xmath185 and @xmath186 are the true and the false values of @xmath0 , respectively . the solid ( dashed ) curves in fig . [ region_resolved1 ] and fig . [ region_resolved2 ] are for cases that we fit the data set generated with @xmath187 under the hypothesis of right hierarchy , @xmath187 ( wrong hierarchy , @xmath209 ) . if we are ignorant about the mass hierarchy the case of worse sensitivity ( wrong hierarchy ) must be considered as the sensitivity region . the fact that the solid and the dashed curves come close to each others proves that the resolution of the @xmath0 octant degeneracy can be carried out independent of the lack of knowledge on the neutrino mass hierarchy , the decoupling of the two degeneracies as argued in sec . [ decoupling ] . the figures indicate that our method of combining reactor measurement of @xmath1 with the accelerator disappearance and appearance experiments allows to resolve the @xmath0 octant degeneracy to a reasonable level . this is highly nontrivial because it is quite a robust degeneracy which is hard to lift by using only relatively short - baseline ( @xmath210 km ) accelerator measurement , as we saw in sec . [ robust ] . we can observe by comparison between fig . [ region_resolved1 ] and fig . [ region_resolved2 ] that improvement of the systematic errors in reactor measurement is the key to the better resolving power of the octant degeneracy . it should also be mentioned that the results in fig . [ region_resolved2 ] are obtained with the extremely small systematic errors . hence , the results may be interpreted as the limit achievable by the present method . so far we have assumed that the input mass hierarchy is normal while examining the true and the false mass hierarchies in analyzing data . in order to verify that resolving power of @xmath0 degeneracy is not affected by the unknown sign of @xmath14 , we tried the converse . namely , we tried to fit the data set generated with the inverted hierarchy @xmath209 under the hypothesis of the right and the wrong hierarchies . we do not show the results because they are very similar to those for the input normal mass hierarchy . of course , this is expected because the experimental setting we consider in this work is not sensitive to the sign of @xmath14 , which in turn implies that our results must be insensitive to the mass hierarchy confusion . this point was discussed in depth in sec . [ robust ] . we note that globally our results in fig . [ region_resolved1 ] and fig . [ region_resolved2 ] are similar to what we can find in fig.7 of ref . @xcite , which was obtained by a much more simplified analysis . however , there are some notable differences . in fig.7 of ref . @xcite , it is always true that the @xmath0 degeneracy is easier to lift in the second octant , @xmath211 . but , we observe that it is true only for relatively small @xmath1 , @xmath212 . for relatively large @xmath1 , @xmath213 , the octant degeneracy is easier to be resolved for the first octant case @xmath109 than for the second octant . it appears that at large @xmath1 , the two effects which were not taken into account in the treatment in @xcite come into play , as indicated in fig . [ 23fig1 ] ; the width of the appearance strip , which comes from @xmath61-dependent terms in the oscillation probability ( see eq . ( [ pmue ] ) in sec . [ analytic ] ) , is narrower at smaller @xmath214 , which is the case of true @xmath0 in the first octant , making rejection of the fake solution easier . also , the two disappearance `` lines '' come closer in the case of true @xmath0 in the second octant , which produces the similar effect . as a result of the two effects which simultaneously act toward the same direction , the degeneracy turns out to be easier to resolve for the case of true @xmath0 in the first octant . we , however , want to note that the feature of first - second octants asymmetry , in particular in fig . [ region_resolved1 ] , depends upon the systematic errors and its full understanding may require much more subtle discussions . in this paper , we have explored the possibility of resolving the @xmath0 octant degeneracy by combining reactor measurement of @xmath1 with the possible highest accuracy accelerator disappearance and appearance measurement , as proposed in @xcite . it utilizes the nature of the reactor experiment as a pure measurement of @xmath1 to resolve the degeneracy . it is nice to see that the reactor measurement can contribute to explore the two small quantities in lepton flavor mixing , @xmath1 and a deviation of @xmath0 from the maximal value . the results of our quantitative analysis indicates reasonably high performance of the method . as shown in the summary figures of the resolving capabilities , fig . [ region_resolved1 ] and fig . [ region_resolved2 ] , the degeneracy is resolved in region of relatively large @xmath1 and a sizable deviation of @xmath0 from the maximal . prior to the quantitative analysis based on our method , we have discussed the robustness of the @xmath0 octant degeneracy and illuminated the difficulty in resolving it only by accelerator experiments . in particular , we have demonstrated by showing a sample figure , fig . [ app - disapp ] , that neither the spectral information nor the matter effect enables us to resolve the degeneracy , this feature was also understood on the basis of analytic treatment using the approximate formulas for appearance and disappearance probabilities valid to first order in matter effect . considering the robustness of the degeneracy , the opportunity of resolution offered by our reactor - accelerator method is highly nontrivial . finally , several remarks are in order : \(1 ) as indicated in fig . [ 23fig1 ] , the sensitivity of our method for resolving the @xmath0 degeneracy is limited mainly by the accuracy of @xmath1 determination by reactor experiments . the region without resolving power which remains in fig . [ region_resolved2 ] must be taken as the intrinsic limitation of our method for resolving the @xmath0 degeneracy in its current form , because we have already assumed a rather optimistic values of systematic errors in the reactor measurement , in addition to the extreme accuracies of accelerator experiments . \(2 ) in this paper , we have considered the setting of t2k experiment originally described in @xcite . it would be interesting to examine how the sensitivity of resolution of @xmath0 degeneracy changes if we adopt the kamioka - korea identical two - detector setting by which the degeneracy related to the neutrino mass hierarchy and the intrinsic one can be resolved @xcite . \(3 ) our method for resolving @xmath0 degeneracy is by no means unique . the other possibilities include : accelerator measurement with silver channel ( @xmath215 ) @xcite which has a different @xmath0 dependence , and detection of the solar oscillation term by either atmospheric neutrino observation , @xcite or very long baseline accelerator experiments @xcite . combination of the accelerator and the atmospheric neutrino experiments can also be pursuit @xcite . the ultimate possibility would , of course , be the `` everything at once '' approach @xcite which combines measurement by superbeam and neutrino factories with golden and silver channels . in this appendix , we provide some detailed informations about how the number of events are computed . for clarity of notation we denote the neutrino energy ( @xmath32 in the text ) as @xmath216 in this appendix . 0.2 cm ( i ) * apparence channel @xmath122 ( or @xmath123 ) * 0.5 cm as we mentioned in sec . iv , since the information on the energy spectrum of the appearance channel is not important in resolving @xmath0 degeneracy , for this channel , we do not try to make binning but use the expected total number of events , which is computed as , @xmath217 where @xmath218 is the ( true ) neutrino energy , @xmath219 is the number of target nucleons in the detector , @xmath220 is the running ( exposure ) time , @xmath221 is the flux spectrum of the 2.5 degree off axis @xmath121 beam @xmath222 is the total cross section and @xmath223 is the detection efficiency , which is given as a function of the neutrino energy . the detection efficiency we used takes into account all the cut imposed to reduce background , which enables us to obtain the signal distribution in the range between 0.35 and 0.85 gev in terms of reconstructed neutrino energy @xcite . we consider two kinds of background events , @xmath224 where @xmath225 comes from neutral current interactions and @xmath226 implies the event induced by the @xmath130 neutrino existed in the original beam . as for the signal , we used the efficiency functions which correspond to the cuts found in ref . @xcite in order to compute the background . 0.5 cm 0.5 cm ( ii ) * disapparence channel @xmath145 ( or @xmath146 ) * 0.2 cm for the disappearance mode , it is essential to make binning in order to take into account the information on the energy spectrum . the expected number of signal events for the i - th bin is computed as follows @xmath227 where @xmath228 implies reconstructed neutrino energy , @xmath229 implies charged current quasi elastic ( ccqe ) reaction cross section , @xmath230 is the detection efficiency as a function of the reconstructed neutrino energy , which corresponds to the one used in ref . @xcite , @xmath231 is the gaussian - like resolution function , @xmath232,\ ] ] where we set @xmath233 = 80 mev . for this mode , we consider two kinds of background , @xmath234 where @xmath235 is the events coming from neutral current reaction and @xmath236 is the one coming from the charged current non - quasi elastic ( cc - nqe ) reactions . we note that @xmath235 do not depend on oscillation parameters , whereas @xmath236 depend on oscillation parameters in a non - trivial way . for the background coming from cc - nqe reactions , @xmath237 is not a good approximation because events induced by higher energy neutrinos mimic the signal induced by lower energy ones . however , we observe that it is a good approximation to take that @xmath238 mev or @xmath239 . this allows us to compute @xmath240 in the presence of oscillation provided that we know the distribution @xmath236 as a function of reconstructed neutrino energy in the absence of oscillation for the 2.5 degree beam . 0.5 cm ( iii ) * disappearance channel @xmath157 * 0.5 cm we compute the expected number of @xmath72 events in the @xmath152-th energy bin , @xmath241 , where @xmath242 @xmath243 where @xmath244 is the number of target protons in the detector fiducial volume , @xmath220 is the exposure time , and @xmath245 is the neutrino flux spectrum from the nuclear power plant ( npp ) expected at its maximal thermal power operation . @xmath246 denotes the averaged operation efficiency of the npp for a given exposure period and it is taken to be 100% here under the understanding that the unit we use gw@xmath247kt@xmath8yr refers the actual thermal power generated , not the maximal value . @xmath248 is the familiar antineutrino survival probability in vacuum , which is given by eq.([pvac_ee ] ) and it explicitly depends on @xmath144 and @xmath28 . @xmath249 is the @xmath131 absorption cross - section on proton , @xmath250 is the detector efficiency and @xmath251 is the energy resolution function , which is assumed to have a gaussian form with @xmath32 ( @xmath252 the observed prompt energy ( total @xmath253 energy ) and @xmath254 - 0.8 mev the true one . two of us ( h.m . and h.n . ) thank takaaki kajita with whom our understanding on parameter degeneracies has been deepened through the collaboration for @xcite . this work was supported in part by the grant - in - aid for scientific research , no . 16340078 , japan society for the promotion of science ( jsps ) , fundao de amparo pesquisa do estado de so paulo ( fapesp ) , fundao de amparo pesquisa do estado de rio de janeiro ( faperj ) and conselho nacional nacional de cincia e tecnologia ( cnpq ) . h.s . thanks jsps for support . r.z.f . is grateful to the abdus salam international center for theoretical physics where part of this work was done . y. fukuda _ et al . _ [ kamiokande collaboration ] , phys . b * 335 * , 237 ( 1994 ) ; y. fukuda _ et al . _ [ super - kamiokande collaboration ] , phys . rev . lett . * 81 * , 1562 ( 1998 ) [ arxiv : hep - ex/9807003 ] ; y. ashie _ et al . _ [ super - kamiokande collaboration ] , phys . rev . lett . * 93 * ( 2004 ) 101801 [ arxiv : hep - ex/0404034 ] ; y. ashie _ et al . _ [ super - kamiokande collaboration ] , phys . d * 71 * , 112005 ( 2005 ) [ arxiv : hep - ex/0501064 ] . b. t. cleveland _ et al . _ , astrophys . j. * 496 * , 505 ( 1998 ) ; j. n. abdurashitov _ et al . _ [ sage collaboration ] , phys . c * 60 * , 055801 ( 1999 ) [ arxiv : astro - ph/9907113 ] ; w. hampel _ et al . _ [ gallex collaboration ] , phys . b * 447 * , 127 ( 1999 ) ; s. fukuda _ et al . _ [ super - kamiokande collaboration ] , phys . b * 539 * , 179 ( 2002 ) [ arxiv : hep - ex/0205075 ] ; m. b. smy _ et al . _ [ super - kamiokande collaboration ] , phys . d * 69 * , 011104 ( 2004 ) [ arxiv : hep - ex/0309011 ] ; q. r. ahmad _ et al . _ [ sno collaboration ] , phys . rev . lett . * 87 * , 071301 ( 2001 ) [ arxiv : nucl - ex/0106015 ] ; _ ibid . _ * 89 * , 011301 ( 2002 ) [ arxiv : nucl - ex/0204008 ] ; b. aharmim _ et al . _ [ sno collaboration ] , phys . c * 72 * , 055502 ( 2005 ) [ arxiv : nucl - ex/0502021 ] . k. eguchi _ et al . _ [ kamland collaboration ] , phys . rev . lett . * 90 * , 021802 ( 2003 ) [ arxiv : hep - ex/0212021 ] ; t. araki _ et al . _ [ kamland collaboration ] , phys . lett . * 94 * , 081801 ( 2005 ) [ arxiv : hep - ex/0406035 ] . m. apollonio _ et al . _ [ chooz collaboration ] , phys . b * 420 * , 397 ( 1998 ) [ arxiv : hep - ex/9711002 ] ; _ ibid . _ b * 466 * , 415 ( 1999 ) [ arxiv : hep - ex/9907037 ] . see also , the palo verde collaboration , f. boehm _ et al . _ , phys . d * 64 * ( 2001 ) 112001 [ arxiv : hep - ex/0107009 ] . e. ma and g. rajasekaran , phys . d * 64 * , 113012 ( 2001 ) [ arxiv : hep - ph/0106291 ] ; e. ma , mod . a * 17 * , 2361 ( 2003 ) [ hep - ph/0211393 ] ; k. s. babu , e. ma , and j. w. f. valle , phys . b * 552 * , 207 ( 2003 ) [ hep - ph/0206292 ] ; w. grimus and l. lavoura , jhep * 0107 * , 045 ( 2001 ) [ arxiv : hep - ph/0105212 ] ; acta . phys . polon . * b34 * , 5393 ( 2003 ) [ arxiv : hep - ph/0310050 ] ; j. kubo , a. mondragon , m. mondragon and e. rodriguez - jauregui , prog . * 109 * , 795 ( 2003 ) [ arxiv : hep - ph/0302196 ] ; j. kubo , phys . b * 578 * , 156 ( 2004 ) [ arxiv : hep - ph/0309167 ] ; k. s. babu and j. kubo , phys . rev . d * 71 * , 056006 ( 2005 ) [ arxiv : hep - ph/0411226 ] ; w. grimus , a. s. joshipura , s. kaneko , l. lavoura and m. tanimoto , jhep * 0407 * , 078 ( 2004 ) [ arxiv : hep - ph/0407112 ] . w. grimus , a. s. joshipura , s. kaneko , l. lavoura , h. sawanaka and m. tanimoto , nucl . b * 713 * , 151 ( 2005 ) [ arxiv : hep - ph/0408123 ] . t. fukuyama and h. nishiura , arxiv : hep - ph/9702253 ; r. n. mohapatra and s. nussinov , phys . d * 60 * , 013002 ( 1999 ) [ arxiv : hep - ph/9809415 ] ; e. ma and m. raidal , phys . * 87 * , 011802 ( 2001 ) [ erratum - ibid . * 87 * , 159901 ( 2001 ) ] [ arxiv : hep - ph/0102255 ] ; c. s. lam , phys . b * 507 * , 214 ( 2001 ) [ arxiv : hep - ph/0104116 ] ; p. f. harrison and w. g. scott , phys . b * 547 * , 219 ( 2002 ) [ arxiv : hep - ph/0210197 ] ; y. koide , phys . d * 69 * , 093001 ( 2004 ) [ arxiv : hep - ph/0312207 ] ; t. kitabayashi and m. yasue , phys . d * 67 * , 015006 ( 2003 ) [ arxiv : hep - ph/0209294 ] . b * 621 * , 133 ( 2005 ) [ arxiv : hep - ph/0504212 ] . r. n. mohapatra , jhep * 0410 * , 027 ( 2004 ) [ arxiv : hep - ph/0408187 ] ; r. n. mohapatra and w. rodejohann , phys . d * 72 * , 053001 ( 2005 ) [ arxiv : hep - ph/0507312 ] ; s. choubey and w. rodejohann , eur . j. c * 40 * , 259 ( 2005 ) [ arxiv : hep - ph/0411190 ] ; k. matsuda and h. nishiura , phys . d * 72 * , 033011 ( 2005 ) [ arxiv : hep - ph/0506192 ] ; arxiv : hep - ph/0511338 . m. raidal , phys . lett . * 93 * , 161801 ( 2004 ) [ arxiv : hep - ph/0404046 ] ; h. minakata and a. y. smirnov , phys . d * 70 * , 073009 ( 2004 ) [ arxiv : hep - ph/0405088 ] . + for further references , see e.g. , h. minakata , arxiv : hep - ph/0505262 . m. h. ahn _ [ k2k collaboration ] , phys . lett . * 90 * , 041801 ( 2003 ) [ arxiv : hep - ex/0212007 ] ; e. aliu _ et al . _ [ k2k collaboration ] , phys . rev . * 94 * , 081802 ( 2005 ) [ arxiv : hep - ex/0411038 ] . s. antusch , p. huber , j. kersten , t. schwetz and w. winter , phys . d * 70 * , 097302 ( 2004 ) [ arxiv : hep - ph/0404268 ] ; p. huber , m. lindner , m. rolinec , t. schwetz and w. winter , phys . d * 70 * , 073014 ( 2004 ) [ arxiv : hep - ph/0403068 ] . h. minakata , phys . rev . * d52 * , 6630 ( 1995 ) [ arxiv : hep - ph/9503417 ] ; phys . * b356 * , 61 ( 1995 ) [ arxiv : hep - ph/9504222 ] ; s. m. bilenky , a. bottino , c. giunti , and c. w. kim , phys . * b356 * , 273 ( 1995 ) [ arxiv : hep - ph/9504405 ] ; k. s. babu , j. c. pati , and f. wilczek , phys . * b359 * , 351 ( 1995 ) [ arxiv : hep - ph/9505334 ] ; g. l. fogli , e. lisi , and g. scioscia , phys . d * 52 * , 5334 ( 1995 ) [ arxiv : hep - ph/9506350 ] . t. kobayashi , j. phys . g29 , 1493 ( 2003 ) ; s. mine , talk presented at the neutrino session of np04 workshop , aug . 2004 , kek , tsukuba , japan ( http://jnusrv01.kek.jp/jhfnu/np04nu/ ) ; t. kobayashi , talk presented at the 8th neutrino workshop ( in japanese ) , nov . 2001 , icrr , kashiwa , japan . o. l. g. peres and a. y. smirnov , phys . b * 456 * , 204 ( 1999 ) [ arxiv : hep - ph/9902312 ] ; nucl . phys . b * 680 * , 479 ( 2004 ) [ arxiv : hep - ph/0309312 ] ; m. c. gonzalez - garcia , m. maltoni and a. y. smirnov , phys . d * 70 * , 093005 ( 2004 ) [ arxiv : hep - ph/0408170 ] .

we present bell - type tests of nonclassicality and non - gaussianity for single - mode fields employing a generalized quasiprobability function . our nonclassicality tests are based on the observation that two orthogonal quadratures in phase space ( position and momentum ) behave as independent realistic variables for a coherent state . taking four ( three ) points at the vertices of a rectangle ( right triangle ) in phase space , our tests detect every pure nonclassical gaussian state and a range of mixed gaussian states . these tests also set an upper bound for all gaussian states and their mixtures , which thereby provide criteria for genuine quantum non - gaussianity . we optimize the non - gaussianity tests by employing a squeezing transformation in phase space that converts a rectangle ( right triangle ) to a parallelogram ( triangle ) , which enlarges the set of non - gaussian states detectable in our formulation . we address fundamental and practical limits of our generalized phase - space tests by looking into their relation with decoherence under a lossy gaussian channel and their robustness against finite data and non - optimal choice of phase - space points . furthermore , we demonstrate that our parallelogram test can identify useful resources for nonlocality testing in phase space .

1601.00279

describing a quantum state in phase space @xcite is a very useful approach to make a comparison between quantum mechanics and classical mechanics . it provides a valuable insight into the phenomenon of quantum - to - classical transition @xcite and a powerful tool to manifest nonclassical effects in quantum optics @xcite and continuous variable ( cv ) quantum informatics @xcite . one of the remarkable distinctions between quantum and classical phase - space distributions is that a negative value is allowed for a quantum state . although the negativity in phase space thus demonstrates nonclassicality immediately , it enables us to detect only a limited subset of nonclassical states . there exist nonclassical states with positive - definite distributions , e.g. , gaussian states with squeezing , which are readily accessible within current technology and provide important practical resources for cv quantum informatics @xcite . it is fundamentally and practically important to have a simple test manifesting nonclassicality @xcite beyond the negativity in phase space . in this respect , there was a seminal work by banaszek and wdkiewicz @xcite ( bw ) , who proposed a method to test the bell nonlocality directly in phase space . unlike the bell test using homodyne detection , which requires the transformation of a gaussian state to a non - gaussian state having a nonpositive wigner function @xcite , bw formalism enables us to detect nonclassical correlation even with a positive - definite wigner function . it has been extended to generalized quasiprobability functions @xcite and multipartite systems @xcite . recently , we have theoretically proposed and experimentally demonstrated a single - mode nonclassicality test using the wigner function @xcite in analogy with bw formalism . we here extend this recent work by using generalized quasiprobability functions . we not only give more details of the proposal in @xcite , but we also investigate other relevant aspects , e.g. , robustness of our tests against experimental imperfections including photon loss , finite data , and a nonoptimal choice of phase - space points . furthermore , we introduce an optimized test of genuine non - gaussianity employing three phase - space points , as an addition to the four - point test in @xcite . we also make a direct connection between our single - mode test and the bw nonlocality test , particularly showing that the single - mode nonclassical states detected under our parallelogram test can be a useful resource to make a two - mode state manifesting nonlocality under the bw test . our starting point is the observation that every pair of orthogonal quadratures in phase space behaves as independent realistic variables for a coherent state . exploiting it , we propose two nonclassicality tests that take four and three points at the vertices of a rectangle and a right triangle , respectively . our tests detect a broad range of nonclassical gaussian states , including all pure states . identifying the upper bounds for all gaussian states and their mixtures , we also propose tests for genuine quantum non - gaussianity . non - gaussian resources are known to be essential for many quantum informatic tasks , including universal cv quantum computation @xcite , entanglement distillation @xcite , quantum error correction @xcite , and cv nonlocality testing @xcite . a simple method to obtain a non - gaussian state would be to prepare a finite mixture of gaussian states . however , we can not claim such a state as a genuine non - gaussian resource . in dealing with non - gaussianity in quantum phase space , it is important to distinguish a genuinely quantum non - gaussian state from a mixture of gaussian states @xcite . we further optimize our non - gaussianity tests by taking points from a parallelogram ( triangle ) instead of a rectangle ( right triangle ) , which essentially realizes a squeezing operation on a given state without actually implementing it . we also discuss the fundamental and the practical limits of our tests . note that there exists a one - to - one correspondence between a @xmath0-parametrized phase - space distribution and a loss mechanism , i.e. , interaction with a vacuum reservoir @xcite . this correspondence might suggest that the limits of the @xmath0-parametrized distributions are the same as those of the wigner function ( @xmath1 ) under a lossy gaussian channel . however , they can yield different results for a nontrivial test . we show that the former can detect more states than the latter when a test sets a bound varying with the parameter @xmath0 , e.g. , the case of our non - gaussianity tests . we demonstrate that our tests are able to detect a nonclassical state reliably even when the number of data is finite and the measurement setting deviates from the optimal setting . we also show that our parallelogram test can detect genuine non - gaussianity for a range of superposition states with loss above 50% , at which the wigner function becomes positive definite . finally , we show that our parallelogram test can identify useful resources for a nonlocality test in phase space . it may open a direction for future works , e.g. , on a deeper understanding of the relation between nonclassicality and nonlocality . the @xmath0-parametrized quasiprobability function of a quantum state @xmath2 is defined as @xcite @xmath3,\ ] ] where @xmath4 is the displacement operator with complex amplitude @xmath5 . the operator @xmath6 is given by @xmath7 which yields , e.g. , a parity operator @xmath8 and a vacuum state @xmath9 for @xmath10 and @xmath11 , corresponding to the well - known wigner and @xmath12 functions , respectively . as the eigenvalues of @xmath6 are @xmath13 ( @xmath14 : non - negative integers ) , the @xmath0-parametrized quasiprobability function is bounded as @xmath15 this shows that the lower bound becomes minimum for the wigner function ( @xmath1 ) and approaches zero with @xmath0 decreasing . the above equation also clearly tells us that the @xmath12 function ( @xmath16 ) is non - negative . while the parameter @xmath0 can have a positive value up to 1 ( @xmath17 function ) , we only deal with a nonpositive @xmath0 throughout the paper as the eigenvalues of @xmath18 become unbounded for @xmath19 . , the remaining three points form a right triangle in eq . . ( b ) optimal choice of rectangle for testing a squeezed state occurs at @xmath20 ( @xmath21 : squeezing axis ) , that is , when the axes of the rectangle are oriented midway between the squeezed and the antisqueezed axes . ] interestingly , the @xmath0-parametrized quasiprobability function of a coherent state @xmath22 is factorized as a product of two gaussian distributions for every pair of orthogonal quadratures , @xmath23 where @xmath24 is a coordinate system rotated by an angle @xmath25 from @xmath26 [ fig . 1 ( a ) ] , with a rotation matrix @xmath27 and @xmath28 $ ] and @xmath29 $ ] . it is thus possible to consider the generalized quasiprobability function of a coherent state as a product of two independent random variables @xmath30 and @xmath31 , that is , @xmath32 with @xmath33 . + we then construct a linear sum of @xmath34parametrized functions at four phase - space points as @xmath35 & \equiv \frac{\pi ( 1-s)}{2 } \ { w_{\rho } ( x_{0 } , y_{0 } ; s ) + w_{\rho } ( x_{1 } , y_{0 } ; s ) \nonumber \\ & + w_{\rho } ( x_{0 } , y_{1 } ; s ) - w_{\rho } ( x_{1 } , y_{1 } ; s ) \ } , \end{aligned}\ ] ] where the points constitutes a rectangle oriented at angle @xmath25 in phase space as depicted in fig . [ fig : rectangle ] . we then obtain @xmath36 for a coherent state , which has the same form as the clauser - horne - shimony - holt ( chsh ) inequality @xcite . using @xmath37 , we obtain @xmath38 for a coherent state as follows . ( i ) for @xmath39 , we have @xmath40 . the minimum is given by , e.g. , @xmath41 and @xmath42 . ( ii ) for @xmath43 , we have @xmath44 . the maximum is given by , e.g. , @xmath45 . we extend the above result to an arbitrary classical state , i.e. a mixture of coherent states , @xmath46 , where @xmath47 is a coherent state . as the coherent amplitude @xmath48 behaves like a hidden variable , @xmath49 we obtain a classicality condition as @xmath50 \leq 2.\ ] ] in other words , the violation of eq . demonstrates the nonclassicality of a single - mode state . at this point , it may be intriguing to ask how many phase - space points should be considered to come up with a meaningful nonclassicality test , particularly to test a positive quasiprobability distribution . can we obtain a useful nonclassicality criterion employing less numbers of points than four in eq . ? of course , verifying nonclassicality from an _ arbitrary _ set of points is impossible without specifications on the chosen points @xmath51 , like a predetermined position ( origin ) with energy constraint in @xcite or a designated shape ( rectangle ) in our case @xcite . for example , if we construct a test exploiting the values at fully arbitrary @xmath52 points without specifying locations , that is , @xmath53 $ ] where @xmath54 with @xmath55 , every result from a set of positive values ( @xmath56 ) is mimicked by a single vacuum state because the same values can be found at @xmath57 for @xmath55 . in this sense , if we intend to introduce a specified shape as a constraint , the least number of points is possibly three with the shape of triangle , whereas the test in eq . adopts a rectangle with four points . + thus we also introduce a three - points ( right triangle ) test as @xmath58 & \equiv \frac{\pi ( 1-s)}{2 } \ { w_{s } ( x_{1 } , y_{0 } ) + w_{s } ( x_{0 } , y_{1 } ) \nonumber \\ & - w_{s } ( x_{1 } , y_{1 } ) \ } , \end{aligned}\ ] ] which excludes one point @xmath59 from eq . we then have a structure @xmath60 for a coherent state . using @xmath33 again , we obtain @xmath61 as follows : ( i ) for @xmath39 , we have @xmath62 . the minimum is achieved by @xmath41 and @xmath42 . ( ii ) for @xmath43 , we have @xmath63 . the maximum is achieved by @xmath45 . therefore , similar to eq . , we obtain another classicality condition as @xmath64 \leq 1.\ ] ] we note that , contrary to the rectangle test in eq . , it has no analogy with a nonlocality test , as @xmath65 is saturated by a hidden variable theory : @xmath66 yields @xmath67 , which are also the quantum bounds due to @xmath68 . note that the optimal values of @xmath69 $ ] and @xmath70 $ ] for a given state @xmath2 are invariant under displacement and rotation . let us assume that a state @xmath2 has an optimal value at points @xmath71 , and then a displaced state @xmath72 has the same optimum at shifted points @xmath73 . this is because the displacement operator only translates the center of the quasiprobability function while preserving its entire profile . similarly , if a state has the optimal value at points @xmath71 where the coordinate system is oriented at angle @xmath25 , a rotated state @xmath74 has the same optimum in the coordinate system now oriented at angle @xmath75 , since the phase - rotation also preserves the profile of the quasiprobability function . these invariance properties can be useful to simplify the analysis of nonclassicality tests for a given state . we first demonstrate how our tests @xmath76 and @xmath77 can detect a wide range of gaussian states . a single - mode gaussian state @xmath78 is fully characterized by its first - order moments ( averages ) @xmath79 and @xmath80 , and second - order moments represented by a covariance matrix @xmath81 with elements @xmath82 where @xmath83 with @xmath84 and @xmath85 . its @xmath0-parametrized distribution is given by a gaussian function as @xmath86 , \end{aligned}\ ] ] where @xmath87 , @xmath88 and @xmath89 . alternatively , a single - mode gaussian state @xmath78 can be represented as a displaced squeezed thermal state , @xmath90 where @xmath91 $ ] is the squeezing operator ( @xmath92 : squeezing strength , @xmath21 : squeezing axis ) , and @xmath93 is a thermal state with mean photon number @xmath94 . for a gaussian state with parameters @xmath95 , its first moments are given by @xmath96 $ ] and @xmath97 $ ] , and the covariance matrix elements by @xmath98 which yields @xmath99 equation shows that the overall factor of @xmath100 in eq . ( 12 ) is bounded by @xmath101 . its maximum @xmath102 is achieved by every pure gaussian state ( @xmath103 ) for @xmath1 , as @xmath104 represents the purity of a gaussian state @xmath105 . on the other hand , only a vacuum state attains the maximum @xmath102 for @xmath106 . in general , with @xmath92 and @xmath0 fixed , @xmath107 increases with purity ( @xmath94 decreasing ) . rewriting eq . using the rotated quadratures @xmath108 , we obtain @xmath109 , \end{aligned}\ ] ] where @xmath110 is the covariance matrix in a rotated frame and @xmath111 , \mathrm{im } [ \alpha e^{-i \theta } ] ) $ ] . from now on , we set @xmath112 , as the displacement operation has no effect on the optimal values ( sec . ii b 1 ) . every gaussian function in the form of eq . can be recast to @xmath113,\ ] ] by introducing rescaled variables , @xmath114 and the parameter @xmath115 from the rescaled distributions in eq . ( 17 ) , we see that two parameters , i.e. the overall factor @xmath116 and the parameter @xmath117 , determine optimal values @xmath76 and @xmath77 for a given gaussian state . in fact , we can show that these optimal values monotonically increase with @xmath117 as well as @xmath116 @xcite . we point out that the parameter @xmath117 for every gaussian state is bounded by @xmath118 . in particular , with @xmath0 and @xmath14 fixed , the parameter @xmath117 is bounded by @xmath119 where the upper bound @xmath120 is obtained at the choice of angle @xmath20 ( see fig . [ fig : rectangle ] ) . this optimal choice of angle intuitively makes sense , as our nonclassicality tests rely on the _ nonfactorizability _ of the quasi - probability distributions in phase space . in contrast , if we take @xmath121 or @xmath122 , the quasiprobability function in eq . is factorized to a form @xmath123 , where the two quadratures @xmath124 and @xmath125 behave as independent variables , yielding no violation of our nonclassicality tests . + for @xmath1 ( wigner function ) , the overall factor @xmath126 depends only on purity ; thus large @xmath127 and @xmath128 occur for a pure state . on the other hand , the optimal @xmath129 in eq . ( 20 ) depend only on the degree of squeezing . in figs . [ fig : gaussian](a ) and [ fig : gaussian](b ) , we show that the optimal @xmath127 and @xmath128 monotonically increase with @xmath130 ( degree of squeezing ) . they rise up to the maximum values @xmath131 and @xmath132 for the rectangle and the right triangle tests , respectively , both achieved at an infinite squeezing @xmath133 ( @xmath134 ) . ( see supplemental matrerial of ref . @xcite for rigorous proofs of the maximal values . ) note that the gaussian bound @xmath135 for the four - points test coincides with the maximal value of a gaussian state for two - mode @xcite and three - mode @xcite nonlocality tests in phase space . on the other hand , for @xmath106 , the overall factor @xmath107 and the ratio @xmath120 involve both the purity @xmath136 and the squeezing strength @xmath92 . with purity and squeezing fixed , both @xmath107 and @xmath120 decrease with the parameter @xmath0 decreasing . we thus observe that maximum values among all gaussian states for rectangle and right triangle tests , respectively , become smaller with the parameter @xmath0 decreasing in fig . [ fig : gaussian](c ) and [ fig : gaussian](d ) , respectively . in addition , although the optimal @xmath137 in eq . ( 20 ) increases with the thermal photon @xmath94 for @xmath138 , the overall factor @xmath116 in eq . ( 15 ) decreases with @xmath94 , which eventually makes the case of a pure state ( @xmath103 ) optimal for given @xmath92 and @xmath0 . we plot the optimal values of @xmath76 [ fig . [ fig : gaussian ] ( e ) ] and @xmath77 [ fig . [ fig : gaussian ] ( f ) ] with respect to parameter @xmath0 and squeezing strength @xmath139 . in these contour plots , we show how the optimal squeezing @xmath130 for each maximum changes with @xmath0 ( black dashed lines ) . interestingly , we note that the optimal violation for a nonzero @xmath138 occurs at a finite squeezing , similar to the case of a two - mode nonlocality test @xcite . with the parameter @xmath0 decreasing , the maximum values and the corresponding squeezing strength become smaller . at @xmath16 and @xmath140 , there is no gaussian state violating the rectangle test and the triangle test , respectively , which suggests that the right - triangle test is more useful practically for detecting gaussian states . in figs . [ fig : gaussian](g ) and [ fig : gaussian](h ) , we also identify the range of mixed gaussian states ( colored region ) that can be detected under our nonclassicality tests . specifically , we plot the critical parameter @xmath141 for each gaussian state with purity @xmath136 and squeezing @xmath142 , above which its nonclassicality can successfully be detected , i.e. , in the range @xmath143 $ ] . as the purity @xmath136 decreases , we see that the squeezing level @xmath142 required for a successful test becomes higher . in addition , the two contour plots ( g ) and ( h ) in comparison show that right triangle test detects more gaussian states than the rectangle test . the range of successful detection may be attributed to the ratio of maximum gaussian bound to classical bound in each test . the rectangle test gives the ratio @xmath144 , whereas the right - triangle test gives @xmath145 , which may account for the resilience of the latter test compared to the former . for instance , violating @xmath146 and @xmath147 becomes impossible if the purity falls below the inverse of the raios , @xmath148 and @xmath149 , respectively , which has been numerically confirmed . + in addition , fig . [ fig : gaussian ] shows that wigner function ( @xmath1 ) is optimal among all @xmath0-parametrized distributions for both tests , which makes sense as the @xmath0-parametrized quasiprobability function is closely related to loss dynamics under a gaussian reservoir . the @xmath0-parametrized quasiprobability functions of a quantum state @xmath2 with two different parameters @xmath150 and @xmath151 @xmath152 are related by a gaussian convolution @xcite , @xmath153 equation is similar to the action of a loss channel @xmath154 on a state @xmath2 , @xmath155 } ( \alpha ; 0 ) \nonumber \\ & = \frac{2}{\pi ( 1 - \eta ) } \int d^{2 } \beta w_{\rho } ( \beta ; 0 ) e^{- \frac{2}{1 - \eta } | \alpha - \sqrt{\eta } \beta |^{2 } } , \end{aligned}\ ] ] where the loss channel @xmath154 is modeled by mixing the input state @xmath2 and a vacuum at a beam splitter with transmittance @xmath156 . it provides a direct connection between the generalized quasiprobability function and loss dynamics as @xcite @xmath157 } ( \alpha ; s ) = \frac{1}{\eta } w_{\rho } \bigg ( \frac{\alpha}{\sqrt{\eta } } ; 1 - \frac{1-s}{\eta } \bigg).\ ] ] inverting eq . by setting @xmath10 and @xmath158 , we obtain @xmath159 } ( \alpha ; 0 ) , \ ] ] which reveals that @xmath76 and @xmath77 in eqs . and correspond to loss dynamics of @xmath127 and @xmath128 , respectively , with @xmath160 . the results in fig . [ fig : gaussian ] thus identify the ultimate limit of our tests for gaussian states under a lossy channel . that is , as the rectangle and the triangle tests for gaussian states have critical values @xmath161 and @xmath162 , respectively , we have 3 db ( @xmath163 ) and 4.77 db ( @xmath164 ) loss limits , below which there exist some gaussian states detectable using our tests . the different limits 3 db and 4.77 db manifest a practical superiority of the triangle test to the rectangle test . + now let us further investigate to what extent our tests can be useful under practical conditions . first , the number @xmath52 of data to construct an average value is always finite , incurring an error of order @xmath165 . we thus require that the degree of violation is large enough to overcome the statistical error as @xmath166 where @xmath167 ( @xmath168 ) and @xmath169 with the classicality bound @xmath170 and 1 for the rectangle and the right - triangle test , respectively . moreover , although we have previously obtained the optimal choice of angle @xmath171 as shown in fig . 1 , it is of practical interest to identify the angle tolerance @xmath172 , i.e. , a successful detection in the range of angles @xmath173 . this is a particularly important issue when there is no information on the phase ( squeezing angle @xmath21 ) of the state . in this case , our choice of angle @xmath25 for a rectangle ( triangle ) in fig . 1 becomes completely random . a worst case would be the choice of @xmath174 or @xmath122 at which no violation occurs due to the factorizability of the phase - space function , as explained below eq . ( 20 ) . for given purity @xmath136 and squeezing @xmath142 , we may take a fixed dimension of rectangle ( triangle ) , like the one used for an optimal test with known phase , but consider the angle @xmath25 randomly distributed over the whole range of @xmath122 . we can then measure a success probability as @xmath175 . in fig . [ fig : fgad ] , we show how faithfully our wigner - function tests @xmath176 in ( a , c , e , g ) and @xmath177 in ( b , d , f , h ) detect gaussian states with unknown phase by evaluating @xmath178 for a data number @xmath179 ( a , b ) , @xmath180 ( c , d ) , @xmath181 ( e , f ) , and @xmath182 ( g , h ) . we see that our tests can confidently detect a range of mixed squeezed states with a practical number @xmath183 . in general , the angle tolerance @xmath172 , and thus the success probability @xmath178 , becomes large by increasing the data number @xmath52 as well as the purity @xmath136 and squeezing strength @xmath184 . as already identified in fig . 2(g ) and 2(h ) , the theoretical limits of purity for a successful test of @xmath185 and @xmath186 are @xmath187 and @xmath188 , respectively , which are achievable with @xmath52 growing . in fig . [ fig : fgad ] , we already have similar levels of critical purity @xmath189 ( e ) and 0.516 ( f ) with a finite data @xmath181 . for the rectangle test , a high level @xmath190 means that our test can be successful unless the randomly chosen angle is too close to the one for the factorized wigner function , @xmath174 or @xmath122 . for the triangle test , the success probability @xmath178 is smaller , however , the range of mixed states detectable is larger than that of the rectangle test . in fig . [ fig : fgadq ] , we also show the results for a triangle test @xmath191 based on the @xmath12 function ( @xmath16 ) . note that we have excluded the rectangle test @xmath192 , since no violation occurs as shown in fig . 2 . quite naturally ( see sec . ii c 2 ) , both the detectable range of mixed gaussian states and the success probability significantly shrink compared to the case of wigner - function tests . for a rectangle test @xmath176 [ ( a ) , ( c ) , ( e ) , and ( g ) ] and a triangle test @xmath177 [ ( b ) , ( d ) , ( f ) , and ( h ) ] to detect the nonclassicality of a gaussian state with purity @xmath136 and squeezing strength @xmath184 when the phase is unknown . we have used the number of data @xmath52 in eq . as @xmath179 ( a , b ) , @xmath180 ( c , d ) , @xmath181 ( e , f ) , and @xmath182 ( g , h ) . ] for a triangle test @xmath193 to detect the nonclassicality of a gaussian state with purity @xmath136 and squeezing strength @xmath184 when the phase is unknown . we have used the number of data @xmath52 in eq . as @xmath180 ( a ) , @xmath181 ( b ) , @xmath182 ( c ) , and @xmath194 ( d ) . ] we now go beyond the gaussian regime and investigate a maximal possible value of @xmath195 among all quantum states beyond gaussian states . to find out the maximum and the minimum values , we solve eigenvalue equations @xmath196 for hermitian operators @xmath197 and @xmath198 that correspond to rectangle and right triangle tests , respectively . these hermitian operators are given by @xmath199 and @xmath200 from now on , without loss of generality , we set @xmath201 since the optimal values in our tests are invariant under displacement as mentioned in sec . ii b 1 . of @xmath202 in eq . that are obtained by truncating @xmath14 and @xmath203 up to @xmath52 in the recurrence relation of eq . . ( b ) expectation values @xmath204 of a truncated superposition of @xmath205 coherent states in eq . compared with @xmath206 in ( a ) . as an example , we plot ( c ) the wigner function and ( d ) its contour plot for a superposition of @xmath207 coherent states with @xmath208 , which achieves a value of @xmath209 \approx 3.70 $ ] . ] for the case of the wigner function ( @xmath1 ) , we can solve the eigenvalue problem on the basis of coherent states . we first construct a trial solution as @xmath210 where the coherent states @xmath211 with integers @xmath14 and @xmath203 form a two - dimensional lattice with points spaced by @xmath212 and @xmath213 along the @xmath214 and @xmath215 axes , respectively . exploiting an identity @xmath216 , we derive a recurrence relation @xmath217 where @xmath218 refers to the area of unit cell in the lattice . in fig . [ fig : qb0 ] , we show the maximum and minimum eigenvalues attainable by taking a truncation number @xmath52 in the recurrence relations with @xmath219 ( @xmath220 : integer ) . it shows that the simple algebraic bounds @xmath221 for a rectangle test are asymptotically obtained by increasing @xmath52 . as a double check , we compare @xmath206 and @xmath222 with @xmath223 , where @xmath224 is given by plugging the coefficients @xmath225 obtained from the recurrence relations . that is , we construct a state with certain numeric coefficients and caluclate its actual average value @xmath226 of @xmath227 . figure [ fig : qb0 ] confirms that the algebraic bounds are actually obtained by the states in eq . . in addition , those states also achieve the algebraic bounds @xmath67 for a right triangle test ( fig . 6 ) . on the other hand , for a generalized distribution @xmath106 , we solve the eigenvalue problems in the basis of fock states . to this aim , we first express @xmath228 as @xmath229 where @xmath230 $ ] for @xmath231 is given by @xmath232 and @xmath233 for @xmath234 is obtained by @xmath235 . using eq . , we construct a density matrix for @xmath197 and @xmath198 , and have obtained the maximum and minimum eigenvalues for each generalized quasiprobability function by extensive numerical optimizations . in fig . [ fig : qb1 ] , the maximal @xmath76 and @xmath77 for quantum states approach classical bounds as @xmath0 decreases , which implies that the whole set of quantum states detectable under our tests shrinks with @xmath0 decreasing . ( @xmath77 ) and ( c , d ) minimal @xmath76 ( @xmath77 ) among all quantum states against the parameter @xmath0 ( orange solid line ) , compared to the algebraic ( black dotted lines ) and gaussian bounds ( brown dashed lines ) , respectively . the quantum bounds reach the algebraic upper bounds at @xmath1 for each case . in ( a ) and ( b ) , the gap between quantum and classical bounds disappears at @xmath16 and @xmath140 for @xmath76 and @xmath77 , respectively . ] for comparison , we also plot the gaussian bounds obtained in the previous sections and the algebraic bounds obtainable from eq . . the algebraically possible ranges of @xmath76 and @xmath77 are given by @xmath236 and @xmath237 while the maximum and the minimum algebraic bounds are saturated by quantum states for @xmath1 , there are gaps between algebraic and quantum bounds for a nonzero @xmath106 . as for the lower bounds of @xmath76 and @xmath77 , the vanishing gap between algebraic and classical bounds clearly indicate that there is no quantum violation below @xmath16 . on the other hand , as for the upper bounds , we observe that the loss limits of @xmath76 and @xmath77 for the whole set of quantum states are identical to the loss limits for gaussian states , 3 db ( @xmath16 ) for @xmath76 and 4.77 db ( @xmath140 ) for @xmath77 , respectively . that is , below those limits , there exist some quantum states , both gaussian ( sec . ii c 2 ) and non - gaussian , violating the classical bounds 2 and 1 , respectively . and @xmath238 for an even cat state @xmath239 with @xmath240 ( blue solid line ) , which are equivalent to @xmath127 and @xmath128 for a decohered state @xmath241 $ ] with loss parameter @xmath242 [ eq . ] , respectively . on the other hand , black dashed lines denote the gaussian bounds in eq . [ the same as fig . 2(c ) and 2(d ) ] for each @xmath243 , above which genuine non - gaussianity is detected . for a non - gaussianity test , the detectable range of @xmath244 based on an @xmath0-parametrzied function for the original pure cat state ( red shaded region ) is larger than the range of @xmath156 based on the wigner function for the decohered state under loss ( brown shaded region ) . see main text . ] as our tests are linear with respect to a convex mixture of quantum states , i.e. , @xmath245 = \sum_{i } p_{i } \mathcal{j}_{s } [ \rho_{i } ] $ ] , the gaps between the quantum and gaussian bounds enable us to detect quantum non - gaussianity . that is , if a given state is a mixture of gaussian states , it must satisfy @xmath246 \leq \max_{\sigma } \mathcal{j}_{s } , \nonumber \\ -1 < & \mathcal{j}_{s}^{\prime } [ \rho_{\mathrm{mg } } = \sum_{i } p_{i } \sigma_{i } ] \leq \max_{\sigma } \mathcal{j}_{s}^{\prime } , \end{aligned}\ ] ] where @xmath247 is an arbitrary gaussian state and the maximum values , @xmath248 and @xmath249 , are shown in figs . 2(c ) and 2(d ) , respectively . in particular , the case of @xmath1 gives @xmath250 \leq \frac{8}{3^{9/8 } } , \nonumber \\ -1 < & \mathcal{j}_{0}^{\prime } [ \rho_{\mathrm{mg } } ] \leq 2 . \end{aligned}\ ] ] in fig . [ fig : cld ] , as an example , we plot the violation of inequalities and for an even cat state @xmath239 under a lossy channel , which is represented by a wigner function , @xmath251 } ( q , p ; s=0 ) \nonumber \\ & = \frac{2}{\pi } \frac{1}{1 + e^{-2\gamma^{2 } } } e^{-2 q^{2 } - 2 p^{2 } } \ { e^{-2\eta \gamma^{2 } } \cosh ( 4 \sqrt{\eta } \gamma q ) \nonumber \\ & + e^{-2(1-\eta ) \gamma^{2 } } \cos ( 4 \sqrt{\eta } \gamma p ) \ } , \end{aligned}\ ] ] with @xmath252 real . we particularly compare ( i ) the wigner - function - based tests @xmath176 and @xmath177 of the decohered state @xmath241 $ ] and ( ii ) the @xmath0-parametrized - function - based @xmath76 and @xmath77 of the original pure state @xmath253 with @xmath243 , respectively . in sec . ii c 2 , we have shown the equivalence of two nonclassicality tests the wigner - function test for the decohered state and the @xmath0-parametrized function test for the original state . in contrast , there exists inequivalence for non - gaussianity tests the latter detects non - gaussianity in a broader parameter regime than the former . it is due to the fact that quantum non - gaussianity bounds , @xmath248 and @xmath249 , vary with the parameter @xmath0 , whereas the nonclassicality bounds are the same regardless of @xmath0 . on a practical side , it implies that the non - gaussianity test for a decohered state under a lossy channel can be harder than expected from the analysis only based on the @xmath0-parametrized function of the original state . and @xmath128 optimized over the points of a rectangle ( right triangle ) for @xmath254 as functions of the vacuum fraction @xmath255 . blue solid , purple dotted , and red dashed lines denote the case of squeezing applied to a given state , with squeezing @xmath256 ( a ) and @xmath257 ( b ) , respectively . we find no violation of eq . for an initial state ( @xmath258 ) , while a state with a sufficiently large squeezing violates eq . ( blue colored region ) even when its wigner function is positive ( @xmath259 ) . ( c , d ) @xmath260 and @xmath261 optimized over the points of a parallelogram ( triangle ) in eqs . ( 41 ) and ( 42 ) for @xmath262 with respect to @xmath255 . black solid , orange dotted , brown dashed lines represent @xmath263 ( c ) and @xmath264 ( d ) , respectively . the wigner function @xmath1 shows the best performance to demonstrate quantum non - gaussianity . ] the test of genuine non - gaussianity in eq . can be further enhanced by applying squeezing @xmath265 on a given state . note that a mixture of gaussian state remains to be a gaussian mixture under gaussian operations . thus , if the state under squeezing shows violation of gaussian bounds , the original state must be genuinely non - gaussian . for example , we show the case of the mixed state @xmath266 in figs . [ fig : examples](a ) and [ fig : examples](b ) , which shows that the squeezing operation on the state enhances the detected region . for the case of a wigner function , we may address the problem by using its interesting property @xcite @xmath267 ; 0 ) , \ ] ] where @xmath268 = \alpha \cosh r + \alpha^ { * } e^{2i\phi } \sinh r$ ] is the transformation of phase - space points due to the squeezing operation @xmath265 . as we illustrate in fig . [ fig : squeezing ] , under the squeezing transformation @xmath268 $ ] , a parallelogram in an unsqueezed profile corresponds to a rectangle in a squeezed profile . importantly , it means that we do not need to implement a squeezing operation on a given state in order to have an enhanced test . instead , we may simply choose the four points at the vertices of the parallelogram corresponding to the squeezing operation and do our tests for the given initial state . on the other hand , for a nonzero @xmath138 , the above argument is not directly applicable . this is because the @xmath0-parametrized function of a state after squeezing is not simply obtained by reshaping of the original @xmath0-parametrized function ( squeezing of the profile ) . instead , we find the following identity ( with its proof in the appendix ) : @xmath269 ; s ) = \frac{\pi}{2 } w_{\mathcal{l}^{\prime } [ \hat{s } \rho \hat{s}^{\dag } ] } \bigg ( \frac{\alpha}{\sqrt{1-s } } ; 0 \bigg),\ ] ] where @xmath270 $ ] represents a beam - splitter interaction with transmittance @xmath271 between a quantum state @xmath2 and a squeezed vacuum @xmath272 . therefore , all @xmath0-parametrized functions taken at the vertices of a paralleogram can be understood as the wigner function taken at the vertices of a rectangle for the decohered state @xmath273 $ ] . as the whole process @xmath273 $ ] is gaussian , it does not create non - gaussianity . we then use the gaussian bounds obtained for @xmath1 in order to detect genuine non - gaussianity under the @xmath0-parametrized functions . we thus propose enhanced non - gaussianity tests for an arbitrary @xmath0 as @xmath274 \leq \frac{8}{3^{9/8 } } , \nonumber \\ -1 < & \mathcal{n}_{s}^{\prime } [ \rho_{\mathrm{mg } } ] \leq 2 , \end{aligned}\ ] ] with @xmath275 & = \frac{\pi ( 1-s)}{2 } \ { w_{\rho } ( s [ q_{0 } + i p_{0 } ] ; s ) + w_{\rho } ( s [ q_{1 } + i p_{0 } ] ; s ) \nonumber \\ & + w_{\rho } ( s [ q_{0 } + i p_{1 } ] ; s ) - w_{\rho } ( s [ q_{1 } + i p_{1 } ] ; s ) \ } , \end{aligned}\ ] ] and @xmath276 & = \frac{\pi ( 1-s)}{2 } \ { w_{\rho } ( s [ q_{1 } + i p_{0 } ] ; s ) + w_{\rho } ( s [ q_{0 } + i p_{1 } ] ; s ) \nonumber \\ & - w_{\rho } ( s [ q_{1 } + i p_{1 } ] ; s ) \ } , \end{aligned}\ ] ] which take into account the vertices of parallelogram and triangle , respectively , with @xmath268 = \alpha \cosh r + \alpha^ { * } e^{2i\phi } \sinh r$ ] . note that any values of @xmath92 and @xmath21 for the squeezing transformation @xmath268 $ ] can be used in eqs . and , as squeezing does not create non - gaussianity . as an illustration , we show the case of the state @xmath266 in figs . [ fig : examples](c ) and [ fig : examples](d ) using the above tests . we here investigate our test of genuine non - gaussianity for noisy non - gaussian states . in particular , when a single - mode state undergoes a lossy gaussian channel , the corresponding wigner function becomes positive definite if its transmission rate @xmath156 is below 50% . it is then interesting to know whether our non - gaussianity test can detect noisy non - gaussian states for @xmath277 . for this purpose , we consider a class of superposition states @xmath278 ( @xmath52 : truncation number ) @xcite as an input to a lossy channel . we then test the genuine non - gaussianity for the output decohered states using the wigner function ( @xmath1 ) or the @xmath12 function ( @xmath16 ) in eqs . and . ( parallelogram test ) , ( b ) @xmath279 ( triangle test ) , and ( c ) @xmath280 ( parallelogram test ) , respectively , optimized over a superposition state @xmath281 for each @xmath156 ( transmission rate ) under a loss channel . the curves from top to bottom represent the case of truncation number @xmath282 ( black dotted ) , @xmath283 ( red dashed ) and @xmath284 ( blue solid ) . @xmath285 and @xmath286 , respectively , represent genuine non - gaussianity of the noisy non - gaussian states . ] in fig . [ fig : ot ] ( a ) , we plot @xmath287 ( parallelogram test ) for the superposition states @xmath281 with varying truncation number @xmath284 ( blue solid ) , @xmath283 ( red dashed ) , and @xmath282 ( black dotted ) . for each @xmath156 , we optimize the coefficients @xmath288 to show a maximal @xmath289 in the figures . we see that @xmath289 is above the gaussian bound @xmath135 even below @xmath290 ( magnified view in the inset ) , demonstrating a successful detection of non - gaussianity for positive wigner functions . the plots indicate that there exist some superposition states for each @xmath52 whose non - gaussianity can be detected under transmission below 50% . we have numerically found that @xmath291 appears at a very low @xmath156 . if we take into account the violation of size @xmath292 , the critical @xmath156 turns out to be @xmath293 ( @xmath284 ) , @xmath294 ( @xmath283 ) , and @xmath295 ( @xmath282 ) , respectively . on the other hand , under the triangle test @xmath296 , the detection becomes less effective as shown in fig . [ fig : ot ] ( b ) . in particular , the critical @xmath156 does not go below 0.5 in contrast to the rectangle test @xmath289 in fig . [ fig : ot ] ( a ) . moving on to the @xmath12-function tests ( @xmath16 ) , they are naturally less powerful than the wigner - function test ( @xmath1 ) . however , they can also manifest genuine non - gaussianity below @xmath290 under the paralleogram test as shown in fig . [ fig : ot ] ( c ) , with critical values @xmath297 ( @xmath283 ) and @xmath298 ( @xmath282 ) at the violation of size 0.001 ( no violation for @xmath284 ) . we do not show the case of triangle test @xmath299 , which does not detect non - gaussianity . thus , for the class of noisy non - gaussian states considered here , we find that the parallelogram test with four phase - space points is more powerful than the triangle test with three phase - space points . finally , we address the connection between our single - mode nonclassicality test and the bw two - mode nonlocality test more directly . we start our discussion with a single - mode state @xmath2 that satisfies a condition @xmath300 > 2 e^{\frac{1}{2 } \max [ | d_{a } | , | d_{b } | ] ^{2}},\ ] ] where @xmath301 $ ] is the quantity defined in eq . ( @xmath1 ) , and @xmath302 - s [ q_{0 } + i p_{0 } ] $ ] and @xmath303 - s [ q_{0 } + i p_{1 } ] $ ] correspond to the diagonal lengths of a parallelogram with the transformation @xmath268 = \alpha \cosh r + \alpha^ { * } e^{2i\phi } \sinh r$ ] . for the case of taking @xmath258 , the four points simply form a rectangle ; then @xmath301 $ ] has a classical bound 2 . in other words , . becomes a nonclassicality test and the quantity @xmath304^{2}}>2 $ ] represents a degree of nonclassicality . on the other hand , for the case of taking @xmath305 , eq . may not be immediately regarded as a nonclassicality test , since a squeezing operation can create nonclassicality to the input state . nevertheless , we show below that if a state @xmath2 satisfies the condition in eq . for any values of @xmath92 and @xmath21 , it manifests nonlocality under the bw test @xmath306 & \equiv \frac{\pi^{2}}{4 } \ { w_{\rho_{ab } } ( \alpha_{0 } , \beta_{0 } ) + w_{\rho_{ab } } ( \alpha_{1 } , \beta_{0 } ) \nonumber \\ & + w_{\rho_{ab } } ( \alpha_{0 } , \beta_{1 } ) - w_{\rho_{ab } } ( \alpha_{1 } , \beta_{1 } ) \}>2 , \end{aligned}\ ] ] by mixing it with a vacuum at a 50:50 beam splitter . - 2 e^{\frac{1}{2 } \max [ | d_{a } | , | d_{b } | ] ^{2}}$ ] and @xmath307 for squeezed states as functions of squeezing strength @xmath92 . ( c , d ) optimal @xmath308 and @xmath307 for states @xmath262 as functions of the vacuum fraction @xmath255 . every @xmath309 in our single - mode test is monotonically connected to the violation of the nonlocality test @xmath310 in phase space . ] it has been known that a single - mode nonclassical state can be turned into an entangled state using a beam - splitter setting @xcite . by mixing a single - mode state @xmath2 with a vacuum at a 50:50 beam splitter , we obtain a two - mode wigner function as @xmath311 where @xmath312 is the wigner function of a vacuum state . let us assume that eq . is satisfied for the state @xmath2 . using a short - hand notation @xmath313\equiv s_{ij}$ ] , we choose the phase - space points for the bw test as @xmath314 and obtain @xmath315 & = \frac{\pi}{2 } \ { w_{\rho } ( s_{00 } ; 0 ) - w_{\rho } ( s_{11 } ; 0 ) \ } e^{- \frac{1}{2 } | d_{b } |^{2 } } \nonumber \\ & + \frac{\pi}{2 } \ { w_{\rho } ( s_{10 } ; 0 ) + w_{\rho } ( s_{01 } ; 0 ) \ } e^{- \frac{1}{2 } | d_{a } |^{2 } } \nonumber \\ & \geq \mathcal{n } [ \rho ] e^{- \frac{1}{2 } \max [ | d_{a } | , | d_{b } | ] ^{2 } } \nonumber \\ & > 2 , \end{aligned}\ ] ] with the identity @xmath316 . we have above used that @xmath317 and @xmath318 are both positive if @xmath301 > 2 $ ] ; if the former is negative , then @xmath301 < 2 $ ] due to the constraint @xmath319 . as a by - product , we now confirm that eq . represents a nonclassicality test for any values of @xmath92 and @xmath21 . this is because nonclassicality is a necessary condition to make a nonlocal resource in the beam - splitter setting . as an illustration , in fig . [ fig : cst ] we plot the optimal value of @xmath320 - 2 e^{\frac{1}{2 } \max [ | d_{a } | , | d_{b } | ] ^{2}}$ ] and the corresponding bw quantity @xmath307 for squeezed states and noisy fock states , respectively , which clearly manifests that our single - mode test can represent a nonlocal resource in phase space . we have proposed bell - type tests of non - classicality and non - gaussianity using generalized phase - space distributions . our rectangle and right - triangle tests are capable of detecting a wide range of nonclassical states including mixed gaussian squeezed states and non - gaussian states that possess even positive - definite distributions in phase space . for nonclassicality tests , we have identified the ultimate limits of 3-db ( transmittance @xmath321 ) and 4.77-db ( @xmath322 ) under a loss channel with our approach employing four and three phase - space points , respectively . furthermore , we have shown that our test can be robust against experimental imperfections , including finite data acquisition and nonoptimal choice of phase - space points . as our nonclassicality tests set bounds for all gaussian states and their mixtures , they can be further used as criteria to detect genuine quantum non - gaussianity , which is known to be a crucial resource for numerous quantum tasks . we have obtained the gaussian bounds for all tests using the @xmath0-parametrized distributions . remarkably , we have shown how the rectangle and the right - triangle tests can be generalized to the parallelogram and the triangle tests , respectively , which is essentially equivalent to a squeezing operation on a given state , thereby enhancing the successful detection of genuine non - gaussianity . importantly from a practical point of view , the optimized parallelogram and triangle tests do not require an actual realization of squeezing operation , as the choice of appropriate phase - space points serves the purpose . employing four and three phase - space point tests have their own advantages in our tests . for instance , we have found that three - point tests are more advantageous in detecting the nonclassicality of mixed gaussian states . on the other hand , four - point tests are more powerful in detecting genuine non - gaussianity of non - gaussian states under a loss channel even below @xmath321 . detecting nonclassicality and quantum non - gaussianity using positive valued points in phase space requires a certain constraint , e.g. , predetermined position ( origin ) with energy constraint @xcite and a designated shape ( tetragon or triangle ) in our tests @xcite . it may be possible to enhance our tests by exploiting more constraints such as the input - state energy , the length of vertices , the area of a polygon , etc . in this respect , we have addressed an example that the diagonal lengths in the parallelogram test can be used as valuable information to connect our single - mode test and the bw nonlocality test in phase space . as nonlocality tests have evolved from simple to more complex tests @xcite , we hope our work can further stimulate efforts for more elaborate tests to detect nonclassical states and find useful resources for genuine multi - mode nonlocality tests . we acknowledge support through nprp , grant no . 7 - 210 - 1 - 032 , from the qatar national research fund . 99 e. wigner , phys . rev . * 40 * , 749 ( 1932 ) . w. h. zurek , rev . mod . phys . * 75 * , 715 ( 2003 ) . s. m. barnett and p. m. radmore , _ methods in theoretical quantum optics _ , oxford university press ( 2003 ) . s. l. braunstein and p. van loock , rev . phys . * 77 * , 513 ( 2005 ) . c. weedbrook , s. pirandola , ral garca - patrn , n. j. cerf , t. c. ralph , j. h. shapiro , and s. lloyd , rev . * 84 * , 621 ( 2012 ) . richter and w. vogel , phys . lett . * 89 * , 283601 ( 2002 ) . a. mari , k. kieling , b. m. nielsen , e. s. polzik , and j. eisert , phys . lett . * 106 * , 010403 ( 2011 ) . k. banaszek and k. wdkiewicz , phys . lett . * 82 * , 2009 ( 1999 ) . j. s. bell , _ speakable and unspeakable in quantum mechanics _ ( cambridge university press ) ( 1988 ) . h. nha and h. j. carmichael , phys . * 93 * , 020401 ( 2004 ) ; r. garcia - patron , j. fiurek , n. j. cerf , j. wenger , r. tualle - brouri , and ph . grangier , _ ibid_. * 93 * , 130409 ( 2004 ) . lee , h. jeong , and d. jaksch , phys . a * 80 * , 022104 ( 2009 ) . j. li , t. fogarty , c. cormick , j. goold , t. busch , and m. paternostro , phys . a * 84 * , 022321 ( 2011 ) . lee , m. paternostro , j. lee , and h. jeong , phys . rev . a * 87 * , 022123 ( 2013 ) . kim , j. kim , and h. nha , phys . a * 88 * , 032109 ( 2013 ) . g. adesso and s. piano , phys . lett . * 112 * , 010401 ( 2014 ) . j. park , j. zhang , j. lee , s .- w . ji , m. um , d. lv , k. kim , and h. nha , phys . lett . * 114 * , 190402 ( 2015 ) . s. lloyd and s. l. braunstein , phys . lett . * 82 * , 1784 ( 1999 ) . j. eisert , s. scheel , and m. b. plenio , phys . rev . lett . * 89 * , 137903 ( 2002 ) ; j. fiurek , _ ibid_. * 89 * , 137904 ( 2002 ) ; g. giedke and j. i. cirac , phys a * 66 * , 032316 ( 2002 ) . j. niset , j. fiurek , and n. j. cerf , phys . lett . * 102 * , 120501 ( 2009 ) . j. park , s .- y . lee , h .- w . lee , and h. nha , j. opt . b * 29 * , 906 - 911 ( 2012 ) . r. filip and l. mita , jr . lett . * 106 * , 200401 ( 2011 ) ; m. jeek , i. straka , m. miuda , m. duek , j. fiurek , and r. filip , _ ibid_. * 107 * , 213602 ( 2011 ) ; i. straka _ _ , _ ibid_. * 113 * , 223603 ( 2014 ) . m. g. genoni , m. l. palma , t. tufarelli , s. olivares , m. s. kim , and m. g. a. paris , phys . a * 87 * , 062104 ( 2013 ) ; c. hughes , m. g. genoni , t. tufarelli , m. g. a. paris , and m. s. kim , _ ibid_. * 90 * , 013810 ( 2014 ) . s. wallentowitz and w. vogel , phys . a * 53 * , 4528 ( 1996 ) . j. f. clauser , m. a. horne , a. shimony , and r. a. holt , phys . * 23 * , 880 ( 1969 ) . h. jeong , w. son , m. s. kim , d. ahn , and c. brukner , phys . a * 67 * , 012106 ( 2003 ) . for its generation , see e.g. d. t. pegg , l. s. phillips , and s. m. barnett , , 1604 ( 1998 ) ; m. dakna , j. clausen , l. knoll , and d. g. welsch , , 1658 ( 1999 ) ; j. fiurek , r. garca - patron , and n. j. cerf , _ ibid . _ * 72 * , 033822 ( 2005 ) ; s .- y . lee and h. nha , _ ibid . _ * 82 * , 053812 ( 2010 ) ; e. bimbard , n. jain , a. macrae , and a. i. lvovsky , nat . photon . * 4 * , 243 ( 2010 ) ; s .- y . lee , s .- w . ji , h .- j . kim , and h. nha , * 84 * , 012302 ( 2011 ) ; j. lee and h. nha , _ ibid . _ * 87 * , 032307 ( 2013 ) . m. s. kim , w. son , v. buek , and p. l. knight , phys . a * 65 * , 032323 ( 2002 ) ; wang xiang - bin , _ ibid_. * 66 * , 024303 ( 2002 ) . m. m. wolf , j. eisert , and m. b. plenio , phys . lett . * 90 * , 047904 ( 2003 ) . j. k. asbth , j. calsamiglia , and h. ritsch , phys . . lett . * 94 * , 173602 ( 2005 ) . h. nha and m. s. zubairy , phys . * 101 * , 130402 ( 2008 ) . r. tahira , m. ikram , h. nha , and m. s. zubairy , phys . a * 79 * , 023816 ( 2009 ) . n. brunner , d. cavalcanti , s. pironio , v. scarani , and s. wehner , rev . mod . phys . * 86 * , 419 ( 2014 ) . applying a linear transformation in eq . , we first have @xmath323 changing the variable @xmath324 to @xmath325 and using the property @xmath326 , we obtain @xmath327 as the interaction between a quantum state @xmath2 and a reservoir @xmath78 is expressed by a beam - splitter with transmittance @xmath156 , we have @xmath328 } ( \alpha ; 0 ) & = \int d^{2 } \beta w_{\rho } ( \sqrt{\eta } \alpha - \sqrt{1-\eta } \beta ; 0 ) w_{\sigma } ( \sqrt{1-\eta } \alpha + \sqrt{\eta } \beta ; 0 ) \nonumber \\ & = \frac{1}{1-\eta } \int d^{2 } \beta^{\prime } w_{\rho } ( \beta^{\prime } ; 0 ) w_{\sigma } \bigg [ \sqrt{\frac{\eta}{1-\eta } } \bigg ( \frac{\alpha}{\sqrt{\eta } } - \beta^{\prime } \bigg ) ; 0 \bigg ] . \tag{a3 } \end{aligned}\ ] ] comparing eqs . and , we see that the reshaping ( squeezing ) of a profile in phase space is related to a beam - splitter interaction between a squeezed quantum state @xmath329 and a squeezed vacuum reservoir @xmath272 with transmittance @xmath271 , @xmath330 } \bigg ( \frac{\alpha}{\sqrt{1-s } } ; 0 \bigg).\ ] ]

stepwise refinement of algebraic specifications is a well known formal methodology for program development . however , traditional notions of refinement based on signature morphisms are often too rigid to capture a number of relevant transformations in the context of software design , reuse , and adaptation . this paper proposes a new approach to refinement in which signature morphisms are replaced by _ logical interpretations _ as a means to witness refinements . the approach is first presented in the context of equational logic , and later generalised to deductive systems of arbitrary dimension . this allows , for example , refining sentential into equational specifications and the latter into modal ones .

1311.7090

the industrial demand for high - assurance software opens a window of opportunity for mathematically based development methods , able to design complex systems at ever - increasing levels of reliability and security . this paper s contribution is placed at a specific corner of the broad landscape of formal methods for software development : that of _ algebraic specification _ @xcite , a family of methods which , having played a pioneering role , constitutes at present a large and mature body of knowledge and active research . such methods have a double origin . on the one hand they can be traced back to early work on data abstraction and modular decomposition of programs @xcite . on the other hand , to research on semantics of program specifications building on results from algebraic logic and model theory . especially relevant in this respect is the original work of the so - called adj group @xcite whose initial algebra semantics was the first , full formal approach to software development put forward . this double origin , temporally located around mid seventies , is not surprising : _ compositionality _ is both a basic requirement in program development and a major asset in algebraic semantics . the whole area flourished rapidly from the outset : not only different approaches to semantics ( final , observational , loose ) emerged , but also the initial tie to many - sorted equational logic was soon extended , first to conditional - equational logic , and later to order - sorted , partial and full first - order among other variants . the emergence of the first effective algebraic specification languages obj @xcite and clear @xcite overlaps another major development : the introduction of institutions by j. goguen and r. burstall @xcite . institution theory , which develops model theory independent of the underlying logical system , made possible to decouple specification methodology from the particularities of whatever semantics one may consider more suitable to a specific problem @xcite . moreover , although for a long time the impact of these methods in industry has been limited , a successful effort has been made in the last 15 years towards convergence on generic frameworks with suitable tool support . the compass and , later , the cofi initiative @xcite , which lead to the development of casl @xcite , are relevant milestones in this process . besides casl , cafeobj @xcite and maude @xcite are currently used in several industrial applications and tool development . actually , research in such methods , either at a foundational or methodological level , found applications in new , unsuspected areas for example , in documenting service interfaces @xcite , characterising contracts in contract - based programming @xcite or test generation for software composition @xcite . for the working software engineer , a software component is documented by an _ interface _ , which provides a language through which it interacts with its environment , and a _ specification _ of the intended meaning of the services provided . this specification is implemented by a concrete piece of software respecting the specified semantics . algebraic specification methods build on the observation that these somehow vague concepts from software engineering can be framed rigorously in terms of well - known mathematical notions . thus , an _ interface _ corresponds to a _ signature _ , i.e. , a set of names for the relevant types , called _ sorts _ , and a family of service or operation names , classified by their arity and input - output sorts . a signature generates a formal language , giving a rigorous meaning to what we have called before the component s interaction language . once fixed a signature , a _ specification _ describes a _ class of models _ for that signature , and an _ implementation _ identifies a specific model within such a class . if functions provide suitable abstractions of the services offered by a software component , this analogy can be made even more concrete by identifying _ interfaces _ with _ algebraic signatures _ , ( denotations of ) _ specifications _ with _ classes of algebras _ , and _ implementations _ with specific _ algebras_. the analogy extends to the entire software development process along which components are _ refined _ by incrementally adding detail and reducing under - specification . formally , this is a process of structural transformations witnessed by _ signature morphisms _ , which map functionally sorts and operations from a signature to another respecting the sort translation of functional types . in such a context this paper raises and discusses the following question : _ can more flexible notions of refinement emerge from replacing signature morphisms by some weaker notion of transformation ? _ the quest for _ weaker notions of transformation _ lead us to a different setting , that of algebraic logic @xcite . the key conceptual tool is that of a _ deductive system _ , i.e. , a formal language generated by a _ signature _ , and a _ consequence relation_. interrelating such systems , through maps connecting logical properties , has been studied from early in the last century . such maps were called _ translations _ and investigated as part of an ambitious programme addressing tools to handle the multiplicity of logics . as a result , several intuitive notions of translation are scattered in the literature . many logicians tailored the notion , for their own purposes , to relate specific logics and to obtain specific results . in general , however , a translation is regarded as a map between sets of formulas of different logics such that the image of a theorem is still a theorem . they were used originally to clarify the relationship between classical and constructive logics . our starting point is the observation that specifications describe classes of models and those can be naturally associated to deductive systems . then , translations that both reflect and preserve consequence relations seem interesting candidates to witness weaker forms of refinement . in this paper we will single out a specific sort of translations based on _ multifunctions _ , , functions mapping an element to a set of elements . such translations are called _ interpretations _ and constitute a central tool in the study of equivalent algebraic semantics ( see , , @xcite ) . a paradigmatic example is the interpretation of the _ classical propositional calculus _ into the _ equational theory of boolean algebras _ ( cf . * example 4.1.2 ) ) . this paper explores interpretations between the deductive systems corresponding to classes of models of specifications as possible witnesses of refinement steps . the notion seems able to capture a number of transformations which are difficult to deal with in classical terms . examples include data encapsulation and the decomposition of operations into atomic transactions . it also seems promising in the context of new , emerging computing paradigms which entail the need for more flexible approaches to what counts as a valid transformation along the development process ( see , for example , @xcite ) . in this context , the contribution of this paper , which combines and extends previous results by the authors reported in @xcite and @xcite , is twofold . on the one hand it puts forward a detailed characterisation of refinement witnessed by interpretations , referred in the sequel as _ refinement by interpretation _ , and exemplifies its potentialities in a number of small , yet illustrative examples . on the other hand , it renders the whole approach at a sufficiently abstract setting to be applicable over _ arbitrary _ , technically _ @xmath0-dimensional _ , deductive systems . the dimension fixes the kind of relationship between terms one is interested in . dimension 2 , for example , encompasses equations , regarded as instances of a binary predicate asserting , for example , term equality , bisimilarity , or observational equivalence . similarly , a unary predicate asserting the validity of a formula is enough to represent a proposition , leading to 1-dimensional deductive systems . refinement by interpretation in a general , @xmath0-dimensional setting provides a suitable context to deal simultaneously with deductive systems arising from classes of models presented in different logics , for example , as a set of equations , propositions or modal formulas . once stated the paper s contributions , it is important to clearly delimit its scope . first of all it should be stressed that the focus of this paper is not placed on _ specifications _ , understood as syntactic entities which describe in a structured , modular way classes of models , but rather on the _ classes of models _ themselves , which constitute their denotations . deductive systems , the basic tool in our approach , correspond to such classes . this means that the whole area of _ specification structuring _ @xcite is , for the moment , left out . our approach is not concerned with the fact that specifications describing the relevant classes of models are _ flat _ , i.e. given by a finite set of sentences , or _ structured _ , _ i.e. _ , built by systematic application of a number of operators , such as _ union _ , _ translation _ or _ hiding _ , all of them well characterised in the literature and implemented in a number of computer - supported modelling tools . this does not deny the fundamental importance of specification structuring . research on this topic started with the introduction of clear @xcite , by the end of the seventies , and its role can not be underestimated . actually , the recursive definition of structured specifications provides basic modular procedures for software composition and architecture . moreover structuring operations allows one to go beyond the specification power of simple , unstructured specifications @xcite . clearly , the approach proposed in this paper can be tuned to specification refinement in a strict sense . in a recent publication @xcite we showed how _ refinement by interpretation _ can be lifted to the level of structured specifications with the usual operators mentioned above . we believe , however , that by focussing on _ classes of models _ this piece of research acquires a broader scope of application and is worth on its own . in particular , it pays off when dealing with requirements that can not be properly formalised in a specification ( for example , the property that a controller has a finite number of states ) . note this does not entail any loss of expressivity : for each specification , one may recursively compute its denotation ( a signature and a class of models ) and work directly with them . on the other hand , the discussion on which operators should be considered in a specification calculus is still active . for example , very recently , reference @xcite introduced two new operators for specification composition in order to deal with non - protecting importation modes . this further justifies the relevance of a semantic approach as proposed here . another concept to make precise is _ refinement_. the word is taken here in the broad sense of a transformation mapping an abstract to a more concrete class of models . as such classes are represented by deductive systems , a refinement will map a deductive system into another , while preserving the consequence relation . this is in line with the usual meaning of refinement : all requirements stated at the original level are still valid after refinement . moreover , it will be shown in the paper that _ refinement by interpretation _ between classes of models boils down to the standard notion of refinement as inclusion of classes of models whenever the witnessing interpretation is simply an identity . finally , a note on the expressivity of deductive systems . actually , deductive systems can play a double role representing both logics , on top of which all the specification machinery can be developed , and classes of models as discussed above . clearly , any institution induces , for each signature , a deductive system through its satisfaction relation , and conversely , deductive systems may be viewed as special cases of institutions as discussed in @xcite . for example , a deductive system may represent the class of boolean algebras ; but the latter can also be specified as a theory in a suitable institution . another example is provided by modal logics which can be regarded as both a deductive system or a theory in the first - order ( fol ) institution through the standard translation . this provides a uniform view of seemingly different settings , enabling us to discuss how essentially the same conceptual tool , that of _ logical interpretation _ , can be used to interrelate logics and refine classes of models both regarded as deductive systems . reciprocally , deductive system can be endowed with an algebraic semantics , as discussed in section [ sc : algsem ] . note that the quest for such a uniform representation of logics and logical theories pops up in other contexts , namely on the design of logical frameworks . a prime example is provided by the logics - as - theories approach proposed by f. rabe in @xcite , resorting to a type theoretical framework , and further developed in the context of the latin project @xcite . section [ sc : pre ] introduces _ @xmath0-dimensional deductive systems _ and their semantics following @xcite . this paves the way to the formulation of refinement by interpretation in a general setting in sections [ sc : gen ] and [ sc : rbi2 ] . before that , however , in section [ sc : rbi1 ] , the approach is instantiated for the case of algebraic specifications over the institution of horn clause logic . this is a popular framework for algebraic specifications which not only deserves attention on its own , but also provides a simpler setting to build up intuitions . finally , section [ sc : conc ] concludes and suggests some problems deserving further attention . _ specifications of complex systems resort to different logics , and even to their combination . consequently a characterisation of _ refinement by interpretation _ needs to be orthogonal to whatever logic is used in specifications . this is achieved through the notion of @xmath0-_dimensional deductive systems _ , of which the equational case is just an instance for @xmath1 . this section reviews such systems and their semantics , following @xcite , to provide the background for the sections to follow . _ roughly speaking , a deductive system is a general mathematical tool to reason about formulas in a language generated by a signature . formally , it is defined as a pair @xmath2 , where @xmath3 is a signature and @xmath4 is a substitution - invariant consequence relation between sets of formulas and individual formulas . clearly , any standard sentential logical system , defined in the usual way by a set of axioms and a set of inference rules ( for instance , classical and intuitionist propositional calculus , referred to in the sequel as cpc and ipc , respectively ) , is a deductive system . first order logic can also be formulated as a deductive system @xcite , which shows how broad the concept is . the formal notion of a deductive system , in this abstract perspective , was originally considered by ukasiewicz and tarski @xcite and intensely studied , from an algebraic point of view , by many logicians . this gave rise to a new , extremely relevant area of mathematics , that of _ abstract algebraic logic _ @xcite . although in some literature on algebraic logic this substitution - invariant consequence relation has been called a _ logic _ ( cf . @xcite ) , we adopt along the paper the designation of _ deductive system _ used by blok and pigozzi @xcite . this terminology allows us to distinguish this concept from the habitual meaning logic has in computer science , typically understood as an abstract framework to express specifications and often abstracted as an _ institution _ @xcite or a @xmath5-institution @xcite . as mentioned above , translation maps were introduced in the early 20th century as a means to interrelate deductive systems . they were first used to understand the relationship between classical and constructive logics . the well - known gdel translation of classical logic into intuitionistic logic has inspired disperse works on comparing different logics by means of translations . illustrative examples include the works of kolmogorov @xcite , glivenko @xcite , and gdel @xcite involving classical , intuitionist , and modal logics . to the best of our knowledge the first general definition of translation between deductive systems is due to prawitz and malmns @xcite . more recently , wjcicki @xcite presented a systematic study of translations between logics , focussing on inter - relations between sentential logics . and the quest goes on ( cf . at the turn of the century , silva , dottaviano and sette @xcite proposed a general definition of translation between logics as maps preserving consequence relations . then , feitosa and dottaviano studied intensively the subclass of translations that preserve and reflect consequence relations and coined the name _ conservative translation _ @xcite . conservative translations which are able to relate a formula to a set of formulas , and are therefore defined as _ multifunctions _ , are called _ interpretations_. those which commute with substitutions were originally used in abstract algebraic logic to define a very important class of deductive systems referred to as _ algebraisable _ in particular , they abstract the strict relationship between classic propositional logic and the class of boolean algebras . a deductive system is said to be _ algebraisable _ whenever there exists a class @xmath6 of algebras such that the consequence relation induced by @xmath6 is equivalent to the consequence in the deductive system . such an equivalence was originally defined by means of two mutually inverse interpretations . since then , this link between logic and universal algebra has been successfully explored . in particular , for an algebraisable @xmath7 , properties of @xmath7 can be related to algebraic properties of its equivalent algebraic semantics . this kind of results , of which many examples exist , are often called _ bridge theorems_. in order to broaden the spectrum of application of deductive systems , blok and pigozzi et al . @xcite introduced consequence relations over @xmath0-tuples of formulas , for @xmath0 a non - zero natural number . _ @xmath0-deductive systems _ , the result of this generalisation , are the higher dimensional version of the well known sentential logics . their theory provides a unified treatment for several such as the ones corresponding to assertional , equational , and inequational logics . this generalisation also allows to regard interpretations witnessing algebraisability as a special kind of translations between @xmath0-deductive systems . an _ equation _ , represented in this paper by a formal expression @xmath8 , can be regarded as a pair of terms ( or formulas ) @xmath9 . this , in turn , is an instance of a binary predicate standing for the equality of two terms . similarly , a _ unary predicate _ asserting the validity of a formula is enough to represent a proposition . the first observation leads to what will be characterised in the sequel as a @xmath10-dimensional deductive system , of which the equational case is a particular instance . the second corresponds , roughly speaking , to sentential logics in a quite broad sense ( to include , for example , first - order predicate logic when suitably formalised ) . in general , adding a @xmath0-ary predicate to a strict universal horn theory without equality , gives rise to a representation of a @xmath0-dimensional , thus providing a suitable context to deal simultaneously with different specification logics . we go even a step further considering @xmath0-deductive systems over _ many sorted _ languages , because , in general , software systems manipulate several sorts of data . almost all notions can be formulated in this broader setting as discussed later . note there are other generalisations that allow the reuse of arguments and tools from abstract algebraic logic in computer science contexts . hidden logics , introduced by pigozzi and martins in @xcite ( see also @xcite and @xcite ) are a prime example . they have been efficiently used to develop specification and verification methodologies for object oriented software systems . examples include the boolean logics , , 1-dimensional multi - sorted logics with bool as the only visible sort , and equality - test operations for some of the hidden sorts in place of equality predicates . the syntactic support for @xmath0-dimensional deductive systems is that of a @xmath0-term . let @xmath11 be a signature and @xmath12 a @xmath13-sorted set of variables . @xmath0-term _ of sort @xmath14 over signature @xmath3 is a sequence of @xmath0 @xmath3-terms , all of the same sort @xmath14 , @xmath15 , abbreviated to @xmath16 whenever references to sorts can be omitted . @xmath0-variable of sort @xmath14 _ is a sequence of @xmath0 variables all of the same sort @xmath14 . @xmath17 is the sorted set of all @xmath0-terms over @xmath3 with variables in @xmath18 , i.e. , @xmath19 where @xmath20 is the set of all terms over @xmath3 , of sort @xmath14 , with variables in @xmath18 . whenever each @xmath21 , for each sort @xmath14 in @xmath3 , is non empty , their union acts as the carrier of the @xmath3-_term algebra _ freely generated from @xmath18 , which we denote by @xmath22 . a _ substitution on _ @xmath22 is just an endomorphism over @xmath22 . let us fix some notation and terminology : if @xmath23 is a @xmath0-term over @xmath3 , @xmath24 is a @xmath3-algebra , and @xmath25 , we denote by @xmath26 the value @xmath16 takes in @xmath24 when variables @xmath27 are instantiated respectively by @xmath28 . more precisely , if @xmath29 then @xmath30 , where @xmath31 is any homomorphism from @xmath22 to @xmath24 such that @xmath32 for all @xmath33 . let @xmath34 be an arbitrary but fixed family of countably infinite disjoint sets @xmath35 of variables of sort @xmath36 . following a typical procedure in similar contexts ( , @xcite ) , we assume in the sequel @xmath37 fixed for each set of sorts @xmath13 and large enough to contain all variables needed . symbols of variables are obviously disjoint of any other symbol in the signature . as usual in sentential logic frameworks , we will refer to formulas ( @xmath0-formulas ) as synonymous to terms ( @xmath0-terms respectively ) . accordingly , we will denote @xmath38 by @xmath39 . moreover , for each nonzero natural number @xmath0 , given a sorted signature @xmath3 , a _ @xmath0-formula _ of sort @xmath14 over @xmath3 is any element of @xmath40 . the set of all @xmath0-formulas will be denoted by @xmath41 . also note that an @xmath13-sorted subset @xmath42 of @xmath0-formulas is identified with the unsorted set @xmath43 , which allows writing @xmath44 to mean that @xmath45 , for some sort @xmath14 . a set @xmath46 is said to be _ globally finite _ when @xmath47 is a finite set for each sort @xmath14 of @xmath3 , equal to @xmath48 except for a finite number of them , , @xmath43 is finite . in this setting , a @xmath0-dimensional deductive system is defined as a substitution - invariant consequence relation on the set of @xmath0-formulas . the following definition generalises the one due to w. blok and d. pigozzi @xcite for the one - sorted case . [ df : klogic ] a _ @xmath0-dimensional deductive system _ is a pair @xmath49 , where @xmath3 is a sorted signature and @xmath50 is a relation such that , for all @xmath51 , the following conditions hold : 1 . @xmath52 for each @xmath53 ; 2 . if @xmath54 , and @xmath55 for each @xmath53 , then @xmath56 ; 3 . if @xmath54 , then @xmath57 for every substitution @xmath58 . a @xmath0- is _ specifiable _ if @xmath59 is _ compact _ ( or _ finitary _ in the terminology of abstract algebraic logic ) , , if , whenever @xmath54 , there exists a globally finite subset @xmath60 of @xmath42 such that @xmath56 . the relation @xmath59 , abbreviated to @xmath4 whenever @xmath61 is clear from the context , is called _ the consequence relation of _ @xmath61 . it is easy to see that , for any @xmath51 , @xmath62 and @xmath63 imply @xmath64 . every consequence relation @xmath4 has a natural extension to a relation between sets of @xmath0-formulas , also denoted by @xmath4 , defined by @xmath65 if @xmath66 for each @xmath67 . finally , the relation of _ interderivability _ between sorted sets is defined by @xmath68 if @xmath69 and @xmath70 . we abbreviate @xmath71 and @xmath72 by @xmath73 and @xmath74 , respectively . let @xmath61 be a ( not necessarily specifiable ) @xmath0- . thm _ of @xmath61 is a @xmath0-formula @xmath16 such that @xmath75 , , @xmath76 . the set of all theorems is denoted by @xmath77 . an _ inference rule _ is a pair @xmath78 where @xmath79 a globally finite set of @xmath0-formulas and @xmath16 a @xmath0-formula , usually represented as @xmath80 a rule such as ( [ sequent ] ) is said to be a _ derivable rule _ of @xmath61 if @xmath81 . a set of @xmath0-formulas @xmath82 closed under the consequence relation , , such that @xmath83 implies @xmath84 , is called a _ theory _ of @xmath61 . the set of all theories is denoted by @xmath85 ; it forms a complete lattice under set - theoretic inclusion , which is algebraic if @xmath61 is specifiable . given any set of @xmath0-formulas @xmath42 , the set of all consequences of @xmath42 , in symbols @xmath86 , is the smallest theory that contains @xmath42 . it is easy to see that @xmath87 . often , a specifiable @xmath0- is presented in the so - called hilbert style , , by a set of axioms ( @xmath0-formulas ) and inference rules . we say that a @xmath0-formula @xmath88 is _ directly derivable _ from a set @xmath42 of @xmath0-formulas by a rule such as ( [ sequent ] ) if there is a substitution @xmath89 such that @xmath90 and @xmath91 . given a set @xmath92 of @xmath0-formulas and a set @xmath93 of inference rules , we say that @xmath88 is _ derivable _ from @xmath42 by @xmath92 and @xmath93 , in symbols @xmath94 , if there is a _ proof _ , , a finite sequence of @xmath0-formulas , @xmath95 such that @xmath96 , and for each @xmath33 , one of the following conditions holds : 1 . @xmath97 , 2 . @xmath98 is a substitution instance of a @xmath0-formula in @xmath92 , 3 . @xmath98 is directly derivable from @xmath99 by one of the inference rules in @xmath93 . it is clear that @xmath100 is a specifiable @xmath0- . moreover , a @xmath0-deductive system @xmath61 is specifiable iff there exist possibly infinite sets @xmath92 and @xmath93 , of axioms and inference rules respectively , such that , for any @xmath0-formulas @xmath88 and any set @xmath42 of @xmath0-formulas , @xmath101 iff @xmath102 ( see @xcite for the one sorted case ) . this justifies that all the examples of specifiable deductive systems introduced in this paper are presented by a set of axioms and inference rules . if a deductive system @xmath61 is equal to @xmath100 , for some sets @xmath92 and @xmath93 with @xmath103 , @xmath61 is said to be _ finitely axiomatisable_. a @xmath0- @xmath104 is an _ extension _ of the @xmath0- @xmath105 if @xmath106 whenever @xmath107 for all @xmath108 ( , @xmath109 ) . a @xmath0- @xmath110 is _ an extension by axioms and rules of _ a specifiable @xmath0- @xmath61 if it can be axiomatised by adding axioms and inference rules to the axioms and rules of some axiomatisation of @xmath61 . typical examples of @xmath0-deductive systems are the ones induced by algebraic specifications ( see section [ sc : rbi1 ] ) . given a signature @xmath3 , they are defined over pairs of @xmath3-terms @xmath9 , representing equations @xmath111 , and have therefore dimension @xmath1 . similarly , @xmath3-conditional equations can be taken as pairs @xmath112 where @xmath42 is a globally finite subset of @xmath113 and @xmath114 . as a particular case , an equation @xmath8 is a conditional equation without premisses , @xmath115 . in general , a conditional equation @xmath116 will be written as @xmath117 . in the sequel we will often use @xmath118 , instead of @xmath113 , for the set of all equations over @xmath37 , and , similarly , @xmath119 , for the set of all @xmath3-conditional equations over @xmath37 . let @xmath120 and @xmath24 be a @xmath3-algebra . we write @xmath121 if , for every homomorphism @xmath122 , @xmath123 for @xmath124 , @xmath125 is abbreviated to @xmath126 . an equation @xmath127 is an _ identity _ of @xmath24 if @xmath128 . similarly , a conditional equation @xmath129 is a _ quasi_-_identity _ of @xmath24 if @xmath130 , which is simply written as @xmath131 when clear from the context . these definitions extend , as expected , to classes of algebras . given a class of @xmath3-algebras @xmath6 , the ( _ semantic _ ) _ equational _ _ consequence _ _ relation _ @xmath132 determined by @xmath6 is defined by @xmath133 in this case @xmath127 is said to be a @xmath6-_consequence _ of @xmath42 . when clear from the context , we simply write @xmath134 , where @xmath129 , for @xmath135 . both @xmath136 and @xmath132 extend to sets of equations @xmath137 : @xmath138 iff @xmath139 for each @xmath140 , and respectively for @xmath132 . for a set @xmath141 of quasi - equations , adopting the notational convention explained above and rather standard in universal algebra , @xmath142 stands for @xmath131 for each @xmath143 ( analogously for a class @xmath6 of @xmath3-algebras ) . the equational consequence relation @xmath132 satisfies the conditions of definition [ df : klogic ] . hence it constitutes an example of a 2- ( perhaps the most important one ! ) often simply denoted by @xmath6 . a class @xmath6 of @xmath3-algebras is axiomatised by a set @xmath141 of conditional equations if @xmath144 . it can be proved that , if @xmath6 is a class of @xmath3-algebras axiomatised by a set @xmath141 of conditional equations , then the relation @xmath145 is specifiable ( see @xcite for the one - sorted case ) . in this case it can be defined in the hilbert style , taking the set of @xmath3-equations in @xmath141 , together with reflexivity , as the set of axioms , and the set of @xmath3-conditional equations in @xmath141 , along with symmetry , transitivity and congruence rules , as inference rules . any specifiable equational deductive system over @xmath3 is the natural extension ( by axioms and rules ) of the ( 2-dimensional ) free deductive system over @xmath3 denoted by @xmath146 and defined in figure [ fig:3 ] . note that the consequence relation associated to @xmath147 coincides with @xmath148 , where @xmath149 is the class of all @xmath3-algebras . recall the discussion of the double role played by deductive systems in subsection [ sc : scope ] . actually , it is well known that any deductive system can be seen as a @xmath5-institution @xcite , and taking care of some extra technicalities , as an institution . however some special finitary ones are , in a more natural way , viewed as theories of institutions , _ i.e. _ , as sets of sentences equipped with extra features the deductive apparatus they induce . for instance , given a quasivariety of algebras , we can define a ( 2-dimensional ) deductive system over the set of equations by using the identities and quasidentities that define the quasivariety , together with the axioms of reflexivity and the inference rules of symmetry , transitivity and congruence rules . this deductive system can naturally be seen as a theory of the institution of horn clause logic . as discussed in @xcite , a semantics for arbitrary @xmath0- needs to go beyond the usual algebraic structures , resorting to algebras endowed with a set of @xmath0-tuples . formally , [ df : kstruct ] a _ @xmath0-structure _ over a signature @xmath150 is a pair @xmath151 where @xmath24 is a @xmath3-algebra and @xmath152 is a sorted set @xmath153 such that @xmath154 in this definition , the sorted set @xmath152 of designated elements of @xmath24 , can be regarded as a set of truth values on @xmath24 : a formula holds if its interpretation is one of these elements . this is why @xmath152 is called a _ filter _ : for a deductive system representing the constructive propositional calculus on a boolean algebra the notion boils down to the familiar , boolean filter . let @xmath155 be a @xmath0-structure . a @xmath0-formula @xmath156 is said to be a _ semantic consequence _ in @xmath157 of a set of @xmath0-formulas @xmath42 , in symbols @xmath158 , if , for every assignment @xmath159 , @xmath160 whenever @xmath161 for every @xmath162 , where @xmath163 is the component @xmath14 of the sorted set @xmath152 . notice that the same notation is used for the assignment and its natural extension to formulas . a @xmath0-formula @xmath164 is _ valid _ in @xmath157 , and conversely @xmath157 is a _ model _ of @xmath16 , if @xmath165 . a rule such as ( [ sequent ] ) is a _ validity _ , or a _ valid rule _ , of @xmath157 , and conversely @xmath157 is a _ model _ of the rule , if @xmath166 . a formula @xmath16 is a _ semantic consequence _ of a set of @xmath0-formulas @xmath42 for an arbitrary class @xmath167 of @xmath0-structures over @xmath3 , in symbols @xmath168 , if @xmath169 for each @xmath170 . it can be proved that @xmath171 is always a @xmath0- , even if not always specifiable . similarly , a @xmath0-formula or a rule is a _ validity of _ @xmath167 if it is a validity of each member of @xmath167 . a @xmath0-structure @xmath157 is a _ model _ of a @xmath0- @xmath61 if @xmath54 always implies @xmath169 , if every consequence of @xmath61 is a semantic consequence of @xmath157 . the special models whose underlying algebra is the formula algebra , _ i.e. _ , models of the form @xmath172 , for @xmath82 a theory , are called _ lindenbaum - tarski models_. the class of all models of @xmath61 is denoted by @xmath173 . if @xmath61 is a specifiable @xmath0- , then @xmath157 is a model of @xmath61 iff every axiom and rule of inference is a validity of @xmath157 . in the equational case the semantics based on @xmath10-structures boils down to the traditional algebraic semantics . more precisely , given a quasi - equational class @xmath6 of @xmath3-algebras ( _ i.e. _ axiomatised by a set of quasi - equations over @xmath3 ) , the algebra of any model of the 2-deductive system induced by @xmath6 with the identity as its filter belongs to @xmath6 . a class of @xmath0-structures @xmath167 is a _ @xmath0-structure semantics _ of @xmath61 if @xmath174 . the class of all models of @xmath61 forms a @xmath0-structure semantics of @xmath61 . this fact is expressed in a specific completeness theorem @xcite , stating that , for any @xmath0- @xmath61 , @xmath54 iff @xmath175 , for every @xmath176 . _ this section introduces , exemplifies and discusses the concept of _ refinement by interpretation _ the core contribution of the paper framed in the specific , but popular , setting of algebraic specification over the institution of horn clause logic . _ this section introduces refinement by logical interpretation for the specific case of algebraic specifications . as usual ( see , @xcite ) , an _ algebraic specification _ @xmath177 is considered here as a structured specification over the institution of horn clause logic ( * hcl * ) @xcite . each @xmath177 denotes a pair @xmath178 where @xmath3 is a signature and @xmath179 is a class of @xmath3-algebras . the class @xmath179 of @xmath3-algebras is called the _ model class of _ @xmath177 , and each @xmath3-algebra in @xmath179 a _ model of _ @xmath177 . if @xmath180 is a conditional equation @xmath181 ( respectively , an equation @xmath182 ) , we write @xmath183 for @xmath184 ( respectively , @xmath185 ) . both cases extend , as expected , to sets of conditional equations ( respectively , equations ) . when an algebraic specification @xmath177 is _ flat _ , or _ basic _ , its model class @xmath179 of algebras is axiomatised by a set @xmath141 of conditional equations . in this case @xmath177 is the pair @xmath186 . if the definition is restricted to formulas over a specific set of @xmath3-variables @xmath187 , @xmath177 is said @xmath18-_flat_. when @xmath141 is a set of equations the flat specification @xmath188 is called a ( flat ) _ equational specification_. recall that a class @xmath6 of @xmath3-algebras axiomatisable by a set of equations is called a _ variety_. a variety can also be characterised as a nonempty class @xmath6 of @xmath3-algebras closed under homomorphic images , subalgebras and direct products ( cf . @xcite , @xcite ) . this famous result , due to birkhoff , turns out to be very useful to show that a given algebraic specification is not an equational specification . in the sequel , for simplicity , when clear from the context , we refer to algebraic specifications simply as specifications . in this context _ stepwise refinement _ @xcite of specifications refers to the process through which a complex design is produced by incrementally adding details and reducing under - specification . this proceeds step - by - step until the class of models becomes restricted to such an extent that a program can be easily manufactured . formally , given a specification @xmath177 , the implementation process builds a correct realisation from a concrete enough class of @xmath3-models @xmath6 such that @xmath6 is a subset of the class of denotations of @xmath177 . during this process , the specification is enriched according to specific design decisions , iteratively approaching the intended meaning for the final program . starting from an initial abstract specification @xmath189 , refinement builds a chain of specifications @xmath190 where , for @xmath191 , @xmath192 represents reverse inclusion of the respective classes of models . transitivity of inclusion assures that @xmath193 . from @xmath194 onwards each element in this chain is the result of a _ refinement step_. in order to deal with more complex requirements along the implementation process , for example to enable the possibility of renaming , adding or grouping together different signature components , refinement steps @xmath195 are traditionally taken up to _ signature morphisms_. recall that a _ signature morphism _ from @xmath196 to @xmath197 is a pair @xmath198 , where @xmath199 and @xmath200 is a @xmath201-family of functions respecting the sorts of operation names in @xmath202 , _ i.e. _ , @xmath203 ( where for @xmath204 given a signature morphism @xmath205 and a @xmath206-algebra @xmath24 , let @xmath207 denote the _ reduct _ of @xmath24 along @xmath58 , , for any @xmath208 , @xmath209 , and for all @xmath210 , @xmath211 . the notation @xmath212 and @xmath213 refer , respectively , to the carrier of sort @xmath14 and the interpretation of operation symbol @xmath214 in the algebra @xmath24 . in this context we say that @xmath215 is a refinement of @xmath177 witnessed by @xmath58 , or simply a @xmath58-refinement , when @xmath216 , where @xmath217 . having fixed the notation and the basic concepts we may now jump to the kernel of this section : representing classes of specification models as 2-deductive systems and introduce interpretations as possible witnesses to the refinement steps . as clarified in the introduction ( subsection [ sc : scope ] ) , our approach is based on the representation of specifications model classes as deductive systems . our focus is essentially semantic , _ i.e. _ , on what specifications denote , and , therefore , the whole theory of refinement by interpretation discussed here is independent from any specification structuring mechanism . actually , any specification @xmath177 denotes a class of algebras its _ model class _ , @xmath218 . this in turn induces a 2- according to the procedure sketched in subsection [ sc : eqqq ] . its consequence relation is @xmath219 . the possibility of this consequence relation being non finitary , for example if arising from non flat specifications , is also covered in this approach . as we will see , it turns out that @xmath0-deductive systems are an efficient universal tool to develop , in this way , a theory about all classes of models over a fixed signature . a most important particular case is that of _ flat _ specifications : each of them can be identified with the deductive system it induces . this observation will be assumed throughout the text and in the examples . moreover , using interpretations between the induced deductive systems , one may go from one model class to another . in a sense to become clearer below , we say that a specification @xmath215 _ refines _ another specification @xmath177 _ by interpretation _ @xmath220 , if @xmath220 is an interpretation of @xmath177 such that @xmath221 for any formula @xmath180 . before jumping to the technical definition , consider the following example which , although elementary , may help to build up some intuitions for this notion of refinement . [ firstex ] the introduction of a two - element boolean algebra together with equality tests for each sort allows the software engineer to formulate arbitrary universal first - order axioms as equations in the boolean sort . the importance of this construction comes from the fact that , although many of the most natural specification conditions take the form of universal first - order sentences , only equational and conditional - equational axioms are guaranteed to possess an initial model . this motivation , as well as equality - test algebras in general , are extensively discussed in @xcite . consider , thus , two flat specifications @xmath222 and @xmath223 . the former has a signature @xmath3 which declares a sort @xmath14 and a function @xmath224 . @xmath222 has an empty set of axioms ; thus , the corresponding class of models consists of all algebras over its signature . on the other hand , specification @xmath223 is depicted in figure [ fig : unica ] . it is not difficult to see , by induction on the structure of proofs , that the translation [ cols="^,^,^,^ " , ] we may now incorporate in the approach proposed in this paper the important tool given in lemma [ th : nsc ] : let @xmath225 be a self - translation between two @xmath226-deductive systems @xmath61 and @xmath110 over the signature @xmath3 . then , if @xmath110 has the conjunction property , @xmath220 is an interpretation iff its associated function is a conservative translation . first we prove that @xmath227 : since for any @xmath228 there is a @xmath229 such that @xmath230 , we have , by the conjunction property of @xmath110 , that @xmath231 and , by @xmath232 of definition [ df : klogic ] , that @xmath233 . analogously , since for each @xmath229 there is a @xmath234 such @xmath235 , we have , by @xmath232 of definition [ df : klogic ] , that @xmath236 . hence , for any interpretation @xmath220 and for all @xmath237 , @xmath238 , which implies that @xmath239 is a conservative translation . in a similar way , if @xmath239 is a conservative translation , @xmath220 is an interpretation . the connection to conservative translations turns out to be very useful in practice . the following theorem , which builds on results in @xcite , provides a sufficient condition for a translation to be an interpretation . [ th : nsc ] let @xmath220 be a @xmath240-translation from @xmath3 to @xmath206 , @xmath61 a @xmath0- over @xmath3 and @xmath110 a @xmath241- over @xmath206 . suppose that @xmath220 is functional and injective . if @xmath242 , for every set of formulas @xmath42 , then @xmath220 interprets @xmath61 in @xmath110 . from the inclusion @xmath243 we have that @xmath244 . suppose now that @xmath245 . hence there is a @xmath246 such that @xmath247 . since @xmath220 is injective @xmath248 , and so , @xmath249 , _ i.e. _ , @xmath54 . the approach to refinement proposed in this paper , in particular when specialised to 2-dimension deductive systems , should also be related to the extensive work of maibaum , sadler and veloso in the 70 s and the 80 s , as documented , for example , in @xcite . the authors resort to interpretations between theories and conservative extensions to define a syntactic notion of refinement according to which a specification @xmath215 refines a specification @xmath177 if there is an interpretation of @xmath215 into a conservative extension of @xmath177 . it is shown that these refinements can be vertically composed , therefore entailing stepwise development . this notion is , however , somehow restrictive since it requires all maps to be conservative , whereas in program development it is usually enough to guarantee that requirements are preserved by the underlying translation . moreover , in their approach , the interpretation edge of a refinement diagram needs to satisfy extra properties . as related work one should also mention @xcite where interpretations between theories are studied in the abstract framework of @xmath5-institutions the first reference is a generalisation of the work of maibaum and his collaborators , whereas the second one generalises the way algebraic semantics on sentential logics is dealt with in abstract algebraic logic to the abstract setting of @xmath5-institutions . similar developments could arise by considering institutions and their ( co-)morphisms @xcite . the work of meseguer @xcite on _ general logics _ , in which a theory of interpretations between logical systems is developed , should also be mentioned . our own approach to refinement by interpretation can be placed between these general works and the original contribution of maibaum . actually , on the one hand , we deal with general @xmath0-deductive systems therefore subsuming all frameworks above which are based on equational or first order logic ( , on specific instances of @xmath0-deductive systems ) . on the other hand , however , our results are formulated in terms of a concrete and intuitive notion of a deductive system ; their scaling to an abstract , institutional level is still to be done . the paper introduced a new notion of refinement and started the development of a corresponding theory of _ refinement by interpretation_. the results obtained and their applications seem promising , in the sense that a number of useful transformations of ( classes of models of ) specifications are captured as refinement steps . in order to clarify the scope of our results we should point out that the development in section 3 can be straightforwardly generalised as to apply to any horn fragment of a structural logic @xmath250 ( _ i.e. _ , a logic whose axioms have the form @xmath251 , where @xmath252 is a subset of the atomic formulas of the logic , with @xmath253 possibly empty ) . all one has to do is to represent such a fragment by the natural equivalent deductive system taking the atomic formulas of @xmath250 as its set of formulas and a presentation given by the axiomatisation of @xmath250 . the generalisation made along sections [ sc : gen ] and [ sc : rbi2 ] turns it relevant to the specification meta - level , , whenever an implementation step requires a change in the underlying logic . this often arises in formal software development with the need for accommodating new requirements ( as in example [ ex : ordbams ] ) or when a particular theorem prover , embodying a specific logic , is to be used for design validation ( as in example [ ex : tool ] ) . our most recent work @xcite is another generalisation effort aiming at reframing this notion of refinement in a categorical setting based on a characterisation of abstract logics as coalgebras for the closure system contravariant functor @xcite upon the category @xmath254 of sets and functions . to conclude we would like to remark again the semantic perspective from which this work was developed , as extensively discussed in the introduction this entails the need for further research on how refinement by interpretation , which is entirely based on properties of arbitrary deductive systems , can be smoothly combined with concrete specification structuring operations . preliminary work on this topic is reported in @xcite in which the emphasis is shifted to specifications . current work in this direction includes the development of a refinement calculus of structured specifications over a @xmath5-institution . as a general remark we would like to stress again that the approach developed in this paper can be applied to any notion of algebraic specification based on any fixed set of specification structuring combinators , further justifying the relevance of the semantic perspective adopted in this paper . a practical limitation of this approach of reducing the specifications to their class of models is that often such classes have infinitely many models . however in some cases subclasses possessing the same theory can be considered instead . for instance , if the class is a finitely generated quasivariety @xmath6 @xcite , and consequently the associated 2-deductive system is finitary with a presentation given by the axiomatisation of the quasivariety , we can replace @xmath6 by the set of its generators which induces the same deductive system . an example is the class of boolean algebras which are generated ( actually , as a variety ) by the two - element boolean algebra . from an application point of view , this semantic approach seems to have its own potentialities which we would like to recall . actually , in a number of cases it is relevant , and even mandatory , to start the implementation procedure from a set of models that does not come from a structured specification . this can be the case when reusing designs ( a recurrent strategy in engineering ) or even to express meta - requirements that can not be easily accommodated within the classic refinement procedure . focussing on _ classes of models _ , on the other hand , makes possible to deal with requirements that can not be properly formalised in a specification . note this does not entail any loss of expressivity . the approach proposed in this paper can be tuned to specification refinement in a strict sense : for each specification , one may recursively compute its denotation ( a signature and a class of models ) and work directly with it . in general , we believe that this approach has a real application potential , namely to deal with specifications spanning through different specification logics . particularly deserving to be considered , but still requiring further investigation , are observational logic @xcite , hidden logic @xcite and behavioural logic @xcite . in all of these cases the satisfaction of requirements is discussed up to some particular satisfaction relation and their verification is checked with respect to relations obtained by replacing strict equality by its underlying notion of satisfaction . in this context , a semantics based on @xmath0-structures paves the way to the unification of all of these approaches . actually , in all of them , models consist of algebras whose @xmath0-structures are of the form @xmath255 , where @xmath256 captures the particular satisfaction relation in each formalism . in particular , the strict models of a ( classical ) algebraic specification @xmath177 consist of algebras @xmath24 whose @xmath0-data structure @xmath257 is a model of @xmath258 . naturally , most of the models of software specifications are not admissible choices as implementations . therefore , the choice of adequate filters along the implementation process becomes a crucial , although not trivial task . this should be driven by the system nature ( for example , adopting observational equality to deal with objects with encapsulated data ) . a similar concern is , moreover , shared by other general approaches to formal development , as , for example , @xcite in the context of behavioral logic . a lot of other questions remain to be answered . one such topic , as mentioned above , concerns _ horizontal composition _ of refinements by interpretation ; _ vertical _ composition raising no special problems as shown in theorems [ th : ver1 ] and [ th : ver2 ] . to illustrate the kind of results we are investigating suppose , for example , that @xmath220 interprets @xmath177 in @xmath215 . the challenge would be to prove that @xmath220 also interprets an enrichment of @xmath177 by axioms in an appropriate sub - specification of @xmath215 . a closely related issue is the extension of this approach to the level of ( structured ) specifications . we believe that this can be captured in a somehow standard way , which will be most relevant in studying the interplay between horizontal ( _ i.e. _ , architectural ) and vertical ( _ i.e. _ , implementation driven ) levels of specification composition . for example , the _ union _ of two specifications will correspond to the union of the corresponding consequence relations . actually , a structured specification also defines a class of models and therefore induces a deductive system . another topic to explore is the equivalence of algebraic specifications up to logical interpretation . as a starting point , it would be worth to explore the relation @xmath259 defined as follows : @xmath260 if there are interpretations @xmath220 and @xmath261 such that @xmath262 and @xmath263 . it is not difficult to see that @xmath183 implies @xmath264 and @xmath265 implies @xmath266 . more challenging seems to be a stronger equivalence , studied in the context of equivalence between deductive systems @xcite , which requires interpretations to be _ mutually _ inverse . last but not least , framing refinement by interpretation in the context of recent works on heterogeneous specification , raises interesting questions and opens the opportunity for computer - based support . actually , classical translations between logics ( , modal into first - order or the latter into equational logic ) are at the basis of hets @xcite , the heterogeneous specification framework . to go further in this direction entails the need to regard interpretations from an institutional point of view @xcite , as some sort of comorphisms , and develop on top of it a calculus of refinements by interpretation . the authors express their gratitude to the anonymous referees who raised a number of pertinent questions entailing a more precise characterisation of the paper s contributions and a clarification of their scope . this work was funded by erdf - european regional development fund through the compete programme ( operational programme for competitiveness ) and by national funds through the fct ( portuguese foundation for science and technology ) within project ` fcomp-01 - 0124-feder-028923 ` ( nasoni ) and the project ` pest - c / mat / ui4106/2011 ` with compete number ` fcomp-01 - 0124-feder-022690 ` ( cidma - ua ) . the first author also acknowledges the financial assistance by the projects getfun , reference ` fp7-people-2012-irses ` , and nociones de completud , reference ` ffi2009 - 09345 ` ( micinn - spain ) . a. madeira was supported by the fct within the project ` norte-01 - 0124-feder-000060 ` . r. m. burstall and j. a. goguen . the semantics of clear , a specification language . in d. bjrner , editor , _ abstract software specifications ( 1979 copenhagen winter school , january 22 - february 2 , 1979 ) _ , volume 86 of _ lecture notes in computer science _ , pages 292332 . springer , 1980 . m. bidoit and r. hennicker . an algebraic semantics for contract - based software components . in j. meseguer and g. rosu , editors , _ algebraic methodology and software technology ( amast 2008 - urbana , il , usa , july 28 - 31 , 2008 _ , volume 5140 of _ lecture notes in computer science _ , pages 216231 . springer , 2008 . m. codescu , f. horozal , m. kohlhase , t. mossakowski , and f. rabe . project abstract : logic atlas and integrator ( latin ) . in james davenport , william farmer , josef urban , and florian rabe , editors , _ intelligent computer mathematics _ , volume 6824 of _ lecture notes in computer science _ , pages 289291 . springer berlin / heidelberg , 2011 . b. dolle and w. dosch . transforming functional signatures of algebraic specifications into object - oriented class signatures . in _ apsec 05 : proceedings of the 12th asia - pacific software engineering conference _ , pages 323332 . ieee computer society , 2005 . j. fiadeiro and t. s. maibaum . generalising interpretations between theories in the context of ( @xmath5- ) institutions . in _ proceedings of the first imperial college department of computing workshop on theory and formal methods _ , pages 126147 , london , uk , 1993 . springer - verlag . j. fiadeiro and a. sernadas . structuring theories on consequence . in d. sanella and a. tarlecki , editors , _ recent trends in data type specification . specification of abstract data types ( papers from the fifth workshop on specification of abstract datac types , gullane , 1987 ) _ , volume 332 of _ lecture notes in computer science_. springer - verlag , berlin , 1988 . k. gdel . an interpretation of the intuitionistic proposicional calculus ( 1933 ) . in s. feferman et al , editor , _ collected works of kurt gdel ( vol . i ) _ , pages 301303 . oxford : oxford university press , 1986 . j. goguen , j. thatcher , and e. wagner . an initial algebra approach to the specification , correctness and implementation of abstract data types . in r. yeh , editor , _ current trends in programming methodology _ , pages 80149 . prentice - hall international , 1978 . j. goguen , t. winkler , j. meseguer , k. futatsugi , and j .- jouannaud . introducing obj . in j. goguen and g. malcolm , editors , _ software engineering with obj : algebraic specification in practice_. cambridge university press , 1996 . j. meseguer . general logics . in h. d. ebbinghaus , j. fernandez - prida , m. garrido , d. lascar , and m. rodriguez - artalejo , editors , _ logic colloqium87 _ , studies in logic and the foundations of mathematics ( volume 129 ) , pages 275330 . elsevier , 1989 . m. a. martins , a. madeira , and l. s. barbosa . refinement via interpretation . in d. v. hung and p. krishnan , editors , _ seventh ieee international conference on software engineering and formal methods ( sefm 2009 , hanoi , vietnam , 23 - 27 november 2009 ) _ , pages 250259 . ieee computer society , 2009 . t. mossakowski , c. maeder , and k. lttich . the heterogeneous tool set , hets . in o. grumberg and m. huth , editors , _ tools and algorithms for the construction and analysis of systems ( tacas 2007 - braga , portugal , march 24 - april 1 , 2007 ) _ , volume 4424 of _ lecture notes in computer science _ , pages 519522 . springer , 2007 . t. s. maibaum , m. r. sadler , and p. a. veloso . logical specification and implementation . in j. mathai and r. k. shyamasundar , editors , _ foundations of software technology and theoretical computer science ( fsttcs , bangalore , india , dec.13 - 15 , 1984 ) _ , volume 4424 of _ lecture notes in computer science _ , pages 1330 , london , uk , 1984 . springer . t. s. e. maibaum , paulo a. s. veloso , and m. r. sadler . a theory of abstract data types for program development : bridging the gap ? in hartmut ehrig , christiane floyd , maurice nivat , and james w. thatcher , editors , _ mathematical foundations of software development ( tapsoft , berlin , germany , march 25 - 29 , 1985 ) _ , volume 186 of _ lecture notes in computer science _ , pages 214230 . springer , 1985 . c. j. rodrigues , m. a. martins , a. madeira , and l. s. barbosa . refinement by interpretation in @xmath5-institutions . in john derrick , eerke a. boiten , and steve reeves , editors , _ proceedings 15th international refinement workshop _ , volume 55 of _ eptcs _ , pages 5364 , 2011 . d. sannella . the common framework initiative for algebraic specification and development of software : recent progress . in m. cerioli and g. reggio , editors , _ recent trends in algebraic development techniques ( revised selected papers of wadt 2001,genova , italy , april 1 - 3 , 2001 ) _ , volume 2267 of _ lecture notes in computer science _ , pages 328344 . springer , 2001 . j. da silva , i. m. dottaviano , and a. m. sette . in _ models , algebras , and proofs : selected papers of the x latin american symposium on mathematical logic , ( bogot , 1995 ) _ , pages 435448 . notes pure appl . math . ( 203 ) , 1999 . d. sannella and a. tarlecki . horizontal composability revisited . in k. futatsugi , j .- jouannaud , and j. meseguer , editors , _ algebra , meaning , and computation , essays dedicated to joseph a. goguen on the occasion of his 65th birthday _ , volume 4060 of _ lecture notes in computer science _ , pages 296316 . springer , 2006 . a. tarlecki . moving between logical systems . in magne haveraaen , olaf owe , and ole - johan dahl , editors , _ recent trends in data type specification ( selected papers of 11th workshop on specification of abstract data types joint with the 8th compass workshop , oslo , norway , september 19 - 23 , 1995 ) _ , volume 1130 of _ lecture notes in computer science _ , pages 478502 . springer , 1996 . b. yu , l. kong , y. zhang , and h. zhu . testing java components based on algebraic specifications . in _ first international conference on software testing , verification , and validation , icst 2008 _ , pages 190199 . ieee computer society , 2008 .

we present a model of selinger and valiron s quantum lambda calculus based on von neumann algebras , and show that the model is adequate with respect to the operational semantics .

1603.02133

in 1925 , heisenberg realised , pondering upon the problem of the spectral lines of the hydrogen atom , that a physical quantity such as the @xmath0-position of an electron orbiting a proton is best described not by a real number but by an infinite array of complex numbers @xcite . soon afterwards , born and jordan noted that these arrays should be multiplied as matrices are @xcite . of course , multiplying such infinite matrices may lead to mathematically dubious situations , which spurred von neumann to replace the infinite matrices by operators on a hilbert space @xcite . he organised these into _ rings of operators _ @xcite , now called _ von neumann algebras _ , and thereby set off an explosion of research ( also into related structures such as jordan algebras @xcite , orthomodular lattices @xcite , @xmath1-algebras @xcite , @xmath2-algebras @xcite , order unit spaces @xcite , hilbert @xmath1-modules @xcite , operator spaces @xcite , effect algebras @xcite , ) , which continues even to this day . one current line of research ( with old roots @xcite ) is the study of von neumann algebras from a categorical perspective ( see e.g. @xcite ) . one example relevant to this paper is kornell s proof that the opposite of the category @xmath3 of von neumann algebras with the obvious structure preserving maps ( i.e. the unital normal @xmath4-homomorphisms ) is monoidal closed when endowed with the spatial tensor product @xcite . he argues that @xmath5 should be thought of as the quantum version of @xmath6 . we would like to focus instead on the category of von neumann algebras and completely positive normal subunital maps , @xmath7 , as it seems more appropriate for modelling quantum computation : the full subcategory of @xmath8 consisting of finite dimensional von neumann algebras is equivalent to selinger s category @xmath9 @xcite , which is used to model first order quantum programming languages . r.3 @xmath10 \ar@{}[r]|\bot & \mathscr{c } \ar@/^1em/[r ] \ar@{}[r]|\bot \ar@/^1em/[l ] & \mathscr{d } \ar@/^1em/[l ] } $ ] on the syntactic side , in 2005 , selinger and valiron @xcite proposed a typed lambda calculus for quantum computation , and they studied it in a series of papers @xcite . a striking feature of this quantum lambda calculus is that functions naturally appear as data in the description of the deutch jozsa algorithm , teleportation algorithm and bell s experiment . although selinger and valiron gave a precise formulation of what might constitute a model of the quantum lambda calculus basically a pair of adjunctions , see figure [ fig : model ] , with some additional properties @xcite the existence of such a model ( other than the term model ) was an open problem for several years until malherbe constructed a model in his thesis using presheaves @xcite . the construction of malherbe s model is quite abstract , and it is ( perhaps because of this ) not yet known whether his model is adequate with respect to the operational semantics defined by selinger and valiron in @xcite ( see also @xcite ) . while several adequate models for variations on the quantum lambda calculus have been proposed in the meantime ( using the geometry of interaction in @xcite , and quantitative semantics in @xcite ) , malherbe s model remains the only model of the original quantum lambda calculus @xcite known in the literature , and so the existence of an adequate model for the quantum lambda calculus is still open . r.45 @xmath11^{\ell^\infty } \ar@{}[r]|(.45){\bot } & { { \mathbf{vna}}_{\mathrm{miu}}}^\mathrm{op } \ar@{^(->}@/^1em/[r ] \ar@{}[r]|\bot \ar@/^1em/[l]^{{\mathrm{nsp } } } & { { \mathbf{vna}}_{\mathrm{cpsu}}}^\mathrm{op } \ar@/^1em/[l ] } $ ] in this paper , we present the model of selinger and valiron s quantum lambda calculus , based on von neumann algebras , see figure [ fig : neumann - model ] , and we show that the model is adequate with respect to the operational semantics . we should note that it is possible to extend the quantum lambda calculus with recursion and inductive types , but that we have not yet been able to include these features in our model . the paper is divided in six sections . we begin with a short review of quantum computation ( in section [ sec : quantum - computation ] ) , and the quantum lambda calculus and its operational semantics ( in section [ sec : quantum - lambda - calculus - op - sem ] ) . we give the denotational semantics for the quantum lambda calculus using von neumann algebras and prove its adequacy in section [ sec : denotational - semantics ] . for this we use several technical results about the categories @xmath3 and @xmath7 of von neumann algebras , which we will discuss in section [ technical - result - vna ] . we end with a conclusion in section [ sec : final - remarks ] . in a nutshell , one gets the quantum lambda calculus by taking the simply typed lambda calculus with products and coproducts and adding a qubit type . this single ingredient dramatically changes the flavour of the whole system e.g. forcing one to make the type system linear , so we will spend some words on the behaviour of qubits in this section . for more details on quantum computation , see @xcite . a state of an isolated qubit is a vector @xmath12 of length @xmath13 in the hilbert space @xmath14 , and can be written as a complex linear combination ( `` superposition '' ) @xmath15 , since the vectors @xmath16 and @xmath17 form an orthonormal basis for @xmath14 . when qubits are combined to form a larger system , one can sometimes no longer speak about the state of the individual qubits , but only of the state of the whole system ( in which case the qubits are `` entangled '' ) . the state of a register of @xmath18 qubits is a vector @xmath12 of length @xmath13 in the @xmath18-fold tensor product @xmath19 , which has as an orthonormal basis the vectors of the form @xmath20 where @xmath21 . for the purposes of this paper there are three basic operations on registers of qubits . 1 . one can add a new qubit in state @xmath22 to a register of @xmath18 qubits in state @xmath12 , turning it to a register of @xmath23 qubits in state @xmath24 . a qubit in state @xmath25 can be added similarly . 2 . one can apply a unitary @xmath26 matrix @xmath27 to a register of @xmath18 qubits in state @xmath12 turning the state to @xmath28 . 3 . one can test the first qubit in the register . if the state of the register is written as @xmath29 where the length of @xmath30 and @xmath31 is @xmath13 , then the test comes out negative and changes the state of the register to @xmath32 with probability @xmath33 , and comes out positive with probability @xmath34 changing the state to @xmath35 . + measurement of the @xmath36-th qubit in the register is also possible and behaves similarly . a predicate on a register of @xmath18 qubits is a @xmath26 matrix @xmath37 such that both @xmath37 and @xmath38 are positive ( which is the case when @xmath37 is a projection ) . the probability that @xmath37 holds in state @xmath12 is @xmath39 . for example , given a state @xmath12 of a qubit , the projection @xmath40 ( which maps @xmath41 to @xmath42 ) represents the predicate `` the qubit is in state @xmath12 '' . thus the predicates on a qubit are part of the algebra @xmath43 of @xmath44 complex matrices . there is also an algebra for the bit , namely @xmath14 . a predicate on a bit is an element @xmath45 with @xmath46 , which is interpreted as `` the bit is true with probability @xmath47 , false with probability @xmath0 , and undefined with probability @xmath48 '' . an operation on a register of qubits may not only be described by the effect it has on states ( schrdinger s view ) , but also by its action on predicates ( heisenberg s view ) . 1 . the operation which takes a bit @xmath49 and returns a qubit in state @xmath50 is represented by the map @xmath51 given by @xmath52 . 2 . the operation which applies a unitary @xmath27 to a register of @xmath18 qubits is represented by the map @xmath53 given by @xmath54 . the operation which tests a qubit and returns the outcome is represented by the map @xmath55 given by @xmath56 . a general operation between _ finite dimensional _ quantum data types is usually taken to be a completely positive subunital linear map ( see below ) between direct sums of matrix algebras , @xmath57 . the category formed by these operations is equivalent to @xmath58 ( * ? ? ? von neumann algebras are a generalisation of direct sums of matrix algebras to infinite dimensions . formally , a von neumann algebra @xmath59 is a linear subspace of the bounded operators on a hilbert space @xmath60 , which contains the identity operator , @xmath13 , is closed under multiplication , involution , @xmath61 , and is closed in the weak operator topology , i.e. the topology generated by the seminorms @xmath62 where @xmath63 ( cf . @xcite ) . we believe that the opposite @xmath8 of the category of von neumann algebras and normal completely positive subunital maps ( definitions are given below ) might turn out to be the most suitable extension of @xmath9 to describe operations between ( possibly infinite dimensional ) quantum data types . indeed , to support this thesis , we will show that @xmath8 gives a model of the quantum lambda calculus . let us end this section with the definitions that are necessary to understand @xmath7 . an element @xmath64 of a von neumann algebra @xmath59 is _ self - adjoint _ if @xmath65 , and _ positive _ if @xmath66 for some @xmath67 . the self - adjoint elements of a von neumann algebra @xmath59 are partially ordered by : @xmath68 iff @xmath69 is positive . any upwards directed bounded subset @xmath70 of self - adjoint elements of a von neumann algebra @xmath59 has a supremum @xmath71 in the set of self - adjoint elements of @xmath59 ( * ? ? ? ( so a von neumann algebra resembles a domain . ) the linear maps between von neumann algebras which preserve the multiplication , involution , @xmath61 , and unit , @xmath13 , are called unital @xmath4-homomorphisms in the literature and _ miu - maps _ by us . a linear map @xmath72 between von neumann algebras is _ positive _ if it maps positive elements to positive elements , _ unital _ if it preserves the unit , _ subunital _ if @xmath73 , and _ if @xmath72 is positive and preserves suprema of bounded directed sets of self - adjoint elements . ( if subunital maps are akin to partial maps between sets , then the unital maps are the total maps . normality is the incarnation of scott continuity in this setting , and coincides with continuity with respect to the @xmath74-weak = ultraweak = weak * topology ( * ? ? 1.13.2 ) . ) given a von neumann algebra @xmath59 on a hilbert space @xmath60 , and a von neumann algebra @xmath75 on a hilbert space @xmath76 , the _ spatial tensor product _ @xmath77 of @xmath59 and @xmath75 is the least von neumann algebra on @xmath78 which contains all operators of the form @xmath79 where @xmath80 for all @xmath81 , @xmath82 , @xmath63 and @xmath83 @xcite . ( the tensor product @xmath84 may be physically interpreted as the composition of the systems @xmath59 and @xmath75 recall that a register of two qubits is represented by the von neumann algebra @xmath85 . ) given normal positive @xmath86 and @xmath87 there might be a normal positive linear map @xmath88 given by @xmath89 . an interesting , and annoying , phenomenon is that such @xmath90 need not exist for all @xmath72 and @xmath91 . this warrants the following definition : if @xmath86 is a positive linear map such that for every natural number @xmath18 the map @xmath92 is positive , then @xmath72 is called _ completely positive _ @xcite . here @xmath93 is the von neumann algebra of @xmath94 matrices with entries drawn from @xmath59 , and @xmath95 for all @xmath96 and @xmath97 . if @xmath72 and @xmath91 are normal and completely positive , then @xmath90 exists , and is completely positive ( * ? ? iv/5.13 ) . we review the quantum lambda calculus for which we will give a denotational semantics . the language and its operational semantics are basically the same as selinger and valiron s ones @xcite , but with sum type @xmath98 @xcite and ` indexed ' terms @xcite , see remark [ rem : indexedterms ] and notation [ notation : sumtype ] below . for space reasons we omit many details , and refer to @xcite . @xmath99[a , b]{m } \mid \tinr[n][a , b]{n } \mid { \mathtt{match}\;{l}\;\mathtt{with } \if\relax\detokenize{n}\relax\;\else^{n}\fi ( { x^a}\;{\mapsto}\;{m } \;|\;{y^b}\;{\mapsto}\;{n } ) } \end{aligned } \\ \text{\emph{value}}\;\ ; v , w \coloneqq { } x^a\mid { \mathop{\mathtt{new}}\nolimits}^a\mid { \mathop{\mathtt{meas}}\nolimits}^a\mid u^a\mid { \mathord{*}}^n \mid { \lambda^{n}x^a.m}\mid { \langlev , w\rangle}^n \mid \tinl[n][a , b]{v } \mid \tinr[n][a , b]{w}\end{gathered}\ ] ] the language consists of _ types _ , _ terms _ and _ values _ defined in table [ tab : types - terms ] . we use obvious shorthand @xmath100 and @xmath101 . the _ subtyping relation _ @xmath102 on types is defined by the rules shown in table [ tab : typing - rules](a ) . in the definition of terms and values , @xmath103 is a natural number ; @xmath0 ranges over variables ; and @xmath27 ranges over @xmath104 unitary matrices for @xmath105 . the ( nullary ) constructors @xmath106 are called _ constants _ and sometimes referred to by @xmath107 . clearly , values form a subclass of terms . as usual , we identify terms up to @xmath108-equivalence . [ rem : indexedterms ] the terms are _ indexed terms _ of @xcite , which have explicit type annotations ( cf . church - style vs. curry - style in the simply - typed lambda calculus ) . a typing derivation for an indexed term is unique in a suitable sense , so that we can more easily obtain lemma [ lem : interpretation - well - defd ] . in fact , for the language of @xcite we can safely remove the type annotations ( * ? ? ? * corollary 1 ) . we conjecture that the same is true for our language , which is left as a future work . [ notation : sumtype ] following @xcite ( and @xcite ) , the language has sum type @xmath98 instead of the @xmath109 type ( which exists in @xcite ) . the @xmath109 type and its constructors are emulated by @xmath110 ; @xmath111[{\top},{\top}]{{\mathord{*}}^n}$ ] ; @xmath112[{\top},{\top}]{{\mathord{*}}^n}$ ] ; and @xmath113 , with fresh variables @xmath114 . the set @xmath115 of _ free variables _ is defined in the usual way . a _ context _ is a list @xmath116 of variables @xmath117 and types @xmath118 where the variables @xmath117 are distinct . we write @xmath119 and @xmath120 . we also write @xmath121 for the context restricted to the free variables of @xmath122 . a _ typing judgement _ , written as @xmath123 , consists of a context @xmath124 , a term @xmath122 and a type @xmath125 . a typing judgement is _ valid _ if it can be derived by the _ typing rules _ shown in table [ tab : typing - rules](b ) . in the rule ( [ rule@ax2 ] ) , @xmath107 ranges over @xmath126 , @xmath127 and @xmath104 unitary matrices @xmath27 ; and the types @xmath128 are defined as follows : @xmath129 , @xmath130 , @xmath131 . * ( a ) * rules for subtyping , with a condition @xmath132 for each rule 2.0ex plus.8ex minus.4ex 2.0ex plus.8ex minus.4ex 2.0ex plus.8ex minus.4ex 2.0ex plus.8ex minus.4ex 2.0ex plus.8ex minus.4ex 2.0ex plus.8ex minus.4ex * ( b ) * typing rules the type system is affine ( weak linear ) . each variable may occur at most once , unless it has a duplicable type @xmath133 . substitution of the following form is admissible . [ lem : subst ] if @xmath134 and @xmath135 , where @xmath136 is a value and @xmath137 , then @xmath138}:b$ ] . note , however , that we need to define the substitution @xmath139}$ ] with care . for example , if @xmath140 , @xmath141 and @xmath142 , then we substitute @xmath143 ( not @xmath144 ) for @xmath145 in @xmath122 . see @xcite or @xcite for details . the operational semantics is taken from @xcite , but is adapted for indexed terms . a _ quantum closure _ is a triple @xmath146}$ ] with @xmath147 where : * @xmath148 is a normalised vector of the hilbert space @xmath149 . * @xmath150 is a list of @xmath151 distinct variables , written as @xmath152 . we write @xmath153 , and @xmath154 for the position of a variable in the list . * @xmath122 is a term with @xmath155 . we say a quantum closure @xmath156}$ ] is _ well - typed of type @xmath125 _ , written as @xmath157 , if the typing judgement @xmath158 is valid . we call @xmath159}$ ] a _ value closure _ if @xmath136 is a value . * proof : * * endproof * @xmath160 } { \to_{1 } } { [ \,\ket{\psi},\psi,{m[v / x]}\ , ] } \\ \tag{$\otimes$}{\label{rule@redu : beta.tens } } { [ \,\ket{\psi},\psi,{\mathtt{let}\;{{\langlex^a , y^b\rangle}^n}={{\langlev , w\rangle}^n}\;\mathtt{in}\;{m}}\ , ] } { \to_{1 } } { [ \,\ket{\psi},\psi,{m[v / x , w / y]}\ , ] } \\ \tag{$\oplus_1$}{\label{rule@redu : beta.sum1 } } { [ \,\ket{\psi},\psi,{\mathtt{match}\;{\tinl[n][a , b]{v}}\;\mathtt{with } \if\relax\detokenize{n}\relax\;\else^{n}\fi ( { x^a}\;{\mapsto}\;{m } \;|\;{y^b}\;{\mapsto}\;{n})}\ , ] } { \to_{1 } } { [ \,\ket{\psi},\psi,{m[v / x]}\ , ] } \\ \tag{$\oplus_2$}{\label{rule@redu : beta.sum2 } } { [ \,\ket{\psi},\psi,{\mathtt{match}\;{\tinr[n][a , b]{w}}\;\mathtt{with } \if\relax\detokenize{n}\relax\;\else^{n}\fi ( { x^a}\;{\mapsto}\;{m } \;|\;{y^b}\;{\mapsto}\;{n})}\ , ] } { \to_{1 } } { [ \,\ket{\psi},\psi,{n[w / y]}\,]}\end{gathered}\ ] ] * ( a ) * classical control @xmath161 } { \to_{1 } } { [ \,\ket{\psi'},\psi,{\langlex_1^{{\mathtt{qbit}}},\dotsc , x_k^{{\mathtt{qbit}}}\rangle}^0\ , ] } \\ \tag{${\mathop{\mathtt{meas}}\nolimits}_0$}{\label{rule@redu : meas0 } } { [ \,\ket{\psi},\ket{x_1\dotso x_m},{\mathop{\mathtt{meas}}\nolimits}^{{\mathtt{qbit}}{\mathbin{\multimap}}{\mathord{!}}^n{\mathtt{bit } } } x_i^{{\mathtt{qbit}}}\ , ] } { \to_{p_0 } } { [ \,\ket{\psi_0},\ket{x_1\dotso x_m},{\mathtt{f\!f}}^n\ , ] } \\ \tag{${\mathop{\mathtt{meas}}\nolimits}_1$}{\label{rule@redu : meas1 } } { [ \,\ket{\psi},\ket{x_1\dotso x_m},{\mathop{\mathtt{meas}}\nolimits}^{{\mathtt{qbit}}{\mathbin{\multimap}}{\mathord{!}}^n{\mathtt{bit } } } x_i^{{\mathtt{qbit}}}\ , ] } { \to_{p_1 } } { [ \,\ket{\psi_1},\ket{x_1\dotso x_m},{\mathtt{t\!t}}^n\ , ] } \\ \tag{${\mathop{\mathtt{new}}\nolimits}_0$}{\label{rule@redu : new0 } } { [ \,\ket{\psi},\ket{x_1\dotso x_m},{\mathop{\mathtt{new}}\nolimits}^{a{\mathbin{\multimap}}{\mathtt{qbit}}}\tilde{{\mathtt{f\!f}}}\ , ] } { \to_{1 } } { [ \,\ket{\psi}\ket{0},\ket{x_1\dotso x_m y},y^{{\mathtt{qbit}}}\ , ] } \\ \tag{${\mathop{\mathtt{new}}\nolimits}_1$}{\label{rule@redu : new1 } } { [ \,\ket{\psi},\ket{x_1\dotso x_m},{\mathop{\mathtt{new}}\nolimits}^{a{\mathbin{\multimap}}{\mathtt{qbit}}}\tilde{{\mathtt{t\!t}}}\ , ] } { \to_{1 } } { [ \,\ket{\psi}\ket{1},\ket{x_1\dotso x_m y},y^{{\mathtt{qbit}}}\,]}\end{gathered}\ ] ] * ( b ) * quantum data if @xmath146}{\to_{p}}{[\,\ket{\psi'},\psi',m'\,]}$ ] , the following are valid reductions ( if well - formed ) . @xmath162 } { \to_{p } } { [ \,\ket{\psi'},\psi',m'n\ , ] } \qquad { [ \,\ket{\psi},\psi , vm\ , ] } { \to_{p } } { [ \,\ket{\psi'},\psi',vm'\ , ] } \\ { [ \,\ket{\psi},\psi,{\langlem , n\rangle}^n\ , ] } { \to_{p } } { [ \,\ket{\psi'},\psi',{\langlem',n\rangle}^n\ , ] } \qquad { [ \,\ket{\psi},\psi,{\langlev , m\rangle}^n\ , ] } { \to_{p } } { [ \,\ket{\psi'},\psi',{\langlev , m'\rangle}^n\ , ] } \\ { [ \,\ket{\psi},\psi,{\mathtt{let}\;{{\langlex^a , y^b\rangle}^n}={m}\;\mathtt{in}\;{n}}\ , ] } { \to_{p } } { [ \,\ket{\psi'},\psi',{\mathtt{let}\;{{\langlex^a , y^b\rangle}^n}={m'}\;\mathtt{in}\;{n}}\ , ] } \\ { [ \,\ket{\psi},\psi,\tinl[n][a , b]{m}\ , ] } { \to_{p } } { [ \,\ket{\psi'},\psi',\tinl[n][a , b]{m'}\ , ] } \\ { [ \,\ket{\psi},\psi,\tinr[n][a , b]{m}\ , ] } { \to_{p } } { [ \,\ket{\psi'},\psi',\tinr[n][a , b]{m'}\ , ] } \\ { [ \,\ket{\psi},\psi,{\mathtt{match}\;{m}\;\mathtt{with } \if\relax\detokenize{n}\relax\;\else^{n}\fi ( { x^a}\;{\mapsto}\;{n } \;|\;{y^b}\;{\mapsto}\;{l})}\ , ] } { \to_{p } } { [ \,\ket{\psi'},\psi',{\mathtt{match}\;{m'}\;\mathtt{with } \if\relax\detokenize{n}\relax\;\else^{n}\fi ( { x^a}\;{\mapsto}\;{n } \;|\;{y^b}\;{\mapsto}\;{l})}\,]}\end{gathered}\ ] ] * ( c ) * congruence rules a ( small - step ) _ reduction _ @xmath163 consists of quantum closures @xmath164 and @xmath165 $ ] , meaning that @xmath37 reduces to @xmath166 with probability @xmath167 . the valid reductions @xmath163 are given inductively by the _ reduction rules _ shown in table [ tab : reduction - rules ] . in the rules , @xmath136 and @xmath168 refer to values . * proof : * * endproof*the ` quantum data ' rules ( b ) correspond to the three basic operations explained in [ sec : quantum - computation ] . in the rule ( [ rule@redu : u ] ) , @xmath169 is the state obtained by applying the @xmath104 unitary matrix @xmath27 to the @xmath170 qubits of the position @xmath171 in @xmath148 . * proof : * * endproof*in the rule ( [ rule@redu : meas0 ] ) , @xmath172 is the probability that we obtain @xmath173 ( ` negative ' in terms of [ sec : quantum - computation ] ) by measuring the @xmath36-th qubit of @xmath148 ; and @xmath30 is the state after that . the rule ( [ rule@redu : meas1 ] ) is similar . * proof : * * endproof*in the rule ( [ rule@redu : new0 ] ) , we denote by @xmath174 any term of the form @xmath175[{\mathord{!}}^k{\top},{\mathord{!}}^h{\top}]{{\mathord{*}}^{n+k}}$ ] ( cf . notation [ notation : sumtype ] ) . the term @xmath176 in ( [ rule@redu : new1 ] ) is similar . reduction satisfies the following properties . if @xmath157 and @xmath163 , then @xmath177 . * proof : * * endproof * [ lem : progress ] let @xmath157 be a well - typed quantum closure . then either @xmath37 is a value closure , or there exists a quantum closure @xmath166 such that @xmath163 . in the latter case , there are at most two ( up to @xmath108-equivalence ) single - step reductions from @xmath37 , and the total probability of all the single - step reductions from @xmath37 is @xmath13 . the next definitions follow @xcite . we define the _ small - step reduction probability _ @xmath178 $ ] for well - typed quantum closures @xmath164 by : @xmath179 if @xmath180 ; @xmath181 if @xmath136 is a value closure ; @xmath182 otherwise . lemma [ lem : progress ] guarantees that @xmath183 is a probabilistic system in a suitable sense . for a well - typed quantum closure @xmath37 and a well - typed value closure @xmath184 , the _ big - step reduction probability _ @xmath185 $ ] is defined by @xmath186 , where @xmath187 and @xmath188 . for each @xmath189 , we define @xmath190 , where @xmath191 is the set of well - typed quantum closures of the form @xmath192}$ ] . we will use a strong normalisation result . the proof is similar to ( * ? ? ? * lemma 33 ) . [ lem : str - norm ] let @xmath157 be a well - typed quantum closure . then there is no infinite sequence of reductions @xmath193 . clearly it suffices to prove the strong normalisation for the underlying ( non - deterministic ) reductions @xmath194 on terms . we add a constant @xmath195 to replace free variables @xmath196 . we then define a translation @xmath197 from the quantum lambda calculus ( with @xmath195 ) to a simply - typed lambda calculus with product , unit , sum types and constants @xmath198 . the translation @xmath197 forgets the @xmath199 modality , and translates the let constructor via @xmath200 . we can prove the strong normalisation for the simply - typed lambda calculus via standard techniques . we need the following notation and facts concerning von neumann algebras . those facts for which we could not find proof in the literature will be discussed in the next section . let @xmath201 be the symmetric monoidal category ( smc ) of von neumann algebras and normal miu - maps ( * ? ? ? 7.2 ) , and @xmath202 the smc of von neumann algebras and normal cpsu - maps ( where @xmath203 is the spatial tensor product ) @xcite . note that the unit @xmath204 is initial in @xmath3 ( but not in @xmath7 ) . both categories have products given by direct sums @xmath98 ( with the supremum norm ( * ? ? ? * def . 3.4 ) ) . to interpret the quantum lambda calculus , we will use the following pair of ( lax ) symmetric monoidal adjunctions , @xmath205_-{{\ell^{\infty } } } \ar@{}[r]|-{\bot } & \ar@/_2ex/[l]_-{{\mathrm{nsp } } } ( { { \mathbf{vna}}_{\mathrm{miu}}},\otimes,{\mathbb{c } } ) \ar@/_2ex/@{^ { ( } ->}[r]_{{\mathcal{j } } } \ar@{}[r]|{\bot } & ( { { \mathbf{vna}}_{\mathrm{cpsu}}},\otimes,{\mathbb{c } } ) \ar@/_2ex/[l]_{{\mathcal{f } } } } \ ] ] where @xmath206 is the opposite of the category @xmath6 of sets and functions , considered as a smc via cartesian products ( i.e. coproducts in @xmath206 ) . the functor @xmath207 is the inclusion functor ; the other functors are explained in the next section . note that @xmath207 is strict symmetric monoidal and strictly preserves products . the following facts are important : * @xmath3 is a co - closed smc @xcite . this means the endofunctor @xmath208 on @xmath3 has a left adjoint @xmath209 . the von neumann algebra @xmath210 is called the _ free exponential _ in @xcite . * the counit of the adjunction @xmath211 is an isomorphism ( see corollary [ cor : first - adjunction - unit - iso ] ) . * the functors @xmath212 and the adjunction @xmath211 are _ strong _ monoidal ( see corollary [ cor : first - adjunction - monoidal ] ) . * moreover , the functors @xmath213 and @xmath214 preserves products ( see cor . [ cor : sp - preserves - coproducts - and - tensors ] and lem . [ lem : ell - tensor ] ) . the tensor product @xmath203 distributes over products @xmath98 in @xmath3 , as @xmath215 , since @xmath216 is a right adjoint and thus preserves products . we denote the canonical isomorphism by @xmath217 . r.45 we define a ` kleisli co - exponential ' @xmath218 by @xmath219 . we have the bijective correspondence as shown on the right . we write @xmath220 for the miu - map @xmath221 corresponding to @xmath72 . we also write @xmath222 for the co - evaluation map , i.e. the cpsu - map corresponding to @xmath223 . then @xmath224 by the naturality of the bijective correspondence . we write @xmath225 for the strong symmetric monoidal monad on @xmath3 induced by the left - hand adjunction of . the unit and multiplication are denoted by @xmath226 and @xmath227 respectively . from the fact that the counit of @xmath211 is an isomorphism , it easily follows that @xmath228 is an idempotent monad , i.e. the multiplication @xmath229 is an isomorphism . note also that @xmath228 preserves products . we denote the structure isomorphisms by : @xmath230 ; @xmath231 ; and @xmath232 . because the adjunction @xmath211 satisfies a dual condition to a linear - non - linear model @xcite ( see also @xcite ) , the monad @xmath228 has a property which is dual to a _ linear exponential comonad_. thus each object of the form @xmath233 is equipped with a map @xmath234 which , with a unique map @xmath235{180}{$!}$}}}_{{\mathcal{l}}{\mathscr{a}}}\colon{\mathbb{c}}\to{\mathcal{l}}{\mathscr{a}}$ ] , makes @xmath233 into a @xmath203-monoid in @xmath3 . [ rem : concrete - model - of - qlc ] one can summarise these facts by saying that the opposite @xmath236 is a _ ( weak ) linear category for duplication _ @xcite ; and moreover @xmath236 is a _ concrete model of the quantum lambda calculus _ defined by selinger and valiron @xcite . although they gave the definition of concrete models of the quantum lambda calculus , results on them ( e.g. how to interpret the quantum lambda calculus ; adequacy of models ) have never been given . in the remainder of the section , therefore , we will give the interpretation of the language in von neumann algebras concretely , and then prove its adequacy . we interpret types as von neumann algebras , i.e. objects in @xmath3 / @xmath7 , as follows . @xmath237 one familiar with fock space might be surprised to realise that @xmath238 , because there is no normal miu - map @xmath239 . the intuition here may be that no part of a qubit can be duplicated , and so the assumption of a duplicable qubit amounts to nothing . this is also the interpretation of @xmath240 intended by selinger and valiron , see @xcite . the interpretation of a function type @xmath241 is obtained by abstract means , and at this point we know very little about it . ( might it be as intangible as an ultrafilter ? ) however , applying @xmath199 makes the function type almost trivial : after 4 , it will be clear that @xmath242 the interpretation of the subtyping relation @xmath243 is a ` canonical ' map @xmath244 in @xmath3 , which exists uniquely by a coherence property for an idempotent ( co)monad ; see @xcite for details . for instance , we have @xmath245 . contexts @xmath116 are interpreted as @xmath246 . we shall treat the monoidal structure @xmath247 as if it were strict monoidal , which is justified by the coherence theorem for monoidal categories . the interpretations @xmath248 , @xmath249 and @xmath250 of constants are defined using the maps @xmath251 , @xmath252 and @xmath253 given in [ sec : quantum - computation ] , as follows . @xmath254 we now give the interpretation @xmath255 of a typing judgement as a map @xmath256 in @xmath7 . the definition is similar to @xcite . first we define a normal cpsu - map @xmath257 ( recall that @xmath121 ) by induction on the derivation of the typing judgement as shown in table [ tab : denotation - typing - judgements ] . we then define @xmath258 . here and in table [ tab : denotation - typing - judgements ] , we use the following notations ( often suppressing subscripts ) . let @xmath259 denote the symmetry isomorphism . for contexts @xmath260 , we write @xmath261 for the ` injection ' map defined via unique miu - maps @xmath235{180}{$!}$}}}_{{\mathscr{a}}}\colon{\mathbb{c}}\to{\mathscr{a}}$ ] . * proof : * * endproof*for a context @xmath262 , we define the map @xmath263 via monoid structures @xmath264 and symmetry maps @xmath265 . the map @xmath266 can be defined using @xmath267 . we write @xmath268 for @xmath269 ; @xmath270 for @xmath271 ; and @xmath272 similarly . projection maps and tupling for direct sums , products in @xmath7 , are denoted by @xmath273 and @xmath274 . * proof : * * endproof * 2.0ex plus.8ex minus.4ex @xmath275 @xmath276 2.0ex plus.8ex minus.4ex @xmath277 2.0ex plus.8ex minus.4ex where : 2.0ex plus.8ex minus.4ex 2.0ex plus.8ex minus.4ex 2.0ex plus.8ex minus.4ex 2.0ex plus.8ex minus.4ex 2.0ex plus.8ex minus.4ex 2.0ex plus.8ex minus.4ex 2.0ex plus.8ex minus.4ex 2.0ex plus.8ex minus.4ex note that the interpretation @xmath255 is defined by induction on typing derivations . because we use indexed terms , it is not hard to prove the following fact by induction on a typing derivation @xmath278 . [ lem : interpretation - well - defd ] suppose that @xmath123 is valid with a derivation @xmath278 , and so is @xmath279 with @xmath280 . then @xmath281 , where @xmath282 is a ( unique by coherence ) isomorphism that permutes @xmath283 to @xmath284 . in particular , @xmath255 is well - defined , not depending on derivations . let @xmath285}:a$ ] be a well - typed quantum closure . the mapping @xmath286 defines a normal cpu - map @xmath287 . the interpretation of the quantum closure is defined by : @xmath288}:a } \;\coloneqq\ ; \sem{a } \xrightarrow{\sem{x_1:{\mathtt{qbit}},\dotsc , x_n:{\mathtt{qbit}}{\mathrel{\vartriangleright}}m : a } } { \mathcal{m}}_2^{\otimes n } \xrightarrow{\braket{\psi}{-}{\psi } } { \mathbb{c}}\ ] ] the next soundness / invariance for the small - step reduction is a key result to obtain adequacy . note that for normal cpsu - maps @xmath289 and @xmath290 $ ] with @xmath291 , the ( convex ) sum @xmath292 of maps is defined in the obvious pointwise manner and is a normal cpsu - map . [ prop : soundness - small - step ] let @xmath157 be a well - typed quantum closure . then @xmath293 . see appendix [ sec : soundness - small - step ] . [ soundness - n - small - step ] let @xmath157 be a well - typed quantum closure . then for all @xmath294 , @xmath295 by induction on @xmath18 . the base case is proposition [ prop : soundness - small - step ] . for the induction step , @xmath296 [ prop : soundness - big - step ] let @xmath157 be a well - typed quantum closure . then @xmath297 , where @xmath184 runs over well - typed value closures . by lemmas [ lem : progress ] and [ lem : str - norm ] , @xmath298 for some @xmath151 it is then easy to obtain @xmath299 by induction on @xmath151 , using proposition [ prop : soundness - small - step ] . let @xmath300 be a quantum closure of type @xmath109 . for the interpretation @xmath301 , we have @xmath302 and @xmath303 . by proposition [ prop : soundness - big - step ] we have @xmath304 . note that for each well - typed value closure @xmath159}:{\mathtt{bit}}$ ] , either @xmath305 or @xmath306 . then the assertion follows since @xmath307}:{\mathtt{bit}}}(\lambda,\rho)=\lambda$ ] and @xmath308}:{\mathtt{bit}}}(\lambda,\rho)=\rho$ ] . let us sketch how we obtained the two monoidal adjunctions in . let @xmath309 denote the von neumann algebra of bounded maps @xmath310 on a set @xmath311 . addition , multiplication , involution , suprema , and so on , are computed coordinatewise in @xmath309 . in fact , @xmath309 is simply the @xmath311-fold product in @xmath3 of @xmath312 with @xmath313 as @xmath0-th projection . we extend @xmath314 to a functor @xmath315 by defining @xmath316 for every map @xmath317 ( in @xmath6 ) and @xmath318 . let @xmath319 be the ` normal spectrum ' of a von neumann algebra @xmath59 , i.e. the set of normal miu - maps @xmath320 . we extend @xmath321 to a functor @xmath322 by defining @xmath323 for every normal miu - map @xmath324 and @xmath325 ( it is simply a hom - functor @xmath326 ) . note that any normal miu - map @xmath327 gives a map @xmath328 by `` swapping arguments '' @xmath329 and with a little bit more work , we get : there is an adjunction @xmath211 . the following two lemmas describe the normal spectrum of direct products and tensors of von neumann algebras , and can be proven using standard techniques . let @xmath330 be a set , and for each @xmath331 , let @xmath332 be a von neumann algebra . for each @xmath333 , there is @xmath331 and @xmath334 with @xmath335 . let @xmath336 and @xmath337 be von neumann algebras . then for every @xmath338 there are unique @xmath339 and @xmath340 with @xmath341 for all @xmath342 . [ cor : sp - preserves - coproducts - and - tensors ] the functor @xmath322 preserves products , and tensors . using that @xmath309 is the @xmath311-fold product of @xmath312 in @xmath3 we get : [ cor : first - adjunction - unit - iso ] the counit of the adjunction @xmath343 is an isomorphism . [ lem : ell - tensor ] let @xmath311 and @xmath344 be sets . there is a normal miu - isomorphism @xmath345 use the proof of proposition 9.2 from @xcite . [ cor : first - adjunction - monoidal ] the adjunction @xmath211 is strong monoidal . let us turn to the second adjunction in . in @xcite it is shown how the following result follows from freyd s adjoint functor theorem ( see theorem v.6.2 of @xcite ) . [ thm : second - adjoint ] the inclusion @xmath346 has a left adjoint . [ cor : second - adjoint ] the category @xmath7 is isomorphic to the co - kleisli category of the comonad @xmath347 on @xmath3 induced by @xmath348 . see theorem 9 of @xcite , or do exercise vi.5.2 of @xcite ( and use the fact that an equivalence of categories which is bijective on objects is an isomorphism ) . [ cor : second - adjoint - monoidal ] the adjunction @xmath349 is symmetric monoidal . clearly , @xmath346 is strict symmetric monoidal . from this fact alone , it follows that the adjunction @xmath349 is symmetric monoidal , see prop . 14 of @xcite . in our model of the quantum lambda calculus the von neumann algebras of the form @xmath350 serve as the interpretation of the _ duplicable types _ ( of the form @xmath351 ) , because @xmath350 carries a @xmath203-monoid structure . among all von neumann algebras @xmath309 is arguably quite special and one might wonder if there is a broader class of von neumann algebras that might serve as the interpretation of duplicable types ( such as the class of all commutative von neumann algebras , which includes @xmath352 $ ] ) . the following result settles this matter : no . due to space constraints , the proof will appear somewhere else . [ thm : duplicable ] for a von neumann algebra @xmath59 the following are equivalent . 1 . there is a _ duplicator _ on @xmath59 , that is , a normal positive unital map @xmath353 such that @xmath354 and @xmath355 for all @xmath356 . [ thm : duplicable-1 ] 2 . [ thm : duplicable-2 ] @xmath59 is isomorphic to @xmath350 for some set @xmath311 . moreover , there is at most one duplicator on @xmath59 . @xmath357 is the free @xmath203-monoid on @xmath59 from @xmath3 . we have given a rather concrete proof of adequacy for the sake of clarity . however , it seems that we only used the fact that @xmath3 is a ` concrete model of the quantum lambda calculus ' ( see remark [ rem : concrete - model - of - qlc ] ) , and that @xmath7 is ` suitably ' enriched over convex sets . thus an abstract result might be distilled from our work stating that any concrete model of the quantum lambda calculus is adequate when suitably enriched over convex sets , but we have not pursued this . we believe selling points of our model are that it is a model for selinger and valiron s original quantum lambda calculus @xcite ( in selinger and valiron s linear fragment @xcite the @xmath199 modality is absent ; in hasuo and hoshino s language @xcite the tensor type @xmath358 does not represent two qubits ; and only function types may be duplicable , @xmath359 , in the language of pagani et al . @xcite ) ; that it is adequate ( malherbe s model @xcite is not known to be ) ; that the interpretation of @xmath360 is rather simple ; and that it is formed using von neumann algebras , a mathematical classic . we believe our model could be improved by a more concrete description of @xmath361 ( as all the other models have ) , and by features such as recursion and inductive types ( present in e.g. hasuo and hoshino s and pagani s models ) , which leaves us with ample material for future research . [ [ acknowledgements ] ] acknowledgements we thank chris heunen for spotting a typo . * proof : * * endproof * we need some results on the denotational semantics . [ lem : denot - subst ] suppose that @xmath362 and @xmath135 , so that @xmath138}:b$ ] by lemma [ lem : subst ] . then the following diagram commute . @xmath363 ^ -{\sem{{\mathord{!}}\delta,\gamma_1,\gamma_2{\mathrel{\vartriangleright}}{m[v / x]}:b } } \ar[d]_{\sem{{\mathord{!}}\delta,\gamma_2,x : a{\mathrel{\vartriangleright}}m : b } } & \sem{{\mathord{!}}\delta,\gamma_1,\gamma_2 } \\ \sem{{\mathord{!}}\delta,\gamma_1}\otimes\sem{a } \ar[r]^-{{\mathrm{id}}\otimes\sem{{\mathord{!}}\delta,\gamma_2{\mathrel{\vartriangleright}}v : a } } & \sem{{\mathord{!}}\delta,\gamma_1}\otimes \sem{{\mathord{!}}\delta,\gamma_2 } \ar[u]_{{{\mathrm{merge } } } } } \ ] ] [ lem : denot - beta - redu ] we have the following equations , when terms @xmath364 and values @xmath365 are appropriately well - typed . @xmath366 } } \\ \sem{{\mathtt{let}\;{{\langlex^a , y^b\rangle}^n}={{\langlev , w\rangle}^n}\;\mathtt{in}\;{m } } } & = \sem{{m[v / x , w / y ] } } \\ \sem{{\mathtt{match}\;{\tinl[n][a , b]{v}}\;\mathtt{with } \if\relax\detokenize{n}\relax\;\else^{n}\fi ( { x^a}\;{\mapsto}\;{m } \;|\;{y^b}\;{\mapsto}\;{n } ) } } & = \sem{{m[v / x ] } } \\ \sem{{\mathtt{match}\;{\tinr[n][a , b]{w}}\;\mathtt{with } \if\relax\detokenize{n}\relax\;\else^{n}\fi ( { x^a}\;{\mapsto}\;{m } \;|\;{y^b}\;{\mapsto}\;{n } ) } } & = \sem{{n[w / y]}}\end{aligned}\ ] ] here we abbreviate @xmath255 to @xmath367 . let @xmath369}:a$ ] be a well - typed quantum closure such that @xmath370 for all @xmath371 . then we define : @xmath372}:a}^{(l ) } = { \mathcal{m}}_2^{\otimes l } \otimes \sem{a } \xrightarrow{{\mathrm{id}}\otimes \sem{x_{l+1}:{\mathtt{qbit}},\dotsc , x_m:{\mathtt{qbit}}{\mathrel{\vartriangleright}}m : a } } { \mathcal{m}}_2^{\otimes m } \xrightarrow{\braket{\psi}{-}{\psi } } { \mathbb{c}}\ ] ] consider the induction step for @xmath375 , and the case where @xmath122 is not a value . then the only possible reductions from @xmath376}$ ] are @xmath377}{\to_{p}}{[\,\psi',\psi',m'n\,]}$ ] when @xmath378}{\to_{p}}{[\,\psi',\psi',m'\,]}$ ] . without loss of generality , which keeps the first @xmath379 variables , with the permutation of the corresponding qubits in @xmath148 , does not change @xmath380 . the same is true for the operational semantics @xcite . ] we may assume that @xmath381 such that @xmath382 and @xmath383 . we will simply write @xmath367 for @xmath384 and @xmath385 similarly . then @xmath386}:b}^{(l ) } \\ & = { \mathcal{m}}_2^{\otimes l}\otimes \sem{b } \xrightarrow{{\mathrm{id}}\otimes\varepsilon } { \mathcal{m}}_2^{\otimes l}\otimes \sem{a{\mathbin{\multimap}}b}\otimes \sem{a } \xrightarrow{{\mathrm{id}}\otimes\sem{m}\otimes\sem{n } } { \mathcal{m}}_2^{\otimes l}\otimes { \mathcal{m}}_2^{\otimes h } \otimes { \mathcal{m}}_2^{\otimes k } \\ & \qquad\xrightarrow{{\mathrm{id}}\otimes\gamma } { \mathcal{m}}_2^{\otimes l}\otimes { \mathcal{m}}_2^{\otimes k } \otimes { \mathcal{m}}_2^{\otimes h } \xrightarrow{\braket{\psi}{-}{\psi } } { \mathbb{c}}\\ & = { \mathcal{m}}_2^{\otimes l}\otimes \sem{b } \xrightarrow{{\mathrm{id}}\otimes\varepsilon } { \mathcal{m}}_2^{\otimes l}\otimes \sem{a{\mathbin{\multimap}}b}\otimes \sem{a } \xrightarrow{{\mathrm{id}}\otimes\gamma } { \mathcal{m}}_2^{\otimes l}\otimes\sem{a}\otimes\sem{a{\mathbin{\multimap}}b } \\ & \qquad\xrightarrow{{\mathrm{id}}\otimes\sem{n}\otimes{\mathrm{id } } } { \mathcal{m}}_2^{\otimes ( l+k)}\otimes\sem{a{\mathbin{\multimap}}b } \xrightarrow{{\mathrm{id}}\otimes\sem{m } } { \mathcal{m}}_2^{\otimes ( l+k ) } \otimes { \mathcal{m}}_2^{\otimes h } \xrightarrow{\braket{\psi}{-}{\psi } } { \mathbb{c}}\end{aligned}\ ] ] let @xmath378}{\to_{p_i } } { [ \,\psi_i,\psi_i , m_i\,]}$ ] ( @xmath331 ) be all the reductions from @xmath378}$ ] . by ih , we have @xmath387}:a{\mathbin{\multimap}}b}^{(l+k ) } \\ & = \sum_{i\in i } p_i\sem{{[\,\psi_i,\psi_i , m_i\,]}:a{\mathbin{\multimap}}b}^{(l+k ) } \\ & = \sum_{i\in i } p_i\bigl ( { \mathcal{m}}_2^{\otimes(l+k)}\otimes\sem{a{\mathbin{\multimap}}b } \xrightarrow{{\mathrm{id}}\otimes\sem{m_i } } { \mathcal{m}}_2^{\otimes(l+k ) } \otimes { \mathcal{m}}_2^{\otimes h_i } \xrightarrow{\braket{\psi_i}{-}{\psi_i } } { \mathbb{c}}\bigr)\end{aligned}\ ] ] it is then straightforward to see that @xmath388}:b}^{(l)}=\sum_i p_i\sem{{[\,\psi_i,\psi_i , m_in\,]}:b}^{(l)}$ ] . next consider the case where @xmath389 and @xmath390 . without loss of generality we may assume @xmath391}$ ] with @xmath392 . the only reduction from @xmath37 is @xmath393}{\to_{1 } } { [ \,\ket{\psi'},\psi,{\langlex_1,\dotsc , x_k\rangle}\ , ] } \eqqcolon q$ ] , where @xmath394 ( @xmath395 denotes the @xmath26 identity matrix ) . we need to show that @xmath396 . note that @xmath397 thus we have @xmath398 on the other hand , we have @xmath399 and hence @xmath400 for each elementary tensor @xmath401 , @xmath402 we conclude that @xmath396 . consider the case where @xmath375 is of the form @xmath403 . only the reduction is @xmath404 } { \to_{1}}{[\,\ket{\psi},\psi,{m[v / x]}\,]}$ ] . the assertion holds immediately by lemma [ lem : denot - beta - redu ] .

we investigate the equilibrium and nonequilibrium properties of the core region of vortices in layered superconductors . we discuss the electronic structure of singly and doubly quantized vortices for both s - wave and d - wave pairing symmetry . we consider the intermediate clean regime , where the vortex - core bound states are broadened into resonances with a width comparable to or larger than the quantized energy level spacing , and calculate the response of a vortex core to an _ a.c . _ electromagnetic field for vortices that are pinned to a metallic defect . we concentrate on the case where the vortex motion is nonstationary and can be treated by linear response theory . the response of the order parameter , impurity self energy , induced fields and currents are obtained by a self - consistent calculation of the distribution functions and the excitation spectrum . we then obtain the dynamical conductivity , spatially resolved in the region of the core , for external frequencies in the range , . we also calculate the dynamically induced charge distribution in the vicinity of the core . this charge density is related to the nonequilibrium response of the bound states and collective mode , and dominates the electromagnetic response of the vortex core . presented at the 2000 workshop on _ microscopic structure and dynamics of vortices in unconventional superconductors and superfluids _ , held at the _ max planck institute for the physics of complex systems _ in dresden .

cond-mat0106546

vortex motion is the principle mechanism for resistive losses in type ii superconductors . vortices also provide valuable information about the nature of low lying excitations in the superconducting state . in clean @xmath1-wave bcs superconductors the low - lying excitations in the core are the bound states of caroli , de gennes and matricon @xcite . these excitations have superconducting as well as normal metallic properties . for example , these states are the source of circulating supercurrents in the equilibrium vortex core , and they are strongly coupled to the condensate by andreev scattering @xcite . furthermore , the response of the vortex core states to an electromagnetic field is generally very different from that of normal electrons . however , in the dirty limit , @xmath2 , the the bardeen - stephen model @xcite of a normal - metal spectrum with the local drude conductivity in the core provides a reasonable description of the dissipative dynamics of the vortex core . the opposite extreme is the `` superclean limit '' , @xmath3 , in which the quantization of the vortex - core bound states must be taken into account . in this limit a single impurity and its interaction with the vortex core states must be considered . electromagnetic response is then controlled by selection rules governing transition matrix elements for the quantized core levels and the level structure of the core states in the presence of an impurity @xcite . in the case of d - wave superconductors in the superclean limit , _ nodes _ in the spectrum of bound states lead to a finite dissipation from landau damping for @xmath4 @xcite . the superclean limit is difficult to achieve even for short coherence length superconductors ; weak disorder broadens the vortex core levels into a quasi - continuum . we investigate the intermediate - clean regime , @xmath5 , where the discrete level structure of the vortex - core states is broadened and the selection rules are broken due to strong overlap between the bound state wave functions . however , the vortex core states remain well defined on the scale of the superconducting gap , @xmath6 . in this regime we can take advantage of the power of the quasiclassical theory of nonequilibrium superconductivity @xcite . the energy required to maintain a net charge density of order an elementary charge per particle within a coherence volume ( or coherence area in two dimensions ) is much larger than the condensation energy . thus , charge accumulation in the vortex core is strongly suppressed . in order to reduce the coulomb energy associated with the charge accumulation an internal electrochemical potential , @xmath7 , develops in response to an external electric field . this potential produces an internal electric field , @xmath8 , which is the same order of magnitude as the external field . even though the external field may vary on a scale that is large compared to the coherence length , @xmath9 , the internal field develops on the coherence length scale . the source of the internal field is a charge density that accumulates inhomogeneously over length scales of order the coherence length . it is necessary to calculate the induced potential self consistently from the spatially varying order parameter , spectral function and distribution function for the electronic states in the vicinity of the vortex core . an order of magnitude estimate shows that to produce an induced field of the order of the external field , the dynamically induced charge is of order @xmath10 , where @xmath11 is the typical energy scale set by the strength of the external field . this charge density accumulates predominantly in the vortex core region and creates a dipolar field around the vortex core . for a pinned vortex the charge accumulates near the interface separating the metallic inclusion from the superconductor . disorder plays a central role in the dissipative dynamics of the mixed state of type ii superconductors . impurities and defects are a source of scattering that limits the mean free path of carriers , thus increasing the resistivity . defects also provide ` pinning sites ' that inhibit vortex motion and suppresses the flux - flow resistivity . however , for _ a.c . _ fields even pinned vortices are sources for dissipation . the magnitude and frequency dependence of this dissipation depends on the electronic structure and dynamics of the core states of the pinned vortex . in the analysis presented below we consider vortices in the presence of pinning centers . we model a pinning center as a normal metallic inclusion which is coupled to the electronic states of the superconductor through a highly transmitting interface . in this model the charge dynamics of the electronic states near the interface between the pinning region and superconductor plays an important role in the electromagnetic response of the core . in the next section we provide a short summary of the nonequilibrium quasiclassical equations , including the transport equations for the quasiparticle distribution and spectral functions , constitutive equations for the order parameter , impurity self - energy and electromagnetic potentials . in section [ sec : electronic_structure ] we present calculations for the the electronic structure of vortices for superconductors with both @xmath1-wave and @xmath12-wave pairing symmetry . the results are based on self - consistent calculations of the order parameter and impurity self energy . for @xmath1-wave superconductors , impurity scattering leads to inhomogeneous broadening of the vortex core bound states , as well as bands of impurity states within the gap . in the case of @xmath12-wave pairing the core states are further broadened by coupling between bound states and the continuum states through impurity scattering . we also discuss the structure of doubly quantized vortices and vortices bound to mesoscopic size metallic inclusions . in the case of the doubly quantized vortex there are two branches of zero - energy bound states centered at finite impact parameter from the vortex center . this leads to a unique signature of a doubly quantized vortex : currents in the core circulate opposite to the supercurrents outside the core . section [ sec : nonequilibrium_response ] summarizes calculations of the vortex core dynamics for @xmath1-wave vortices in the presence of impurity scattering . we describe the charge dynamics of the vortex core for both pinned and unpinned vortices , and calculate the local a.c . conductivity that results from the coupled dynamics of the order parameter collective mode and the quasiparticle bound states in the vortex core . we discuss energy transport by the core states and the absorption features in the conductivity spectrum , which we interpret in terms of absorption within the bound - state band centered at the fermi level and resonant transitions involving the bound and continuum states . the quasiclassical theory describes equilibrium and nonequilibrium properties of superconductors on length scales that are large compared to microscopic scales ( i.e. the lattice constant , fermi wavelength , @xmath13 , thomas - fermi screening length , etc . ) and energies that are small compared to the atomic scales ( e.g. fermi energy , @xmath14 , plasma frequency , conduction band width , etc . ) . thus , there are small dimensionless parameters that define the limits of validity of the quasiclassical theory . in particular , we require @xmath15 , @xmath16 and @xmath17 , where the _ a.c . _ frequencies of interest are typically of order @xmath18 , or smaller , and the length scales of interest are of order the coherence length , @xmath19 , or longer . hereafter we use units in which @xmath20 , and adopt the sign convention @xmath21 for the electron charge . in quasiclassical theory quasiparticle wavepackets move along nearly straight , classical trajectories at the fermi velocity . the classical dynamics of the quasiparticle excitations is governed by semi - classical transport equations for their phase - space distribution function . the quantum mechanical degrees of freedom are the `` spin '' and `` particle - hole degree of freedom '' , described by @xmath22 density matrices ( nambu matrices ) . the quantum dynamics is coupled to the classical dynamics of the quasiparticles in phase space through the matrix structure of the quasiclassical transport equations . the nonequilibrium quasiclassical transport equations @xcite are formulated in terms of a quasiclassical nambu - keldysh propagator @xmath23 , which is a matrix in the combined nambu - keldysh space , and is a function of position @xmath24 , time @xmath25 , energy @xmath26 , and momenta @xmath27 on the fermi surface.,@xmath28,@xmath29 ) is related to the phase - space description in @xmath30 space by a transformation , @xmath31 , with @xmath32 and @xmath33 @xcite . ] we denote nambu - keldysh matrices by a `` check '' , and their 4@xmath344 nambu submatrices of _ advanced _ ( @xmath35 ) , _ retarded _ ( @xmath36 ) and keldysh - type ( @xmath37 ) propagators by a `` hat '' . the nambu - keldysh matrices for the quasiclassical propagator and self - energy have the form , [ nambu - keldysh ] =( ) , = ( ) , where @xmath38 are the retarded ( r ) , advanced ( a ) and keldysh ( k ) quasiclassical propagators , and similarly for the self - energy functions . each of these components of @xmath39 and @xmath40 are @xmath22 nambu matrices in combined particle - hole - spin space . for a review of the methods and an introduction to the notation we refer to refs . in the compact nambu - keldysh notation the transport equations and the normalization conditions read @xmath41_{\otimes } + i\vec{v}_f\cdot\grad \check g\\ = \ 0\enspace , \ ] ] @xmath42 where the commutator is @xmath43_\otimes=\check a\otimes\check b-\check b\otimes\check a$ ] , @xmath44 the vector potential , @xmath45 , includes @xmath46 which generates the static magnetic field , @xmath47 , as well as the non - stationary vector potential describing the time - varying electromagnetic field ; @xmath48 is the mean - field order parameter matrix , @xmath49 describes diagonal mean fields due to quasiparticle interactions ( landau interactions ) , and @xmath50 is the impurity self - energy . the electrochemical potential @xmath51 includes the field generated by the induced charge density , @xmath52 . the coupling of quasiparticles to the external potential involves virtual high - energy processes , which result from polarization of the non - quasiparticle background . the interaction of quasiparticles with both the external potential @xmath53 and the polarized background can be described by coupling to an effective potential @xmath54 @xcite . the high - energy renormalization factor @xmath55 is defined below in eq . ( [ renormalization ] ) . the coupling of the quasiparticle current to the vector potential in eq . ( [ transpqcl ] ) is given in terms of the quasiparticle fermi velocity . no additional renormalization is needed to account for the effective coupling of the charge current to the vector potential because the renormalization by the non - quasiparticle background is accounted for by the effective potentials that determine the band structure , and therefore the quasiparticle fermi velocity . equations ( [ transpqcl]-[normalize ] ) must be supplemented by maxwell s equations for the electromagnetic potentials , and by self - consistency equations for the order parameter and the impurity self - energy . we use the weak - coupling gap equation to describe the superconducting state , including unconventional pairing . the mean field self energies are then given by , @xmath56 @xmath57 @xmath58 the impurity self - energy , @xmath59 is specified by the impurity concentration , @xmath60 , and impurity scattering @xmath25-matrix , which is obtained from the the self - consistent solution of the @xmath25-matrix equations , [ tmatrix ] ( _ f,_f , ; , t ) & = & ( _ f,_f ) + & + & n_f . the nambu matrix @xmath61 is the off - diagonal part of @xmath62 , while @xmath63 is the diagonal part in particle - hole space . the fermi surface average is defined by @xmath64 where @xmath65 is the average density of states on the fermi surface . the other material parameters that enter the self - consistency equations are the dimensionless pairing interaction , @xmath66 , the dimensionless landau interaction , @xmath67 , the impurity concentration , @xmath60 , the impurity potential , @xmath68 , and the fermi surface data : @xmath27 ( fermi surface ) , @xmath69 ( fermi velocity ) . we eliminate both the magnitude of the pairing interaction and the cut - off , @xmath70 , in favor of the transition temperature , @xmath71 , using the linearized , equilibrium form of the mean - field gap equation ( eq . ( [ gapequation ] ) ) . the quasiclassical equations are supplemented by constitutive equations for the charge density , the current density and the induced electromagnetic potentials . the formal result for the non - equilibrium charge density , to linear order in @xmath72 , is given in terms of the keldysh propagator by @xmath73\big > -2e^2n_fz_0\phi(\vec{r};t ) \,,\ ] ] with the renormalization factors given by [ renormalization ] z(_f)=1-a(_f,_f ) , z_0= z(_f ) . the high - energy renormalization factor is related to an average of the scattering amplitude on the fermi surface by a ward identity that follows from the conservation law for charge @xcite . the charge current induced by @xmath74 , calculated to leading order in @xmath75 , is also obtained from the keldysh propagator , @xmath76 there is no additional high - energy renormalization of the coupling to the vector potential because the quasiparticle fermi velocity already includes the high - energy renormalization of the charge - current coupling in eq . [ current1 ] . furthermore , the self - consistent solution of the quasiclassical equations for @xmath77 ensures the continuity equation for charge conservation , @xmath78 is satisfied . an estimate of the contribution to the charge density from the integral in eq . ( [ density ] ) leads to the condition of `` local charge neutrality '' @xcite . a charge density given by the elementary charge times the number of states within an energy interval @xmath79 around the fermi surface implies @xmath80 . such a charge density can not be maintained within a coherence volume because of the cost in coulomb energy . the coulomb energy is suppressed by requiring the leading order contribution to the charge density vanish : i.e. @xmath81 . thus , the spatially varying renormalized electro - chemical potential , @xmath54 , is determined by [ localneutral ] 2ez_0(;t)= _ -_c^+_cd4i . the continuity equation implies @xmath82 . we discuss violations of the charge neutrality condition ( [ localneutral ] ) , which are higher order in @xmath72 , in sec . [ chargeresp ] . finally , ampere s equation , with the current given by eq . ( [ current1 ] ) , determines the vector potential in the quasiclassical approximation , @xmath83 equations ( [ transpqcl])-([tmatrix ] ) and ( [ localneutral])-([vecpot ] ) constitute a complete set of equations for calculating the electromagnetic response of vortices in the quasiclassical limit . for high-@xmath84 superconductors we can simplify the self - consistency calculations to some degree . since quasiparticles couple to the vector potential via @xmath85 , eq . ( [ vecpot ] ) shows that this quantity is of order @xmath86 , where @xmath87 is the magnetic penetration depth . thus , for @xmath88 , as in the layered cuprates , the feedback effect of the current density on the vector potential is small by factor @xmath89 . for sufficiently weak fields we can calculate the electromagnetic response to linear order in the external field . the propagator and the self - energies are separated into unperturbed equilibrium parts and terms that are first - order in the perturbation , @xmath90 and similarly for the electromagnetic potentials , @xmath91 , @xmath92 . the equilibrium propagators obey the matrix transport equation , @xmath93 + i\vec{v}_f\cdot\grad \check g_0\\ = \ 0 \,.\ ] ] these equations are supplemented by the self - consistency equations for the mean fields , eqs . ( [ gapequation])-([nuequation ] ) , the impurity self energy , eqs . ( [ born])-([tmatrix ] ) , the local charge - neutrality condition for the scalar potential , eq . ( [ localneutral ] ) , ampre s equation for the vector potential , eq . ( [ vecpot ] ) , the equilibrium normalization conditions , @xmath94 and the equilibrium relation between the keldysh function and equilibrium spectral density , @xmath95 \,.\ ] ] the first - order correction to the matrix propagator obeys the linearized transport equation , @xmath96_{\otimes } + i\vec{v}_f\cdot\grad \delta\check g\\ = \ \bigl [ \delta\check{\delta}_{mf}+\delta\check{\sigma}_{i}+\delta\check{v}\,,\ \check{g}_0]_{\otimes}\enspace , \ ] ] with source terms on the right - hand side from both the external field ( @xmath97 ) and the internal fields ( @xmath98 , @xmath99 ) . in addition , the first - order propagator satisfies the `` orthogonality condition '' , @xmath100 obtained from linearizing the full normalization condition . : @xmath101 , @xmath102 . ] the system of linear equations are supplemented by the equilibrium and first - order self - consistency conditions for the order parameter , @xmath103 @xmath104 and the impurity self - energy , [ respborn0 ] _ 0 ( _ f , ;) & = & n_i_0(_f,_f , ; ) , + [ tmatrix0 ] _ 0 ( _ f,_f,;)&= & ( _ f,_f)+ n_f , + [ tmatrix1 ] _ i ( _ f , ; , t)&= & n_i n_f . in general the diagonal mean fields also contribute to the response . however , we do not expect landau interactions to lead to qualitatively new phenomena for the vortex dynamics , so we have neglected these interactions in the following analysis and set @xmath105 ( i.e. @xmath106 ) . as a result the local charge neutrality condition for the electro - chemical potential becomes , [ resplocalneutral ] 2e(;t)=_-_c^+_cd4i . in what follows we work in a gauge in which the induced electric field , @xmath107 , is obtained from @xmath108 and the uniform external electric field , @xmath109 , is determined by the vector potential @xmath110 . for @xmath111 we can safely neglect corrections to the vector potential due to the induced current . thus , in the nambu - keldysh matrix notation the electromagnetic coupling to the quasiparticles is given by @xmath112 the validity of linear response theory requires the external perturbation @xmath113 be sufficiently small and that the induced vortex motion responds to the external field at the frequency set by the external field . at very low frequencies frictional damping of the vortex motion , arising from the finite mean free path of quasiparticles scattering from impurities , gives rise to a nonlinear regime in the dynamical response of a vortex . this regime is discussed extensively in the literature @xcite , and is not subject of our study . however , for sufficiently small field strengths the vortex motion is nonstationary over any time interval , although it may be regarded as quasi - stationary at low enough frequencies . the nonstationary motion of the vortex can be described by linear response theory if @xmath114 for @xmath115 , and @xmath116 for @xmath117 . note that the frequency of the perturbation , @xmath118 , is not required to be small compared to the gap frequency ; it is only restricted to be small compared to atomic scale frequencies , e.g @xmath119 . self - consistent solutions of eqs . ( [ lgapequation1 ] ) , ( [ tmatrix1 ] ) and ( [ resplocalneutral ] ) for the self - energies and scalar potential are fundamental to obtaining a physically sensible solution for the electromagnetic response . the dynamical self - energy corrections are equivalent to `` vertex corrections '' in the kubo formulation of linear response theory . they are particularly important in the context of nonequilibrium phenomena in inhomogeneous superconductors . in our case these corrections are of vital importance ; the self - consistency conditions enforce charge conservation . in particular , eqs . ( [ respborn0])-([tmatrix1 ] ) imply charge conservation in scattering processes , whereas ( [ lgapequation0 ] ) and ( [ lgapequation1 ] ) imply charge conservation in particle - hole conversion processes ; any charge which is lost ( gained ) in a particle - hole conversion process is compensated by a corresponding gain ( loss ) of condensate charge . it is the coupled quasiparticle and condensate dynamics which conserves charge in superconductors . neglecting the dynamics of either component , or using a non - conserving approximation for the coupling leads to unphysical results . self - consistent calculations for the equilibrium order parameter , impurity self - energy and local excitation spectrum ( spectral density ) are necessary inputs to the linearized transport equations for the dynamical response of a vortex . the equilibrium spectral function also provides key information for the interpretation of the dynamical response . because of particle - hole coherence the spectral density is sensitive to the phase winding and symmetry of the order parameter , as well as material properties such as the transport mean - free path and impurity cross - section . in the following section we present results for the low - energy excitation spectra of singly- and doubly - quantized vortices in layered superconductors with @xmath1-wave and @xmath12-wave pairing symmetry the local density of states for excitations with fermi momentum @xmath27 is obtained from the retarded and advanced quasiclassical propagators , n(_f,;)= n_f . this function measures the local density of the quasiparticle states with energy @xmath26 at the point @xmath27 on the fermi surface . the local density of states ( ldos ) is obtained by averaging this quantity over all momentum directions of the quasiparticles , n(;)= . the product of the angle - resolved density of states and the fermi velocity , @xmath120 , determines the current density carried by these states ; @xmath120 also defines the direction of a quasiclassical trajectory passing through the space point @xmath24 . we introduce the angle - resolved _ spectral current density _ @xcite , ( _ f,;)= 2e_f(_f)n(_f , ; ) , which measures the current density carried by quasiparticle states with energy @xmath26 moving along the trajectory defined by @xmath120 . the local spectral current density is then defined as ( ; ) = , and the total current density is obtained by summing over the occupied states for each trajectory , ( ) = df ( ) ( ; ) , where @xmath121 . self - consistent calculations of the equilibrium structure and spectral properties of vortices are relatively straight - forward computations . below we present results for @xmath1-wave and @xmath12-wave pairing symmetry with impurity scattering included . the calculations reported are carried for a circular fermi surface , with an isotropic fermi momentum @xmath27 and fermi velocity @xmath120 . the elastic scattering rate is chosen to represent the intermediate - clean regime , @xmath122 . the pairing potential can be represented as a sum over invariant products of basis functions @xmath123 for the irreducible representations of the crystal point group labeled by @xmath124 , n_f v(_f,_f)= _ , i v_^_i^(_f)_i(_f ) . the pairing interaction , @xmath125 , and the cutoff , @xmath70 , are eliminated in favor of the instability temperature for pairing in symmetry channel @xmath124 . we limit the discussion here to even - parity , one - dimensional representations , which for tetragonal symmetry includes the ` @xmath1-wave ' ( identity ) representation , a@xmath126 , two ` @xmath12-wave ' representations , b@xmath126 and b@xmath127 , and a ` @xmath128-wave ' representation , a@xmath127 . the corresponding basis functions we use are listed in table [ table : basis_functions ] . .symmetry classes and model basis functions for the 1d even - parity representations of @xmath129 . the angle @xmath130 is the angular position of @xmath28 on the fermi surface with respect to the crystallographic @xmath131-axis ( = @xmath132-axis ) . [ cols="^,^,^",options="header " , ] [ table : basis_functions ] the results for the order parameter , impurity self energy and spectral properties of vortices that follow are calculated self consistently in the @xmath25-matrix approximation for point impurities ( pure @xmath1-wave scattering ) , i.e. @xmath133 . the quasiparticle scattering rate , @xmath134 , and normalized impurity cross section , @xmath135 , are then given by , @xmath136 the born limit corresponds to @xmath137 , while unitary scattering corresponds to @xmath138 or @xmath139 . in the calculations that follow the temperature is set at @xmath140 , and the mean free path is is chosen to represent the intermediate - clean regime ; @xmath141 . for isotropic @xmath1-wave pairing the equilibrium order parameter for an isolated vortex with winding number @xmath142 has the form , ( _ f , ) = ( ) e^ip , where the amplitude @xmath143 is isotropic and @xmath144 is the azimuthal angle of @xmath24 in the plane . angle - resolved density of states for an @xmath1-wave vortex for quasiparticles propagating along trajectories parallel to the @xmath132-direction , at points along the @xmath145-axis spaced by @xmath146 . the left panel is for born scattering , and the right panel is for unitary scattering with the same mean free path of @xmath147 . the temperature is @xmath140 . , title="fig : " ] angle - resolved density of states for an @xmath1-wave vortex for quasiparticles propagating along trajectories parallel to the @xmath132-direction , at points along the @xmath145-axis spaced by @xmath146 . the left panel is for born scattering , and the right panel is for unitary scattering with the same mean free path of @xmath147 . the temperature is @xmath140 . , title="fig : " ] the angle - resolved local density of states spectra for a singly quantized vortex is shown in fig . [ spec1a ] for space points @xmath148 along the @xmath145-axis with a spacing of @xmath149 , and for trajectories parallel to the x - axis , @xmath150 . the vortex center is at @xmath151 , and the phase winding is such that the direction of the superflow is in @xmath132-direction for spectra with negative @xmath145 coordinate . far outside the core the angle - resolved density of states resembles the bcs density of states with a gap in the spectrum roughly between @xmath152 , and peaks in the spectrum near the continuum edges . careful inspection of fig . [ spec1a ] shows that the coherence peak for positive energy at @xmath153 is not at @xmath154 , but is shifted to higher energy by the doppler effect , @xmath155 , where @xmath156 is the condensate momentum @xcite . in a homogeneous superflow field the spectrum is the doppler - shifted bcs spectrum ; the doppler shift increases with condensate momentum until pairbreaking sets in at the bulk critical momentum , @xmath157 . however , in the vortex core _ nonlocal effects _ associated with the inhomogeneous flow field lead to a redistribution of the spectral weight near the gap edge . the positive energy continuum edge is broadened considerably compared to the square - root singularity in the absence of the doppler effect . the continuum starts at @xmath158 even for the doppler - shifted spectra near the maximal current regions . in contrast , the negative energy continuum edge shows sharp structures due to the accumulation of spectral weight in the region between @xmath159 and @xmath160 . the sharp structure corresponds to a bound state that is separated from the continuum edge . the density of states at the continuum edge drops precipitously at @xmath161 . we emphasize that nonlocal effects lead to qualitative differences in the spectrum near the gap edges compared to the widely used approximation of doppler - shifted quasiparticles in a locally homogeneous superflow @xcite . in clean superconductors nonlocal effects dominate the spectrum . for a homogeneous superflow the current is carried mainly by the states that are doppler shifted the region between @xmath159 and @xmath79 . the spectral current density shows that contributions to the current density from states outside of this region nearly cancel . in the case of a vortex , the bound state that splits off from the continuum not only robs the doppler - shifted continuum edge of its spectral weight , but the bound state also carries most of the supercurrent in the vortex core region . at distances approaching the vortex center the bound state is clearly resolved and disperses through @xmath162 at zero impact parameter . as shown in the left panel of fig . [ spec1a ] the bound state also broadens considerably for born scattering as it disperses towards the fermi level , but remains a sharp resonance in the limit of unitary scattering , as shown in fig . [ spec1a ] . local density of states ( ldos ) in the center of an @xmath1-wave vortex for different effective impurity scattering cross sections @xmath135 from born to unitary limit , with a fixed mean free path @xmath163 , and temperature @xmath140 . from top to bottom : @xmath164 . ] the coherence peaks are completely suppressed at the vortex center , both in the born and unitary limits . note the difference in the evolution of the spectral weight for the bound and continuum states : for positive energies the spectral weight of the coherence peak is shifted to higher energies as one approaches the vortex core , with a continuously diminishing intensity at the coherence peak . in contrast , the spectral weight of the negative energy coherence peak is transferred to the bound state which splits off from the continuum . in principle there could be additional , secondary bound states , which would split off from the continuum if the vortex core were wider . however , a self - consistent calculation of the order parameter suppresses the secondary bound state in the vortex core . we return to the discussion of secondary bound states when we discuss the spectrum of pinned vortices . for trajectories with @xmath165 the structure of the spectrum for positive and negative energies is reversed because the superflow is now counter to the quasiparticle velocity , leading to negative doppler shifts and bound states at positive energies below the continuum edge . the small spectral features that appear at the energies corresponding to the _ negative _ of the bound state energies are due to mixing of with trajectories of opposite velocity by backscattering from impurities . finally , consider the differences in the spectral features for unitary versus born scattering . in fig . [ unitary ] we show the local density of states at the center of the vortex for several scattering cross sections ranging from the born limit ( @xmath137 ) to the unitary limit ( @xmath166 ) . in the born limit the broadening is a maximum . for small , but finite cross section two bands of impurity bound states split off from the zero - energy resonance and remove spectral weight from the central peak . the impurity bands evolve towards the continuum edges as the cross - section increases and merge with them in the unitary limit . when the impurity bound - state bands no longer overlap the central peak the zero - energy resonance sharpens dramatically with the width remaining constant as the unitary limit is approached . the overlap between the impurity bands and the continuum edges in the unitary limit is determined by the scattering rate , and is increasing with increasing scattering rate , @xmath167 . as shown in the right panel of fig . [ spec1a ] , the impurity bands are localized in the vortex core region ; their existence depends on impurity scattering in a region where the phase changes rapidly over length scales of order the coherence length . for @xmath12-wave pairing symmetry the order parameter has the form , ( _ f , ) = _ ( _ f ) ( ) , where @xmath168 changes sign on the fermi surface at the points , @xmath169 . these nodal points lead to strong anisotropy and gapless excitations in the quasiparticle spectrum , which feeds back to produce anisotropy in the spatial structure of the order parameter in the core region of of a vortex . the spatial part of the order parameter , ( ) = |()|e^i ( ) , for a vortex , at distances far from the core , approaches that of an isotropic vortex : @xmath170 and @xmath171 . however , in the core region the current density is comparable to the critical current density and develops a four - fold anisotropy as a result of the backflow current concentrated near the nodes @xcite . this current - induced pairbreaking effect is dominant for flow parallel to the nodal directions and leads to weak anisotropy of the order parameter in the core region . the electronic structure of the @xmath12-wave vortex , presented in fig . [ spec1d ] , shows distinct differences from that of an @xmath1-wave vortex . the resonances in the vortex core are broader than the core states of vortex in an @xmath1-wave superconductor . mixing with extended states in nodal direction broadens the peaks near the continuum edges . also note the effect of impurity scattering on the spectra far from the vortex core region , which show a broadened continuum rather than a sharp continuum edge . in the right panel of fig . [ spec1d ] we show the angle - resolved density of states for a @xmath12-wave vortex for unitary impurity scattering . as in the @xmath1-wave case there is sharpening of the zero resonance , but note that the width of the resonance is much broader than in the case of @xmath1-wave pairing because impurity scattering provides a coupling and mixing of the bound state resonance with the low - energy extended states for momenta near the nodal directions , even for a trajectory in the antinodal direction . angle resolved density of states for a @xmath12-wave vortex for a quasiparticle trajectory along the @xmath132-direction ( antinodal ) as a function of impact parameter ( @xmath145-direction ) . the impact parameter spacing is @xmath172 . the left panel corresponds to impurity scattering in the born limit , and the right panel is for the unitary limit . the temperature is @xmath140 . , title="fig : " ] angle resolved density of states for a @xmath12-wave vortex for a quasiparticle trajectory along the @xmath132-direction ( antinodal ) as a function of impact parameter ( @xmath145-direction ) . the impact parameter spacing is @xmath172 . the left panel corresponds to impurity scattering in the born limit , and the right panel is for the unitary limit . the temperature is @xmath140 . , title="fig : " ] in fig . [ nofepss ] we show the local density of states ( ldos ) for both @xmath1-wave ( top two panels ) and @xmath12-wave pairing ( bottom two panels ) as a function of position @xmath24 along an anti - nodal direction ( bottom left ) and along a nodal direction ( bottom right ) . the ldos is obtained by averaging the angle - resolved density of states over the fermi surface at a particular space point @xmath29 . the averaging , together with the dispersion of the bound states and resonances as a function of angle leads to one - dimensional bands with characteristic van hove singularities , which are clearly visible for @xmath1-wave symmetry , but considerably smeared and broadened for @xmath12-wave symmetry . the ldos for a vortex with @xmath1-wave pairing symmetry is shown for impurity scattering in both the born ( top left ) and unitary limits ( top right ) . the spectra show the characteristic bound state bands , van hove singularities and the dramatic reduction in the width of the zero bias resonance for unitary scattering . local density of states for an @xmath1-wave vortex ( top panels ) and a @xmath12-wave vortex ( bottom panels ) , as a function of distance from the vortex center with a spacing @xmath173 . for @xmath1-wave pairing the born limit is shown on the left and unitary scattering is on the right . for @xmath12-wave pairing the ldos is shown only for born scattering , but along axes parallel to a anti - nodal direction ( left ) and parallel to the nodal ( right ) . the temperature is @xmath140 . , title="fig : " ] local density of states for an @xmath1-wave vortex ( top panels ) and a @xmath12-wave vortex ( bottom panels ) , as a function of distance from the vortex center with a spacing @xmath173 . for @xmath1-wave pairing the born limit is shown on the left and unitary scattering is on the right . for @xmath12-wave pairing the ldos is shown only for born scattering , but along axes parallel to a anti - nodal direction ( left ) and parallel to the nodal ( right ) . the temperature is @xmath140 . , title="fig : " ] + local density of states for an @xmath1-wave vortex ( top panels ) and a @xmath12-wave vortex ( bottom panels ) , as a function of distance from the vortex center with a spacing @xmath173 . for @xmath1-wave pairing the born limit is shown on the left and unitary scattering is on the right . for @xmath12-wave pairing the ldos is shown only for born scattering , but along axes parallel to a anti - nodal direction ( left ) and parallel to the nodal ( right ) . the temperature is @xmath140 . , title="fig : " ] local density of states for an @xmath1-wave vortex ( top panels ) and a @xmath12-wave vortex ( bottom panels ) , as a function of distance from the vortex center with a spacing @xmath173 . for @xmath1-wave pairing the born limit is shown on the left and unitary scattering is on the right . for @xmath12-wave pairing the ldos is shown only for born scattering , but along axes parallel to a anti - nodal direction ( left ) and parallel to the nodal ( right ) . the temperature is @xmath140 . , title="fig : " ] calculations of the ldos in the superclean limit for a vortex with @xmath12-wave pairing are discussed by schopohl and maki @xcite and by ichioka _ the self - consistent calculations shown in fig . [ nofepss ] include impurity scattering in the born limit for space points , @xmath24 , along two different directions ; the left panel corresponds the ldos measured as a function of distance along the anti - nodal direction and the right panel is the ldos measured along the nodal direction . the nodes of the order parameter for @xmath12-wave pairing lead to continuum states with energies down to the fermi level . these states are visible in the ldos as the smooth background extending to zero energy from both positive and negative energies , even for distances far from the core . in the vortex core region several broad peaks disperse as a function of distance from the vortex center . these peaks correspond to broadened van hove singularities resulting from averaging the vortex core resonances over the fermi surface for at a fixed position @xmath24 . the differences in the @xmath12-wave spectra for the two directions reflects the weak fourfold anisotropy of the ldos around the vortex at fixed energy . the calculations discussed above describe the average effects of atomic scale impurity disorder on the spectral properties of a vortex . a specific defect can also act as a pinning site for a vortex . we model such a defect as a mesoscopic size , normal metallic inclusion in the superconductor . the defect is assumed to be a circular inclusion , with a radius , @xmath174 , of order the coherence length of the superconductor . for simplicity we assume that the metallic properties of the inclusion , e.g. fermi surface parameters , are the same as those of the normal - state of the host superconductor . the calculations presented below neglect normal reflection processes at the interface between the inclusion and the host metal , but include andreev reflection . the analysis and calculations can be generalized for more detailed models of a pinning center . angle resolved density of states for a quasiparticle trajectory along the @xmath132-direction , as a function of impact parameter ( @xmath145-axis ) with a spacing of @xmath175 for a vortex centered on a pinning center of radius of @xmath176 . the left panel is for @xmath1-wave symmetry and right panel is for @xmath12-wave symmetry . impurity scattering is included for born scattering with @xmath147 . the temperature is @xmath140 . , title="fig : " ] angle resolved density of states for a quasiparticle trajectory along the @xmath132-direction , as a function of impact parameter ( @xmath145-axis ) with a spacing of @xmath175 for a vortex centered on a pinning center of radius of @xmath176 . the left panel is for @xmath1-wave symmetry and right panel is for @xmath12-wave symmetry . impurity scattering is included for born scattering with @xmath147 . the temperature is @xmath140 . , title="fig : " ] figures [ ados_pin_s ] and [ dos_pin_s ] show the angle - resolved local density of states and the ldos for a vortex pinned on a metallic inclusion of radius @xmath176 . the order parameter , impurity self energy and spectral densities were calculated self consistently for impurity scattering in the born limit . qualitative changes resulting from the inclusion occur inside the pinning center . the shape of the bound state resonance lines are asymmetric in energy . the asymmetry arises from multiple andreev reflection processes by the interface between the pinning center and the superconductor , which leads to additional bands of resonances that overlap the vortex core resonances . in addition to the asymmetry in the linewidth of the resonances the zero - energy bound state at the vortex center has a peculiar spectral shape , shown in more detail in fig . [ pincompare ] for @xmath1-wave pairing symmetry , but also visible in the right panel of fig . [ dos_pin_s ] for @xmath12-wave pairing symmetry as well . in contrast to the spectra for a vortex without a pinning center , the coherence peaks at the continuum edges are present at the vortex center . local density of states as a function of distance from the vortex center , with a spacing of @xmath173 , for a vortex pinned to an inclusion of radius of @xmath177 . the left panel corresponds to @xmath1-wave symmetry and the right panel is for @xmath12-wave symmetry along the anti - nodal direction . the temperature is @xmath140 . , title="fig : " ] local density of states as a function of distance from the vortex center , with a spacing of @xmath173 , for a vortex pinned to an inclusion of radius of @xmath177 . the left panel corresponds to @xmath1-wave symmetry and the right panel is for @xmath12-wave symmetry along the anti - nodal direction . the temperature is @xmath140 . , title="fig : " ] figure [ pincompare ] shows the peaks near the continuum edges in the vortex center for a pinned vortex . the spectral weight near the continuum edge is taken from the resonance at the fermi level . the enhanced weight at the continuum edges is a precursor to the formation of a secondary bound state that splits off from the continuum . this can be seen in the evolution of the spectral density at the vortex center as a function of the radius of the pinning center . there is a zero energy resonance for all pinning radii , however , increasing the radius of the pinning center transfers spectral weight from the zero - energy resonance to the continuum edge . a coherence peak develops , splits off from the continuum edge , strengthens and evolves to energies within the gap as the pinning radius changes from @xmath178 to @xmath179 . thus , the appearance of the coherence peak for pinning centers the size of a coherence length or so is a precursor to the formation of a secondary bound state within the gap . the spectral weight comes at the expense of states just above the continuum edge and the zero - energy bound state , which is diminished in intensity with increasing pinning radius . local density of states at the center of a vortex pinned in the center of a normal inclusion . the spectra ( from top to bottom ) correspond to different pinning radii : @xmath180 , @xmath181 , @xmath182 , @xmath183 , @xmath184 , @xmath185 , and @xmath186 . the temperature is @xmath140 . ] vortices with winding numbers larger than @xmath187 generally have line energies per unit winding number that are larger than that of a singly quantized vortex . nevertheless , doubly quantized vortices once formed are metastable and in principle it should be possible to observe the rare doubly quantized vortex using an atomic probe such as a scanning tunneling microscopic @xcite . the spectrum of a doubly quantized vortex differs in a fundamental way from that of a singly quantized vortex . the singly quantized vortex , has a single branch of states that disperse through zero energy at the vortex center . the zero mode is guaranteed in the quasiclassical limit by @xmath188 change along trajectories that pass through the vortex center . in contrast , there is no phase change along a trajectory through the vortex center for a doubly quantized vortex , and thus no topological requirement enforcing a zero energy bound state at the center of a doubly quantized vortex . nevertheless , there is a spectrum of bound states in the cores of doubly quantized vortices which lead to characteristic structures in the ldos and current spectral density of a doubly quantized vortex . dispersion of the bound states ( below the continuum edge ) and of the coherence peaks ( above the continuum edge ) for quasiparticle trajectories in @xmath132 direction as a function of @xmath145 , for a singly quantized vortex ( left ) and for a doubly quantized vortex ( right ) . the center of the vortex is at @xmath151 . born impurity scattering with a mean free path of @xmath189 is assumed . the temperature is @xmath140 . the thin lines show the energies of the continuum edges and the doppler shifted energies of the coherence peaks , assuming the london form for the condensate momentum @xmath190 , where @xmath142 is the winding number . the dispersion is shown for s - wave pairing ; the corresponding data for d - wave pairing is similar . , title="fig : " ] dispersion of the bound states ( below the continuum edge ) and of the coherence peaks ( above the continuum edge ) for quasiparticle trajectories in @xmath132 direction as a function of @xmath145 , for a singly quantized vortex ( left ) and for a doubly quantized vortex ( right ) . the center of the vortex is at @xmath151 . born impurity scattering with a mean free path of @xmath189 is assumed . the temperature is @xmath140 . the thin lines show the energies of the continuum edges and the doppler shifted energies of the coherence peaks , assuming the london form for the condensate momentum @xmath190 , where @xmath142 is the winding number . the dispersion is shown for s - wave pairing ; the corresponding data for d - wave pairing is similar . , title="fig : " ] this structure was discussed for a doubly quantized vortex in the superclean limit in ref . figure [ double_ados ] shows the angle - resolved density of states for a doubly quantized vortex in an @xmath1-wave superconductor for trajectories parallel to the @xmath132-axis at different impact parameters along the @xmath145-direction . _ two _ branches of vortex bound states cross the fermi level at distances of order a coherence length from the vortex center . thus , zero - energy bound states exist in the core but they are localized ( for @xmath1-wave pairing symmetry ) on trajectories at finite impact parameter from the vortex center . the locus of these trajectories forms a circle of radius @xmath191 around the vortex center . also note that the doppler shift of the continuum spectrum is twice that for singly quantized vortices . angle resolved density of states for a doubly quantized vortex along quasiparticle trajectories parallel to the @xmath132-direction , as a function of impact parameter , @xmath145 , with a spacing @xmath192 . the left panel corresponds to @xmath1-wave symmetry and the right panel is for @xmath12-wave symmetry . impurity scattering is included self - consistently in the born limit with a mean free path @xmath193 . the temperature is @xmath140 . , title="fig : " ] angle resolved density of states for a doubly quantized vortex along quasiparticle trajectories parallel to the @xmath132-direction , as a function of impact parameter , @xmath145 , with a spacing @xmath192 . the left panel corresponds to @xmath1-wave symmetry and the right panel is for @xmath12-wave symmetry . impurity scattering is included self - consistently in the born limit with a mean free path @xmath193 . the temperature is @xmath140 . , title="fig : " ] this can be seen by comparing the spectra far from the core in figs . [ double_ados ] and [ spec1a ] . the distinctive features in the spectrum of a doubly quantized vortex are present for either @xmath1- and @xmath12-wave pairing symmetry . the doubling of bound state branches that cross the fermi level is observed for both pairing symmetries . the main difference in the spectra , as in the case of singly quantized vortices , is in the width of the resonances . the most distinguishing feature of doubly quantized vortices is that the supercurrents near the center of the vortex flow _ counter _ to the asymptotic superflow associated with the phase winding around the vortex @xcite . the counter circulating currents in the core , shown in the right panel of fig . [ current_reversal ] , are due to the bound states interior to the radius defined by the zero - energy bound state ( note that the zero - energy bound state itself does not carry current ) . this structure is revealed in figs . [ double_dos ] and [ current_reversal ] , which show the ldos and spectral current density for a doubly quantized vortex with @xmath1-wave and @xmath12-wave pairing symmetry . the left panel of fig . [ current_reversal ] also shows the cumulative spectral current density and the reversal of the current for as one branch of bound states disperses below the fermi level for @xmath194 . local density of states for doubly quantized vortices along trajectories parallel to the @xmath132-axis as a function of impact parameter along the @xmath145-axis with a spacing of @xmath195 . the left panel corresponds to @xmath1-wave pairing symmetry , and the right panel is for @xmath12-wave pairing along the anti - nodal direction . impurity scattering is included in the born limit with a mean free path of @xmath193 and the temperature is @xmath140 . , title="fig : " ] local density of states for doubly quantized vortices along trajectories parallel to the @xmath132-axis as a function of impact parameter along the @xmath145-axis with a spacing of @xmath195 . the left panel corresponds to @xmath1-wave pairing symmetry , and the right panel is for @xmath12-wave pairing along the anti - nodal direction . impurity scattering is included in the born limit with a mean free path of @xmath193 and the temperature is @xmath140 . , title="fig : " ] these spectra show that the current density near the vortex core is carried mainly by the bound states , and that the reversal of the current direction near the vortex center is due to the branch of counter - flowing bound states dispersing below the fermi level for impact parameter @xmath196 . in this region of the core the co - moving bound state is above the fermi level so the asymmetry in the level occupation produces a counter flowing current . the spectral current density for a doubly quantized vortex is shown along a trajectory parallel to the @xmath132-axis as a function of impact parameter ( @xmath145-axis ) with a spacing @xmath195 , starting at @xmath197 . the thick curve is the cumulative spectral current density as a function of @xmath26 , obtained by integrating @xmath198 from @xmath199 to @xmath26 . the right panel is the x - component of the current density as a function of impact parameter for @xmath1-wave ( solid line ) and d - wave ( dashed line ) pairing . , title="fig : " ] the spectral current density for a doubly quantized vortex is shown along a trajectory parallel to the @xmath132-axis as a function of impact parameter ( @xmath145-axis ) with a spacing @xmath195 , starting at @xmath197 . the thick curve is the cumulative spectral current density as a function of @xmath26 , obtained by integrating @xmath198 from @xmath199 to @xmath26 . the right panel is the x - component of the current density as a function of impact parameter for @xmath1-wave ( solid line ) and d - wave ( dashed line ) pairing . , title="fig : " ] the dynamics of the electronic excitations of the vortex core play a key role in the dissipative processes in type ii superconductors . except in the dirty limit , @xmath200 , the response of the core states to an electromagnetic field is generally very different from that of normal electrons . it is energetically unfavorable to maintain a charge density of the order of an elementary charge over a region with diameter of order the coherence length . instead an electrochemical potential is induced which ensures that almost no net charge accumulates in the core region . however , a dipolar - like charge distribution develops which generates an internal electric field in the core . the internal field varies on the scale of the coherence length . this leads to a nonlocal response of the quasiparticles to the total electric field , even when the applied field varies on a much longer length scale and can considered to be homogeneous . the dynamical response of the vortex core includes the collective mode of the inhomogeneous order parameter . this mode couples to the electro - chemical potential , @xmath201 , in the vortex core region . this potential is generated by the charge dynamics of vortex core states and gives rise to internal electric fields which in turn drive the current density and the order parameter near the vortex core region . the induced electric fields in the core are the same order of magnitude as the external field . the dynamics of the core states are strongly coupled to the charge current and collective mode of the order parameter . thus , the determination of the induced order parameter , as well as the spectrum and distribution function for the core states and non - equilibrium impurity scattering processes requires dynamical self consistency . numerous calculations of the a.c . response neglect the self - consistent coupling of the collective mode and the spectral dynamics , or concentrate on the @xmath202 limit @xcite . quasiclassical theory is the only formulation of the theory of nonequilibrium superconductivity presently capable of describing the nonlocal response of the order parameter and quasiparticle dynamics in the presence of mesoscopic inhomogeneities and disorder . the numerical solution to the self - consistency problem was presented for unpinned vortices in ref . here we report results for the the electromagnetic response of isolated vortices bound to a pinning center in a superconductor with @xmath1-wave pairing symmetry . the charge dynamics of layered superconductors has two distinct origins . the c - axis dynamics is determined by the josephson coupling between the conducting planes . here we are concerned with the in - plane electrodynamics associated with the response of the order parameter and quasiparticle states bound to the vortex core . we assume strong josephson coupling between different layers , and neglect variations of the response between different layers . this requires that the polarization of the electric field be in - plane , so that there is no coupling of the in - plane dynamics to the josephson plasma modes . the external electromagnetic field is assumed to be long wavelength compared to the size of the vortex core , @xmath203 . in this limit we can assume the a.c . electric field to be uniform and described by a vector potential , @xmath204 . we can also neglect the response to the a.c . magnetic field in the limit @xmath205 . in this case the spatial variation of the induced electric field occurs mainly within each conducting layer on the scale of the coherence length , @xmath9 . poisson s equation implies that induced charge densities are of order @xmath206 , where @xmath201 is the induced electrochemical potential in the core . this leads to a dynamical charge of order @xmath207 in the vortex core . once the electrochemical potential is calculated from eq . ( [ resplocalneutral ] ) we can calculate the charge density fluctuations of order @xmath208 from poisson s equation , ^(3)(;t)=-^2(;t ) . left : total electric field in the center of a pinned @xmath1-wave vortex as a function of frequency @xmath209 of an external a.c . electric field with polarization vector in the @xmath132-direction and wavelength large compared to @xmath9 . right : the corresponding in phase charge response around the pinning site for frequency @xmath210 . gray corresponds to negative charge and black to positive charge . the arrows denote the total electric field vectors . the pinning center is a circular normal metallic inclusion with a radius @xmath211 , shown as the black circle . impurity scattering is included in the born limit with a mean free path of @xmath212 . the temperature is @xmath140 and the calculations are carried out in the high-@xmath84 limit . , title="fig : " ] left : total electric field in the center of a pinned @xmath1-wave vortex as a function of frequency @xmath209 of an external a.c . electric field with polarization vector in the @xmath132-direction and wavelength large compared to @xmath9 . right : the corresponding in phase charge response around the pinning site for frequency @xmath210 . gray corresponds to negative charge and black to positive charge . the arrows denote the total electric field vectors . the pinning center is a circular normal metallic inclusion with a radius @xmath211 , shown as the black circle . impurity scattering is included in the born limit with a mean free path of @xmath212 . the temperature is @xmath140 and the calculations are carried out in the high-@xmath84 limit . , title="fig : " ] in fig . [ feld ] we show the total electric field ( external plus induced ) acting on the quasiparticles in units of the external field . for @xmath213 the total field is approximately equal to the external field . however , at frequencies @xmath214 an out - of - phase response develops . for @xmath215 , the total field in the pinning region approaches twice the external field , and the out - of - phase component vanishes . in the intermediate frequency region , @xmath216 , both in - phase and out - of - phase components are comparable . the right panel of fig . [ feld ] shows the charge distribution for @xmath217 which oscillates out of phase with the external field . a dipolar charge distribution accumulates at the interface between the superconductor and the normal inclusion , oscillating at the frequency of the external field . at the center of the pinning site the out - of - phase component of the field is nearly zero at low frequencies ( see also the left panel ) . the induced charge which accumulates is of order of @xmath218 within a region of order @xmath219 in each conducting layer . this charge is a factor of @xmath220 larger than the static charge of a vortex that arises from particle - hole asymmetry @xcite . because the total electric field varies on the scale of a coherence length , the current response expressed in terms of the the total field is nonlocal in the intermediate clean regime . however , we can define a local conductivity tensor in terms of the response to the external field , provided the external field varies on a length scale large compared to the coherence length , j _ ( , ) = _ ( , ) e^ _ ( ) . imaginary ( left ) and real part ( right ) of the local conductivity ( multiplied with @xmath209 for convenience ) in the center of an unpinned @xmath1-wave vortex ( circles ) , and at different distances from the center along the @xmath145-axis : @xmath221 ( squares ) , and @xmath222 ( triangles ) , as a function of frequency @xmath209 of an external a.c . electric field with polarization vector in @xmath132-direction and wavelength large compared to @xmath9 . impurity scattering is taken into account in born limit with a mean free path of @xmath212 . the temperature is @xmath140 . calculations are done in the high-@xmath223 limit . also shown are the response for a homogeneous @xmath1-wave superconductor ( dotted , denoted ` bcs ' ) , and the drude conductivity of the normal metal ( dashed ) . , title="fig : " ] imaginary ( left ) and real part ( right ) of the local conductivity ( multiplied with @xmath209 for convenience ) in the center of an unpinned @xmath1-wave vortex ( circles ) , and at different distances from the center along the @xmath145-axis : @xmath221 ( squares ) , and @xmath222 ( triangles ) , as a function of frequency @xmath209 of an external a.c . electric field with polarization vector in @xmath132-direction and wavelength large compared to @xmath9 . impurity scattering is taken into account in born limit with a mean free path of @xmath212 . the temperature is @xmath140 . calculations are done in the high-@xmath223 limit . also shown are the response for a homogeneous @xmath1-wave superconductor ( dotted , denoted ` bcs ' ) , and the drude conductivity of the normal metal ( dashed ) . , title="fig : " ] figures [ cond_unpin ] and [ cond_pin ] show results for the conductivity , @xmath224 , in the vortex core region as a function of frequency for both unpinned and pinned vortices . for the pinned vortex the radius of the pinning center is @xmath225 . first consider the unpinned vortex . the absorptive part of the conductivity ( right panel of fig . [ cond_unpin ] ) is strongly enhanced at the vortex center compared to the normal - state drude conductivity . the reactive response exhibits a maximum at a frequency determined by the impurity scattering rate . a few coherence lengths away from the vortex center the real part of the conductivity changes sign at low frequencies . this is a signature that energy is transported by vortex - core excitations away from the vortex center producing `` hot spots '' outside the core . the net dissipation is determined by inelastic scattering processes in the region around the vortex core . at distances of order a coherence length or so from the vortex center there is also structure in the conductivity spectrum at higher frequencies reflecting absorptive transitions between quasiparticle excitations with energies corresponding to the van hove singularities in the local density of states . the maxima in the absorptive part of the conductivity at @xmath226 and @xmath227 from the center correspond to the energy level separation between the van hove peaks above and below the fermi level shown in fig . [ nofepss ] . like fig . [ cond_unpin ] for a pinned vortex with a pinning radius @xmath211 . as pinning site a circular normal conducting inclusion located at the center of the vortex was assumed ( see fig . [ feld ] ) . , title="fig : " ] like fig . [ cond_unpin ] for a pinned vortex with a pinning radius @xmath211 . as pinning site a circular normal conducting inclusion located at the center of the vortex was assumed ( see fig . [ feld ] ) . , title="fig : " ] the conductivity spectra for a vortex pinned by a metallic inclusion at its center is shown in fig . [ cond_pin ] . the most significant difference compared to the unpinned vortex occurs at frequencies @xmath228 . the absorptive part of the conductivity ( right panel ) is reduced compared to that of the unpinned case at low frequencies . the three broad peaks in the absorption spectrum correspond to scattering and dissipation within the zero - energy resonance ( the dominant low - frequency peak ) , transitions between the zero - energy resonance and the continuum ( the peak near @xmath229 ) , and pair - breaking transitions from the negative energy to positive energy continuum states ( broad peak above @xmath230 ) . the other notable feature is the reactive response at low frequencies which becomes _ negative _ in the low frequency limit , corresponding to superflow in the core that is counter to the induced supercurrent outside the vortex core and pinning center . this counterflow is required in order to satisfy the conductivity sum rule . the counterflow in the vortex center is also present for unpinned vortices , but at smaller frequencies for this particular impurity scattering rate . the low - frequency counterflow is also related to characteristic current patterns associated with low - frequency vortex dynamics which we discuss below . _ _ current density pattern for an pinned @xmath1-wave vortex with a pinning radius @xmath211 . as pinning site a circular normal conducting inclusion located at the center of the vortex was assumed . the border of the pinning center ( gray ) is shown as black circle . the frequency of the external electric field with polarization vector in @xmath132-direction is chosen @xmath210 . left picture for out of phase ( reactive ) response , right picture for in phase ( absorptive ) response . impurity scattering is taken into account in born limit with a mean free path of @xmath212 . the temperature is @xmath140 . calculations are done in the high-@xmath223 limit . , title="fig : " ] _ a.c . _ current density pattern for an pinned @xmath1-wave vortex with a pinning radius @xmath211 . as pinning site a circular normal conducting inclusion located at the center of the vortex was assumed . the border of the pinning center ( gray ) is shown as black circle . the frequency of the external electric field with polarization vector in @xmath132-direction is chosen @xmath210 . left picture for out of phase ( reactive ) response , right picture for in phase ( absorptive ) response . impurity scattering is taken into account in born limit with a mean free path of @xmath212 . the temperature is @xmath140 . calculations are done in the high-@xmath223 limit . , title="fig : " ] results for the _ a.c . _ component of the current density near a pinned vortex are shown in fig . [ curr ] for @xmath217 . in addition to the a.c . current there is the time - independent circulating supercurrent around the vortex center which adds to the current shown in fig . the current response shows a dipolar pattern , which is also observed for unpinned vortices . the in - phase current response ( right panel ) indicates a region of strong absorption within the pinning region ( @xmath231 ) , and emission ( @xmath232 ) in the region roughly perpendicular to the direction of the applied field several coherence lengths away from the pinning center . calculations of the energy transport current show that energy absorbed in the core is transported away from the vortex center by the vortex core excitations in directions predominantly perpendicular to the applied field . the net absorption is ultimately determined by inelastic scattering and requires integrating the local absorption and emission rate over the vortex array . note that the long - range dipolar component does not contribute to the total dissipation . far from the vortex core the current response is out of phase with the electric field and predominantly a non - dissipative supercurrent . also note that at low frequency we clearly observe the counterflowing supercurrent within the pinning center . the electrodynamics of the vortex state in the intermediate - clean regime is nonlocal and largely determined by the response of the vortex - core states . transitions involving the vortex - core states , and their coupling to the collective motion of the condensate requires dynamically self - consistent calculations of the order parameter , self energies , induced fields , excitation spectra and distribution functions . the results of these calculations provide new insight into the dynamics of vortex cores in conventional and unconventional superconductors .

motivated by recent work indicating that the uv continuum in agn may be produced by reradiation of energy absorbed from x - rays irradiating an accretion disk , we present a calculation of the vertical structures and ultraviolet spectra of x - ray irradiated accretion disks around massive non - rotating black holes . after finding the radial dependence of vertically - integrated quantities for these disks , we solve the equations of hydrostatic equilibrium , energy balance , and frequency - dependent radiation transfer as functions of altitude . to solve the last set of equations , we use a variable eddington factor method . we include electron scattering , free - free , and hi , hei , and heii bound - free opacities and the corresponding continuum cooling processes . the incident x - ray flux heats a thin layer of material 3 - 4 scale heights above the midplane of the disk . this x - ray heated skin has two layers : a radiation pressure supported region in which the uv flux is created , and , immediately above this layer , a warmer zone , optically thin to uv radiation , formed where the x - ray ionization parameter is large . in the lower layer the gas pressure is nearly independent of altitude but the temperature increases upward . the fraction of the incident hard x - ray flux which emerges in the uv falls with increasing ( the accretion rate in eddington units ) . at frequencies below the lyman edge the slope of the continuum ( ) varies from -1.6 to 0.8 as increases from 0.001 to 1 . here is the mass of the central black hole in units of . in all cases examined ( and ) , the lyman edge appears in emission . the amplitude of the lyman edge feature increases with but is relatively independent of . the amplitude of the lyman edge emission feature increases with disk inclination . compton scattering in disk coronae can smooth the lyman edge feature only if , where is the thomson depth of the coronae . while the overall spectral shape predicted by x - ray irradiation may be compatible with observations , the lyman edge emission feature it predicts is not . this finding raises questions for many otherwise plausible models in which x - ray irradiation plays a major role . # 1#2#3#4#4 19#3 phys . rev . d , # 1 , # 2 # 1#2#3#4#4 19#3 phys . lett . , # 1 , # 2 # 1#2#3#4#4 19#3 phys . rev . lett . , # 1 , # 2 # 1#2#3#4#4 19#3 phys . rev . , # 1 , # 2 # 1#2#3#4#4 19#3 phys . rep . , # 1 , # 2 # 1#2#3#4#4 19#3 phys . fluids , # 1 , # 2 # 1#2#3#4#4 19#3 proc . phys . soc . , # 1 , # 2 # 1#2#3#4#4 19#3 nucl . phys . , # 1 , # 2 # 1#2#3#4#4 19#3 mod . phys . lett . , # 1 , # 2 # 1#2#3#4#4 19#3 ap . j. , # 1 , # 2 # 1#2#3#4#4 19#3 astr . j. , # 1 , # 2 # 1#2#3#4#4 19#3 acta astr . , # 1 , # 2 # 1#2#3#4#4 19#3 rev . mod . phys . , # 1 , # 2 # 1#2#3#4#4 19#3 nuovo cimento c , # 1 , # 2 # 1#2#3#4#4 19#3 sov . phys . jetp , # 1 , # 2 # 1#2#3#4#4 19#3 sov . ast . aj , # 1 , # 2 # 1#2#3#4#4 19#3 pub . ast . soc . japan , # 1 , # 2 # 1#2#3#4#4 19#3 pub . ast . soc . pacific , # 1 , # 2 # 1#2#3#4#4 19#3 ann . phys . ( ny ) , # 1 , # 2 # 1#2#3#4#4 19#3 yad . fiz . , # 1 , # 2 # 1#2#3#4#4 19#3 sov . j. nucl . phys . , # 1 , # 2 # 1#2#3#4#4 19#3 ast . ap . , # 1 , # 2 # 1#2#3#4#4 19#3 ann . rev . astr . ap . , # 1 , # 2 # 1#2#3#4#4 19#3 m.n.r.a.s . , # 1 , # 2 # 1#2#3#4#4 19#3 j. de physique , # 1,#2 # 1#2#3#4#4 19#3 j. quant . spec . rad . transfer , # 1,#2 # 1#2#3#4#4 19#3 j.e.t.p . lett . , # 1,#2 # 1#2#3#4#4 19#3 ap . j. ( letters ) . , # 1,#2 # 1#2#3#4#4 19#3 ap . j. ( supp . ) . , # 1,#2 # 1#2#3#4#4 19#3 ap . lett . , # 1,#2 # 1#2#3#4#4 19#3 ap . sp . sci . , # 1,#2 # 1#2#3#4#4 19#3 nature , # 1,#2 # 1#2#3#4#4 19#3 sp . sci . rev . , # 1,#2 # 1#2#3#4#4 19#3 adv . sp . res . , # 1,#2

astro-ph9609094

the optical and uv emission of radio - quiet active galactic nuclei ( agns ) is dominated by a quasi - thermal component , the big blue bump " ( shields 1978 , malkan & sargent 1982 , malkan 1983 ) . the big blue bump is usually interpreted as thermal emission from a geometrically thin , optically thick , accretion disk around a massive black hole ( malkan 1983 , sun & malkan 1989 , laor & netzer 1989 : ln89 ) . this model makes two key predictions . first , the large jump in the absorptive opacity at the lyman edge combined with the vertical temperature gradients in the disk insure that some feature will be present at the lyman edge ( kolykhalov & sunyaev 1984 , ln89 ) . for the same reason , some have predicted that the polarization of the disk should decrease at frequencies above the lyman edge ( laor , netzer & piran 1990 ; but see blaes & agol 1996 for another view ) . second , variations in different portions of the optical / ultraviolet band can be associated with fluctuations in the conditions at particular radii . if fluctuations move radially through the disk , there should be time - delays between variations seen at different wavelengths which correspond to signal travel times within the disk . unfortunately , while this picture is very attractive , it suffers from severe problems , both theoretical and observational . from the theoretical point of view , it is generally expected that the viscous stress is proportional to the pressure ( the @xmath9-model " introduced by shakura & sunyaev 1973 : ss73 ) , including the radiation pressure when the disk is optically thick . unfortunately , if this is true , when the accretion rate is more than a small fraction of the eddington rate , the disk is thermally unstable throughout its inner region ( shakura & sunyaev 1976 ) . for this reason , the most detailed disk spectral calculations made to date ( ln89 ) supposed that the stress was instead proportional to the geometric mean of the gas and radiation pressures . in addition , neither of the two key predictions just described is confirmed by observations . partial lyman edges have been detected in very few agn ( antonucci , kinney & ford 1989 ; koratkar , kinney & bohlin 1992 ) and the polarization of two quasars with partial absorption edges _ rises _ across the lyman edge ( koratkar , 1995 ) . intensive monitoring of the type 1 seyfert galaxy ngc 5548 revealed that variations from the optical to the far - uv are tightly correlated . the signal travel speed through a conventional disk that would be required to coordinate these variations was found to be at least @xmath10 , far higher than any of the expected signal speeds ( krolik 1991 ) . moreover , variations in the soft x - ray flux have been found to be almost as tightly correlated with ultraviolet fluctuations ( clavel , 1992 ) . these two findings suggest that the big blue bump may actually be due to reprocessed x - ray emission ( rokaki , collin - souffrin , & magnan 1993 ) . other evidence also points to the importance of x - ray reprocessing in the formation of agn continua . in many type 1 seyfert galaxies , there is a spectral feature in the vicinity of 10 30 kev which is readily interpreted as compton reflection " of x - rays from a cool , optically thick surface occupying roughly half of the solid angle around the x - ray source ( pounds 1990 ) . it would be very natural to interpret this reflector as an accretion disk ( lightman & white 1988 ) ( but a number of other interpretations are also possible : nandra & george 1994 ; krolik , madau , & @xmath11 1994 ; ghisellini , haardt & matt 1994 ) . further support for the thought that the disk is subjected to intense x - ray irradiation comes from the discovery that the fe k@xmath9 profiles in many agn are extremely broad ( as summarized by nandra 1997 ) . recently popular models for the production of the x - rays are also consistent with this picture in which a large fraction of the emitted x - rays strike a nearby cool surface , possibly the accretion disk ( haardt & maraschi 1993 : hm93 ; haardt , maraschi & ghisellini 1994 ; pietrini & krolik 1995 ; stern 1995 ) . these models suppose that a large part of the disk s dissipation takes place in a small amount of mass at or above the disk s surface . the very large heating rate per unit mass in an optically thin gas creates a very hot corona which cools by inverse compton scattering . roughly half the x - ray flux generated strikes the disk , and is reprocessed into uv photons which then return to the corona as seeds for inverse compton scattering . fits to the broad - band x - ray spectra of agn are at least qualitatively consistent with this sort of thermal comptonization model ( e.g. zdziarski 1995 ) . in its simplest form , the corona model predicts that the observed uv flux should be roughly equal to the observed intrinsic x - ray flux ( i.e. the x - ray flux after subtracting any components due to reprocessing ) . whether this prediction is consistent with the observations is uncertain . the uv / x - ray flux ratio is conventionally parameterized by @xmath12 , the spectral index of a power - law interpolated between 2500 and 2 kev , as measured in the agn rest - frame . depending on the sample definition , the observed distribution of @xmath12 varies somewhat , but typically it ranges from @xmath13 to @xmath14 , with a tendency for lower luminosity agn to have smaller @xmath12 ( stocke et al . 1991 ; wilkes et al . 1994 ; green et al . 1995 ) . the flux at at 2500 is @xmath15 times the flux at 2 kev so that , if we compare at these two frequencies alone , we would conclude that the uv luminosity exceeds the x - ray luminosity in most agn . however , to find the ratio of the total uv and x - ray fluxes , substantial and very uncertain bolometric corrections must be applied to both . we can not observe low - redshift agn in the rest - frame euv , but composite high - redshift quasar spectra ( e.g. zheng et al . 1996 ) suggest that the spectrum rolls over sharply at wavelengths shorter than @xmath16 . if this shape is general , the uv bolometric correction ( relative to @xmath17 at 2500 ) might be as small as @xmath18 , but there is likely to be a considerable dispersion from case to case . the x - ray bolometric correction is similarly uncertain . those low luminosity agn for which osse has been able to obtain spectra typically show power - laws with indices @xmath19 across the hard x - ray range , supplemented by a substantial bump at a few tens of kev which is generally believed to be due to reprocessing ( zdziarski et al . if that power - law extends from 0.5 to 200 kev , the bolometric correction relative to @xmath17 at 2 kev is 6.5 . taking these bolometric corrections at face value , we would conclude that the uv and x - ray fluxes in low luminosity agn ( where the evidence for the corona model is best ) are indeed close to being equal , while the uv flux is typically rather greater than the intrinsic x - ray flux in higher luminosity agn . however , it is clear from the character of this estimate that it is very weakly based . the situation is made still cloudier by the possibility that relativistic effects may direct the majority of the emitted x - rays toward the disk , reducing the observed ratio between x - rays and uv ( martocchia and matt 1996 ) . it is our goal to begin the process of testing these coronal models by comparing the ultraviolet spectrum they predict with observations . in this paper we develop the techniques necessary to compute the predicted uv spectra , and apply them to the simplest version of the model : that in which _ all _ the dissipation takes place in a slab - like corona resting directly on top of the accretion disk . in future work we will examine other variations , such as models in which the dissipation is shared between the disk proper and the corona , and models with more complicated coronal geometry . an immediate consequence of this picture in which most of the dissipation takes place in a corona , and not inside the disk proper , is that radiation pressure support in the bulk of the disk is far weaker than in a conventional disk . the radiation force is simply the product of the radiation flux and the opacity ; where there is no outgoing flux ( because there is no internal heat generation ) , there can be no radiation force . as a result , such disks collapse to a state of much greater density ( and also column density ) than would be predicted by conventional models ( svensson & zdziarski 1995 : sz95 ) . they are then supported primarily by gas pressure , and ( in the @xmath9-model ) are thermally stable . it is also an immediate corollary that these disks are very nearly isothermal zero outward flux in an optically thick environment implies constant temperature . some analogous work has been done in the past . a number of groups ( e.g. ross & fabian 1993 ; @xmath20 1994 ) have computed the spectra of x - ray irradiated disks , but all have assumed the disks are radiation pressure supported . because disks in which most dissipation is in a corona are much denser than conventional radiation pressure - supported disks , the character of the uv spectrum they produce is quite different from what has been predicted by assuming a conventional disk structure . others have tried to predict the uv spectra radiated by dense lumps heated by x - rays ( guilbert & rees 1988 , ferland & rees 1988 ) . in [ sec : isothermal disk ] we solve the structure equations for x - ray irradiated disks . the solution of the radiative transfer and vertical structure equations in the x - ray heated skin is described in [ sec : xray skin ] and our results are summarized in [ sec : results ] the gravitational stability of the cold disk is discussed in [ sec : gravitational instability ] we conclude in [ sec : conclusions ] when there is no internal dissipation ( sz95 treat the case in which the dissipation is shared between the disk and the corona ) , the main body of the disk is rigorously isothermal , as explained in 1 . the radial structure of geometrically thin isothermal accretion disks is defined by two equations : the conservation of angular momentum @xmath21 and the inter - ring stress prescription , for which we use the @xmath9-model : @xmath22 vertical structure is determined by hydrostatic equilibrium : @xmath23 and the isothermal equation of state . the symbols in these equations represent the mass of the central black hole ( @xmath24 ) , the mass accretion rate ( @xmath25 ) , the keplerian angular velocity ( @xmath26 ) , the vertically integrated shear stress ( @xmath27 ) , the radial coordinate ( @xmath28 ) , the vertical coordinate ( @xmath29 ) , the gas density ( @xmath30 ) , the gas pressure ( @xmath31 ) , the sound speed in the gas ( @xmath32 ) , the surface mass density @xmath33 and the dimensionless stress @xmath34 . the general relativistic correction factors are taken from the standard references . the subscript nt refers to the novikov and thorne ( 1973 , page & thorne 1974 ) expressions and the subscript rh refers to the recent corrections presented by rifferts & herrold ( 1995 ) . these equations are easily solved . the integrated surface mass density is @xmath35 and the density profile is gaussian @xmath36 with a scale height of @xmath37 where @xmath38 is the gas temperature in the accretion disk . the normalization of the density , @xmath39 is set by assuming that all the mass lies in the isothermal inner disk . it is instructive to express the disk solution in terms of nondimensional parameters @xmath40 where we have introduced the eddington accretion rate @xmath41 and the schwarzschild radius @xmath42 the electron scattering opacity is @xmath43 . the numerical factors have been calculated by assuming an accretion efficiency of @xmath44 , which is appropriate for a non - rotating black hole . using these scalings we find that @xmath45 @xmath46 and @xmath47 there are two important things to note about these results . first , the total surface mass density is many orders of magnitude larger than the penetration depth of the hard x - rays . consequently , supposing the bulk of the disk to be isothermal and gas pressure - supported is an excellent approximation . second , the gas density of the isothermal disk is very much larger than the @xmath48 gm @xmath49 typical of conventional radiation pressure supported disks ( ss73 ) . few x - rays can penetrate much greater than the compton depth @xmath50 gm @xmath51 . on the other hand , as shown by equation 12 , these disks are extremely optically thick to compton scattering . consequently , there are essentially no x - rays below @xmath52^{1/2 } \simeq 3 $ ] the exact number of scale heights above the midplane where the x - ray - heated skin begins depends only slightly on parameters . in this estimate , and all the numerical work , we set @xmath54 for definiteness , but it hardly affects the results at all . inside the x - ray - heated skin , the temperature , pressure , and radiation intensity are determined as functions of altitude by solving the differential equations for radiative transfer , @xmath55 hydrostatic equilibrium @xmath56 and the surface mass density @xmath57 subject to the local constraints of radiative equilibrium @xmath58 and charge conservation @xmath59 equations [ eq : radiative transfer ] , [ eq : hydrostatic equilibrium ] , [ eq : surface mass density ] , [ eq : radiative equilibrium ] , [ eq : charge conservation ] are solved for the mean ultraviolet intensity ( @xmath60 ) , gas pressure ( @xmath31 ) , gas temperature ( @xmath61 ) , electron density ( @xmath62 ) and vertical coordinate ( @xmath29 ) as functions of the surface mass density . the other symbols represent the x - ray heating rate ( @xmath63 , see [ subsec : heating rate ] ) , total opacity ( @xmath64 ) , the absorptive opacity ( @xmath65 ) , the emissivity ( @xmath66 ) , the variable eddington factors ( @xmath67 ) , the number density of ion species @xmath68 in ionization state @xmath69 ( @xmath70 ) and the maximum ionization state of species @xmath68 ( @xmath71 ) . we assume that the disk is geometrically thin and neglect self - gravity of the disk . under these assumptions , the gravitational acceleration of the material in the disk is ( rifferts & herrold 1995 ) @xmath72 the gravitational force of the underlying isothermal disk can contribute to the net acceleration of material in the skin but the contribution is small ( see the discussion in [ sec : isothermal disk ] ) . the momentum of the absorbed x - rays is transferred to the gas , resulting in an additional force directed towards the disk midplane . the gradient of the hard x - ray radiation pressure is @xmath73 in the eddington approximation , which we have used to solve for @xmath63 ( [ subsec : heating rate ] ) . the total opacity includes bremsstrahlung , photoionization and electron scattering . we neglect all line opacities . the expression for the free - free opacity is taken from rybicki & lightman ( 1979 ) and the general expression for the photo - ionization opacity is from mihalas ( 1978 ) . the bound - free oscillator strengths of hi and heii are calculated using the analytic formulae for hydrogen - like atoms ( mihalas 1978 ) and the hei oscillator strengths are from bell & kingston ( 1967 ) . we calculate the number densities of electrons , hi , hii , hei , heii and heiii assuming saha equilibrium , local thermodynamic equilibrium , and hydrogen and helium abundances of x=0.9 and y=0.1 , respectively . explicit photoionization and recombination are not included in the calculation of the ionization state . we compute the emissivity using the corresponding continuum cooling processes : bremsstrahlung and hi , hei , and heii radiative recombination . the gas density in the x - ray heated layer is high enough that we expect resonance line cooling to be suppressed . resonance lines are thermalized when the collisional deexcitation rate is greater than the product of the radiation rate and the photon escape probability . neglecting stimulated emission , this occurs when the fractional abundance of the lower state @xmath74 satisfies the inequality @xmath75 where @xmath76 is the electron density in units of @xmath77 @xmath49 , @xmath78 is the column density of h atoms in units of @xmath79 @xmath51 , and @xmath80 the masses of the atom and proton , respectively . these scalings were chosen because the gas density at 3.5 scale heights from the midplane is typically @xmath81 @xmath49 and 2 kev photons are absorbed at a column density @xmath82 @xmath51 . when the temperature is @xmath83 k uv and euv lines are the important potential coolants , but the only lines which contribute to the cooling are those with very low abundance lower states . by contrast , the density in the atmosphere of an irradiated radiation pressure supported disk is characteristically four orders of magnitude lower . this means that the ionization parameter is so much greater that a thermal runaway is triggered , and soft x - ray lines are the most important transitions for cooling . due to the combination of lower density and higher radiation rates , emission lines are much more important in the thermal balance in this context than in our limit of the zero internal dissipation disk . in principle , the complete angular dependent radiative transport problem can be solved by introducing variable eddington factors ( mihalas 1978 ) . the computational overhead associated with an iterative calculation of the eddington factors is fairly large and , given the other approximations we have made , probably unwarranted . instead , we adopt a simple parametrization @xmath84 where @xmath85 which has the correct limiting behavior ( g. s. miller , private communication ) . we perform an iterative calculation of @xmath67 for one set of disk parameters in [ subsubsec : angular distribution ] and use this solution to test the accuracy of the analytic prescription . the complete system of equations is solved by transforming the differential equations into finite - difference equations , linearizing with respect to all variables , and relaxing from an initial solution . we solve the structure equations for a grey atmosphere with a shooting procedure ( press , 1992 ) and use this solution as a first guess for the relaxation method . the only significant difference between our work and the standard stellar atmosphere calculation ( mihalas 1978 , rybicki 1971 ) is the non - zero heating rate ( see eq . [ eq : radiative equilibrium ] ) . we assume that all the energy dissipated due to accretion at radius @xmath28 is released as heat in the corona . the rate per unit area is @xmath86 and this is converted into an x - ray flux by inverse compton scattering of the uv photons emerging from the disk ( hm93 ) . one - half of this flux illuminates the underlying disk . in terms of the dimensionless parameters defined in eq . [ eq : disk parameter scalings ] @xmath87 the spectrum of the illuminating flux is taken to be @xmath88 where @xmath89 . a self - consistent calculation of the x - ray spectrum is beyond the scope of this paper , so we simply assume @xmath90 , as is commonly observed ( zdziarski 1995 ) . the exponential cut - off is meant to mimic the sharp rise in the x - ray albedo above 10 kev ( lightman & white 1988 ) . as a result of the high albedo for harder photons , they are mostly reflected and do not contribute to the heating rate . still harder photons ( @xmath91 kev ) lose much of their energy in recoil , but because there are many uncertainties in our knowledge of the high energy spectrum in agn ( zdziarski 1995 ) , and because the expected contribution to the net heating is small and depends sensitively on the high energy spectrum ( p. @xmath92 , private communication ) , we neglect this contribution . we also suppose that there is no sub - structure to the corona ; the local heating rate could vary substantially if the dissipation is concentrated in active regions ( haardt , maraschi & ghisellini 1994 ) . these complications will be addressed in later work . given this simple parametrization , the heating rate is a well - defined function of the surface mass density @xmath93 where the effective x - ray absorption opacity is defined as ( rybicki & lightman 1979 ) @xmath94 ^{1/2}.\ ] ] and @xmath43 is the electron scattering opacity . the x - ray opacity in this energy range is dominated by photoionization of oxygen and iron ( london , mccray & auer 1981 ) . the threshold energy for k - shell photoionization depends upon the ionization state of the gas ( daltabuit & cox 1972 ) which is largely determined by the incident x - ray flux ( lightman & white 1988 , @xmath92 , 1994 ) . a careful calculation of the ionization balance of the heavy elements is beyond the scope of this paper . instead , we assume threshold energies of 0.8 kev and 8.0 kev for oxygen and iron , respectively , and solar abundances of these elements . the x - ray opacity can then be estimated as @xmath95 using absorption cross - sections for hydrogen - like atoms ( rybicki & lightman 1979 ) . this approximate description slightly underestimates the absorptive opacity in most of the x - ray - heated atmosphere , where most medium - z atoms retain some valence shell electrons . the expression for @xmath63 ( eq . [ eq : local heating rate ] ) is equivalent to a two - stream solution for the radiative transfer equation in the eddington approximation . this approximation is reasonable because the incident hard x - rays are nearly isotropic in the lower hemisphere and electron scattering in the disk will tend to isotropize the x - rays . the heating rate at lower @xmath96 is probably underestimated in this approximation because x - rays incident at large angles will be absorbed at smaller @xmath96 . the local heating rate is calculated by integrating equation [ eq : local heating rate ] numerically and the results for the assumed x - ray spectrum are plotted in fig [ fig : xray heating rate ] . the overall scale of the heating rate depends upon the accretion parameters , but the @xmath96 dependence is the same for all the models . at low column densities , the heating rate is almost independent of depth because all x - ray frequencies are optically thin . when @xmath97gm @xmath98 gm @xmath51 , @xmath99 because photons of energy @xmath100 are absorbed at the depth @xmath101 . at @xmath102 gm @xmath51 , the x - ray flux is exponentially attenuated and the heating rate falls rapidly . the analytic approximation @xmath103 is also plotted in fig . [ fig : xray heating rate ] . here @xmath96 is measured in units of gm @xmath51 . the boundary between the x - ray heated skin and the isothermal disk is set at @xmath104 . the x - ray heating rate at the boundary is negligible ( @xmath105 ) and we assume no dissipation in the cold disk , so @xmath106 at the inner boundary . the gas temperature is taken to be continuous across the boundary and the inner boundary conditions on the gas pressure and vertical coordinate are fixed by requiring that the numerical solution for the skin match the analytic solution for the isothermal disk at @xmath107 . the x - ray heated skin has a sharp outer boundary because a thermal runaway occurs at that point ( see 4.1 for a discussion of its physical origin ) . we designate the column density at this runaway @xmath108 . its location is found by bisection . we guess @xmath108 and attempt a solution . if the code converges to a thermally stable solution , @xmath108 is decreased ; if the code fails to converge , it is increased . this is repeated until the fractional change in @xmath108 is less than a few percent . we tested a number of methods for determining the location of the thermal runaway and found that the most sensitive test was the convergence of the code . the outer boundary condition on the mean uv intensity is usually set by assuming that there is no incident intensity at the boundary ( mihalas 1978 ) . however , in our situation this is not a valid approximation because @xmath108 ( or the compton depth of the corona ) may be large enough to reflect a significant amount of radiation back towards the skin . instead , we assume that the uv flux remains constant at @xmath109 and that @xmath110 . we neglect any contribution from the hot corona to the thomson depth at @xmath109 . we also neglect ( except in the global sense that this is the origin of the x - rays ) any frequency change in the uv photons caused by scattering off gas at @xmath109 . under these assumptions , there is an analytic relation between the mean intensity and the flux ( mihalas 1978 ) which provides our boundary condition @xmath111 where @xmath112 @xmath113 and @xmath114 are exponential integrals . in practice , the final value of @xmath115 and eq . [ eq : surface eddington factor ] reduces to the free streaming value , @xmath116 . however , @xmath117 for many of the intermediate steps . the inner edge of the x - ray heated skin is 3 - 4 scale heights from the midplane of the isothermal disk . the thickness of the skin increases slightly with radius but remains @xmath118 of the height of the inner edge ( fig . [ fig : structure]a ) . the thin disk approximation is valid at all radii and the geometry of the disk can be determined from the isothermal disk solution ( [ sec : isothermal disk ] ) . the density profile in the isothermal disk is gaussian so the gas pressure at the inner edge of the skin is much smaller ( typically by @xmath119 four orders of magnitude ) than the central disk pressure . this pressure is still much larger than the pressure in a fully radiation pressure - supported disk , however . throughout most of the skin , the radiation force of the uv photons is great enough to balance gravity and the force of the incident x - ray flux . the gas pressure is therefore nearly constant for @xmath120 ( see fig . [ fig : structure]b ) . both the x - ray and uv fluxes drop to zero at @xmath121 and a gas pressure gradient forms to support the disk against gravity . the gas pressure profile joins smoothly to the isothermal disk solution at @xmath107 . the radiative cooling rate is equal to the x - ray heating rate when the gas is in radiative equilibrium . therefore , the gas temperature must be greater than the uv radiation temperature because the cooling rate is @xmath122 , where @xmath123 and @xmath124 is the frequency integrated mean intensity of the uv radiation field . the actual gas temperature profile is determined by three competing effects . first , the mean intensity , and @xmath125 , decrease from the bottom of the skin to the surface because of the net outward flux . second , the x - ray heating rate increases from zero at the bottom of the skin to its maximum value at the top . therefore , @xmath126 at @xmath107 and the difference between the two temperatures increases steadily until @xmath127 at the upper edge of the skin . third , the gas temperature gradient changes when an opacity jump , such as the lyman edge , becomes transparent ( mihalas 1978 , p. 206 ) . the gradient can even change sign in some cases , but not in all cases ( see fig . [ fig : structure ] ) . the relative importance of these three effects depends upon the local gas pressure and temperature . we find that @xmath38 generally increases from the midplane outward but the increase is not always monotonic . an example of a typical temperature profile is presented in fig . [ fig : structure]c . the gas is optically thick at all frequencies when @xmath121 and the temperature gradient @xmath128 the gradient is slightly greater than 0 ( @xmath61 increases inward ) at @xmath129 , and slightly less than 0 ( @xmath61 increases outward ) for @xmath130 . the x - ray heating rate increases rapidly at @xmath131 and @xmath38 rises to maintain thermal equilibrium . in this example , the photosphere for frequencies below the lyman edge is at @xmath132 , so the temperature gradient changes sign at this point . the sign flips back at the photosphere for frequencies just above the lyman edge . the gradient remains positive at smaller @xmath96 and the thermal runaway starts at @xmath133 . the upper boundary of the uv reprocessing region is set by the thermal runaway . the temperature runs away , causing the code to fail to converge , when the x - ray ionization parameter @xmath134 where @xmath135 is the x - ray flux at @xmath96 . when @xmath136 exceeds this critical value , bremsstrahlung and recombination can not cool the gas , and its temperature rises toward the compton temperature ( krolik , mckee & tarter 1981 ; voit & shull 1988 ; ferland & rees 1988 ) . the exact mass column of the thermal runaway ( @xmath108 ) depends upon @xmath0 , @xmath3 and @xmath137 , but in general @xmath138 . @xmath108 is never smaller than @xmath139 gm @xmath51 because the skin is optically thin to x - rays above that point and the gas pressure does nt change much over such a small column density . on the other hand , @xmath108 is always less than about one compton depth because the x - rays are exponentially attenuated beyond that point . typically @xmath108 rises sharply at larger radii . a good example of this effect is seen in the model @xmath140 and @xmath141 ( see fig . [ fig : rsig mdot ] ) . the reason for this sharp rise is that heii recombination provides much of the cooling at columns @xmath142 at smaller radii , but the lower temperatures at larger radii mean that little heiii is formed there . within the framework of the model it is possible to estimate @xmath143 , the fraction of the incident x - ray flux which is reradiated in the uv this quantity is determined by how much x - ray flux is either absorbed or reflected by the material in the thermal runaway layer at column densities smaller than @xmath108 . we estimate @xmath143 by the somewhat crude prescription of assuming that the opacity of the warm layer is unchanged by the thermal runaway . a more careful treatment would allow for the reduction in photoelectric opacity due to the higher ionization states found there . nonetheless , we believe our global estimate of @xmath143 is not too bad because the shielding effect is only of major importance when @xmath108 is at least @xmath144 gm @xmath51 . when @xmath108 is that large , it hardly matters whether the disk proper is shadowed by photoelectric opacity or compton scattering . the general trend of @xmath143 is to decrease with increasing @xmath0 because @xmath108 increases . the gas pressure in the skin is @xmath145 on the other hand , @xmath146 at fixed @xmath137 . consequently , @xmath147 at the upper boundary of the skin . thus , as @xmath0 increases , the x - ray flux is able to trigger a thermal runaway at larger column densities and over a broader range of radii . when @xmath148 the x - ray flux is never large enough to cause a runaway ( see fig . [ fig : rsig mdot ] ) . in table [ table : uv fraction ] we list @xmath149 where @xmath150 is the x - ray flux at a mass column @xmath96 in the ring at radius @xmath137 . we find that @xmath143 is 1.0 at @xmath151 and decreases to 0.62 at @xmath141 . the column density of the runaway is insensitive to @xmath3 and @xmath143 varies only slightly when @xmath3 changes by three orders of magnitude . we caution that these numerical values also depend on the specific x - ray spectral shape we have chosen . the material above @xmath108 might influence the emergent spectrum by radiating soft x - rays and by comptonizing the uv flux leaving the disk atmosphere . we can estimate the gas temperature there as @xmath152 ( begelman , mckee & shields 1988 ) , where @xmath153 are the mean intensities of the uv and incident x - rays , respectively , and @xmath154 are the corresponding compton temperatures . although we discount the effect of hard x - rays for heating the disk , they are effective for compton heating , at least up to @xmath155 kev , beyond which the klein - nishina reduction in the scattering rate curtails their contribution . if the spectrum is @xmath156 over many orders of magnitude in photon energy , @xmath157 . as we have already estimated , @xmath158 . the uv and x - ray intensities are comparable in the skin , so the additional cooling provided by scattering of the uv photons will reduce the gas temperature to @xmath159 , roughly independent of radius out to the point where the heating is insufficient to maintain a corona at high temperature . the emission measure ( @xmath160 ) of the soft x - ray corona is dominated by the regions at larger radius where @xmath161 gm @xmath51 . we estimate @xmath162 where @xmath163 is the thickness of the corona . we expect @xmath164 and so the associated bremsstrahlung luminosity ( @xmath165 erg s@xmath166 ) is much smaller than the total disk luminosity unless @xmath167 . similarly , the importance of comptonization to the spectrum depends on the compton @xmath168-parameter , @xmath169 , where @xmath170 is the thomson depth of the layer . because @xmath170 is at most @xmath144 and we expect @xmath171 , this , too , is unlikely to alter the emergent spectrum . we elaborate on the effects of comptonization on the uv spectrum in [ subsubsec : comptonization ] we calculated integrated spectra for face - on disks over a range of @xmath0 and @xmath3 chosen to facilitate comparisons with previous calculations of disk spectra , namely luminosities ranging from 0.003 to 0.3 of the eddington luminosity and central masses of @xmath172 to @xmath173 . in each case , we divided the disk into annuli spaced at ( roughly ) equal intervals in @xmath174 and the local structure and spectrum were calculated . the radii ranged from the inner radius @xmath175 to some outer radius at which the gas temperature was so low that the contribution from the outermost radius to the flux near the lyman edge became negligible . this occurs at @xmath176 where @xmath177 ( eq . [ eq : teff ] ) . finally , the local spectra are added using the method described by ln89 , which accounts for limb - darkening , doppler boosting , and relativistic aberration of the emergent flux . we neglect the general relativistic bending of light rays in the gravitational field of the central black hole ( cunningham 1975 ) . a variety of face - on spectra are plotted in ( figs . [ fig : mass spectra ] , [ fig : mdot spectra ] ) . note that we define @xmath178 as @xmath179 the luminosity per solid angle at an angle @xmath180 from the disk axis . in the simplest treatment of predicted accretion disk spectra , the dynamics are approximated as newtonian , and the spectra of individual rings are assumed to be blackbodies . these approximations lead to a temperature profile @xmath181 and an integrated spectrum whose shape is roughly @xmath182 , where @xmath183 is the temperature of the innermost ring . it is clear from these figures that such an approximation is not a very good description of our predicted spectra . at low frequencies , there is considerable spectral curvature , and strong lyman edge features substantially modify the shape of the exponential cut - off at high frequencies . three different effects alter the spectral shape at frequencies below the lyman edge . first , relativistic effects flatten the temperature profile in the inner rings ( novikov & thorne 1973 ) . this softens the spectrum relative to the newtonian approximation . second , the presence of a large emission feature at the lyman edge implies that a smaller fraction of the reprocessed flux emerges at frequencies below the lyman edge . this depresses the general level of the near - uv continuum . third , at frequencies below the lyman edge , the mass column of the uv photosphere increases with increasing frequency . the flux at lower frequencies then comes from hotter gas because the gas temperature typically increases as @xmath96 decreases . this softens the continuum spectrum relative to the black body approximation . the low - frequency spectrum becomes harder with increasing hard x - ray flux , @xmath2 . the disk temperature is roughly proportional to the effective temperature of the x - ray flux , so the peak of the disk spectrum shifts from @xmath184 hz in the coldest disks to @xmath185 hz in the hottest . to quantify this effect , we define @xmath186 at @xmath187 . the value of this quantity for each model we computed is displayed in table [ table : uv fraction ] . the smallest @xmath188 we found was -1.6 ( @xmath189 ) ; the largest was 0.8 ( @xmath190 ) . as the previous paragraphs explain , some , but not all , these effects are due to the external irradiation . in particular , the temperature gradient effect acts the opposite way when the dissipation is inside the disk rather than outside . an improved calculation of the spectrum of a conventional disk with internal dissipation methods yields a predicted @xmath191 ( sincell & krolik 1996 ) . we found a strong lyman edge emission emission feature over the entire parameter range we studied ( see figs . [ fig : localspec2 ] , [ fig : mass spectra ] and [ fig : mdot spectra ] ) . this , too , is due to the temperature increasing upward ( [ fig : structure]d , [ fig : structure]c ) . the larger opacity above the edge than below causes the lyman continuum photosphere to lie higher in the atmosphere than the photosphere at frequencies just below the edge . the higher temperature at the lyman continuum photosphere naturally leads to a greater emergent flux , i.e. an emission feature ( fig . [ fig : localspec2 ] ) . additional emission features appear at the hei and heii edges in some of the integrated spectra . the smearing of the edges is caused by doppler boosting of radiation emitted at small @xmath137 . the amplitude of the lyman edge feature increases with the central mass ( fig . [ fig : mass spectra ] ) because @xmath192 ( eq . [ eq : skin gas pressure ] ) . low pressure gas cools less efficiently because both the bremsstrahlung and photoionization opacities are proportional to @xmath193 . inefficient cooling leads to larger temperature inversions in the skin . the ratio @xmath194 remains roughly constant because the ratio of the opacities is independent of @xmath193 . therefore , @xmath195 increases with @xmath3 and the amplitude of the lyman edge feature increases . the amplitude of the edge feature is nearly independent of @xmath0 ( fig . [ fig : mdot spectra ] ) . we incorporated an iterative calculation of the variable eddington factors ( auer & mihalas 1970 ) into the @xmath196 , @xmath140 model . the solution for the eddington factors ( @xmath67 ) and the mean intensity is equivalent to the exact solution for the angular distribution of the spectral intensity . the angular distribution of the emergent flux can be determined by a formal solution of the transfer equation with the known source function . the surface eddington factors ( eq . [ eq : surface eddington factor ] ) can not be computed iteratively because we do not have a solution at @xmath109 . the accuracy of the analytic approximation for the eddington factors ( eq . [ eq : analytic fedd ] ) was tested by comparison with the exact solution . the radiation field is isotropic when @xmath197 and @xmath198 , in good agreement with the analytic approximation . however , when @xmath199 we find that @xmath200 . the approximate @xmath67 assumes that the radiation is collimated perpendicular to the accretion disk so that @xmath201 . fortunately , the errors introduced by our assumed @xmath67 are small . the approximate solution underestimates the true gas temperature by @xmath202 at @xmath203 and overestimates the temperature by @xmath118 at larger @xmath96 . the gas pressure is underestimated by @xmath118 at all @xmath96 . the differences in the uv spectrum depend on frequency and never exceed @xmath204 . the approximate solution overestimates the mean intensity below the lyman edge and above the heii edge and underestimates @xmath60 between these frequencies . as a consequence , the amplitude of the lyman edge feature is larger in the exact solution . the observed flux from an accretion disk around a massive black hole depends upon the disk inclination angle . first , as discussed in the previous paragraph , the emergent flux has an intrinsic angular distribution in the local rest frame ( see below ) . second , there are a number of relativistic effects acting upon photons emitted from material orbiting near the black hole . relativistic orbital motion boosts their frequencies and collimates their directions , while the strong gravitational field bends their trajectories and imposes a gravitational redshift on their frequencies as viewed at large distance ( cunningham 1975 ) . the boosting and beaming due to orbital motion tend to be stronger than the gravitational effects , so the net result is to strengthen the radiation emitted close to the equatorial plane from the innermost rings of the disk ( cunningham 1975 , laor , netzer & piran 1990 ) . these effects are stronger in rotating black holes than in non - rotating black holes because the disk extends closer to the event horizon in the former case ( laor , 1990 ) . for this reason , the impact of general relativistic effects is small in our ( non - rotating black hole ) models . we calculated the integrated spectrum of the @xmath205 , @xmath140 disk for three disk inclination angles . for each inclination , the disk is divided into several annuli ( [ subsubsec : face on disk ] ) and each annulus subdivided into several sectors . the radiative transfer equation is solved for the spectral intensity at the appropriate rest - frame inclination angle using the exact source function and the results are summed over all annuli and sectors . as before , doppler boosting , relativistic aberration and the gravitational redshift are included , but we neglect the general relativistic bending of the light rays by the central black hole . at low frequencies , the emergent intensity is limb - darkened ( fig . [ fig : limb disk ] ) . due to the low opacity below the lyman edge , the photosphere for these frequencies tends to lie relatively deep inside the atmosphere , where the temperature gradient is small or positive ( i.e. temperature decreases upward ) . consequently , the angular dependence is determined largely by the usual limb - darkening influence of scattering . however , at frequencies well above the lyman edge , the intensity is more nearly isotropic . several effects combine to explain this change . because of the greater opacity , the photosphere in this range lies higher in the atmosphere , where the temperature tends to increase outward . therefore , photons traveling well away from the axis can only depart from relatively high altitude , where the temperature is comparatively high . in addition , because these frequencies are radiated almost exclusively by the innermost rings of the disk , relativistic boosting and beaming are especially important . the decreasing degree of limb - darkening with increasing frequency causes the feature at the lyman edge to strengthen with disk inclination . we expect the relative limb - brightening of the ionizing continuum to increase with central mass and , if @xmath206 is approximately constant , disk luminosity . this is because temperature gradients increase with @xmath3 ( [ subsubsec : face on disk ] ) . doppler broadening of the lyman edge increases with disk inclination . the typical line of sight to an edge - on disk is nearly parallel to the disk rotation velocity so the doppler boost is maximized . the lyman edge is formed at @xmath207 , where @xmath208 , so the edge is spread over a range @xmath209 ( fig . [ fig : limb disk ] ) . the peak of the emergent flux shifts upward by the same amount . an edge - on disk will not appear to have an emission feature at the lyman edge , but the continuum slope becomes harder going from frequencies below the lyman edge to higher frequencies . this result is consistent with the conclusions of laor ( 1992 ) . the reprocessed uv radiation is inverse compton scattered by the relativistic electrons in the coronae . we calculate the comptonized uv spectrum for scattering by a relativistic maxwell - boltzmann distribution of electrons @xmath210 where @xmath211 , the coronal temperature is @xmath212 and @xmath213 is the second order modified bessel function . this distribution is normalized so that @xmath214 and we assume a thomson depth , @xmath8 , for the corona . we further assume that each photon has a probability @xmath215 of being scattered in the corona and that each photon is scattered at most one time . the spectrum of the scattered radiation is calculated using the formulae for isotropic electron and photon distributions ( rybicki & lightman 1979 ) . the small corrections due to anisotropy of the incident uv radiation field were neglected ( haardt 1993 ) . higher order scatterings determine the shape of the high energy spectrum ( pozdnyakov , sobol & sunyaev 1976 ) so we plot the results over a limited range of frequencies near the lyman edge ( figs . [ fig : compton t ] and [ fig : compton tau ] ) . the spectrum of the comptonized uv flux is determined primarily by the thomson depth of the corona , @xmath8 . inverse compton scattering in an optically thin corona has a negligible effect on the spectrum of the uv continuum because few photons are scattered ( fig . [ fig : compton t ] ) . comptonization in an optically thick corona ( @xmath7 ) smooths the spectrum at all frequencies ( fig . [ fig : compton tau ] ) , preferentially scattering photons from low to high energy . as a consequence , the spectrum of the non - ionizing flux becomes steeper and the ionizing continuum hardens to an approximate power - law . because the width of the lyman edge is dominated by relativistic boosting , inverse compton scattering in an optically thick corona may reduce its amplitude but does not significantly broaden it ( fig . [ fig : compton tau ] ) . the amplitude of the emission feature reaches a minimum when @xmath216 . the actual value of @xmath217 depends upon the parameters of the corona and the input uv spectrum but the range of values is easily understood . each input photon has a probability @xmath218 of being scattered in the corona and the scattered photons are redistributed over a frequency range @xmath219 . thus , very little redistribution occurs when @xmath220 and the scattered photons emerge at much higher energies when @xmath221 . the total optical depth of the corona is approximately the sum of the contributions from the soft x - ray corona ( discussed in [ subsec : vertical structure ] ) and the hard x - ray corona , where the gravitational potential energy is released . in [ subsec : vertical structure ] we demonstrated that a thermal runaway occurs for @xmath222 for @xmath223 and @xmath224 at larger radii . this range of column densities corresponds to @xmath225 . the lyman edge is formed at @xmath226 , where @xmath227 . the optical depth ( @xmath228 ) and geometry of the hard x - ray corona are controversial , but may be estimated through the use of thermal comptonization models . hm93 find @xmath229 assuming a uniform slab ; zdziarski ( 1995 ) found @xmath230 for the same geometry ; pietrini & krolik ( 1995 ) confirmed the zdziarski ( 1995 ) model , but also suggested that either clumping of the corona or a physical separation between it and the reprocessing surface would permit a match to the observed x - ray slope with larger @xmath228 . stern , ( 1995 ) prefer a model in which the corona is clumped , but close to the accretion disk , and has @xmath231 . if the corona is highly clumped , it can have little effect on the bulk of the uv photons . these modeling efforts indicate that a smooth corona must be rather optically thin , so it seems unlikely that the mean total compton depth overlying the rings producing the lyman edge can be any greater than a few tenths in most cases . accretion disks are self - gravitating and unstable when ( toomre 1964 , goldreich & lynden - bell 1965 , sakimoto & coroniti 1981 ) @xmath232 or @xmath233 where @xmath234 is the keplerian angular frequency . the range of length scales for the growing perturbations is @xmath235 ( toomre 1964 ) . comparing eqs . [ eq : isothermal gas density ] and [ eq : self gravity criterion ] , we see that an isothermal accretion disk with no internal dissipation is gravitationally stable if @xmath236 the temperature of the isothermal disk is reasonably well approximated by the effective temperature of the irradiating flux ( eq . [ eq : teff ] ) because the thomson depth of the skin is near unity . in contrast , the central temperature of the standard optically thick disk can be more than an order of magnitude larger than @xmath237 . inserting @xmath237 into the stability criterion ( eq . [ eq : isothermal disk instability ] ) , we find that the isothermal disk is unstable at radii @xmath238 the disk is self - gravitating at all radii ( @xmath239 ) when @xmath240 of the disks we have considered , only the @xmath205 and @xmath241 model satisfies this criterion over the entire range of radii . the case @xmath242 and @xmath243 is stable for @xmath244 and the remainder are unstable at all radii . although a strictly isothermal disk is unstable , a very small amount of internal dissipation will stabilize an irradiated disk . the gas pressure dominated solution for the disk structure described in sz95 applies when there is enough internal dissipation ( and optical depth ) to significantly elevate the central temperature above @xmath237 ( sz95 and appendix [ app : cold disk structure ] ) . this occurs when @xmath245 where @xmath246 is the electron scattering optical depth of the disk and @xmath247 is the fraction of the gravitational potential energy which is dissipated in the corona ( appendix [ app : cold disk structure ] and sz95 ) . as a result of the warmer internal temperature , the scale height increases and the density falls . we find that the central density drops below the critical density ( eqs . [ eq : isothermal gas density ] and [ eq : self gravity criterion ] ) and the disk is stabilized when @xmath248 all of our models are stable at @xmath249 with small values of @xmath250 . the largest value of @xmath251 required for stability occurs in the @xmath205 and @xmath252 case at @xmath253 . substantially smaller values of @xmath251 are needed to stabilize the disk at @xmath244 , where most of the lyman edge radiation is produced , and for smaller values of @xmath0 and @xmath3 . in general , @xmath254 increases , and the value of @xmath251 required to stabilize the disk decreases , with increasing luminosity and @xmath3 ( eq . [ eq : self gravity criterion ] ) . from the scaling behavior we have already discovered , we may make some educated guesses about how the disk spectra will change if @xmath250 changes from zero to the small values required for stability . because the density throughout the disk will fall , while x - ray heating will keep the gas temperature in the heated layer near @xmath237 , we expect the gas pressure to decrease with increasing @xmath250 . therefore , the ionization parameter in the heated skin will increase , and more of the incident x - ray flux will emerge as soft x - rays rather than in the ultraviolet . ( [ subsec : vertical structure ] ) . in addition , we expect the amplitude of the lyman edge feature will increase as @xmath251 increases because of the lower gas pressure in the atmosphere ( [ subsubsec : face on disk ] ) . the disk is supported by radiation pressure at @xmath255 when @xmath256 and over a larger range of radii for smaller @xmath247 ( sz95 ) . radiation pressure support will inflate the disk , suppressing the gravitational instability , but may cause the disk to become thermally unstable ( piran 1978 ) . however , for our range of model parameters , the value of @xmath247 needed to stabilize the disk is much smaller than this value in all but the highest luminosity disk ( @xmath141 and @xmath140 ) . we have calculated the vertical structure and uv spectrum of an accretion disk irradiated by hard x - rays from a corona , as predicted by many models for the x - ray emission in agn ( pounds 1990 ; hm93 ) . consistent with these x - ray production models , we assume there is no viscous dissipation in the accretion disk . the gravitational potential energy of the accreting gas is dissipated in a corona and radiated as hard x - rays with a power law spectrum . photoionization of oxygen and iron by the coronal x - rays heat a geometrically thin skin at the surface of the accretion disk . because there is no dissipation inside the disk proper , its temperature is constant as a function of altitude and the gas density and optical depth are much larger than in a conventional dissipative disk ( sz95 ) . to find the structure of the x - ray heated skin , we have solved the differential equations of radiative transfer , hydrostatic equilibrium and surface mass density subject to the constraints of radiative equilibrium and charge conservation . the inner boundary conditions for this numerical solution were chosen to match an analytical solution for the underlying isothermal accretion disk . we tested our method by computing the structure and spectrum of a conventional accretion disk and comparing to the previous state - of - the - art calculation ( ln89 , sincell & krolik 1996 ) . the inner edge of the x - ray heated skin is 3 - 4 scale heights from the midplane of the isothermal disk . in the skin , but not the body of the disk , the force of the uv flux supports the gas against gravity and the momentum of the absorbed x - rays . the gas pressure in the skin is orders of magnitude smaller than the pressure at the midplane of the disk and is nearly constant throughout the skin . the gas temperature typically increases from the inner edge of the skin to the surface , although the temperature gradient occasionally changes sign . the changes in the temperature gradient correspond to changes in the ionization state and opacity of the gas . the gas above the column density @xmath108 is unable to cool because cooling by thermal radiation processes is no longer able to balance heating by x - ray photoionization . this transition takes place at a characteristic value of the ionization parameter @xmath257 . compton cooling was not included in this model so we could not find a solution for @xmath109 . however , we estimate the gas temperature in the soft x - ray corona to be @xmath258 . we find that @xmath108 increases with disk radius , because the cooler gas at large radii radiates less efficiently , and the average value of @xmath108 for a given disk increases with the accretion rate , because @xmath147 at the surface of the heated skin . for our specific x - ray spectrum and set of approximations , the fraction of the incident hard x - ray flux which is reprocessed into uv photons ranges from @xmath259 at @xmath151 to 0.62 when @xmath141 . the exact value of @xmath143 depends on the shape of the x - ray spectrum , and our method ( see 4.1 ) can be expected to slightly underestimate @xmath143 . nevertheless , the decrease in @xmath143 with accretion rate is a direct consequence of our assumption that all the gravitational potential energy is dissipated in the corona ( see eq . [ eq : skin gas pressure ] ) and should be a general property of irradiated disks . the reprocessed uv continuum depends significantly on the specific parameters of the model ( accretion rate , central mass ) , but certain features are generic : it has strong spectral curvature , in the sense that the spectrum softens with increasing frequency ; and there is a significant lyman edge emission feature . the low frequency continuum ( parameterized by the spectral index at @xmath260 ) hardens with increasing x - ray flux ( @xmath2 ) and ranges from -1.6 when @xmath261 to 0.8 when @xmath190 . the continuum slope in this frequency range is only weakly dependent on viewing angle . the lyman edge is in emission because the photosphere is higher in the atmosphere at frequencies above the edge , and the temperature increases upward . additional emission features at the hei and heii edges may also occur in the spectra of the hotter disks . the amplitude of the lyman edge feature increases with central mass and is insensitive to the accretion rate . this is because both the photoionization and bremsstrahlung opacities are proportional to @xmath192 , so the gas cools less efficiently as @xmath3 increases . this results in larger temperature inversions which increase the strength of the emission feature . doppler boosting spreads the lyman edge over a range @xmath262 . the amplitude of the lyman edge increases with disk inclination because the non - ionizing continuum is limb - darkened and the ionizing flux is ( relatively ) limb - brightened . irradiated disks are self - gravitating , and unstable , unless there is a small amount of internal dissipation . we find that the strictly isothermal disk is gravitationally unstable at @xmath263 , so that strictly isothermal disks with luminosities @xmath264 ergs s@xmath166 are unstable at all radii . if , as is likely , there is a small amount of internal dissipation , and this stabilizes the disk , we expect @xmath143 to decrease , and the amplitude of the lyman edge feature to increase relative to what our models show for the limiting case of @xmath265 . these effects are likely to become stronger for larger luminosity and central mass . some , but not all , of these predictions are consistent with observations . the average optical - uv power - law index for quasars is @xmath266 ( laor 1990 ) but a single power law is a poor description of the continuum spectrum ( francis , 1991 ) . the continuum of the composite quasar spectrum , formed by combining all the quasars in the large bright quasar survey , softens with increasing frequency ( francis , 1991 ) . both the mean optical - uv slope and the spectral curvature are consistent with our predicted uv spectrum of an irradiated disk if @xmath267 . on the other hand , our finding that a lyman edge emission feature is a general property of x - ray - irradiated disks is difficult to reconcile with recent observations of agn . lyman edge features are rarely seen , and then usually as partial absorption edges ( antonucci , kinney & ford 1989 , koratkar , kinney & bohlin 1992 ) . emission features at the lyman edge have been detected only in polarized light , and in these two quasars the total flux spectrum shows the lyman edge in absorption ( koratkar , 1995 ) . the polarization of this emission is quite high and detailed polarization dependent calculations ( blaes & agol 1996 ) will be required to determine if these features could be produced by an irradiated disk . internal dissipation might , in some cases , reduce the amplitude of the lyman edge feature . conventional disks ( sincell & krolik 1996 ) sometimes produce emission features at the lyman edge , and sometimes absorption , depending primarily on @xmath0 . however , both the fraction of the total gravitational energy dissipated in the disk and @xmath0 must be very finely tuned to eliminate lyman edge features from all agn . inverse compton scattering in the disk coronae can smooth out the lyman edge feature only if the hot corona producing the hard x - rays covers most of the disk and has an optical depth @xmath7 . because most recent treatments of x - ray production suggest that either the optical depth is rather smaller than this , or else the hot corona is highly clumped , it is unlikely that compton scattering smooths the lyman edge . the only escape we see from this contradiction makes use of the fact that the evidence for x - ray irradiation comes almost entirely from low luminosity agn ( seyfert galaxies ) , while spectra of the lyman edge region exist only for high luminosity agn ( quasars ) . the reason for this separation is that only the seyfert galaxies , which are relatively nearby , are bright enough for detailed x - ray spectroscopy , while in order for us to see the lyman edge , we must look at objects with redshifts great enough to bring it below the lyman cut - off imposed by our galaxy s interstellar medium . perhaps , then , seyfert galaxies have lyman edges in emission ( if we could only see them ) , while quasar disks are not so thoroughly irradiated by x - rays . x - ray spectroscopy of quasars by future , more sensitive instruments , will be able to test this idea . thus , our calculation of the uv spectrum expected from irradiated disks raises serious questions for accretion disk models . while the x - ray production and x - ray reflection models which motivated this study have had substantial success in describing agn x - ray spectra , we now see that they also predict strong lyman edge emission features , a prediction that is contradicted by the simplest interpretation of the observations . we would like to thank ari laor for helpful conversations and for providing many disk spectra . we also thank omer blaes and eric agol for ongoing conversations . mws also thanks the observatoire de meudon for hospitality during part of this work . mws received support for this work from nasa grants nagw-3129 , 1583 and nag 5 - 2925 , and nsf grant ast 93 - 15133 . jhk was partially supported by nasa grant nagw-3156 . blaes , o. m. & e. agol , 1996 , submitted to ap . j. letters ( hm93 ) ( ln89 ) laor , a. 1992 , in testing the agn paradigm : aip conference proceedings 254 , american institute of physics , new york martocchia , a. and matt , g. 1996 , m.n.r.a.s . in press mihalas , d. m. , 1978 , stellar atmospheres , w. h. freeman & co. , san francisco nandra , k. 1997 , in iau coll . 159 , emission lines in active galactic nuclei : new methods and techniques , eds . b. peterson , f .- z . cheng , and a.s . wilson novikov , i. & k. s. thorne , 1973 in black holes , p. 422 , eds : dewitt , c. & b. dewitt , gordon & breach , new york pounds , k. a. & t. j. turner , 1988 proceedings of the iau symposium 134 active galactic nuclei " , eds . j. s. miller & d. e. osterbrock , kluwer , dordrecht press , w. h. , s. a. teukolsky , w. t. vetterling & b. p. flannery , 1992 , numerical recipes in c , cambridge university press rybicki , g. b. & a. p. lightman , 1979 , radiative processes in astrophysics , john wiley & sons , inc . , new york ( ss73 ) sincell , m. w. & j. h. krolik , 1996 , in preparation ( sz95 ) zheng , w. , kriss , g.a . , telfer , r.c . , grimes , j.p . , and davidsen , a.f . 1996 , ap.j . in press the radial structure equations for the cold gas pressure dominated disk are ( sz95 ) @xmath269 @xmath270 and the perfect gas equation of state . the gas temperature of the disk is determined by the x - ray heating rate ( eq . [ eq : dissipation rate ] ) @xmath271 when @xmath268 . sz95 assume that the gas temperature is proportional to the effective temperature of the flux dissipated _ internally _ and neglect the x - ray heating . consequently , the sz95 solution for the radial structure is singular ( @xmath272 ) when @xmath265 . the solution for the radial structure is straightforward . we find that the scale height is @xmath273 where @xmath274 the midplane gas density is @xmath275 where we have inserted the scalings in eq . [ eq : disk parameter scalings ] . these equations reduce to the isothermal disk solution ( eqs . [ eq : disk scale height ] and [ eq : isothermal gas density ] ) for a disk temperature of @xmath276 when @xmath265 , as expected . the introduction of a new definition of the disk temperature removes the singularity of the sz95 solution at @xmath265 . the cold disk temperature will never fall to zero , as predicted by sz95 , because of the x - ray heating by the corona . as a result , the scale height is non - zero and the density remains finite for @xmath265 . in addition , we find that the scale height and midplane density of a gas pressure dominated disk are insensitive to the actual value of @xmath247 , so long as it is close enough to unity that the central temperature is determined primarily by irradiation ( _ cf . _ 5 ) . figure [ fig : structure ] . the structure of the x - ray heated skin at @xmath277 for @xmath205 and @xmath140 . the panels show the vertical coordinate scaled to the value at the inner edge of the skin ( @xmath278 ) ( a ) , the gas pressure ( b ) , the gas temperature ( c ) and the location of the photosphere ( d ) . figure [ fig : localspec2 ] . the spectrum at @xmath277 for @xmath196 and @xmath140 . the crosses are the results of the simulation and the dashed line represents a black body spectrum at the effective temperature of the disk .

using monte carlo simulation , we analyse the behaviour of two - dimensional hard rods in four different types of geometric confinement : ( i ) a slit pore where the particles are confined between two parallel walls with homeotropic anchoring ; ( ii ) a hybrid slit pore formed by a planar and a homeotropic wall ; square cavities that frustrate the orientational order by imposing either ( iii ) homeotropic or ( iv ) planar wall anchoring . we present results for the state diagram as a function of the packing fraction and the degree of confinement . under extreme confinement , unexpected states appear with lower symmetries than those of the corresponding stable states in bulk , such as the formation of states that break the anchoring constraints or the symmetry imposed by the surfaces . in both types of square cavities the particles form disclinations at intermediate densities . at high densities , however , the elastic stress is relaxed via the formation of domain walls where the director rotates abruptly by ninety degrees .

1502.03583

a wealth of new phenomena arises when a liquid crystal is confined in a pore , even in the simplest of geometries where a slit pore is formed by two identical walls that are parallel to each other . in that case the isotropic - nematic transition inside the pore is shifted with respect to the bulk phase transition and terminates in a capillary critical point at a specific value of the wall separation distance . the capillary binodal or capillary nematization line @xcite forms the analogue of the capillary condensation line in simple fluids . when confining a smectic phase , layering transitions @xcite and the suppression of the nematic - smectic transition for specific pore widths @xcite occur due to a commensuration effect between the size of the pore and the smectic layer spacing . in a hybrid pore , formed by two parallel walls with two antagonistic anchoring conditions @xcite , a balance between the anchoring strengths of the two walls and the elastic energy of the liquid crystal determines the director configuration . the director field can either gradually rotate or generate a step - like defect . in the latter , the director rotates abruptly by @xmath0 . if one surface imposes much stronger anchoring than the other , then the director can be approximately constant in the entire capillary . disclinations are another genuine effect of confinement of liquid crystals . disclinations can appear by curvature of the underlying space @xcite , e.g. by confining a liquid crystal on the surface of a sphere @xcite . the geometry and the dimensionality of the system restrict the topology of the defects that can arise in a given system . the equilibrium director configuration can be highly non - trivial as a result of the balance between the inner forces of the system and the external interaction with the surfaces . in this paper we study the corresponding two - dimensional system of confined liquid crystals . we investigate whether the above phenomenology of confinement effects persists in two dimensions . we model the particles as hard rods with rectangular shape . this model has been previously investigated . the bulk phase behaviour was analyzed with monte carlo ( mc ) simulation and density functional theory ( dft ) @xcite . the confinement in a planar slit pore , i.e. a slit pore in which both walls promote planar anchoring , has been studied with dft @xcite using the restricted - orientation approximation ( the orientation of the particles is restricted to two perpendicular axes ) . triplett and fichthorn used orientational - bias mc simulation @xcite to study a planar slit pore with selected system sizes . in @xcite gonzlez pinto et . al . analysed the confinement in square cavities with planar anchoring using dft in the restricted - orientation approximation . finally , very recently , the orientational ordering in circular cavities of selected radii has been analyze with mc simulation @xcite . here , we confine the particles in homeotropic and hybrid slit pores and square cavities . we study the phase behaviour by means of monte carlo simulation . for all the geometries investigated , we present the state diagram in the plane of system size and density . we find the expected phenomena , such as capillary nematization and smectization and the formation of disclinations in closed cavities . moreover , we find unexpectedly domain walls and states that break the anchoring or the symmetry imposed by the surfaces . we explain qualitatively the stability of these new states as a balance between different contributions to the free - energy . we expect these new states to appear also in three - dimensional systems under extreme confinement , a regime that as not been investigated yet . despite the simplicity of the model , the results can be relevant to understand a variety of systems where packing plays a crucial role such as e.g. experiments on vibrated granular rods @xcite , the adsorption of colloids on substrates @xcite , the confinement of actin filaments @xcite and colloids @xcite in quasi - two - dimensional geometries , the assembly of anisotropic nanoparticles at liquid - liquid @xcite or liquid - air @xcite interfaces , and the confinement of magnetic nanorods @xcite . we consider a system hard rectangular particles of length - to - width ratio @xmath1 that interact through excluded volume interactions . the position vector @xmath2 of the center of mass of the @xmath3-th particle and the unit vector along the long particle axis @xmath4 determine the configuration of rod @xmath3 . we confine @xmath5 of such particles between two parallel walls or in a square cavity . the interaction between the rods and the surfaces is modelled via an external hard potential @xmath6 . we use hard walls ( all corners of a particle can not penetrate the wall ) to promote planar anchoring such that the long particle axis aligns preferentially parallel to the walls . we use hard center - walls ( the center of mass of a particle can not penetrate the wall ) to promote homeotropic anchoring , that is , the long particle axis aligns preferentially perpendicular to the wall : @xmath7 @xmath8 where @xmath9 , with @xmath10 the boltzmann constant and @xmath11 the absolute temperature . a hard wall induces planar anchoring even for a single particle because a rod sufficiently close to the wall must adopt a planar configuration . in contrast , homeotropic alignment at a hard center wall emerges from collective behaviour ( a single particle sufficiently close to the wall can adopt any orientation ) . the homeotropic alignment promoted by hard center walls have been previously shown in mc simulation @xcite and dft studies in two @xcite and three @xcite dimensions using different particle shapes . given the different origin of both types of anchoring , we expect the planar wall to promote stronger alignment of the particles than the homeotropic wall . the different geometries that we have analysed are schematically represented in fig . [ fig1 ] . in the homeotropic cell ( fig . [ fig1]a ) the rods are confined between two parallel hard center - walls on the @xmath12 axis . we set periodic boundary conditions in the @xmath13-direction . the length of the simulation box in the @xmath13-direction is @xmath14 . in the hybrid cell ( fig . [ fig1]b ) one of the walls promotes homeotropic anchoring and the other induces planar alignment of the particles . periodic boundary conditions are again applied along the @xmath13 axis . we also study confinement in square cavities with planar ( fig . [ fig1]c ) and with homeotropic ( fig . [ fig1]d ) walls . in all cases we fix the origin of coordinates in the middle of the simulation box . @xmath15 designates the side length of the square cavities and the distance between the parallel walls of the homeotropic and hybrid cells . in order to compare between different geometries , we use an effective distance @xmath16 ( see fig . [ fig1 ] ) . it takes into account that the center - walls ( homeotropic anchoring ) can be penetrated by the particles by a distance @xmath17 . hence , for homeotropic pores and homeotropic square cavities @xmath18 , for hybrid pores @xmath19 , and for planar square cavities @xmath20 . we define the packing fraction @xmath21 as the ratio between the area covered by the particles and the total area of the simulation box , i.e. @xmath22 with @xmath23 for the square cavities and @xmath24 in the case of slit pores . is the distance between two parallel walls and @xmath16 is the effective distance that accounts for the extra space that can be occupied by the corners of the particles if homeotropic walls are present . in the middle of the hybrid cell ( b ) we show the particle geometry : a hard rectangle of length @xmath25 and width @xmath26 . the unit vector along the main axis is @xmath27 and @xmath28 is the angle between @xmath27 and the horizontal axis.,width=307 ] we study the equilibrium configurations of the confined rod systems by means of ( standard ) mc simulations at fixed number of particles and system area . temperature is irrelevant for hard core systems . following the ideas in @xcite , we initialize the system at low densities ( @xmath29 ) with the particles randomly located and oriented in the simulation box . then we equilibrate the system and perform @xmath30 monte carlo steps ( mcs ) to obtain equilibrium configurations . here , a mcs is defined as @xmath5 single particle trial moves , consisting on translation and rotation . once the simulation ends we insert a few particles and run a new mc simulation . we repeat the insertion of particles until the desired packing fraction is reached or until no new particles can be added . finally , in order to rule out metastable states , we repeat the whole process but now starting with the last configuration of the simulation ( high packing fraction ) and removing a few particles each time . in this way we can compare the configurations obtained by increasing and decreasing the number of particles . they should be the same because gradual transitions between states are expected giving the dimensionality of the system . in order to insert new particles we randomly choose one particle and create a parallel replica displaced by @xmath31 in the direction of the short particle axis . then we move and rotate the new particle a few thousands times . when the insertion of a new particle leads to overlap with other particles we choose a new rod to create the replica . when decreasing the number of particles we just select one particle at random and remove it . the number of inserted / removed particles from one simulation to another is chosen such that the change in packing fraction is small , between @xmath32 ( at low densities ) and @xmath33 ( at higher densities ) . the probability @xmath34 of accepting one particle move depends on the maximum displacement @xmath35 and maximum rotation angle @xmath36 each particle is allowed to perform in one mcs . we aim for @xmath37 . both @xmath35 and @xmath36 are calculated each time we change @xmath5 . for each @xmath5 we first find @xmath36 in order to accept half of the rotations . next we find @xmath35 in order to achieve the desired value of @xmath34 . we characterize the structural properties of the confined rods with three local fields : the density @xmath38 , the uniaxial order parameter @xmath39 and the tilt angle @xmath40 . @xmath39 is defined as the largest eigenvalue of the local order tensor @xmath41 , where @xmath42 is the unit vector along the main axis of the @xmath43particle , @xmath44 denotes the kronecker delta , and @xmath45 is a canonical and spatial average . the tilt angle @xmath40 is the angle between the local director ( given by the eigenvector of @xmath39 ) and the @xmath12 axis . the local quantities @xmath38 , @xmath39 and @xmath40 are defined , at each @xmath46 , as an average over @xmath47 different configurations at intervals of @xmath48 mcs . due to the symmetry in the case of slit pores , the dependence of the local fields is only on the @xmath12 axis ( the axis perpendicular to the walls ) . we divide the @xmath12 axis in @xmath49 equidistant bins and for each bin we calculate the local fields by including all the particles with its center of mass located inside the bin . for the square cavities ( fig . [ fig1]c and [ fig1]d ) we study the full @xmath12 and @xmath13 dependence of the local fields . in order to obtain spatially smooth fields we calculate the local order tensor by including , for each @xmath50 , all the particles whose center of mass is located in a circle of radius @xmath51 centered at @xmath46 . in bulk a fluid of hard rectangles of aspect ratio @xmath1 undergoes a phase transition from an isotropic to a nematic phase upon increasing the density @xcite . the bulk transition is continuous , presumably of the kosterlitz - thouless type @xcite . the aim of the present work is to study confinement properties , and hence we have not analysed in detail the bulk properties . however we have run bulk simulations using a square box of side length @xmath52 with periodic boundary conditions and found a continuous isotropic - nematic transition at @xmath53 , in agreement with the predictions of the scale particle theory @xcite . in what follows we show the states that occur under confinement . for each geometry we group the states in a diagram as a function of the system size and the packing fraction . we use the local fields ( density , uniaxial order parameter and tilt angle ) to distinguish between distinct states . however , one should bear in mind that given the confined geometries analyzed here and the dimensionality of the system , one expects gradual transitions between the different states . in addition , the distinction between states , such as nematic , smectic , or isotropic , in confined geometry is not clearcut . and packing fraction @xmath21 for hard rectangles ( @xmath1 ) confined in a homeotropic cell . the color indicates the average uniaxial order parameter @xmath54 . white circles show the state points where @xmath55 ( the white - dashed line is a guide to the eye ) . the black lines delimit approximate boundaries between different states ( no simulation data is available for the regions depicted in white at high packing fractions ) . ] we first consider hard rectangles ( @xmath1 ) confined between two parallel planar walls inducing homeotropic anchoring ( see fig . [ fig1]a ) . the state diagram in the plane of packing fraction as a function of pore width is depicted in fig . we have calculated about @xmath56 state points with @xmath16 varying between @xmath57 and @xmath58 . the color map shows the average of the uniaxial order parameter inside the pore @xmath59 * isotropic , nematic and smectic states . * first we focus on the larger pores that we have investigated , @xmath60 . we identified three distinct states : isotropic ( i ) , nematic ( n ) , and smectic ( s@xmath61 , with @xmath3 the number of smectic layers inside the pore ) . examples of the particle configurations and the density profiles for each state in a pore with @xmath62 are shown in figs . [ fig3 ] and [ fig4 ] , respectively . at low densities the particles form an isotropic state ( fig . [ fig3]a ) . the density profile ( fig . [ fig4 ] , top panel ) is rather constant with a small amount of adsorption of particles close to the walls . the uniaxial order parameter ( middle panel ) is zero except in a small region near the walls where it shows incipient orientational order due to the walls . in this region the particles are ( slightly ) aligned with their long axes perpendicular to the walls , as the tilt angle ( bottom panel ) shows . the maximum in density occurs at contact with the surfaces . however , the maximum of the uniaxial order parameter is shifted @xmath63 away from the walls . this is a general feature of a hard center - wall that has been previously reported in three- @xcite and two- @xcite dimensional systems on the basis of mc simulation and dft . by increasing the density , the capillary nematization ( i.e. the formation of a nematic state inside the pore ) occurs in a continuous fashion , as expected . in the nematic state all the particles are oriented , on average , perpendicular to the walls ( fig . [ fig3]b , and fig . [ fig4 ] bottom panel ) . the uniaxial order parameter is positive in the whole capillary ( fig . [ fig4 ] middle ) and there is incipient positional order that propagates into the pore from the walls ( fig . [ fig4 ] top ) . by further increasing the density the particles form a smectic state ( fig . [ fig3]c ) with well - defined layers ( fig . [ fig4]a ) , the number of which is the result of commensuration between the size of the pore and the smectic period . in the range of pore sizes investigated here , we have found smectic states with @xmath64 to @xmath65 layers . the smectic layers are slightly tilted , especially those in the center of the pore ( see the tilt angle profile in fig . [ fig4]c ) . the reason is that the commensuration is not perfect , i.e. the ratio between the pore size and the smectic period is not an integer number . by reducing the size of the pore , the nematization occurs at lower packing fractions . in order to visualize this effect we have depicted a line of constant average uniaxial order parameter , @xmath66 , in the state diagram ( fig . [ fig2 ] ) . this line monotonically increases with @xmath16 and asymptotically tends to the bulk packing fraction at which @xmath66 . therefore , confinement promotes nematic order . this is an expected behaviour because even at low densities the walls induce some homeotropic anchoring . similarly , confinement promotes smectic order as oscillations in the density profile start to appear at lower packing fractions in smaller pores . of the states in a homeotropic cell with @xmath62 . ( top ) scaled density profile @xmath67 , ( middle ) uniaxial order parameter profile @xmath68 and ( bottom ) tilt angle profile @xmath69 . dotted line : isotropic state , @xmath70 . dashed line : nematic state , @xmath71 . solid line : smectic state ( s@xmath72 ) , @xmath73 . snapshots of the particle configurations corresponding to these profiles are depicted in fig . [ fig3 ] . ] the most interesting phenomenology arises when the particles are strongly confined in narrow pores and at high packing fractions . in this regime new states with symmetries different than those of the stable bulk phases appear . * smectic c*. for pores with @xmath74 the particles form a nematic state at intermediate densities . by further increasing the density , the rods align into two well - defined layers as already present in the s@xmath75 state . here , however , the particles are strongly tilted with respect to the direction perpendicular to the layers . we call this the smectic c state , s@xmath76 , where @xmath64 indicates the number of layers . the particle configurations and the order parameter profiles of the s@xmath76 state are depicted in fig . [ fig5 ] panels ( a ) and ( b ) , respectively . the tilt profile ( bottom of panel b ) presents two minima shifted from the location of the the maximum density . the smectic c state appears because the size of the pore is not commensurate with the smectic period and the number of layers is reduced . as a result the particles tilt in order to fill efficiently the available space . by increasing the density , the tilt angle decreases . this is consistent with the fact that the smectic period monotonically decreases with the density . in fig . [ fig5]c ) we plot the tilt angle at contact with the wall , @xmath77 , as a function of @xmath16 . for each packing fraction , above a certain threshold , there is a critical pore size below which the particles start to tilt . the smaller the pore is the more tilted the particles are . increasing the number of particles : @xmath78 , nematic state ( left ) ; @xmath79 , smectic c state ( middle ) ; @xmath80 , smectic c state ( right ) . ( b ) local fields as a function of @xmath12 : density profile ( top ) , uniaxial order parameter ( middle ) , and tilt angle ( bottom ) . the different sets correspond to the states showed in panel ( a ) : @xmath78 , nematic state ( dotted line ) ; @xmath79 , smectic c state ( dashed line ) ; @xmath80 , smectic c state ( solid line ) . ( c ) smectic c tilt angle ( absolute value ) as a function of the pore width for different packing fractions , as indicated . ] we have also found a smectic c state with three layers , s@xmath81 , in a small region of the state diagram around @xmath82 ( see fig . [ fig2 ] ) . this region is significantly smaller than the stability region of the s@xmath83 state , and it appears at higher packing fractions than @xmath84 does . at very high packing fractions the smectic period is sufficiently small such that three non - tilted layers fit inside the capillary and the s@xmath81 state is replaced by a s@xmath85 state . we have not found smectic c states with more than three layers although we can not rule out their existence in regions of the state diagram at very high packing fractions . nevertheless , we are confident that those regions shrink rapidly by increasing the pore size and eventually might cease to exist . this can be understood as follows . there is a minimum @xmath86 and a maximum @xmath87 layer spacing between which the formation of non - tilted smectic layers is stable . consider a capillary with @xmath88 smectic layers inside . as a rough estimate the layers will tilt if there is no sufficient space to accommodate @xmath88 non - tilted layers , i.e. if the condition @xmath89 is satisfied . in addition @xmath90 should hold as well , because otherwise an s@xmath91 state , and not an s@xmath92 state , would be stable . both conditions together roughly set the limits in pore size for a smectic c phase with @xmath88 layers as @xmath93 the above equation also shows that the range in pore size in which the smectic c is stable decreases with increasing @xmath88 . indeed , there is a maximum pore size above which no tilted smectic is expected : @xmath94 which results from taking the equality on both sides of eq . ( [ minmax ] ) . for hard rectangles @xmath95 ( because the particles can not overlap ) . we can express @xmath96 , with @xmath97 being the maximum expansion of the layer spacing for the smectic state in units of @xmath86 . then , using eq . ( [ pinpanpum ] ) , @xmath98 . we did not find sm@xmath99 states for @xmath100 . it implies @xmath101 , which seems to be a reasonable value . . ( a ) snapshot of characteristic configurations : nematic state , @xmath102 ( left ) ; nematic - brush , @xmath103 ( second ) ; brush , @xmath104 ( third ) ; smectic - brush , @xmath105 ( right ) . ( b ) local fields as a function of @xmath106 of the states showed in panel ( a ) : density profile ( top ) , uniaxial order parameter ( middle ) , and tilt angle ( bottom ) . the different sets are : nematic ( dotted line ) , nematic - brush ( red dotted - dashed line ) , brush ( dashed line ) , smectic brush ( solid line ) . ( c ) average of the tilt angle profile in a pore with @xmath107 as a function of the packing fraction . results obtained by increasing and decreasing the number of particles . simulation parameters : @xmath108 , @xmath14 ( black - dashed line ) ; @xmath109 , @xmath14 ( blue - dotted line ) ; @xmath108 , @xmath110 ( red - solid line ) . ] * brush states . * the remaining states in the diagram depicted in fig . [ fig2 ] are the brush nematic b@xmath61 and the brush smectic b@xmath111 , both with @xmath3 homeotropic layers . for pores with size in the vicinity of @xmath112 and high packing fractions there is a region where b@xmath75 and b@xmath113 are stable . in fig . [ fig6 ] we show the corresponding order parameter profiles and characteristic particle configurations . by increasing the density from a stable nematic state , some of the particles located in the middle of the pore rotate by ninety degrees , placing their long axis parallel to the walls ( nematic - brush state ) . a further increase in the number of particles results in the pure brush nematic state b@xmath75 , with one layer of particles with homeotropic anchoring adjacent to each wall . the particles at the center of the cavity are aligned parallel to the walls . to rule out the possibility that this state is an artefact of our method of increasing the number of particles , we have initialized a system with @xmath114 and @xmath112 in a nematic state with all the particles perpendicular to the walls . after an equilibration stage of about @xmath115 mcs the particles formed the brush state . hence we are confident that the brush state is indeed stable . the extent of the central region where the particles are oriented parallel to the walls grows by increasing the density ( see e.g. the tilt angle profile in fig.[fig6 ] , bottom of panel b ) . in the b@xmath75 state those particles in the central region posses orientational but no positional order . however at sufficiently high packing fractions smectic layers occur also at the center . this additional order constitutes the smectic brush state , b@xmath113 . the transitions between the different states is gradual and hence no differences should appear if we e.g. fix the size of the pore and track the order parameters by first increasing and then decreasing the density . this is actually what we have found for all the states discussed throughout of the paper except for the nematic - brush transition : we plot in fig . [ fig6]c the average of the tilt angle profile , @xmath116 as a function of the packing fraction , as obtained by either increasing or decreasing the number of particles . @xmath117 is approximately zero in the nematic state and is different than zero in the brush - nematic state due to the particles aligning perpendicular to the walls . we show in the figure three sets of data corresponding to simulations with different numbers of mcs and different lateral pore sizes . in the three cases there is strong hysteresis , most likely related to a finite size effect because our system can be effectively considered as a one - dimensional system with short range interactions where no first order transitions are expected according to the so - called van hove theorem . nevertheless , the van hove theorem does not apply in the presence of external fields ( see @xcite for a discussion about the exceptions to the van hove theorem ) . ( a , b ) and @xmath118 ( c ) at an average packing fraction @xmath105 . ] brush nematic and smectic states with three layers of particles that are perpendicular to the walls also appear in the region of the state diagram where @xmath119 ( see the labels b@xmath120 and b@xmath121 in fig . [ fig2 ] ) . an example of the b@xmath121 state is shown in fig . the @xmath122 and @xmath123 states are in general not symmetric in @xmath12 . the region of particles aligned parallel to the walls is not located in the middle of the cell . in order to test the stability of the unexpected symmetry breaking of these states , we have initialized the system in a symmetric configuration where two small regions of rods parallel to the walls are placed between layers of particles with homeotropic configuration . the resulting configurations after running more than @xmath124 mcs are shown in panels ( b ) and ( c ) of fig . [ fig7 ] . as we had to initialize the system at very high packing fractions we were not able to recover the asymmetric brush profile shown in ( a ) . however , the states in ( b ) and ( c ) are again asymmetric . they resemble coexistence states between the state represented in ( a ) and its mirror image . hence , we conclude that the symmetric b@xmath120 and b@xmath121 phases are not stable . this could have been anticipated because the symmetric state has four interfaces between parallel and perpendicular rods , whereas the asymmetric one contains only two . asymmetric profiles in symmetric pores have been previously found in three - dimensional mixtures of hard rods @xcite and monocomponent and binary mixtures with soft interactions , see e.g. @xcite . the states shown in ( b ) and ( c ) are probably not stable because they have larger interfaces than the one obtained by gradually increasing @xmath5 ( a ) . in addition in both cases the central layer is distorted , which increases the elastic energy of the system . nevertheless , due to the finite - time simulations and the finite lateral pore size , the system may show a bimodal behaviour oscillating between states ( b ) and ( c ) as if it were a genuine phase transition . as for the smectic c states , the regions in the state diagram where the brush states are stable move to high packing fractions and shrink with the size of the pore . we could not find brush states above b@xmath120 but their existence in narrow regions and high packing fractions can not be ruled out . instead of forming a brush state , the particles could tilt and form a smectic c state reaching very high packing fractions , similar to those in the brush state . hence , an interesting question is : why are there regions of the state diagram where the brush states are more stable than the smectic c ? to answer this question we compare the excess in free energy of both states with respect to non - distorted nematic or smectic states . in the brush state the excess in free energy is dominated by the two interfaces of perpendicularly aligned particles . in the smectic c state there are two important contributions : the violation of the anchoring imposed by the surfaces and the formation of tilted layers . both contributions increase by increasing the tilt angle . the tilt angle in the smectic c state , and hence the excess in free energy , increases by reducing the size of the pore . as a consequence the brush states appear replacing the smectic c when the size of pore is reduced . this simple argument explains not only the appearance of the brush states but also their relative position to the s@xmath99 in the state diagram , cf . [ fig2 ] . the regions in the state diagram ( fig . [ fig2 ] ) where smectic c and brush states become stable possess a smaller average uniaxial order parameter than the surrounding regions . when these states start to form there are regions in the pore where the particles align according to the incipient state and other regions where the particles remain in the nematic state . in addition , the order parameter profiles of the brush states depend on both the vertical and the horizontal coordinates . however we calculate them only as a function of the horizontal coordinate . both effects result in an artificially reduced uniaxial order parameter that , on the other hand , is useful to distinguish the boundaries between states in the state diagram . next we investigate the behaviour of hard rectangles with the same aspect ratio @xmath1 as before , but confined between two parallel walls that promote antagonistic anchoring , the so - called hybrid cell . the left wall induces homeotropic anchoring and the right wall promotes planar alignment of the particles ( see a schematic of the geometry in fig . [ the state diagram is depicted in fig . [ fig8 ] in the plane of packing @xmath21 fraction and scaled - effective pore width @xmath125 . as in the previous case the color map indicates the value of the averaged uniaxial order parameter inside the pore . we have run more than @xmath126 simulations with pore widths @xmath127 to generate the diagram . and the packing fraction @xmath21 for hard rectangles ( @xmath1 ) confined in a hybrid planar cell . the color map represents the value of the averaged uniaxial order parameter @xmath54 . the empty circles connected with a dashed white line show the state points for which @xmath66 . the empty squares connected via a solid white line show the approximate boundary between the linear and the uniform states . ] for any pore width the isotropic state is stable at low densities . in this state there is a small layer of particles oriented perpendicular ( parallel ) to the left ( right ) wall . the remaining particles do not show orientational order . as discussed the anchoring imposed by the planar wall is stronger than that of the center - hard wall . a manifestation of this is the the value of uniaxial order parameter in the isotropic state ( not shown ) , which is higher close to the planar wall than close to the homeotropic one . confinement in a hybrid cell promotes , as in the homeotropic cell , orientational order of the particles ( see e.g. the line of constant uniaxial order parameter in the state diagram ) . first we focus on the regime of large pore sizes . by increasing the number of particles the following sequence of states appears : isotropic ( i ) , step ( st ) , linear ( l ) , uniform nematic ( u ) , and uniform smectic ( u@xmath128 ) . examples of the configuration of the particles and the order parameter profiles in the intermediate and high density states for a pore with @xmath62 are shown in figs . [ fig9 ] and [ fig10 ] , respectively . * step state . * also known as director - exchange phase or biaxial phase , the step phase was proposed by schopohl and sluckin @xcite and by palffy - muhoray et al . it has been study in three - dimensional systems with landau - de gennes theory @xcite , simulation @xcite , and density functional theory @xcite . in the st state there are two nematic regions with uniform and opposite directors following the anchoring imposed by the surfaces ( see fig . [ fig9]a ) . the interface between both regions is sharp ; the director rotates by ninety degrees in a region of about two molecular lengths ( see the tilt profile in fig . [ fig10 ] dotted line ) . at the interface the uniaxial order parameter drops to zero . for large pores the st state is stable in a very narrow region of packing fractions contiguous to the isotropic state . actually , as suggested in @xcite , the st state could be a manifestation of the isotropic state at densities close to the capillary nematization in sufficiently narrow pores . of the states in a hybrid cell with @xmath62 . ( top ) density profile , ( middle ) uniaxial order parameter profile , ( bottom ) tilt angle profile . dotted line : step state , @xmath129 . red dotted - dashed line : linear nematic state , @xmath130 . dashed line : uniform nematic state , @xmath131 . solid line : uniform smectic state , @xmath132 . snapshots of the particle configurations corresponding to these profiles are depicted in fig . [ fig9 ] . ] * linear state . * increasing the packing fraction from the st state in large pores gives rise to the formation of the linear state ( see fig . [ fig9]b ) . here the director rotates continuously from homeotropic to planar anchoring ( see fig . [ fig10 ] , red dotted - dashed line ) . far enough from both substrates the tilt profile varies linearly with the distance across the pore . in this way the elastic free energy is minimized . the formation of the l state is the analogue to the capillary nematization in a symmetric pore . the l state is compatible with the anchoring imposed by both substrates and at the same time minimizes the elastic free energy . * uniform nematic state . * by further increasing the packing fraction there is a configurational change from the linear state to the uniform nematic state . the u state is a nematic with uniform director parallel to the wall except for the first layer of particles adsorbed at the homeotropic wall , where the particles are perpendicular to the substrate ( see figs . [ fig9]c and [ fig10 ] ) . this first layer is most likely a consequence of the peculiarities of the hard center - wall , which allows for a high packing fraction of particles only in the case that rods are aligned perpendicular to the wall . the linear - uniform transition is a consequence of the stronger anchoring induced by the planar wall in comparison to the hard - center wall . although it occurs gradually as we increase the packing fraction , the range in @xmath21 at which the transition occurs is small , enabling us to draw a line in the state diagram that approximately indicates its location ( see fig . [ fig2 ] ) . the density at the l - u configurational change and , therefore , the range in packing fractions at which the l state is stable , increases with the pore width . actually , the l state may replace the uniform nematic state in the regime of very large pores . to understand this , consider the excess in free energy of the l and u states over a bulk undistorted nematic , @xmath133 . in the l state @xmath134 , where @xmath135 and @xmath136 are due to the anchoring imposed by the right and the left walls , respectively , and @xmath137 is the elastic energy due to the deformations of the director field . for very large pores the director varies linearly and rotates by ninety degrees in the pore . hence , the divergence of the director is @xmath138 and the elastic energy @xmath139 , with @xmath140 the splay elastic constant . in the uniform state both anchoring constraints are satisfied and contribute to the excess in free energy as in the linear state . the director is not distorted ( @xmath141 ) but there is an interface generated by the first layer of particles with homeotropic alignment . hence , @xmath142 , where @xmath143 is the free energy of the nematic - nematic interface . the elastic contribution in the l state decreases with @xmath15 but @xmath143 does not . therefore we expect the l state to replace the u state for sufficiently wide pores . note that the same argument is valid if instead of an interface between two nematics with opposite directors in the u state there is a violation of the anchoring imposed by one of the substrates . in that case the anchoring energy in the u state would be higher than that in the l state and would not decrease with the size of the pore . we have performed simulations in a pore with @xmath144 , and the l - u transition occurs at @xmath145 , considerably higher than e.g. the case @xmath62 ( @xmath146 ) . this scenario , in which the l state replaces the u state in very wide pores , is therefore plausible . nevertheless , we can not rule out another scenario in which the u state is stable for any pore width , as it has been found in @xcite . in @xcite a system of spherocylinders confined in a hybrid cell is analyzed with dft , and the lu transition persists at any pore length due to an anchoring transition at one of the substrates , i.e. the type of anchoring induced by the wall changes by varying the density . in our case , however , such an anchoring transition is not expected as we deal with hard core potentials . note , nevertheless , that the first layer of particles adsorbed on the hard - center wall could effectively act as a hard wall for the second layer if the density is sufficiently high , which in practice could be viewed as an anchoring transition . in this second scenario the u state would be stable even for very wide pores . simulations for pores wider than those considered here could help to elucidate this point . * uniform smectic state . * finally at very high packing fractions the particles parallel to the walls form smectic layers . the resulting state is similar to the uniform nematic but with positional ordering . an example of the particle configurations is presented in fig . [ fig9]d . the corresponding order parameter profiles are shown in fig . [ fig10 ] ( solid lines ) . the formation of layers by increasing the density from the u state takes place very gradually and we could not identify the packing fraction of the u - u@xmath128 transition in the state diagram . we find that the direction of the layers is not perpendicular to the walls . the director is tilted with respect to the direction perpendicular to the layers , like it is in a smectic c. we did not find any relation between the tilt angle and the size of the pore . the fact that the layers are tilted could be a finite size effect related to the vertical size of the pore , or it could also be related to high fluctuations in the tilt angle . in contrast to the homeotropic case , the state diagram of the hybrid cell does not show additional states in the regime of small pores . the only significant difference in the region of small pores with respect to the region of large pores is that the linear state disappears . we could not find the linear state in pores with @xmath147 . ) confined in a square cavity with planar anchoring in the packing fraction - side length plane . the color map represents the average of the uniaxial order parameter inside the cavity , @xmath54 . empty circles indicate the position where @xmath66 . empty squares roughly show the boundary between the elastic and the bridge smectic states . lines are guides to the eye . ] we next consider confinement of the rods in a square cavity favouring planar alignment of the particles ( see a sketch of the geometry in fig.[fig1]c ) . confinement in all spatial directions adds additional constraints on the orientational ordering of the particles that might result in e.g. the formation of topological defects . the state diagram in the plane of packing fraction and side length is depicted in fig . [ fig11 ] . we found three distinct states : isotropic ( i ) , elastic ( e ) , and bridge smectic ( br@xmath128 ) . representative results of these states are shown in fig . [ fig12 ] for a cavity with side length @xmath148 . at low densities the isotropic state is stable , here the fluid is disordered . only a thin layer of particles close to the walls shows some degree of orientational order ( see the uniaxial order parameter in fig . [ fig12]a ) . the density ( not shown ) is rather uniform in the whole cavity , showing only a small desorption of particles close to walls , especially near the corners of the cavity . the uniaxial order parameter is also smaller in the vicinity of the corners . the nematization occurs by increasing the number of particles . the result is a gradual transition from the isotropic to the elastic state ( see fig . [ fig12]b ) . in the e state the nematic can not adopt a uniform configuration due to the surfaces and six disclinations arise in the cavity . four disclinations are located in the corners of the cavity . in the middle of the cavity the rods align along one of the diagonals . this leads to the formation of two further disclinations with topological charge @xmath149 located along the other diagonal , at a distance of about @xmath150 from the corners . the disclinations are clearly visible as a drop of the uniaxial order parameter ( see fig . [ fig12]b ) . the density profile ( not shown ) also reveals a depletion of particles close the defect cores . the position of the cores of the @xmath149 defects fluctuates during the simulation but they always stay away from each other . the inner @xmath149 defects are connected with the adjacent corner defects , see the uniaxial order parameter in fig . [ fig12]b ) . the smaller the cavity becomes the stronger this effect is . the packing fraction at which the capillary nematization occurs increases monotonically with the size of the cavity and tends asymptotically to the bulk value ( see for example the line of constant average uniaxial order parameter depicted in the state diagram , fig . [ fig11 ] ) . the average order parameter @xmath54 depends nontrivially on @xmath16 and @xmath21 . in the i region , e.g. , @xmath151 , @xmath54 decreases by increasing @xmath16 because the walls induce order in a small region close to them and the ratio between this region and the whole cavity decreases as @xmath16 is increased . once the nematic is formed the trend is reversed . for instance , at @xmath152 the smaller the cavity becomes the lower @xmath54 is . here the whole cavity is in a nematic state , except in those regions where the disclinations appear , and the ratio between the surface occupied by the disclinations and the whole cavity decreases with @xmath16 . next we focus on the regime of high packing fraction . by increasing @xmath21 from the e state the particles show incipient positional order , forming smectic layers without changing their director field ( not shown ) . then , at higher packing fractions there is a complete structural change to the bridge smectic state ( see fig . [ fig12]c ) . in the br@xmath128 state the particles that were oriented along one diagonal in the e state rotate by @xmath153 generating three domains where the director is almost uniform . the domains are separated by domain walls where the director rotates by @xmath154 ( the uniaxial order parameter vanishes at the domain walls , see fig . [ fig12]c ) . the domain walls connect two corners and divide the cavity in three regions with uniform director . the size of the domains fluctuates but the central domain is always bigger than the others . the domain walls become more rigid as the density is increased . the same state has been predicted recently using density functional theory in a system of rectangles with restricted orientations ( zwanzig approximation ) confined in the same geometry @xcite . the authors of @xcite classify the br@xmath128 state according to the number of smectic layers in the central domain . such a criterion is not applicable in our case due to the large fluctuations of the domain walls , but obviously the number of smectic layers in the cavity varies with the side length . in order to estimate the packing fraction of the e - br@xmath128 transition we have made a histogram of the global tilt angle @xmath155 inside the cavity , i.e. the tilt angle resulting of a diagonalization of the tensorial order parameter formed by all the particles . in the isotropic state @xmath155 fluctuates between @xmath156 and @xmath157 . as soon as the the elastic state arises , @xmath155 fluctuates between the values for both diagonal directions ; the histogram shows two peaks at @xmath158 and @xmath159 . at high densities , but still in the e state , the system stops fluctuating between the diagonals ( during the available simulation time ) and the histogram shows only one peak either at @xmath158 or @xmath159 . finally , in the br@xmath128 state there is a single peak centered at @xmath160 or @xmath161 governed by the particles of the main domain . again at these packing fractions the particles can not fluctuate between both equivalent states with @xmath160 or @xmath161 during the available simulation time . the behaviour of @xmath155 allows us to estimate the e - br@xmath128 transition as the packing fraction at which @xmath155 changes from @xmath158 or @xmath159 to @xmath156 or @xmath161 . the result is plotted in the state diagram , fig . [ fig11 ] . the bigger the cavity is the higher the packing fraction at the e - br@xmath128 transition is . we can rationalize the transition as follows . let @xmath162 be the excess in free energy of the confined system over a bulk undistorted state . in the br@xmath128 state @xmath163 with @xmath164 being the anchoring free energy due to the interaction with the walls and @xmath165 the contribution due to the domain walls . in the e state @xmath166 , with @xmath167 accounting for the elastic deformations of the director field , and @xmath168 for the disclination cores . @xmath164 is similar in both cases because the anchoring is satisfied in both states . @xmath165 is proportional to the length of the domain walls and hence to @xmath16 . @xmath168 does not depend on the size of the cavity and , finally , the elastic energy is @xcite @xmath169,\ ] ] where @xmath170 is the director field and @xmath171 and @xmath172 are the splay and bend elastic constants , respectively . for rods confined in a circular cavity , the elastic energy grows logarithmically with the radius of the cavity @xcite . here , we have computed numerically the divergence and the rotational of the director in the e state and we have found that the dependence of the elastic energy with the cavity size is also weak , increasing slower than linear in @xmath16 . on the other hand , in the br@xmath128 state , the size of the domain walls is proportional to the size of the cavity , and hence @xmath173 . therefore , for a fixed @xmath21 we expect the bridge state to be replaced by the elastic state at sufficiently big cavity sizes due to different dependence of @xmath162 with @xmath16 in both states . the increase of the packing fraction at the e - br@xmath128 can be understood given the behaviour of the elastic constants with the packing fraction ; both @xmath171 and @xmath172 monotonically increase with @xmath21 . in a very recent study @xcite garlea et al . have simulated a quasi - monolayer of hard spherocylinders confined in a square prism as well as the two - dimensional limit of discorectangles in a square cavity . the authors observe a state very similar to the elastic state in which the inner @xmath149 defect and its adjacent corner defect form a kind of line defect . actually , in our case , for the smaller cavities it is difficult to say whether those defects are actually two independent defects or whether they form a single structure . garlea et al . have also found smectic ordering in their simulations , but in contrast to our findings they did not observe domain walls at high packing fractions , although they state in @xcite `` ... we do sometimes observe particles trapped perpendicularly to the smectic layers , invariably next to the wall '' . the differences at high packing fractions between both systems are probably due to the slightly different geometry of the particles ( spherocylinders vs rectangles ) . in contrast to hard spherocylinders ( or discorectangles in two dimensions ) , hard rectangles posses degenerate close packing states and have a higher tendency to cluster @xcite . this may explain the presence of domain walls in a system of hard rectangles and its absence in a system of hard spherocylinders . it is unlikely that the dimensionality plays a dominant role because garlea et al . have studied both the quasi two dimensional system and the strict two - dimensional limit , and found no differences between them . confined in a square cavity with homeotropic anchoring : packing fraction - side length plane . the color map indicates the average of the uniaxial order parameter @xmath54 . empty circles mark the packing fraction at which @xmath66 . empty squares roughly indicate the elastic - bridge nematic transition . lines are guides to the eye . ] finally we investigate the confinement of rods in a square cavity that promotes homeotropic anchoring ( a schematic of the geometry is shown in fig . [ fig1]d ) . the state diagram and representative states for @xmath148 are shown in figs . [ fig13 ] and [ fig14 ] , respectively . here , as in the case of the planar cavity , the elastic state consists of particles aligned along one diagonal ( see fig . [ fig14]b ) . however , in contrast to the planar cell , the alignment of the particles leads to the formation of only two disclinations with topological charge @xmath174 ( see the drop of the uniaxial order parameter in panel b of fig . [ fig14 ] ) . the fluctuations in the position of the disclinations is high , much higher than in the planar cavity . this is most likely related to the dominant elastic deformations of the director involved in each disclination : splay - like deformations in the case of @xmath174 disclinations and bend - like in @xmath149 disclinations . as @xmath175 ( see e.g. , @xcite ) we expect more fluctuations in the positions of @xmath174 disclinations than in @xmath149 disclinations . as an example of the high fluctuations of the @xmath174 disclination cores , we show in fig . [ fig14]a ) a state where both disclinations have merged forming a single @xmath176 disclination . this state is a variation of the elastic state that we observe sometimes , especially at low densities . this configuration is metastable because it involves higher elastic deformations and the energy of one @xmath176 disclination core is higher than that of two @xmath174 disclination cores ( the energy of a disclination core increases with the square of its topological charge ) . by further increasing the density , we find a gradual transition from the elastic to the bridge nematic state ( fig . [ fig14]c ) . in the bridge nematic state there are three domains of particles with uniform director . in contrast to the planar cavity , here the bridge state is not accompanied by positional order because it appears at lower density ( compare the position of the elastic - bridge transition in the states diagrams of figs . [ fig11 ] and [ fig13 ] ) and the particles remain in a nematic state . another difference involves the domain walls that stay always at a distance of about one molecular length from the ( effective ) walls . the position of the domain walls fluctuates less than in the case of a planar cavity . at higher packing fractions the rods in the main domain form well - defined layers . we call this the bridge smectic state , b@xmath111 , where @xmath3 indicates the number of layers of the main domain . the number of smectic layers is well defined due to the stable positions of domain walls in the cell . the number of layers is the result of commensuration between the side length of the cavity and the smectic layer spacing . the approximate regions of the distinct b@xmath111 states are indicated in the state diagram of fig . [ fig12 ] . in summary , we have performed a systematic analysis of the behaviour of two - dimensional hard rods confined in slit pores and in square cavities . in the case of slit pores we have shown that our simple hard core model contains much of the phenomenology observed in corresponding confined three dimensional systems . examples are the capillary nematization and smectization in homeotropic pores , and the formation of linear and step states that occurs in hybrid planar cells . in addition we have found new states that have not been experimentally observed or theoretically predicted . an example is the smectic c and the brush state that we have observed in homeotropic cells . both states break the anchoring imposed by the surfaces . the asymmetric brush state breaks also the symmetry of the cell . in all cases we have rationalized the stability by comparing the excess in free energy to the corresponding undistorted bulk phase . in recent experiments on vertically vibrated monolayers of rods confined in a circular cavity @xcite the same textures were found as mc studies predict for equilibrium rods @xcite . actually , the elastic state we have found in the square planar cavity has been observed in vibrated granular rods @xcite . granular materials flow and diffuse anomalously @xcite . although being non - thermal fluids , under certain circumstances such systems form steady states with the textures of thermal fluids . a comparison between mc simulation of confined rods ( thermal fluid ) and vibrated granular rods ( non - thermal fluid ) would help to find the analogies between both systems . the elastic state of the planar square cell has been also observed experimentally in confined actin filaments @xcite , colloidal particles @xcite , and predicted using onsager - like density functional theory @xcite . the authors of @xcite found , using experiments on confined colloids and oseen frank elastic theory , that the elastic state is metastable with respect to another state that contains two corner disclinations and is free of bulk disclinations ( diagonal state ) . in the diagonal state the total deformation of the director is higher than in the elastic state , but on the other hand , the diagonal state has no bulk disclinations . the total elastic energy decreases with the size of the cavity , and the energetic cost associated to a disclination is independent of cavity size . hence , we expect the diagonal state to replace the elastic state in our system for cavities much bigger than the ones studied here . the rate between the splay and bend elastic constant also plays a role determining the relative stability of the confined states . in @xcite the case of equal elastic constants is analysed , whereas in our system we expect the bend elastic constant to be much higher than the splay one . although we have analysed a two - dimensional model , our results may be of relevance to gaining a better understanding of three - dimensional systems where similar phenomenology has been already found . for example , capillary nematization @xcite and smectization @xcite have been studied in confined rods and platelets between two parallel walls . the hybrid cell has been also analysed in three dimensions @xcite , and phases with the same symmetry as those found here appear . our results indicate that other states , not observed yet in three - dimensional systems , can arise under extreme confinement . for example , states that break the anchoring , like the smectic c or the brush states found here , or states with symmetry breaking ( i.e. , asymmetric states in confined symmetric pores ) such as the asymmetric brush state . those states might be difficult to find in e.g. density functional studies where one typically assumes that the symmetry of the order parameter profiles is the same as the one imposed by the surfaces . it is interesting to compare the confinement of rods in square cavities , secs . [ sp ] and [ sh ] , with the recent study of confined rods in circular cavities @xcite . in both cases at high densities the system form domain walls in an attempt to reduce the elastic distortions of the director field . although the domain walls will probably disappear in larger cavities , they might be a general mechanism to reduce elastic stresses under extreme confinement . some of the states found in the slit - pore geometry show lateral ordering , such as for example the brush smectic states . for selected pore sizes , we have performed simulations varying the lateral size of the cell , @xmath177 from @xmath178 to @xmath179 and no differences have been found . we are , therefore , confident that the lateral ordering is not induced by the applied boundary conditions . nevertheless , monte carlo simulations in the isothermal isobaric ensemble ( npt ) might help to elucidate the role that the lateral size of the pore plays in the stability of such states . we have restricted the analysis to hard rectangles with length - to - width ratio of @xmath180 . we expect a similar phenomenology for particles with aspect ratio higher than @xmath181 because the bulk behaviour is qualitatively the same . however , for particles with shorter aspect ratios completely new phenomenology will presumably appear because states with tetratic correlations are stable in bulk @xcite and might modify the phase behaviour presented here . we thank e. velasco for useful discussions . + t. geigenfeind and s. rosenzweig contributed equally to this work . 58ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1080/002689797170716 [ * * , ( ) ] link:\doibase 10.1103/physreve.63.051703 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.94.017801 [ * * , ( ) ] link:\doibase 10.1103/physreve.74.011709 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1080/02678299408036543 [ * * , ( ) ] link:\doibase 10.1080/026782997207731 [ * * , ( ) ] link:\doibase 10.1103/physreve.79.011712 [ * * , ( ) ] \doibase 10.1051/jp2:1992133 [ * * , ( ) ] link:\doibase 10.1103/physreve.62.5081 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.101.037802 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.108.057801 [ * * , ( ) ] link:\doibase 10.1039/c3sm52394b [ * * , ( ) ] link:\doibase 10.1080/002689798168790 [ * * , ( ) ] link:\doibase 10.1063/1.1849159 [ * * , ( ) ] link:\doibase 10.1103/physreve.80.011707 [ * * , ( ) ] link:\doibase 10.1103/physreve.77.011707 [ * * , ( ) ] link:\doibase 10.1103/physreve.75.051708 [ * * , ( ) ] link:\doibase 10.1103/physreve.88.032506 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevlett.105.168001 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1126/science.1140414 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.109.108303 [ * * , ( ) ] link:\doibase 10.1039/c1sm06060k [ * * , ( ) ] link:\doibase 10.1039/c4sm01123f [ * * , ( ) ] link:\doibase 10.1039/c1sm05602f [ * * , ( ) ] @noop * * , ( ) \doibase http://dx.doi.org/10.1063/1.4867790 [ * * , ( ) ] link:\doibase 10.1080/00268979909483083 [ * * , ( ) ] link:\doibase 10.1103/physreve.79.061703 [ * * , ( ) ] link:\doibase 10.1063/1.1646374 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1063/1.481637 [ * * , ( ) ] link:\doibase 10.1103/physreva.42.2126 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.1588994 [ * * , ( ) ] \doibase 10.1023/b : joss.0000022373.63640.4e [ * * , ( ) ] link:\doibase 10.1039/c003448 g [ * * , ( ) ] link:\doibase 10.1103/physrevlett.59.98 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.2715934 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.2895747 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/j.cis.2010.01.008 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.59.2582 [ * * , ( ) ] link:\doibase 10.1103/physreve.60.1821 [ * * , ( ) ] link:\doibase 10.1103/physreve.79.011709 [ * * , ( ) ] link:\doibase 10.1103/physreve.83.041701 [ * * , ( ) ] link:\doibase 10.1103/physreve.64.051704 [ * * , ( ) ] link:\doibase 10.1039/df9582500019 [ * * , ( ) ] link:\doibase 10.1039/c4sm02087a [ ( ) ] link:\doibase 10.1063/1.2209000 [ * * , ( ) ] link:\doibase 10.1080/02678290903362840 [ * * , ( ) ] http://link.aps.org/doi/10.1103/physrevlett.96.028002 [ * * , ( ) ] link:\doibase 10.1140/epje / i2012 - 12104 - 0 [ * * , ( ) ] link:\doibase 10.1103/physreve.75.051301 [ * * , ( ) ] link:\doibase 10.1039/c3sm51991k [ ( ) ] @noop * * , ( ) \doibase http://dx.doi.org/10.1063/1.1849159 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.2209000 [ * * , ( ) ]

by comparing the orbital period distributions of black hole and neutron star low mass x - ray binaries ( lmxbs ) in the ritter - kolb catalogue we show that there is statistical evidence for a dearth of black hole systems at short orbital periods ( h ) . this could either be due to a true divergence in orbital period distributions of these two types of system , or to black hole lmxbs being preferentially hidden from view at short orbital periods . we explore the latter possibility , by investigating whether black hole lmxbs could be concealed by a switch to radiatively inefficient accretion at low luminosities . the peak luminosity and the duration of x - ray binary outbursts are related to the disc radius and , hence , the orbital period . at short periods , where the peak outburst luminosity drops close to the threshold for radiatively inefficient accretion , black hole lmxbs have lower outburst luminosities , shorter outburst durations and lower x - ray duty cycles than comparable neutron star systems . these factors can combine to severely reduce the detection probability of short period black hole lmxbs relative to those containing neutron stars . we estimate the outburst properties and orbital period distribution of black hole lmxbs using two models of the transition to radiatively inefficient accretion : an instantaneous drop in accretion efficiency ( ) to zero , at a fraction ( ) of the eddington luminosity ( ) and a power - law efficiency decrease , , for l l . we show that a population of black hole lmxbs at short orbital periods can only be hidden by a sharp drop in efficiency , either instantaneous or for . this could be achieved by a genuine drop in luminosity or through abrupt spectral changes that shift the accretion power out of a given x - ray band . [ firstpage ] stars : binaries x - rays : binaries black hole physics accretion accretion discs

1310.8303

low mass x - ray binaries ( lmxbs ) , comprising a black hole or neutron star primary ( @xmath8 ) and a low mass main sequence or evolved secondary ( @xmath9 ) , produce x - rays through the accretion of matter onto their primaries ( see reviews by e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? all black hole and many neutron star lmxbs are transient , experiencing occasional outbursts during which their mass accretion rate and x - ray luminosity increase by several orders of magnitude . these outbursts have typical durations spanning days to months . the cycle of outburst and quiescence in lmxbs is explained by the disc instability model ( see e.g. the review by * ? ? ? mass transferred from the secondary builds up in an accretion disc around the primary . as the disc density increases , the temperature rises to the point at which hydrogen ionizes . the change in ionization state is associated with a switch to higher viscosity , increasing mass accretion onto the primary and causing an outburst . this enhanced accretion eventually lowers the disc density and temperature to the point at which the viscosity returns to its original value . unlike cataclysmic variables , where outbursts are terminated by a cooling wave after a few days , the outbursts of lmxbs are prolonged by x - ray irradiation , which heats the disc and traps it in the hot , ionized state @xcite . in sufficiently short period systems the whole disc is irradiated and the majority of the disc mass is accreted . the accretion rate during an outburst is known to decay exponentially ( e.g. * ? ? ? * ; * ? ? ? * ) , unless the disk is entirely irradiated ( see section 3.2 ) . it was characterised by @xcite as : @xmath10 here @xmath11 is the disc density and @xmath12 is the viscosity , where @xmath13 is the hot state viscosity parameter , @xmath14 is the sound speed , and h is the vertical scale height of the disk . the maximum extent of the disc , @xmath15 , is limited by the size of the roche lobe of the primary and can be related to the orbital period ( @xmath16 ) . for typical mass ratios ( @xmath17 ) , @xmath18 cm , where @xmath19 is the orbital period in hours and @xmath20 is the primary mass in solar masses . both the peak accretion rate ( @xmath21 ) and outburst duration ( @xmath22 , where @xmath23 is the disc mass ; c.f . section 3 ) depend inherently on the orbital period . the radiative efficiency of accretion in an lmxb is given by @xmath24 , where @xmath25 for a radiatively efficient flow through a thin accretion disc ( see e.g , * ? ? ? * ) . at low accretion rates cooling becomes inefficient and a radiatively inefficient , advection dominated accretion flow ( adaf ) can occur @xcite . the adaf model is a solution to the hydrodynamical equations of viscous differentially rotating flows with low sub - eddington accretion rates . the accreting gas has a very low density , leading to an optically thin flow which can not cool efficiently within an accretion time . the viscous energy is stored in the gas as thermal energy rather than being radiated away , and is advected onto the central compact object . the transition from radiatively efficient to inefficient flow is expected to take place once the accretion luminosity reaches a few percent of the eddington luminosity ( @xmath3 ) . for black hole primaries , which lack a hard surface , @xmath26 and the accretion energy is carried with the mass flow into the hole or transferred elsewhere , e.g as radio jets or mechanical outflows . neutron star primaries do not experience a drop in accretion efficiency because the advected energy must always be radiated from the stellar surface . hence , at sufficiently short periods ( and thus small disc radii and peak accretion rates ) black hole lmxbs will undergo fainter , shorter outbursts than comparable neutron star systems , making them more difficult to detect . the relationship between orbital period and peak outburst luminosity for lmxbs is well established ( e.g. * ? ? ? * ; * ? ? ? @xcite studied the outburst luminosities of a sample of transient lmxbs observed by the rossi x - ray timing explorer ( _ rxte _ ) . they showed that there was no distinguishable difference in the orbital period - peak outburst luminosity relation of black holes compared to neutron star primaries , when luminosities were measured in eddington units . however , they suggest that the two populations may diverge at short periods . along similar lines , note that the low peak outburst luminosities in short period black hole lmxbs can cause them to remain in a low luminosity - hard spectral state , rather than entering the high luminosity - soft state expected for radiatively efficient accretion . this breed of outburst , where sources remain in the low - hard state throughout , has been observed in several sources @xcite in addition to establishing a relation between orbital period and peak outburst luminosity , @xcite point out that the absence , in their dataset , of black holes with orbital periods of @xmath27hr , may be caused by radiatively inefficient accretion lowering the peak outburst luminosities of these systems . in this paper , we investigate this hypothesis , using the ritter - kolb catalogue @xcite to show that there is statistical evidence for a dearth of black holes in lxmbs at short orbital periods . we suggest that this is caused by the increasing importance of radiatively inefficient accretion in black hole systems lowering not only peak outburst luminosities but also outburst durations and x - ray duty cycles . additionally , we investigate the nature of the transition to radiatively inefficient accretion , modelling it as an instantaneous change to @xmath1 = 0 at a fraction @xmath2 of the eddington luminosity ( @xmath3 ) and as a power law decrease , @xmath4 , below @xmath28 . in section 2 we present the sample of systems and the statistical evidence for the lack of black hole lmxbs at short orbital periods . in section 3 we study the effect of a radiative efficiency switch on the peak luminosities and outburst timescales of black hole lmxbs . we determine the conditions under which these effects can hide a short orbital period black hole population in section 4 . this is followed by a discussion of our findings and our conclusions . [ cols="<,^,^,^",options="header " , ] [ tab : ad ] if a switch to inefficient accretion in black hole lmxbs at low luminosities reduces their outburst luminosities and timescales , then black holes will be more difficult to detect than neutron stars with comparable accretion rates . such an effect may be observed as a divergence in the populations of black hole systems compared to those of neutron stars at short orbital periods . we searched for this within the catalogue of cataclysmic variables , lmxbs and related objects compiled by ritter and kolb @xcite , by selecting 3 lmxb samples . * sample 1 : all black hole lmxbs , excluding any unconfirmed candidate systems . ( 17 systems ) * sample 2 : all neutron star lmxbs , both transient and persistent , but excluding any globular cluster systems ( which are likely to have very different evolutionary histories to galactic binaries ) . ( 51 systems ) * sample 3 : only the transient neutron star lmxbs from sample 2 ( 25 systems ) while we see no significant difference in the period distribution of transient and persistent lmxbs ( the anderson - darling test of the two distributions gives a 0.60 probability that they are the same ) , we compare both samples 2 and 3 with the black hole population as a control to test for sampling biases caused by the lack of persistent black hole sources . in all samples , extragalactic sources were removed . in addition , we removed the following systems which are mislabelled in the catalogue : * globular cluster sources ( not labeled gc in catalogue ) : j1748 - 2446 @xcite , j1748 - 2021 # 1 @xcite , j1748 - 2021 # 2 @xcite , ngc 104-x-5 @xcite , j1623 - 2631 @xcite , ngc 104-x7 @xcite , j1910 - 5959 # 1 @xcite and ngc 104-w37 @xcite , * black hole candidates ( not labeled bh ? in catalogue ) : v1408 @xcite , v4134 sgr @xcite j1242 + 3232 @xcite , j1752 - 0127 @xcite in the upper panel of figure [ fig : histo ] we plot the normalised histogram of the log[@xmath16 ] distribution of samples 1 , 2 and 3 . there is no discernible difference between the two neutron star samples , but the black hole sample shows a distinct lack of sources with orbital periods below @xmath30 days . there are in fact 25 ( 12 ) neutron star lmxbs in sample 2 ( 3 ) with orbital periods below the minimum black hole period of 0.17 days ( 4.1 hr ) . the cumulative histogram in the bottom panel of figure [ fig : histo ] further highlights this difference . the vertical dashed line in figure [ fig : histo ] separates sources with orbital periods above and below 80 minutes . below this ( conservative ) limit , the mass - period relation requires that these systems have either sub - stellar or compact secondaries @xcite . we compare the black hole sample to the ns sample including and excluding sources below this limit , as a check for differences caused by system evolution . we use the anderson - darling test to determine whether the populations of the black hole and neutron star lmxbs are different . the a - d test is similar in purpose to the more familiar kolmogorov - smirnov test , and can be used to test whether two samples are consistent with being drawn from the same parent distribution . however , the a - d has better sensitivity to differences between distributions , especially near their tails , compared with the k - s test @xcite.the a - d test results are shown in table [ tab : ad ] . the 4 rows represent comparisons of the black hole sample with ( a ) sample 2 , ( b ) sample 2 for @xmath31 min , ( c ) sample 3 and ( d ) sample 3 for @xmath31 min . the a - d @xmath32-values are shown in the left hand columns ; these correspond to the probability of observing a difference as ( or more ) extreme than actually observed , on the assumption the two parent populations are the same . in addition to the @xmath32-values for the full samples we show those for sources with orbital periods above and below @xmath33 days ; this breaks the black hole population roughly in half ( 8 and 9 sources respectively ) . results with @xmath34 are highlighted in bold . these results suggest that all 4 neutron star samples have different orbital period distributions to the black hole sample across the full period range . this holds when we compare systems with @xmath35 days . however , if we remove all sources with periods @xmath36 0.5 days , we find no significant difference in the period distributions . we conclude that the black hole and neutron star samples differ at the short end of the orbital period distribution . we tested the robustness of this conclusion by adding / removing a few short period systems from each sample . in order to reduce the difference between the samples to the point where the a - d test reports no statistically significant result ( i.e. @xmath37 ) , we need to remove the eight shortest period ns systems ( from the ns transient sample ) or include four additional bh systems with periods @xmath38d . consequently , the make up of the samples has to change quite considerably in order to influence this result . at high luminosities , a radiative efficiency of @xmath39 is a reasonable approximation to black hole and neutron star lmxb outbursts . however , below a fraction @xmath2 of @xmath3 , accretion onto black holes becomes radiatively inefficient and @xmath26 . the efficacy of this switch to radiatively inefficient accretion , in hiding a population of short period black hole lmxbs , depends on the nature of the transition . here , we investigate two possible forms : a : : a sharp , instantaneous switch to @xmath40 at @xmath41 . b : : a gradual reduction in @xmath1 , where below @xmath41 : @xmath42 based on theory by @xcite . we use @xmath43 in subsequent sections , as used in @xcite , but discuss larger values of @xmath44 in section 5.2 . a reduction in @xmath1 at low luminosities will reduce the observed duration of an outburst and , in the case of short orbital period systems , may render the entire outburst undetectable . in the early stages of lmxb outbursts , the accretion rate is known to decay exponentially , following ( [ eqn : mdot ] ) . we approximate the disc radius as @xmath45 , where @xmath46 is the roche radius of the primary : @xmath47 @xcite , and @xmath48 is the binary separation , given by : @xmath49 the secondary mass is fixed at @xmath50 , based on typical values in the ritter - kolb catalogue ( see section 4 for further discussion of secondary masses ) . following @xcite , we take @xmath11 to be @xmath51 g @xmath52 . they show that @xmath11 is independent of radius , meaning that this value is suitable for short period systems . for neutron stars , we use a value of @xmath53 as @xmath11 depends on the surface density ( see equation [ eqn : sigma ] ) which is proportional to @xmath54 . the viscosity is taken as @xmath55 where @xmath56 0.1 , @xmath57 km s@xmath58 with t , the hydrogen ionisation temperature , @xmath59 6500k . the alpha prescription above suggests that @xmath60 is linearly proportional to @xmath15 . however , observationally , there is significant uncertainty in its value and its dependence on @xmath15 . @xcite choose a value of @xmath61 @xmath62s@xmath58 for a disc radius of @xmath63 cm . at the orbital periods @xmath64 days , @xmath15 @xmath65 cm . adopting this as a typical radius , we use @xmath66 @xmath67 s@xmath58 . this choice , while consistent with earlier work , also yields peak luminosities in reasonable agreement with the observational results in @xcite : @xmath68 ( a few @xmath69 ) erg s@xmath58 for periods of a few hours . at t = 0 , the mass accretion rate , @xmath70 , so the peak luminosity is @xmath71 . since we wish to compare our estimates to x - ray observations , we convert this bolometric luminosity into the 2 - 10 kev value , by introducing a correction factor , @xmath72 , such that @xmath73 . for both neutron star and black hole systems , we calculated a simple spectrum , from an accretion disc of specified inner / peak temperature absorbed by a column density of 5@xmath7410@xmath75 @xmath76 ; a typical level of galactic absorption to an x - ray binary . we used this to compute the ratio of bolometric flux ( 0.002 - 30 kev ) after removing the effect of the absorption , to the flux in the 2 - 10 kev band including absorption . for this calculation we used xspec v12.8 @xcite . in the neutron star case , we used a tbabs absorption model @xcite , whereas we modelled the black hole spectrum with a diskbb model , since , in outburst , its spectrum is dominated by the disc and needs to be treated as a multi - temperature blackbody . a black hole lmxb in outburst has an inner disc temperature of 0.5 - 1.0 kev , giving @xmath77 . for neutron star lmxbs , with higher disk temperatures and thermal emission from their surface and boundary layer , @xmath78 . based on these estimates we use a correction factor of 4 for black holes ( corresponding to a temperature of @xmath79 kev ) , and 1.3 for neutron stars . the uppermost panels of figure [ fig : varyf ] show the peak luminosities ( @xmath80 ) as a function of orbital period for black holes of mass @xmath81 using three different values of @xmath2 : 0.01 , 0.03 and 0.05 , plotted as dark blue dotted , purple dotted and light blue dot - dashed lines respectively . both a neutron star ( solid red line ) and a black hole system ( green dashed line ) with ( @xmath82 ) are shown for comparison on all panels . our choice of 8@xmath83 is based on the result of @xcite , who find a narrow black hole mass distribution at @xmath84 @xmath85 case a , where @xmath1 drops sharply to 0 at @xmath86 , is plotted in the left panel . here , @xmath87 0.03 produces a cut - off in the peak outburst luminosity at orbital periods below @xmath30 days , matching the observed cut - off of low - orbital period black hole systems in the observational data . the gradient before the cut - off is @xmath88 , since @xmath89 . the gradients are consistent with the observed correlation in @xcite . the different normalisations of the neutron star and black hole @xmath80 - @xmath16 relations are due to our alternate choices of @xmath72 for the two types of system . case b , with a power - law decrease in @xmath1 below @xmath90 , with @xmath44 = 1 , is plotted in the righthand panel . in this case , the peak luminosity - orbital period relation steepens at short orbital periods when the transition to radiatively inefficient accretion sets in . here , the factor of @xmath91 in @xmath1 means that @xmath92 @xmath93 . while a gradual switch to radiatively inefficient accretion causes the peak luminosities of black hole systems to fall more rapidly at short periods , it does not produce large differences between the two types of system at periods @xmath64 days , unless @xmath94 . in this case , the black hole and neutron star gradients diverge below 0.3 days . observations show a similar gradient and normalisation of peak luminosities down to @xmath64 days , although the predicted effect could be hidden . since our simple analytic models only track emission from the disk , we note that the sharp change in luminosity at @xmath90 does not necessarily produce an equally strong observable signature . sources with peak luminosities below the efficiency threshold are likely to exhibit low - hard outbursts , with power - law spectra that are not dominated by the disk blackbody . in this case , our choice of spectrum , and hence , @xmath72 is no longer valid and it is likely to be larger by factor of @xmath95 . this change in @xmath72 may go some way towards shielding the drop in disk luminosity from the overall bolometric luminosity , which does not change dramatically as the disk moves into a hard state @xcite from a time @xmath96 onwards , after the initial exponential fall off , @xcite show that the mass infall rate follows a linear decay , obeying : @xmath97\ ] ] where @xmath96 is the time taken for the irradiated radius to drop below the disc radius , and is given by : @xmath98 and @xmath99 is the irradiated mass at time @xmath96 : @xmath100 @xmath101 is defined by @xmath102 where @xmath103 is the irradiation radius of the disc . the value of @xmath101 depends on @xmath104 ; 1 for neutron stars with a hard surface and 2 for black holes without . in our simple calculations we use a value of @xmath105 for both systems . we note that the uncertainties in our chosen parameters affect the exact results of our calculations , but not the main conclusions of our study ; these are addressed in section 5.1 . the outburst ends when @xmath106 ; with @xmath107 . however , our observations are flux limited , and therefore the outburst appears to end earlier , once its signal drops below a limiting flux , @xmath108 . we define the observable outburst duration , @xmath109 to be the time at which the luminosity drops below @xmath110 . since the stellar density is highest at the galactic centre , it is reasonable to assume that this is the location of the highest concentration of lmxbs , so that d @xmath111 8kpc . using the daily exposure sensitivity of the _ rxte _ , we define @xmath108 as @xmath112 mcrab , and calculate @xmath113 to be @xmath114 @xmath115 . our use of the _ rxte _ survey to define @xmath108 is explained in section 4.1 . in the middle panels of figure [ fig : varyf ] we plot the predicted observerable outburst durations . for case a , outbursts are terminated abruptly when the luminosity falls below either @xmath113 or @xmath28 . as well as the sharp cut - offs corresponding to the peak luminosity graphs in the top panels , we also see the timescales diverging away from the expected duration by an increasing factor as the orbital period drops . for case b transitions , a decrease in outburst duration is only evident at orbital periods @xmath116 days . here a more gradual divergence from the expected outburst timescales occurs . at the minimum observed black hole orbital period of @xmath59 0.1 days , the expected duration of a black hole outburst is approximately halved for @xmath117 . however , it is still longer than the duration of a neutron star outburst , and should not significantly affect detectability . the relative observable outburst duration of black hole and neutron stars is affected by their distance from us , which determines the minimum luminosity , @xmath113 at which they can be detected . this is illustrated in figure [ fig : vary_dist ] , where the observable timescales ( @xmath109 ) are plotted for a black hole lmxb ( green dashed line ) and a neutron star lmxb ( solid red line ) as a function of orbital period , at distances of 5 , 10 and 15 kpc in the upper , middle and lower panels respectively . these correspond to limiting luminosities of @xmath118 , @xmath119 and @xmath120 erg / s respectively . at orbital periods greater than @xmath59 0.2 days , the outburst duration for black hole systems is longer than that for neutron star systems in all three cases . however , at low orbital periods , neutron star lmxbs may be visible for longer . this is because neutron stars have larger observable peak luminosities ( smaller @xmath72 ) . over time , the neutron star luminosity falls more rapidly than the black hole case ( neutron stars have a shorter e - folding times as they have smaller disk radii ) . if the outburst luminosity of a neutron star lmxb at a given orbital period drops below @xmath113 before it drops below the outburst luminosity of an equivalent black hole system , then the neutron star system will be observed for longer . in the lower panels of figure [ fig : varyf ] we estimate the outburst duty cycle , which we define as @xmath121 , where @xmath122 and @xmath123 are the outburst and quiescent timescales respectively . the quiescent timescale is estimated as @xmath124 , where @xmath125 is the mass transfer rate from the secondary . we take @xmath23 , the disc mass , as : @xmath126 @xcite give @xmath127 as : @xmath128 where @xmath129 is the cold state viscosity parameter . using ( [ eqn : sigma ] ) and integrating we obtain : @xmath130 using the entire disc mass in the calculation of the quiescent timescale is a good approximation at short orbital periods , where x - ray irradiation is expected to result in the accretion of all or most of the disc . the rate at which the disc is replenished with mass after an outburst is given by @xmath131 , which is calculated using the formulae of @xcite : @xmath132 the equations above represent mass transfer driven by gravitational radiation ( @xmath133 ) and magnetic breaking ( otherwise ) . it is clear from the lower panels of figure [ fig : varyf ] that in case a the x - ray duty cycle of black hole lmxbs @xmath134 at periods below 0.1 day for all @xmath135 . in case b the duty cycle is also noticeably reduced at short orbital periods . in section 4.3.3 . , we compare this model to one with a constant @xmath125 and find no differences in our conclusions . we can test the effect of the switch to adaf in hiding a population of short orbital period black hole systems by predicting the relative detection probabilities of neutron star to black hole lmxbs as a function of orbital period . the all sky monitor on board _ rxte _ constantly surveyed the sky from march 1996 until january 2012 and discovered many of the known lmxbs in our galaxy @xcite . the instrument cycled through a series of @xmath136 s dwells on different areas of the sky in a stochastic pattern , so that a randomly chosen source was scanned typically 5 - 10 times per day @xcite . the limiting flux , below which a source is undetectable , depends on the instrument used and the exposure time . an all - sky survey is more likely to see a transient object than an instrument with a narrow field of view . for this reason , we use the _ rxte _ for all the observational comparisons in this study . the approximate daily sensitivity of the asm is @xmath112 mcrab ( where @xmath137 @xmath138 @xmath139 @xmath140 ) . we therefore assume an outburst is detectable above @xmath108 = @xmath112 mcrab , and that the observed outburst duration is @xmath141 . for a transient source to have been detected by the _ rxte _ , it must have had at least one outburst during the lifetime of the survey . this depends on the recurrence time @xmath142 = @xmath123 and the length of the survey ( @xmath143 = 15 years ) , and can be estimated as : @xmath144 provided an outburst occurred during the survey time , it must have been bright enough , and have lasted long enough , for the survey to have detected it . for an outburst to have been unambiguously detected it must have been visible for at least 1 day ( private communication , a. levine @xmath145 r. remillard ) . if @xmath146 day the likelihood of it being observed ( @xmath147 ) decreases with the decay timescale . we therefore parameterise @xmath147 as : @xmath148 the total probability of having an outburst observed by the asm is therefore : @xmath149 and depends on both the mass of the system ( i.e. whether it is a black hole or neutron star ) and on its orbital period . a further constraint on objects forming part of our analysis is whether they have been followed up optically to measure orbital parameters and identify the nature of the primary . we make the assumption here that the fraction of candidate objects which are currently unidentified or poorly studied is independent of the type of primary and its orbital period . in this case , this additional factor will not affect our results . initially we make our detection probability calculations for lmxbs located at the galactic centre , approximately 8 kpc away , as assumed in section 3 . in this case , the outburst timescales @xmath150 correspond to those in the middle panels of figure [ fig : varyf ] . for orbital periods where @xmath151 @xmath152 , @xmath153 and @xmath154 . in the upper panel of figure [ fig : p_outburst ] , outburst probabilities for black hole and neutron star lmxbs are plotted as a function of orbital period , using solid red and green dashed lines respectively . @xmath155 is a function of @xmath156 which is plotted in the panel below . @xmath155 drops below 1 only when @xmath156 increases above @xmath143 ( blue dashed line ) . the discontinuity in @xmath156 at 2 hours marks the switch in mass transfer process from magnetic breaking to gravitational radiation in ( [ eqn : mdot2 ] ) . above this value , @xmath157 , and below , @xmath158 . although the discontinuity in @xmath156 produces a negative spike in the black hole outburst probability curve , this feature does not affect our results ( see section 4.3.3 for a discussion . ) in figure [ fig : nosim ] ( upper panels ) we plot the relative detection probabilities with @xmath159 ( left panel ) and @xmath160 ( right panel ) . the solid red line shows the ratio of black hole to neutron star detection probabilities without a switch to adaf , and the green dashed and dark blue dotted lines show the results for sharp and smooth switches ( case a and case b ) respectively . at large orbital periods , when @xmath153 , the ratios plotted are simply the ratios of @xmath155 shown in the upper panels of figure [ fig : p_outburst ] . when there is no efficiency switch , the black hole detection probability drops to zero at @xmath59 0.01 days , since @xmath146 day . when an efficiency switch is included , this drop in detection probability ratios occurs at a higher orbital periods , of @xmath59 0.05 ( 0.07 ) days with a smooth switch and @xmath59 0.2 ( 0.3 ) days with a sharp switch when f = 0.03 ( 0.05 ) . we can now use this detection probability approximation to estimate how the switch to radiatively inefficient accretion affects the orbital period distributions of black hole lmxbs . by convolving the relative detection probability with the observed period distribution of neutron star lmxbs we can predict an expected distribution for black hole systems . we use sample 2 , weighting each neutron star lmxb by the relative detection probability at that orbital period , assuming the population distribution of periods is the same as that observed for neutron stars . these predicted distributions are shown in the bottom panels of figure [ fig : nosim ] alongside the observed distributions for black hole and neutron star lmxbs from figure [ fig : histo ] . it is clear that the case a efficiency switch is able to reproduce the absence of black hole lmxbs below 0.1 days when f = 0.03 , as well as to some extent , predicting the shape of the black hole orbital period distribution . as a simple confirmation , we use these results to estimate the number of black holes expected to have orbital periods below 0.17 days , in a sample the same size as sample 1 ( 17 sources ) . we find it to be zero for a sharp switch , provided f @xmath161 0.03 . it is particularly interesting that this transition lies in the expected range of a few percent of @xmath3 , as any other reason for the relative population difference would give arbitrary values of @xmath2 . switch b does not produce a significant difference between the black hole and neutron star orbital period distributions , predicting 7 ( 6 ) black holes below 0.17 days , for f = 0.03 ( 0.05 ) . if all lmxbs are 8 kpc from the earth , @xmath147 is either 0 or 1 , depending on their orbital period . in our galaxy however , the probability of a source being detected while in outburst depends on its distance , since closer sources will be visible for longer . therefore , when observing the lmxbs in our galaxy , @xmath147 may range between 0 and 1 , depending on the fraction of sources which are detectable . to address this issue , we now repeat our calculation using a galactic distribution of lmxbs . the milky way stellar population comprises a disk , bulge and spheroidal component . studied the population of lmxbs in our galaxy , and modelled them using these three components , and found a ratio of 2:1:0.8 of the mass of lmxbs in the disk : bulge : spheroid respectively . the spheroidal component takes account of globular cluster lmxbs , which are not part of our study , so we do not include this in our calculations . we distribute lmxbs in a disk : @xmath162 where @xmath163 and a bulge , following the parameterisation of @xcite @xmath164 where @xmath165 we use parameters of @xmath166 = 6.5kpc , @xmath167 = 3.5kpc , @xmath168 = 0.41 kpc , @xmath169 = 1.9kpc , @xmath170 = 0.1kpc , @xmath171 kpc and @xmath172 0.6 kpc . we normalise both distributions by containing them within 16 kpc , the outer edge of our galaxy , and by requiring a ratio of 2:1 of binaries in the galactic disk , and bulge , assuming direct proportionally between number ratios and the mass ratios calculated in on the left panel of figure [ fig : sim ] we plot this distribution , with 500 binaries , where @xmath173 , and @xmath174 are the cartesian coordinates in the plane of the galaxy , and on the right panel this is converted into a normalised histogram of their distance from the earth ( located at @xmath175 , @xmath176 ) the vertical lines on this plot represent the distances out to which black hole lmxbs with orbital periods of 0.01 ( green dashed line ) and 0.1 ( purple dotted line ) days are guaranteed to be observed by _ rxte _ when in outburst , i.e. @xmath153 . these two lines bracket a significant proportion of the distribution , implying that short orbital period systems in this distance range are unlikely to be observable . in particular it is clear that an efficiency switch , shortening the outburst time , may significantly affect the likelihood of observing short orbital period systems . in order to recreate the detection probabilities and predictions of figure [ fig : nosim ] using a distribution in lmxb distances , we run a monte carlo simulation of lmxbs in the galaxy . we create 200 000 lmxbs ( 100 000 each of black hole and neutron stars ) with the same orbital period . these systems are distributed in 3 dimensions over the galactic disc and bulge by randomly selecting @xmath177 , @xmath178 and @xmath179 so that they follow the distributions in ( [ eqn : disk_distribution ] ) and ( [ eqn : bulge_distribution ] ) respectively . the detection probability is then evaluated for each system and averaged to give an overall detection probability of lmxbs at that orbital period . we run models for orbital periods ranging from 0.01 to 100 days , until a convergence of @xmath180 has been reached . the results are plotted in figure [ fig : mcsim ] , in the same style as figure [ fig : nosim ] . additionally , in the upper panels , we plot the detection probability ratios with no efficiency switch for disk lmxbs only , as a faint red line . this relation differs from the equivalent ratios plotted in figure [ fig : nosim ] ( solid red line ) , because @xmath147 now takes fractional values . below @xmath59 0.2 days , the most distant black hole lmxbs in the galaxy become undetectable , because they have lower peak luminosities than neutron stars ( cf . figure [ fig : varyf ] ) . this causes the black hole lmxb detection probabilities , and hence the relative detection probabilities , to decrease . however , below @xmath59 0.04 days , the ratios begin to increase again . this is because the neutron star detection probability ratio drops more steeply than the black hole one . this effect is visible in figure [ fig : vary_dist ] . from 15kpc ( bottom panel ) , to 5 kpc ( top panel ) , the observable outburst timescale for short orbital period systems increases for both neutron star and black hole lmxbs , as the limiting luminosity becomes lower . however , the rate of increase for black holes is more significant . therefore , as systems out to greater distances , like 15kpc and 10 kpc become undetectable , the black hole lmxbs become more observable than neutron star systems . the same shape can be seen in the disk + bulge result ( solid red line ) , but here , another feature is present ; a sharp decrease in the relative detection probabilities at an orbital period of @xmath181 days . this is due to the black hole bulge population becoming undetectable . while the results for the sharp switch case are the same as in figure [ fig : nosim ] , the smooth efficiency switch now alters the predicted black hole orbital period distribution more significantly . however , these results predict 6 ( 5 ) black hole systems below 0.17 days when f = 0.03 ( 0.05 ) , so a smooth switch can not explain the absence of observed short orbital period black hole lmxbs . while the sharp switch is still more successful at reproducing the observed distribution , it appears that at higher values of @xmath2 ( @xmath182 0.05 ) the smooth switch may have a similar effect . in the previous plots , @xmath183 has been fixed at 0.4 @xmath83 . however , the actual dependence of @xmath183 on the systems other properties is uncertain , and it is likely that @xmath183 increases with orbital period . for this reason , we repeat our models with secondary masses that adhere to the main - sequence , roche lobe filling relation , @xmath184 @xcite , with the results plotted in figure [ fig : varm2 ] . here the relative detection probability curves have a different shape at high orbital periods , leading to minor differences in the predicted black hole population distributions in the lower panels . however , despite this difference , the actual effect of the efficiency switches does not change ; a sharp switch , with predictions of 0 ( 0 ) observable black hole systems below 0.17 days when f = 0.03 ( 0.05 ) is still able to reproduce the divergence in black hole and neutron star systems , whereas a smooth switch , with corresponding predictions of 5 ( 4 ) does not . the discontinuity ( at @xmath186 days ) in the duty cycles plotted in the bottom panels of figure [ fig : varyf ] , is due to the switch in accretion mode from magnetic braking to gravitational radiation as described in ( [ eqn : mdot2 ] ) , and also produces a feature in the relative detection probability plots . since this is in the orbital period range we are interested in , we also checked that our results were not sensitive to this switch by redoing our models with ( a ) a constant @xmath185 and ( b ) no gravitational radiation mode at short orbital periods . these changes alter p(@xmath161 1 outburst ) , which describes the shape of the relative detection probability plots . they have no effect on @xmath147 which depends on the accretion efficiency switch . therefore our results , and the effect of different types of accretion switch on the predicted black hole orbital period distributions , are unaltered since observations do not show consistent recurrence times for single sources , we have tested the resilience of our results to allowing a range of recurrence times at each orbital period . in the extreme case , the time spent in quiescence after each outburst is independent of any previous event . in this case we can assume a poisson distribution of outbursts so that : @xmath187 where @xmath188 = @xmath143/ @xmath156 . this affects the shape of the black hole to neutrons star detection ratio plots ( upper panels of figure [ fig : nosim ] ) , but not enough to change the results in the lower panels , so does not affect our conclusions . within the framework of the simple outburst model above , it is clear that the lack of known black hole lmxbs at short orbital periods can be explained by a transition to radiatively inefficient accretion , provided the switch to inefficient accretion is sharp . we assume a fixed primary mass as well as constant disc densities and viscosities . all of these are likely to vary between observed systems and would produce differences in the details of the orbital period distributions . a more robust treatment aimed at reproducing the observed period distributions would require detailed population synthesis calculations incorporating these factors . some of the major causes of uncertainty are discussed below . the results above are sensitive to our choices of @xmath11 , @xmath60 , and the black hole @xmath72 , which we investigate in figures [ fig : varyrho ] , [ fig : varynu ] and [ fig : varycorr ] . in figure [ fig : varyrho ] we plot the same 6 panels as in figure [ fig : varyf ] . the solid red line represents an @xmath189 black hole system assuming no switch to inefficiency at low luminosities . the other 3 lines show the effect of varying @xmath11 . each line assumes f = 0.03 , and the same @xmath60 as used previously ( @xmath190 @xmath62 s@xmath58 ) . the values of @xmath11 shown are , in order of decreasing density , @xmath191 g @xmath52 ( dark blue dotted line ) , @xmath192 g @xmath52 ( purple , fine dotted line line ) and @xmath193 g @xmath52 ( light blue dot - dashed line ) . both the peak luminosity relation and the orbital period at which the sharp switch ( switch a ) produces a cut - off in the distribution , are affected by changes in density . the smooth switch ( case b ) also varies . by requiring that the gradient and peak luminosity normalisation match observational expectations it is possible to rule out the higher density cases . similarly in figure [ fig : varynu ] , we plot the same systems , with f = 0.03 , but this time we keep the density at its original value ( @xmath192 g @xmath52 ) and allow @xmath60 to vary . we choose values of @xmath194 @xmath67 s@xmath58 ( dark blue dotted line ) , @xmath195@xmath67 s@xmath58(purple fine dotted line ) and @xmath196 @xmath67 s@xmath58 . ( light blue dot - dashed line ) . varying @xmath60 has a similar effect to varying @xmath11 for the peak luminosities . however , in addition , the timescales , which are inversely proportional to @xmath60 , are extended by decreasing @xmath60 . in figure [ fig : varycorr ] , @xmath60 and @xmath11 are fixed at their original values , and we vary @xmath72 , the correction factor applied to the bolometric luminosity of the black hole systems , to estimate the x - ray luminosity . this depends on the inner disc temperature , which is likely to range from 0.5 to 1 kev . we plot values of 2.5 ( dark blue dotted line ) , 4 ( purple fine dotted line ) and 8.5 ( light blue dot - dashed line ) , which correspond to disk temperatures of 1 , 0.7 and 0.5 kev respectively . varying @xmath72 alters the luminosities of the sharp switch case , and all three panels corresponding to the smooth switch case , in a similar way to varying @xmath60 and @xmath11 . since the inefficiency switch luminosity ( @xmath59 2 @xmath74 @xmath197 ergs s@xmath58 ) is higher than the limiting luminosity at all values of @xmath72 shown , varying this parameter does not affect the timescale or duty cycle plots when a sharp efficiency switch is implemented . it is clear that our choices of @xmath11 and @xmath60 , which are inherently uncertain , significantly affect the feasibility of the sharp switch to inefficiency explaining a short orbital period black hole deficit . in addition , @xmath72 , which depends on the black hole temperature , is likely to affect the normalisation of our results . however , in all cases the smooth switch scenario only produces a `` turn - off '' the known peak luminosity - orbital period relation at higher orbital periods than observed . we now investigate how sharp the switch needs to be , to produce the observed cut - off . in figure [ fig : ratios2 ] we vary the dependence of @xmath1 on @xmath91 such that : @xmath42 a larger value of @xmath44 more closely approximates the sharp switch scenario . here we show the relative detection probabilities on the top panel for n = 1 ( green dashed ) , n = 3 ( dark blue dotted ) and n = 5 ( purple dotted ) for f = 0.03 on the left panel and f = 0.05 on the right - hand panel . indices of @xmath7 appear to fit the black hole distribution as well as the sharp switch itself , provided f = 0.05 , with a prediction of 1(0 ) observable black hole systems below 0.17 days for n = 3(5 ) . the above discussion is based on detections possible with the _ rxte _ asm , which discovered several new black hole transients during its 15 year mission ( e.g. xte j1118@xmath198480 , xte j1650 - 500 , xte j1550@xmath19956 ) . however , there have been , and still are , many other surveys capable of discovering bhb transients , each with different sensitivity levels , energy coverage , time resolution and sampling . rxte _ also routinely performed scans of the galactic bulge region with the pca detectors , capable of detecting fainter sources but reduced angular and temporal coverage compared to the asm . these scans discovered the bh candidate xte j1752@xmath199223 among others . presently , the burst alert telescope ( bat ) on the @xmath200 mission detects known and new transients in the 15 - 150 kev range , making it sensitive to the bright low / hard states of black hole x - ray binaries ( bhb ) during outburst in addition to its higher sensitivity in crab units . this has been operating since 2004 and discovered several new systems including @xmath200 j1753.5@xmath1980127 , a short period ( 3.24 hr ) black hole candidate @xcite . since 2009 the @xmath201 monitor has been scanning almost the entire sky in the 2 - 20 kev x - ray band every 96 minutes and has detected several new transients . both @xmath200 and @xmath201 detected a new transient @xmath201 j1659@xmath198152 , later identified as another short period bhb . these , together with other past and current missions , increase our sensitivity to bhbs compared to using the _ rxte _ asm alone , increasing the chance of detecting outbursts in galactic bhbs except at very low peak luminosity and/or very long recurrence times ( @xmath18210 years ) . we also note that in calculating the relative detection probabilities , we have neglected some of the bias that is likely to exist in the ritter - kolb catalogue . systems with confirmed orbital periods must have had an optical follow - up after their initial x - ray detection . such a follow - up may be unsuccessful if the x - ray flux is very uncertain , or if the optical / ir counterpart is too faint . both of these factors are biased against short period systems . we justify neglecting these issues by arguing that they are likely to affect black hole and neutron star systems equally , so would not enter into the ratio calculations . the fact that the values of @xmath2 fall in the expected theoretical range provides good support for the premise that the transition to radiatively inefficient accretion underpins the lack of black hole lmxbs at short periods . alternative arguments for the lack of black hole systems ( based on system evolution , selection effects etc ) would produce arbitrary values of @xmath2 , and the fact that the cut off period for black hole lxmbs happens to agree with theoretical expectations for radiatively inefficient accretion would be purely coincidental . despite this support for a sharp change in @xmath1 , observations do not show a sudden change in bolometric luminosity at the efficiency switch boundary . in addition , observational evidence ( see e.g. * ? ? ? * ; * ? ? ? * ) suggests a smooth switch between efficient and inefficient flows . these factors suggest that , rather than being lost to the black hole , accretion power may be shifted out of the x - ray band , through a sharp change in @xmath72 . this may mean that systems become detectable in other wavelengths , making it harder for short orbital period black hole lmxbs to `` hide '' through inefficient accretion . we will consider the detection of hard x - ray dominated lmxb spectra in future work . it is worth noting that black hole lmxbs with @xmath186 day would be expected to produce low luminosity ( @xmath59 few % @xmath3 ) and short ( @xmath59 1 day ) outbursts . these systems would have low detection probabilities and would be rare , but may be observable in locations where formation conditions were favourable . it may be possible that short period black hole lxmbs form some fraction of the low - luminosity , short - duration very fast x - ray transients observed in the galactic centre ( e.g. * ? ? ? * ; * ? ? ? we would like to thank the anonymous referee for their helpful and constructive comments . we are also grateful to alan levine and ron remillard , for their advice on asm detections . gk acknowledges the support of an stfc studentship . this research is supported by an stfc consolidated grant . archibald a. m. , stairs i. h. , ransom s. m. , kaspi v. m. , kondratiev v. i. , lorimer d. r. , mclaughlin m. a. , boyles j. , hessels j. w. t. , lynch r. , van leeuwen j. , roberts m. s. e. , jenet f. , champion d. j. , rosen r. , barlow b. n. , dunlap b. h. , remillard r. a. , 2009 , science , 324 , 1411 casares j. , orosz j. a. , zurita c. , shahbaz t. , corral - santana j. m. , mcclintock j. e. , garcia m. r. , martnez - pais i. g. , charles p. a. , fender r. p. , remillard r. a. , 2009 , the astrophysical journal supplement , 181 , 238 cornelisse r. , davanzo p. , campana s. , casares j. , charles p. a. , israel g. , muoz - darias t. , obrien k. , steeghs d. , stella l. , torres m. a. p. , 2012 , monthly notices of the royal astronomical society , 420 , 3538 krimm h. a. , markwardt c. b. , deloye c. j. , romano p. , chakrabarty d. , campana s. , cummings j. r. , galloway d. k. , gehrels n. , hartman j. m. , kaaret p. , morgan e. h. , tueller j. , 2007 , the astrophysical journal , 668 , l147 e. , kouveliotou c. , belloni t. , cadolle bel m. , chenevez j. , daz trigo m. , homan j. , ibarra a. , kennea j. a. , muoz - darias t. , ness j .- u . , parmar a. n. , pollock a. m. t. , van den heuvel e. p. j. , van der horst a. j. , 2013 , astronomy and astrophysics , 552 , a32 + object name & ra & dec & type & transient & p@xmath202 & source & sample + v1487 aql & @xmath203 @xmath204 @xmath205 & @xmath206 @xmath207 @xmath208 @xmath137 & bh & yes & 30.8 & @xcite & ( i ) + v404 cyg & @xmath209 @xmath210 @xmath211 & @xmath212 @xmath213 @xmath214 @xmath137 & bh & yes & 6.4714 & @xcite & ( i ) + v4641 sgr & 18 19 21.6 & @xmath215 24 25 1 & bh & yes & 2.8173 & @xcite & ( i ) + v1033 sco & 16 54 00.1 & @xmath216 50 45 1 & bh & yes & 2.62120 & @xcite & ( i ) + bw cir & 13 58 09.9 & @xmath217 44 05 1 & bh & yes & 2.54451 & @xcite & ( i ) + v821 ara & 17 02 49.4 & @xmath218 47 23 1 & bh & yes & 1.7557 & @xcite & ( i ) + v381 nor & 15 50 58.8 & @xmath219 28 35 1 & bh & yes & 1.542033 & @xcite & ( i ) + il lup & 15 47 08.3 & @xmath220 40 11 1 & bh & yes & 1.116407 & @xcite & ( i ) + v2107 oph & 17 08 14.1 & @xmath215 05 32 1 & bh & yes & 0.521 & @xcite & ( i ) + gu mus & 11 26 26.6 & @xmath221 40 32 1 & bh & yes & 0.432602 & @xcite & ( i ) + qz vul & 20 02 49.4 & @xmath222 14 11 1&bh & yes & 0.344087 & @xcite & ( i ) + v616 mon & 06 22 44.4 & @xmath223 20 45 1 & bh & yes & 0.323014 & @xcite & ( i ) + j1650 - 4957 & 16 50 00.8 & @xmath224 57 45 1 & bh & yes & 0.3205 & @xcite & ( i ) + mm vel & 10 13 36.3 & @xmath225 04 32 1 & bh & yes & 0.285206 & @xcite & ( i ) + v406 vul & 18 58 41.7 & @xmath226 39 30 1 & bh & yes & 0.274 & @xcite & ( i ) + v518 per & 04 21 42.8 & @xmath227 54 27 1 & bh & yes & 0.212160 & @xcite & ( i ) + kv uma & 11 18 10.9 & @xmath228 02 13 1 & bh & yes & 0.16995 & @xcite & ( i ) + j1744 - 2844 & 17 44 33.1 & @xmath229 44 27.1 & ns & yes & 11.8367 & @xcite & ( ii ) , ( iii ) + 0042 + 3244 & 00 44 48.8 & @xmath212 00 33.2 & ns & yes & 11.6 & @xcite & ( ii),(iii ) + v1333 aql & 19 11 15.9 & @xmath230 35 06 1 & ns & yes & 0.78950 & @xcite & ( ii),(iii ) + j0556 - 3310 & 05 56 46.3 & @xmath231 10 26 1 & ns & yes & 0.684 & @xcite & ( ii),(iii ) + v822 cen & 14 58 21.9 & @xmath232 40 07 1 & ns & yes & 0.629052 & @xcite & ( ii),(iii ) + qx nor & 16 12 42.9 & @xmath233 25 23 1 & ns & yes & 0.5370 & @xcite & ( ii),(iii ) + j1749 - 2808 & 17 49 31.8 & @xmath229 08 05 2 & ns & yes & 0.367369 & @xcite & ( ii),(iii ) + j1745 - 2901 & 17 45 35.7 & @xmath234 01 34 11 & ns & yes & 0.347960 & @xcite & ( ii),(iii ) + v2134 oph & 17 02 06.4 & @xmath234 56 44 1 & ns & yes & 0.296505 & @xcite & ( ii),(iii ) + lz aqr & 21 23 14.5 & @xmath235 47 53 1 & ns & yes & 0.24817 & @xcite & ( ii),(iii ) + v1727 cyg & 21 31 26.2 & @xmath236 17 25 1 & ns & yes & 0.218259&@xcite & ( ii),(iii ) + ay sex & 10 23 47.7 & @xmath230 38 41 1 & ns & yes & 0.198096 & @xcite & ( ii),(iii ) + v5511 sgr & 18 13 39.0 & @xmath231 46 22 1 & ns & yes & 0.178110 & @xcite & ( ii),(iii ) + j1749 - 2919 & 17 49 55.4 & @xmath234 19 20 1 & ns & yes & 0.160134 & @xcite & ( ii),(iii ) + uy vol & 07 48 33.7 & @xmath237 45 08 1 & ns & yes & 0.159338 & @xcite & ( ii),(iii ) + j1751 - 3057 & 17 51 08.7 & @xmath238 57 41 1 & ns & yes & 0.144531 & @xcite & ( ii),(iii ) + j1710 - 2807 & 17 10 12.5 & @xmath229 07 51 1 & ns & yes & 0.136711 & @xcite & ( ii),(iii ) + v1037 cas & 00 29 03.1 & @xmath239 34 19 1 & ns & yes & 0.102362 & @xcite & ( ii),(iii ) + v4634 sgr & 18 29 28.2 & @xmath240 47 49 2 & ns & yes & 0.093725 & @xcite & ( ii),(iii ) + v4580 sgr & 18 08 27.6 & @xmath241 58 43.1 & ns & yes & 0.083902 & @xcite & ( ii),(iii ) + j1900 - 2455 & 19 00 08.7 & @xmath242 55 14 1 & ns & yes & 0.057815 & @xcite & ( ii ) , ( iii ) + j1756 - 2506 & 17 56 57.2 & @xmath215 06 28 1 & ns & yes & 0.037990 & @xcite & ( ii ) , ( iii ) + j1748 - 3607 & 17 48 13.1 & @xmath241 07 57 1 & ns & yes & 0.036 & @xcite & ( ii),(iii ) + bw ant & 09 29 20.2 & @xmath232 23 03 1 & ns & yes & 0.030263 & @xcite & ( ii),(iii ) + j1751 - 3037 & 17 51 13.5 & @xmath238 37 23 1 & ns & yes & 0.029460 & @xcite @xmath243 @xmath243 @xmath243 & ( ii),(iii ) + v5512 sgr & 18 14 31.0 & @xmath244 09 26 1 & ns & no & 25.07 & @xcite & ( ii ) + v1341 cyg & 21 44 41.1 & @xmath245 19 18 1 & ns & no & 9.8445 & @xcite & ( ii ) + v395 car & 09 22 34.7 & @xmath246 17 41 1 & ns & no & 9.0026 & @xcite & ( ii ) + j1538 - 5542 & 15 38 18.3 & @xmath247 42 12 4 & ns & no & 1.23 & @xcite & ( ii ) + v1101 sco & 17 05 44.4 & @xmath241 25 23 1 & ns & no & 0.938 & @xcite & ( ii ) + 1624 - 4904 & 16 28 02.8 & @xmath224 11 55 1 & ns & no & 0.86991 & @xcite & ( ii ) + 1813 - 1403 & 18 16 01.4 & @xmath248 02 11 1 & ns & no & 0.813 & @xcite & ( ii ) + v818 sco & 16 19 55.0 & @xmath249 38 25 1 & ns & no & 0.787313 & @xcite & ( ii ) + mm ser & 18 39 57.6 & @xmath250 02 10 1 & ns & no & 0.54 & @xcite & ( ii ) + lu tra & 16 01 02.2 & @xmath251 44 18 1 & ns & no & 0.46229 & @xcite & ( ii ) + v691 cra & 18 25 46.6 & @xmath252 06 18 1 & ns & no & 0.232109 & @xcite & ( ii ) + v926 sco & 17 38 58.1 & @xmath253 27 00 1 & ns & no & 0.193834 & @xcite & ( ii ) + v2216 oph & 17 31 44 .1 & @xmath254 57 41 1 & ns & no & 0.174827 & @xcite & ( ii ) + gr mus & 12 57 37.2 & @xmath255 17 19 1 & ns & no & 0.163889 & & ( ii ) + v801 ara & 16 4 55.6 & @xmath256 45 05 1 & ns & no & 0.158047 & @xcite & ( ii ) + 1822 - 0002 & 18 25 22.0 & @xmath223 00 44 1 & ns & no & 0.133 & @xcite & ( ii ) + 1323 - 6152 & 13 26 37.0 & @xmath257 08 09 1 & ns & no & 0.122580 & @xcite & ( ii ) + uw crb & 16 05 45.7 & @xmath258 51 45 1 & ns & no & 0.0771 & @xcite & ( ii ) + j0043 + 4112 & 00 42 08.7 & @xmath259 12 48 1 & ns & no & 0.074 & & ( ii ) + 1705 - 4402 & 17 08 54.5 & @xmath253 06 07 1 & ns & no & 0.054 & @xcite & ( ii ) + v1405 aql & 19 18 47.8 & @xmath235 14 17 1 & ns & no & 0.034730 & @xcite & ( ii ) + v1055 ori & 06 17 07.3 & @xmath260 08 13 1 & ns & no & 0.033713 & @xcite & ( ii ) + kz tra & 16 32 16.7 & @xmath237 27 42 1 & ns & no & 0.02876 & @xcite & ( ii ) + j1806 - 2924 & 18 06 59.8 & @xmath234 24 30 1 & ns & no & 0.027829 & @xcite & ( ii ) + qu tra & 15 47 54.3 & @xmath257 34 11 1 & ns & no & 0.01264 & @xcite & ( ii ) + 0918 - 5459 & 09 20 27.0 & @xmath247 12 26 3 & ns & no & 0.01207 & @xcite & ( ii ) +

we present a study of the recent star formation of 30 doradus in the large magellanic cloud ( lmc ) using the panchromatic imaging survey hubble tarantula treasury project ( http ) . in this paper we focus on the stars within 20 pc of the center of the massive ionizing cluster of 30 doradus , ngc 2070 . we recovered the star formation history by comparing deep optical and nir color - magnitude diagrams ( cmds ) with state - of - the - art synthetic cmds generated with the latest parsec models , which include all stellar phases from pre - main sequence to post - main sequence . for the first time in this region we are able to measure the star formation using intermediate and low mass stars simultaneously . our results suggest that ngc2070 experienced a prolonged activity . in particular , we find that the star formation in the region : i ) exceeded the average lmc rate myr ago ; ii ) accelerated dramatically myr ago ; and iii ) reached a peak value 1 - 3 myr ago . we did not find significant deviations from a kroupa initial mass function down to 0.5 m . the average internal reddening e(bv ) is found to be between 0.3 and 0.4 mag .

1505.04799

the large magellanic cloud ( lmc ) harbors the nearest giant extragalactic hii region @xcite , the tarantula nebula ( 30 doradus ) . the central region of the nebula is dominated by the cluster ngc 2070 , a collective of several dense sub - clusters ( @xcite ) whose light is dominated by massive ob stars . the most prominent of these sub - clusters is the very central super star cluster ( ssc ) r 136 . this region has a number of characteristics that make it extraordinary . first of all , it displays an extreme rate of star formation ( sf ) , with one quarter of the total massive recent ( @xmath4 myr ) star formation in the lmc contained within 15@xmath5 from 30 doradus @xcite . moreover , with a stellar mass of at least @xmath6 concentrated within a radius of 4.7 pc @xcite , this cluster can be classified as a relatively low - mass clone of more distant starburst clusters , likely building blocks of starburst galaxies . over the years , there have been debates over whether or not ngc 2070 is a young analog of old globular clusters . recent arguments seem to suggest a positive answer . using photometry @xcite found that ngc2070 s initial mass function ( imf ) is salpeter - like down to @xmath7 , giving the first direct evidence that solar - like stars are formed in a starburst cluster . using multi - epoch spectroscopy of o - type stars , @xcite measured a velocity dispersion between 4 and 5 kms@xmath8 , close to the expected value if it were in virial equilibrium . in this case the velocity dispersion would be low enough to allow the cluster to stay bound . due to the proximity of the lmc , 30 doradus can be imaged with the hubble space telescope ( hst ) on scales down to @xmath9 pc , resolving stars down to the sub - solar regime . by coupling hst imaging with ground - based spectroscopy diagnostics , we know that the region has undergone continuous sf activity or a superposition of multiple bursts of sf . indeed , over the whole region , there is evidence of at least three events : i ) an on - going off - center activity , as documented by the presence of embedded o - type stars and luminous infrared protostars along an arc of molecular gas and warm dust around 30 doradus @xcite ; ii ) a recent event , represented by the ssc r 136 itself , 2 myr old or less , as inferred from spectroscopy of the o stars ( @xcite ) , 3 - 4 myr old from hst imaging ( @xcite ) ; iii ) an `` old '' event , documented by the presence of hodge 301 , a 25 myr old cluster ( @xcite ) located only 3 arcmin northwest of r 136 . however , most of high resolution studies are limited to a few bands and cover only small patches of the 30 doradus complex . the hubble tarantula treasury project ( http ; sabbi et al . 2015 , in prep . ) , an hst survey covering a @xmath10 wide area around 30 doradus at high resolution , fills this gap . this unique data - set , which combines resolution and spatial coverage , can be exploited in varied ways , including studies of massive sf , low mass sf , the imf , the reddening distribution , etc .. in our current paper , we take the opportunity to explore the sf during the last 50 myr for the central 40 pc of 30 doradus , a.k.a . the starburst cluster ngc 2070 , by comparing the observational color - magnitude diagrams ( cmds ) with state - of - the - art synthetic cmds . these simulations incorporate the latest ( v.1.2s ) set of padova and trieste stellar evolution code ( parsec ) isochrones ( see @xcite and @xcite ) , the first theoretical library to include homogeneously all stellar phases from pre - main sequence ( pms ) to post - main sequence for all masses between @xmath11 and @xmath12 . for the first time , we derive the history of the region using low and intermediate mass stars , either in the pms or main sequence ( ms ) phases , _ simultaneously_. in particular , we exploit magnitudes and colors of the pms turn ons ( hereafter ton ; see section [ tons ] ) , i.e. the cmd loci where the pms phase joins to the ms . this `` hook '' has been demonstrated to be particularly sensitive to age ( see , e.g. , @xcite ) . we use both the optical f555w vs f555w@xmath3f775w and the near infrared ( nir ) f110w vs f110w@xmath3f160w cmds of the http sample , which offer complementary advantages in terms of higher spatial resolution ( the optical ) and lower reddening sensitivity ( the nir ) . to recover the sf rate of ngc 2070 since the beginning of its activity , there are three important factors that must be accounted for : differential reddening , lmc field contamination , and stellar crowding . the first causes the cmd features to appear redder and more scattered , mimicking older populations and lowering the age resolution . field contamination mostly adds low mass stars to the sample , mimicking much older populations and a steeper imf . finally , stellar confusion affects completeness and , therefore , the reachable look - back time . moreover , unaccounted incompleteness can also mimic mass segregation . the paper is structured as follows . in section 2 we briefly describe the observations . section 3 is dedicated to the physics of the ton clock . in section 4 we identify the general properties of the stellar populations in the http data - set , with special emphasis on ngc 2070 . in section 5 we perform artificial star tests , the only way to take crowding into account , and we recover the sfh of ngc 2070 using the synthetic cmd approach . field contamination is also carefully discussed . results are presented in section 6 and compared to the literature in section 7 . section 8 compares ngc 2070 with other starburst clusters . our conclusions ( section 9 ) close the paper . throughout the paper , for the sake of simplicity , we will refer to the entire http data - set as `` 30 doradus '' , the starburst cluster as ngc 2070 and its core as r 136 . we observed 30 doradus with the hst wide - field camera 3 ( wfc3 ) and the hst advanced camera for surveys ( acs ) as part of proposal go-12939 ( pi : e. sabbi ) . we built this program on an existing hst monochromatic survey in the f775w filter ( go-12499 , pi : lennon ) , designed to measure proper motions of runaway candidates . in both data - sets we used the wide field channel ( wfc ) of acs in parallel with either the uvis or the ir channels of wfc3 to maximize the efficiency of the observations . the images were taken with the filters f275w , f336w , f555w , f658n , f110w and f160w between december 2012 and september 2013 . the survey utilized 60 orbits . figure [ 30d_full ] shows the f775w image of 30 doradus with indicated the major stellar concentrations , ngc 2070 , hodge 301 and ngc 2060 . bright and faint sources are respectively identified using the packages img2xym_wfc.09x10 ( @xcite ) and ks2 , an evolution of the program described in @xcite . a detailed description of the photometric analysis can be found in sabbi et al . ( 2015 , in preparation ) . we culled the catalog of detected objects to only include sources with high quality in the psf - fitting , q@xmath13 . the final catalog contains @xmath14 detected in the filter f275w , @xmath15 in f336w , @xmath16 in f555w , @xmath17 in f658n , @xmath18 in f775w , @xmath19 stars in f110w and @xmath20 in the f160w . one of the characteristics of 30 doradus that makes it a particularly interesting object is the high concentration of massive stars that coexist with a plethora of coeval intermediate and low mass pms stars . this enables us to reconstruct the past history of 30 doradus using the ms turn - off ( msto ) , whose ms evolutionary times are shorter than @xmath21 myr . ] , the locus of the cmd where ms stars exhaust their core hydrogen , and the pms ton , where low and intermediate mass pms stars ignite hydrogen in their cores . in terms of stellar mass , the msto mass is the mass of the most massive star still on the ms at an evolutionary time corresponding to the age of the cluster , while the ton mass is the mass of the least massive star that has reached the ms at the age of the cluster . in the following we discuss pros and cons of the two clocks . _ massive stars : _ the main appeal of using the msto of massive stars ( m@xmath22 ) stems from the high luminosity of such sources , which translates into high quality photometry and complete samples . the optical and infrared spectra of these objects are well approximated by the rayleigh - jeans tail of a black body with temperature t@xmath23 . hence , given that the spectral energy distribution shape is almost unchanged as a function of wavelength , the optical and infrared colors are nearly constant ( and around 0 mag ) . this is clearly visible in figure [ 3iso ] , where we overlaid young isochrones of different ages ( 0.5 , 1 , 3 , 7 , 15 myr ) on to the entire http catalog ( shaded grey area ) in the uv ( left panel ) , optical ( middle panel ) and nir ( right panel ) cmds . the adopted distance modulus ( m@xmath3m)@xmath24 and reddening e(b@xmath3v ) are 18.5 ( @xcite ) and 0.3 ( chosen by eye based on the fit of the optical cmd ) , respectively . in the nir cmd the upper ms ( ums ) looks like a vertical line and isochrones of different age have mstos almost indistinguishable from one another . this degeneracy is attenuated in the optical and almost lifted in the uv cmd . observationally , studies of massive stars has always been hampered by their rarity , due to the short timescales involved in their evolution and the steepness of the imf . on the theoretical side , models of massive stars are still affected by major uncertainties . in particular , physical mechanisms like mass loss , rotation and binary evolution are not well understood ( see , e.g. , @xcite ) . _ intermediate and low mass stars : _ for ages younger than 20 - 30 myr , the pms ton is another valuable stellar chronometer which involves intermediate and low mass stars instead of massive stars have no pms phase at all . ] . in analogy with the msto , the ton properties are directly related to the age of the stellar population , but with evolutionary times much shorter than the corresponding ms times . in fact , the age of a cluster is equal to the time spent in pms phase by its most massive star still in pms phase . by definition , this star is at the ton . hence , when the intrinsic luminosity of the ton is detected , it is straightforward to associate it with the age of the cluster . from the cmd point of view , the potential strength of the ton is apparent from the morphology of the isochrones in fig . [ 3iso ] . in the optical and nir cmds the isochrone portion just before the ms has a hook and then is significantly flatter than the ms . the ton is at the vertex of the hook , quite easy to recognize . for ages older than 20@xmath330 myr the pms phase is much closer to the ms and the ton visibility declines . theory also predicts that the pms phase ( recognizable as the portion of the isochrones at the right of the ms in fig . [ 3iso ] ) itself is a valuable age indicator : older pms isochrones are fainter and closer to the ms than younger ones . however , poorly understood phenomenons like residual mass accretion and magnetic fields ( whose interplay is responsible for the appearing of irregular photometric variability and uv to infrared excesses ) , and observational uncertainties like differential reddening from the circumstellar material can dislocate these stars ( especially in the first few myr ) from their theoretical positions in the cmd ( see @xcite for a review ) . an advantage of the ton with respect to the msto derives from the evolution at nearly constant luminosity of pms stars near the ms , corresponding to an almost horizontal track in the optical and nir cmds . this leads to a luminosity(ton ) - age relation . on the other hand , the ms evolution of massive stars near the msto is rather vertical ( except in the uv cmd ) , hence age is not uniquely related to luminosity . among the drawbacks , the intrinsic faintness of tons compared to the mstos makes tons prone to photometric errors and incompleteness , issues that are exacerbated in the uv cmd because older tons tend to be also redder . from the theoretical side , the ton visibility is intimately connected with the pms evolutionary times which are still model dependent ( see , e.g. , @xcite and references therein ) . in this paper we aim to study the sfh of 30 doradus with the tons , hence we focus our analysis on the optical and nir cmds . as shown in the optical cmd ( middle panel of fig . [ 3iso ] ) , the most massive star that is relevant to this study is around @xmath25 . before closing this section , we emphasize another interesting feature that emerges from fig . [ 3iso ] . whereas the isochrones reddened with e(b@xmath3v)= 0.3 fit well the ums in the optical and nir cmds ( middle and right panel ) , the same isochrones are clearly too blue in the uv cmd ( left panel ) . the explanation for this effect resides in the different populations that these cmds trace : most of the uv stars are very young 30 doradus stars , with minimal contamination from field stars ( the rc is hardly visible at f275w@xmath26 ) , while the optical and nir stars can be both members and field stars . as a consequence , the average reddening of the young population in 30 doradus must be _ higher _ than the adopted value e(b@xmath3v)= 0.3 , while the average reddening of the field can be as low as the foreground mw reddening ( e(b@xmath3v)= 0.07 ; @xcite ) . the next section is dedicated to a cursory cmd inspection of the stellar populations in the whole http data - set . the cmd is the most powerful tool to recover the history of a resolved stellar population , since different parts of the cmd are populated by different masses with different evolutionary times . in the following analysis we use the cmd to investigate which populations are present in the whole http catalog and the role of the ngc 2070 region . for this task we use the optical cmd , which offers better spatial resolution than the nir one . a more quantitative analysis will be the subject of section [ sfr ] . the f555w vs f555w@xmath3f775w cmd of the entire http data - set is shown in fig . [ cmd_tot ] , with overlaid isodensity contours ( left panel ) and stellar isochrones ( parsec ) of different ages ( right panel ) . the first remarkable feature of this cmd is an extended ums , populated by a plethora of intermediate and high mass stars . as the overlaid isochrones show , the width of the ums is hardly explained by a difference in age . moreover , given the average youth of these stars , plausibly much younger than @xmath27 myr , a metallicity spread is unlikely . as widely discussed in other works ( see , e.g. , @xcite ) , differential reddening is the main cause of this effect , although stellar rotation may also have a role in widening the ums . to the right of the ums , the next striking feature is the very elongated red clump ( rc ; see also section [ rc_sec ] ) and a broad red giant branch ( rgb ) . these phases are populated by stars older than 1 gyr and belonging to the general field population of the lmc . as for the ums width , most of the elongation and broadening is due to severe differential reddening affecting the region ( see @xcite ) . finally , the lower ms shows a huge color dispersion ( more than 1 mag ) at relatively bright magnitudes ( f555w@xmath26 ) . as usual , a possible explanation comes from the comparison with the isochrones . although differential reddening contributes to this effect , pushing part of the lower ms ( lms ) to the red , the existence of short - lived massive stars requires that a large fraction of these red stars are genuine pms stars ( coeval with the massive stars ) . more information on the nature of these populations can be inferred from their spatial distribution . in fact , the spatial distribution of stars in different evolutionary stages yields important information on the star formation processes across the region . assuming a velocity dispersion of 21 - 27 kms@xmath8 ( measured from rgb stars ; @xcite ) , stars older than 1 gyr are expected to be diffused over scale lengths of several kpc , hence their distribution should be rather uniform over the http field of view ( fov ; @xmath28200 pc ) . on the other hand , stars like those in ngc 2070 , young and with very low velocity dispersion ( 4@xmath35 kms@xmath8 ; @xcite ) , are expected to be close to their birthplace . the right panel of fig . [ cmd_tot ] highlights cmd regions of selected ums ( cyan box ) , rc ( green box ) , low mass ms ( lms ) ( pink box ) and pms ( blue box ) stars . ums and pms stars are young objects a few tens of myr old , rc stars are intermediate age stars ( @xmath29 gyr ) , lms are ms stars older than few tens of myr . figure [ all_xy ] shows the corresponding spatial density ( stars per pc@xmath30 ) . populations grow more compact as one moves toward younger ages ( see also fig . 5 and 6 in @xcite ) . other interesting features are : \1 ) the ums stars ( bottom left panel in fig . [ all_xy ] ) appear very clustered . three major concentrations are visible : ngc 2070 , the most prominent , about 40 pc wide ( encircled in blue ) , hodge 301 ( encircled in green ) and ngc 2060 ( encircled in cyan ) . lighter over - densities are also visible throughout the fov . in other words , although most of the ongoing and recent ( i.e. , in the last 50 myr ) sf is concentrated in few dense loci , a minor recent activity is present in the entire area ; \2 ) as compared to the ums distribution , rc distribution ( bottom - right panel ) is , as expected , rather uniform . rc stars are a pure sample of field stars . their age , older than 700 myr , rules out that these objects are associated with 30 doradus , which is , at most , a few myr old . the small over - densities are probably contamination of intermediate mass pms stars or very reddened massive stars from the youngest regions . from a theoretical point of view , the intrinsic position of the rc in the cmd could be affected by factors like binaries , differences in age and metallicity , etc . ( see , e.g. , @xcite ) . however , these effects are insufficient to account for the observed rc elongation ( up to 1.5 mag in f555w@xmath3f775w ) , which is mostly due to differential reddening . the source of this reddening is the gas / dust located between the closest and farthest rc star along the line of sight . more specifically , background rc stars suffer the highest degree of absorption , which is caused by the combination of milky way ( mw ) and 30 doradus extinction , while foreground stars will suffer only from mw ( e(b@xmath3v)@xmath31 ) extinction ; \3 ) the distributions of lms ( top - left panel ) and pms ( top - right panel ) stars appear mutually exclusive : the top - left corner of the fov shows a paucity of pms stars , while lms stars are clearly overabundant there . the main cause for this effect is reddening . most of the lms stars belong to the lmc field . this is because 30 doradus is younger than few myr , hence its ms stars are necessarily brighter than our lms box of fig . [ cmd_tot ] ( right panel ) ( at ages @xmath32 myr the 30 doradus low - mass stars are mainly still in pms , as suggested by the isochrones in fig . [ cmd_tot ] ) . under these circumstances , the only way to remove lms stars from the lms box is the action of reddening ( see also the map of fig . [ rc_spread ] ) . doing so , these reddened ms stars are likely to fill the pms box , eventually producing the observed anti - correlation lms / pms stars ( a similar behavior is found by @xcite in ngc 346 ; see their fig . in addition to this , the tendency of young pms stars to be concentrated where the optical depth is higher exacerbates this effect . the only exceptions are the center of ngc 2070 , where the lms stars are missed because of the severe incompleteness , and hodge 301 , whose high concentration of lms is due to its higher age ( so hodge 301 s ton gets into the lms box ) . finally , the blue circle in fig . [ all_xy ] is the region we have used to recover the sfh of ngc 2070 . it is immediately clear from the maps that the region harbors most of the ums stars of the entire 30 doradus complex , as well as a remarkable concentration of pms stars . on the other hand the region is quite deficient in lms stars , probably lost because of the extreme crowding conditions . in this section we discuss the broad cmd features of the stellar population of ngc 2070 . our selection includes all stars within 2000 pixels ( @xmath33 20 pc ) from the cluster center . figure [ cmd_c ] shows the corresponding cmd for the f555w vs f555w@xmath3f775w filters with overlaid isodensity contours ( left panel ) and parsec isochrones of 1 , 7 and 15 myr ( right panel ) . the ms contours of ngc 2070 differs from those of 30 doradus as a whole ( fig . [ cmd_tot ] ) . the peak density is around f555w@xmath34 , with minor peaks down to f555w@xmath35 , while the 30 doradus cmd shows a smooth profile down to f555w@xmath36 . in figure [ cmd_c_lf_radial ] we investigate the radial distribution of stars in ngc 2070 . the top panels show cmds of stars in concentric annuli ( 1 - 2 - 3 ) of equal area ( the radius of the innermost circle is about 12 pc ) around the center of the cluster ( from left to right , progressively further out from the center ) , the bottom panel shows the corresponding luminosity functions ( lfs ) . we find that : \1 ) the lf of the innermost region 1 shows two clear peaks , located at f555w@xmath34 and f555w@xmath36 . the bright peak is well fitted by an isochrone of 7 myr ( see the figure ) , evidence of a young ton . stars brighter than this magnitude are mostly ms cluster members , while fainter stars are members only if they are on the pms ( by the definition of ton , fainter members have not reached the ms yet ) . indeed , a visual inspection of the cmd shows a clear color bimodality below f555w@xmath34 , which is likely due to field ms stars ( mostly not - members ) on the blue side , and pms stars ( members ) on the red side . the dip after the bright peak is caused by the short evolutionary timescale of the pms phase compared to the ms . after the dip the lf rises again following the imf , which increases at lower masses . eventually the incompleteness wins , creating the second decline at f555w@xmath36 . the apparent lack of lower ms stars in the region 1 is probably due to the higher crowding , which causes more severe incompleteness than in the more external annuli 2 and 3 . in section [ sfr ] we will use the synthetic cmd approach , combined with artificial star experiments , to test if these lower ms stars are compatible with lmc field contamination or hide some older tons . to this aim , we also need to estimate the lmc contamination in a reference field ( see next section ) . \2 ) in contrast to the innermost region 1 , annular regions 2 and 3 show a monotonic increase towards fainter magnitudes , with no intermediate peak before the final drop due to incompleteness . in terms of age , the lack of obvious tons means that regions 2 and 3 are not dominated by as young stars like in region 1 . nonetheless , at the magnitude of the region 1 peak , region 2 s lf has a mild excess of stars compared to region 3 , which would point to a poorly populated ton . as a further proof , the region 2 cmd also shows an excess of pms stars relative to region 3 . \3 ) stars brighter than @xmath37 in the region 3 show a larger color spread than in regions 1 and 2 . the rc is clearly visible in regions 1 and 2 ( at magnitudes 20.0@xmath320.3 and colors 1@xmath31.5 ) , while it is much more dispersed in 3 . all of these suggest a differential reddening that is higher in region 3 than in regions 1 and 2 . the technique of recovering the sfh of spatially resolved populations from their cmds ( e.g. @xcite ) has undergone continuous refinement and can now provide reliable sfhs for any resolved population within @xmath0 mpc ( see , e.g. , @xcite and references therein ) . these advances have been achieved because of improved models of stellar evolution , which are now computed for fine grids of stellar masses and metallicities and faster multi - cpu computing facilities , which allow one to perform extensive artificial star tests on the real images and to fully explore a wide parameter space . a widely used approach consists of populating a 2d array of basic synthetic cmds generated from stellar models . each basic cmd is a `` fuzzy '' isochrone , with duration @xmath38 , fixed metallicity and an assumed imf . to compare the basic cmds and the observational counterparts , the models are convolved with photometric errors and incompleteness as derived from artificial star tests performed on the real images . the best superposition of the basic cmds defines the sfh and the age - metallicity relation , and is the one that minimizes the residuals from observational cmd . the approach adopted here is described in more details in the appendix . despite the advances in this field , the derivation of the sfh from the cmd is often affected by systematic errors that are difficult to assess . from a theoretical point of view , several stellar phases are still uncertain ( thermally pulsing asymptotic giant branch , pms , and post - ms for massive stars ) , while observationally differential reddening and highly variable incompleteness can be hard to treat . concerning ngc 2070 , and the general 30 doradus region , we face three major problems in trying to estimate the sfh : 1 ) membership errors due to field interlopers from the lmc field that mimic older populations ; 2 ) the extreme crowding conditions exacerbate the incompleteness , shortening the reachable look - back time ; 3 ) the high level of differential reddening spreads and dims the cmd , blending together young pms stars and older ms stars , which introduces further age ambiguities . to measure the sfh of ngc 2070 we need to estimate the local field contamination . a typical approach is to decontaminate the cluster by subtracting the star - counts from a reference field . however , given reddening and incompleteness variations across the whole 30 doradus , it is impossible to find another direction replicating the observational conditions of ngc 2070 . a way out is to find a field that resembles as much as possible the lmc field _ we would observe without _ ngc 2070 , which means low or negligible sf activity in the last 50 myr , minimal differential reddening and high completeness down to f555w@xmath39 . this reference lmc field could be then artificially corrected for the more severe incompleteness and photometric errors of ngc 2070 . the resulting field would differ only in terms of normalization and differential reddening from the real lmc field contaminating ngc 2070 . finally , reddening and normalization of this field could be tuned together with the sfh of ngc 2070 until an adequate match of ngc 2070 s cmd is found . as a first step we searched for low reddening areas around 30 doradus . a good tracer for extinction is the rc color . figure [ rc_spread ] shows the spatial distribution of rc stars , color coded according to the f555w@xmath3f775w color ( larger and redder symbols correspond to redder f555w@xmath3f775w ) . we find that , at odds with the smooth distribution of the entire rc sample ( see the bottom - right panel of fig . [ all_xy ] ) , redder rc stars are very concentrated along filaments and arcs , most likely tracing gas and dust in 30 doradus . this allows us to select extended regions , like the wide region just above ngc 2070 ( see the black box in the figure ) , with relatively blue rc stars . because these stars can not all be in the foreground , these regions must be low - extinction windows in the gas / dust layers of 30 doradus . interestingly , the aforementioned region is also poorly populated by pms stars ( see top - right panel of fig . [ all_xy ] ) , which implies minimal sf activity in the last 50 myr . after a careful inspection of all regions with low extinction and pms number , we found the best compromise in the region delineated by a black line in fig . [ rc_spread ] . the corresponding cmd is shown in fig . [ cmd_low_red ] , here overlaid on the 30 doradus cmd ( grey symbols ) . this sample was adopted to represent our reference field . as expected , its rc and lower ms are much tighter than the general cmd , signatures of scarce differential reddening and young sf activity , respectively . it is worth noting , however , that residual differential reddening is still present . to test the level of completeness of our photometric data and to have reliable estimates of photometric errors , we ran extensive artificial star experiments . the experiments consist of adding `` fake '' sources for each of the eight pass - bands , modeled with the psf used in the photometric analysis of the frames , onto the actual images . we then applied the same source detection routines used for our science images to the fields containing the combined actual images and the fake sources . we then determined the completeness fractions , defined as the ratio of recovered artificial stars to the number of the injected ones . we considered an artificial star lost if it is not recovered or if it is recovered being 0.75 mag brighter than its input magnitude . in fact , this means that it has fallen either on a real star brighter than the artificial star or one of its same brightness . thus , we are not recovering or measuring the artificial star but a real one instead . to preserve the crowding conditions of the data , fake stars were arranged in a spatial grid such that the separation of the centers in each star pair was larger than two psf radii . we found that a separation of 20 pixels guarantees that the probability of recovering a fake star is the same as it would be if we added only one fake star . the experiment is then repeated using a series of slightly shifted grids . as an example of our fake stars procedure , the top panel of figure [ map50 ] shows the impressive 50% completeness map in the f555w band for a region 2000 pixel ( @xmath33 20 pc ) away from the cluster center ( 5 million fake stars ) . the bottom panel shows the corresponding real image . the most striking feature is the remarkable completeness variation , up to 8 magnitudes , in different locations within the region . regions like the broad area around x=10750 , y=13500 are photometrically complete at 50% level down to @xmath40 , whereas more crowded sub - regions , like the central area , are 50% complete only for @xmath41 . + such variations are determined by the interplay of four major effects : photon scattering off dust in 30 doradus , continuum / line nebular emission , pixel saturation and stellar crowding . dust scattering and gas emission raise the background flux and , in turn , make it more difficult to resolve stars . the net result is visible in the diffuse `` nebulous '' areas of the completeness map ( see figure [ map50 ] ) . more localized than the dust and gas effects , pixel saturation produces the visible `` streaks '' ( charge - bleeding along the rows of the ccd ) extending off of many bright stars . finally , the extreme crowding conditions near r 136 are responsible for the central completeness `` hole '' ( where most of the injected fake stars are lost ) . we are not assuming any a priori reddening distribution . in principle we do not know where populations of different age are located _ inside _ ngc 2070 , so reddening is an unknown function of age . naively speaking , young massive stars may have carved the gas with their ionizing flux , hence lowering the extinction . on the other hand , high levels of sf could be sustained only where the gas density is higher , which would lead to the opposite situation . besides , normalization and reddening of the reference field are also free parameters . furthermore , pms stars may appear reddened due to their local circumstellar material . when dealing with such a large parameter space ( age , reddening and field contamination ) there is uncertainty on whether the best solution is local or global . to cope with this we combined a genetic algorithm with a local search procedure ( hybrid - genetic algorithm , hga ; see appendix [ app ] for details ) which is more effective to avoid local trapping than local search alone . in our approach the first step is to store a library of `` basic '' synthetic cmds , where each cmd is a monte carlo synthetic population generated with a step - wise sf and reddening distribution . when the cmds are generated other population parameters like metallicity , distance modulus ( m@xmath3m)@xmath24 , slope of the imf and binary fraction q are kept fixed at 0.008 ( e.g. , @xcite ) , 18.5 ( e.g. , @xcite ) , @xcite and 30% , respectively . the extinction coefficients @xmath42 are taken from @xcite . the explored ages range from now up to 50 myr ago . since older isochrones tend to be more tightly packed in the cmd than younger isochrones , the duration of each age step increases with age from 1 myr ( at the present time ) to 20 myr ( 50 myr ago ) . the reddening is allowed to vary between 0 and 1 mag with a step of 0.05 mag . all `` basic '' synthetic cmds ( see left panel of fig . [ basic_cmds ] ) are then degraded using the photometric errors and incompleteness as derived from the artificial star tests discussed in the previous section . once this basis is generated , any complex synthetic cmd can be constructed as a linear combination of the `` basic '' synthetic cmds . in order to take into account field contamination , the cmd of the reference field ( see section [ rc_sec ] ) is artificially reddened with steps of 0.05 mag between e(b@xmath3v)=0 and e(b@xmath3v)=1 to produce a complete basis of field cmds . likewise any kind of contamination can be simulated by linearly combining these `` basic '' field cmds ( see right panel of fig . [ basic_cmds ] ) . the only difference is that the `` basic '' field cmds are corrected for the _ difference _ in photometric errors and incompleteness as estimated in the field itself and in the ngc 2070 region , while the `` basic '' synthetic cmds are only corrected for the latter . however , since in the reference field completeness and photometric errors at f555w@xmath39 are 100% and @xmath43 of the error in sub - region a1 , the first correction is almost negligible . the combination of `` basic '' cmds ( both synthetic and field ) which minimizes the residuals from the observational cmd ( in terms of poissonian likelihood ) is searched with the hga code . the best coefficients tell us the most likely : 1 ) star formation rate as a function of time ( i.e. , the sfh ) ; 2 ) total reddening ( foreground @xmath44 internal ) as a function of time ; and 3 ) field contamination ( reddening and normalization ) . _ which part of the cmd _ not all of the cmd is used to recover the sfh . our analysis is limited on the bright end by the saturation magnitude and on the faint end by the 50% completeness magnitude level . nonetheless , even with these conservative selections , the large variations of completeness with magnitude in some regions ( see upper plot in fig . [ map50 ] ) force us to derive the sfh in sub - regions within which the completeness is reasonably uniform . in fact , our artificial stars are uniformly distributed across the region , whereas real stars are concentrated in clumps and filaments , structures covering a minor fraction of the total area . as a consequence , most of the stars will suffer a real incompleteness which is worse than the average measured with artificial stars . the division in `` iso - complete '' sub - regions mitigates this bias . [ map3a ] shows our selection : sub - regions of ngc 2070 where the completeness in f555w is 50% for f555w @xmath45 ( a1 ) , @xmath46f555w@xmath47 ( a2 ) and @xmath48f555w@xmath49 ( a3 ) are indicated in red , orange and blue , respectively . the stars found in both the f555w and f775w filters inside these sub - regions are used to recover the optical sfhs . overall , our sub - regions trace the stellar density , from the lowest ( a1 ) to the highest ( a3 ) . this analysis leaves out the very center of ngc 2070 , a 3 pc region mostly represented by the ssc r 136 . here stellar crowding is so severe that most of the intermediate and low mass stars are lost , hence little , if any , information is available from tons . to further validate the sfh , we independently analized the nir data . to do this we used all stars detected in both the f110w and f160w images inside each sub - region ( defined using the f555w data ) . in this case however the faint limiting magnitude was anchored to the 50% completeness level in f110w . figures [ sfra1 ] and [ sfra2 ] show the recovered sf rate ( in @xmath50 ; top panel ) and reddening as a function of age ( bottom panel ) for sub - regions a1 and a2 respectively , as predicted using the optical ( green curve ) and nir ( red curve ) cmds . in principle our code provides the full distribution of reddening as a function of age , but to ease the visualization we only provide the mass - weighted average reddening as a function of age . the shaded bands indicate one ( darker ) and two ( lighter ) standard deviations . such uncertainties are the quadrature sum of a statistical error ( obtained by bootstrapping the data and re - deriving the solutions ) and systematic error ( obtained by re - deriving the solutions with different age - binnings and cmd binning scheme ) . despite the different spatial resolution and reddening sensitivity , optical and nir predictions are in good agreement . both results predict that mild sf activity started throughout the whole ngc 2070 region @xmath0 myr ago , and that about 7 - 8 myr ago the birth rate accelerated . interestingly , the activity in the last 1 myr has been relatively quiet compared to the average in the last 5 myr . concerning the activity that commenced 20 myr ago , it is important to note that our sfhs have been obtained by taking into account field contamination , hence , unless we have been very unlucky in our field selection , it is reasonable to assume that this activity is a local extras over the lmc field . whatever the origin , its significance is low ( zero activity is still within 1-sigma error bars ) . focusing on the specific sub - regions , the sfh of region a1 shows a mild bimodality , with a minor peak around 5 - 7 myr ago and a major peak 1 - 4 myr ago ( 1 - 3 myr ago in the optical sfh , 2 - 4 myr ago in the nir one ) . the measured reddening distribution anti - correlates with sf : the average e(b@xmath3v ) is @xmath51 for stars with ages older than 7 myr , when the sf activity was lower , and @xmath52 for stars of younger ages , when the sf was stronger . optical and nir reddening derivations agree within errors , both in confirming the presence of notable differential reddening . the large oscillations of the nir solution are mostly due to the low sensitivity to reddening in these filters . figure [ full_red ] shows an example of the full reddening solution ( number of considered stars vs e(b@xmath3v ) ) for region a1 in three age bins ( 2 , 6 and 20 myr ) . as it can be seen , the full distributions are asymmetric with a long tail at high reddening . the sf in sub - region a2 ( fig . [ sfra2 ] ) is about two times higher than in sub - region a1 . the most important feature in the a2 sf is a pronounced peak in both optical and nir solutions 1 - 3 myr ago . like in sub - region a1 , the reddening of sub - region a2 is anti - correlated with the sf activity . the average reddening values are also very similar . because of the high crowding we used only the optical cmd to derive the sfh of the most internal sub - region a3 . [ sfra3 ] shows the result ( blue ) compared to the other two sfhs over the last 15 myr . the bottom panel shows the corresponding reddening distribution . the peak activity is almost three times higher than in a2 , and at odds with other sub - regions , is clearly located between 1 and 2 myr ago . the total mass assembled in the regions a1 , a2 , and a3 , in the last 7 myr , is about @xmath53 , @xmath54 , and @xmath55 , respectively . finally , our analysis does not involve the central core of ngc 2070 , where the activity is presumably much higher . a rough estimate can be made by re - scaling the sfh of region a3 to match the number of stars in the magnitude range f555w 16 - 18 , where the sample is fairly complete even in the core . in this magnitude range the core contains two times the stars of region a3 , but concentrated in an area four times smaller . this leads to a scaled peak rate of the order of @xmath56 and a total mass of @xmath57 , in excellent agreement with @xcite s @xcite estimate . adding this mass estimate to the mass of sub - regions a1 , a2 and a3 , we get a total mass for ngc 2070 of @xmath58 increases the total by a factor 1.6 . ] . this value is compatible with the results of @xcite , who found @xmath59 within 14 pc from r 136 ( using a salpeter imf down to 0.5 m@xmath2 ) , @xcite , who found @xmath60 ( using a salpeter imf down to 0.1 m@xmath2 ) and @xcite , who found @xmath61 using dynamical considerations . figure [ cmd_synth_vi ] compares the observed optical cmds ( top panels ) in the three sub - regions ( from left to right , a1-a2-a3 ) to synthetic cmds ( middle panels ) generated from our best solutions . the bottom panels show the corresponding lfs . the red dashed line corresponds to the faintest limit used to fit the data . figure [ cmd_synth_jh ] shows the same analysis for the nir cmds ( regions a1 and a2 only ) . simulations for sub - regions a1 and a2 show a good agreement with both the optical and nir cmds . within the portion of the cmds used to fit the data ( see red dashed lines ) , observational and model lfs show deviations that are compatible to the errors ( computed as square root of the observational count rates ) . this suggests that a kroupa imf , which is very similar to a salpeter imf above 0.5 m@xmath2 , is consistent with the data . at fainter magnitudes our simulations systematically underestimate the observational star counts ( more in the optical than the nir ) . this effect is minor in sub - region a1 , but becomes significant in sub - region a2 . one possible explanation for this mismatch is the 0.75 mag selection threshold that we used to reject artificial stars that are actually blended with real stars . relaxing this condition alleviates the issue , but also increases the scatter in the synthetic cmds more than what we see in the data . however , the impact of this effect should be minor in the magnitude range used for fitting . the simulation for sub - region a3 is in less good agreement with the observations . as is visible from the lfs ( see the bottom right panel of fig . [ cmd_synth_vi ] ) , the observed star counts systematically outnumber synthetic counts even at bright magnitudes ( f555ws@xmath62 ) . the straightforward explanation is to attribute the discrepancy to the incompleteness of the region , which is too inhomogeneous to be accurately modeled by our tests . however , from a physical point of view , it is also conceivable that the transition from pms to ms for intermediate / massive stars is not well reproduced by models . indeed , the major discrepancy is in the range f555w 18 - 19 , which corresponds to ton ages younger than 1 myr . if the pms evolutionary time for these objects is overestimated ( so more pms stars were predicted than observed ) or the pms birth - line ( the region in the cmd where stars are still embedded in their gas cocoons and , therefore , still invisible in optical bands ) is closer to the ms than predicted , any attempt to simultaneously fit pms and ms star counts will end up with a bias . in this case , the deficiency of ms stars could suggest that the sf rate for stars younger than 1 myr might be underestimated . finally , the number and colors of rc stars are well reproduced . this indicates that field contamination and reddening are correctly modeled . to test the robustness of the solution against variations in the assumed imf , binary prescription ( fraction and mass ratio ) and distance , we re - derived the sfh of region a1 using alternative values as an example . in figure [ imf_bin ] we compare the results ( shaded orange histograms ) with the standard solution ( shaded green histogram ) . the top - left panel shows the sfh obtained using a shorter distance modulus , ( m@xmath3m)@xmath63 , as suggested by some cepheids studies ( e.g. @xcite ) . the top - right panel shows the sfh obtained with a binary fraction of 60% ( e.g. @xcite ) and mass ratio randomly drawn from a constant distribution between 0.5 and 1 . the bottom - left and bottom - right panels show the sfh obtained using an imf exponent @xmath64 , above @xmath7 , of 1.9 and 2.7 , respectively . overall , all recovered solutions are qualitatively and quantitatively ( within 2-sigma ) consistent , suggesting that our findings are robust . in particular , the onset of the major activity 7 myr ago is unchanged . relatively larger differences are found for imf changes . the predicted rate for s@xmath65 is significantly stronger than the @xmath66 case for ages @xmath67 myr , while for s@xmath68 it is stronger for ages @xmath69 myr . these differences are principally due to the sf rate - imf degeneracy . the lack of stars of lower mass ( caused by the flatter imf ) is compensated with a higher early sf rate , whereas the lack of intermediate and massive stars ( caused by the steeper imf ) is compensated with a higher recent activity . we also found that the synthetic cmds corresponding to the @xmath70 and @xmath71 solutions systematically underestimate and overestimate , respectively , the star - counts for f555w@xmath72 ( the reverse for f555w@xmath73 ) . on the other hand , changing the binary fraction and mass ratio does not affect either the sfh or the fit quality . @xcite recovered the sfh of ngc 2070 using a bayesian approach applied to the ubv photometry down to @xmath74 , corresponding to stars more massive than @xmath75 . the sfh was found to be dominated by three episodes , namely a young peak at @xmath76 myr , an intermediate - age peak at @xmath77 myr , and an old peak @xmath78 myr ago . these three bursts appear to be spatially disjoint , with the youngest stars concentrated towards the center while the intermediate - age stars appear spherically distributed over a 6 pc radius , slightly off - center . the observations are consistent with a sf that propagated inward . @xcite obtained hst / nicmos f160w band images of the central 14 pc @xmath79 14.25 pc around r 136 and combined them with archival wfpc2 f555w and f814w observations ( @xcite ) . by fitting the f160w vs f555w@xmath3f160w cmd with stellar models above @xmath80 @xcite and stellar models below @xmath80 @xcite , they constrained the age of the low mass population to be @xmath81 myr old . @xcite derived the age distribution for a region of @xmath82 arcseconds@xmath30 , at a distance of 4 arcseconds from r 136 , using a synergy of nir photometry with adaptive optics and hst photometry . their investigation was limited to stars more massive than 12 m@xmath83 . the resulting age distribution ( see their fig . 13 ) grows from 7 myr ago up to the present time with three visible bursts at 5 - 6 myr , 3 - 4 myr and 1 myr ago . moreover , they did not find red giants and red super - giants in their fov , concluding that the age of r 136 must be less than 5 myr . according to @xcite the central region can be divided into a core , r 136 , @xmath84 myr old , a peripheral triggered population @xmath85 1 myr old , and a group of late - o and early - b stars 4 - 5 myr old . @xcite used deep hst observations including h@xmath86 for a field of @xmath87 arcmin @xmath88 arcmin enclosing ngc 2070 . they inferred that a significant fraction ( @xmath28 35% ) of the pms stars were formed prior to 12 myr ago , while a similar fraction is younger than 4 myr . in terms of sf , they found that 1 myr ago the region was about 30 times more active than 16 myr ago . using isochrone fitting , @xcite found that the majority of the stars in the `` northeast clump '' , a group of stars a few pc away from r 136 , have ages between 2 and 5 myr ago , while stars in r 136 are at most 2 myr old . except for @xcite and @xcite , our data are generally much deeper than the others in the literature and , therefore , more sensitive to older activity . indeed , our prediction about the beginning of activity in ngc 2070 , about 20 myr ago , is only accessible with our data . moreover , our analysis implements the parsec evolutionary models , which cover consistently the entire evolution from the pms phase to the post - ms phase , whereas all previous analyses had forced to combine models for the pms phase ( @xcite ) and other models for later phases . despite these differences , our sfh is broadly consistent with the aforementioned results . of particular interest is the 7 myr epoch , when our solution predicts a significant sf enhancement . this event is generally consistent , or slightly older , with what is found in all other studies . similarly to @xcite we find evidence of some activity prior to 10 myr ago , whose rate is , at most , one order of magnitude lower than the recent one . from a spatial point of view , the activity shifts to younger ages moving from our region a1 to a3 , resembling the inward progression found by @xcite . this is interesting , given that @xcite and this work use complementary mass ranges , above @xmath75 in the case of @xcite and below @xmath25 in our case . the inward scenario is also broadly consistent with the @xcite , @xcite and @xcite findings . @xcite found that stars younger than 4 myr are more concentrated towards r 136 than stars older than 12 myr . these authors also found a remarkable lack of old activity ( @xmath89 myr ) near r 136 , whereas the old component is not significantly different in our sub - regions . however , the data in our sub - regions a2 and a3 are not deep enough to allow for a conclusive argument . our results predict that stars younger and older than about 10 myr are on average reddened by e(b@xmath3v)@xmath90 and @xmath91 , respectively . because young stars are spatially more concentrated , this also translates into a negative reddening gradient towards the center . this trend is in apparent contrast with findings of @xcite , who found that the young populations ( star forming regions few myr old ) in the lmc are more reddened than the older ones ( @xmath92 1 gyr ) . however , we point out that the timescales here are much shorter , our `` old '' population is only few myr older the young one . besides , all results indicate a large reddening dispersion ( at least 0.1 mag ) . looking at the literature , we find a general consensus for highly variable reddening in the ngc 2070 neighborhood . more specifically , @xcite used ums and rc stars to study extinction around ngc 2070 . once corrected for the foreground contribution ( e(b@xmath3v)=0.07 ) , our average prediction for the young component is very close to the peak of their reddening distribution obtained from ums stars ( see the blue histogram in their fig . 9 ) , while our average prediction for the old component is well within their reddening dispersion . our reddening for the young population is also in good agreement with the results of @xcite . these authors used stars more massive than @xmath75 and found average extinctions in the range @xmath93 ( e(b@xmath3v)=0.35 - 0.5 ) . overall , ngc 2070 s sfh is similar to that in ngc 346 , the largest star forming region in the small magellanic cloud ( smc ) , whose major activity started about 6 - 8 myr ago and peaked about 3 myr ago before dropping to a lower level ( see @xcite ) . if we exclude the core of ngc 2070 , which is ten times more active than ngc 346 , even the peak rate of region a3 is similar in amplitude to the peak rate in ngc346 ( about @xmath94 between 4 and 5 myr ago ; @xcite ) . @xcite analysed the cmds of eight young massive lmc clusters and derived upper limits on potential age spreads by fitting gaussian profiles to their sfhs . they found age spreads smaller than a few myr for the youngest clusters ( 20 60 myr old ) , which is consistent with our much more detailed analysis for ngc 2070 . outside the local group , resolving older populations in sscs becomes very challenging . @xcite analysed seven young massive star clusters ( 5 50 myr old ) in five nearby galaxies , located at distances of 3 5 mpc . they found that the simulated cmds generated with padova isochrones @xcite and single burst sfh have problems to reproduce : 1 ) the observed separation in the cmds between ms and he - burning `` blue loop '' stars , 2 ) the ratio between red and blue supergiants and , in some clusters , 3 ) the scatter in the luminosities of the supergiant stars ( scatter that can not be explained by observational errors alone ) . the authors could improve the fit by including an age spread of 1030 myr in model clusters . we have presented a detailed analysis of the star formation history in the starburst cluster ngc 2070 , located in the heart of 30 doradus in the lmc , using deep optical and nir cmds from the hubble tarantula treasury project . we used a new synthetic cmd approach combined with the latest padova models ( parsec ) , the first to cover homogeneously all stellar phases from pms to post - ms . + in our implementation we encountered a number of interesting challenges . we summarize here our main conclusions : + _ sfh : _ we found that ngc 2070 experienced prolonged activity , starting at least 7 myr ago . we identify three major events in the history of this cluster : * @xmath0 myr ago - this epoch demarcates the commencement of the first significant period of sf . prior to this epoch , local activity is not distinguishable from the average activity in the lmc field ; * 7 myr ago - the sf accelerated throughout the entire region ; * 1 - 3 myr ago the activity reached a peak . in this time range , the sf moves from the periphery to the central regions . our innermost region ( a3 ) shows a maximum activity 1 - 2 myr ago ; _ stellar mass : _ we estimate the stellar mass of ngc 2070 out of 20 pc to be @xmath95 . this value is close to the median mass of galactic globular clusters ( @xmath96 ; @xcite ) . _ reddening : _ concerning the reddening distribution , we find an average e(b@xmath3v)@xmath52 mag for the young population ( @xmath8510 myr old ) , and @xmath51 mag for the old one . an explanation could be that only in the last few myr the sf has been vigorous enough to sweep away part of the gas through stellar winds . another possibility is that the old activity took place on the far side of the ngc 2070 nebula as seen from us , therefore those stars experience most of the line of sight optical depth . _ imf : _ except the innermost few pc , where the incompleteness is too severe to allow firm conclusions , a kroupa imf down to @xmath97 is compatible with the data . this corroborates and extends the result of @xcite to sub - solar masses . to the level we can measure low - mass stars can form in starburst clusters in the same way they form in low density environments . we would like to thank mario gennaro , nino panagia , martha boyer , chris evans , jay gallagher , karl gordon , anton koekemoer , soren larsen and selma de mink for helpful comments , discussions , and contributions to other aspects of http . e.k.g . gratefully acknowledges funding via the collaborative research center `` the milky way system '' ( sfb881 ) of the german research foundation ( dfg ) , particularly via subproject b5 . . has been partially funded by the italian prin - miur grant 2010ly5n2 t . d.a.g . kindly acknowledges financial support by the german research foundation ( dfg ) through grant go1659/3 - 2 . support for program # 12939 was provided by nasa to the us team members through a grant from the space telescope science institute , which is operated by aura , inc . , under nasa contract nas 5 - 26555 . as done previously ( @xcite ) , we parametrize the synthetic cmd as a linear superposition of basic cmds generated with step - function - like star formation , metallicity and reddening . in the specific case of this paper we deal with 19 steps in age and 20 steps in reddening , for a total of 380 parameters . to explore this wide parameter space we have combined a well tested genetic algorithm ( ga ) , pikaia , with a local search routine . as shown in various papers ( see , e.g. , @xcite , @xcite , @xcite ) , gas allow to find global optimum more efficiently than a local search alone . in our approach the synergy of the ga and a local search combines the advantages of both worlds . in the next sections we describe the synthetic population code and the optimization routine ( ga@xmath44local search ) finally , the capabilities of the approach are tested with artificial data . the basic synthetic cmds@xmath98 are populated through the following monte carlo procedure : 1 ) synthetic masses and ages are extracted from the assumed imf and the j - th sf step , respectively ; and 2 ) synthetic masses and ages are converted to absolute synthetic magnitudes and colors by using a fine grid of isochrones . for our calculations we used the latest ( v.1.2s ) parsec isochrones , covering the entire mass spectrum of @xmath99 from the pms phase to the early - agb phase ; 3 ) a fraction q of synthetic stars is randomly chosen to have a companion . the masses of companions are extracted from the same imf and their flux is added to the flux of the primaries ; 4 ) the absolute synthetic photometry is put at the distance of the lmc and reddened with the k - th reddening . to produce realistic simulations , all basic cmds are degraded with photometric errors and incompleteness as estimated from artificial star tests ( see section [ artifix ] ) . to have all stellar phases well populated , the synthetic cmds are generated with a large number of stars ( @xmath100 ) . once constructed , the basic cmds@xmath98 are binned in n bins of color and m bins of magnitude . the final result is a library of j@xmath79k 2d histograms cmd@xmath101 and any cmd can be expressed as a linear combination of these 2d histograms ( see equation [ eq1 ] ) . the coefficients @xmath102 that multiply each of the cmd@xmath101 are the star formation rate at the time step j and reddening step k. the sum over j and k of @xmath103 provides the total star - counts predicted @xmath104 in the cmd bin ( m , n ) by the star formation @xmath102 . @xmath105\end{aligned}\ ] ] including the reference field in the models corresponds to changing equation [ eq1 ] into equation [ eq2 ] : @xmath106+\\+s_{f}(k)\times f_{m , n}(k)]\nonumber\end{aligned}\ ] ] where s@xmath107(k ) regulates the number of field stars with reddening k - th , while f@xmath108 is the actual number of stars in the reference field ( reddened with the k - th reddening ) in the cmd bin ( m , n ) . once the observational cmd is binned as well , the next step is to search for the combination of basic cmds that minimizes the cmd residuals between data and model . for this task we implemented a likelihood distance , whose minimization is not biased by low count statistics . the combination of basic cmds that minimizes the likelihood corresponds to the most likely sfh behind the data . the uncertainty around the recovered best solution will be the sum in quadrature of a statistical error , obtained through a data bootstrap , and a systematic error , obtained by re - deriving the sfh using different age - binnings and cmd - binning . in our approach the likelihood is minimized with an hybrid - genetic algorithm ( hga ) , which combines a classical ga with a local search . pure gas are iterative probabilistic algorithms for solving a problem that mimic processes found in the natural biological evolution . compared to local search algorithms , gas explore the search space in more points simultaneously , hence they are far less sensitive to the initial conditions and show remarkable ability to escape from local minima . shortcomings of gas are the weak ability for local exploration and the slow convergence rate . the proposed hga aims to overcome both by alternating two phases : a ga , whose goal is to search for a quasi - global solution , and a local search , whose goal is to increase solution accuracy . the synergy of the two incorporates the exploration ability of gas and the exploitation ability of local search algorithm . for this study we implemented both _ parallel _ and _ serial _ hybridization . the parallel one is applied to each iteration and aims at enhancing offspring likelihood by means of a local search before moving to the next generation . the serial one consists in applying a final local search after which the ga population has evolved to the region containing the global solution . to test our approach we have generated synthetic data with features resembling ngc 2070 ( 10000 stars brighter than f555w@xmath109 ) . to simulate the observational conditions , this fake population is convolved with photometric uncertainties and incompleteness from the artificial star tests of ngc 2070 . the input sfh is a sequence of five bursts of different ages and duration 1 myr , while the input reddening distribution is built with 50% of the stars with e(b@xmath3v ) between 0.20 and 0.25 , and 25% between 0.25 and 0.30 , 25% between 0.15 and 0.20 . figure [ test ] shows the result of the reconstruction . the input sfh and reddening distribution are in black ; the best reconstructed counterparts in red . the recovered sfh looks like a smoothed version of the input sfh , due to the finite time resolution associated with the discrete nature of the data , and the ill - conditioned nature of mathematical inverse problems . however , overall , input functions are successfully recovered , with most of the differences within the uncertainties . moreover , a bursty sfh is recovered more accurately if the time gap between the individual bursts is longer .

let be a general primitively polarized surface of genus and let be the arithmetic genus of we prove the existence in of curves with a triple point and-singularities . in particular , we show the existence of curves of geometric genus in with a triple point and nodes as singularities and corresponding to regular points of their equisingular deformation locus , for every and our result is obtained by studying the versal deformation space of a non - planar quadruple point .

1111.3051

let @xmath10 be a complex smooth projective @xmath1 surface and let @xmath11 be a very ample line bundle on @xmath10 of sectional genus @xmath12 the pair @xmath0 is called a _ primitively polarized @xmath1 surface of genus @xmath13 _ it is well - known that a ( very ) general such pair satisfies @xmath14.$ ] we denote by @xmath15 the _ zariski closure _ of the severi variety of @xmath16-nodal curves , defined as the locus of _ irreducible _ and reduced curves with exactly @xmath16 nodes as singularities . the non - emptiness of these varieties for every @xmath17 where @xmath3 is the arithmetic genus of @xmath18 has been established in @xcite . like the severi variety of @xmath16-nodal plane curves , @xmath19 has several good properties . by @xcite , we know that it is smooth of expected dimension at every point @xmath20 $ ] corresponding to a @xmath21-nodal curve . this implies that every irreducible component @xmath22 of @xmath19 has codimension @xmath21 in @xmath23 and it is contained in at least one irreducible component of the severi variety @xmath24 moreover , it is also known that @xmath19 coincides with the variety @xmath25 defined as the zariski closure of the locus of reduced and irreducible curves of geometric genus @xmath7 ( cf . * lemma 3.1 ) and ( * ? ? ? * remark 2.6 ) ) . unlike the severi variety of @xmath16-nodal plane curves , nothing is known about the irreducibility of @xmath26 classically , the irreducibility problem of severi varieties is related to the problem of the description of their boundary ( cf . @xcite ) . _ let @xmath27 be an irreducible component and let @xmath28 be the locus of @xmath16-nodal curves . what is inside the boundary @xmath29 ? does @xmath22 contain divisors @xmath30 , @xmath31 , @xmath32 and @xmath33 whose general element corresponds to a curve with @xmath34 nodes , a tacnode and @xmath35 nodes , a cusp and @xmath36 nodes and a triple point and @xmath37 nodes , respectively ? _ because of the literature about severi varieties of plane curves ( cf . @xcite , @xcite and @xcite ) , the divisors @xmath30 , @xmath31 , @xmath32 and @xmath33 of @xmath27 , when non - empty , are expected to play an important role in the description of the picard group @xmath38 or , more precisely , of the picard group of a `` good partial compactification '' of the locus @xmath28 of @xmath16-nodal curves . proving the existence of these four divisors in _ every _ irreducible component @xmath22 of @xmath19 is a very difficult problem . the first progress towards answering the previous question has been made in @xcite . in that paper , the authors prove the existence of irreducible curves in @xmath23 of every allowed genus with a tacnode or a cusp and nodes as further singularities . this article is devoted to the existence of irreducible curves in @xmath23 of geometric genus @xmath8 with a triple point and nodes as further singularities . before introducing our result , we make some observations concerning problem 1 , describing the type of singularity of @xmath22 along @xmath30 , @xmath31 , @xmath32 and @xmath39 whenever these loci are non - empty . let us denote by @xmath40 , @xmath41 and @xmath42 the zariski closure of the locus in @xmath23 of reduced and irreducible curves of geometric genus @xmath7 with a tacnode , a cusp and a triple point , respectively , and nodes as further singularities . let @xmath43 be an irreducible component of any of these varieties or of @xmath44 then , by ( * ? ? ? * lemma 3.1 ) , we have that @xmath45 thus @xmath43 is a divisor in at least one irreducible component @xmath22 of @xmath46 and , by using the same arguments as in ( * ? ? ? * section 1 ) , we also know what @xmath22 looks like in a neighborhood of the general point of @xmath47 if @xmath43 is an irreducible component of @xmath48 then @xmath49 has an ordinary multiple point of order @xmath34 at the general point of @xmath47 in particular , at least in principle , there may not be a unique irreducible component @xmath22 of @xmath49 containing @xmath47 on the contrary , if @xmath43 is any irreducible component of @xmath40 , @xmath41 or @xmath50 then @xmath43 is contained in only one irreducible component @xmath22 of @xmath49 . in particular , @xmath49 is smooth at the general point of every irreducible component of @xmath40 and @xmath42 and it looks like the product of a cuspidal curve and a smooth @xmath51-dimensional variety in a neighborhood of the general point of an irreducible component of @xmath52 we finally observe that , unlikely the case of severi varieties of plane curves , if @xmath53 is an irreducible component , a priori , there may exist divisors in @xmath22 parametrizing curves with worse singularities than an ordinary triple point , a tacnode or a cusp and nodes . we now describe the results of this paper . our theorem about non - emptiness of @xmath42 is based on the following local problem . _ let @xmath54 be a smooth family of surfaces with smooth general fiber @xmath55 and whose special fiber @xmath56 is reducible , having two irreducible components intersecting transversally along a smooth curve @xmath57 what kind of curve singularities on @xmath58 at a point @xmath59 may be deformed to a triple point on @xmath55 ? _ in the first part of section [ sect : getting triple point ] , we prove that , if a triple point on @xmath55 specializes to a general point @xmath59 along a smooth bisection @xmath60 of @xmath61 then the limit curve singularity on @xmath58 is a space quadruple point , union of two nodes having one branch tangent to @xmath62 the result is obtained by a very simple argument of limit linear systems theory , with the same techniques as in @xcite . in lemma [ lemma : analytic - type ] we find the analytic type of this quadruple point . this allows us to compute the versal deformation space of our limit singularity and to prove that , under suitable hypotheses , it actually deforms to an ordinary triple point singularity on @xmath63 see theorem [ th : triple - point ] . in the last section , we consider the case that @xmath55 is a general primitively polarized @xmath1 surface and @xmath64 and @xmath65 are two rational normal scrolls . in lemma [ limiti ] , we prove the existence on @xmath66 of suitable curves with a non - planar quadruple point as above , tacnodes ( of suitable multiplicities ) and nodes and that may be deformed to curves in @xmath67 with an ordinary triple point and nodes as singularities . in particular , this existence result is obtained as a corollary of the following theorem , which is to be considered the main theorem of this paper . [ main - theorem ] let @xmath0 be a general primitively polarized @xmath1 surface of genus @xmath68 then , for every @xmath69 and for every @xmath70-tuple of non - negative integers @xmath71 such that @xmath72 if @xmath73 is even , or @xmath74 if @xmath75 is odd , there exist reduced irreducible curves @xmath76 in the linear system @xmath23 on @xmath10 such that : * @xmath76 has an ordinary triple point , @xmath77 nodes and @xmath78 singularities of type @xmath79 for every @xmath80 and no further singularities ; * @xmath76 corresponds to a regular point of the equisingular deformation locus @xmath81 equivalently , @xmath82}es(c))=0.$ ] finally , the singularities of @xmath76 may be smoothed independently . in particular , under the hypotheses and , for every @xmath83 and for every @xmath84 there exist curves @xmath76 in the linear system @xmath23 on @xmath10 with an ordinary triple point , @xmath85 singularities of type @xmath79 for every @xmath86 and @xmath16 nodes as further singularities and corresponding to regular points of their equisingular deformation locus . by corollary [ primitive - corollary ] , the family in @xmath87 of curves with a triple point and @xmath85 singularities of type @xmath88 is non - empty if it has expected dimension at least equal to one or it has expected dimension equal to zero and @xmath89 i would like to thank m. giusti for offering to send me personal notes of @xcite during the holiday time of my library . my intention to write this paper became stronger after a conversation with c. ciliberto about related topics . i also benefited from conversation with t. dedieu and a. l. knutsen . finally , i would like to thank the referee for many helpful comments and corrections . throughout the paper an irreducible curve @xmath76 will be a reduced and irreducible projective curve , even if not specified . we will be concerned here only with reduced and locally complete intersection curves @xmath76 in a linear system @xmath90 on a ( possibly singular ) surface @xmath91 having , as singularities , space singularities , plane ordinary triple points or plane singularities of type @xmath92 we recall that an ordinary triple point has analytic equation @xmath93 while an @xmath6-singularity has analytic equation @xmath94 singularities of type @xmath95 are nodes , @xmath96-singularities are cusps and an @xmath97-singularity is a so called @xmath98-tacnode . whenever not specified , a tacnode will be a @xmath99-tacnode , also named a simple tacnode . for curves with ordinary plane triple points and plane singularities of type @xmath100 equisingular deformations from the analytic point of view coincide with equisingular deformations from the zariski point of view ( cf . * definition ( 3.13 ) ) ) . because of this , given a curve as above , we define the equisingular deformation locus @xmath101 as the zariski closure of the locus of analytically equisingular deformations of @xmath76 , without discrepancy with classical terminology . finally , we recall that , if @xmath102 is a smooth surface and @xmath76 has as singularities an ordinary triple point and @xmath78 singularities of type @xmath100 then the dimension of the tangent space to @xmath103 at the point @xmath20 $ ] corresponding to the curve @xmath76 is given by @xmath82}es(c))\geq\dim|d|-4-\sum_kd_k k$ ] ( cf . if equality holds , we say that _ @xmath103 is regular at _ @xmath20.$ ] the regularity of @xmath103 is a sufficient and necessary condition for the surjectivity of the standard morphism @xmath104 in this case , we say that _ the singularities of @xmath76 may be smoothed independently _ , with obvious meaning because of the versality properties of @xmath105 let @xmath54 be a smooth family of projective complex surfaces with smooth general fiber @xmath55 and whose special fiber @xmath56 has two irreducible components , intersecting transversally along a smooth curve @xmath57 let @xmath106 be a cartier divisor and let us set @xmath107 we want to find sufficient conditions for the existence of curves with a triple point in @xmath108 we first ask the following question . assume that , for general @xmath109 , there exists a reduced and irreducible divisor @xmath110 degenerates to a curve @xmath111 in such a way that the triple point of @xmath112 comes to a general point @xmath113 we ask for the type of singularity of @xmath114 at @xmath115 actually we do not want to find all possible curve singularities at @xmath116 that are limit of a triple point on @xmath117 we only want to find a suitable limit singularity . we first observe that , since @xmath116 belongs to the singular locus of @xmath58 and @xmath118 is smooth at @xmath119 there are no sections of @xmath54 passing through @xmath115 thus the triple point of the curve @xmath120 as @xmath121 must move along a multisection @xmath122 of @xmath123 in order to deal with a divisor @xmath10 in a smooth family of surfaces @xmath124 with a triple point at the general point of a section of @xmath125 we make a base change of @xmath123 let @xmath124 be the smooth family of surfaces @xmath126\ar[dr ] & \mathcal x^\prime \ar[d]\ar[r ] & \mathcal x\ar[d]\\ & \mathbb a^1\ar[r]^{\nu_2}&\mathbb a^1 } \ ] ] obtained from @xmath54 after a base change of order two totally ramified at @xmath127 and by normalizing the achieved family . now the family @xmath128 has general fibre @xmath129 and special fibre @xmath130 , where , by abusing notation , @xmath131 and @xmath132 are the proper transforms of @xmath131 and @xmath132 in @xmath118 and @xmath133 is a @xmath134-bundle on @xmath62 in particular , @xmath135 and @xmath136 are two sections of @xmath133 isomorphic to @xmath62 denote by @xmath137 the fibre @xmath133 corresponding to the point @xmath138 and let @xmath60 be a section of @xmath128 intersecting @xmath137 at a general point @xmath139 assume there exists a divisor @xmath140 having a triple point at the general point of @xmath141 _ step 1 . _ let @xmath142 be the blow - up of @xmath128 along @xmath60 with new exceptional divisor @xmath143 and special fibre @xmath144 where @xmath145 is the blow - up of @xmath133 at @xmath139 still denoting by @xmath137 the proper transform of @xmath146 in @xmath147 we have that @xmath137 has self - intersection @xmath148 on @xmath149 morever , if @xmath150 is the proper transform of @xmath10 in @xmath147 we have that @xmath151 we deduce that @xmath152 and hence @xmath153 _ step 2 . _ let now @xmath154 be the blow - up of @xmath155 along @xmath137 with new exceptional divisor @xmath156 and new special fibre @xmath157 where @xmath158 and @xmath159 are the blow - ups of @xmath131 and @xmath132 at @xmath160 and @xmath161 respectively . denoting again by @xmath137 the proper transform of @xmath162 in @xmath163 we have that @xmath137 is the @xmath164-curve of @xmath165 moreover , if @xmath166 is the proper transform of @xmath150 in @xmath163 by , we deduce that @xmath167 where @xmath168 is the fibre of @xmath169 and @xmath170 is the multiplicity of @xmath150 along @xmath171 furthermore , since @xmath172 is an effective divisor , we have that @xmath173 in particular , if @xmath174 then @xmath175 contains @xmath137 in the base locus with multiplicity @xmath176 hence @xmath177 with @xmath178 using again that @xmath166 is a cartier divisor , we find that @xmath179 ( resp . @xmath180 ) has two smooth branches intersecting @xmath181 transversally at @xmath182 and @xmath183 ( resp . @xmath184 and @xmath185 ) , as in figure 1 . now let @xmath186 be the image of @xmath187 if @xmath109 is a general point of @xmath188 and @xmath189 , then the fibre of @xmath190 over @xmath109 is @xmath191 where @xmath192 is the fibre of @xmath10 over @xmath193 while the special fibre @xmath194 of @xmath190 is the image curve , counted with multiplicity @xmath195 of the special fibre @xmath196 of @xmath10 under the contraction of @xmath133 ( see figure 2 ) . the curve @xmath197 belongs to the subvariety @xmath198 defined as the zariski closure of the locus of curves @xmath199 where @xmath200 and @xmath201 have a node at @xmath202 with one branch tangent to @xmath62 notice that every such curve @xmath76 has a non - planar quadruple point at @xmath116 and that @xmath203 is a linear subspace of @xmath204 of codimension at most @xmath205 [ lemma : analytic - type ] let @xmath20\in q_p$ ] be a point corresponding to a curve @xmath206 with a non - planar quadruple point at @xmath116 as in the previous remark . then the analytic equations of @xmath76 at @xmath116 are given by @xmath207 we want to observe that , as for plane curve singularities , the classification of simple complete intersection space singularities is known , ( cf . the singularity defined by is not a simple singularity . before proving lemma [ lemma : analytic - type ] , we need to recall a basic result about space curve singularities . we refer to @xcite and use the same notation and terminology . let @xmath208 be the set of positive rational numbers . a polynomial @xmath209 $ ] is said to be quasi - homogeneous of type @xmath210 if @xmath211 is a linear combination of monomials @xmath212 such that : @xmath213 the @xmath214-tuple @xmath215 is called a system of weights . we shall also say that @xmath211 has degree @xmath216 if every variable @xmath217 has weight @xmath218 for every @xmath219 for every fixed system of weights @xmath220 we will denote by @xmath221 the induced graduation @xmath222=\oplus_{d\geq 0}g_d^{\underline a}\ ] ] on @xmath223,$ ] where @xmath224 is the set of quasi - homogeneous polynomials of type @xmath225 an element @xmath226)^p$ ] is quasi - homogeneous of type @xmath227 if , for every @xmath228 , the component @xmath229 of @xmath230 is of type @xmath231 the @xmath232-tuple @xmath227 is called a system of degrees and weights . for every fixed system of degrees and weights @xmath233 one defines a graduation @xmath234 on @xmath235)^p$ ] by setting @xmath236)^p=\oplus_{\nu\in\mathbb z}g^{(\underline d,\underline a)}_\nu,\ ] ] where @xmath237)^p|\,g_i\in g^{\underline a}_{d_i+\nu}\}.$ ] moreover , we denote by @xmath238 the valuation naturally associated to @xmath221 and by @xmath239 the valuation associated to @xmath240 and defined by @xmath241,\ ] ] for every @xmath242)^p.$ ] finally , we recall that , if @xmath243 denotes the set of germs of applications @xmath244 such @xmath245 is a germ of a complete intersection analytic variety with an isolated singularity at the origin , then we have the versal deformation space @xmath246 ( cf . @xcite or @xcite ) and , by @xcite , a graduation is naturally induced on @xmath247 [ merle ] let @xmath248 be a quasi - homogeneous element of type @xmath249 . let @xmath250)^p$ ] be any element such that : @xmath251 where @xmath252 then the germs of singularities @xmath253 and @xmath254 are analytically equivalent . let @xmath255\in q_p$ ] be a point associated to a curve @xmath256 where @xmath257 and @xmath258 have a node at @xmath59 with one branch tangent to @xmath57 let @xmath259 be analytic coordinates of @xmath118 at @xmath260 in such a way that @xmath261 and @xmath262 then the analytic equations of @xmath263 at @xmath116 are given by @xmath264 where @xmath265 is the equation of an analytic surface @xmath10 in @xmath266 having a singularity of multiplicity @xmath99 at @xmath267 moreover , the tangent cone of @xmath10 at @xmath116 must contain the line @xmath268 thus , up to a linear transformation , we may assume that the analytic equations of @xmath263 at @xmath116 are given by @xmath269 where @xmath270 is a polynomial of degree at least @xmath271 with no @xmath272-term . we want to prove that the complete intersection curve singularity given by is analytically equivalent to the curve singularity given by the equations . we first observe that the element @xmath273 is quasi - homogeneous of type @xmath274 moreover , every term of the polynomial @xmath275 has degree strictly greater than @xmath276 if the variables @xmath277 have weights @xmath278 in order to apply proposition [ merle ] , we need to compute @xmath279 let @xmath76 be a reduced cartier divisor on @xmath56 having a singularity of analytical equations given by at @xmath116 . then , since @xmath76 is a local complete intersection subvariety of @xmath280 we have the following standard exact sequence of sheaves on @xmath76 @xmath281 & \theta_c \ar[r ] & \theta_{\mathcal x}|_c \ar[r]^{\alpha } & \mathcal n_{c|\mathcal x } \ar[r ] & t^1_c \ar[r ] & 0 , } \ ] ] where @xmath282 is the tangent sheaf of @xmath283 defined as the dual of the sheaf of differentials of @xmath283 @xmath284 is the tangent sheaf of @xmath118 restricted to @xmath283 @xmath285 is the normal bundle of @xmath76 in @xmath118 and @xmath286 is the first cotangent sheaf of @xmath76 , which is supported at the singular locus @xmath287 of @xmath76 and whose stalk @xmath288 at every singular point @xmath289 of @xmath76 is the versal deformation space of the singularity . we also recall that the global sections of the image sheaf @xmath290 of @xmath291 are , by the versality properties of @xmath292 the infinitesimal deformations of @xmath76 in @xmath118 preserving singularities of @xmath76 and their analytic type . for this reason , @xmath290 is called the _ equisingular deformation sheaf of @xmath76 in @xmath293 _ now observe that @xmath294 in order to compute @xmath295 we use the following standard identifications : * of the local ring @xmath296 of @xmath76 at @xmath116 with @xmath297/(f_1,f_2)$ ] , where @xmath298 and @xmath299 + * of the @xmath300-module @xmath301 with the free @xmath302-module @xmath303 generated by morphisms @xmath304 and @xmath305 , defined by @xmath306 and , finally , * of the @xmath300-module @xmath307 with the free @xmath302-module generated by the derivatives @xmath308 with these identifications , the localization map @xmath309 at @xmath116 of the sheaf map @xmath291 in is defined by @xmath310 and , similarly , for @xmath311 and @xmath312 in particular , we have that @xmath313 it follows that the versal deformation space of the non planar quadruple point of @xmath76 at @xmath116 is the affine space @xmath314 in particular , @xmath315 ( in accordance with ( * ? ? ? * proposition on p. 165 ) ) and , if we fix affine coordinates @xmath316 on @xmath317 then the versal deformation family @xmath318 has equations @xmath319 furthermore , by the equality @xmath320 we deduce that the graduation on @xmath321 induced by @xmath322 is given by @xmath323 in particular , we obtain that @xmath324 by proposition [ merle ] , using that @xmath325 we get that the singularity defined by is analytically equivalent to the singularity of equations , as desired . [ eq - eq - in ] let @xmath326 be a curve as in lemma [ lemma : analytic - type ] . by the proof of the lemma , we may deduce the equations at @xmath116 of an equisingular infinitesimal deformation of @xmath76 in @xmath293 indeed , by the equalities , we have that , if @xmath327 then the equations of @xmath328 at @xmath116 are given by @xmath329 where @xmath330 now @xmath331 is the equation of an infinitesimal deformation @xmath58 vanishing along the singular locus @xmath62 by ( * ? ? ? * section 2 ) , we know that these infinitesimal deformations are the infinitesimal deformations of @xmath58 preserving the singular locus . since @xmath58 is the only singular fibre of @xmath118 , we have that @xmath332 in . since this does not depend on the type of singularity of @xmath76 at @xmath119 we deduce that @xmath333 moreover , we obtain that @xmath334 and @xmath335 are polynomials with no @xmath336-terms , for every @xmath337 and such that @xmath338 and no @xmath339-terms ( resp . @xmath340-terms ) appear in @xmath335 ( resp . @xmath334 ) . thus the equations of @xmath341 at @xmath116 are given by @xmath342 where we set @xmath343 @xmath344 and @xmath345 with @xmath346 finally , @xmath347 is a linear space of codimension @xmath348 in @xmath349 contained in the linear system @xmath350 of curves @xmath351 with @xmath352 @xmath353 and @xmath354 and @xmath355 tangent to @xmath356 at @xmath115 we may now provide sufficient conditions for the existence of curves with a triple point and possibly further singularities in the linear system @xmath357 let @xmath118 be a family of surfaces as above and let @xmath358 be a cartier divisor . let @xmath359 be the locally closed subset defined as follows @xmath360,t)\mid\,\mathcal x_t\,\textrm{is smooth,}\ , d^\prime\cap \mathcal x_t:=d_t^\prime\in\,\mid\mathcal o _ { \mathcal x_t}(d_t)\mid\ , \textrm{is irreducible}\\ \textrm{of genus}\,g,\,\textrm{with a triple point and nodes as singularities}.\}\end{aligned}\ ] ] there is a naturally defined rational map @xmath361 where @xmath362 is the relative hilbert scheme of @xmath123 we will denote by @xmath363 the zariski closure of the image of @xmath364 and we will name it the universal severi variety of curves of genus @xmath7 in @xmath90 with a triple point and nodes . the restriction of this variety @xmath365 where @xmath366 is general , is the severi variety of genus @xmath7 curves in @xmath367 with a triple point and nodes as singularities . we observe that , if @xmath368 is an irreducible component , then @xmath22 coincides with the equisingular deformation locus @xmath369 of the curve @xmath76 corresponding to the general point @xmath20\in v,$ ] defined in section [ notation - terminology ] . [ th : triple - point ] with the notation above , let @xmath370 be a reduced divisor in the linear system @xmath204 such that @xmath200 and @xmath201 have a node at a general point @xmath116 of @xmath371 with one branch tangent to @xmath62 suppose that @xmath372 and @xmath373 are smooth and they intersect @xmath356 transversally outside @xmath115 assume , moreover , that : * @xmath374 for every @xmath375 * @xmath376 for a general @xmath375[liscezza_schema_hilbert ] * @xmath377[expected_dimension ] then the universal severi variety @xmath378 is non - empty . more precisely , the point @xmath20\in\mathcal h^{\mathcal x|\mathbb a^1}$ ] corresponding to the curve @xmath76 belongs to an irreducible component @xmath379 of the special fibre @xmath380 of @xmath381 contained in @xmath380 with multiplicity @xmath382 finally , @xmath383 is smooth at @xmath20 $ ] and the irreducible component @xmath384 of the general fibre @xmath385 of @xmath386 specializing to @xmath387 has expected dimension in @xmath357 we want to obtain curves in the linear system @xmath388 with a triple point as deformations of @xmath389 the scheme parametrizing deformations of @xmath76 in @xmath118 is an irreducible component @xmath390 of the relative hilbert scheme @xmath362 of the family @xmath293 in particular , by ( * ? ? ? * proposition 4.4.7 ) and the hypotheses @xmath391 and @xmath392 , we have that @xmath390 is smooth at the point @xmath20\in\mathcal h$ ] corresponding to @xmath393 now , by hypothesis , @xmath76 has a non - planar quadruple point at @xmath116 , nodes on @xmath394 and no further singularities . moreover , it is well - known that , no matter how we deform @xmath76 to a curve on @xmath63 the nodes of @xmath76 on @xmath356 are smoothed ( see , for example , ( * ? ? ? * section 2 ) ) . we want to prove that @xmath76 may be deformed to a curve on @xmath55 in such a way that the non - planar quadruple point of @xmath76 at @xmath116 is deformed to a triple point . this will follows from a local analysis . first recall that , by lemma [ lemma : analytic - type ] , we may choose analytic coordinates @xmath259 of @xmath118 at @xmath116 in such a way that the equations of @xmath76 at @xmath116 are given by and the versal deformation family @xmath395 has equations given by , where @xmath316 are the affine coordinates on @xmath396 by versality , denoting by @xmath397 the universal family parametrized by @xmath398 there exist tale neighborhoods @xmath399 of @xmath20 $ ] in @xmath398 @xmath400 of @xmath116 in @xmath401 and @xmath402 of @xmath403 in @xmath288 and a map @xmath404 so that the family @xmath405 is isomorphic to the pull - back of @xmath406 , with respect to @xmath407 , @xmath408 & \mathcal c_p |_{v_p } \ar[d ] \ar[l ] & u_p\times_{v_p}\mathcal c_p |_{v_p } \ar[l ] \ar[r]^{\hspace{0.5cm}\simeq } \ar[dr ] & \mathcal d|_{u_p}\cap u_p^\prime \ar[r ] \ar[d ] & \mathcal d \ar[d ] \\ t^1_{c , p } & v_p \ar[l ] & & \ar[ll]_{\phi_p } u_p \ar[r ] & \mathcal h.}\ ] ] we need to describe the image @xmath409 first we want to prove that @xmath410 where @xmath411 obviously , @xmath412 to prove that the equality holds , it is enough to show that the differential @xmath413}:h^0(c,\mathcal n_{c|\mathcal x_0})\to t_{\underline 0}\gamma$ ] is surjective . by standard deformation theory , by identifying the versal deformation space of a singularity with its tangent space at @xmath414 the differential of @xmath407 at @xmath20 $ ] can be identified with the map @xmath415}:h^0(c,\mathcal n_{c|s})\to h^0(c , t^1_c)\to t^1_{c , p},\ ] ] induced by the exact sequence . in particular , using that , by remark [ eq - eq - in ] , @xmath416 and that the nodes of @xmath76 on @xmath356 are necessarily preserved when we deform @xmath76 on @xmath417 we have that @xmath418}=h^0(c,\mathcal n^\prime_{c|\mathcal x}).\ ] ] now , again by remark [ eq - eq - in ] , we know that @xmath419 actually , by the hypothesis @xmath420 we have that @xmath421 has the expected dimension and hence the equality is verified . using that @xmath422 is a hyperplane in @xmath423 and that @xmath424 this implies , in particular , that @xmath425 is a subvariety of dimension @xmath426 smooth at @xmath427 we want to determine the equations of tangent space @xmath428 for this purpose , it is enough to find the image by @xmath429 of the infinitesimal deformation @xmath430 having equations @xmath431 the image of @xmath432 is trivially the vector @xmath433 we deduce that @xmath434 now we want to prove that the locus in @xmath435 of points corresponding to curves with an ordinary triple point as singularity is not empty and its zariski closure is a smooth curve @xmath436 tangent to @xmath437 at @xmath427 this will imply the theorem by versality and by a straightforward dimension count . because of smoothness of @xmath438 at @xmath414 it is enough to prove that the locus of points @xmath439 with @xmath440 and corresponding to a curve with an ordinary triple point is not empty and its zariski closure is a smooth curve @xmath436 tangent to @xmath437 at @xmath427 let @xmath441 be a point with @xmath442 then the fibre @xmath443 of the versal family @xmath395 has equation @xmath444 or , equivalently , @xmath445 now a point @xmath446 is a singular point of multiplicity at least three if and only if @xmath447 by the hypothesis @xmath440 and the equalities and , we find that @xmath448 by substituting @xmath449 in the equalities , , , and @xmath450 we find that @xmath451 conversely , for every point @xmath441 such that @xmath452 @xmath453 and @xmath454 we have that the corresponding curve has a triple point at @xmath455 with tangent cone of equation @xmath456 and no further singularities . the curve @xmath457 is smooth and tangent to @xmath458 at @xmath427 the following corollary is a straight consequence of the proof of theorem [ th : triple - point ] . [ ingenerale ] let @xmath118 be a family of regular surfaces and @xmath358 a cartier divisor as in the statement of theorem [ th : triple - point ] . let @xmath459 be any reduced curve with a space quadruple point of equations at a point @xmath59 and possibly further singularities . then , using the same notation as in the proof of the previous theorem , the image @xmath460 of the morphism @xmath461&t^1_{c^\prime , p}}\ ] ] is contained in the @xmath462-dimensional plane of equations @xmath463 if @xmath464 then there exist deformations @xmath465 of @xmath76 on @xmath55 with a triple point , obtained as deformation of the singularity of @xmath76 at @xmath115 this section is devoted to the proof of theorem [ main - theorem ] . we will prove the theorem by using the very classical degeneration technique introduced in @xcite . let @xmath0 be a general primitively polarized @xmath1 surface of genus @xmath466 in @xmath467 we will degenerate @xmath10 to a union of two rational normal scrolls @xmath468 on @xmath469 we will prove the existence of suitable curves @xmath470 with a space quadruple point given by equations , tacnodes and nodes . finally , we will show that the curves @xmath76 deform to curves on @xmath10 with the desired singularities . we first explain the degeneration argument , introducing notation . fix an integer @xmath471 and set @xmath472 . let @xmath473 be a smooth elliptic normal curve of degree @xmath474 . consider two general line bundles @xmath475 . we denote by @xmath476 and @xmath477 the ( unique ) rational normal scrolls of degree @xmath478 in @xmath479 defined by @xmath480 and @xmath481 , respectively . notice that @xmath482 if @xmath2 is odd whereas @xmath483 if @xmath2 is even . moreover , @xmath476 and @xmath477 intersect transversally along the curve @xmath484 which is anticanonical in each @xmath485 ( cf . * lemma 1 ) ) . more precisely , for odd @xmath2 , where @xmath486 , we let @xmath487 and @xmath488 on @xmath485 be the generators of @xmath489 , with @xmath490 . for even @xmath2 , where @xmath483 , we let @xmath491 be the section of negative self - intersection and @xmath492 be the class of a fiber . then the embedding of @xmath485 into @xmath493 is given by the line bundle @xmath494 for @xmath490 . let now @xmath495 and let @xmath496 be the component of the hilbert scheme of @xmath493 containing @xmath469 . then we have that @xmath497 and , by ( * ? ? ? * theorems 1 and 2 ) , the general point @xmath498\in\mathcal u_{\p}$ ] represents a smooth , projective @xmath1 surface @xmath10 of degree @xmath499 in @xmath493 such that @xmath500=\mathbb z[h].$ ] we denote by @xmath501 a general deformation of @xmath502 over a one - dimensional disc @xmath503 contained in @xmath496 . in particular , the general fiber is a smooth projective @xmath1 surface @xmath504 in @xmath479 with @xmath505 $ ] . now @xmath506 is smooth except for @xmath507 rational double points @xmath508 lying on @xmath62 in particular , @xmath508 are the zeroes of the section of the first cotangent bundle @xmath509 of @xmath510 determined by the first order embedded deformation associated to @xmath511 @xcite . by blowing - up @xmath190 at these points and by contracting the corresponding exceptional components ( all isomorphic to @xmath512 ) on @xmath513 we get a smooth family of surfaces @xmath514 such that @xmath515 and @xmath516 where @xmath517 is the blowing - up of @xmath477 at the points @xmath518 with new exceptional curves @xmath519 we will name @xmath508 _ the special points of _ @xmath62 1 . @xmath526 and @xmath527 where @xmath528 @xmath529 and @xmath530 are smooth rational curves with : * @xmath531 @xmath532 and @xmath533 if @xmath75 is odd , * @xmath534 @xmath535 and @xmath536 if @xmath73 is even , + for every @xmath537 and @xmath538 + 2 . there exist distinct points @xmath539 where @xmath116 is a general point and @xmath540 are determined by the following relations : * if @xmath75 is odd , @xmath490 and @xmath541 then @xmath542 * if @xmath73 is even , @xmath490 and @xmath541 then @xmath543 3 . the singularities of @xmath76 on @xmath544 are nodes , @xmath76 has a quadruple point analytically equivalent to at @xmath59 and tacnodes and nodes on @xmath545 in particular , @xmath76 has a @xmath546-tacnode at @xmath289 and nodes at @xmath547 if @xmath548 @xmath76 has a @xmath549-tacnode at @xmath550 a simple tacnode at @xmath551 and nodes at @xmath552 and @xmath553 for every @xmath554 if @xmath73 . we first consider the case @xmath75 odd . recall that , in this case , we have that @xmath555 @xmath556 where @xmath557 and @xmath558 are the two rulings on @xmath559 and the linear equivalence class of @xmath356 on @xmath485 is @xmath560 @xmath561 let @xmath116 and @xmath289 be two distinct points of @xmath562 and let us denote by @xmath563 the family of divisors tangent to @xmath356 at @xmath116 and @xmath289 with multiplicity @xmath99 and @xmath564 respectively . then @xmath563 is a linear system of dimension @xmath565 moreover , it is very easy to see that the general element of @xmath566 corresponds to an irreducible and reduced smooth rational curve . by applying ( * ? ? ? * proposition 2.1 ) , one shows that , in fact , @xmath567 we deduce that the variety @xmath568 parametrizing divisors tangent to @xmath356 at two distinct points with multiplicity @xmath99 and @xmath569 respectively , has dimension @xmath382 furthermore , looking at the tangent space to @xmath570 at its general element , it is easy to prove that , if @xmath116 and @xmath289 are two general points of @xmath571 then the unique divisor @xmath572 tangent to @xmath356 at @xmath116 with multiplicity @xmath99 and to @xmath289 with multiplicity @xmath564 intersects @xmath356 transversally at a point @xmath573 now let @xmath116 be a general point of @xmath356 and let @xmath574 be the fibre @xmath535 passing through @xmath115 because of the generality of @xmath116 and the fact that @xmath575 we may assume that @xmath576 and @xmath577 intersect @xmath356 transversally at a further point . again by the generality of @xmath119 we may assume that the fibre @xmath578 ( resp . @xmath579 ) passing through the point of @xmath580 ( resp . @xmath581 ) , different from @xmath119 intersects transversally @xmath356 at a further point . let @xmath582 ( resp . @xmath583 ) be the fibre of @xmath584 ( resp . @xmath585 ) passing through this point . we repeat this argument @xmath586 times , getting @xmath587 points @xmath588 of @xmath356 and fibres @xmath589 with @xmath541 verifying relations and . from what we proved above , by using that * the family of divisors in @xmath590 tangent to @xmath356 at @xmath119 with multiplicity @xmath195 and at a further point , with multiplicity @xmath564 has dimension @xmath591 * the point @xmath116 is general and the points @xmath588 are determined by @xmath116 by the argument above ; we find that there exists a point @xmath289 such that the unique divisor @xmath592 passes through @xmath593 now @xmath594 and hence @xmath595 this implies that there exists on @xmath477 a divisor @xmath596 cutting on @xmath356 the divisor @xmath597 moreover , @xmath481 is uniquely determined by the equality @xmath598 now , if @xmath599 then there exist only finitely many curves like @xmath600 and passing through a fixed general point @xmath601 if @xmath116 varies on @xmath571 then the curves @xmath283 constructed in this way , move in a one parameter family of curves @xmath602 by construction and by lemma [ lemma : analytic - type ] , the curve @xmath76 has a space quadruple point of analytic equations at @xmath116 and nodes or tacnodes at points @xmath603 as in the statement . it remains to prove that , if @xmath116 is general or , equivalently , if @xmath20\in\mathcal w$ ] is general , then the singularities of @xmath76 on @xmath544 are nodes . since @xmath604 for every @xmath605 and @xmath606 this is equivalent to showing that @xmath530 intersects transversally @xmath528 for every @xmath605 and @xmath607 if @xmath608 there is nothing to show . just to fix ideas , we prove the statement for @xmath609 our argument trivially extends to the general case . let @xmath20\in \mathcal w\subset |\mathcal o_r(2)|$ ] be a general element . first observe that , since @xmath610 is contained in the equisingular deformation locus @xmath611 of @xmath283 it follows that , if @xmath612 is the curve corresponding to a point @xmath613 $ ] of the tangent space @xmath614}\mathcal w,$ ] then @xmath612 has at @xmath289 a @xmath615-tacnode ( cf . * proof of theorem 3.3 ) ) . now assume that @xmath480 intersects @xmath616 at @xmath617 points @xmath618 let @xmath619 be the intersection multiplicity of @xmath480 with @xmath616 at @xmath620 then , the analytic equation of @xmath76 at @xmath217 is @xmath621 and , by ( * ? ? ? * proposition ( 5.6 ) ) , the localization at @xmath217 of the equisingular deformation ideal of @xmath76 is @xmath622 thus @xmath612 must be tangent to @xmath623 at every point @xmath217 with multiplicity @xmath624 similarly , @xmath612 must contain the intersection point of @xmath616 with @xmath625 it follows that the cardinality of intersection of @xmath626 and @xmath616 is given by @xmath627 we deduce that @xmath628 using again that @xmath612 passes through every node of @xmath76 and that , by lemma [ lemma : analytic - type ] and remark [ eq - eq - in ] , the curve @xmath629 must be tangent to @xmath356 at @xmath119 for @xmath630 we obtain that @xmath612 contains the points @xmath631 and @xmath116 of @xmath632 thus , since the intersection number @xmath633 we have that @xmath634 it follows that the analytic equations of @xmath358 at @xmath116 are given by , where @xmath635 and @xmath636 in particular , we find that @xmath626 has a node at @xmath116 with one branch tangent to @xmath356 and the other one tangent to @xmath625 again , we find that @xmath626 contains at least three points of @xmath637 counted with multiplicity , and hence @xmath638 this implies , by repeating the same argument , that @xmath639 thus , for every @xmath630 we have that @xmath640 where @xmath641 @xmath642 is tangent to @xmath356 at @xmath116 with multiplicity @xmath99 and at @xmath289 with multiplicity @xmath643 and , finally , @xmath642 contains @xmath644 but there is a unique divisor in @xmath645 with these properties . we deduce that @xmath646 for every @xmath630 @xmath647 and @xmath614}\mathcal w=\{[c]\},$ ] getting a contradiction . this completes the proof in the case @xmath75 . we now consider the case @xmath648 where @xmath649 @xmath556 with @xmath650 and @xmath651 and @xmath652 for every @xmath561 the proof of the lemma works as in the previous case , except for the construction of the curves @xmath653 let @xmath59 be a general point and let @xmath535 be the fibre passing through @xmath119 for @xmath561 because of the generality of @xmath119 the curve @xmath529 intersects @xmath356 at a further point , say @xmath654 if @xmath655 and @xmath656 if @xmath657 now the curves in @xmath658 passing through @xmath654 cut out on @xmath571 outside @xmath659 a @xmath660 having , because of the generality of @xmath119 four simple ramification points . let @xmath661 be one of the four curves in @xmath658 passing through @xmath654 and simply tangent to @xmath356 at a further point @xmath662 then , denote by @xmath663 the unique curve tangent to @xmath356 at @xmath664 and let @xmath665 be the further intersection point of @xmath583 with @xmath62 now repeat the same argument until you get curves @xmath666 with @xmath490 and @xmath667 and points @xmath588 satisfying relations - . again by the generality of @xmath119 there exists a point @xmath289 such that the unique ( smooth and irreducible ) divisor @xmath668 tangent to @xmath356 at @xmath116 with multiplicity @xmath99 and at @xmath289 with multiplicity @xmath669 passes through @xmath593 it follows that there exists a unique ( smooth and irreducible ) divisor @xmath670 passing through @xmath656 and tangent to @xmath356 at @xmath116 and @xmath289 with multiplicity @xmath99 an @xmath671 respectively . the curve we constructed has tacnodes , nodes and a space quadruple point of analytic equations on @xmath356 , as desired . to see that the singularities of this curve outside @xmath356 are nodes , argue as in the case @xmath672 let @xmath509 be the first cotangent bundle of @xmath510 defined by the standard exact sequence @xmath673 & \theta_r \ar[r ] & \theta_{\mathbb p^{\p}}|_r \ar[r ] & \mathcal n_{r|\mathbb p^{\p } } \ar[r ] & t^1_r \ar[r ] & 0 . } \ ] ] since @xmath469 is a variety with normal crossings , by ( * ? ? ? * section 2 ) , we know that @xmath509 is locally free of rank one . in particular , @xmath674 is a degree @xmath507 line bundle on @xmath62 fix a general divisor @xmath675 since the family of curves constructed in the previous lemma is a one parameter family , we may always assume that there exists a curve @xmath676 with @xmath527 as in the statement of lemma [ limiti ] , such that @xmath677 and @xmath678 for every @xmath679 and @xmath680 now , by ( * ? ? ? * corollary 1 ) , we have that the induced map @xmath681 is surjective . by ( * ? ? ? * theorems 1 and 2 ) and related references ( precisely , ( * ? ? ? * remark 2.6 ) and ( * ? ? ? * section 2 ) ) and because of the generality of @xmath682 it follows that there exists a deformation @xmath501 of @xmath502 whose general fiber is a smooth projective @xmath1 surface @xmath504 in @xmath493 with @xmath683\cong \mathbb z [ h]$ ] and such that @xmath190 is singular exactly at the points @xmath684 let @xmath685 be the smooth family of surfaces obtained by blowing - up @xmath686 and by contracting the corresponding exceptional components on @xmath513 in such a way that @xmath516 where @xmath517 is the blowing - up of @xmath477 at the points @xmath518 with new exceptional curves @xmath519 let us denote by @xmath687 the proper transform of @xmath76 and by @xmath688 the pull - back of @xmath76 with respect to @xmath689 now @xmath690 has one more node at the point @xmath691 on @xmath692 we want to prove the existence of irreducible curves @xmath693 with the desired singularities by deforming of the curve @xmath694 the irreducibility of @xmath112 easily follows from the fact that @xmath695 is indivisible . * @xmath697 nodes @xmath698 on @xmath699 arising from the intersection of the curves @xmath700 , for @xmath701 with @xmath702 for every @xmath703 * a node at @xmath704 * @xmath705 nodes @xmath706 @xmath630 arising from the intersection of @xmath529 with @xmath528 for every @xmath707 and @xmath538 * @xmath587 nodes at @xmath708 where now @xmath709 * a @xmath546-tacnode at @xmath289 and * a space quadruple point of analytic equations at @xmath115 now , as we already observed in the proof of the previous lemma , the tangent space @xmath710}es(\pi^*(c))$ ] to the equisingular deformation locus of @xmath690 in @xmath711 is contained in the linear system of divisors @xmath712 passing through the nodes of @xmath690 on @xmath713 ( @xmath714 included ) ; having a @xmath615-tacnode at the point @xmath289 of @xmath690 and having local analytic equations given by at @xmath116 . this implies that , if @xmath715}es(\pi^*(c)),$ ] then @xmath716 contains the irreducible component of @xmath687 passing through @xmath717 it follows that @xmath718 and so on , until , doing the same local analysis of @xmath612 at @xmath116 as in the proof of the previous lemma , we find that @xmath719}es(\pi^*(c ) ) ) = \{[\pi^*(c)]\}.\ ] ] using that the nodes of @xmath690 at the points @xmath720 are trivially preserved by every section of @xmath721 we deduce the injectivity of the standard morphism @xmath722 where @xmath723 in particular , we have that @xmath724 has image of dimension @xmath725 morever , by corollary [ ingenerale ] and by ( * ? ? * corollary 3.6 ) , we know that the image of the morphism @xmath724 must be contained in @xmath726 where @xmath727 and @xmath728 are linear spaces of dimension @xmath462 and @xmath569 respectively . we first study the map @xmath724 in the case that @xmath729 i.e. for @xmath730 and @xmath731 or @xmath608 and @xmath732 in this case the curve @xmath690 has a @xmath733-tacnode , i.e. a node , at @xmath734 and @xmath735 using again that every node of @xmath690 on @xmath356 is trivially preserved by every section of @xmath736 we obtain that the induced morphism @xmath737 is injective . in fact this morphism is also surjective by virtue of the equality @xmath738 by versality and by the proof of theorem [ th : triple - point ] , it follows that we may deform @xmath690 to an irreducible curve @xmath739 such that @xmath740 by preserving all nodes of @xmath690 on @xmath741 and by deforming the singularity of @xmath690 at @xmath116 to an ordinary triple point . now we study the morphism @xmath724 under the assumption that @xmath742 in this case @xmath724 is not surjective . more precisely , we have that @xmath743 and @xmath744 is a hyperplane in @xmath745 containing the image @xmath746 of @xmath747 as a codimension @xmath733 subspace . now , using on @xmath748 the affine coordinates @xmath749 introduced in the proof of lemma [ lemma : analytic - type ] , by the proof of theorem [ th : triple - point ] , by ( * ? ? ? * proof of theorem 3.3 ) and by a straightforward dimension count , we have that @xmath746 coincides with the @xmath750-plane @xmath751 where @xmath752 is the @xmath753-plane of equations @xmath754 and @xmath755 is the @xmath615-plane parametrizing @xmath546-nodal curves . in particular , we have that @xmath756 where @xmath757 is a hyperplane in @xmath758 such that @xmath759 it trivially follows that the projection maps @xmath760 and @xmath761 are surjective . now , by the proof of theorem [ th : triple - point ] , we know that the locus of curves with a triple point in @xmath762 is the smooth curve @xmath503 having equations @xmath763 and intersecting @xmath764 only at @xmath403 . similarly , by ( * ? ? ? * proof of theorem 3.3 ) , for every @xmath70-tuple of non - negative integers @xmath71 such that @xmath765 the locus @xmath766 of points corresponding to curves with @xmath78 singularities of type @xmath79 for every @xmath767 is a reduced ( possibly reducible ) curve intersecting @xmath768 only at @xmath427 it follows that , for every @xmath70-tuple of non - negative integers @xmath71 such that @xmath765 the locus @xmath769 is a reduced ( possibly reducible ) curve whose parametric equations may be explicitly computed by arguing exactly as in ( * ? ? ? * proof of lemma 4.4 , p.381 - 382 ) . this proves , by versality , that we may deform @xmath690 to an irreducible curve @xmath770 by preserving all nodes of @xmath690 at @xmath771 @xmath772 and @xmath773 by deforming the singularity of @xmath690 at @xmath116 to an ordinary triple point and the @xmath546-tacnode of @xmath690 to @xmath774 singularities of type @xmath88 for every @xmath98-tuple of integers @xmath78 such that @xmath775 in particular , if @xmath776 and @xmath777 the corresponding curve @xmath112 is an elliptic curve in @xmath67 with an ordinary triple point and nodes as singularities . finally , by the injectivity of the morphism @xmath778 we have that , if @xmath779 is the equisingular deformation locus of @xmath112 in @xmath780 then @xmath781}es(c_t))=0 $ ] and this implies that the singularities of @xmath112 may be smoothed independently , i.e. that the standard morphism @xmath782 is surjective . this proves the theorem for @xmath672 * @xmath783 nodes @xmath698 on @xmath699 arising from the intersection of the curves @xmath700 , for @xmath701 with @xmath702 for every @xmath703 * a node at @xmath704 * @xmath705 nodes @xmath706 @xmath630 arising from the intersection of @xmath529 with @xmath528 for every @xmath707 and @xmath538 * @xmath586 simple tacnodes at @xmath784 and nodes at @xmath785 where now @xmath709 * a @xmath549-tacnode at @xmath786 * a space quadruple point of analytic equations at @xmath115 with the same argument as above , one may prove that the curve @xmath690 may be deformed to a curve @xmath770 by preserving all nodes of @xmath690 at @xmath771 @xmath772 and @xmath773 by deforming the singularity of @xmath690 at @xmath116 to an ordinary triple point , every simple tacnode of @xmath690 on @xmath356 to a node and the @xmath549-tacnode of @xmath690 to @xmath78 singularity of type @xmath88 for every @xmath98-tuple of non negative integers @xmath78 such that @xmath787 the curve @xmath112 obtained in this way has the desired singularities , is a reduced point for the equisingular deformation locus and its singularities may be smoothed independently . finally , if we choose @xmath776 and @xmath788 then @xmath112 is an elliptic irreducible curve with a triple point and nodes as singularities . [ main - corollary ] let @xmath0 be a general primitively polarized @xmath1 surface of genus @xmath466 as above . then , for every @xmath8 and for every @xmath789 there exist reduced and irreducible curves in @xmath23 of geometric genus @xmath7 with an ordinary triple point and nodes as singularities and corresponding to regular points of their equisingular deformation locus . in accordance with @xcite , we do not expect the existence of rational curves in @xmath5 with a triple point . when @xmath608 and @xmath522 , theorem [ main - theorem ] implies that the family in @xmath87 of curves with a triple point and @xmath85 singularities of type @xmath88 is non - empty whenever it has expected dimension at least equal to one . the precise statement is the following . [ primitive - corollary ] let @xmath0 be a general primitively polarized @xmath1 surface of genus @xmath790 then , for every @xmath70-tuple of non - negative integers @xmath71 such that @xmath791 there exist reduced irreducible curves @xmath76 in the linear system @xmath87 on @xmath10 such that : * @xmath76 has an ordinary triple point , a node and @xmath78 singularities of type @xmath79 for every @xmath80 and no further singularities ; * @xmath76 corresponds to a regular point of the equisingular deformation locus @xmath81 equivalently , @xmath82}es(c))=0.$ ] the case @xmath792 and @xmath608 is the only case where the existence of elliptic curves with a triple point is expected but it is not treated in this paper . in this case , with the notation above , it is easy to show the existence of a unique curve @xmath793 with a space quadruple point analytically equivalent to . because of the unicity , the argument we used in the proof of theorem [ main - theorem ] to compute the dimension of the tangent space to the equisingular deformation locus of @xmath76 does not apply . actually , we expect that @xmath82}es(c ) ) > 0.$ ] nevertheless , it is easy to prove that @xmath76 deforms to curves @xmath794 with a triple point as singularity on the general @xmath1 surface @xmath117 but describing the equisingular deformation locus @xmath779 from the scheme - theoretic point of view seems to us to be a very difficult problem . the case @xmath792 and @xmath608 will be treated in detail in an upcoming article on related topics .

rotational spectra in four new excited vibrational levels of the linear carbon chain radical ch radical were observed in the millimeter band between 69 and 364 ghz in a low pressure glow discharge , and two of these were observed in a supersonic molecular beam between 19 and 38 ghz . all have rotational constants within 0.4% of the ground vibrational state of ch and were assigned to new bending vibrational levels , two each with and vibrational symmetry . the new levels are tentatively assigned to the and bending vibrational modes ( both with symmetry ) , and the and combination levels ( symmetry ) on the basis of the derived spectroscopic constants , relative intensities in our discharge source , and published laser spectroscopic and quantum calculations . prior spectroscopic constants in the and levels were refined . also presented are interferometric maps of the ground state and the level obtained with the sma near 257 ghz which show that ch is present near the central star in irc+10216 . we found no evidence with the sma for the new vibrationally excited levels of ch at a peak flux density averaged over a synthesized beam of jy / beam in the and ghz range , but it is anticipated that rotational lines in the new levels might be observed in irc+10216 when alma attains its full design capability .

1501.01262

carbon - rich asymptotic giant branch ( agb ) stars are a principal source of carbonaceous gas and dust in the interstellar medium , and some of the material from which planets are ultimately formed @xcite . the dust is believed to form in the poorly explored inner envelope at @xmath14 where the temperatures and densities are high ( @xmath15 k and @xmath16 @xmath17 ; * ? ? ? * ; * ? ? ? * ; * ? ? ? during the past 40 years , astronomers have studied the chemistry and dynamics in the outer envelope of the prototypical carbon - rich agb star irc+10216 , but it is only in the last few years that they have begun to observe a few small stable molecules in the inner envelope at high angular resolution with millimeter - wave interferometers.m @xcite . ] in the first wide band interferometric spectral line survey of irc+10216 , a large population of narrow molecular emission lines were observed whose expansion velocities ( @xmath18 km s@xmath19 ) are about three times lower than the terminal velocity ( 14.5 km s@xmath19 ; * ? ? ? subsequent measurements at still higher angular resolution than the initial @xmath10 confirmed that the emission from the narrow lines arises in the dust formation zone close to the star @xcite . in all , more than 200 narrow lines were observed in a 61 ghz wide spectral line survey in the 345 ghz band with the submillimeter array ( sma ) , of which the carriers of approximately 50% remain unassigned @xcite . when the same source was observed at somewhat higher angular resolution and sensitivity with alma , the number of unassigned lines is over 10 times greater ( e.g. , see the 1.9 ghz wide band near 267 ghz observed with a @xmath20 diameter beam in alma cycle 0 by * ? ? ? detection of such a large number of unassigned lines confirms that the supporting laboratory spectroscopy is far from complete , and the chemistry in the inner envelope of carbon - rich agb stars is not well understood . after h@xmath21 and co , acetylene ( hcch ) is the most abundant species in the inner envelope of irc+10216 @xcite . both on earth and in space , long carbon chain molecules are formed by reactions of highly reactive linear carbon chain c@xmath22h radicals with closed shell acetylenic molecules of the type c@xmath21h + c@xmath22h@xmath21 @xmath23 c@xmath24h@xmath21 + h c@xmath21h + hc@xmath25n @xmath23 hc@xmath26n + h @xcite . rotational lines of seven c@xmath22h radicals ( c@xmath21h @xmath27 c@xmath28h ) in the ground vibrational state have been studied extensively in the outer envelope of irc+10216 , and a few rotational lines have been observed in low - lying ( bending ) vibrational levels of c@xmath21h , c@xmath29h , c@xmath0h , and c@xmath30h at an angular resolution of @xmath31 with single antennas ( see table [ cnhspace ] , and references therein ) . chemical equilibrium calculations show that at temperatures near 1500 k , the abundance of the prototypical c@xmath21h radical is comparable to that of hcch @xcite , but c@xmath22h radicals have not yet been identified in the inner envelope . in the course of measuring the rotational spectra of reactive acetylenic radicals , carbenes , and anions in a laboratory discharge through hcch and a rare gas , we have observed rotational lines in four vibrationally excited levels of c@xmath0h that had not previously been studied . following the initial laboratory measurements of the c@xmath0h radical in the ground vibrational state @xcite , rotational lines were observed in irc+10216 and a laboratory discharge which were tentatively assigned to the @xmath32 vibrational level @xcite . at about the same time , @xcite in a comprehensive laboratory study measured the rotational spectra in the @xmath33 and @xmath34 excited vibrational levels , derived a full set of accurate spectroscopic constants for the @xmath8 and @xmath9 levels , obtained an approximate estimate of the @xmath35 bending vibrational frequency , and assigned over 20 previously unidentified lines in the outer envelope of irc+10216 to vibrationally excited c@xmath0h in @xmath8 and @xmath9 . the spectroscopic constants derived from the rotational spectra and the symmetry of the new vibrational levels are well determined , but owing to the absence of either supporting quantum calculations or spectroscopic measurements in other wavelength bands , the assignments of the rotational spectra to particular vibrational levels were uncertain . however , with the recent laser spectroscopic measurements of previously unexplored bending vibrational levels in the ground @xmath36 electronic state of c@xmath0h @xcite , we are now able to assign the rotational spectra to four vibrational levels . the assignments are based in part on the relative intensities of the rotational lines in our discharge source , and theoretical energies of the vibrational levels @xcite , whose accuracy was confirmed in part by the laser spectroscopic measurements @xcite . here we present laboratory measurements of the rotational spectra in four new vibrational levels of c@xmath0h in the millimeter band between 69 and 364 ghz ( two each with @xmath2 and @xmath3 vibronic symmetry ) , and the two @xmath2 levels in the centimeter band between 19 and 38 ghz . also described are interferometric maps with the sma of emission from the ground state and @xmath8 level near 257 ghz , which show for the first time that c@xmath0h is present near the central star . further interferometric observations at higher angular resolution and sensitivity , might help establish whether other vibrationally excited c@xmath22h radicals and small carbon bearing reactive species ( such as the cumulene carbenes h@xmath21c@xmath22 , etc . ) are the carriers of some of the many unassigned lines that have been observed in the inner envelope of irc+10216 . observations such as these might in turn help advance the chemistry of the inner envelope of carbon rich agb stars in the region where dust begins to form . the c@xmath0h radical was observed at millimeter wavelengths with a free space gas discharge spectrometer used to study reactive radicals , carbenes , and molecular ions of astrophysical interest @xcite . rotational lines in vibrationally excited c@xmath0h were measured under conditions which optimized the corresponding transitions in the ground vibrational state : a dc discharge ( 0.5 a , 1.4 kv ) through a flowing mixture of helium ( 1015% ) and acetylene ( hcch ) at a total pressure of 25 mtorr with the walls of the discharge cell cooled to 150 k. although the most intense lines of c@xmath0h were observed near 300 ghz , measurements with the free space spectrometer were extended to frequencies as low as 69 ghz in the three most intense new vibrational levels , and to 44 ghz in @xmath37 . zeeman chopping with a 50 g magnetic field helped discriminate between paramagnetic species and interfering closed shell species present in the discharge . the frequencies of the rotational lines were derived from a simultaneous least - squares fit of the explicit expression for the second derivative of a theoretical lorentzian profile , and an n@xmath38 order polynomial to spectra observed at 512 points in a 10 mhz bandwidth ( see * ? ? ? * ) . the s / n in the raw unsmoothed spectra were about 10 in approximately 15 min of integration ( figure [ fig - mmspectra ] ) . because the line widths were typically about 800 khz ( fwhm ) , the second derivative line profile was sampled at approximately 70 points between the two horns ( or satellites ) . from prior work on similar systems , we estimate that the uncertainties in the measured frequencies are about 20 khz , or less than 0.1 km s@xmath19 for lines up to 363 ghz . three rotational transitions were also observed in the centimeter band with our fourier transform microwave ( ftm ) spectrometer @xcite in @xmath39 and the two new ones with @xmath2 symmetry ( see sample spectra in figure [ fig - ftmspectra ] ) . the discharge conditions were optimized on the corresponding rotational transition of c@xmath0h in the ground vibrational state : a 1050 v dc discharge in the throat of the supersonic nozzle of the spectrometer through a mixture of acetylene ( 0.1% ) heavily diluted in neon . typical gas flow rates were 25 @xmath17 min@xmath19 at standard temperature and pressure ( stp ) , a pulse repetition rate of 6 hz , pulse duration of @xmath40s , and stagnation pressure behind the nozzle of 2.5 ktorr . under these conditions , lines in the @xmath8 and @xmath9 levels were @xmath41 times less intense than those of the ground state . spectra and maps of c@xmath0h in irc+10216 were observed with seven antennas of the sma in the compact configuration on 6 march 2008 , at baselines between 16.4 and 77.0 m. the synthesized beam was @xmath42 at a position angle ( p.a . ) of @xmath43 , and the field of view ( size of primary beam ) was @xmath44 . the system temperature ranged from 200 to 350 k , @xmath45 , and the total on source integration time was about 10 hours . the correlator was configured for a uniform frequency resolution of 0.64 mhz per channel in the lower and upper 2 ghz wide sidebands with band centers at 256.4 and 266.4 ghz . the quasar 0854 + 201 was observed every 20 min for gain calibration . the spectral bandpass was calibrated by observations of 3c273 and jupiter , and absolute flux calibration by observations of mars and titan . four new series of satellite lines were assigned to rotational transitions in vibrationally excited c@xmath0h because : ( 1 ) the rotational and centrifugal distortion constants ( @xmath46 and @xmath47 ) are within 0.4% of those in the ground state ( see figure [ fig - scheme ] ) ; and ( 2 ) the analogous satellite lines were observed in c@xmath0d , when hcch was replaced by dccd in our discharge . the symmetries of the four new vibrational levels were established from the following considerations . two quartets , symmetrically displaced about a central frequency , were assigned to two new @xmath3 levels ( designated @xmath48 and @xmath49 ) . the quartets , consisting of four lines with comparable intensities in each rotational transition , precluded levels with @xmath2 symmetry and the well - resolved @xmath50-doubling implied that @xmath51 . the remaining two series were assigned to @xmath2 levels ( @xmath52 and @xmath53 ) , because only two lines were found in each series and the coupling case remained hund s case ( @xmath54 ) at all observed frequencies . although doublets ( rather than quartets ) are frequently observed in levels with @xmath55 because the @xmath50-doubling is often unresolved ( e.g. , in @xmath56 levels ) , there is a significant departure from case ( @xmath54 ) coupling in the lower rotational transitions in vibrational levels with @xmath55 owing to the spin - orbit interaction . listed in tables [ tab - nu7pi][tab - sib ] are the measured rotational lines in the four new vibrational levels , and some new measurements in the previously studied @xmath57 ( @xmath3 ) and @xmath58 ( @xmath2 and @xmath56 ) levels . in a closed shell molecule , @xmath59-type doubling is described by only one constant ( @xmath60 ) , but two constants ( @xmath61 and @xmath60 ) are needed to describe @xmath50-type doubling in molecules with vibrational and electronic contributions @xcite . in the well studied c@xmath21h radical , the @xmath62 excited electronic state is 3600 cm@xmath19 above the @xmath64 ground state and @xmath61 is much smaller than @xmath60 , implying that the vibrational contribution to the @xmath50-type doubling is dominant @xcite . however , @xmath61 and @xmath60 are comparable in c@xmath0h , because the @xmath62 state is much closer to ground ( 222 cm@xmath19 ; * ? ? ? * ) , and the vibronic interaction between the two electronic states in c@xmath0h is much greater than in c@xmath21h . the rotational spectra in the vibrationally excited levels of c@xmath0h were analyzed by essentially the same method as previously applied to vibrationally excited c@xmath21h ( * ? ? ? * and references therein ) . in this approach , the rotational structure in each vibrational level was described by an effective hamiltonian used to analyze an isolated electronic state in a diatomic molecule . in open shell molecules , the electron orbital ( @xmath65 ) and vibrational ( @xmath59 ) angular momentum are coupled , and @xmath65-type and @xmath59-type doubling combine to produce @xmath50-type doubling . it has been shown in work similar to that here that the rotational spectra in excited vibrational levels of open shell molecules are reproduced to very high accuracy by this approach . some rotational levels in @xmath53 and @xmath49 were perturbed by nearby vibrational levels , requiring two additional terms in the hamiltonian to reproduce the spectrum of @xmath53 ( see section [ sec : perturbations ] ) . we have adhered to the sign convention in @xcite , in which the sign of @xmath50-doubling constant @xmath60 was determined on the assumption that the hyperfine doubling constant @xmath66 is positive . following this convention , @xmath60 is negative in @xmath37 . although @xmath66 is not well determined in the @xmath49 level , @xmath60 is assumed to be negative in @xmath49 as well . because the hyperfine constants were not determined in the @xmath48 level , we assumed by analogy with @xmath37 and @xmath49 that @xmath61 is also negative in the @xmath48 level . the signs of @xmath67 and the hyperfine coupling constants @xmath68 and @xmath69 are the same as those in the ground state as expected , because the spin - rotation and hyperfine interactions are sufficiently decoupled from vibrational effects . the spin - orbit constant @xmath70 is negative in the @xmath37 , @xmath34 , @xmath48 , and @xmath49 vibrational levels of the @xmath36 ground electronic state , because @xmath70 is negative in the low - lying @xmath62 excited electronic state @xcite . the @xmath53 and @xmath49 levels exhibit pronounced @xmath71-dependent perturbations by vibrational levels that were not observed in this study . in @xmath53 , the interacting levels cross between @xmath72 and 25 , resulting in very large shifts in the transition frequencies of one of the spin - rotation components , whereas the other component is not perturbed . to fit this pattern , a standard coriolis - type contribution , @xmath73 was added to the energies of the perturbed component . the @xmath71-dependence of the numerator depends on the specific spin - rotation components in the interaction . we chose the pair that gave the lowest standard deviation of the fit , but @xmath74 differs by @xmath75% for other choices and the constants differ by less than one standard deviation . in @xmath49 , all four components are weakly perturbed and the magnitude of the perturbation increases with increasing @xmath71 . analyzing the perturbation in @xmath49 would require adding several empirical parameters with no clear interpretation . instead , the two most strongly perturbed transitions in @xmath49 were removed from the fit , and @xmath49 was treated as though it were an isolated @xmath3 state ( see table [ tab - pib ] ) . [ subsec : bsigma ] the strong perturbation of the @xmath76 component of the spin - rotation doublets in @xmath53 implies that there is an interaction with a level with either @xmath3 or @xmath77 vibronic symmetry . to first order in perturbation theory , @xmath78 and @xmath79 in eq . ( 1 ) , where @xmath80 is the difference in band origins @xmath81 , @xmath82 , the perturbing level is labeled @xmath83 , and @xmath84 is the coefficient of the coriolis coupling matrix element . both @xmath74 and @xmath85 in @xmath53 are negative ( table [ newconst ] ) , indicating that the perturbing level has a smaller @xmath46 and the band origin is higher than @xmath53 . the @xmath49 level is perturbed at high @xmath71 by a level that has not yet been identified . when the two uppermost rotational lines from any one of the four parity / spin - orbit branches are removed from the fit , the rms is reduced by a factor of two and the higher order distortion terms @xmath86 and @xmath87 are not needed . in table [ newconst ] , transitions with @xmath88 in @xmath89 were omitted because doing so yielded the lowest rms , but other fits with comparable rms were obtained if one of the other branches was neglected instead . the addition of higher - order terms in the hamiltonian such as @xmath90 , implies that more than one branch is significantly perturbed . a perturbation analysis similar to that for @xmath53 was not attempted for @xmath49 , primarily because the frequency shifts are much smaller in @xmath49 . in the future , measurements of the perturbing levels near the resonance interactions in @xmath53 and @xmath49 might yield accurate coupling constants and frequency separations of the perturbing levels . the vibrational modes of c@xmath0h may be loosely divided into c h local modes ( @xmath91 stretch and @xmath92 bend ) , and c@xmath0 normal modes ( @xmath93 and @xmath94 symmetric stretches , @xmath95 antisymmetric stretch , and @xmath96 and @xmath35 bends ) . for the most likely ones observed here ( @xmath97 , and combinations thereof ) , accurate estimates of the band origins of three ( @xmath94 , @xmath35 , and @xmath98 ) were recently derived from laser spectroscopic double resonance four - wave mixing measurements of the @xmath99 electronic band near 24,000 cm@xmath19 @xcite . theoretical frequencies of all seven vibrational modes were calculated independently in the harmonic approximation @xcite . because the theoretical frequencies of @xmath94 and @xmath35 agree to within 6% or better with those measured by mazzotti et al . , the theoretical frequencies of @xmath100 and @xmath96 are likely known to comparable accuracy ( see table [ tab - hc3n ] ) . owing to the precision of the frequency measurements and degree of symmetry in the rotational spectra here , the vibronic symmetries and spectroscopic constants of the four new levels in c@xmath0h were determined to a very high level of confidence . because the rotational information needed to guide the assignment of the new vibrational levels of c@xmath0h was not available either from theoretical calculations or ir spectroscopic measurements , the assignments to particular vibrational levels is tentative . rotational spectra in vibrational satellites of closed shell polyatomic molecules are routinely assigned on the basis of relative intensities of the lines , vibronic symmetry , the magnitude and sign of the vibration - rotation interaction @xmath101 ( where @xmath102 ) , and the @xmath59-doubling constant @xmath60 , where the approximations @xmath103 and @xmath104 are normally accurate to within 20% for low vibrational excitation . for example , in the well studied linear cyanopolyynes hcn and hcccn @xcite , the magnitude of @xmath101 increases as the fundamental frequency of the mode decreases , it is negative in bending levels ( @xmath105 ) , and positive in stretching levels ( @xmath106 ) as expected . unfortunately , the standard procedure for assigning rotational spectra to specific vibrational levels in closed shell polyatomic molecules is not necessarily applicable in the open shell c@xmath22h radicals . for example , @xmath101 in the @xmath8 level of c@xmath0h is 1/5 that in @xmath58 , rather than 1/2 as expected from eq . [ alpha ] . similarly , the @xmath101 s in the bending vibrational levels in c@xmath21h are anomalous : @xmath101 in the @xmath107 level is 1/12 that in @xmath39 , rather than 1/2 as expected @xcite . assignments of the new vibrationally excited levels of c@xmath0h were restricted to levels with @xmath2 and @xmath3 symmetry , and to ones whose magnitude of @xmath101 are among the smallest . before considering higher excitations with @xmath108 and higher - lying combination levels , we examined the fundamental modes ( @xmath109 ) and combinations with the @xmath35 bend . the requirement that the rotational frequencies in the new vibrational levels are close to those of the ground state , implies that they are hund s case ( @xmath54 ) coupled in the frequency range where most of the measurements were made ( i.e. , at @xmath110 ) . because the spin - orbit interaction is small , the vibration - rotation coupling increases with increasing vibrational excitation ( as indicated by eq . [ alpha ] , and observed in hcccn ; * ? ? ? * ) , implying that rotational lines in the lower vibrational levels lie in the frequency band that we had covered . there are many vibrational levels in the ground @xmath36 and excited @xmath62 electronic states of c@xmath0h below 1000 cm@xmath19 ( see figure 6 in * ? ? ? considering only those in the @xmath36 electronic state , ( excluding @xmath8 and @xmath9 ) there are eight possible candidates . electronic state are ( @xmath111 , ( 0004 ) , ( 0010 ) , ( 0100 ) , ( 1000 ) , ( 0011 ) , ( 0101 ) , and ( 1001 ) . ] three of these , all with excitation energies between 400 and 600 cm@xmath19 , were considered as candidates for the two new vibrational levels with @xmath112 symmetry : @xmath4 , @xmath5 , and @xmath113 . we have assumed on the basis of the qualitative form of eq . [ alpha ] , that @xmath101 in @xmath113 is significantly larger than in @xmath9 . therefore the @xmath113 level was ruled out , because the rotational lines in the levels we observe are closer in frequency to the ground state lines than those in @xmath9 . as a result , the @xmath49 level was assigned to @xmath4 , and the less intense @xmath48 level to the higher lying @xmath5 level . there are seven candidates ( between 600 and 1000 cm@xmath19 above ground ) for the two new levels with @xmath114 symmetry . the @xmath115 level was discounted , owing to the large @xmath101 and intensity considerations ( see section [ subsec : intensity ] ) . the @xmath52 level was assigned to the @xmath98 combination level , even though the magnitude of @xmath101 is very small for a combination level and the @xmath56 and the second @xmath2 level were not observed , because @xmath98 has the lowest energy of the remaining levels with @xmath114 symmetry . the @xmath53 level was assigned last , because it was made primarily on the basis of the intensity information discussed in section [ subsec : intensity ] . assignments of the vibrational levels of c@xmath0h on the basis of relative intensities might be secure if the vibrational temperature ( @xmath116 ) in our hcch discharge were known precisely . our best independent estimate of @xmath116 is from prior work on the c@xmath21h radical because : ( 1 ) the electronic and vibrational level structures of c@xmath21h and c@xmath0h are qualitatively similar ; ( 2 ) both radicals are observed under comparable excitation conditions in the same hcch discharge ; and ( 3 ) there is a large body of ir measurements and extensive quantum calculations of the vibrational level structure in c@xmath21h ( * ? ? ? * and references therein ) . we found that @xmath116 in the bending vibration ( 010 ) , and the combination levels ( 110 ) and ( 011 ) of c@xmath21h is @xmath117 k higher than the temperature of the wall of our discharge cell ( 150 k ) , whereas @xmath116 in the stretching levels ( 100 ) and ( 001 ) is about 5 times higher ( @xmath118 k ; * ? ? ? * ) . by analogy with c@xmath21h , it was anticipated that @xmath119 k would also apply to c@xmath0h in @xmath120 , and in levels in combination with the bending vibrations . as a test , we first examined the excitation of @xmath8 and @xmath9 in c@xmath0h , because the assignments and the vibrational energies are well established @xcite . the derived vibrational temperature for the @xmath8 and @xmath9 levels in c@xmath0h ( @xmath121 k ) is in good agreement with that of c@xmath21h ( @xmath122 k ) and c@xmath21d ( @xmath123 k ; * ? ? ? * ) , therefore the analogy with c@xmath21h appears justified . encouraged by the similarity between the vibrational excitation of c@xmath21h and the lowest bending vibrational levels of c@xmath0h , we then considered the intensity information when assessing the assignments of the new levels that have been made on the basis of the spectroscopic constants ( section [ subsec : spectroscopic ] ) . the @xmath124 level at 930 cm@xmath19 ( 1337 k ; * ? ? ? * ) was ruled out because @xmath125 k is much too low for a stretching vibration , and the intensity is much too high if @xmath53 were a combination level with a bending vibration ( e.g. , @xmath126 at 1579 k above ground ) . on the assumption that @xmath116 is comparable to that of the @xmath8 and @xmath9 bending vibrational levels , the most plausible assignment of @xmath53 is the combination level @xmath127 . the vibrational energy of @xmath53 derived from figure [ vtdiag ] ( @xmath128 k , or @xmath129 cm@xmath19 ) is in good agreement with the theoretical frequency ( 756 cm@xmath19 ; * ? ? ? * ) . with these assignments , the relative intensities of all six vibrational levels of c@xmath0h observed in our discharge are characterized by a vibrational temperature ( @xmath130 k ; figure [ vtdiag ] ) that is consistent with the bending vibrational levels of c@xmath21h . shown in figure [ chanmaps ] , are channel maps of the @xmath131 rotational transition of c@xmath0h near 257 ghz in the ground vibrational state , and the @xmath37 excited vibrational level . the rotational transitions observed here are about 2.5 times higher in frequency , and 150 k higher in energy ( @xmath132 ) than those observed earlier in the 3 mm band with the iram plateau de bure interrferometer ( pdbi ) by @xcite . the peak flux density is 0.63 jy / beam for the average of both spin components in the ground state , and 0.34 jy / beam for three of the four rotational transitions in the vibrationally excited level . also shown are integrated intensity maps of the same transitions in the ground state and excited vibrational level , for a velocity interval corresponding to the middle two rows in the channel maps ( from @xmath133 km s@xmath19 in figure [ chanmaps ] ) that is roughly centered on the systemic velocity of @xmath134 km s@xmath19 ( see figures [ v0mom0 ] and [ v1mom0 ] ) . as noted earlier by @xcite , the ground and excited emission of c@xmath0h follow each other very closely in the outer envelope . here for the first time , both ground and vibrationally excited c@xmath0h are observed near the central star . the exact nature of the emission near the central star remains unclear from our sma observations i.e. , whether it is from a smaller expanding inner shell that has not reached the terminal velocity ( @xmath135 km s@xmath19 ) , or an asymmetrical structure connecting the outer shell to the inner envelope . interpretation of the central emission will require further interferometric observations at higher s / n and with a much better @xmath136-@xmath137 coverage . the clumpy features in the expanding shell structure illustrates the inherent uncertainties when comparing the interferometric measurements with published chemical models ( which typically show only radial dependence ) . as expected , the flux density in the ground and vibrationally excited emission in the central region ( 0.6 versus 0.4 jy / beam ) is comparable at the higher kinetic temperature of the gas @xcite . we also examined the 345 ghz spectral line survey with the sma @xcite for possible evidence of the four new excited bending vibrational levels of c@xmath0h , but none was found at peak flux densities averaged over a @xmath10 synthesized beam of @xmath138 jy in the @xmath12 and @xmath13 ghz range . it has generally been assumed that c@xmath0h ( and other c@xmath22h radicals ) is confined to the outer envelope of irc+10216 , because that is what is implied from the spectral line profiles of the ground and @xmath8 , @xmath32 , and @xmath34 excited vibrational levels observed at low angular resolution with single antennas . furthermore , interferometric maps of rotational transitions of c@xmath0h in the ground vibrational and the lowest excited vibrational level @xmath8 in the 3 mm band , show that the emission is confined to a thin ( @xmath10 thick ) shell in the outer envelope with a radius ( @xmath139 ) of @xmath140 @xcite just as most chemical models predict . however , @xcite recently showed that when density - enhanced dust shells are included in the chemical models , c@xmath0h is predicted to be present in a thick inner ring ( @xmath141 ) in addition to the outer ring at @xmath140 . rotational lines in the four new vibrational levels of c@xmath0h will most likely be observed in the inner envelope of irc+10216 , because that is where these higher lying levels will be most populated owing to the higher densities and temperatures closer to the star . although we did not find any evidence for lines of the new excited bending vibrational levels of c@xmath0h , it is plausible that the new vibrationally excited levels of c@xmath0h will be observed when millimeter - wave interferometers attain higher sensitivity now that we have established that c@xmath0h is present near the central star . direct observations of hcch in the ir and hnc in the millimeter band , provide a template for assigning and analyzing rotational spectra in vibrationally excited levels of c@xmath22h radicals among the many unassigned features recently observed in the inner envelope of irc+10216 with interferometers . these include : ( 1 ) a detailed analysis of the kinetic , vibrational , and rotational temperatures of hcch at @xmath142 derived from rotationally resolved vibrational lines in the @xmath143 m band @xcite ; ( 2 ) observations of hnc in the dust formation zone near 270 ghz with alma cycle 0 , and an accompanying detailed radiative transfer analysis @xcite ; and ( 3 ) a related observation of hcccn in the dense core of the protoplanetary nebula crl 618 with the sma in the 345 ghz band at @xmath144 from the central star where @xmath145 k @xcite . sample spectra obtained in alma cycle 0 with sixteen 16 m diam antennas and a @xmath20 synthesized beam in the @xmath146 ghz band @xcite are about 7 times more sensitive than those in the sma survey in the 345 ghz band @xcite . times greater with the sma . the synthesized beam in alma cycle 0 was 5 times smaller , but the diameter of the inner envelope ( @xmath147 ) is small with respect to the angular resolution in both observations . ] referring to figure 1 in @xcite , the peak flux densities of most of the unassigned narrow lines observed with alma ( @xmath148 jy / beam ) are below the detection limit with the sma . when alma attains full design sensitivity with 66 antennas it will be about 4 times more sensitive still , offering great promise for the identification of key reactive molecular intermediates such as the c@xmath22h radicals in the inner envelope of the prototypical carbon rich agb star irc+10216 , and elucidation of the chemical processes crucial to the formation of carbonaceous dust in the interstellar gas . agndez , m. , & cernicharo , j. 2006 , , 650 , 374 brown , j. m. 1975 , j. mol . spectrosc . , 56 , 159 caris , m. , giesen , t. f. , duan , c. , mller , h. s. p. , schlemmer , s. , & yamamda , k. m. t. 2009 , j. mol . spectrosc . , 253 , 99 cernicharo , j. , gulin , m. , agndez , m. , mccarthy , m. c. , & thaddeus , p. 2008 , , 688 , l83 cernicharo , j. , daniel , f. , castro - carrizo , a. , agndez , m. , marcelino , n. , joblin , c. , goicoechea , j. r. , & gulin , m. 2013 , , 778 , l25 cherchneff , i. 2000 , , 456 , 1001 cordiner , m a. , & millar , t. j. 2009 , , 697 , 68 fonfra , j. p. , cernicharo , j. , richter , m. j. , & lacy , j. h. 2008 , , 673 , 445 gail , h .- , & sedlmayr , e. 1988 , , 206 , 153 gottlieb , c. a. , gottlieb , e. w. , thaddeus , p. , & kawamura , h. 1983 , , 275 , 916 gottlieb , c. a. , mccarthy , m. c. , & thaddeus , p. 2010 , , 189 , 261 gottlieb , c. a. , myers , p. c. , & thaddeus , p. 2003 , , 588 , 665 graf , s. , geiss , j. , & leutwyler , s. 2001 , , 114 , 4542 gulin , m. , cernicharo , j. , navarro , s. , woodward , d. r .. gottlieb , c. a. , & thaddeus , p. 1987 , , 182 , l37 gulin , m. , neininger , n. , lucas , r. , & cernicharo , j. 1999 , in proc . 3rd cologne - zermatt symp . , the physics and chemistry of the interstellar medium , ed . v. ossenkopf ( herdecke : cga - verlag ) , 326 killian , t. c. , gottlieb , c. a. , & thaddeus , p. 2007 , , 127 , 114320 lee , c .- f . , yang , c .- h . , sahai , r. , & snchez contreras , c. 2013 , , 770 , 153 mazzotti , f. j. , raghunandan , r. , esmail , a. m , tulej , m. , & maier , j. p. 2011 , , 134 , 164303 mbosei . l. , fayt , a. , dran , p. , & coslou , j. 2000 , j. mol . struct . , 517 - 518 , 271 mccarthy , m. c. , chen , w. , travers , m. j. , & thaddeus , p. 2000 , , 129 , 611 millar , t. j. , herbst , e. , & bettens , r. p. a. 2000 , , 316 , 195 molster , f. j. , & waters , l. b. f. m. 2003 , in _ astromineralogy _ , ed . by t. henning ( springer , new york , 2003 ) , pp . 121 - 170 ott , u. 2010 , in _ astromineralogy _ , by t. henning . lecture notes in physics , vol . 815 ( springer , berlin , 2010 ) , pp . 277 - 311 patel , n. a , young , k. h. , brnken , s. , et al . 2009a , , 692 , 1205 patel , n. a , young , k. h. , brnken , s. , menten , k. m. , thaddeus , p. , & wilson , r. w. 2009b , , 691 , l55 patel , n. a , young , k. h. , gottlieb , c. a. , et al . 2011 , , 193 , 17 sharp - williams , e. n. , roberts , m. a. , & nesbitt , d. j. 2011 , , 134 , 064314 speck , a. , barlow , m. j. , & skinner , c. j. 1997 , , 288 , 431 tenenbaum , e. d. , dodd , j. l. , milam , s. n. , woolf , n. j. , & ziurys , l. m. 2010 , , 190 , 348 woods , p. m. , millar , t. j. , herbst , e. , & zijlstra , a. a. 2003 , , 402 , 189 yamamoto , s. , saito , s. , gulin , m. , cernicharo , j. , suzuki , h. , & ohishi , m. 1987 , , 323 , l149 yamamoto , s. , saito , s. , suzuki , h. , deguchi , s. , kaifu , n. , ishikawa , s .- ohishi , m. 1990 , , 348 , 363 yasuda , y. , & kozasa , t. 2012 , , 745 , 159 c@xmath21h & @xmath149 & 370 & & 29 & 0.3 & d & e + c@xmath29h & @xmath150 ) & 27 & & 28 & 0.4 & d & f + c@xmath0h & @xmath37 & 170 & & 27 & 2.9 & d & g + & @xmath32 & 375 & & 27 & 1.5 & d & g + & @xmath34 & 375 & & 27 & 0.6 & d & g + c@xmath30h & @xmath151 & 14 & & 28 & 0.1 & h & i + & @xmath152 & 73 & & 28 & 0.04 & h & i + @xmath153 & @xmath154 & @xmath155 & 44712.367 & 0.036 & & & & & + & @xmath154 & @xmath153 & 44712.913 & 0.007 & & & & & + & @xmath156 & @xmath155 & 44830.238 & @xmath157 & & & & & + & @xmath156 & @xmath153 & 44830.966 & 0.010 & & & & & + @xmath158 & & & & & & @xmath156 & @xmath159 & 68553.686 & 0.025 + & & & & & & @xmath156 & @xmath158 & 68554.209 & 0.022 + & & & & & & @xmath154 & @xmath159 & 68716.586 & @xmath160 + & & & & & & @xmath154 & @xmath158 & 68717.020 & 0.004 + @xmath159 & @xmath154 & @xmath158 & 74141.426 & 0.001 & & @xmath156 & @xmath161 & 77833.064 & @xmath162 + & @xmath154 & @xmath159 & 74141.841 & 0.003 & & @xmath156 & @xmath159 & 77833.510 & @xmath163 + & @xmath156 & @xmath158 & 74352.312 & 0.015 & & @xmath154 & @xmath161 & 78029.136 & 0.020 + & @xmath156 & @xmath159 & 74352.817 & @xmath164 & & @xmath154 & @xmath159 & 78029.504 & 0.024 + @xmath161 & @xmath154 & @xmath159 & 83878.977 & @xmath165 & & @xmath156 & @xmath166 & 87142.553 & 0.005 + & @xmath154 & @xmath161 & 83879.350 & @xmath167 & & @xmath156 & @xmath161 & 87142.954 & @xmath164 + & @xmath156 & @xmath159 & 84122.980 & 0.021 & & @xmath154 & @xmath166 & 87371.839 & @xmath168 + & @xmath156 & @xmath161 & 84123.434 & 0.013 & & @xmath154 & @xmath161 & 87372.159 & @xmath165 + @xmath166 & @xmath154 & @xmath161 & 93585.949 & @xmath169 & & & & & + & @xmath154 & @xmath166 & 93586.275 & @xmath167 & & & & & + & @xmath156 & @xmath161 & 93863.168 & @xmath170 & & & & & + & @xmath156 & @xmath166 & 93863.572 & @xmath171 & & & & & + [ tab - nu7pi ] @xmath172 & @xmath172 & 19157.642 & 0.002 & & @xmath173 & 19100.146 & @xmath174 + @xmath173 & @xmath173 & 28721.504 & @xmath167 & & @xmath173 & 28664.320 & @xmath164 + & @xmath172 & 28721.625 & @xmath175 & & @xmath176 & 28664.476 & @xmath177 + @xmath176 & @xmath176 & 38285.562 & 0.000 & & @xmath176 & 38228.637 & @xmath164 + & @xmath173 & 38285.583 & 0.002 & & @xmath178 & 38228.734 & 0.002 + [ tab - nu7sig ] @xmath158 & & & & & @xmath159 & 69192.595 & 0.020 + & & & & & @xmath158 & 69193.060 & 0.007 + @xmath159 & @xmath158 & 74161.246 & 0.017 & & @xmath161 & 78519.431 & 0.009 + & @xmath159 & 74161.662 & 0.019 & & @xmath159 & 78519.898 & 0.033 + @xmath161 & @xmath159 & 83947.326 & @xmath1790.002 & & @xmath166 & 87870.583 & @xmath1790.013 + & @xmath161 & 83947.724 & 0.003 & & @xmath161 & 87870.967 & @xmath1790.033 + @xmath166 & @xmath161 & 93708.740 & 0.001 & & & & + & @xmath166 & 93709.107 & 0.003 & & & & + @xmath180 & @xmath156 & 95502.793 & 0.032 & & @xmath154 & 95557.501 & 0.028 + & @xmath154 & 95542.080 & 0.031 & & @xmath156 & 95609.099 & 0.014 + @xmath181 & @xmath156 & 124182.153 & @xmath1790.006 & & @xmath154 & 124198.118 & 0.012 + & @xmath154 & 124235.243 & @xmath1790.024 & & @xmath156 & 124263.526 & @xmath1790.008 + @xmath182 & @xmath156 & 143293.623 & @xmath1790.030 & & @xmath154 & 143295.568 & @xmath1790.009 + & @xmath154 & 143356.034 & @xmath1790.004 & & @xmath156 & 143370.299 & 0.019 + @xmath183 & @xmath156 & 171954.051 & 0.000 & & @xmath154 & 171942.999 & @xmath1790.028 + & @xmath154 & 172030.463 & @xmath1790.008 & & @xmath156 & 172031.797 & 0.032 + @xmath184 & @xmath156 & 181506.072 & @xmath1790.037 & & & & + & @xmath154 & 181587.188 & @xmath1790.056 & & @xmath156 & 181585.504 & @xmath1790.016 + @xmath185 & @xmath156 & 286540.183 & 0.012 & & @xmath154 & 286510.241 & @xmath1790.079 + & @xmath154 & 286674.782 & 0.010 & & @xmath156 & 286657.242 & 0.004 + @xmath186 & @xmath156 & 362881.183 & 0.024 & & @xmath154 & 362847.399 & 0.015 + & @xmath154 & 363057.084 & 0.053 & & @xmath156 & 363035.520 & @xmath1790.053 + [ tab - pia ] @xmath187 & & & & & & @xmath154 & @xmath188 & 69557.920 & 0.032 + & & & & & & @xmath154 & @xmath187 & 69558.338 & @xmath157 + & & & & & & @xmath156 & @xmath188 & 69689.931 & 0.026 + & & & & & & @xmath156 & @xmath187 & 69690.356 & 0.029 + @xmath188 & @xmath156 & @xmath187 & 73404.509 & 0.034 & & & & & + & @xmath156 & @xmath188 & 73404.924 & 0.021 & & & & & + & @xmath154 & @xmath187 & 73600.331 & 0.008 & & @xmath156 & @xmath189 & 79038.104 & @xmath190 + & @xmath154 & @xmath188 & 73600.807 & 0.004 & & @xmath156 & @xmath188 & 79038.541 & 0.027 + @xmath189 & @xmath156 & @xmath188 & 83138.919 & 0.008 & & @xmath154 & @xmath180 & 88203.982 & 0.003 + & @xmath156 & @xmath189 & 83139.341 & 0.032 & & @xmath154 & @xmath189 & 88204.406 & 0.014 + & @xmath154 & @xmath188 & 83366.506 & 0.006 & & @xmath156 & @xmath180 & 88401.284 & @xmath191 + & @xmath154 & @xmath189 & 83366.936 & @xmath171 & & @xmath156 & @xmath189 & 88401.641 & @xmath171 + @xmath180 & @xmath156 & @xmath189 & 92858.025 & @xmath167 & & & & & + & @xmath156 & @xmath180 & 92858.378 & @xmath165 & & & & & + & @xmath154 & @xmath189 & 93118.952 & @xmath164 & & & & & + & @xmath154 & @xmath180 & 93119.392 & 0.019 & & & & & + @xmath192 & @xmath156 & @xmath180 & 102562.162 & @xmath170 & & & & & + & @xmath156 & @xmath192 & 102562.488 & @xmath163 & & & & & + & @xmath154 & @xmath180 & 102857.575 & @xmath165 & & & & & + & @xmath154 & @xmath192 & 102857.964 & @xmath165 & & & & & + @xmath193 & @xmath156 & @xmath192 & 112251.885 & @xmath194 & & @xmath154 & @xmath181 & 116272.310 & @xmath171 + & @xmath156 & @xmath193 & 112252.194 & @xmath169 & & @xmath154 & @xmath193 & 116272.661 & 0.009 + & @xmath154 & @xmath192 & 112582.623 & 0.022 & & & & & + & @xmath154 & @xmath193 & 112582.992 & 0.020 & & & & & + @xmath181 & & & & & & @xmath154 & & 125653.515 & @xmath170 + & & & & & & @xmath156 & & 125992.970 & @xmath195 + @xmath182 & @xmath156 & & 141242.289 & @xmath162 & & @xmath154 & & 144448.893 & @xmath196 + & @xmath154 & & 141680.954 & 0.009 & & @xmath156 & & 144860.942 & 0.014 + @xmath197 & & & & & & @xmath154 & & 153861.682 & @xmath198 + & @xmath154 & & 151356.927 & 0.00 & & @xmath156 & & 154309.767 & @xmath199 + @xmath183 & @xmath156 & & 170131.947 & @xmath200 & & @xmath154 & & 172713.383 & @xmath201 + & @xmath154 & & 170678.204 & @xmath202 & & @xmath156 & & 173232.979 & @xmath203 + @xmath184 & @xmath156 & & 179743.519 & 0.073 & & @xmath154 & & 182150.890 & 0.071 + & @xmath154 & & 180325.149 & 0.033 & & @xmath156 & & 182705.875 & 0.078 + @xmath204 & @xmath156 & & 237266.740 & @xmath205 & & @xmath154 & & 238895.750 & @xmath206 + & & & & & & @xmath156 & & 239656.612 & 0.018 + @xmath185 & @xmath156 & & 285066.869 & @xmath207 & & @xmath154 & & 286286.672 & @xmath165 + & @xmath154 & & 286016.697 & & & @xmath156 & & 287211.598 & @xmath198 + @xmath186 & @xmath156 & & 361388.690 & 0.044 & & @xmath154 & & 362204.016 & @xmath175 + & @xmath154 & & 362584.458 & & & @xmath156 & & 363380.662 & 0.000 + [ tab - pib ] @xmath172 & @xmath172 & 19052.195 & 0.002 & & @xmath173 & 19034.551 & @xmath174 + & & & & & @xmath172 & 19034.559 & @xmath174 + @xmath173 & @xmath173 & 28573.715 & @xmath167 & & @xmath176 & 28556.273 & @xmath164 + & @xmath172 & 28574.092 & @xmath175 & & @xmath173 & 28556.305 & @xmath177 + @xmath176 & @xmath176 & 38095.249 & @xmath174 & & @xmath178 & 38077.887 & @xmath164 + & @xmath173 & 38095.401 & 0.002 & & @xmath176 & 38077.915 & 0.002 + @xmath182 & & 142822.059 & 0.059 & & & 142805.027 & 0.017 + @xmath197 & & 152341.151 & 0.036 & & & 152324.186 & 0.023 + @xmath183 & & 171378.323 & 0.027 & & & 171361.426 & @xmath175 + @xmath204 & & 237995.258 & 0.061 & & & 237978.662 & @xmath199 + @xmath185 & & 285563.150 & @xmath191 & & & 285546.928 & @xmath208 + @xmath186 & & 361637.717 & @xmath209 & & & 361622.319 & 0.057 + [ tab - sia ] @xmath172 & & & & & @xmath173 & 19000.323 & 0.005 + @xmath173 & @xmath172 & 28558.709 & @xmath164 & & @xmath210 & 28511.978 & @xmath164 + & & & & & @xmath176 & 28512.176 & 0.007 + @xmath176 & @xmath173 & 38070.179 & @xmath211 & & @xmath176 & 38023.823 & @xmath212 + & @xmath176 & 38070.204 & @xmath160 & & @xmath178 & 38023.924 & @xmath212 + @xmath159 & & 76114.910 & 0.035 & & & & + @xmath161 & & 85625.489 & @xmath157 & & & 85580.757 & @xmath213 + @xmath166 & & 95135.697 & @xmath214 & & & 95091.505 & @xmath215 + @xmath192 & & 104645.517 & @xmath216 & & & 104602.169 & @xmath217 + @xmath193 & & 114154.716 & @xmath218 & & & 114112.540 & @xmath219 + @xmath181 & & & & & & 123622.556 & @xmath203 + @xmath182 & & & & & & 142641.738 & @xmath220 + @xmath197 & & & & & & 152150.770 & @xmath221 + @xmath183 & & & & & & 171167.819 & @xmath222 + @xmath184 & & & & & & 180675.774 & @xmath164 + @xmath223 & & & & & & 228208.117 & @xmath169 + @xmath204 & & 237750.637 & @xmath224 & & & 237713.112 & 0.077 + @xmath225 & & 256763.157 & 0.093 & & & 256720.966 & @xmath226 + @xmath185 & & 285272.924 & 0.168 & & & 285228.362 & @xmath196 + @xmath227 & & 323274.033 & 0.131 & & & 323228.432 & 0.044 + @xmath228 & & & & & & 332726.510 & 0.011 + @xmath229 & & 342269.653 & 0.050 & & & 342223.785 & @xmath163 + @xmath186 & & 361261.840 & @xmath230 & & & 361215.873 & @xmath231 + [ tab - sib ] @xmath70 & & @xmath17989812.6(10 ) & & @xmath17950575.69(39 ) + @xmath46 & 4758.6557(7 ) & 4762.85047(26 ) & 4782.16713(28 ) & 4778.57763(32 ) + @xmath232 & 0.8627(10 ) & 0.89320(41 ) & 0.91475(57 ) & 0.90125(21 ) + @xmath233 & & 0.93(24 ) & @xmath234 & + @xmath67 & @xmath17938.555(2 ) & @xmath17938.149(38 ) & @xmath235 & @xmath17937.225(17 ) + @xmath236 & 0.127(9 ) & 0.030(14 ) & 0.709(26 ) & + + @xmath61 & & @xmath237 & & + @xmath238 & & 0.1809(95 ) & & + @xmath60 & & @xmath239 & & + @xmath240 & & 0.11794(36 ) & & + + @xmath241 & & 1.62(49 ) & & 1.25(34 ) + @xmath242 & @xmath243 & @xmath244 & @xmath245 & @xmath246 + @xmath69 & 12.435(10 ) & 13.6(16 ) & 12.29(21 ) & 11.1(1.9 ) + @xmath66 & & 5.07(72 ) & & + + & & 0.020 & 0.023 & 0.020 + [ tab - const ] @xmath70 & @xmath247 & @xmath179152601.3(69 ) & & + @xmath46 & 4778.03697(39 ) & 4770.99746(38 ) & 4760.85127(32 ) & 4755.9431(14 ) + @xmath232 & 0.80158(87 ) & 0.91582(76 ) & 0.89191(62 ) & 0.9762(17 ) + @xmath233 & 1.04(36 ) & 11.65(33 ) & 2.28(28 ) & 3.94(72 ) + @xmath67 & @xmath17934.371(32 ) & @xmath17928.94(19 ) & @xmath17917.266(8 ) & @xmath17946.674(33 ) + @xmath236 & & 1.653(62 ) & 0.409(10 ) & 0.038(18 ) + + @xmath61 & @xmath248 & @xmath249 & & + @xmath238 & & @xmath250 & & + @xmath60 & 2.26473(50 ) & @xmath251 & & + @xmath240 & 0.04516(36 ) & 0.1889(20 ) & & + @xmath252 & & @xmath253 & & + + @xmath74 & & & & @xmath254 + @xmath85 @xmath255 & & & & @xmath256 + + @xmath241 & & 0.0 & & + @xmath242 & & @xmath257 & @xmath258 & @xmath259 + @xmath69 & & 18.9(22 ) & 13.65(25 ) & 6.0(8 ) + @xmath66 & & 3.2(20 ) & & + + & 0.032 & 0.034 & 0.030 & 0.035 + [ newconst ] @xmath35 & & & 168 & 178 + @xmath96 & & @xmath260 & & 379 + @xmath92 & & @xmath261 & & 578 + @xmath94 & & & 930 & 924 + @xmath95 & & & & 2131 + @xmath93 & & & & 2301 + @xmath91 & & & & 3608 + @xmath98 & & @xmath52 & 565 & 557 + @xmath127 & & @xmath53 & & 756 + h radical . _ upper : _ the @xmath262 transition in b@xmath3 , and the @xmath263 transition in @xmath8 near 256.4 ghz . the carriers of features labeled u are unidentified . _ lower : _ the two spin rotation components of the @xmath264 transition of a@xmath2 , and the @xmath265 transition in @xmath8 near 257.0 ghz . also present are the two lambda components of the @xmath266 transition of the c@xmath267h radical . the line shape is the second derivative of a lorentzian profile owing to the detection scheme employed . the integration time for each spectrum was approximately 15 min . ] rotational transition in the a@xmath2 excited vibrational level of c@xmath0h observed in a supersonic molecular beam . partially resolved hyperfine structure is observed in the @xmath268 component . the integration time was 12 min at 19034 mhz and 20 min at 19052 mhz . ] rotational transition in the ground and excited vibrational levels of c@xmath0h near 171 ghz . _ red : _ three vibrationally excited levels identified previously by @xcite , and the @xmath2 electronic ground state ground state ( gs ) . _ black : _ the four new vibrational levels a@xmath3 , b@xmath3 , a@xmath2 , and b@xmath2 ( this work ) . the relative intensities were measured in a low pressure discharge through helium and acetylene ( see section [ subsec : laboratory ] ) . ] , @xmath9 , and four new levels of c@xmath0h in the @xmath36 electronic state . the relative populations in the ground ( @xmath269 ) and excited vibrational levels ( @xmath270 ) were determined from measurements of the @xmath271 rotational transition near 143 ghz in a low pressure discharge through he and hcch cooled to 150 k. the vibrational levels are labeled as @xmath272 . see sections [ subsec : spectroscopic ] and [ subsec : intensity ] for a discussion of the assignments of the new levels . ] rotational transition of c@xmath0h in the ground and @xmath37 excited vibrational level . _ color half - tone : _ average of the two spin doublets in the ground vibrational state at 256880 and 256919 mhz at @xmath273 k. _ white contours : _ average of three of the four lambda components of the @xmath37 level at 256414 , 257017 , and 257218 mhz at @xmath274 k. the contours are plotted every @xmath275 , starting with @xmath275 and an rms noise of 20 mjy / beam . the actual rms noise varies in each channel map , but is approximately 20 to 50 mjy / beam . the maps show the close correspondence between the ground and excited state emission , and the azimuthal asymmetry and clumpiness in the expanding outer shell . ] h ( see figure [ chanmaps ] ) . the velocity range of integration ( @xmath276 km s@xmath19 ) is approximately centered on the systemic velocity . the emission near the center shows that c@xmath0h is not confined solely to the outer envelope . ] h of the @xmath8 excited vibrational level ( see figure [ chanmaps ] ) . the velocity range of integration ( @xmath276 km s@xmath19 ) is approximately centered on the systemic velocity . the map here of the integrated intensity of the @xmath8 level confirms that c@xmath0h is present close to the star . the clumps seen in the outer ring correlate well with those seen in the ground state emission in figure [ v0mom0 ] . ]

we present the x - ray source catalog in the subaru/_xmm - newton _ deep survey . a continuous area of 1.14 deg centered at r.a . = 02h18 m and dec . = 05d is mapped by seven pointings with _ xmm - newton _ covering the 0.210 kev band . from the combined images of the epic pn and mos cameras , we detect 866 , 1114 , 645 , and 136 sources with sensitivity limits of , , , and erg s in the 0.52 , 0.54.5 , 210 , and 4.510 kev bands , respectively , with detection likelihood ( corresponding to a confidence level of 99.91% ) . the catalog consists of 1245 sources in total including 32 extended - source candidates . the averaged log - log relations are in good agreement with previous results , bridging the flux range between _ chandra _ deep surveys and brighter surveys . the log - log relations show significant spatial variation among pointings on a scale of 0.2 deg . analyzing the auto correlation function , we detect significant clustering signals from the 0.52 kev band sample , which can be fit with a power law form with a correlation length of when the integral constraint term is included . in the 210 kev band , however , the clustering is not significant with a 90% upper limit of .

0806.2846

the subaru/_xmm - newton _ deep survey ( sxds ; @xcite ) is , along with cosmos @xcite , one of the largest multi - wavelength survey projects with an unprecedented combination of depth and sky area over a contiguous region of @xmath15 1 deg@xmath0 . the main aims of the sxds are to make an accurate measurement of the global properties of the universe without being affected by cosmic variance and to reveal the evolution of the large scale structure . the sxds consists of a wealth of multi - wavelength data taken by the most modern observing facilities ; x - ray imaging / spectroscopic data in the 0.210 kev band taken by the european photon imaging camera ( epic ; @xcite ) onboard _ xmm - newton _ @xcite , multi - color ( b , v , r , i@xmath16 , z@xmath16 ) deep optical images by suprime - cam on the subaru telescope @xcite , deep near infrared maps ( j , h , k ) observed as the ultra deep survey ( uds ) in the united kingdom infrared deep sky survey ( ukidss ; @xcite ) , mid- and far - infrared data ( 3.6160 @xmath17 m ) taken with the _ spitzer _ space observatory , the submillimeter ( 850 @xmath17 m ) map by the scuba half degree extragalactic survey ( shades ; @xcite ) , and the deep radio image ( 1.4 ghz ) by the very large array @xcite . the sxds field is centered at r.a . = 02h18 m and dec . = 05d and the total field of the subaru suprime - cam and _ xmm - newton _ epic images covers an area of 1.3 and 1.14 deg@xmath0 , respectively . the overall survey design and details of each survey are summarized in @xcite . the data of _ xmm - newton _ constitute a major component of the sxds project . x - ray surveys are a powerful tool to trace the cosmological evolution of active phenomena in the universe , including active galactic nuclei ( agns ) and clusters / groups of galaxies . the main constituents of x - ray sources that make up the x - ray background ( xrb ) are agns ( for a recent review , see @xcite ) . their dominant populations are `` obscured '' agns @xcite , where the central engine is blocked by dust and/or gas in the line of sight . in particular , surveys by hard x - rays above 2 kev are the most efficient approach to detect these obscured agn populations of various luminosity classes with least bias , thanks to their strong penetrating power against the photo - electric absorption of matter and small contamination from stars . in fact , the surface density of agns detected in the most sensitive x - ray surveys @xcite far exceeds that detected in the optical bands ( e.g. , * ? ? ? * ) , thus providing us with the most complete and clean samples for agn studies including heavily obscured , low luminosity ones that dominate the whole agn populations by number . the tight correlation between the mass of a super massive black hole ( smbh ) in the galactic center and that of the galactic bulge @xcite indicates strong links between the growth of smbhs and star formation in the past . since agns are phenomena that mark the process of growth of smbhs in galactic centers , to elucidate the evolution of agns is a fundamental issue for understanding the history of the universe . combination of ultra - deep pencil beam surveys and large area surveys have revealed the cosmological evolution of the x - ray luminosity function of agns @xcite . the global accretion history of the universe is similar to that of star formation ( see e.g. , * ? ? ? * ; * ? ? ? an important finding is that the number density of more luminous agns have a peak at higher redshifts compared with less luminous ones . this behavior is called `` down - sizing '' or `` anti - hierarchical '' evolution , which is opposite to a naive expectation from the standard structure formation theory of the universe . similar trends were also obtained in the star formation history ( e.g. , * ? ? ? * ; * ? ? ? * ) . these facts imply two modes of `` co - evolution '' of galaxies and smbhs for different masses . it could be explained by the feedback from supernova and agns ( e.g. , * ? ? ? deep multi - wavelength surveys enable us to observe various populations in different evolutionary stage , giving the best opportunities to investigate how galaxies and smbhs co - evolved with cosmic time . despite this major progress on agn evolution in recent years , there are several critical issues to be resolved by current and future x - ray surveys , even if we limit the scope only to x - ray detected agns . immediate objectives include ( 1 ) whether or how the fraction of obscured agns evolves with redshift , ( 2 ) the number density of compton thick agns ( those whose line - of - sight absorption exceeds @xmath18 @xmath5 ) , ( 3 ) the evolution of the agn luminosity function at high redshifts ( @xmath19 ) to be compared with the results from optical qso surveys , and ( 4 ) the evolution of clustering properties of agns as a function of luminosity and type . the sxds x - ray survey , one of the few `` wide '' and `` deep '' surveys , will give important steps to answer these questions and to establish the average properties of the agn populations . population synthesis models of the xrb utilize the results from deep surveys in a limited area at the faintest flux levels @xcite , which would be subject to cosmic variance when we discuss the xrb spectrum with @xmath2010% accuracy . this is crucial when we discuss a relatively minor contribution of some x - ray sources , such as compton thick agns . to constrain the statistical properties of rare objects , such as high redshift qsos , much larger cosmic volume is necessary than in the currently available deep surveys . by surveying a continuous area with a large depth , we can also study the evolution of the large scale structure probed by agns . it helps us to understand the physical conditions that triggers agn activity for different luminosity class , type ( 1 or 2 ) , and redshift . at the flux limits of the sxds , we can detect the most dominant populations of x - ray sources that contribute to the xrb . in all of these studies , high completeness of identification ( redshift determination ) is crucial . the multi - wavelength data are particularly useful in identifying x - ray sources even if optical spectroscopy is difficult for a population of `` optically faint '' agns @xcite . in this paper , we describe the x - ray data of the sxds based on the _ xmm - newton_survey performed in 2000 and 2003 . we present the whole x - ray catalog , and basic statistical properties of the detected x - ray sources . the x - ray data have been used in a number of studies in the sxds projects , detection of ultra luminous x - ray sources in nearby galaxies @xcite , mid - ir and radio selected compton - thick agns @xcite , radio sources @xcite , lyman @xmath21 sources @xcite , and optical variability selected agns @xcite . a series of papers on optical identification of the x - ray sources are forthcoming ( akiyama et al . , in preparation ) . the paper is organized as follows . 2 summarizes the observation of the _ xmm - newton_. in 3 we describe the detailed procedure of data analysis , and present the source list and its statistics . 4 and 5 shows the results of log @xmath8 - log @xmath9 relations and auto correlation functions , respectively . in 6 long - term time variability of x - ray sources is studied in a selected field . 7 summarizes the conclusions . throughout the paper , the cosmological parameters ( @xmath22 , @xmath23 , @xmath24)= ( 100@xmath25 km s@xmath6 mpc@xmath6 , 0.3 , 0.7 ) are adopted . the sxds field , centered at r.a . = 02h18 m and dec . = 05d , was mapped with _ xmm - newton _ in the 0.210 kev band by seven pointings , one deep ( nominal exposure 100 ksec ) observation in the center surrounded by six shallower ( 50 ksec each ) ones . _ xmm - newton _ carries three epics , pn , mos1 , and mos2 , each has a field of view ( fov ) of about @xmath26 . the pointing positions are arranged so that the combined x - ray map , having a `` flower petal '' pattern , covers essentially the whole region of the optical images taken with the subaru telescope , mosaic of 5 suprime - cam images in each band @xcite . there are small overlapped regions between these pointings , making the x - ray image continuous without gaps over an area of about 1.14 deg@xmath0 , although the sensitivity is not completely uniform over the entire map due to different exposures and instrumental effects , such as vignetting ( see 4.1 for details ) . table 1 gives the observation log , including the pointing position , observation time , and net exposure obtained after screening out high background time . we designate the pointings as sds-1 for the central field and sds-2 through sds-7 for the surrounding ones in a clockwise direction . the observations were performed in 3 discrete epochs , the first from 2000 july 31 to august 8 ( for sds-1 , 2 , 3 , 4 ) , the second from 2002 august 8 to 12 ( for sds-5 , 6 , 7 ) , and the third on 2003 january 7 to supplement the unfulfilled exposure of sds-4 . the observation of each pointing were performed continuously except for short intervals ( several hours ) in sds-1 and sds-3 and a long interval ( two and half years ) in sds-4 . the `` thin '' filters were used for the three cameras in all the observations . rrrrrrrr sds-1 & 0112370101 & 34.47838 & -4.98117 & 73.6 & 2000/07/31 22:38 & 08/01 14:04 & 40361 + & 0112371001 & 34.47883 & -4.98050 & 74.5 & 2000/08/02 21:13 & 08/03 13:15 & 42473 + sds-2 & 0112370301 & 34.87900 & -4.98047 & 74.8 & 2000/08/04 21:05 & 08/05 14:00 & 40220 + sds-3 & 0112370401 & 34.67971 & -4.63422 & 75.7 & 2000/08/06 07:53 & 08/06/17:38 & 14341 + & 0112371501 & 34.67958 & -4.63414 & 75.7 & 2000/08/06 20:58 & 08/06/23:20 & 4080 + sds-4 & 0112371701 & 34.27946 & -4.63394 & 76.0 & 2000/08/08 07:57 & 08/08 15:02 & 21277 + & 0112372001 & 34.32404 & -4.66858 & 245.7&2003/01/07 04:41 & 01/07 11:56 & 25940 + sds-5 & 0112370601 & 34.07854 & -4.98164 & 76.1 & 2002/08/12 06:06 & 08/12 18:52 & 34377 + sds-6 & 0112370701 & 34.27754 & -5.32797 & 74.9 & 2002/08/08 15:28 & 08/09 04:43 & 46802 + sds-7 & 0112370801 & 34.67792 & -5.32856 & 74.9 & 2002/08/09 05:52 & 08/09 19:02 & 36803 the reduction of the epic data was performed by using the science analysis system ( sas ) software . we used the sas version 6.1.0 for image creation and all the subsequent analysis , while an earlier version of sas ( that very similar to version 5.3.3 ) was utilized to produce the event files from which we start the image analysis , processed in the _ xmm - newton _ survey science center ( ssc ) pipeline processing system ( pps ) facility in university of leicester . as the first step , we created sky images with a bin size of 4 arcsec in four energy bands , 0.30.5 , 0.52 , 24.5 , 4.510 kev , separately for each observation i d and detector . the 7.98.3 kev band was excluded to avoid the instrumental background of cu k - lines @xcite . with the standard event selection , we accumulated good photon events with `` patterns '' of 0 through 12 ( mos1 and mos2 ) or 0 through 4 ( pn ) . the time region of background flare was excluded by examining the light curve . this leaves a net exposure of about 6080% of the allocated observing time , as listed in table 1 . to achieve the best positional accuracy , we corrected the absolute astrometry of the epic image of each obs . i d with respect to the subaru r - band image in the following way . we performed source detection from each observation , and cross correlated the obtained ( tentative ) source list with the optical catalog @xcite to calculate the mean positional offsets . then we corrected the original images for these offsets , which were found to be 0@xmath12 1.3@xmath12 , consistent with the estimated error in the absolute astrometry of _ xmm - newton_. this correction also minimizes systematic errors caused by summing up multiple observations for the case of sds-1 , 3 and 4 . we produce source lists separately from different pointings ( sds-1 through 7 ) . later , these are merged into one list by excluding overlapping regions to define a statistically independent sample from the whole data ( 3.3 ) . the overall flow of the source detection is similar to the one adopted in the ssc pps used to produce the 2xmm catalog ( watson et al . , in preparation ) , although several improvements are employed here . the main differences are as follows ; ( 1 ) we perform source detection to the summed image of pn , mos1 , and mos2 , not to individual ones . ( 2 ) we optimize several parameters of the spline interpolation in making background maps . ( 3 ) special care is paid to separate neighboring sources . for each observation , energy band , and detector , we create a background map from the image by spline interpolation with the sas program _ esplinemap _ after excluding sources found with _ eboxdetect _ , which perform simple source detection based on a cell detection algorithm . after optimization of parameters , we verify that the obtained models well represent the profiles of the background ( i.e. , the unresolved xrb plus non x - ray background ) over the whole fov without large deviation from the data . we sum up the pn , mos1 , mos2 images and background maps for each energy band and each `` pointing '' , which means also combining those of multiple observations for sds-1 , 3 and 4 . this enables us to achieve the best sensitivity from the whole available data in the simplest manner . at the same time , we can avoid technical problems in the position and flux determination that would be caused by image gaps between ccd chips in a single detector . to obtain exposure maps , we first calculated them for each instrument in narrow energy bands of every 0.5 kev , using the sas task _ eexpmap_. then , we took their count - rate weighted average , normalized to the pn count rate in a given survey band . the count rate distribution is calculated through the energy response of each instrument assuming a power law photon index of 1.5 above 2 kev and 2.0 below 2 kev . the dependence on assumed photon indices is negligibly small in the analysis . the cross - calibration of effective area between the pn and mos cameras is known to be accurate within 5% level , and its uncertainty does not affect the results . with this procedure , we are able to treat the summed images and exposure maps as if they were created from a single pn detector regardless of the detected position . in the following analysis , we present the flux of a source in terms of a `` pn - equivalent '' count rate . strictly speaking , this treatment may not be perfectly accurate in the image fitting process described below , because the positions of the optical axis in the combined image are not common for different detectors and observations , which could affect a precise modeling of the combined point spread function ( psf ) of the mirrors . to estimate _ systematic errors caused by this effect , we make the same analysis by changing the position of the optical axis to that of a mos camera . we find that the fluxes of sources obtained in the two analyses match each other by 1.3% level , confirming that this approximation is justified at this accuracy . figure 1 shows the pseudo three - colored x - ray image of the sxds field combined from the 7 pointings . this image is made by the _ export _ command in the iraf package . as the inputs , we use the exposure - corrected , background subtracted pn+mos1+mos2 images in the 0.52 , 24.5 , and 4.510 kev bands , corresponding to red , green , and blue colors , respectively . they are smoothed by a gaussian profile with a 1@xmath27 width of 4 arcsec over the entire field . larger color fluctuation is apparently seen in outer regions compared with the central region due to the shorter exposure . the source list from each pointing is obtained by the maximum likelihood fit applied to the detector co - added images in the four energy bands by using the sas task _ emldetect_. the _ emldetect_program reads images , background maps , and exposure maps in a single or multiple energy bands , and makes a simultaneous fit to the images with a model consisting of psfs over the background map , based on the input list of source candidates . it returns the fitted position , vignetting - corrected count rate , and detection likelihood ( hereafter called `` ml '' standing for maximum likelihood ) of each source . the obtained count rate corresponds to the total flux in the entire psf . to make the list complete , this process is iterated twice in the following manner . we first perform cell detection by _ eboxdetect _ with a cell size of 20@xmath12@xmath28 20@xmath12 to obtain a list of source candidates with minimum likelihood of 3.2 , which is used as an input to the first run of _ emldetect_. in both programs we allow the flux and position of a source to be free parameters . we find that in some cases the above cell size is too large to completely detect sources in a high density region . to supplement this , we also perform cell - detection with a 12@xmath12@xmath28 12@xmath12cell size , and add new source candidates detected with high significance into the source list obtained above . in this stage , by human inspection , we pay attention not to include obviously fake sources such as a part of a psf tail from nearby bright sources , diffuse emission , and false detection close to gaps between ccd chips . the merged list is again input to _ emldetect _ , producing a final source list . through the fitting process , we find that there are sources that are likely to be confused by the psf of neighboring ones in crowded regions . to measure the fluxes of these sources with the best accuracy , we perform an image fit in each region by allowing the positions and fluxes of the multiple sources ( up to three ) to vary simultaneously in order to solve the coupling of the fluxes self - consistently . their angular separation is typically 12@xmath12 24@xmath12 . their fluxes are replaced with the new values obtained here , and the ml values are calculated for each source at the fixed position . the number of pairs of such possibly confused sources are about 10 per pointing , and are marked by flag `` c '' in the last column of the source catalog ( table 2 ) . the statistic that _ emldetect _ uses to fit the source parameters is the @xmath29 statistic defined by @xcite . for a poissonian probability distribution , appropriate to the measurement of x - ray events by _ xmm - newton _ , cash s statistic takes the form @xmath30 where @xmath31 is the expected number of x - ray events in the @xmath32th pixel and @xmath33th energy band , @xmath34 is the detected number of events in that pixel / band , and the sum is over the @xmath8 pixels within the detection region and the @xmath35 energy bands used ( here = 4 as stated ) . the value of @xmath31 is obtained by adding the value of the source model at that pixel / band to the expected contribution from the background ; @xmath31 is thus a function of the source model parameters . after arriving at those values of the source parameters which minimize @xmath29 , the detection likelihood ( formally , the probability of the null hypothesis ) for those optimum values is then calculated . cash s prescription for this is to form the difference @xmath36 where @xmath37 is @xmath29 calculated by setting the amplitude of the source model to zero and @xmath38 is the minimum result returned by the fitting routine . according to cash s theory , @xmath39 is distributed approximately as @xmath40 for @xmath41 degrees of freedom , where @xmath41 is the number of fitted parameters . the probability @xmath42 of obtaining the calculated value of @xmath39 or greater by chance fluctuations of the detected background can therefore be obtained in terms of the complementary incomplete gamma function @xmath43 as follows : @xmath44 the value of ml is set within _ emldetect _ to equal @xmath45 . note that the statistics must be treated with caution when the total number of photon counts used for the fit are very small ( @xmath46 9 ) , which is not the case in our analysis . the ml value as calculated in the above way is viewed as more sensitive than the simple box - detection statistic used by _ eboxdetect _ , because it uses information about the source psf to help reject random fluctuations of the detected background . offsetting this is a possible bias introduced by the fact that _ emldetect _ is not given free rein over the whole field , but is applied only to a small number of restricted areas of the field , at locations of candidate sources already found by _ eboxdetect_. if the detection threshold of _ eboxdetect _ is set to too high a value , there is a risk that a real source with a good shape , but not many counts , will be missed by the combined detection process . in order to avoid this bias , one ought therefore to run the preceding _ eboxdetect _ with a detection threshold set deliberately low . on the other hand , since _ emldetect _ must go though the computationally - intensive fitting process with each of its candidates , it will be impractical to provide it with too long a list of candidates . the respective values of 3.2 and 7 adopted here for the _ eboxdetect _ and _ emldetect _ likelihood cutoffs were chosen with this necessary compromise in mind . an additional advantage of using _ emldetect _ to determine the final source parameters is its ability to add the models of already fitted sources to an internal background map . since the sources are processed in the order of their brightness , it is possible to take into account the background introduced by bright sources when fitting the fainter sources . up to this stage , all sources are assumed to be point like . to constrain the spatial extent of the detected sources , we perform an image fit by allowing the source extent to be a free parameter assuming a gaussian profile . the source positions are fixed at the input ones . this procedure yields a list of extended source candidates from the sxds field ( table 3 ) . for these sources , the fluxes listed in table 2 correspond to those obtained in this process with consideration of the source extent . however , as our procedure is essentially dedicated to detection of point like sources in the earlier stage , this list must be regarded to be incomplete . a more extensive approach to search for extended sources will be presented elsewhere ( finoguenov et al . , in preparation ) . lcc 0034 & 5.1 & 2.8@xmath47 0.5 + 0051 & 11.4 & 17.9@xmath47 1.2 + 0140 & 6.8 & 22.0@xmath47 3.3 + 0153 & 9.2 & 4.7@xmath47 0.4 + 0156 & 7.4 & 4.2@xmath47 0.9 + 0239 & 4.4 & 3.9@xmath47 1.0 + 0280 & 4.5 & 19.3@xmath47 3.3 + 0285 & 36.6 & 16.0@xmath47 1.8 + 0287 & 4.9 & 4.8@xmath47 1.3 + 0292 & 5.5 & 4.1@xmath47 1.3 + 0396 & 4.0 & 5.9@xmath47 1.8 + 0441 & 12.4 & 12.7@xmath47 1.4 + 0453 & 7.2 & 12.4@xmath47 2.2 + 0514 & 30.9 & 21.0@xmath47 2.0 + 0552 & 6.5 & 7.1@xmath47 1.4 + 0621 & 6.5 & 6.6@xmath47 1.4 + 0622 & 4.3 & 4.6@xmath47 1.4 + 0625 & 5.7 & 2.9@xmath47 1.8 + 0646 & 9.1 & 3.5@xmath47 0.8 + 0647 & 10.3 & 9.4@xmath47 1.6 + 0712 & 6.6 & 4.3@xmath47 0.8 + 0784 & 14.3 & 33.6@xmath47 3.3 + 0796 & 20.5 & 3.0@xmath47 0.3 + 0829 & 48.6 & 11.5@xmath47 0.6 + 0852 & 4.7 & 3.3@xmath47 0.2 + 0876 & 19.6 & 17.4@xmath47 1.6 + 0889 & 5.7 & 17.2@xmath47 3.1 + 0934 & 10.0 & 4.7@xmath47 0.4 + 1152 & 5.0 & 11.8@xmath47 3.9 + 1168 & 4.6 & 18.4@xmath47 2.8 + 1169 & 5.2 & 3.7@xmath47 1.1 + 1176 & 31.4 & 15.3@xmath47 1.1 thus , we obtain seven source lists separately from different pointings ( sds-1 through 7 ) , containing the information of position and fluxes with the ml values in the four energy bands , 0.30.5 , 0.52 , 24.5 , and 4.510 kev . in these tentative lists , we keep all sources whose summed ml from the four bands is equal to or exceeds 5 , which will be further screened in the following way . to make scientific research in comparison with previous studies , it is quite useful to define a sample selected in standard bands such as 0.54.5 kev ( hereafter `` xid band '' ) and 210 kev band ( `` hard band '' ) . although the combined flux and ml can be calculated from those in the two individual bands ( i.e. , 0.52 and 24.5 kev for the xid band , and 24.5 and 4.510 kev for the hard band ) , its statistical treatment would become quite complex ; for instance , the sensitivity limit can not be uniquely determined as a function of ml at a given position , depending on the source spectrum . hence , to obtain a well defined sample selected in the xid or hard band in the same way as in the four narrow bands , we calculate their ml values by repeating the likelihood fit using the 0.54.5 or 210 kev image with the background map summed from the two narrow bands . in the fit , the positions of the sources are fixed value is smaller than those obtained in the four individual bands for a given ml , because the degree of freedom @xmath41 is reduced by 2 . this is also the case for the `` confused '' sources whose ml values are obtained by fixing the position ( see [ sec : mlf ] ) . we properly take these effects into account in the analysis of log @xmath8 - log @xmath9 relations and auto correlation function . ] and no new sources are considered . in this paper we adopt a threshold for the detection likelihood of 7 in a single band . thus , from the tentative source lists , we select only those detected with ml@xmath77 either in the 0.30.5 kev ( ultra - soft ) , 0.52 kev ( soft ) , 24.5 kev ( medium ) , 4.510 kev ( ultra - hard ) , 0.54.5 kev ( xid ) , or 210 kev ( hard ) band , to be included in the final list . further , for statistical analysis using this sample , it is convenient to have a single source list merged from the seven pointings ; as mentioned above , there are overlapping regions between different pointings in the outer edge of the fov , where same sources are repeatedly detected . to exclude such duplication , we only refer to the results of a single pointing that achieves the highest sensitivity at a given position , based on sensitivity maps in the xid band ( see below ) . table 2 gives the source list in the sxds field complied in this way , sorted by right ascension ( ra ) and declination ( dec ) : ( col . 1 ) source number ; ( cols . 2 and 3 ) the x - ray source position as determined by _ xmm - newton _ ( ra and dec ) ; ( col . 4 ) the statistical error in the position estimated through the maximum likelihood fit ( root sum square of the 1@xmath27 error in each direction ) ; ( cols . 5 through 10 ) the ml value in each energy band ; ( cols . 11 through 14 ) the vignetting - corrected count rate in each energy band ; ( cols . 15 trough 17 ) the hardness ratios defined as @xmath48 , @xmath49 , @xmath50 , where @xmath51 , @xmath9 , @xmath35 , and @xmath52 are the count rates in the 0.30.5 , 0.52 , 24.5 , 4.510 kev band , respectively ; ( col . 18 ) the pointing i d at which the source is detected ; ( col . 19 ) the offset angle from the mean optical axis in the corresponding pointing ; ( col . 20 ) the total pn - equivalent exposure ( sum of pn , mos1 , and mos2 ) at the source position , corrected for the vignetting in the 0.5 - 4.5 kev band ; ( col . 21 ) the background rate in the 0.5 - 4.5 kev band at the source position ; ( col . 22 ) the flag if the flux is determined by multiple - source fit to take into account possible source confusion with nearby sources ( see 3.2.3 ) . the exposure and background rate ( cols . 20 and 21 ) are averaged over the source extraction region with a fixed encircled energy fraction ( 68% ) of the psf . from the results of optical identification ( akiyama et al . , in preparation ) , we confirm that the positional errors given in this list are reasonable . in table 3 we list the spatial extent and its likelihood of all the extended sources contained in table 2 . we detect 1245 sources in total . the numbers of detected sources in a single or two bands in any combination from the six energy bands are summarized in table 4 . according to simulation study , the number of fake sources due to statistical fluctuation is estimated to be about 1.5 per pointing in each energy band at our likelihood threshold ( ml@xmath77 ) , hence @xmath53 in the sum sample . their fraction in the total number of detected sources depends on the selection band ; it becomes the smallest ( 0.6% ) in the 0.54.5 kev but largest ( 5% ) in the 4.510 kev band because of different sensitivities . ccccccccc 0.30.5 kev & * 357 * & 339 & 235 & 73 & 341 & 253 & 16 + 0.52 kev & & * 866 * & 412 & 113 & 853 & 466 & 13 + 24.5 kev & & & * 487 * & 121 & 474 & 461 & 4 + 4.510 kev & & & & * 136 * & 125 & 133 & 3 + 0.54.5 kev & & & & & * 1114 * & 550 & 169 + 210 kev & & & & & & * 645 * & 78 we calculate the conversion factors from a count rate into a flux in each energy band by using the energy response of the pn . galactic absorption of @xmath54 = @xmath55 @xmath5 @xcite is taken into account . the values are summarized in table 5 for a power law spectrum with various photon indices ( @xmath56 ) . figure 2(a ) shows their dependence on the photon index assuming no absorption , whereas in figure 2(b ) we change the absorption column density at zero redshift for a fixed photon index of 1.8 . in the same figures we also plot the hardness ratios @xmath57 , @xmath58 , and @xmath59 as a function of spectral parameters . for the flux conversion , hereafter we assume @xmath60 ( with no absorption ) for the 24.5 kev , 4.510 kev , 0.54.5 kev , and 210 kev band , and @xmath61 for the 0.30.5 and 0.52 kev band , unless otherwise mentioned . ccccccc 1.0 & 0.164 & 0.171 & 0.574 & 1.718 & 0.305 & 1.057 + 1.1 & 0.167 & 0.171 & 0.571 & 1.696 & 0.295 & 1.026 + 1.2 & 0.171 & 0.170 & 0.568 & 1.675 & 0.285 & 0.996 + 1.3 & 0.174 & 0.170 & 0.564 & 1.653 & 0.276 & 0.967 + 1.4 & 0.177 & 0.169 & 0.561 & 1.632 & 0.267 & 0.939 + 1.5 & 0.181 & 0.168 & 0.558 & 1.612 & 0.259 & 0.912 + 1.6 & 0.184 & 0.168 & 0.555 & 1.593 & 0.250 & 0.886 + 1.7 & 0.187 & 0.167 & 0.551 & 1.573 & 0.243 & 0.862 + 1.8 & 0.190 & 0.167 & 0.548 & 1.554 & 0.236 & 0.838 + 1.9 & 0.194 & 0.167 & 0.545 & 1.536 & 0.229 & 0.816 + 2.0 & 0.197 & 0.166 & 0.541 & 1.517 & 0.223 & 0.794 in this subsection , we summarize spectral properties of the sxds sources based on hardness ratio analysis using the above source list . figures 3(a ) shows the flux distribution of sources detected in the 0.54.5 and 210 kev bands . as described below , our sample covers x - ray sources that constitute a major fraction of the xrb below 10 kev . figure 3(b ) shows the histogram of hardness ratio @xmath58 and @xmath59 for those detected in the 0.54.5 and 210 kev bands , respectively . the flux versus hardness ratio plots are given in figures 4 ( 0.54.5 kev flux versus @xmath58 ) and 5 ( 210 kev flux versus @xmath59 ) . for clarity we do not attach error bars to each point in these figures ; the typical @xmath62 errors in @xmath58 and @xmath59 are 0.22 and 0.15 at flux levels of @xmath63 erg @xmath5 s@xmath6 ( 0.54.5 kev ) and @xmath64 erg @xmath5 s@xmath6 ( 210 kev ) , respectively . to examine the mean spectral properties as a function of flux , we produce flux - weighted spectra in the four bands from flux - range limited samples , by summing up the count rate of the individual sources in each energy band . table 6 summarizes their @xmath58 and @xmath59 values with corresponding power - law photon indices . we find evidence that the spectra of the x - ray sources becomes harder at fainter fluxes with a confidence level of 99.99% ( @xmath58 ) and 96% ( @xmath59 ) for the 0.54.5 kev sample but with a lower confidence level of 84% ( @xmath58 ) and 90% ( @xmath59 ) for the 210 kev sample . their mean slope is similar or even harder than that of the xrb , @xmath65 over the 0.510 kev band at these flux levels . this result is consistent with the fact that we already resolved a major parts of the xrb and that at brighter fluxes where 2030% of the xrb is produced , the mean spectra are much softer than the xrb spectrum @xcite . cccccccc 0.54.5 kev & 0.52.0&406 & & @xmath66 & @xmath67&@xmath68 & @xmath69 + 0.54.5 kev & 2.08.0&107 & & @xmath70 & @xmath71&@xmath72 & @xmath73 + 210 kev & 1.04.0&334 & & @xmath74 & @xmath75&@xmath76 & @xmath77 + 210 kev & 4.016&38 & & @xmath66 & @xmath67&@xmath78 & @xmath79 figure 6 shows the color - color plot between @xmath58 and @xmath59 by using 208 sources detected in the 0.54.5 and/or 210 band whose statistical errors in both hardness ratios are smaller than 0.2 . the solid curve tracks the color of a power law spectrum with varying photon indices from 0 to 4 , while the dashed curve does that with varying absorption column densities at @xmath80 ( a typical redshift of our sample ; see figure 14 ) from log @xmath54 = 0 to 24 for a fixed photon index of 1.8 . as noticed , the spectra of some sources are not simply represented by an absorbed power law model . in particular , there is a population of sources that are hard in the 210 kev band ( e.g. , @xmath81 ) but soft in the 0.54.5 kev band ( @xmath82 ) , indicating the presence of additional soft components . detailed spectral analysis using information of redshift will be presented in forthcoming papers . to make statistical analysis using the source list , such as determination of log @xmath8 - log @xmath9 relation , it is crucial to have reliable sensitivity maps for each pointing and energy band , i.e. , we need to know a flux ( count rate ) limit as a function of position at a given detection criterion ( i.e. , ml @xmath83 ) . since we utilize the maximum likelihood algorithm , however , it is not trivial to calculate sensitivities by employing an analytic formula , unlike in the case of cell detection . ideally , only a detailed simulation can give the correct estimate of a sensitivity at every position , since it depends on the background , exposure map , and psf in a complex way . to save computing time , we here take an empirical approach by utilizing both simulation and analytical calculation , as described below . first , we directly estimate the flux limit by * simulation * at 2-dimensional grid points with 40@xmath12 spacing over the fov . for a given position , we produce a simulated image where a point source is superimposed on an input background map , using the _ emldetect_program . since we are looking simply for a relation between the emldetect ml value and the flux of the sources , we do not taken into account poisson counting noise in our simulation . we then perform a likelihood fit with _ emldetect _ to the simulated image and derive the ml value for the input source . repeating the simulation assuming several different fluxes for the point source , we obtain an empirical relation between the input flux and output ml value . this relation enables us to calculate the sensitivity for a given threshold of ml at that position . next , we produce full resolution ( 4@xmath12 pixel ) sensitivity maps by * interpolating * between the grid points in the following manner . the flux limit could change quite sensitively with the position , being affected by data gaps between ccd gaps or hot pixel regions . a simple interpolation of the flux limit based on the rough position sampling in the above simulation may not be sufficiently accurate . to estimate the precise position dependence , we utilize analytical sensitivity maps calculated by the _ esensmap _ program , which is based on the cell detection algorithm . generally , the detection likelihood obtained by _ emldetect _ ( ml ) differs from that defined in the cell detection ( ml@xmath84 ) for the same flux . the ratio between ml and ml@xmath85 should be a function of position , primarily depending on the size and shape of the psf . hence , we calculate the ml@xmath84 value corresponding to the flux limit at each grid point , and interpolate them between the grid points as a linear function of position . we finally convert ml@xmath84 into a `` flux limit '' at each position , by referring to a set of sensitivity maps calculated by _ esensmap _ with different thresholds of ml@xmath84 , thus producing the final sensitivity maps . to evaluate systematic errors in the sensitivity map , we compare the results when the positions of the grid points are shifted by 20@xmath12 in each direction . we find that the difference in the flux limit is mostly within a few percent and @xmath8610% at maximum . we regard this ( 10% ) as the maximum relative systematic error in the sensitivity map , which will be considered in the analysis of the spatial distribution of sources ( [ sec : acf ] ) . superposing the sensitivity maps of the seven pointings in the common sky coordinates , we examine , position by position , in which pointing data the best sensitivity is achieved in the 0.54.5 kev band in overlapping regions . as mentioned above , we only refer to the results of a single pointing , at given sky coordinates , to obtain the combined source list and sensitivity maps to eliminate duplication . the resultant sensitivity map in the 0.54.5 kev band is shown in figure 7 . figure 8 shows the area curves @xmath87 for ml@xmath77 ( i.e. , survey area where the detection completeness is guaranteed above the given flux ) obtained from the combined sxds field in the four energy bands , 0.52 kev , 0.54.5 kev , 210 kev , and 4.510 kev . dividing the observed flux distribution by @xmath87 , we first obtain the log @xmath8 - log @xmath9 relations in the differential form , @xmath88 . here , we discard sources whose flux falls below the sensitivity limit at the give position because we consider that such detection may not be reliable . figure 9 shows @xmath88 in the four bands in units of number per square degree per @xmath89 erg @xmath5 s@xmath6 , where the errors represent the @xmath62 poisson error in the number of sources in each flux bin . following previous work , we fit them with a broken power law form , expressed as @xmath90 the model has a break flux @xmath91 , above and below which the slope is @xmath92 and @xmath93 , respectively . to estimate the parameters , we utilize a maximum likelihood method , so that the model best reproduces the observed flux distribution expected from the area curve . we fix @xmath94 at 2.5 in the 0.52 kev , 0.54.5 kev , and 210 kev bands , while in the 4.510 kev band , we adopt a single power law by setting @xmath95 and @xmath96 erg @xmath5 s@xmath6 because the break is not evident in our flux range . the best - fit models are shown in solid lines in figure 9 . table 7 summarizes the best - fit parameters , which are found to be consistent with previous results within uncertainties ( e.g. , * ? ? ? * ; * ? ? ? ccccc 0.52 kev & 2.5 ( fixed ) & @xmath97 & @xmath98 & @xmath99 + 0.54.5 kev & 2.5 ( fixed ) & @xmath100 & @xmath101 & @xmath102 + 210 kev & 2.5 ( fixed ) & @xmath103 & @xmath104 & @xmath105 + 4.510 kev & @xmath106 & & 1.0 ( fixed ) & @xmath107 the differential log @xmath8 - log @xmath9 relation are then integrated toward lower fluxes to produce the log @xmath8 - log @xmath9 relations in the integral form , @xmath108 , where the number density of sources with fluxes above @xmath9 is plotted ( figure 10 ) . the achieved sensitivities , defined as the fluxes at which the sky area falls below 1% of the maximum area ( 1.14 deg@xmath0 ) , are @xmath1 , @xmath2 , @xmath3 , and @xmath4 erg @xmath5 s@xmath6 in the 0.52 kev , 0.54.5 kev , 210 kev , and 4.510 kev band : above these fluxes @xmath109% , @xmath110% , @xmath111% , and @xmath112% of the xrb are resolved , when we adopt the xrb intensity of @xmath113 , @xmath114 , @xmath115 , and @xmath116 erg @xmath5 s@xmath6deg@xmath117 , respectively , taken from table 6 of @xcite with an appropriate flux conversion for the 4.510 kev band . for the estimate of the contribution of bright sources that is not well constrained from the sxds data , we have used the formula by @xcite , who complied the results from _ chandra _ deep surveys , _ xmm - newton _ serendipitous surveys , and the _ asca _ medium sensitivity survey . figure 11 shows the same log @xmath8 - log @xmath9 relations scaled by @xmath118 to stress the deviation from a euclidean slope . the dashed curves represent the best - fit formula by @xcite ; they are plotted in the figure covering only the region where the fit is performed . our results are in good agreement with the previous work within @xmath2010% , indicating systematic errors , if any , is confined within this level . the red points are the results of the cosmos survey @xcite . here we convert the result by @xcite in the 4.57.5 kev band and that by @xcite in the 510 kev band to the 4.510 kev band assuming a photon index of 1.5 ( the flux conversion factor is 1.686 and 1.124 , respectively ) . as seen from the figure , the cosmos source counts in the 210 kev are 2030% smaller than ours and the @xcite results . this is most probably because while our results are truly based on the 210 kev survey , @xcite make this plot from the 24.5 kev survey by converting the flux into the 210 kev band assuming a photon index of 1.7 . this could easily miss a population of sources with hard spectra or underestimate their fluxes . in the 0.52 kev band , our source counts at @xmath119 erg @xmath5 s@xmath6 are systematically larger than both @xcite and @xcite but are consistent with the latest results from the _ xmm - newton _ lockman hole survey @xcite . investigating the evolution of the large scale structure of the universe is one of the main scientific objectives of the sxds . from x - ray surveys , we can study the clustering properties of agns including both unobscured and obscured populations over a wide redshift range as a tracer of the large scale structure . this also helps us to understand the agn phenomena and their environment by measuring the mass of underlying dark matter halo where the agn activity took place . many studies have shown the presence of the significant large scale structure in the spatial ( both 2-dimensional and 3-dimensional ) distribution of x - ray sources ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? this leads pencil beam surveys to be inevitably subject to cosmic variance . the sxds , thanks to its wide and continuous area coverage , provides us with an ideal opportunity to investigate this issue . figure 12 shows the location of sources detected in the 0.52 , 0.54.5 , or 210 kev band ( with ml@xmath77 ) in our source catalog . figure 13 shows the log @xmath8 - log @xmath9 relation separately derived from the seven pointings . as done in the analysis of 3.3 , here we excluded any data with shorter exposures in overlapping regions of multiple pointings , and hence the results are statistically independent one another . it can be seen that the source counts show significant variation among the pointings , depending on the flux limit . in particular , those of sds7 in the 0.52 and 0.54.5 kev band are by more than 30% smaller than the other fields in the medium to high flux range . these results indicate that the cosmic variance indeed exists over an area scale of one _ xmm - newton _ fov , @xmath1200.2 deg@xmath0 , which is much larger than the fov of _ chandra _ observatory . the most straightforward approach to quantify the clustering properties of sources is to calculate the angular two - point correlation function ( or auto - correlation function ; hereafter acf ) . the acf , @xmath121 , is defined as @xmath122d\omega_1 d\omega_2,\ ] ] where dp gives the probability of finding a pair of two objects located at the differential solid angle of @xmath123 and @xmath124 with a separation angle of @xmath125 , and @xmath126 is the mean number density . to estimate the acf , we essentially follow the same procedure as done in @xcite . we adopt the @xcite estimator , @xmath127 where @xmath128 , @xmath129 , @xmath130 are the normalized number of data - data pairs , data - random pairs , and random - random pairs , respectively . in calculating the acf , it is critical to ensure that the sensitivity map is correctly modeled at every position . hence , considering possible position - dependent systematic errors in the sensitivity maps created in 4.1 , we introduce a `` safety factor '' , @xmath131 , to multiply the nominal sensitivity limits and select only sources whose observed flux is higher than the ( corrected ) sensitivity limit at the detected position . the choice of @xmath131 is quite conservative , and indeed we obtain essentially the same results even when @xmath132 is adopted . we make a random sample whose number is 10 times that of the actual data . the flux distribution of the random sample is taken to be the same as in the actual source list , and their positions are randomly distributed by satisfying that the flux must be higher than the sensitivity limit at the allocated position . figure 14 show our results of the acf obtained in the 0.52 and 210 kev bands . the flux range and number of sources used in the analysis are summarized in the 2nd and 3rd columns of table 8 , respectively . the attached error is estimated by simulation : we produce 100 sets of random samples with the same source number as in the data , obtain a standard deviation of the acf signal obtained by the same procedure as applied for the actual data , and then multiplied it by @xmath133 . we fit the acf with a power law form @xmath134 in the range of @xmath135 @xmath136 , considering the size of the psf of _ xmm - newton_. we fix @xmath137 @xcite , which is difficult to constrain by our data , and derive the correlation length @xmath138 as a free parameter . to take into account the coupling of the acf between different bins , we utilize the `` covariance matrix '' , which is also obtained from the simulation of the random sample described above ; we refer the readers to 3.4 of @xcite for details . in the fit , we also consider an offset produced by the integral constraint , @xmath139 , assuming that the power law form of the acf holds over the whole area . the obtained values of @xmath138 are listed in the 6th column of table 8 . ccccccccc 0.52 kev & 0.0622 & 765 & & 0.510 & 1.8 & 5.9@xmath140 & 1.3 & 14.9@xmath471.1 + 210 kev & 0.3233 & 573 & & 0.510 & 1.8 & 0.1 ( @xmath141 ) & 1.1 & 2.3 ( @xmath142 ) we detect an acf signal from the 0.52 kev selected sample with @xmath11@xmath12@xmath13 . the signal is not significantly found from the 210 kev sample , however , with a 90% confidence upper limit of @xmath143 1.5@xmath12 . by limber s equation @xcite , the acf can be transformed to the 3-dimensional correlation function @xmath144 once the redshift distribution of the sample is known . here we assume the comoving clustering model where the correlation length @xmath145 is constant in the comoving coordinates , hence @xmath146 . figure 15 shows the agn redshift distribution for the 0.52 and 210 kev samples estimated from the model by @xcite . from the @xmath138 values and redshift distribution , we finally obtain @xmath147 @xmath148 mpc and @xmath149 @xmath148 mpc with a medium redshift of 1.3 and 1.1 for the 0.52 and 210 kev selected sample , respectively . the bias parameter , the ratio of the rms fluctuation amplitude @xmath150 @xcite between agns and underlying mass , is found to be about 5.8@xmath151 for the 0.52 kev selected agns at a median redshift of 1.3 . the correlation length in the 0.52 kev is consistent with the previous results from soft x - ray surveys at similar redshift range reported by e.g. , @xcite , @xcite , and @xcite , although it is larger than those obtained from the optically selected qsos @xcite . by contrast to several previous works that reported strong correlation of hard - band ( @xmath152 kev ) selected samples @xcite , we find no evidence for significant clustering signals in the hard band . the detection of acf in the 0.52 kev band but not in the 210 kev band is similar to what was found by @xcite and @xcite . it is curious that the first result from the xmm - cosmos reports a relatively large correlation length for the 4.510 kev band sample but smaller for the 24.5 kev band sample @xcite . our 210 kev result is consistent with the cosmos 24.5 kev result . we do not find significant acf signals from the 4.510 kev sample , although the result is more subject to the statistical fluctuation due to the small number of sources detectable in this energy band . we investigate the long term variability of x - ray sources detected in the sds-4 field , for which two observations were performed with a time separation of about 2.5 year . the exposures of both observations are similar and are sufficiently long ( @xmath152 ksec ) . using the x - ray catalog produced from the combined images of the two observations , we examine the fluxes of the cataloged sources separately in each epoch by fixing their positions . to ensure high signal - to - noise ratio , we only use sources whose summed ml value from the 0.52 kev and 24.5 kev bands exceeds 15 in the combined catalog and vignetting - corrected exposure is longer than 15 ksec in both observations . figure 16 shows the comparison of the 0.54.5 kev flux between the first and second epochs for this sample . as noticed from the figure , many sources at intermediate fluxes that have sufficiently small error bars show a significant variability , demonstrating the impact of time variability in studying agn properties . the results are consistent with previous studies @xcite reporting that the fraction of variable sources on time scale of months to years is 8090% or higher in _ chandra _ and _ xmm - newton _ deep fields . ] we have presented the first x - ray source catalog and their basic x - ray properties in the sxds field , based on the seven _ xmm - newton _ pointings performed between 2000 and 2003 that cover a continuous area of 1.14 deg@xmath0 in the 0.210 kev band . the catalog consists of 1245 sources in total , consisting of those detected either in the 0.30.5 kev , 0.52 kev , 24.5 kev , 4.510 kev , 0.54.5 kev , and 210 kev with detection likelihood larger than 7 . from empirical sensitivity maps derived by simplified simulation , we obtain log @xmath8 - log @xmath9 relations in various energy bands bands , which bridge the flux range between _ chandra _ deep surveys and brighter surveys . clustering properties of x - ray sources are also studied by means of auto correlation function . we detect significant signals that can be fit by @xmath10 with @xmath11@xmath12@xmath13 for the 0.52 kev selected sample at an estimated median redshift of @xmath153 , and an 90% upper limit of @xmath154@xmath12 for the 210 kev selected sample at @xmath155 . two pointing data separated by 2.5 years indicate the importance of flux variability on a time scale of years . our results establish the average properties of x - ray sources at fluxes where a majority of the xrb is produced , being least affected by cosmic variance . the data presented here constitute a major component of the sxds project and shall be used for a number of research programs in combination with other wavelengths data . this work was done as a part of the large international collaboration of the sxds project . we thank the members of the _ xmm - newton _ ssc and the subaru observatory project team . we also thank dr . takamitsu miyaji for his invaluable advise in the calculation of the auto correlation function . part of this work was financially supported by grants - in - aid for scientific research 17740121 , and by the grant - in - aid for the 21st century coe `` center for diversity and universality in physics '' from the ministry of education , culture , sports , science and technology ( mext ) of japan . lrrcccccccccccccc ccccc 0001 & 33.85387 & @xmath1564.92870&2.36 & 0.0 & 19.9 & 3.2 & 0.4 & 26.5 & 5.1 & 0.00@xmath47 0.09 & 3.45@xmath47 0.79 & 1.39@xmath47 0.63 & 0.86@xmath47 0.81 & 1.00@xmath470.05&@xmath1560.42@xmath470.21&@xmath1560.24@xmath470.49 & 5 & 15.6 & 9.6 & 2.6 & + 0002 & 33.85392 & @xmath1564.90221&2.24 & 16.1 & 33.7 & 2.5 & 0.8 & 37.1 & 6.2 & 2.69@xmath47 0.74 & 6.33@xmath47 1.25 & 1.38@xmath47 0.72 & 1.18@xmath47 0.90 & 0.40@xmath470.14&@xmath1560.64@xmath470.16&@xmath1560.08@xmath470.46 & 5 & 16.0 & 5.5 & 2.6 & + 0003 & 33.86346 & @xmath1565.03354&2.88 & 0.3 & 0.0 & 0.0 & 6.8 & 0.4 & 7.5 & 0.24@xmath47 0.27 & 0.18@xmath47 0.31 & 0.09@xmath47 0.25 & 2.72@xmath47 0.82 & @xmath1560.13@xmath471.01&@xmath1560.33@xmath471.43&0.94@xmath470.17 & 5 & 14.8 & 12.2 & 2.6 & + 0004 & 33.87124 & @xmath1564.96292&2.99 & 5.2 & 14.1 & 0.0 & 0.0 & 11.1 & 0.0 & 0.79@xmath47 0.34 & 2.33@xmath47 0.57 & 0.00@xmath47 0.11 & 0.00@xmath47 0.28 & 0.49@xmath470.19&@xmath1561.00@xmath470.09&&5 & 14.2 & 15.5 & 2.6 & + 0005 & 33.87257 & @xmath1564.89099&1.85 & 2.5 & 11.8 & 6.7 & 0.7 & 22.4 & 8.6 & 0.68@xmath47 0.38 & 2.85@xmath47 0.73 & 1.76@xmath47 0.55 & 0.88@xmath47 0.69 & 0.61@xmath470.19&@xmath1560.24@xmath470.19&@xmath1560.33@xmath470.38 & 5 & 15.2 & 9.7 & 2.6 & + 0006 & 33.87878 & @xmath1564.98760&2.84 & 0.0 & 2.6 & 2.9 & 0.0 & 8.7 & 2.8 & 0.00@xmath47 0.04 & 0.89@xmath47 0.41 & 0.61@xmath47 0.29 & 0.00@xmath47 0.38 & 1.00@xmath470.10&@xmath1560.19@xmath470.32&@xmath1561.00@xmath471.24 & 5 & 13.6 & 17.9 & 2.5 & + 0007 & 33.89418 & @xmath1565.06675&0.56 & 131.3&554.7 & 5.5 & 0.0&484.2 & 4.3 & 6.86@xmath47 0.88&25.08@xmath47 1.56 & 1.20@xmath47 0.43 & 0.03@xmath47 0.25 & 0.57@xmath470.05&@xmath1560.91@xmath470.03&@xmath1560.94@xmath470.40 & 5 & 13.6 & 14.1 & 2.8 & + 0008 & 33.90476 & @xmath1564.94904&1.25 & 24.9 & 41.6 & 6.7 & 0.0 & 50.9 & 5.9 & 1.56@xmath47 0.33 & 3.57@xmath47 0.56 & 0.85@xmath47 0.27 & 0.00@xmath47 0.21 & 0.39@xmath470.11&@xmath1560.62@xmath470.11&@xmath1560.99@xmath470.49 & 5 & 12.3 & 16.8 & 2.2 & + 0009 & 33.90754 & @xmath1565.08932&2.64 & 0.0 & 6.8 & 3.4 & 0.0 & 13.6 & 3.0 & 0.00@xmath47 0.06 & 1.75@xmath47 0.56 & 0.77@xmath47 0.34 & 0.14@xmath47 0.32 & 1.00@xmath470.06&@xmath1560.39@xmath470.23&@xmath1560.69@xmath470.61 & 5 & 13.4 & 17.3 & 2.7 & + 0010 & 33.91171 & @xmath1565.01450&1.30 & 10.6 & 72.1 & 6.0 & 2.3 & 75.5 & 11.0 & 0.89@xmath47 0.25 & 4.53@xmath47 0.59 & 0.73@xmath47 0.25 & 0.90@xmath47 0.44 & 0.67@xmath470.09&@xmath1560.72@xmath470.09&0.11@xmath470.29 & 5 & 11.8 & 17.4 & 2.2 &

we present multi - frequency vla , multicolor ccd imaging , optical spectroscopy , and _ rosat _ hri observations of the giant fr ii radio galaxy nvss 2146 + 82 . this galaxy , which was discovered by the nrao vla sky survey ( nvss ) , has an angular extent of nearly 20 from lobe to lobe . the radio structure is normal for an fr ii source except for its large size and regions in the lobes with unusually flat radio spectra . our spectroscopy indicates that the optical counterpart of the radio core is at a redshift of , so the linear size of the radio structure is mpc , . this object is therefore the second largest fr ii known ( 3c 236 is mpc ) . optical imaging of the field surrounding the host galaxy reveals an excess number of candidate galaxy cluster members above the number typically found in the field surrounding a giant radio galaxy . wiyn hydra spectra of a sample of the candidate cluster members reveal that six share the same redshift as nvss 2146 + 82 , indicating the presence of at least a `` rich group '' containing the fr ii host galaxy . _ rosat _ hri observations of nvss 2146 + 82 place upper limits on the x - ray flux of ergs s for any hot igm and ergs s for an x - ray agn , thereby limiting any x - ray emission at the distance of the radio galaxy to that typical of a poor group or weak agn . several other giant radio galaxies have been found in regions with overdensities of nearby galaxies , and a separate study has shown that groups containing fr iis are underluminous in x - rays compared to groups without radio sources . we speculate that the presence of the host galaxy in an optically rich group of galaxies that is underluminous in x - rays may be related to the giant radio galaxy phenomenon .

astro-ph0002033

the `` giant '' radio galaxies ( grgs ) , which we define as double radio sources whose overall projected linear extents exceed 2@xmath2 mpc , are interesting as extreme examples of radio source development and evolution . members of this class , which comprise only a few percent of all powerful extragalactic radio sources , have been documented for almost 25 years ( e.g. , @xcite ) . they have been used to constrain the spectral aging and evolution of radio sources and as tests for the evolution of conditions in intergalactic environments on mpc scales ( @xcite ; @xcite ; @xcite ) . their 1.4 ghz radio powers are generally in the regime @xmath8 @xmath9 w hz@xmath6 , just above the transition between fanaroff - riley types i ( plumed ) and ii ( lobed ) radio structures ( @xcite ) . it is unclear whether the giant sources are examples of unusually long - lived ( and directionally stable ) nuclear activity in radio - loud systems , or of the development of sources in unusually low - density environments . because of their large angular sizes , nearby giant radio galaxies can in principle be studied in great detail , but their largest - scale structures may be over - resolved and undersampled by interferometers . they have traditionally been discovered through sky surveys with compact interferometers or single dishes at relatively low frequencies , where angular resolution is modest but large fields of view and diffuse steep - spectrum structures can be imaged more easily . the source nvss 2146 + 82 was noted as a candidate giant radio galaxy when it appeared in the first 4@xmath10 by 4@xmath10 field surveyed by the nrao vla sky survey ( nvss : @xcite ) , a northern - hemisphere survey at 1.4 ghz using the vla d configuration at 45 ( fwhm ) resolution . figure [ fig : nvsscontours ] shows contours of the nvss image at 45 resolution . there are two symmetric , extended lobes ( d and e ) on either side of an unresolved component c , plus an unusually large number of other radio sources within 10 of c. two of these ( a and b ) are also symmetrically located around c. comparison with the digital sky survey ( dss ) showed that source c coincides with an @xmath118@xmath11 mag elliptical galaxy to within the uncertainties in the nvss and dss positions . if the elliptical galaxy is the host of an unusually large radio source ( c+d+e ) , then the apparent magnitude suggests that the whole structure may be similar in linear scale to 3c236 . the dss also shows a nearby image that might be another galactic nucleus , and a faint extended feature suggesting a possible `` tail '' or interaction . we have undertaken several observational studies of the radio and optical objects in the field to determine their nature and to clarify the relationships between the optical and radio sources . these studies include : 1 . high resolution radio imaging at 4.9 and 8.4 ghz to locate any compact flat - spectrum radio components in the field , and thus to identify any agn that could be responsible for some or all of the other radio emission , 2 . a search for fainter diffuse radio emission between the d and e components that might link them together or to other sources in the field and thus clarify their physical relationship , 3 . higher - resolution radio imaging of the other radio sources in the field to explore whether they might be physically related to the diffuse components , or to each other by gravitational lensing , 4 . optical spectroscopy of both optical `` nuclei '' and other galaxies in the field , 5 . ubvri optical photometry of the field , and 6 . x - ray imaging using _ rosat _ hri observations to search for any hot x - ray emitting gas which might be associated with an overdensity of galaxies or non - thermal x - ray emission from an agn . throughout this paper , we assume a hubble constant @xmath3 km s@xmath6 mpc@xmath6 . at a redshift of @xmath12 , the angular diameter distance to the radio galaxy is 708.4@xmath2 mpc , the luminosity distance is 928.7@xmath2 mpc , and @xmath13 corresponds to 206@xmath2 kpc . table [ vlaobslog ] gives a journal of our vla observations . the observations in the a configuration were designed to locate any compact radio components in the field . those in the b , c , and d configurations were intended to image the largest scale emission in enough detail to reveal any relationships and connections between the extended components , as well as to determine their spectral and faraday rotation / depolarization properties . the bnc configuration data were designed as a sensitive search for connections , such as jets , between the central radio source and the extended features . the flux density calibration was based on 3c48 and 3c286 . the on - axis instrumental polarization corrections were determined from observations of the unresolved synthesis phase calibrator 2005 + 778 , and the absolute polarization position angle scale from observations of 3c286 . multiple observations of 3c286 and other polarized sources were used to detect problems with ionospheric faraday rotation , but none was noted in any of the sessions . the data were calibrated using the source 2005 + 778 as an intermediate phase reference , then self - calibrated using aips software developed by w. d. cotton for the nvss survey . due to the large size of this source , 1.4 and 1.6 ghz observations used three pointings ; one on the central source c , and one near the center of each putative lobe . the b , c , and d vla configuration observations were made at 1.365 and 1.636 ghz to measure rotation measure and spectral index . the data from these frequencies were calibrated and imaged separately . data taken in the bnc configuration were in two adjacent 50 mhz bands centered on 1.4 ghz . since the source extent is comparable to that of the antenna pattern and the bandwidth used was relatively large , the deconvolution ( clean ) and self calibration applied corrections for the frequency dependence of the antenna pattern . data from each of the three pointings were imaged independently and combined into a single image by interpolating the images onto a common grid , averaging weightings by the square of the antenna power pattern , and correcting for the effects of the antenna pattern . the 0.3 ghz observations were of limited use owing to interference . the most sensitive image of nvss 2146 + 82 is derived from our bnc configuration data at 1.4 ghz which has a resolution of 13(fwhm ) . figure [ fig : nvsslhi ] shows logarithmic contours of the total intensity in the region around the source in this image ; the rms noise is 20 @xmath14jy per clean beam area . a gray scale representation of the same image showing the filamentary structure of the lobes is given in figure [ fig : bncgray ] . figure [ fig : nvssjets ] shows the inner region of this image contoured to lower levels using an initially linear contour interval . the structures of the extended features d and e shown in figures [ fig : nvsslhi ] and [ fig : bncgray ] are entirely consistent with their being associated with each other as the two lobes of a large frii double source of overall angular size 195 . both features are brightest in the regions furthest from c , contain bright ( but resolved ) substructure near their outer edges resembling the hot spots of frii sources , and have their steepest brightness gradients on their outer edges . the overall length of the two lobes is the same to within 5% . although features a and b in figure [ fig : nvsscontours ] appear symmetric around feature c , the higher resolution vla images ( figures [ fig : nvsslhi ] and [ fig : bncgray ] ) reveal them to be background sources , unrelated to nvss 2146 + 82 . the northern feature ( d ) contains a region of enhanced emission ( hot spot ) at its northern extremity with about 65 mjy in an area 30by 18 and an l - shaped extension to the west . the southern feature ( e ) has 75 mjy in a region of enhanced emission 50 by 30 ( a `` warm spot '' ) recessed by 10% of the distance from the core and sharp brightness gradients around its southern and western boundaries . both regions of enhanced emission show evidence of finer , but resolved , structure in our data taken in the b configuration ( see contour plots in figure [ fig : hotspots ] ) . figure [ fig : bncgray ] clearly shows that the internal brightness distributions of both lobes are non - uniform , and suggest the presence of filamentary structures , again a common characteristic of frii radio lobes at this relative resolution . most importantly , figures [ fig : nvsslhi ] , [ fig : bncgray ] , and [ fig : nvssjets ] also show that these lobes are linked to the central compact feature c by elongated features that are plausibly the brightest segments of a weak jet - counterjet system . these features are labeled in figure [ fig : nvssjets ] . we interpret the following features as belonging to the jet in the south lobe . * j1*. this feature is clearly part of a jet that points towards the south lobe but not directly at the peak of feature e. * j2*. this feature ( 15 from c ) and feature k ( 14 to the north of c ) are roughly symmetric in distance from c and in intensity but are not quite collinear with c. on both sides of the source the jet becomes harder to trace further into the lobe . j2 appears to be south of the c j1 direction , suggesting a southward bend , however . * j3*. this feature is plausibly a knot in the continuation of the jet into the south lobe . the lobe brightens beyond j3 and contains a diffuse ridge that is a plausible continuation of the ( possibly decollimated ) jet in the direction of the `` warm spot '' e. the north lobe also brightens at about the same distance from c although there is no feature corresponding to j3 in the north . table [ table : fluxes ] gives flux density estimates for the main features of the source . we estimate that the jet and counterjet together comprise about 1% of the total flux density of the extended lobes , a typical jet prominence " for radio galaxies slightly above the fri ii transition . the higher - resolution radio images provide no evidence that sources a , b , or f in figure 1 are physically related to each other , or to c , d and e. although none can be optically identified , we consider it likely that these are three ( or more ) unrelated background sources . the symmetrical alignment of a and b around c is apparently coincidental , and there is no evidence for any radio `` bridge '' between these sources and component c. the polarization structure derived from the sensitive bnc configuration observations is shown in figure [ fig : polarization ] . the 1.4 and 1.6 ghz data are sufficiently separated in frequency to enable us to measure faraday rotation but still maintain comparable surface brightness sensitivity . the derived rotation measure images of the two lobes are shown in figure [ fig : rm ] . the rotation measure distribution over the north lobe is featureless but several filamentary rotation measure structures can be seen over the southern lobe . the average rotation measure is about the same in the two lobes , -9 rad m@xmath15 in the north and -8 rad m@xmath15 in the south . the faraday rotation measure in the south lobe has a somewhat larger root mean square variation , 8 rad m@xmath15 compared to 5 rad m@xmath15 in the north . figure [ fig : wenss ] shows the 0.35 to 1.4 ghz spectral index distribution inferred from comparing the wenss ( @xcite ) image with our bnc configuration image convolved to the same resolution . the northern and southern warm spots have spectral indices , as used here is given by @xmath16 . ] @xmath17 of -0.6 and -0.55 , not unusual for the hot spots of fr ii sources in this frequency regime . the background sources also exhibit spectral indices that are quite typical of extragalactic sources ( a , -0.68 ; b , -1.0 ; f , -0.7 ) . near the centers of the north and south lobes of nvss 2146 + 82 , however , this comparison shows regions of unusually `` flat '' spectral index ( @xmath18 in the north lobe , @xmath19 in the south lobe ) . the spectral index variations across the lobes can also be studied from our 1.36 and 1.63 ghz data . due to the low surface brightness the data were tapered to 55@xmath20 resolution before imaging for this comparison . to eliminate any complication from the mosaicing technique , only data derived from the pointing on a given lobe were used to determine the spectral index variations for that lobe . thus , the data from two pointings were imaged independently at 1.36 and 1.63 ghz , corrected for the antenna power pattern , and spectral index images were derived independently for the two lobes . these results are shown in figure [ fig : si ] . the close spacing of the frequencies makes determining the spectral index more difficult ; but this is compensated to some extent by the nearly identical imaging properties at the two frequencies , which reduce systematic errors . these data sets are fully independent of those used for the spectral index image in figure [ fig : wenss ] , but also reveal symmetric regions of unusually flat spectral index , @xmath21 -0.3@xmath22 , in both lobes . we conclude that four independent data sets show evidence for regions with @xmath23 -0.3 in regions of relatively high signal to noise ratio . these regions are not artifacts of `` lumpiness '' in the zero levels of the images . nvss 2146 + 82 is not aligned along a single axis . the two warm regions ( e and d ) and the core ( c ) are not collinear . the jet in the south appears to have several bends ; one near the end of j1 ( see figure [ fig : nvssjets ] ) where it bends toward j2 , a change in position angle from -150@xmath10 to -170@xmath10 . beyond j3 , the ridge line of the lobe is fairly well defined and is again at position angle -150@xmath10 , consistent with a second bend ( apparently @xmath24 20@xmath10 ) in the neighborhood of j3 . the jet is not so prominent in the north but feature k , which may be the brightest part of a counterjet , is elongated along position angle of -169@xmath10 . the general `` c '' shape of the source suggests that the overall misalignment is due to environmental effects that have bent the jets , rather than to a changing initial jet direction which is likely to produce overall `` s '' symmetry . we consider it beyond doubt that c , d and e comprise a single large frii radio source with weak radio jets , whose parent object is the galaxy identified with c. optical photometric and spectroscopic observations were obtained to identify the host galaxy of the radio emission and to measure its redshift . we began the search for the optical counterpart to the radio source using the digitized sky survey ( @xcite ; hereafter dss ) . the radio core is aligned with an elliptical galaxy on the dss image to within the astrometric accuracy of the radio and optical positions from the nvss and dss . there is also a second , equally bright object a few arcseconds east of the galaxy at the radio core position . finally , in the dss image , there appears to be s shaped diffuse emission that passes through both bright `` nuclei '' . therefore , our initial assumption was that the host galaxy of nvss 2146 + 82 was possibly a disturbed , double nucleus galaxy . in the following sections , we summarize the optical imaging of the field surrounding the candidate host galaxy and the spectroscopic observations of this host galaxy and its candidate galactic companions . u , b , v , r , and i ccd observations were obtained at the 1.52-m telescope at palomar observatory on the nights of 7 - 9 january 1997 . in addition , u , b , v , and i ccd observations were made at kitt peak national observatory on 4 april 1997 . the palomar 1.52-m observations were made with a 2048 @xmath25 2048 ccd with a pixel scale of 037 per pixel , resulting in a 1263 field of view . though photometric , the seeing was poor ( @xmath26 on 7 january , @xmath27 on 8,9 january ) during the palomar run , so higher resolution ( @xmath28 seeing ) images were obtained with the kpno 4-m telescope in april . the kpno observations were made with the prime focus t2 kb ccd with a pixel scale of @xmath29 per pixel , resulting in a 16@xmath30 field of view . because the kpno data were not taken in photometric conditions , the palomar data remained useful for calibration . data from both observing runs were reduced using the standard iraf ccdred reduction tasks . after the initial reduction , aperture photometry was performed on the host galaxy of nvss 2146 + 82 using the iraf package apphot . unfortunately , due to the poor seeing on the first night of the palomar run and the proximity of the foreground star ( see 3.3 ) to the agn host , it was impossible to photometer nvss 2146 + 82 without significant flux contamination from the foreground star . therefore , we used the daophot ii package ( @xcite ) to psf fit and subtract stars from the palomar nvss 2146 + 82 images . after the foreground star was subtracted , photometry of the galaxy was performed identically to the photometry of several @xcite standard stars . approximately 20 stars were selected from each frame containing the agn host galaxy . a circular aperture 2.5 times the average fwhm of these stars was used to measure the flux of the host galaxy . this aperture was chosen to be consistent with the standard star photometry and because it completely enclosed the host without including contaminating flux from other nearby objects . once instrumental magnitudes for the galaxy were determined , they were transformed to the standard system using transformation equations incorporating an airmass and color term that were determined for the landolt standard stars . the results of our u , b , v , r , & i photometry of the host galaxy are listed in table [ table : magnitudes ] . optical spectra of nvss 2146 + 82 were obtained at kitt peak national observatory on 9 december 1996 . the spectroscopic observations were made with the rc spectrograph on the kpno mayall 4-meter telescope . the detector in use was the t2 kb ccd in a 700 @xmath25 2048 pixel format . all exposures were made with a @xmath31 slit width and a 527 lines / mm grating . the spectral resolution , measured using unresolved night sky lines , is @xmath13.4 . the data were reduced using the standard iraf reduction tasks . the extracted spectra were wavelength calibrated using a solution determined from the spectrum of a henear comparison source . finally , spectrophotometric calibration was applied using a flux scale extrapolated from several standard star spectra . spectra of candidate galactic companions to nvss 2146 + 82 ( see 3.5 below ) were obtained with the hydra multi - fiber positioner and the bench spectrograph as part of the wiyn queue experiment over the period of 14 - 22 september 1998 . the t2kc ccd was used as the spectrograph detector in its spatially binned 1024 @xmath25 2048 pixel mode . all exposures were made with the red fibers , the simmons camera , and a 400 lines / mm grating . the spectral resolution in this configuration is @xmath14.5 . we calculated an astrometric solution for the kpno 4-m frame of the nvss 2146 + 82 field using positions for stars in the frame taken from the usno a1.0 catalog ( @xcite ) . using this solution , we derived positions with the accuracy required by the hydra positioner for our target galaxies . due to fiber placement restrictions and the density of our target galaxies on the sky , we were only able to place 46 fibers on targets . the remaining 50 fibers were randomly placed on blank sky , and they were used during the reduction process for night sky subtraction . the nine 30 minute program exposures were reduced using the iraf dohydra script . the weather conditions during the last two nights were poor , and the spectra from these nights were not usable . therefore the final spectra were obtained by co - adding only the data from nights one and two , a total of two hours of integration . in figure [ fig : oldfig2 ] , we present a contour plot of the v band surface brightness from the central @xmath32 region of the kpno 4-m image after smoothing with a 3 pixel by 3 pixel boxcar kernel . although we find that the elliptical galaxy at the radio core position ( @xmath33 , @xmath34 j2000.0 ) has a narrow line agn emission spectrum with a redshift of @xmath0 , we find that the object just to the east , which was assumed to be potentially a second nucleus , has a zero - redshift stellar spectrum , indicating it is a foreground star . figure [ fig : oldfig3 ] shows two plots of the wavelength and flux calibrated spectrum of the host galaxy of nvss 2146 + 82 . an unusual feature of the spectrum ( figure [ fig : oldfig3 ] ) of the agn is that all of the emission lines appear to be double peaked . the second panel in figure [ fig : oldfig3 ] shows an expanded view of the @xmath35 $ ] doublet clearly showing the double peaked profile of the emission lines . each emission line was easily fit with a blend of two gaussians , indicating that agn line emission is coming from two sources with a velocity separation of @xmath1450 km s@xmath6 . since the agn emission line spectrum gives two different velocities , we have decided to take the velocity of the stellar component of the galaxy as the systemic velocity of the galaxy . the stellar absorption line redshift , calculated by cross - correlating the host galaxy spectrum with the spectrum of the star immediately to the east , is 0.1450@xmath360.0002 . table [ table : linedata ] lists the properties of the observed emission features in the spectrum of nvss 2146 + 82 . the redshifts of the agn emission line components were calculated by identifying features and taking the average redshift of all of the identified features . in this way , the two agn emission line components have been measured to be at velocities of 40070@xmath3650 km s@xmath6 and 40520@xmath3650 km s@xmath6 , which corresponds to redshifts of 0.1440@xmath360.0002 and 0.1456@xmath360.0002 respectively . this indicates that the gas which is giving rise to the bluer component of the agn emission line spectrum is moving relative to the stars in the agn host galaxy at @xmath37 km s@xmath6 and the gas emitting the redder lines is moving at 170 km s@xmath6 with respect to the stars . each emission feature identified in table [ table : linedata ] was fit with a blend of two gaussian components ( except for the two weak lines @xmath38@xmath39 $ ] @xmath40 and @xmath38@xmath39 $ ] @xmath41 , where a single gaussian was used ) to determine the line flux . the fluxes listed in table [ table : linedata ] were measured after the spectrum of nvss 2146 + 82 was flux calibrated using the average of four measurements of the calibrator feige 34 . the flux of the calibrator varied significantly among our four separate exposures , and we therefore estimate our spectrophotometry is only accurate to about 20% . in addition to calibration error , there is an additional error in the profile fitting , and therefore the errors listed for the fluxes include both calibration and measurement error . we derived an extinction of @xmath42 ( the large error is due mostly to the calibration error in the fluxes ) using the standard balmer line ratios for case b recombination ( @xcite ) and the extinction law of @xcite . the galactic extinction at the position of nvss 2146 + 82 is given as @xmath43 on the reddening maps of @xcite . this value is consistent with our balmer line derived value , but possibly indicates that there may be some dust in the host galaxy itself . we decided to correct the measured line fluxes for reddening using the mean value we derived of @xmath44 . the errors listed in table [ table : linedata ] for the fluxes do not include the error in the extinction determination . @xcite found that the optical luminosity function of radio galaxy hosts was similar to that of first ranked cluster members , and he noted that their optical morphology was similar to bright e galaxies . although it was therefore generally believed that the hosts of all radio galaxies were ge types , subsequent large surveys of radio galaxies showed a good deal of evidence for peculiar morphologies ( e.g. , @xcite ) . we find that the host galaxy of nvss 2146 + 82 is likely typical , i.e. it is a ge galaxy , but with evidence of some peculiar morphological features . the broadband colors of nvss 2146 + 82 are typical of bright fr ii host galaxies . the absolute magnitude we derive for the host is @xmath45 at @xmath0 if we adopt a k correction of 0.46 magnitudes in the v passband ( @xcite ) . this magnitude is consistent with the host being a ge galaxy , and also is very similar to the mean v magnitude for 50 low redshift fr iis of -22.6 ( @xcite ) . similar to other fr ii host galaxies , we find the optical morphology of the host elliptical of nvss 2146 + 82 to be disturbed . in figure [ fig : oldfig2 ] , the four distinct objects besides the host galaxy and foreground star have been identified as having non - stellar morphologies with the faint object classification and analysis system ( focas , @xcite ) . if these four galaxies share the same redshift as the ge host of nvss 2146 + 82 , they all lie 50100 kpc away from its nucleus , a distance that implies that they may be dynamically interacting with it . figure [ fig : oldfig2 ] also shows what appears to be a bridge of diffuse optical light that almost connects nvss 2146 + 82 to the galaxy to the southwest . this bridge may indicate that this smaller galaxy has recently passed close enough to nvss 2146 + 82 to interact with it gravitationally . there is also a fifth object @xmath46 to the southeast of the center of nvss 2146 + 82 , which could be in the process of merging with the ge galaxy . however , due to the faintness of this object and its proximity to the nucleus of 2146 + 82 , we are unable to classify this object definitively as a galaxy with the focas software . although we can not conclude based on this image that nvss 2146 + 82 is undergoing a merger , its outer isophotes do show evidence that it has been disturbed . correlations between the radio power and optical emission line luminosities in radio galaxies have been established in several studies ( e.g. , @xcite ; @xcite ; @xcite ) . these radio / optical correlations are assumed to arise primarily due to the fact that both the radio jet and the ionization source originate in the central engine . the radio core power at 5 ghz ( @xmath47 = 23.85 $ ] ) and the h@xmath48@xmath39 $ ] luminosity ( @xmath49 = 35.2 $ ] ) for nvss 2146 + 82 lie well within the dispersion in the correlation in these quantities found for low redshift fr iis ( @xcite ) . this apparently indicates that the physical conditions that cause this radio / optical correlation to arise may be similar in this grg and in `` normal '' fr iis . the shape of the emission line profiles in nvss 2146 + 82 are not unique ; emission line profiles and narrow band imaging of seyfert galaxies and radio galaxies have shown evidence for interaction between the radio synchrotron emitting plasma and the optically emitting ionized gas ( see e.g. , @xcite ) . although the majority of objects that show kinematic evidence for interactions between the radio jets and ionized gas clouds tend to have more compact radio structures , the double peaked line profiles seen in nvss 2146 + 82 appear similar to those seen in radio galaxies with jet / cloud interactions . a recent model ( @xcite ) for interactions between the nuclear radio emission and nlr gas in seyferts produces @xmath38@xmath39 $ ] profiles for objects near the plane of the sky that are very similar to the double peaked profiles seen in nvss 2146 + 82 . the model of taylor et al . ( 1992 ) produces double peaks in the line profiles of objects oriented close to the plane of the sky because the emission lines are postulated to arise from gas that is being accelerated as a bowshock expands into the ionized medium surrounding the nucleus . they model the bowshock as a series of annuli , and each annulus contributes most of its luminosity at the two extreme radial velocities found along the line of sight . although the specifics of the model of taylor et al . ( 1992 ) , such as the discrete plasmon emission from the radio nucleus , may not necessarily apply in the case of nvss 2146 + 82 , it suggests that the narrow line profiles observed for this fr ii ( which is assumed to be very near the plane of the sky ) can be produced plausibly in a model where the ionized gas is in a cylindrical geometry around the radio jet . double peaked _ broad _ lines have been observed in radio galaxies ( e.g. , pictor a [ @xcite ] ) , however the model that is typically invoked to explain the broad line profiles requires the radio galaxy to be oriented close to the line of sight . since nvss 2146 + 82 does not show a broad line component and is unlikely to be oriented close to the line of sight , the accretion disk model relied on to fit double peaked broad lines in agn is probably unrelated to the emission line profiles observed in nvss 2146 + 82 . although a jet / cloud interaction appears to be the most reasonable explanation for the double peaked narrow emission lines observed in the spectrum of nvss 2146 + 82 , it is also plausible that a gravitational interaction between the fr ii host galaxy and its nearest companions may be the source of the @xmath1450 km / sec separation between the blue and red emission line peaks . higher spatial resolution long slit spectroscopy is necessary to determine which cause is more likely . deep ccd imaging of the region surrounding the host galaxy of nvss 2146 + 82 has revealed a large number of nearby galaxies . these galaxies are near the limiting magnitude of the poss / dss images , so nvss 2146 + 82 appears to lie in a sparsely populated region of the sky in the dss . however , photometry from the deeper palomar 1.52-m images gives @xmath50 for these nearby galaxies if they also lie at @xmath12 , indicating a possible association with nvss 2146 + 82 . in figure [ fig : oldfig4 ] , we present a region of the kpno 4-m image of nvss 2146 + 82 that is 0.5 mpc on a side and that has all identified galaxies with @xmath51 ( corresponding to @xmath52 at @xmath0 ) circled . these images do not go deep enough to allow accurate identification and photometry of all galaxies to @xmath53 , so this sample is not complete . however , even though the sample shown in figure [ fig : oldfig4 ] is probably only complete to @xmath54 , we have identified 34 candidate galaxies surrounding nvss 2146 + 82 . although there are no previous identifications of the cluster around nvss 2146 + 82 ( at @xmath55 , it is too close to the galactic plane to have been included in the abell [ 1958 ] catalog ) , there is a zwicky cluster to the north , with nvss 2146 + 82 lying only @xmath56 south of the southern border of the zwicky cluster . the zwicky cluster 2147.0 + 8155 ( b1950.0 coordinates ) is a compact group with 56 members classified as `` extremely distant '' or @xmath57 ( @xcite ) . while this gives a redshift for the zwicky cluster larger than that of nvss 2146 + 82 , it is close enough to @xmath12 ( @xmath58 mpc more distant ) that we may be seeing nvss 2146 + 82 in projection against a background rich cluster . in september of 1998 wiyn / hydra spectra were obtained of 46 candidate galactic companions of nvss 2146 + 82 to determine their redshifts . the sample of 46 was selected in the following way : ( 1 ) we selected all objects morphologically classified as galaxies in the kpno 4-m frame by focas with aperture magnitudes @xmath59 , resulting in an initial sample of 205 galaxies . ( 2 ) we divided this group into two subdivisions : the first being all galaxies within 0.5 mpc of 2146 + 82 in projected radius , and the second being all those outside of the 0.5 mpc radius . however , due to exposure time limitations , the available sample taken from the 34 galaxies identified in figure [ fig : oldfig4 ] within 0.5 mpc of the host was reduced to the 17 brightest galaxies . fiber placement restrictions allowed us to observe only 11 of these 17 galaxies . objects from the sample outside of the 0.5 mpc radius from nvss 2146 + 82 were assigned to 35 of the remaining fibers , leaving about 45 fibers on blank sky to allow accurate sky subtraction . unfortunately , as mentioned in 3.2 above , the weather conditions during some of the queue observing were poor , and this limited the success of the program . there was enough signal - to - noise to identify features in the spectra of only 24 of the 46 objects successfully . we found that 7 of the 24 objects with good spectra were actually misidentified stars . nonetheless , from the remaining 17 spectra of galaxies in the field surrounding nvss 2146 + 82 , we were successful in identifying what we believe to be a true cluster that contains the radio source host galaxy . figure [ fig : oldfig5 ] presents an image with the 17 galaxies with measured redshifts marked . the positions , redshifts , and magnitudes for these objects are listed in table [ table : redshifts ] . a quality factor is assigned for each redshift using the 0 ( unreliable ) to 6 ( highly reliable ) scale of @xcite . the quality is determined using : @xmath60 $ ] , where @xmath61 is the number of spectral features that are accurately identified ( less than 5% chance of being incorrect ) and @xmath62 is the number of spectral features that are probably correct ( about a 50% chance of being correct ) . if @xmath63 is adopted as the requirement for a reliable redshift , 5 of the 17 galaxies have unreliable redshifts . the histogram plotted in figure [ fig : oldfig6 ] is a redshift distribution for the 17 galaxies , and it shows that 50% ( 6 ) of the reliable redshifts fall in the range of @xmath64 , with 5 of those having redshifts of @xmath65 . extrapolating the redshift distribution for the sample of galaxies identified around nvss 2146 + 82 from the redshift distribution of the 17 reliable galaxy spectra suggests that the 2146 + 82 cluster may be abell richness class 0 or 1 . of course , the statistics are very uncertain . of the 11 galaxies within a projected distance of 0.5 mpc of nvss 2146 + 82 that were in the wiyn / hydra sample , redshifts were measured for three of them . two of these have @xmath66 , while the third has @xmath67 . we identified features in 21 of the remaining 35 spectra that were measured for objects outside of the projected 0.5 mpc radius . we found that 7 were misclassified stars , and 3 of the 14 galaxies with reliable redshifts had @xmath68 . abell s ( 1958 ) richness criterion was based on the number of cluster galaxies within the range @xmath69 to @xmath70 ( @xmath69 is the magnitude of the third brightest cluster member ) . for the nvss 2146 + 82 cluster , @xmath69 should be @xmath71 , since the third brightest galaxy of the 7 ( which includes nvss 2146 + 82 ) we have found at @xmath12 has @xmath72 . of the 205 galaxies originally found in the kpno 4-m field containing nvss 2146 + 82 , 123 of these fall within the @xmath69 to @xmath73 range used for estimating the abell richness . if we apply the percentages above to this sample of 123 galaxies , then @xmath74 might be at the same redshift as nvss 2146 + 82 . to this point , we have been considering the cluster richness inside of 0.5 mpc , for comparison with the @xmath75 richnesses of allington - smith et al . ( 1993 ) and zirbel ( 1997 ) , and also within an area @xmath13.8 mpc on a side , which is the size of the kpno 4-m field at @xmath0 . however , we must note that the original richness criterion for abell class 1 clusters was that 50 or more galaxies were contained in a radius of 3 mpc for @xmath76 km sec@xmath6 mpc@xmath6 ( @xcite ) . a circle of radius 3@xmath2 mpc at @xmath0 subtends 507 square arcminutes on the sky , nearly twice the amount of area covered in our image . if the calculated optical richness from the 4-m image galaxy sample is taken as a lower limit to the number of galaxies within an abell radius , the richness class of the group surrounding nvss 2146 + 82 appears to be at least abell class 0 . richness class 0 clusters of galaxies typically have luminosities with l@xmath77 ergs s@xmath6 ( @xcite ) , while x - ray agn range from l@xmath78 ergs s@xmath6 ( @xcite ) , so a cluster or bright agn will easily be seen with a medium length exposure with _ rosat_. nvss 2146 + 82 was observed with the _ rosat _ high resolution imager ( hri ) between 1998 february 24 and 1998 march 13 for a duration of 30.3 ksec to search for any hot gas that might be associated with the apparent overdensity of galaxies or for an x - ray luminous agn . the data were analyzed with the iraf post - reduction off - line software ( pros ) . the hri data were filtered for periods of high background and corrected for non - x - ray background , vignetting , and exposure using the computer programs developed by snowden ( @xcite ; @xcite ) . after filtering , the live exposure was 29.8 ksec . the resulting x - ray image was convolved with a gaussian beam with @xmath79 to recover diffuse x - ray emission . the contours of the image are shown superposed on the dss image in figure [ fig : xray ] . a few sources were visible near the edge of the field , but there seem to be no significant sources of x - ray emission associated with any optical or radio sources within the @xmath80 extent of nvss 2146 + 82 ( figure [ fig : xray ] ) . we derived upper limits on both the agn or cluster emission by extracting the x - ray counts from the corrected x - ray image using circular regions centered on the host galaxy of @xmath81 and @xmath82 , respectively . the region sizes were chosen simply because @xmath81 represents the size of a typical hri point source and @xmath82 is roughly 1 - 2 times the typical size of a cluster core at the distance of the radio galaxy . the x - ray background was determined by extracting the x - ray counts from an annulus of @xmath83 centered on the nucleus of the radio host and removing 3 point sources using @xmath81 circular regions . we used pimms ( @xcite ) to convert the hri count rate into an unabsorbed flux in the 0.1 - 2.0 kev band , assuming an emission model and a galactic photoelectric absorption column of @xmath84 @xmath85 ( @xcite ) . for the agn , we assumed a power law with a photon index , @xmath86 , of 2.0 and derived an upper limit at the @xmath87 confidence level of @xmath7 ergs @xmath5 s@xmath6 , or @xmath88 ergs s@xmath6 at the distance of the radio galaxy . similarly for the cluster , we assumed a raymond - smith thermal emission spectrum characterized by @xmath89 kev which yielded an upper limit of @xmath4 ergs @xmath5 s@xmath6 , or @xmath90 ergs s@xmath6 . unfortunately , our limit on the x - ray emission from the radio galaxy is not very stringent . @xcite studied the x - ray properties of several 3cr radio galaxies with the _ einstein _ observatory . they found that the fr ii s radio and x - ray luminosities are strongly correlated . thus with a radio flux of 6.8 mjy at 5 ghz , nvss 2146 + 82 should have a nuclear x - ray flux of a few times @xmath91 ergs s@xmath6 . this flux is comparable to our upper limit . taking into account the intrinsic scatter in the radio / x - ray correlation , our non - detection of the agn is quite reasonable . our upper limit on the x - ray emission from hot cluster gas provides a much stronger constraint . most abell richness class 0 clusters have x - ray luminosities of @xmath92 erg / s ( @xcite ) . therefore any cluster of galaxies associated with the radio galaxy must be either intrinsically weak in x - rays or must be poorer than our optical estimate . @xcite studied the x - ray emission of low - redshift fr ii galaxies and found that poor clusters that contain fr ii sources are underluminous in x - rays compared to similar clusters that do not contain fr iis . the median x - ray luminosity for low-@xmath93 clusters with fr iis was found to be @xmath94 ergs s@xmath6 while it is @xmath95 ergs s@xmath6 for a sample of low-@xmath93 clusters without fr iis ( wan & daly 1996 ) . assuming that the group surrounding nvss 2146 + 82 is similar to that of other clusters found around low-@xmath93 fr iis and is underluminous in x - rays , the optical richness estimate is probably correct . our observations of nvss 2146 + 82 clearly show that it is an unusually large fr ii radio galaxy . its angular distance from the north lobe to the south lobe gives an unusually large extent of @xmath96 . for our assumed cosmology and our measured redshift of @xmath0 , the linear extent of the radio structure is 4@xmath2 mpc , placing it in the giant radio galaxy ( grg ) class , which we define as sources larger than 2@xmath2 mpc . nvss 2146 + 82 is therefore the second largest fr ii known , surpassed only by 3c236 which is @xmath16@xmath2 mpc in extent . fr ii galaxies of this size are extremely rare ; a literature search by @xcite of 540 fr iis contains only 27 objects with sizes greater than 1@xmath2 mpc . of this sample of 27 large fr iis , only 5 are larger than 2@xmath2 mpc . for comparison , the other known giant radio sources are shown in table [ table : giants ] . the log radio luminosity of nvss 2146 + 82 at 1.4 ghz is 25.69 , in the middle of the range for giant radio sources . it remains unclear if there are fundamental differences between grgs and `` normal '' radio galaxies . the relative paucity of known grgs may be in part due to observational selection effects in past radio surveys . an alternative reason for the rarity of giant radio galaxies may be that the physical conditions necessary for the creation of a grg are uncommon in the universe . although the similarity between nvss 2146 + 82 and other fr iis suggests that it is a typical fr ii radio galaxy at the extreme end of the size distribution , a study of a complete sample of radio galaxies that includes grgs will have to be made to determine if grgs are part of a continuous distribution in size of normal radio galaxies or if there are fundamental differences between grgs and smaller fr iis . if the usual equipartition assumptions are made , then it is possible to estimate the magnetic field strength and pressure in the lobes . assuming that the observed spectral index is maintained from 10 mhz to 100 ghz , that there are equal energies in the radiating electrons and other particles , and that the filling factor is unity , the derived magnetic field is b@xmath97 gauss and p@xmath98 @xmath99k for the hot spots . at the midpoint of the lobes these values are b@xmath100 gauss and p@xmath101 @xmath99k . at this redshift , the 3 k microwave background has an equivalent magnetic field of 4.2@xmath102 gauss so the energy loss in the lobes should be dominated by inverse compton scattering of this background , and the time for the electrons radiating at 1400 mhz to lose half of their energy will be @xmath103 years . the mean faraday rotation of @xmath24 @xmath104 rad m@xmath15 shown in figure [ fig : rm ] is consistent with the results of @xcite for other extragalactic sources seen through this region of the galaxy ( @xmath105 ) . it is therefore likely that the rotation measure screen seen in figure [ fig : rm ] is primarily the foreground screen of our galaxy . the low apparent rotation measure and the smoothness of the polarization structure shown in figure [ fig : polarization ] suggests that the magnetic field in this source is well ordered . the field configuration is entirely typical of older extended fr ii sources , with the e vectors lying approximately perpendicular to the ridge line of the radio emission in most features . we note that the greater variance and evidence for organized structure in the faraday rotation of the southern lobe is the opposite of what would be expected if the jet sidedness were due to doppler favoritism and the faraday rotating medium were local to the source . we think it more likely that the faraday rotation structure arises along the line of sight in our galaxy . the spectral index variations shown in figure [ fig : si ] indicate that there are regions @xmath106 back towards c from the brightest region in each lobe that have unusually flat spectra ( @xmath23 -0.3 ) , flatter even than the hot spots . the only extended synchrotron sources known with spectra this flat are a few galactic supernova remnants ( @xcite ) . the spectral index structure in nvss 2146 + 82 is unlike the systematic steepening of the spectrum away from the hot spots that is usually interpreted as an effect of spectral aging in extended lobes . in such interpretations , electrons are presumed to be injected into a high field region in or around the hot spots , and their energy spectrum steepens with distance as they diffuse into lower field regions of the extended lobes . clearly no such interpretation can be made here . these flatter spectrum regions occur in the transition zone from the featureless parts of the lobes ( closer to the core ) to the parts near the regions of enhanced emission that contain significant filamentary structure . the anomalous regions are near the midline of the lobes ; the southern region is centered on the path of the jet and the northern region is at one end of a prominent filament ( the path of the jet is uncertain ) . the relative symmetry of the flatter spectrum regions of the lobes suggests that they might be produced by an intrinsic property of the source , such as a variable spectral index in the injection spectrum of the relativistic electrons from the jet , rather than local environmental effects . if the magnetic field has values near those estimated by the equipartition calculations given above , then the energy loss of the radiating electrons is dominated by inverse compton scattering against the cosmic microwave background . in the low density , low magnetic fields in these lobes , the aging effects will be slow and the history of a variable electron spectrum could be maintained along the length of the lobe . there are three size scales on which symmetries appear or change in the radio structure : the first is @xmath107 kpc . the jets appear to become symmetric on this scale but are asymmetric on smaller scales . if the j2 and k components ( figure 4 ) are symmetric features in the jet and counterjet , any doppler boosting from relativistic motion must have disappeared by this point in the jet . the second scale is @xmath108 kpc . on this scale , there is a dramatic brightening of both lobes . the third scale is @xmath109 kpc . at this distance , the lobes become even brighter and strong filamentary structure appears . this is the distance at which regions of spectral anomaly appear in the extended emission . the largest scale symmetries thus suggest a symmetric overall environment , apart from the slight non - collinearity ( c - symmetry ) of the structure . the small scale brightness asymmetries of the jet and counterjet might be attributed to doppler boosting and dimming by relativistic motion which effectively disappears by @xmath110 kpc , i.e. on a scale more typical of a `` non - giant '' fr ii source . we reiterate however that the small asymmetry in rotation measure dispersion ( variance ) between the lobes is opposite in sign to that expected on this interpretation . this asymmetry seems more likely to reflect an intrinsic asymmetry ( or gradient ) in the foreground magnetoionic medium . one possibility for the origin of grgs is that they are otherwise normal fr ii sources that reside in extremely low density gaseous environments . the environments in which radio galaxies reside have been studied in depth ( e.g. @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) because the gas density and pressure in the host galaxy s ism , any intracluster medium , and the igm are at least partly responsible for determining the resulting radio morphology . an intriguing result of recent studies ( @xcite ; @xcite , @xcite ) is that fr ii galaxies are found in a range of cluster richnesses at moderate redshifts , but they are only found in poor to very poor groups at low redshift . the `` richness '' of the cluster associated with a radio galaxy can be estimated in a statistical sense in the absence of redshift data on nearby galaxies . allington - smith et al . ( 1993 ) define the richness parameter @xmath75 as the number of galaxies within a projected radius of 500 kpc and with @xmath111 assuming the same redshift as the agn . the number counts are corrected for contamination by foreground and background galaxies by subtracting number counts from a field offset from the radio galaxy . zirbel ( 1997 ) gives a conversion of @xmath75 to abell class as @xmath112 . with this conversion , the thresholds for abell classes 0 and 1 are @xmath113 and 26 respectively . using this richness estimation technique , zirbel ( 1997 ) found that of a sample of 29 low redshift ( @xmath114 ) fr iis : ( 1 ) 41% of the sample of low @xmath93 fr iis reside in very poor groups ( @xmath115 ) , and ( 2 ) more importantly , no low redshift fr ii was found in a rich group with @xmath116 . based on the results given in 3.5 , nvss 2146 + 82 appears to reside in a group with an anomalously high galaxy richness compared to other low redshift fr iis . although the galaxy counts from the field surrounding nvss 2146 + 82 were not calculated identically to those of zirbel ( 1997 ) , the value of @xmath75 is likely @xmath117 for nvss 2146 + 82 . the upper limit on the cluster x - ray emission is consistent with the nvss 2146 + 82 group being at the low end of the x - ray luminosity distribution for poor clusters . @xcite found that in a comparison of low redshift clusters with and without fr ii sources , clusters that contained fr iis were underluminous in x - rays compared to clusters without fr iis . although the cluster surrounding nvss 2146 + 82 may be abell class 0 , its lack of associated x - ray gas suggests that the pressure in the surrounding medium is low enough for a giant radio source to form with little disruption of the fr ii jet . curiously , several other grgs listed in table [ table : giants ] also appear to lie in regions with overdensities of nearby galaxies . the grg 0503 - 286 appears to lie in a group of 30 or so galaxies ( @xcite ) . these companions are concentrated to the northeast of the host galaxy of 0503 - 286 , and may have caused the asymmetric appearance of the northern lobe of the radio structure . overdensities of nearby galaxies are also reported for 1358 + 305 ( @xcite ) and 8c 0821 + 695 ( @xcite ) ; however , in both cases there is no spectroscopic confirmation of the redshifts of the candidate cluster galaxies . in a recent study of the optical and x - ray environments of radio galaxies , @xcite find that for a sample of fr i sources , all have extended x - ray emission and overdensities of optical galaxies . however of their sample of seven fr ii sources , none have overdensities of optical galaxies or extended x - ray emission except for the grg da240 , which has no extended x - ray emission but does have a marginally significant excess of optical companions . perhaps for at least some of the grg population , the presence of the host galaxy in an optically rich group with little associated x - ray gas is related to the formation or evolution of the radio source ? we have presented multi - wavelength observations of the unusually large fr ii radio galaxy nvss 2146 + 82 . the overall size of the radio source is @xmath118 mpc , making it the second largest known fr ii source . we have found the host galaxy to be similar in both luminosity and morphology to a sample of other low redshift fr ii galaxies . emission line profiles seen in the spectrum of the host galaxy are double peaked , which may indicate that the ionized gas may be being accelerated by the bipolar radio jet . we have also found evidence for an anomalously rich group of galaxies at the same redshift as nvss 2146 + 82 that has little associated x - ray emitting gas . though unusual in having a rich environment , this source is similar to other low redshift fr iis in clusters ; the nvss 2146 + 82 group is underluminous in x - rays compared to clusters of similar richness that contain no fr ii . the large radio size , lack of significant faraday rotation and non detection of x - rays all suggest that in spite of the richness of the cluster in which this galaxy resides , it has a low gas density . there is some morphological evidence that the host galaxy of nvss 2146 + 82 may be undergoing tidal interaction with one or more of its nearest companions . also , an interaction may be responsible for the double - peaked emission line profiles , however the spatial resolution of the spectrum of the nucleus is not high enough to distinguish between a merger origin or radio jet / cloud interaction origin for the peculiar profiles . apart from the radio spectral index anomaly , the radio properties of this source are like a normal fr ii source scaled up by a factor of ten , preserving the standard overall morphology and polarization structure . in the outer regions of the source the magnetic field is likely to be so weak that inverse compton losses to the cosmic microwave background dominate synchrotron losses . we are grateful to mark whittle for many helpful conversations . we are grateful to matt bershady , randy phelps , and mike siegel for either sharing observing time or taking observations in support of this research . cp acknowledges the support of a grant - in - aid of research from sigma xi , the scientific research society . this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration . we acknowledge the use of nasa s _ skyview _ facility ( http://skyview.gsfc.nasa.gov ) located at nasa goddard space flight center . monet , d. , bird , a. , canzian , b. , harris , h. , reid , n. , rhodes , a. , sell , s. , ables , h. , dahn , c. , guetter , h. , henden , a. , leggett , s. , levison , h. , luginbuhl , c. , martini , j. , monet , a. , pier , j. , riepe , b. , stone , r. , vrba , f. , walker , r. 1996 , usno - sa1.0 , ( washington d.c . : u.s . naval observatory ) vla & observing & center frequencies & bandwidth & number & integration + configuration & date & ( mhz ) & ( mhz ) & of fields & ( min ) + a & 1995 jul 08 & 8415 , 8465 & 50 & 6 & 5 + a & 1995 jul 08 & 4835 , 4885 & 50 & 6 & 6 + b & 1995 dec 23 & 1365 , 1636 & 12.5 & 3 & 13 + c & 1996 feb 15 & 1365 , 1636 & 25 & 3 & 22 + d & 1996 sep 02 & 1365 , 1636 & 25 & 3 & 18 + b & 1995 dec 23 & 327.5 , 333 & 3.1 & 1 & 69 + c & 1996 feb 15 & 327.5 , 333 & 3.1 & 1 & 30 + d & 1996 sep 02 & 327.5 , 333 & 3.1 & 1 & 7.5 + bnc & 1997 jun 17 & 1365 , 1435 & 50 & 3 & 185 + & 0.35 ghz & 1.4 ghz & 4.9 ghz & 8.4 ghz & @xmath17 + total & 0.99 @xmath36 0.02 jy & 0.53 @xmath36 0.1 jy & & & -0.45 @xmath36 0.06 + n lobe & 0.43 @xmath36 0.01 jy & 0.24 @xmath36 0.05 jy & & & -0.42 @xmath36 0.07 + s lobe & 0.52 @xmath36 0.01 jy & 0.27 @xmath36 0.05 jy & & & -0.47 @xmath36 0.06 + c & 23 @xmath36 2 mjy & 13.6 @xmath36 0.5 mjy & 6.8 @xmath36 0.2 mjy & 3.4 @xmath36 0.2 mjy & -0.38 @xmath36 0.03 + j1 & & 1.3 @xmath36 0.2 mjy & & & + j2 & & 0.3 @xmath36 0.2 mjy & & & + j3 & & 0.8 @xmath36 0.2 mjy & & & + k & & 0.7 @xmath36 0.2 mjy & & & + l c c c c c c species & @xmath119 & @xmath120 & @xmath121 & @xmath122 & flux & luminosity + & & & & & @xmath123 erg / sec/@xmath85 & @xmath124 erg / sec @xmath9 + @xmath38@xmath39 $ ] @xmath125 & 4262.5 & 0.1436 & 4269.6 & 0.1455 & 7.4@xmath360.5 & 7.6@xmath360.5 + @xmath38@xmath39 $ ] @xmath126 & 4425.3 & 0.1439 & 4432.6 & 0.1458 & 2.8@xmath360.3 & 2.9@xmath360.3 + @xmath38@xmath39 $ ] @xmath40 & & & 4545.1 & 0.1456 & 0.9@xmath360.1 & 0.9@xmath360.1 + h@xmath127 & 4693.7 & 0.1443 & 4698.2 & 0.1454 & 0.3@xmath360.1 & 0.3@xmath360.1 + h@xmath128 & 4963.9 & 0.1436 & 4971.9 & 0.1455 & 1.4@xmath360.2 & 1.4@xmath360.2 + @xmath38@xmath39 $ ] @xmath41 & & & 4998.8 & 0.1457 & 0.8@xmath360.1 & 0.8@xmath360.1 + @xmath129 & 5360.8 & 0.1440 & 5367.5 & 0.1454 & 0.8@xmath360.2 & 0.8@xmath360.2 + h@xmath130 & 5560.5 & 0.1438 & 5569.2 & 0.1456 & 3.0@xmath360.4 & 3.1@xmath360.4 + @xmath38@xmath39 $ ] @xmath131 & 5672.4 & 0.1439 & 5681.6 & 0.1457 & 12.3@xmath361.1 & 12.7@xmath361.1 + @xmath38@xmath39 $ ] @xmath132 & 5727.2 & 0.1439 & 5736.4 & 0.1457 & 35.7@xmath363.2 & 36.9@xmath363.3 + @xmath38@xmath39 $ ] @xmath133 & 7206.2 & 0.1438 & 7217.4 & 0.1456 & 1.6@xmath360.4 & 1.7@xmath360.4 + @xmath38@xmath39 $ ] @xmath134 & 7493.7 & 0.1444 & 7505.4 & 0.1462 & 1.7@xmath360.3 & 1.8@xmath360.3 + h@xmath135 & 7508.9 & 0.1442 & 7520.3 & 0.1459 & 8.9@xmath361.0 & 9.2@xmath361.0 + @xmath38@xmath39 $ ] @xmath136 & 7530.6 & 0.1438 & 7542.8 & 0.1457 & 5.1@xmath360.6 & 5.3@xmath360.6 + c c c c c c galaxy i d & @xmath137 & @xmath138 & z & q & @xmath139 + 1 & 21:42:18.5 & 81:55:34 & 0.242 & 5 & 20.0 + 2 & 21:42:56.3 & 81:48:29 & 0.378 & 3 & 20.2 + 3 & 21:42:58.5 & 81:57:40 & 0.350 & 5 & 19.4 + 4 & 21:44:47.8 & 81:56:15 & 0.145 & 6 & 18.8 + 5 & 21:45:24.5 & 81:55:05 & 0.144 & 6 & 19.3 + 6 & 21:45:27.7 & 81:57:54 & 0.267 & 6 & 19.4 + 7 & 21:45:54.8 & 81:53:23 & 0.135 & 6 & 18.4 + 8 & 21:46:08.8 & 81:48:08 & 0.123 & 6 & 20.1 + 9 & 21:46:24.3 & 81:57:43 & 0.243 & 6 & 18.4 + 10 & 21:46:48.6 & 82:01:46 & 0.183 & 2 & 19.8 + 11 & 21:47:05.7 & 81:52:35 & 0.144 & 6 & 18.3 + 12 & 21:47:07.2 & 81:55:36 & 0.145 & 6 & 18.4 + 13 & 21:47:20.4 & 81:51:40 & 0.149 & 1 & 19.7 + 14 & 21:47:24.7 & 81:50:40 & 0.173 & 6 & 20.0 + 15 & 21:47:53.7 & 81:53:43 & 0.208 & 2 & 19.4 + 16 & 21:48:19.9 & 82:01:03 & 0.143 & 1 & 17.7 + 17 & 21:48:44.3 & 81:56:59 & 0.148 & 6 & 18.9 + c c c c c c iau name & other name & z & las & log@xmath140 & lls + & & & ( arcsec ) & ( @xmath9 w hz@xmath6 ) & ( @xmath2 mpc ) + 1003 + 351 & 3c236 & 0.0989 & 2478 & 26.37 & 6.04 + * 2146 + 82 * & * nvss 2146 + 82 * & * 0.1450 * & * 1175 * & * 25.69 * & * 3.91 * + 0821 + 695 & 8c 0821 + 695 & 0.5380 & 402 & 26.30 & 2.94 + 1637 + 826 & ngc6251 & 0.0230 & 4500 & 24.73 & 2.89 + 0319 - 454 & & 0.0633 & 1644 & 25.83 & 2.72 + 1549 + 202 & 3c326 & 0.0885 & 1206 & 26.08 & 2.67 + 1358 + 305 & b2 1358 + 305 & 0.2060 & 612 & 25.93 & 2.64 + 1029 + 570 & hb13 & 0.0450 & 2100 & 24.57 & 2.54 + 0503 - 286 & & 0.0380 & 2400 & 25.23 & 2.48 + 1452 - 517 & mrc 1452 - 517 & 0.08 & 1218 & 25.66 & 2.48 + 0114 - 476 & pks 0114 - 476 & 0.1460 & 702 & 26.51 & 2.36 + 1127 - 130 & pks 1127 - 130 & 0.6337 & 297 & 27.53 & 2.30 + 0707 - 359 & pks 0707 - 359 & 0.2182 & 492 & 26.71 & 2.21 + 1910 - 800 & & 0.3460 & 366 & 26.65 & 2.18 + 0745 + 560 & da240 & 0.0350 & 2164 & 25.39 & 2.07 + 0313 + 683 & wenss 0313 + 683 & 0.0902 & 894 & 25.64 & 2.01 +

degenerate geometrical configurations in quantum gravity are important to understand if the fate of classical singularities is to be revealed . however , not all degenerate configurations arise on an equal footing , and one must take into account dynamical aspects when interpreting results : while there are many degenerate spatial metrics , not all of them are approached along the dynamical evolution of general relativity or a candidate theory for quantum gravity . for loop quantum gravity , relevant properties and steps in an analysis are summarized and evaluated critically with the currently available information , also elucidating the role of degrees of freedom captured in the sector provided by loop quantum cosmology . this allows an outlook on how singularity removal might be analyzed in a general setting and also in the full theory . the general mechanism of loop quantum cosmology will be shown to be insensitive to recently observed unbounded behavior of inverse volume in the full theory . moreover , significant features of this unboundedness are not a consequence of inhomogeneities but of non - abelian effects which can also be included in homogeneous models . aei2005133 + gr qc/0508118 + degenerate configurations , singularities and + the non - abelian nature of loop quantum gravity + martin bojowald + max - planck - institut fr gravitationsphysik , albert - einstein - institut , + am mhlenberg 1 , d-14476 potsdam , germany

gr-qc0508118

by now , several discussions of the singularity issue have appeared in the context of loop quantum cosmology @xcite and loop inspired approaches @xcite to quantum cosmology . as an initial step , one often considers operators for inverse powers of the determinant of the metric such as @xmath0 as they are needed for matter hamiltonians and can be used for a first guess as to the fate of classical singularities in a quantization . in fact , many models lead to bounded operators for inverse volume @xcite despite of the classical divergence , following methods of full loop quantum gravity @xcite . this is the case generically , for instance , in isotropic cosmology @xcite where bounded inverse volume expressions play a central role in the non - singular evolution as it follows from a difference equation for the wave function . the issue is already different , however , in anisotropic models @xcite where the same quantization gives matter densities or curvatures which are not necessarily bounded when considered on full minisuperspace . in those cases , a non - singular evolution exists nonetheless , which demonstrates that , by itself , boundedness of kinematical operators such as inverse volume is not essential and that its role can only be seen when combined with dynamical information such as that given by the hamiltonian constraint . there are also other cautionary points one has to keep in mind when using inverse volume properties to understand the singularities . for instance , implicit in using inverse volume for statements about singularities is the classical relation between curvature divergence and inverse metric components . even if all inverse powers of the volume could be shown to be finite at the quantum level , one would still have to discuss their relation to space - time curvature , so again dynamical properties , which is more complicated . while classically , at least in isotropic cases , extrinsic curvature is proportional to an inverse power of the volume , this follows only after equations of motion are solved . thus , also at the quantum level one has to use equations of motion to relate properties of inverse volume expressions to the singularity issue . nevertheless , in order to obtain a more complete picture it is important to analyze the behavior of inverse volume in different models and also , as far as possible , in the full theory . recently @xcite , partial information from the full theory has been derived for gauge - invariant 3-vertices of spin network states which are always annihilated by the volume operator @xcite . here , inverse volume operators have , as functions of the edge spins , unbounded kinematical expectation values in 3-valent vertex states . the new situation , unlike any example used in isotropic , anisotropic or even some inhomogeneous models so far @xcite , is that unboundedness occurs already on states which have zero volume eigenvalue . ( which still is an improvement compared to the classical behavior which would predict diverging inverse volume . that this is not happening in the loop quantization is a direct consequence of the techniques introduced in @xcite . ) while it would be possible to construct operators with similar properties in models , in loop quantum cosmology they would break parity invariance of the theory . ( parity , i.e. @xmath1 for the densitized triad @xmath2 , has not been studied in the full calculations done so far since only configurations of vanishing determinant are considered . ) when a difference between symmetric models and the full theory is discovered , the first reflex is often to blame it , without further corroboration , on a breakdown of the minisuperspace approximation . so also in this case , e.g. inhomogeneities have quickly been named as a potential culprit . however , there are many technical and conceptual differences between models and the full theory , and so several reasons for discrepancies can be imagined . for a reasoned and informed judgment a deeper investigation is thus warranted . to illustrate this we mention atomic spectra as an example , where characteristic properties of the hydrogen atom obtained from a spherically symmetric potential remain intact but only become more complicated in details if realistic interactions are switched on . in particular , the atom remains stable unlike its classical counterpart . this is , of course , only true if interactions are restricted ( in this case from other theoretical investigations or observations ) and not allowed to be arbitrary . under arbitrary interactions , properties can certainly change dramatically , and indeed do so for atoms of high central charge where the electromagnetic interaction is stronger . similarly , in quantum gravity it is not sufficient to look at arbitrary degenerate geometrical configurations unless their meaning is known . there must be an analog of switching on the _ correct _ gravitational interaction , e.g. between inhomogeneities , which means that constraint equations or observables need to be considered . while in @xcite the main aim was to derive and point out potential differences ( also emphasizing , as oftentimes before in loop quantum cosmology , that dynamical information has to be included ) , here we analyze and contrast different geometrical configurations so as to facilitate developing a better idea for the relation between models and the full theory . we will then first review dynamical information about the singularity issue and discuss properties of inverse volume . new considerations start with sec . [ abel ] , which contains aspects of the regularization procedures of loop quantum gravity and differences to analogous procedures in an abelian truncation which play a role for geometrical operators but have not been recognized thus far . this indicates new sources of contributions to effective equations from non - abelian properties which can play a role for cosmological phenomenology . an analysis of inverse volume operators in different models and truncations to be discussed in sec . [ deg ] reveals a possible technical origin of bounded behavior which , however , is not directly related to physical information such as degrees of freedom . after comparing the behavior in models with the full theory and its abelian truncation in sec . [ comp ] , we discuss in sec . [ evol ] aspects of dynamics which in any situation ultimately has to be studied for a conclusion about singularity removal . we will find potential dynamical effects of those properties of the full theory studied in @xcite . the main problem posed by a singularity is the breakdown of dynamical equations at such a point . in any complete discussion of quantum removal of classical singularities one thus has to address three points : 1 . first , one has to isolate potential singularities , i.e. find conditions under which singularities must occur for the classical equations . one then has to determine the quantum evolution equations for the dynamics around those classically singular states . 3 . finally , this dynamics has to be shown not to break down when quantum behavior is taken into account . all three steps , even the first purely classical one , are too complicated to approach in full generality but can be dealt with in symmetric models which classically also show a wealth of different singularities . when comparing with the full theory , one then focuses so far on the issue of boundedness of quantities related to curvature , motivated by the fact that often curvature diverges at classical singularities . more importantly , such expressions , e.g. the inverse square root of the determinant of the metric , also enter matter hamiltonians and thus appear in parts of the dynamical equations . boundedness then is still not required but may simplify the analysis . moreover , only the second and third steps are avoided while the first one of isolating singularities is still present at full force . in this section , we will collect the currently available information about evolution and inverse volume before entering more detailed aspects of a comparison between symmetric and the full situations later on . the boundedness of inverse volume operators on superspace has proven not to be relevant for the general mechanism of singularity removal in models of loop quantum gravity as long as there is a well - defined expression at all . the isotropic case , where these developments have started , is however quite special which has led to some confusion and , occasionally , an overemphasis of certain aspects . there is only one parameter characterizing an isotropic spatial geometry and thus only one way to approach the classical singularity on minisuperspace . in this case , a well - defined behavior of inverse powers of volume would imply boundedness which indeed is realized automatically @xcite . the most crucial aspect of non - singular behavior , however , is a unique extension of evolution beyond the classical singularity which is _ not _ guaranteed even for bounded inverse volume ( see , e.g. , the discussion in @xcite ) . on the other hand , this extension of evolution has been generalized to non - isotropic cases ( making use of difference equations representing the hamiltonian constraint in a triad representation ) even when inverse powers of volume or curvature components are not bounded on the respective superspaces . the key reason for using models is that for them the singularity structure is often clear and the first task above can indeed be performed . other simplifications may then arise which facilitate a direct analysis of the evolution but are not necessarily crucial for the result . for the evolution one has to specify initial values and usually also boundary values for the wave function in the _ non - singular _ part of superspace , such that one obtains a well - posed initial value problem . in non - singular situations , these data uniquely give the solution to the constraint not only on one classically connected part of superspace , but also on other parts separated by classical singularities . in this sense , one can evolve beyond the classical singularity and the singular boundary is removed . note that , although the well - posedness of initial / boundary value problems for difference equations does play an important role , this is not simply an issue of counting the number of solutions through initial values and that the existence of an extension is non - trivial . singular evolution in this setting arises when values of the wave function at the singular boundary are not determined ( due to vanishing leading coefficients in a difference equation ) but would be needed for evolving further . one can , of course , simply include those values as data to be specified , but this does not solve the singularity problem . it would even be possible classically , e.g. in isotropic cosmology where one can specify the scale factor and matter fields at , say , @xmath3 as well as @xmath4 when the singularity is at coordinate time @xmath5 , and then solve the friedmann equation with both sets of initial values and simply glue the solutions together . in the classical as well as quantum case this would give a ( non - unique ) evolution beyond the classical singularity but would only remove a breakdown of evolution by putting in missing information by hand . for singularity removal , it is necessary to demonstrate a complete extension of wave functions given values in only one connected non - singular part of superspace . crucial ingredients are a quantum equation for the wave function on a suitable configuration space which requires an explicit construction from quantum gravity and involves analytical techniques . this equation is often called `` evolution equation '' even though it is not necessary to have a global time evolution picture . one then needs to identify classically singular parts of the configuration space and suitable transversal directions along which classical evolution breaks down . transversality means that the singularity is not simply approached asymptotically but is reached after a finite coordinate distance on configuration space . these directions would , in an evolution picture , be parameterized by internal time variables and thus a singularity will be reached after a finite amount of internal time has passed . it is then possible to ask and investigate what can happen in a more complete and non - singular theory after that time , or beyond the classical singularity . in this step one uses classical gravity and geometry . all ingredients then have to be combined in order to see if a wave function can be extended uniquely along the transversal directions , which would correspond to evolution through the classical singularity ( see also @xcite ) . one can conclude non - singular behavior only if the extension is possible for any state or , if one has information on the physical inner product , any state relevant for the physical hilbert space . in this way , the explicit construction of observables and a reconstruction of the corresponding geometry is by - passed , but nonetheless models can be concluded to be _ physically _ non - singular : if an extension for the wave function to a new region is known , one can extract information from it on both sides of a classical singularity . this is in particular true for loop quantum cosmology where kinematically both sides differ only by their orientation . if extracting an effective geometry at one side is well - defined , as is required for the correct semiclassical limit , one can do so equally at the other side for any parity invariant observable . what is not strictly necessary but often useful for an explicit analysis are the existence of a triad representation , a global time variable , or bounded inverse volume on superspace ( in general , it is enough that inverse volume remains well - defined along any transversal direction relevant for the quantum evolution ; see fig . [ fig ] ) . if a triad representation exists , one obtains a more intuitive picture through evolution by a difference equation . bounded inverse volume or curvature can be helpful since any state will be in their domain of definition . the action of the matter hamiltonian can then be computed in each iteration step of a recurrence scheme . if such an operator is not bounded and only densely defined , it could happen that one needs to act on a state arising during the evolution on which the operator is not defined . this could be a reason for a breakdown of evolution related to unbounded operators . with such a property the analysis would be more complicated , but it does not immediately imply singular behavior . the dynamical behavior would have to be analyzed in a detailed manner , and all initial values leading to such a breakdown must be identified . even if one would obtain a unique extension for generic initial values ( which is likely because operators in any case have dense domains ) , it would be impossible to claim singularity freedom unless one can show that the singular states are removed from the physical hilbert space . if , on the other hand , arbitrary initial values lead to a uniquely extended state , it does not matter for the singularity issue how exactly the physical hilbert space is obtained from the space of solutions to the constraint . having discussed the dynamical aspects , we briefly recall now which operators have been studied for quantizations of @xmath0 and their main properties . all these constructions follow the general scheme of classically expressing inverse metric components through poisson brackets between connection components @xmath6 and a positive power of volume based on @xcite @xmath7 ( @xmath8 is the barbero immirzi parameter @xcite and @xmath9 the gravitational constant ) . instead of the spatial metric @xmath10 , a densitized triad @xmath2 is used which is canonically conjugate to the connection @xcite . this relation is then quantized using holonomies of @xmath6 , the volume operator , and turning the poisson bracket into a commutator . the resulting operator is densely defined , and in some models it was found to be even bounded . a first explicit expression for a spectrum was computed in the isotropic context @xcite where , e.g. , @xmath11 on states @xmath12 with a single label @xmath13 whose geometrical meaning is given through the densitized triad component @xmath14 with eigenvalue @xmath15 ( @xmath16 is the planck length and @xmath17 a volume eigenvalue ) . the absolute value of @xmath18 thus determines the volume , while its sign is spatial orientation encoded geometrically in the handedness of the triad . the fact that the isotropic inverse scale factor is bounded has played a major role in further developments leading to a demonstration of singularity - free evolution in various models . already in anisotropic but still homogeneous models the situation is different since volume can become small even if some metric components are large . the single quantum number @xmath18 is replaced in diagonal homogeneous models @xcite by three labels @xmath19 , @xmath20 , which determine the three diagonal components of a densitized triad by @xmath21 , the volume by @xmath22 and the orientation by @xmath23 . volume can then become small not only when all @xmath19 go to zero but also when , say , two of them go to zero while the third diverges at a suitable rate . this can , and generically does , lead to unbounded behavior of inverse volume eigenvalues at smaller volume unless cancellations occur . for an inverse volume as above , or more generally a power of the volume parameterized by @xmath24 , one arrives at an expression of the form @xmath25 where @xmath26 is a function behaving as @xmath27 for @xmath28 but cutting off the classical divergence at small @xmath18 . this is obtained using , analogously to eq . ( 2.5 ) of @xcite and its generalization to @xmath29 , the classical relation @xmath30 where @xmath31 are homogeneous connection components conjugate to the triad components @xmath32 , @xmath33 . one quantizes the right hand side by expressing the connection components through `` holonomies , '' are not represented directly but only their exponentials @xmath34 in a manner not continuous in @xmath35 @xcite . ] e.g. @xmath36 with su(2 ) generators @xmath37 , using the volume operator and turning the poisson bracket into a commutator . this leads to functions @xmath38 since there are still positive powers of @xmath19 in ( [ vdiag ] ) , unboundedness could be possible when , e.g. , @xmath39 becomes large . while in this case we can use the large-@xmath18 behavior of @xmath26 and see that @xmath39 does not lead to unboundedness for an inverse power ( i.e. @xmath40 ) , other quantizations or different curvature components , where not a product as above occurs but instead a sum , lead to unbounded behavior at small volume . with the above quantization , however , inverse volume or curvature operators are always _ bounded , and in fact zero , on states of zero volume _ since @xmath41 . the latter property is not realized in the full theory where a non - zero and unbounded expectation value in 3-valent vertex states , equation ( 4.19 ) of @xcite , has recently been derived . the expression is again available for any positive value of @xmath42 if one only replaces the powers @xmath43 of that formula by @xmath44 . it can thus not only be used for inverse volume ( @xmath45 ) but also for other powers such as the volume itself ( @xmath46 ) . in all these cases the function of the three spins @xmath47 is _ unbounded on zero - volume eigenstates _ , showing clearly a difference to homogeneous models . this behavior on zero - volume states has been used in @xcite to argue that the singularity structure of full loop quantum gravity may not be captured reliably by models . what should make one suspicious , though , is the fact , unnoticed in @xcite , that not only the inverse volume ( @xmath45 ) is unbounded on zero - volume states but also the determinant of the metric ( @xmath46 ) , which can be taken as the volume of the vertex . thus , even though the volume eigenvalue is zero on any such state , there is a quantization of the volume which can be arbitrarily large . ) , which would give zero expectation values of the new operator in any zero - volume state @xcite . however , this property would equally apply to inverse powers which then also become zero and thus bounded . though maybe valuable , we will not consider this quantization choice further in this paper because it would remove the differences between models and the full theory related to boundedness of inverse volume on zero - volume states . ] this should not come as a big surprise since any such state must be considered deeply quantum and quantization ambiguities can have large effects . but it also shows that the geometrical , let alone dynamical , meaning of these configurations must be understood better before the results can be used for definitive conclusions . in particular the geometrical meaning of the spin labels around a vertex should be better known , similarly to the situation in homogeneous models where the parameters can directly be associated with quantized metric components . unlike in models , a suggestive identification of those states of zero volume , or other states containing those vertices , as `` big bang states '' is premature . one particularly important aspect of a geometrical configuration used for establishing non - singular evolution in models is parity . since loop quantum gravity is based on a triad formulation of gravity , the orientation of space ( not through the background manifold but through the basic variables of the theory ) is encoded in @xmath48 . the classical formulation is not much different from a metric formulation since the two orientations are not connected by physical trajectories but instead separated by singularities . this changes in loop quantum cosmology where the quantum evolution equation connects a wave function uniquely from one parity side to the other . for this , a region of reversed parity is required since otherwise it would not be possible to extend evolution . this is the main scheme for non - singular evolution , and its main technical ingredient is the parity degree of freedom and the structure of the evolution equation @xcite . it is clear from those considerations that it is not enough to consider only zero - volume states in isolation , i.e. static properties . the main part to a conclusion of singularity removal is a consideration of transitions where zero - volume states appear , which can only come from whatever evolution equation one is using . these are thus dynamical properties , which are crucial for any singularity statement . after providing more details about degenerate configurations , we will come back to dynamics in sec . since exact calculations in the full theory are usually quite involved , approximations other than symmetry reduction are necessary if a wider picture is to be developed . in the present context , it can be helpful to employ an abelianization by replacing the structure group su(2 ) by @xmath49 and using suitable definitions of the volume operator and other objects @xcite . many calculations then simplify , which has already been used to approach zero - volume states in a coherent state and to demonstrate that at least in this case expectation values of inverse volume can remain bounded along isotropic classical trajectories @xcite ( crucially relying on classical and _ symmetric _ properties such as how an isotropic singularity is approached ) . as we will show now , however , the boundedness issue for inverse volume operators is quite different in @xmath49 compared to su(2 ) . nevertheless , @xmath49 calculations are not only much simpler but also provide additional information which is of interest for comparison with symmetric models . in fact , many symmetric models turn out to be related to @xmath50-vertices . the labels @xmath19 of a diagonal homogeneous model , for instance , can be viewed as corresponding to a 6-valent @xmath49-vertex with edges given by straight lines entering and leaving the vertex and having charges @xmath51 , @xmath52 and @xmath53 , respectively . different holonomies in the model are still non - commuting and thus take su(2)-properties into account , but as far as geometrical expressions such as the volume are concerned the model can be formulated in an effectively abelianized manner , which is the main reason for simplifications . ( non - abelian properties , however , enter the dynamics as given by the hamiltonian constraint . ) looking at @xmath49-configurations thus provides means to distinguish between different aspects of a symmetry reduction and to see which of them are responsible for special behavior in models . however , since introducing @xmath49 also presents an approximation to the full theory , one must as well understand its range of validity and possible deviations to full su(2)-calculations . there is indeed one prominent place in our context where the non - abelian nature plays a role . in the rewriting procedure one makes use of relations of the form @xmath54 where @xmath55 is a holonomy along an edge @xmath56 of parameter length @xmath57 and @xmath58 the volume of a region ( @xmath59 are generators of the structure group ) . the left bracket can thus be expressed by holonomies up to terms of order @xmath60 which vanish in a continuum limit . alternatively , at fixed @xmath57 the right hand side can be seen as an approximation or regularization of the left hand side . irrespective of the interpretation of @xmath57-terms , the quantization does not depend on this parameter . the poisson bracket gives then rise to a commutator in a vertex contribution to an operator , and the holonomy can be elongated or shortened at the end away from the vertex without changing the action on states . as mentioned before in homogeneous models ( see sec . [ invvol ] ) , the application of identities such as ( [ commut ] ) is necessary because the hilbert space obtained through a background independent quantization does not allow the action of individual connection components but only of holonomies . this is a characteristic property of the quantization and is present irrespective of details of a regularization or its interpretation . it is one possible origin of correction terms in , e.g. , an effective hamiltonian compared to the classical expression just as higher curvature terms often arise in effective actions @xcite . for the arguments to follow it is important to notice that there are two different kinds of corrections on the right hand side of ( [ commut ] ) , which usually are both subsumed in the symbol @xmath61 : first , one has to approximate the connection component in a single point by a suitable edge integral and , secondly , that integral is exponentiated . if the edge length is small , both modifications imply only small changes of the indicated order . what is different in these two types of corrections is their dependence on the phase space variables . the left hand side of the equation , even though it contains connection components in the poisson bracket , depends only on triad components after the bracket is computed . similarly , if the connection component is replaced by an integral which is still linear in @xmath6 , the bracket is just evaluated in different points along the edge and still depends only on triad components . when the integral is exponentiated , however , the situation changes since in a non - abelian theory the cancellation between @xmath55 outside the bracket and @xmath62 inside is not complete : we have , schematically , @xmath63 where @xmath64 denote the holonomies running from the starting point of the edge to @xmath65 and from @xmath65 to the end point , respectively . only the contribution from @xmath66 cancels in ( [ commut ] ) , but we are left with @xmath67 . again , this gives only small correction terms , but _ they do depend on connection components _ , not just on the triad . in an abelianization , on the other hand , the second type of correction terms does not appear because now the cancellation between @xmath55 and @xmath62 is complete . there are still @xmath61-corrections of the first kind , but they are independent of connection components . thus , in the abelian case a commutator as vertex contribution is a more direct quantization of the classical expression on the left hand side of ( [ commut ] ) which is not the case in a non - abelian framework where regularization has stronger effects . a further consequence is that the commutators obtained in a non - abelian setting do not commute with the volume operator even though they are supposed to quantize objects which classically do not depend on connection components . again , this can be seen as resulting from correction terms as above , which do depend on the connection . indeed , in abelian quantizations such commutators commute with the volume operator , even though holonomies appear in intermediate stages . the two different correction terms also have different implications for homogeneous models . the first one , originating in replacing connection components by integrals , is clearly a consequence of a field theory . it does not arise in homogeneous models where we can simply take exponentials such as @xmath68 as used before . thus , if the first correction would be crucial for the properties of @xcite , it would clearly be an indication that homogeneous models do not reliably capture the full situation . the second type of correction term , on the other hand , is a consequence of non - abelian behavior . it disappears in an abelianization irrespective of whether it is introduced as a truncation to the full theory or effectively a consequence of a particular symmetry reduction . moreover , it even occurs in non - diagonal homogeneous models where not all derivatives with respect to connection components of , e.g. , holonomies of the form @xmath69 ( as explained in more detail in sec . [ models ] ) are proportional to the same internal direction @xmath70 . these properties , which we will discuss further in what follows , thus allow one to _ distinguish between non - abelian effects and effects of inhomogeneities_. the above remarks on regularization are by no means saying that the quantization would be wrong or suspect . the classical limit is not affected since the relation is valid in regions of phase space where semiclassical behavior is expected . in this limit , one obtains similar behavior of abelian and non - abelian calculations ( which in some cases can also be demonstrated explicitly @xcite ) . but when regimes close to potential singularities are studied , one should pay utmost attention to such correction terms . in fact , the rewriting procedure introduces such corrections which are expected to be strong close to singularities and can significantly influence even qualitative behavior . in addition , quantization ambiguities appear since rewriting is possible in different ways . one example for such quantization ambiguities in action is the already observed discrepancy , based on @xcite , between unbounded behavior of a full quantization of @xmath71 in a vertex even though the basic volume operator annihilates that state . since both operators correspond to the same classical expression , this different behavior is caused by a quantization ambiguity which can be traced back to the relation ( [ commut ] ) discussed above . in this case , it is clear that the volume operator should be considered more basic since it can be quantized directly using only fluxes @xcite , and as such it is indeed essentially unique @xcite . then again , the fact that it annihilates all 3-vertices is very special , resting solely on gauge invariance , and so considering only 3-vertices is potentially misleading . such properties of different operators quantizing volume show that a geometrical interpretation of vertex configurations can be complicated . unless other information is available , it is safest to use only regimes where the volume operator is close to other quantizations of @xmath72 , which in the case of 3-valent vertices means small spins . as for inverse volume operators , they are clearly bounded on any bounded range of spins even though the volume eigenvalue vanishes . moreover , on those states , the quantized @xmath0 is smaller than @xmath71 even though this corresponds to states of small volume . there is thus a cut - off of classical divergences on degenerate vertices of the full theory . in light of the two types of correction terms , the difference of the two volume operators on 3-vertices must be a consequence of the second type of corrections since it does not occur , for the same valence of a vertex , in an abelian truncation ( see the following subsection ) . it is thus not a consequence of inhomogeneities . it is , rather , a consequence of additional correction terms which only appear in a non - abelian setting and can lead to more distinguishing features between otherwise similar operators . the same conclusions then apply to the behavior noticed in @xcite where commutators such as inverse volume or the alternative volume are unbounded on states which are annihilated by the basic volume . it is then crucial to understand which operator among different choices is relevant for the identification of singular states . fortunately , the presence of quantization ambiguities does not preclude definitive conclusions . in the present case , where different quantizations of volume lead to different degenerate states , one can use dynamical input as additional information . after all , the singularity issue is a dynamical problem , and isolating classical geometries or quantum states corresponding to a singularity has to be done with knowledge of the evolution equations . in a canonical quantization , the evolution equations are given by the hamiltonian constraint operators . they do indeed contain the volume operator , but only in combination with holonomies as a commutator @xcite . where explicit evolution equations have been obtained @xcite , their coefficients are determined by matrix elements of these commutators . it is the potential vanishing of these coefficients which signals the possibility of a singularity in the quantum evolution , and knowing these coefficients is a prerequisite for understanding the removal of singularities at the quantum level . the conclusion is that , from a dynamical point of view , one should not look at states annihilated by the basic volume operator , but at states annihilated by commutators between the volume operator and holonomies as they appear in the constraints . these commutators also make use of the relation ( [ commut ] ) and are thus much closer to expressions used for inverse volume . in particular , all such commutators , whether they quantize positive or inverse powers of volume or other metric components , have generically the same kernel , smaller than that of the basic volume operator , since they differ only in containing different powers of the volume operator . issues of unboundedness on degenerate states , as in @xcite , then do not arise . the situation is simpler in models studied so far since they have volume eigenstates which are identical to eigenstates of commutators of volume and holonomies ( see also sec . [ models ] ) . to shed more light on the relation between the full non - abelian theory and those models , we study different degenerate configurations in @xmath49 and in particular inverse volume operators . since , e.g. , diagonal homogeneous models can be viewed as special cases of 6-valent @xmath50-vertices , one can directly see where potential differences come from . in su(2 ) one typically has objects such as @xmath73 $ ] where holonomies in the commutator when acting on an edge of spin @xmath74 lead to contributions with higher spin @xmath75 and lower spin @xmath76 . in a symmetric setting such as loop quantum cosmology , whose composite operators are modeled on su(2 ) expressions of the full theory , one can take the trace explicitly and obtain products of the form @xmath77 where @xmath78 is a connection component appearing in holonomies @xmath79 @xcite . since , e.g. , @xmath80 with @xmath81 , the resulting operators are close to the behavior in the non - abelian setting because @xmath82 as a multiplication operator increases the label @xmath18 of a state @xmath83 while @xmath84 decreases it . nevertheless , since one can simply interpret @xmath81 as an abelian holonomy , one can also view this expression as obtained in an abelian setting . however , in a directly abelianized model one would rather use an expression such as @xmath85 $ ] , simply replacing the non - abelian holonomies by abelian ones @xcite . the action is then quite different because an abelian holonomy either increases the charge or decreases it , but does not give two such contributions at the same time as happens with su(2 ) . the more symmetric treatment which does model some non - abelian behavior is precisely what leads to an inverse volume vanishing on zero - volume states in isotropic loop quantum cosmology as can easily be seen from the resulting eigenvalues . in an isotropic model , the purely abelian version @xcite results in eigenvalues of the form @xmath86 with @xmath87 while the symmetric form of loop quantum cosmology gives @xmath88 as in ( [ iso ] ) . at zero volume , @xmath89 , the first result is non - zero unlike the latter one . thus , while in loop quantum cosmology the inverse volume is automatically zero on zero - volume states , not by hand but by following constructions of the full theory , the loop inspired treatment of @xcite and also the abelianized version of @xcite can give non - zero results . @xcite also considers vertices of higher valence where even the symmetric treatment can give non - zero results on zero volume , as discussed below . however , those higher valent configurations are not contained in a homogeneous setting . ) this demonstrates the importance of following full expressions as closely as possible in models , as advocated in loop quantum cosmology . while basic operators can directly be derived from the full theory @xcite , which is the crucial difference between loop quantum cosmology and an ordinary minisuperspace quantization , this is more complicated ( and so far unfinished ) for composite operators such that for them guidance from the full theory must be employed in their constructions . we can now discuss the issue of zero - volume states in the @xmath50 setting , still using our 6-vertex but now with unrestricted charges collected in a @xmath90 matrix @xmath91 . the volume of such a configuration is given by @xmath92 , such that zero - volume configurations span an 8-parameter subset of the 9-dimensional configuration space . following loop quantum cosmology , powers of metric components expressed through commutators will take eigenvalues of the form @xmath93 where only one coefficient @xmath94 changes in the volume labels . for any matrix @xmath91 , not just diagonal ones , this can easily be seen to be zero using multi - linearity of the determinant since the volume eigenvalues depend only on the absolute value of the determinant of @xmath91 . thus , a whole row of components @xmath95 vanishes in those cases and so do all volume or inverse volume operators @xmath96 constructed from them . the main difference to diagonal models is that cancellations as in ( [ vdiag ] ) do not happen ; but this is not problematic since they have not been made use of in loop quantum cosmology , anyway . the advantage of using a 6-vertex of the above form is that the geometrical meaning of its labels is clear from the identification of @xmath91 with eigenvalues of densitized triad components @xmath97 through flux operators . as we have demonstrated , boundedness can be a direct consequence of effectively abelian behavior , but it is not directly related to symmetry in the form of diagonalization or isotropy which only leads to more special matrices @xmath91 . the situation for a 6-vertex in su(2 ) of a similar type , which could still be interpreted as corresponding to a homogeneous model , is unfortunately more complicated and so far unknown . one can generalize the abelian setting by introducing more edges ( and thus more degrees of freedom , albeit still finitely many ones ) . volume eigenvalues are then given not by a single determinant but by a sum of determinants of different matrices ( for each non - planar triple of edges ) . in many cases one can reduce this to a single determinant using multi - linearity and gauge invariance , but in general this is not possible and there are explicit examples with unboundedness @xcite . however , it is not clear what this means since also the geometrical interpretation of labels would be lost : we would have more independent labels than the nine values contained locally in a classical triad . moreover , such an abelian situation can not come from a geometrically motivated symmetry reduction where any vertex which is more than 6-valent ( or 6- or less - valent but without straight edges ) could not be embedded in a @xmath49-vertex . in diagonal homogeneous models . this is no longer possible if there are new independent connection components from additional edges . ] a different generalization brings us to inhomogeneities : so far we assumed edge labels to be identical for opposite edges of the 6-vertex , which can be interpreted as diagonal homogeneous models . when this condition is dropped , one leaves homogeneity but the configurations can still be interpreted as vertices of inhomogeneous models . ( indeed , also in the su(2)-setting the 3-vertices of @xcite can be viewed as extreme forms of such inhomogeneous 6-vertices with three vanishing labels . ) for instance , if we have an abelian vertex with only one such inhomogeneous edge with opposite labels @xmath98 and of diagonal form , it can be viewed as a general vertex of a polarized cylindrical wave model @xcite . volume eigenvalues are then of the form @xmath99 . the behavior of inverse volume can easily be seen to be unchanged since we simply replace a label by a sum of two labels . also in non - diagonal cases ( which include the configurations used in @xcite without any symmetry assumptions ) , the previous conclusions about boundedness do not change . thus , even inhomogeneity does not lead directly to unbounded behavior . from the preceding expressions and discussion it is clear that a direct technical reason for vanishing inverse volume operators on zero volume eigenstates lies in volume eigenvalues which depend on the edge labels only through the absolute value of a multi - linear function . examples are the multi - linear functions @xmath18 in the isotropic case or @xmath100 for a @xmath49 6-vertex . this is , however , not a necessary condition as demonstrated by the isotropic model where alternative volume eigenvalues @xmath101 , somewhat closer to su(2 ) expressions , have been used @xcite which also give zero inverse volume eigenvalues . nevertheless , the appearance of multi - linear functions is quite common , as they can occur e.g. in isotropic , diagonal homogeneous and some inhomogeneous models . it is , however , difficult to find a geometrical reason for unboundedness since many different possibilities are realized . linking unbounded behavior to , say , inhomogeneities is impossible because there are ( non - diagonal ) homogeneous models whose volume eigenvalues do not depend on edge labels through a multi - linear function @xcite , but also inhomogeneous models which do have such a dependence @xcite . there are different ways to break the multi - linearity by considering configurations with additional parameters . the simplest way is by adding more edges and their labels as discussed in the abelian setting . sometimes there are also alternative volume operators in models , e.g. @xcite , which lead to a different dependence on labels and thus can possibly change the qualitative behavior of eigenvalues . this situation is similar to the full theory in that the volume operator does not commute with its commutator with holonomies . finally , there are models or the full theory where non - abelian effects play a role . here , new parameters are in fact provided by using general non - abelian holonomies while in symmetric models often only the killing norm of su(2 ) elements is relevant . additional correction terms are automatically switched on when non - abelian degrees of freedom are added , as we already saw in sec . [ abel ] with additional connection dependent terms in identities such as ( [ commut ] ) . additional parameters appear only if one takes into account more degrees of freedom than realized in a given symmetric model . however , the physical role of these additional degrees of freedom is not always clear since the geometrically and physically relevant ones have already been picked out by the symmetry reduction . with additional degrees of freedom , one thus has always a new physical situation which can not immediately be compared to the original , e.g. symmetric one . it is important to keep in mind that , in contrast to a minisuperspace quantization , the selection of degrees of freedom in loop quantum cosmology is not done solely by hand . common to a minisuperspace reduction is the choice of a symmetry to be imposed , which specifies the physical situation of interest and the corresponding reduced classical phase space . symmetric states , in the connection representation , are then defined as distributions in the full theory whose support contains all relevant invariant connections as a dense subset @xcite . since the support of a distribution is by definition a closed set , it is not automatic that symmetric states are not supported on some connections which do not appear as invariant ones at the classical level . at this point , crucial information from the full configuration space of generalized connections enters the basic construction of symmetric models . as the analysis reveals , the classical set of invariant connections is indeed extended to a bigger ( compactified ) space of generalized invariant connections , which can also be introduced at the minisuperspace level @xcite . however , one can prove @xcite that this space of generalized invariant connections is a closed subset of the full space of generalized connections and indeed the support of symmetric distributions . thus , reducing at the kinematical quantum level does not add new degrees of freedom , which shows that kinematically the symmetry reduction followed in loop quantum cosmology is consistent within the full theory . this justifies the treatment along the lines of a minisuperspace quantization , an issue which could never have been addressed without a detailed relation between models and the full theory . more complicated , and still open , is the question of how the dynamics of models and the full theory are related . in addition to geometrical , static properties of degenerate configurations one also needs to know their dynamical role as recalled in sec . [ general ] . this is achieved in models by using difference equations in internal time representing the hamiltonian constraint in the triad representation , which show how degenerate configurations are approached during physical evolution . the existence of such difference equations which are globally defined , i.e. on the full mini- or midisuperspace of the model , relies on the removal of some non - abelian effects . general non - abelian behavior makes fluxes , representing triad components , non - commutative as a consequence of ( [ commut ] ) . in this case , correction terms lead to non - commuting operators even when they correspond to metric components smeared along the same surface . ] such that not all of them can be sharp at the same time and no triad representation exists @xcite . nevertheless , local versions of the difference equation , as indeed used in inhomogeneous models @xcite , can still be possible . in models , as discussed before , one can then see that unboundedness , even if it occurs such as in anisotropic models , is no obstruction to non - singular evolution . originally , a non - symmetric ordering had been used which ( together with its associated dynamical initial conditions @xcite at least in their most straightforward incarnation ) would not work with non - zero matter densities at zero volume . such an ordering , however , is already ruled out in inhomogeneous models @xcite even in the vacuum case . with the symmetric ordering currently preferred it is not necessary for non - singular behavior that inverse volume expressions are zero or bounded on zero - volume states @xcite . we can now ask what dynamical effects one can expect from properties of inverse volume in the full theory . so far the observed unboundedness of inverse volume on tri - valent vertices of the full theory has no direct implications for the singularity removal mechanism of loop quantum cosmology . those states can be considered zero - volume states ( even though also this depends on quantization ambiguities , see sec . [ abel ] ) but it is not clear how they occur in a physical transition . for this , one would also have to consider states of non - zero volume which in some sense are close to tri - valent vertices . this could be conceivable to do in a 4-valent setting , or on a 6-vertex using @xcite , where one would have to determine the volume spectrum on the intertwiner space . parity reversal , which is important in models , will then be controlled not by spins of outer edges but by intertwiner labels . changing from positive to negative eigenvalues of @xmath102 would not affect outer edge spins and thus not result in an intermediate 3-vertex . this could only arise after using the full hamiltonian constraint , which does change edge spins . after finding suitable such transitions , one has to see if they are required by the dynamical evolution and whether or not evolution breaks down . this could possibly be done locally by a difference equation with an intertwiner spin as clock variable even though due to the non - abelian nature no global difference equation for the constraint on superspace exists . a mechanism alternative to using dynamical information from difference equations , which would be much more complicated to obtain in the full theory , was suggested by using coherent states and following their behavior when they are peaked on geometries of ever smaller volume @xcite . one can then follow the expectation value of matter energy , which indeed in a @xmath49 calculation has been shown to remain bounded provided that fluxes are bounded along the classical evolution ( as realized , e.g. , for vacuum bianchi class a models even when approaching the classical singularity ) . this coincides with the expectation from anisotropic models . but it also shows the potential dangers of full considerations with our current limited knowledge : somewhat hidden in the calculations of @xcite is the _ assumption that fluxes remain bounded _ during the approach to the singularity . if this is not the case , there is no boundedness result from the @xmath49 calculation . relevant for boundedness is not the fact that coherent states are used in contrast to volume eigenstates in the su(2 ) calculation , but the fact that _ dynamical _ information is put in as a key additional assumption . the danger is that this assumption is based on classical dynamical properties which moreover are known only in symmetric situations . thus , although the calculations of @xcite are done in the full setting , crucial for boundedness results are properties of symmetric solutions . it would be a large step forward in our understanding of singularities if the boundedness of densitized triad components close to classical singularities could be shown more generally . this is also of importance for the mechanism of loop quantum cosmology , as emphasized in @xcite . a similar calculation for su(2 ) would be much more involved , but to some degree one can rely on arguments that expectation values in @xmath103 coherent states are close to what one gets with su(2 ) . nevertheless , potential differences can occur particularly when unbounded behavior is to be studied . in @xmath49 , the expectation value is independent of the connection where the coherent state is peaked since it appears only in a phase factor , but this is not the case in a non - abelian setting . moreover , as discussed above , in a non - abelian calculation there are correction terms to the basic identity which are small for small connection components . in contrast to an abelian calculation we thus expect an expectation value of inverse volume operators in a coherent state which is also connection dependent , and moreover correction terms which in general are small only for small connections entering the coherent state ( a coherent state requires a gauge choice to specify the peak position , thus making it meaningful to talk about small connections ) . accordingly , the abelian calculation should be reliable in semiclassical regimes , but not necessarily for strong curvature . while one can give arguments for the boundedness calculation of @xcite and as discussed above , this only refers to boundedness of the expectation value as a function of spin labels _ provided that all fluxes remain bounded_. these arguments can thus be used to illustrate the expected effective semiclassical behavior , but say nothing about boundedness of operators and the more complicated issue of extending evolution beyond a classical singularity . ] of expectation values of inverse volume even in su(2 ) coherent states @xcite , based on the fact that holonomies are bounded functions of the connection , correction terms are important for more detailed aspects and in particular cosmological phenomenology . since classically connection components , entering non - abelian correction terms , become arbitrarily large when a singularity is approached , the @xmath49 truncation of the full theory has to be supported by additional arguments . similarly , potential correction terms to symmetric models and their dynamical role are still to be studied . a general extension of wave functions beyond classical singularities has been found to be realized in all models so far where singularities have been investigated @xcite . we now give some details in the isotropic case which also show how bringing in additional parameters of su(2 ) elements could potentially change the picture . similarly to inverse volume operators , the hamiltonian constraint @xcite is constructed from holonomies and the volume operator which in the flat case gives @xcite @xmath104)\,.\ ] ] holonomies are of the form @xmath105 with the isotropic connection component @xmath78 and an so(3 ) matrix @xmath106 specifying internal directions which are pure gauge . using @xmath107 with @xmath108 , the two contributions are @xmath109 and @xmath110&= & \hat{v}-\cos { \textstyle\frac{c}{2}}\hat{v}\sin { \textstyle\frac{c}{2}}- \sin { \textstyle\frac{c}{2}}\hat{v}\cos { \textstyle\frac{c}{2}}\nonumber\\ & & - 2\lambda_k ( \sin { \textstyle\frac{c}{2}}\hat{v}\cos { \textstyle\frac{c}{2}}-\cos { \textstyle\frac{c}{2}}\hat{v}\sin { \textstyle\frac{c}{2}})\nonumber\\ & & - 2\sin { \textstyle\frac{c}{2}}\cos { \textstyle\frac{c}{2 } } [ \lambda_k,\hat{v}]- 4\lambda_k\sin^2{\textstyle\frac{c}{2}}[\lambda_k,\hat{v}]\,.\end{aligned}\ ] ] in the isotropic model , @xmath106 commutes with the volume operator and its columns are orthogonal to each other . the constraint then reduces to @xmath111 ( or its symmetrization ) which has been used for the proof of non - singular behavior @xcite . here , one uses the action @xmath112 and derives a difference equation in the triad representation for a wave function @xmath113 . we can now bring in new parameters by allowing arbitrary @xmath106 except that their norm must still be one ( since otherwise @xmath78 would be redundant ) . this allows six rather than three angles in specifying the internal directions , and three of them remain after removing gauge freedom . analyzing the canonical structure of the new model and the meaning of the additional parameters ( which can be related to shape parameters following @xcite ) is beyond the scope of this paper , but we can already see potential extra terms in the constraint . the following statements can be interpreted in the context where these additional parameters are switched on perturbatively such that the basic representation of an isotropic model remains unchanged except that states now are also functions of the new angles . there are two conditions of the isotropic model which are no longer satisfied : ( i ) @xmath114 for @xmath115 and ( ii ) @xmath116=0 $ ] . dropping the first condition implies that also the second term in ( [ hols ] ) contributes to the constraint which changes the difference equation . however , there is no crucial change in structure since the equation remains of the same order ( the number of trigonometric functions in each term is the same ) . leading coefficients of the difference equation could a priori be vanishing in different places with the new term , but this does not happen since the new contribution is imaginary ( the additional sine implies an additional factor of @xmath117 in a difference operator from ( [ sin ] ) ) . the only change is thus in coefficients becoming complex . dropping the second condition is more important since it leads to a higher order of difference equations . ( note that we always have @xmath118=\lambda_k\cdot[\lambda_k,\hat{v}]+ [ \lambda_k,\hat{v}]\cdot\lambda_k$ ] using the fact that the @xmath119 are normalized . but an individual commutator @xmath116 $ ] can still be non - zero . ) moreover , the last two terms in ( [ comm ] ) , unlike the first two lines , do not commute with the volume operator . this is different from other homogeneous models where inverse volume operators commute with the volume operator , but it would be analogous to behavior in the full theory as noticed in sec . indeed , the extra terms in ( [ comm ] ) can also contribute to non - zero expectation values of inverse volume operators in zero - volume eigenstates . if we use @xmath120)$ ] for an inverse power , we will have @xmath121 in the isotropic zero volume eigenstate @xmath122 since @xmath123=0 $ ] and @xmath124 . but for @xmath115 we can have @xmath125)|0\rangle\not=0 $ ] from the last term in ( [ comm ] ) which also corresponds to some inverse power of volume . the behavior is thus closer to that in the full theory and provides an explicit example for the role of non - abelian effects in unboundedness as discussed in general in sec . still , features of the isotropic model are recognizable . in particular , there is a difference equation which , compared to the isotropic one , is of higher order and has different coefficients . deriving implications for the singularity issue requires an interpretation of new degrees of freedom in @xmath126 in relation to classical metric components as well as an understanding of the classical singularity structure . this requires a more detailed analysis of the canonical and geometrical structure , but the above considerations already show that kinematical properties such as those discussed in @xcite can be studied in models without jumping directly to the full theory . discussing the singularity problem is made complicated by the fact that there are several different issues to be kept in mind . we have presented several technical points in this paper , so far mainly in separation from each other . before concluding we now collect these observations . information used for the singularity issue includes * kinematical properties : eigenvalues of inverse volume or intrinsic curvature operators , or their expectation values in zero volume states or kinematical coherent states , * dynamical properties : the form of difference equations derived from the constraint , effective equations or classical dynamical properties in coherent states . as inverse volume may or may not be bounded , its relevance can be seen only in conjunction with dynamics , e.g. from how it appears in an evolution equation . kinematically , it does not matter which kinds of states are used to compute expectation values of inverse volume , and also in coherent states does one generally still have unboundedness . the advantage of using coherent states , however , is that it is easier to bring in dynamical information through the classical peak position . one can then show that for bounded fluxes also inverse volume is bounded , requiring an additional assumption motivated by classical properties of symmetric solutions . so also here symmetric properties are being used , even though the intention is to go beyond symmetry . ( strictly speaking , the argument in @xcite does not only use available classical information because it assumes boundedness of _ all _ fluxes around a given vertex . this still has to be related to classical behavior , in particular considering the fact that more independent labels are required for an @xmath49 vertex with unbounded inverse volume than one has locally in a classical triad . ) this illustrates the current state of understanding of singularities not just from the direction of quantum gravity but even classically . precise statements do not exist without assuming at least some degree of symmetry . one can try to find perturbative behavior around a symmetric solution , but in particular for the singularity issue this can be very misleading since it is not clear how perturbations would grow . ( also from a technical perspective , this has been explicitly observed in @xcite . ) difference equations provide alternative dynamical input where the situation can be decided fully at the quantum level . classical information is then required only to identify classical singularities . so far , however , this procedure is restricted to symmetric situations . as already mentioned in the beginning , isotropy is very special with minisuperspace having the same dimension as trajectories . there is then only one way to approach a classical singularity , and inverse volume expressions are completely bounded . anisotropic models are more general and display almost all of the explicitly known characteristic behavior of classical singularities . here , curvature is in general unbounded on minisuperspace . there are some inhomogeneous models which are not necessarily more complicated as far as unboundedness is concerned , but dynamical properties are more difficult to disentangle . in other models or in the full theory , even the boundedness issue can be problematic as a consequence of non - abelian behavior . the latter may be a potential source for differences between models and the full theory , but its relevance requires a better understanding of non - abelian degrees of freedom . this is of importance for the role of additional ambiguities and for finding the relation between degenerate and singular states . from this point we come to the final issue : non - abelian properties clearly have to be taken into account in the full theory , but also in several models such as non - diagonal homogeneous models @xcite or unpolarized cylindrical waves @xcite . inhomogeneities certainly appear in the full theory but also in models such as spherical symmetry or cylindrical waves . here , one should remember that the role of all the degrees of freedom provided by loop quantum gravity or inhomogeneous models is not yet clear ; in particular , the relation between mathematical degrees of freedom and physical inhomogeneities , i.e. fields on a given homogeneous background , is not fully understood . this is relevant especially when vertices of arbitrarily high valence are considered in the full theory which provides a potentially infinite number of degrees of freedom but of undetermined geometrical role . this issue is the most difficult to analyze because in general there is no clear separation between non - abelian effects , inhomogeneities and the valence of vertices . high - valent vertices can occur only in the full setting or in models whose symmetry orbits have at most dimension one . however , there are abelian as well as non - abelian homogeneous models , and also both kinds of inhomogeneous models . to avoid having to deal with arbitrarily high valent vertices we can restrict ourselves to three - valent ones , or six - valent if there are three straight edges entering and leaving a vertex , since this is anyway the only calculation available from the full point of view so far . it also represents the combinatorial situation of models where singularities have been studied so far . then , a distinction between non - abelian effects and inhomogeneities becomes possible as discussed in sec . [ abel ] : while the non - abelian volume operator does not commute with inverse volume , the latter then having unbounded expectation values in states annihilated by the former , volume and inverse volume do commute in the abelian case and inverse volume vanishes on zero volume states . here , inhomogeneity does not play any role whatsoever and thus can not be responsible for unboundedness . in an abelian calculation , one obtains unboundedness only at vertices of high valence for which the geometrical role of their labels is unknown . we emphasize again that inverse volume is considered here only because it is the best understood operator in this context . otherwise , unboundedness properties of curvature already arise when introducing anisotropy , not even inhomogeneities . we have demonstrated that , from currently available information , there is no contradiction between models and the full theory of loop quantum gravity . instead , a consistent picture emerges when all cases , isotropic as well as anisotropic or inhomogeneous models and the full theory , are considered . there is a technical difference between inverse volume in the full theory and similar operators in models , in that the expectation values in the full theory are unbounded on states of vanishing volume eigenvalue , while they are zero on such states in models considered so far . however , in the full theory the distinction of singular states , in particular their identification with certain degenerate states , is blurred due to non - abelian effects . there is then no crucial difference between a model where curvature is unbounded _ close to singular states _ ( i.e. for quantum labels close to those of a singular state ) and the full theory where inverse volume is unbounded _ on degenerate states _ which can be argued to be close to , but not necessarily identical with singular states . any claim of a contradiction between models and the full theory , based on these properties , is thus unsubstantiated . nevertheless , there are certainly differences by design since models capture behavior only in a particular , geometrically selected sector of physical interest . what is clear from the considerations is that so far inhomogeneities can not be made responsible for any explicit discrepancy between models and the full theory . to judge what implications particular properties of inverse volume operators have , their geometrical and dynamical roles must be clear before rushing to conclusions . the latter can be derived , e.g. , through difference equations representing the hamiltonian constraint or observables . but also from the classical side one needs to provide knowledge on the general singularity structure , which becomes exceedingly complicated when symmetry assumptions are dropped ( see , e.g. , @xcite ) . the main issue to be checked in further investigations is implications of non - abelian effects not studied so far in the different sectors of loop quantum gravity ( including not only cosmological situations but also black hole horizons and other models or semiclassical states , all of which often exploit possible eliminations of non - abelian terms in explicit calculations ) . in models , understanding the meaning of degenerate configurations is achieved by the selection of an appropriate sector of the theory displaying the configurations of interest explicitly . we emphasize again that at this point crucial information from the full theory enters through distributional states . the symmetry is specified for the physical context , and then the relevant quantum degrees of freedom result through derivation . ( similarly , imposing black hole horizons selects the appropriate degrees of freedom , again in an effectively abelian manner , through physical conditions for an isolated horizon @xcite relating horizon degrees of freedom to full flux operators . ) this suggests that non - abelian degrees of freedom are not always crucial physically , which can also be seen from the fact that diagonal homogeneous models , which in contrast to non - diagonal ones can effectively be abelianized , show the complete behavior of cosmological evolution @xcite . still , additional correction terms in a non - abelian situation can provide characteristic effects . better understanding non - abelian behavior is thus not just important for the relation between models and the full theory , i.e. as a test of approximations , but can also provide new physical insights . the non - commutative behavior of quantum geometry @xcite so far has not been made use of in cosmological investigations in loop quantum gravity , while non - commutative geometry itself has given rise to several cosmological applications ( see , e.g. , @xcite ) . combining loop cosmological phenomenology with non - commutative behavior thus has the potential of providing further scenarios for the very early universe . the author is grateful to abhay ashtekar , johannes brunnemann , ghanashyam date and thomas thiemann for discussions . m. bojowald and h. a. morales - tcotl , cosmological applications of loop quantum gravity , in _ proceedings of the fifth mexican school ( dgfm ) : the early universe and observational cosmology _ , _ lect . notes phys . _ 646 ( 2004 ) 421462 ( springer - verlag ) , [ gr - qc/0306008 ] m. bojowald , loop quantum cosmology : recent progress , in : proceedings of the international conference on gravitation and cosmology ( icgc 2004 ) , cochin , india , _ pramana _ 63 ( 2004 ) 765776 , [ gr - qc/0402053 ]

as the title says we want to answer the question ; how and why does statistical mechanics work ? as we know from the most used prescription of gibbs we calculate the phase space averages of dynamical quantities and we find that these phase averages agree very well with experiments . clearly actual experiments are not done on a hypothetical ensemble they are done on the actual system in the laboratory and these experiments take a finite amount of time . thus it is usually argued that actual measurements are time averages and they are equal to phase averages due to ergodicity . aim of the present review is to show that ergodicity is not relevant for equilibrium statistical mechanics ( with tolman and landau ) . we will see that the solution of the problem is in the very peculiar nature of the macroscopic observables and with the very large number of the degrees of freedom involved in macroscopic systems as first pointed out by khinchin . similar arguments are used by landau based upon the approximate property of `` statistical independence '' . we review these ideas in detail and in some cases present a critique . we review the role of chaos ( classical and quantum ) where it is important and where it is not important . we criticise the ideas of e. t. jaynes who says that the ergodic problem is conceptual one and is related to the very concept of ensemble itself which is a by - product of frequency theory of probability , and the ergodic problem becomes irrelevant when the probabilities of various micro - states are interpreted with laplace - bernoulli theory of probability ( bayesian viewpoint ) . in the end we critically review various quantum approaches ( quantum - statistical typicality approaches ) to the foundations of statistical mechanics . the literature on quantum - statistical typicality is organized under four notions ( 1 ) kinematical canonical typicality , ( 2 ) dynamical canonical typicality , ( 3 ) kinematical normal typicality , and ( 4 ) dynamical normal typicality . analogies are seen in the khinchin s classical approach and in the modern quantum - statistical typicality approaches . `` unless the conceptual problems of a field have been clearly resolved , you can not say which mathematical problems are the relevant ones worth working on , and your efforts are more than likely to be wasted '' - e. t. jaynes .

1103.4003

we start the introduction by considering the following views of great men : @xmath0``although nobody is in doubt today of the validity of the remarkable interpretation of thermodynamics with which statistical mechanics , following the efforts of boltzmann and gibbs , has recently provided us , it still remains extremely difficult to give a completely accurate justification for it '' -louis de broglie@xcite . @xmath0``the problem [ defining the ensemble distribution depending upon the external conditions imposed on a system ] was solved by gibbs , although a rigorous justification of the distributions obtained is a complicated problem that is still not completely solved at the present time . it is not even clear to what extent this rigorous justification is possible '' -d . n. zubarev@xcite . @xmath0``there is probably no other well - established field of theoretical physical science that is as much plagued by paradox and criticism of its fundamental logic as is statistical mechanics '' -joseph edward mayer and maria goeppert mayer@xcite @xmath0 `` if we are describing only a state of knowledge about a single system , then clearly there can be nothing physically real about frequencies in the ensemble ; and it makes no sense to ask , ' ' which ensemble is the correct one ? `` ... gibbs understood this clearly ; and that , i suggest , is the reason why he does not say a word about ergodic theorems , ... '' e. t. jaynes@xcite @xmath0 there is lot of confusion regarding ergodic hypothesis in the literature ( regarding whether it is necessary for statistical mechanics or not ) and in the usual discussions of people . also the reasons how and why statistical mechanics works are not properly understood . the problem appears complicated as one has to take into consideration various ingredients of vary different nature ( from the character of experimental measurements , large number of degrees of freedom , and the microscopic dynamical properties ) involved in the statistical approach for dynamical systems . in the present review we try to dis - entangle various ingredients and present the resolution of the ergodic problem in a simpler and clear way . a detailed study of literature shows that there are mainly three camps at the foundations of statistical mechanics ( 1 ) ergodic school , ( 2 ) non - ergodic school , and ( 3 ) quantum foundations of statistical mechanics . we critically analyse all the three approaches to reach on a broader understanding of the foundations of statistical mechanics or we attempt to see the `` complete picture '' . aim of the present review is to show that ergodicity is not relevant for equilibrium statistical mechanics of macroscopic systems ( with tolman and landau ) . we will see that the solution of the problem is in the very peculiar nature of the macroscopic observables ( they are sum functions ) and with the very large number of the degrees of freedom involved as first pointed out by khinchin . similar arguments are used by landau based upon the approximate property of `` statistical independence '' . we critically review these ideas . we also review the role of chaos ( classical and quantum ) in the foundations of statistical mechanics , i.e. , where it is important and where it is not important . for students and beginners the purpose of the article is best served if they first understand the first chapter of landau - lifshitz s book on statistical physics and read the little book of khinchin on the mathematical foundations of statistical mechanics . _ again our motivation is to have a broader view of the subject and a pedagogical presentation . _ in the present section we present a brief summary of the essential topics in the form of questions and answers . in section ii we review the ergodic approach to the foundations of statistical mechanics by dividing it into further subsections . first defining the ergodic problem and then giving a historical account of how the ergodic problem came up with the works of maxwell and boltzmann . in subsection ii(c ) the reformulation of the ergodic problem by birkhoff is given and the resolution of the ergodic problem by khinchin is given in subsection d. the role of chaos , integrability , and non - integrability is discussed in subsection e. in section iii we present the non - ergodic approach . this section consists of two subsections one on landau - lifshitz s approach and another on e. t. jaynes approach with a critique . we also advance the plausibility arguments for equal - a - priori probability hypothesis . the last section is devoted to the quantum foundations of statistical mechanics . in this we review the eigenstate thermalization hypothesis , the quantum ergodic theory of von neumann , and other recent quantum - statistical typicality approaches . we will see that von neumann s quantum ergodic theorem is a general statement applicable to systems with many degrees of freedom and the eigenstate thermalization hypothesis is a consequence of quantum chaos , other approaches falling under typicality properties are also analyzed . the issue of compatibility of microscopic time reversibility and macroscopic time irreversibility is not considered here . this has been clarified in detail in the beautiful papers of jeol lebowitz@xcite(for irreversibility in quasi - isolated systems see@xcite ) . in their famous book@xcite lev landau wrote : `` in the discussion of the foundations of statistical physics , we consider from the start the distribution of small subsystems ...... this allows the complete avoidance of ergodic or similar hypotheses , which are in fact not essential as regard to these aims [ i.e. , the foundations of physical statistics ] '' @xmath0 as we know from the most used prescription of gibbs we calculate the phase space averages of dynamical quantities using appropriate ensembles and we find that these phase averages agree very well with experiments . clearly actual experiments are not done on a hypothetical ensemble they are done on the actual system in the laboratory and these experiments take a finite amount of time . thus it is usually argued that actual measurements are time averages and they are equal to phase averages due to ergodicity , @xmath1 now we will give the following reasons against ergodicity . \(1 ) if one assumes that during the measurement time the system samples all the microstates and @xmath2 limit of equation ( 1 ) is reached during the measurement time , then ones assumption is wrong . it is a well know fact that time taken by the system to sample all the available phase is fantastically large ( more than the age of the universe ) . for example@xcite , if we consider a nanoparticle with @xmath3 nuclei each with spin half and consider only the nuclear spin system . the total number of quantum state are @xmath4 . consider that these spins continuously flip from up spin to down spin and vice versa and this spin flipping is caused by phonons from the system and bath in which the nanoparicle is present . at room temperature this frequency is about @xmath5 cycles per second . let us assume that each spin makes a transition during this period . thus the total number of transitions per second is @xmath6 . the time taken by the system to go over all the states is @xmath7 , and the age of universe is @xmath8 . _ thus clearly during the laboratory measurement time the full ensemble is not realized , and only a very tiny fraction of the total number of the members of the ensemble is realized . _ note that the time scale diverges nearly exponentially with the number of atoms in the sample . \(2 ) if one accepts ergodic hypothesis ( and do not accept explanation ( 1 ) ) , that the laboratory measurements are time averages ( measurement time scale being much large as compared to microscopic dynamical time scale ( time taken to flip a spin @xmath9 ) , then one accepts equation ( 1 ) and accepts @xmath2 ) , then time average value of an observable ( or a result of measurement ) in a series of experiments ( repeating the same experiment again and again ) on the same system will agree with ensemble average . this is in fundamental contradiction with what we observe i.e. thermodynamical measurements are not strictly reproducible@xcite . fluctuations in thermodynamic measurements are the unavoidable consequence of our lack of control of microscopic dynamics and thus our lack of control of the initial micro - condition at the start of the measurement . the above discussion against ergodicity opens the question `` but how does statistical mechanics work ? '' and the theory has been so successful . the answer is : in statistical mechanics we concern with macroscopic systems and with macroscopic observables which are very special ones . special in the sense that they are `` sum functions '' i.e. , their value for the whole system is equal to the sum of equivalent functions for the small parts of the body ( for example , energy of the whole system is equal to the sum of energies of the small parts of the body ) . due to the property called `` statistical independence '' of small parts of the body , the sum functions take almost constant value on the energy hypersurface ( details are given in section ii(d ) ) . @xmath0the microscopic time taken by the measurement produces that `` same '' value . thus equality of time average and phase space average is imposed due to the fact of `` temporal constancy of special macroscopic observables in equilibrium '' . the magnitude of fluctuations goes as @xmath10 . thus these are very small for macroscopic systems . put differently , for an extremely large number micro - states ( called typical micro - states ) the macro - state of the system is the same . only very exceptional micro - states ( `` bad states '' with fantastically small probability ) give non - typical behaviour . to explain these lines let us consider a macroscopic system and measure an observable . repeat the experiment many many times , each time the starting micro - state is different but important point is that each time we get approximately the same value for the observable ( since huge number of micro - states are equilibrium micro - states ) . it is extremely rare that we get a value during a measurement which is very different from the value in other measurements . this `` typicality '' is at the heart of why statistical mechanics explain thermodynamic behaviour . here the probability theory inters into statistical description . the typicality explains its high predictability for macroscopic systems ( even though statistical mechanics has probabilistic basis ) . as the macro - observables are extremely insensitive to micro - condition and microscopic dynamics , the role of microscopic dynamics is very insignificant in equilibrium statistical mechanics . see section ii . d. for a detailed explanation , and the role of fermi - pasta - ulam problem in this connection ( section ii . also it is good idea to abandon 19th century philosophy of explaining everything we see around with microscopic dynamics called `` reductionism '' . boltzmann initially tried to explain 2nd law of thermodynamics with microscopic molecular dynamics but later on realized the need of of statistical laws . at each level of complexity new laws emerge not explainable `` purely '' by shroedinger equation@xcite . in standard practice , for example for a system of volume @xmath11 with @xmath12 particles in equilibrium with very large heat bath at temperature @xmath13 , one use s gibbs canonical ensemble @xmath14 here @xmath15 is the probability to find the system in the infinitesimal phase volume @xmath16 around the phase point @xmath17 [ @xmath18 for a system of @xmath19 degrees of freedom ] . @xmath20 with @xmath21 called boltzmann constant . the negative logarithm of partition function @xmath22 is proportional to the free energy of the system @xmath23 from the free energy and its derivatives all the needed information is extracted . each time we get the gibbs ensemble averaging , for example , pressure is given as @xmath24 with hamiltonian @xmath25 as a function of the volume @xmath11 ( through potential function with boundaries etc . ) . thus our computational algorithm involves phase space averaging . but the actual experiments are done on the given system in the laboratory ( not on the hypothetical ensemble ) . measurements during the experimentation take finite amount of time and thus what we measure in laboratory is the time averages not the ensemble averages . so the immediate question arises how to justify the replacement of time averages with ensemble average . this is called the ergodic problem . there is a considerable confusion regarding the precise meaning of ergodic hypothesis . thus we will present a brief historical account about the origin and its mis - interpretations by ehrenfests@xcite . in this analysis we will better understand the ergodic program for justifying statistical mechanics . boltzmann s first attempt to reduce the second law of thermodynamics to a theorem in mechanics appeared in 1866 and his last papers in columns of nature in 1890s regarding his debates on irreversibility . during his scientific career ( 1866 - 1895 ) boltzmann tried to understand the kinetic theory of gases and thermodynamics with many different approaches based on atomic constitution of matter . favouring one approach at one time and then rejecting it for the another and then returning back again to the previous one . this is a very distinctive character of boltzmann@xcite . roughly , from 1866 - 1871 he was interested in deriving second law of thermodynamics purely from mechanics ( in his 1866 paper no probabilistic argument is present ) , in 1867 he reads maxwell s work and his subsequent papers from 1868 to 1871 uses probabilistic notions after maxwell s velocity distribution function . in these papers he extends maxwell s results to a gas in an external potential ( maxwell - boltzmann law ) . in the paper@xcite of 1868 he first introduces his ergodic hypothesis . he was considering an isolated gas in an arbitrary initial state ( see ehrenfests@xcite ) , then he argues based on the empirical fact that systems tend to equilibrium and permanently stay there that average behaviour over a long time interval will the thermal equilibrium behaviour ( note that this long before his h - theorem ) . according to ehrenfests here he introduces the concept of ensembles ( much before gibbs ) . in justifying the equivalence of ensemble averages and temporal averages he introduces a hypothesis ( note that he do not use the word ergodic at this point , word ergoden appears much later in his 1884 paper in a different context ) . thus the present day terminology is largely due to ehrenfests . note also that this is the birth of modern statistical mechanics . the older treatments in which statistical aspects are related to single molecules are called ( by ehrenfests ) the `` kineto - statistics of the molecule '' and the 1868 paper treats gas model as a whole and ehrenfests call it `` kineto - statistics of the gas model '' . important point to be noted here is that the version of ergodic hypothesis which is attributed to boltzmann is stronger than what the founding father has thought : `` the great irregularity of the thermal motion , and the multiplicity of forces that act on the the body from outside , make it probable that the atoms themselves , by virtue of motion that we call heat , pass through all possible positions and velocities consistent with the equation of kinetic energy , and that we can therefore apply the equations previously developed to the coordinates and velocities of the atoms of warm bodies '' now the present strong form ( trajectory passing through each and every point of the energy surface for an isolated system ) of ergodic hypothesis can not be attributed to boltzmann , it is the mis - interpretation on the ehrenfests side . the representative phase point of the system passing through every point of the phase space is impossible . in fact later in 1913 the impossibility of ergodic systems ( stronger form ) was proved by independently by rosenthal and pancherel on measure - theoretic arguments . ehrenfests after mis - interpreting the boltzmann ergodic hypothesis proposed some what weaker form and they called it the quasi - ergodic hypothesis , which meant that the trajectory of phase point covers the energy surface densely even though not actually passing through every point of it . an excellent discussion of this mis - interpretation is given in the work of s. g. brush@xcite . in the period 1872 - 1878 he wrote two most important papers of his life , the 1872 paper contains what we now call `` boltzmann equation '' and h - theorem . here he claimed that his h - theorem provided the desired theorem from mechanics to explain irreversibility and second law of thermodynamics . this came under a serious objection due to loschmidt in 1876 who insisted on the incompatibility of the time - asymmetric behaviour ( shown by h - theorem ) and the time - symmetric behaviour of microscopic equations of motion ( these debates are now well understood and documented@xcite ) . as a result boltzmann rethought the basis of his approach and presented a very different approach in 1877 what we now call permutational argument and his famous result @xmath26 ( see appendix a ) written in modern form by max planck . equilibrium is now conceived as most probable macro - state instead of stationary macro - state ( in time ) . it is highly probable for a system initially in a non - equilibrium state to move towards equilibrium state ( real space density distribution in accord with external conditions(potentials ) and maxwellian velocity distribution ) as the phase space volume of the equilibrium macro - state is fantastically large as compared to non - equilibrium state . thus the equilibrium is most probable state but arbitrary deviations from it are also probable , but it turns out that the probability is fantastically small . this originates the notions of typicality . third period , in 1880s he left the `` purely '' probabilistic approach and again went back to his mechanical explanation of the laws thermodynamics ( we see here that boltzmann did not stick to one approach ) . he wrote an important paper in 1884 largely forgotten by now ( see the review by gallavotti@xcite ) . this paper is a generalization of a paper by helmholtz on the mechanical models of thermodynamics based on monocyclic systems . boltzmann proves a theorem called heat theorem , i.e. , @xmath27 . here @xmath13 is the time average of the kinetic energy @xmath28 , @xmath29 is the total energy , and @xmath30 the potential energy which depend on an experimentally controllable parameter @xmath11 . the motivation here was to find mechanical analogues of thermodynamic entropy . result was proved for monocyclic systems by helmholtz and boltzmann generalize this for high dimensional systems under the assumption of ergodicity ( note that this is the second time he introduces ergodic considerations ) . in his discrete world - view he imagined the phase trajectory visiting all the discrete cells of the phase space . with the introduction of some stationary phase space distribution ( he calls it `` monode '' ) , the calculation of difficult time averages can be replaced with much simpler phase averages . he considers an ensemble of systems in exactly same macroscopic conditions and a family of stationary distributions which he calls `` monode '' ( note that this the introduction of ensembles much before gibbs ) . a monode which verifies the heat theorem is called by him the `` orthode '' . here ( first time ) he considers two orthodes ( 1 ) ergode ( @xmath31 microcanonical ensemble ) and ( 2 ) holode ( @xmath32 canonical ensemble ) , see fig 1 . as we have seen in section i(a ) , in real thermodynamic measurements we do not have infinite time averages , and thus we are not justified in equating the finite time averages with infinite time averages and then using the ergodic hypothesis ( which has never been proved in generality ) . the answer to this was already clear to boltzmann . the answer lies in the peculiarities of the thermodynamic observables and the large number of degrees of freedom involved . we will discuss this in section ii(d ) . we also note that the original ergodic hypothesis as envisaged by boltzmann was a weaker statement as compared to modern version of that . ehrenfests 1911 encyclopedia article@xcite generated lot of interest about ergodic systems in the community of mathematicians ( paul ehrenfest was the student of boltzmann ) . @xmath0if one is able to prove the ergodic hypothesis for a given system i.e. , @xmath33 , then 42 one eliminates the necessity of determining initial state of the system and of solving hamilton s equations . one justifies the dynamical foundation of statistical mechanics . clearly , then statistical mechanics reduces to a branch of mechanics . but story is not so simple . it is a very difficult problem . in 1931 , g. d. brikhoff@xcite reduced the ergodic property of a dynamical system to an equivalent property called `` metric transitivity '' . he proves the following theorems : 45 @xmath34 exists for almost every @xmath35 . a necessary and sufficient condition for the system to be ergodic is that the phase space be metrically transitive . a system is `` metrically in - transitive '' iff there exists regions @xmath36 and @xmath37 of phase space such that @xmath38 and @xmath39 , which are invariant under the system s dynamics : @xmath40 and @xmath41 for all @xmath42 . in simple words phase point wanders all the available phase space if and only if the system is metrically transitive . @xmath0this property again can not be experimentally verifiable . thus the implications of brikhoff s theorems for physical statistical mechanics are inconclusive . @xmath0but luckily or unluckily , we have seen that ergodicity is not required for doing statistical mechanics of macroscopic systems . khinchin@xcite(see also the 4th chapter of the book@xcite ) approached the ergodic problem from physical point of view . he pointed out the importance of the following points 1 . in statistical mechanical systems of interest the number of the degrees of freedom is very large . 2 . in statistical mechanical systems the observables of interest are very special ones they are `` sum functions '' ( @xmath43 ) . with these considerations he proves the following two theorems : @xmath0theorem ( 1 ) : consider a physical quantity @xmath44 pertaining to a single molecule and define phase coefficient of correlation : @xmath45 . if @xmath46 for @xmath47 the function @xmath48 is ergodic ( @xmath49 ) . the physical assumption that goes into this theorem is : `` because of the fact that the given system consists of a very large number of molecules it is natural to expect that knowledge of the state of a single molecule at a certain moment does not permit us to predict anything ( or almost anything ) about the state in which this molecule will be found after a sufficiently long time '' khinchin . @xmath50this is equivalent to the idea of molecular chaos . next he proves a much more relevant result ( to physical statistics):@xmath50theorem ( 2)@xcite : @xmath51 here @xmath52 is the actual value of the observable ( no time averaging ! ) pertaining to the whole system , and @xmath53 and @xmath54 are @xmath55 . he proves that the correlation co - efficient between phase functions is actually very small for large @xmath12 . @xmath0this implies that the physically relevant observable are self averaging ! the above statement is the resolution of the whole problem _ ( for macroscopic systems and sum - function observables ) . _ as far as one is concerned with macroscopic systems ( large @xmath12 ) and sum function observables chaos , integrability , and non - integrability does not play any role in the foundations of statistical mechanics . as usual statistical mechanical systems in the laboratory are never isolated , the molecular chaos assumption of khinchin seems to be true and the consequences are clear to us . but for `` isolated '' low dimensional systems and non - sum - function observables the case is not so simple . before 1955 there was a general consensus that weak extraneous interactions makes the statistical mechanical systems ergodic ( phase point wandering the whole of phase space ) . in 1955 , fermi and his collaborators did a numerical experiment on a chain of non - linearly coupled harmonic oscillators . their expectation was that the small non - linear coupling will make the system ergodic ( energy stored initially in one of the normal modes will go over in time to all other modes ) . but the result was quite surprising . consider the hamiltonain of the system as @xmath56 . for @xmath57 system is completely integrable and the energy @xmath58 of each normal mode @xmath59 is conserved . it was expected that when @xmath60 energy stored initially in one normal mode will `` spread '' to all other modes . but this did not happen , the energy periodically comes back to the original mode showing no sign of equipartition of energy . this `` paradox '' can be explained with kam theorem . which states that if @xmath61 is small enough then on the constant energy surface , invariant tori survive in a region whose measure tends to @xmath62 as @xmath63 . there is a threshold @xmath64 called kam threshold , if ( 1 ) @xmath65 kam tori play a major role and system does not follow equipartition , and ( 2 ) if @xmath66 then kam tori has a minor role and equipartition happens . in 1987 , livi , pettini , ruffo , and vulpiani published an important paper@xcite on the relevance of chaos regarding the predictions of canonical ensemble . they divided the chain into coupled subsystems containing large number of particles ( see figure 3 ) . let @xmath67 is the per particle time averaged energy of the subsystem @xmath68 and @xmath69 it s heat capacity @xmath70 . it was found that the time averages limit ( private communication with roberto livi ) ] agrees well with the predictions of the canonical ensemble , even though the system was having the kam tori . however the story is not so simple . they also considered another model of coupled rotators in which only @xmath67 showed canonical behaviour not @xmath69 as the integrability - to- non - integrability parameter was changed ( see for details chapter 4 of @xcite ) . note also that @xmath69 is not a sum function . thus one can loosely say that _ `` more coarse your observables , less you depend on chaos '' . _ as expressed at the beginning of section i ( a ) , lev landau did not believe in the need of ergodic or similar hypotheses in justifying the foundations of statistical mechanics@xcite . from the beginning they consider a system always present in a bath , i.e. , subsystem in a given system which is in itself present in bigger system ( as in nature no system is ideally isolated ) . they argue that the extreme complexity of the interactions of the system with the surrounding bodies makes the system s phase point to wander in the phase space . let in a sufficiently long interval of time @xmath13 the phase point spends @xmath71 amount of time in a given volume @xmath72 of the phase space , then , @xmath73 is the probability that , if the system is observed at any arbitrary time , its phase point will be found in the volume @xmath72 . going to the infinitesimals @xmath74 . here @xmath75 is the density of that temporal probability distribution , i.e. , @xmath76 is the probability to find the system , observed at any time , in the infinitesimal phase volume @xmath16 . the important point to be noted here is that we have `` only one '' system under consideration ( no statistical ensemble is considered at this point ) . the distribution defined @xmath75 is a temporal distribution for that system . with the introduction of statistical distribution function one can calculate the mean value of any dynamical quantity @xmath77 as @xmath78 it is obvious by the definition ( 6 ) of probability that the statistical averaging ( equation ( 7 ) ) is exactly equivalent to the infinite time averaging @xmath79 thus avoiding any ergodic hypothesis , as there is no ad - hoc introduction of any probability distribution . the @xmath75 is the actual probability distribution in the system s phase space from its temporal behaviour . but it is non - trivial to prove that @xmath80 this has never been proved in generality ( it turns out that this is a very difficult problem for a system with large number of degrees of freedom ) . we will see from landau s argument and with the assumption of `` statistical independence '' that @xmath81 when number of degrees of freedom involved becomes very large . then they argue@xcite that predictions of statistical mechanics are very reliable due to the fact that the relevant observables take almost constant value on the energy surface . they also consider `` sum '' functions @xmath82 . their considerations are based on a very general fact of `` statistical independence '' of various parts ( pertain to subsystems ) of the macroscopic observables ( it is not true for long range interacting systems ) . on `` statistical independence '' they prove : @xmath83 . one can say @xmath0@xmath84 1 . the distribution function of two sub - systems is equal to the product of individual sub - systems functions ( statistical independence ) , i.e. , @xmath85 ( for simplicity of notation here @xmath86 ) . thus log of the distribution function is an additive integral - of - motion . 3 . due to liouville s theorem ( hamiltonian dynamics ) distribution function is an integral - of - motion ( temporal invariance of the phase space distribution of an ensemble ) . we have only seven independent additive integrals of motion @xmath88 , and @xmath89(from mechanics ) . @xmath90 6 . from which canonical ensemble ensues @xmath91 , say , for a system in a box . the important point to be noted is that there are @xmath92 other integrals of motion ( excluding the additive ones , i.e. , energy , three components of linear and three components of angular momentum ) . these remaining integrals of motion are not additive thus they do not play any role in the landau s argument . thus we see in lev landau s program for the foundations of statistical mechanics , if one starts from the beginning with a system present in a bigger system and one assumes the property of statistical independence and take note of the additive integrals of motion then one can reach the canonical distribution and no ergodic hypothesis is required as we have not constructed any ensemble . we observe that the property of statistical independence is equivalent to the hypothesis of equal - a - priori probabilities . we give two reasons : reason ( 1 ) : there are two views at the foundations of statistical mechanics ( 1 ) is based on the principle of equal - a - priori probabilities ( kubo s book , tolman s etc ) and the ( 2 ) is based on quasi - closed subsystems and the additive integrals of motion ( due to l. d. landau , see their book ) . if one analyze them carefully one sees that both view - points are consistent with each other , first one comes from the second . the principle of equal - a - priori probability is just a consequence of the fact that there are mechanical invariants of motion proportional to log of the distribution function for a quasi - closed subsystem @xmath93 ( due to statistical independence ) , which is another integral of motion due to liouville s theorem , thus one has @xmath94 . the right hand side of this equation is the content of dynamics ( conservation laws ) . the left hand side consists of a more subtle quantity related to statistical properties and a consequence of the theorem of dynamics and the property of statistical independence . since @xmath95 is constant no matter where is phase point is in the available phase space . consequently log of @xmath96 , thus @xmath96 is same for all possible @xmath17 . this is an equal - a - priori probabilities ( see figure 4 ) . as tolman put it microstate has no preference to be in this or that part of phase space , but with the help of statistical independence it is more clearly seen . @xmath0reason ( 2 ) : if one assume `` statistical independence '' then @xmath97 . let @xmath98 , where @xmath99 . now it is easy to prove that @xmath100 . thus @xmath101 is almost constant on the energy surface @xmath102 for @xmath103 . which again force us to say that @xmath104 take almost constant value on the energy surface a statement of equal - a - priori probabilities . thus we see that the hypothesis of statistical independence directly gives us equal - a - priori probabilities . the important point here to be noted is that the above reasons are only qualitative observations based on statistical independence and these are not rigorous " mathematical theorems . most fundamental ingredient : the property of `` statistical independence '' . @xmath0if one assume that various parts(macroscopic ) of the body are statistically independent from each other , then one can construct canonical and microcanonical formulation without any ad - hoc and extra hypothesis ( only based on statistical independence and some additive mechanical invariants of motion ) . consider a macroscopic system in equilibrium , all its parts ( macroscopic ) or subsystems of the whole system are in equilibrium with each other . let @xmath105 is the energy of the subsystem . it remains constant when the whole system is in equilibrium and the subsystem although small as compared to the whole system is still macroscopic and the relative fluctuations in @xmath105 are very small @xmath106 . one can argue that after a sufficiently long time intervals the subsystem can not be considered quasi - closed , as the effect of interaction of subsystems , however weak , will ultimately appear , and this weak interaction ultimately leads to the establishment of statistical equilibrium . we will see below that in equilibrium this effect does not apply but in non - equilibrium it applies and we do not have any `` non - equilibrium canonical '' formulation . let us consider the subsystem for @xmath71 amount of time ( much less than the relaxation time of the subsystem under consideration ) . the dynamics is hamiltonian and due the liouville s theporem @xmath107 is constant of motion . with landau s argument @xmath108 or @xmath109 now let us wait for a time interval sufficiently long as compared to the relaxation time . after this , the subsystem is no longer quasi - closed and say the distribution is @xmath110 at the end of this long time interval . again consider an interval of time @xmath111 much less than the relaxation time , during this interval system is again closed and landau s argument again applies @xmath112 or @xmath113 . from the above two equations we have @xmath114 . now thermodynamic considerations show that @xmath115 is to be identified with inverse temperature of any subsystem of the whole system . as the whole system is in equilibrium thus all the subsystems of the whole system are in equilibrium with each other thus @xmath116 . also normalization conditions demand that @xmath117 . this gives us @xmath118 and the distribution remains stationary . thus in equilibrium landau s argument applies at all times . moreover @xmath119 is constant , as rhs of this equation is constant both in time and phase ( we neglect the fluctuations in @xmath105 ) . thus @xmath120 , which is microcanonical ensemble . on similar lines with landau s argument we can proceed to show the canonical distribution @xmath119 . here we allow fluctuations in @xmath105 but keep @xmath115 same for all subsystems . _ the most important point to be noted here is that we do not `` mentally construct '' an ensemble and we do not assume equal - a - priori probability hypothesis . all the statistical formulation comes out from the property of statistical independence " and the few additive integrals of motion ( note that statistical independence does not apply to long range interacting systems ) . all other integrals of motion does not play any role in the landau s argument because they are not additive . _ ( 1)temporal definition of probability : in a sufficiently long interval of time @xmath13 the phase point spends @xmath71 amount of time in a given volume @xmath72 of the phase space , thus @xmath121 is the probability that , if the system is observed at any arbitrary time , its phase point will be found in the volume @xmath72 . in the above definition , non - stationary case can not be defined , as the probability to find the system in @xmath122 itself change with time . thus the above definition excludes ubiquitous non - equilibrium cases . \(2 ) as landau - lifshitz s whole analysis is based upon quasi - closed subsystems , the relation @xmath123 holds good only for not too long intervals of time `` . as after a sufficiently long time intervals the subsystem can not be considered quasi - closed . in their own words ' ' over a sufficiently long interval of time ( compared to the relaxation time ) , the effect of interaction of subsystems , however weak , will ultimately appear . moreover , it is just this relatively weak interaction which leads finally to the establishment of statistical equilibrium " . we have seen that in equilibrium landau s argument applies at all times , but when the system is not settled to equilibrium ( temperature equilibrium ) we can not apply landau s argument due to the above reason . there are two schools at the foundations of probability theory ( 1 ) frequentists ( probability from many random experiments ) , and ( 2 ) bernoulli - bayes - laplacians or simply bayesians ( principle of indifference ) . jaynes standpoint : the ergodic problem is a conceptual problem related to the very concept of ensemble itself which is a byproduct of frequency theory of probability . ergodic problem becomes devoid of any meaning when the probabilities of various micro - states are interpreted with bernoulli - bayes - laplace theory of probability . in the bayesian theory of probability , probabilities of occurrence of events are independent of the frequency concept it is a more general viewpoint in which frequency theory is a special case and is based upon the principle of indeference . thus if we do not visualize the probability of occurrence of a micro - state with frequency ( or construct ensemble ) there is no question of ensemble averaging and ergodic problem is completely bypassed . then the statistical mechanical theory is attacked with statistical inference ( shannan entropy approach ) . in jaynes words@xcite : @xmath0 `` we can have our justification for the rules of statistical mechanics , in a way that is incomparably simpler than anyone had thought possible , if we are willing to pay the price . the price is simply that we must loosen the connections between probability and frequency , by returning to the original viewpoint of bernoulli and laplace . the only new feature is that their principle of insufficient reason is now generalized to the principle of maximum entropy . '' for more details see his detailed account@xcite . the present author feels that it does not matter ( atleast in the computational problems of statistical mechanics ) that which theory of probability one is accepting . but there is a disadvantage if we attack statistical mechanical problem with bayesian viewpoint . if one accepts the bayesian viewpoint , then one abandons the concept of ensembles . the probabilities of the microstates are obtained by maximizing the shannon s entropy ( with the given amount of macroscopic information ) . but this also means that one is neglecting the dynamical properties of constituents of matter ( although they are not so important for sum - functions , a - la khinchin ) . obviously molecules of a gas obey newton s laws or schroedinger s equation . the phase point of an isolated system has no preference for this region or that region of phase space ( the phase point of an isolated system visits all the accessible regions of phase space , _ with equal frequency _ although time involved are fantastically large ) . it is also the fact that in macroscopic observables ( which are so coarse on microscopic scale ) the microscopic dynamical properties do not reflect _ in all its detail _ , only very general microscopic dynamical properties reflect . for example , additive mechanical invariants play the essential role ( in landau s approach ) . thus neglecting the dynamical properties of the constituents altogether is not a valid starting point . thus it seems difficult to describe the integrable to non - integrable transition in dynamical systems with jaynes approach . as we have seen that the sum - function macroscopic observables are so weakly depend on the dynamics of the phase point and thus this is the resolution ( with khinchin ) , not with the bayesian viewpoint , which makes ergodic problem devoid of any meaning . historically it is to be noted that prof . g. uhlenbeck objected jaynes ideas@xcite . since in the quantum case position and momentum do not commute with each other , the classical concept of phase space that helps us to envisage the dynamics of the phase point on the energy surface breaks down . however an equation analogous to equation ( 1 ) can be written in the quantum case also . the equivalent of `` phase space energy shell '' in the quantum case is the sub - hilbert space @xmath124 of the total hilbert space @xmath125 . @xmath126 is defined as the space spanned by all @xmath127 ( eigenstates of the system s hamiltonian ) such that their corresponding eigenvalues @xmath128 are in a specified interval of energy from @xmath129 to @xmath130 . the quantum equivalent of the classical ergodic hypothesis ( equation ( 1 ) ) is written as @xmath131 for a system in pure quantum state ( for details see equation ( 11 ) below ) . for a system in mixed quantum state replace @xmath132 with @xmath133 as the diagonal element of the density matix in the energy representation . _ here again we encounter ( lhs of the above equation ) the problem of infinite time averaging , and all the arguments of section i(a ) are valid , but we will see below that the `` resolving results '' analogous to the khinchin s self averaging theorem hold in quantum case also either based upon quantum chaos or based on von neumann s `` scale separation '' ideas . we again see the fundamental role of typicality and large dimensional hilbert spaces_. eigenstate thermalization hypothesis(eth ) implies that the thermalization happens at the level of individual eigenstates . consider a system with hamiltonian @xmath25 and eigensystem @xmath134 prepared in some initial state . if system has unitary evolution , then at any later time @xmath42 @xmath135 quantum mechanical mean of an operator @xmath136 pertaining to the system is : @xmath137 consider infinite time average @xmath138 @xmath139 is also called the diagonal ensemble in@xcite . now the thermodynamic universality demands : long time average @xmath32 average over appropriate statistical ensemble ( microcanonical or canonical etc . ) @xmath140 this is an universality relation with lhs depends on initial conditions but rhs does not ! _ now eigenstate thermalization hypothesis ( eth ) says that there are no eigenstate to eigenstate fluctuations in @xmath141 s for eigenstates close in energy _ @xmath142 this is called the eigenstate thermalization hypothesis ( eth ) by deutsch ( 1991 ) and srednicki ( 1994)@xcite : @xmath0``the quantum expectation value @xmath143 in a large interacting many - body system is equal to the thermal average '' @xmath144 if this is valid in generality , then one explain thermodynamic universality . also if the system obeys berry s conjecture ( eigenfunctions are random gaussian variables for a classically chaotic quantum system ) the off - diagonal elements @xmath145 are negligible and _ then one obtains thermal behaviour without any time averaging at all_@xcite . also it has been numerically shown@xcite that the magnitude of the off - diagonal elements of the momentum distribution operator are very small as compared to the diagonal elements @xmath146 . clearly much work is needed to show eth for other operators and _ importantly for sum - function operators _ ( across the integrability - to-(non)integrability transition ) . for detailed discussion see@xcite and the ref@xcite for the behaviour of diagonal and off - diagonal elements of various other observables . study of literature shows that eth has been proved analytically only in special circumstances ( see figure 5 ) . _ figure shows that some kind of chaos is necessary for eth to hold_. this is also evident in the numerical experiment of rigol etal@xcite for a small system ( five hard core bosons on a lattice ) that eth hold good in the non - integrable system but fails in the integrable one ( for marginal momentum distribution as the observable ) . clearly systems with very small number of degrees - of - freedom can not be fit into khinchin s program ( which requires macroscopic systems ) . this implies that the statistical mechanical universality ( irrelevance of chaos ) that we enjoy with macroscopic systems and sum function observables may no longer holds good for small isolated quantum systems . we present here the qualitative statement of von neumann s quantum ergodic theorem , for a precise definition see@xcite . our aim here is to understand the physical basis of the theorem , to draw some analogies with the approach by khinchin , and to see the connection with the eth ( eigenstate thermalization hypothesis ) . setup : consider a system with hamiltonian @xmath25 and total hilbert space @xmath125 . consider that the total hilbert space is partitioned into mutually orthogonal sub - hilbert - spaces @xmath126 , with family @xmath147 ( @xmath148 ) . let @xmath149 and @xmath150 . let total number of partitions is @xmath151 i.e. , @xmath152 . the physical basis of this partitioning ( according to von neumann ) is that each @xmath126 belongs to a different macro - state ( or macro - observer ) . this is the crucial insight of von neumann , as in quantum theory all operators corresponding to observables do not commute among themselves . von neumann take a physical stand point that the measurement process is coarse on microscopic variables , thus operators can be taken `` approximately '' commuting ( measurement errors are large as compared to bounds of uncertainty relations ) . this is called `` rounding '' of operators . the aim of the quantum ergodic theorem ( qet ) is to tell us some kind of universality or `` normality '' that the quantum expectation values of operators are close to the microcanonical averaging for most of the time ( under some conditions , see below ) . it is a kind of quantum h - theorem which says that for all initial states the time evolution take the system from initial non - equilibrium state to the equilibrium state ( provided the system satisfy some conditions ) . thus the total hilbert space is partitioned in sub - spaces @xmath126 and the macroscopic observables take specific values in each @xmath126 ( values taken in @xmath126 are different from those taken in @xmath153 when @xmath154 ) . the `` rounded '' operators commute with each other but they do not commute with the hamiltonian ( this is analogous to non - integrals of motion in classical gibbs picture ) . qualitative statement of qet : let @xmath155 be the projection to @xmath126 with macro - state @xmath156 having some values of observables . @xmath0then qet says : for _ all _ initial wavefunctions @xmath157 ( note that this includes the special initial non - equilibrium states ) , we have for most of the time @xmath158 in words : probability distribution over macro - states is approximately equal to the ratio of the dimension of hilbert space corresponding to that macro - state to the dimension of the total hilbert space . thus the macro - state with largest @xmath159 is most probable ( which is in accord with common sense , equilibrium state has the largest @xmath160 and it is typically observed ) . conditions involved in qet are : 1 . hamiltonian is free of resonances i.e. , @xmath161 , unless either @xmath162 or @xmath163 . the quantity @xmath164 is small for all @xmath156 . the meaning of small here is that the quantity @xmath165 is smaller than @xmath166 . here @xmath61 and @xmath167 are small positive numbers and @xmath151 is the number of partitions @xmath168 . for more details see@xcite and english translation of the original von neumann s article@xcite . this is qet . this property should be better called `` normality '' after gltz . gltz also give a stronger bound on the deviations from the average@xcite . the notion of `` normality '' is a special case of a broader notion of `` typicality '' . there are various kinds of normality properties as discussed in the next subsection . strict determinism of classical mechanics is replaced by the notion of `` typicality '' of statistical mechanical systems . the notion of typicality captures the _ almost _ deterministic behaviour of statistical mechanical systems ( although atypical behaviour is possible in principle but highly improbable ) . 1 . qet is a typicality result ( typicality due to `` scale separation''(see dnt below ) ) valid for large systems ( i.e. , containing large number of particles ) but eth ( again a typicality result typicality due to quantum chaos ) has been applied even for few boson atoms in an optical lattice . eth involves quantum chaos and qet does not involve quantum chaos ! eth has been proved ( analytically ) only in special circumstances based on the assumption of quantum chaos@xcite , see figure(5 ) . general theory is seriously lacking . recent literature can be organized under the following four notions ( see figure 6 ) : qualitative definition : we know that a small system weakly coupled to a large bath is described by canonical ensemble when the composite system ( system + bath ) is described by microcanonical ensemble . kinematical canonical typicality says that even if the state of a composite system is pure ( rather than a mixture as in the traditional case ) the reduced density matrix of the system is canonical . kct has been announced in@xcite extending the works of schrdinger@xcite . related works appeared in@xcite . in @xcite the basic assumption is that the probability distribution is uniform over all normalized wavefunctions @xmath169 with energy in the shell @xmath170 $ ] of the composite system . the system density matrix ( when the composite is in pure state ) @xmath171 is obtained by tracing out bath degrees of freedom . the crucial role in the tracing is played by the expansion co - efficients @xmath172 in @xmath173 which were assumed gaussian random variables . it is shown that @xmath174 is approximately canonical ( see for details@xcite ) . a rather general treatment of kct is given in@xcite . they prove kct using levy s lemma which is analogous to the law of large numbers . they prove that @xmath175 , where @xmath176 is the trace distance @xmath177 between @xmath174 ( reduced density matrix of the system when the composite is in the pure state ) and @xmath178 ( the traditional ( when composite is in the mixed state ) reduced density matrix of the system ) . the average @xmath179 is over the environment states with the standard unitarily invariant measure . @xmath180 is the dimension of the system s hilbert space and @xmath181 is a measure of the effictive size of the environment @xmath182 ( @xmath183 is the dimension of the environment s hilbert space ) . thus when bath or environment is much larger than the system i.e. , @xmath184 then @xmath185 and if @xmath186 is very small then @xmath174 and @xmath178 are close to each other , @xmath175 , which is a statement of kct ( for details and related results see@xcite ) . qualitative definition : when the composite system ( system + bath ) is in a pure state , the expectation value of an operator pertaining to the system , _ after sufficiently large time _ , will be almost equal to the canonical expectation value . dct has been proved in@xcite and there were numerical experiments@xcite for its evidence . in @xcite dct has been proved with an assumption of weak coupling between system and bath with @xmath187 ( @xmath188 is the minimum spacing between the energy levels of the system , @xmath189 is the magnitude of system - bath coupling , and @xmath190 is the maximum spacing between the energy levels of the bath ) . also a hypothesis of `` equal weights for eigenstates '' is used which makes the energy expansion co - efficients small . the statement of dct goes like this . suppose that the composite - system is in the pure state @xmath191 at any time @xmath42 , the expectation value of an operator of the system is written as @xmath192 . then dct says , after sufficiently long time , @xmath193 . for proof see@xcite . dct has been proved in a much more general setting in@xcite and also in@xcite using an assumption that matrix elements of the interaction hamiltonian has random phases and then constructing markovian master equations . in @xcite , the authors extend their previous kinematical results ( last subsection ) . they prove that @xmath194{\rangle}_t \leq \frac{1}{2}\sqrt{\frac{d_s}{d^{eff}(\omega_b ) } } \leq \frac{1}{2}\sqrt{\frac{d_s^2}{d^{eff}(\omega ) } } $ ] . here $ ] is the system density matix ( @xmath196 is the total density matrix ( system + bath ) ) and @xmath197 . in words , theorem says that the time average of the trace distance ( see previous subsection ) between instantaneous system s density matrix and its time averaged density matrix is small if @xmath198 . this means that system spends almost all of its time near to @xmath199 . this is called equilibration ( weakly fluctuating about a steady state this steady state need not be an equilibrium state ) . the important point is that this result is completely general ( except the assumption that the hamiltonian is free of resonances condition no ( 1 ) in von neumann s theorem ) . it does not depend upon the nature of system - bath coupling , condition of bath ( whether it is in equilibrium or not ) . they also treat the problem of thermalization ( i.e. , initial state independence , as the steady state reached may depend on the initial conditions ) . for a detailed account and further results , see their very nice paper@xcite . our purpose here is to introduce various typicality theorems and to give the excitement to the reader . interested reader should go to the original literature . qualitative definition : for an isolated quantum system in an arbitrary pure state , the quantum expectation value of an operator in that state _ hardly _ deviates from the ensemble average . the meaning of `` hardly '' is clarified below . a clear account of knt is given in@xcite . the statement of knt theorem there goes like this ; consider that the system is in some pure state @xmath200 , with @xmath201 as eigenstates . the quantity @xmath202 is small . here the average @xmath203 , where @xmath204 is the probability distribution over the expansion co - efficients . the meaning of `` small '' in the above statement is that , @xmath205 , here @xmath206 s are positive numbers typically of order one , and @xmath207 is the difference between the maximum and minimum eigenvalues of @xmath208 . for @xmath28 operators one has @xmath209 \leq \frac{k ( max_{n } q_n ) ( tr \rho^2)}{{\epsilon}}$ ] . where for @xmath28 operators @xmath210 runs from @xmath62 to @xmath28 , and @xmath211 is the difference between the maximum and minimum eigenvalues of @xmath212 . the smallness of @xmath213 ( which is must for typicality ) is imposed due to the conditions involved in the above theorem ; 1 . the expansion co - efficients @xmath214 are statistically independent and @xmath215 are equally likely for arbitrary phase @xmath216 . thus @xmath217 . 2 . the mixed state @xmath218 , with ( overbar means average over @xmath219 has low purity @xmath220 . the author@xcite justify these assumptions in a qualitative way that these are valid in `` practice '' if not `` in principle '' for a system with large number of degrees of freedom . the rigorous justification is serious lacking . the role of quantum chaos in these theorems is not clear at present . knt is also discussed in @xcite where it is termed as `` thermodynamic normality '' the von neumann s quantum ergodic theorem is the first known theorem of dynamical normal typicality . this theorem is based upon the observer s inability to further resolve @xmath126 by macroscopic means and the bound on fluctuations of the quantum expectation values of the operators from microcanonical averaging due to the structure imposed on the hilbert space by macroscopic perspective ( the large @xmath221 and large @xmath222 , see subsection b ) . important point is that _ no _ chaos or disorder assumption was invoked by neumann . in his own words@xcite `` ... we emphasize that the true state ( about which we do calculations ) is a wavefunction , i.e. , something microscopic to introduce a macroscopic description of the state would mean to introduce disorder assumptions , which is what we definitely want to avoid '' . _ his work is a result of `` scale separation '' ( i.e. , observer s inability to further resolve @xmath126 by macroscopic means , the large @xmath221 and large @xmath222 etc)_. spirit of his work resembles with that of khinchin and more recent works@xcite , in contrast to eth which involves quantum chaos . this state of affairs ( un - necessity or necessity of disorder ) make the situation complicated and no satisfactory resolution is available at present . we again consider the role of quantum chaos ( previous example was eth ) . the important work in this direction ( quantum chaos in the foundations of statistical mechanics ) was done by asher peres in 1980s@xcite . he discusses dnt based upon his definition of quantum chaos that the observables are represented by pseudorandom matrices when the hamiltonian is diagonal . due to this , expectation values of these observables tend to equilibrium values and fluctuations around these equilibrium values are , on the average , very small . his argument for quantum ergodicity goes like this ; the quantum expectation value of an observable is @xmath223 for non - degenerate spectrum , time average is @xmath224 ( here @xmath225 is the density matrix in energy representation ) . here comes the peres argument ; consider that the system has very many energy levels in the range @xmath129 to @xmath226 , ( 1 ) distribution of these energy levels is very very different in regular and chaotic case@xcite , ( 2 ) the observable @xmath227 too behave very differently for regular and chaotic case ( regular system have selection rules , most @xmath227 vanish and only a few are large , and in chaotic case @xmath227 is pseudorandom ( his definition of quantum chaos ) and very numerous ) , and ( 3 ) for a chaotic system , the pseudorandom @xmath227 are _ statistically independent _ from @xmath225 in that energy range and _ their average does not appreciably depend on the energy interval @xmath228_. as @xmath229 , one has @xmath230 . this he calls quantum ergodicity . the fundamental assumption used is the assumption of statistical independence of @xmath227 from @xmath225 . then he discuss more relevant property , called mixing in quantum case , that the fluctuations of @xmath231 about its equilibrium value @xmath232 are , on the average , small i.e. , @xmath233 ( @xmath234 is the large number of energy levels in the interval @xmath129 to @xmath130 ) . to prove this he again used the assumption of statistical independence . thus we see from peres program that dnt is a consequence of quantum chaos ( observables as pseudorandom matrices ) and _ the assumption of statistical independence . _ this is in sharp contrast with the approach of von neumann ( as von neumann did not use quantum chaos ) , still we do not have a coherent picture that when chaos is important ( peres ) and when chaos is not important ( von neumann ) ? various typicality approaches are summarized in figure ( 6 ) . 1 . justification of gibbs ensembles is a complicated problem . 2 . ergodic hypothesis is not necessary for the workings of statistical mechanics as far one is concerned with sum - function observables and macroscopic systems . statistical mechanics explain thermodynamic behaviour because of the property of statistical independence , or , more accurate approach of khinchin which is the cause of typicality and canonical formalism can be obtained from landau s program without using any ad - hoc hypothesis except the property of statistical independence . the issues of chaos , integrability or non - integrability becomes important when one deals with small isolated systems , for example , @xmath235 or so bosonic or fermionic atoms in an optical lattice . they are irrelevant for a macroscopic system and sum - function observables . clearly von neumann s quantum ergodic theorem and other quantum - statistical typicality theorems are analogous to khinchin s self averaging theorem . these become more accurate for a system with large number of degrees of freedom and these say that for such systems the expectation values of the observables ( sum - functions in khinchin s case ) stay close to the microcanoncal / canonical predictions . in principle , khinchin s approach should come out as a special case of von neumann s theory . clearly much work is needed in this direction . one can contemplate _ classical _ kct ( this traditional statistical mechanics ) , _ classical _ dct , and so on . 7 . the above presented consolidation or amalgamation of various quantum - statistical typicality theorems singles out eth and peres dnt . eth and peres dnt involves quantum chaos , whereas others like von neumann s do not involve quantum chaos . still we do not have a coherent picture ; when chaos is important ( peres ) and when chaos is not important ( von neumann ) ? i would like to end this manuscript with the following aphoristic words of prof . jeol lebowitz about the ergodic hypothesis ; @xmath0``now it is the real time to recognize that the ergodic hypothesis is not a necessary and is not a sufficient condition for the foundation of statistical mechanics '' author would like to thank to prof . sheldon goldstein and prof . marcos rigol for many critical discussions , to prof . angelo vulpiani and dr . jitesh bhatt for pointing out many corrections and suggestions . he is also thankful to prof . michael kastner for giving him the opportunity to present this work at 2nd stellenbosch workshop on statistical physics from 7th march to 18th march 2011 at stellenbosch , south africa , and also for the partial financial help . boltzmann introduced the notion of `` komplexions '' of molecules as follows . consider the kinetic energies of the molecules take the following discrete values @xmath237 , and let @xmath238 number of molecules posses a kinetic energy @xmath239 . number of komplexions for a given distribution is @xmath241 . he noted that maximization of @xmath242 with the above constraints leads to maxwellian kinetic energy distribution . @xmath0boltzmann then defines `` degree of permutability '' @xmath243 , and finds that ( by direct computation ) that @xmath244 is equal to the entropy of an ideal gas in a reversible process up to an additive constant . @xmath245.\ ] ] it was max planck who gave the general foundations to the fundamental principle @xmath246 , ( using @xmath247 in place of @xmath242 ) introducing first time the constant @xmath21 called the boltzmann constant . he proposed the following fundamental hypothesis called planck s thermodynamic probability hypothesis . `` the entropy of physical system in a definite state depends solely on the thermodynamic probability @xmath247 of this state '' , @xmath248 . @xmath0considering two systems in equilibrium with each other . from second law , one has additivity of entropy @xmath249 , thus @xmath250 and then by differentiating one can obtain @xmath251 . it gives @xmath252 the important point here is that the equilibrium is conceived as the most probable state rather than the temporally constant state . but as we know that the fluctuations in observables are extremely small for macroscopic systems both definitions are approximately consistent with each other . goldstein s. , lebowitz j. , tumulka r. , zanghi n. , long - time behaviour of macroscopic quantum systems : commentary accompanying the english translation of john von neumann s 1929 article on the quantum ergodic theorem " john von neumann , `` proof of the ergodic theorem and the h - theorem in quantum mechanics '' , eur . j. h * 35 * : 201 - 237 ( 2010 ) , arxiv:1003.2133 , [ english translation by roderich tumulka of `` beweis des ergodensatzes und des h - theorems '' in zeitschrift fuer physik * 57 * : 30 - 70 ( 1929 ) ] . sheldon goldstein , joel l. lebowitz , christian mastrodonato , roderich tumulka , nino zanghi , `` normal typicality and von neumann s quantum ergodic theorem '' , proceedings of the royal society a * 466 * , 3203 ( 2010 ) . r. v. jensen and r. shanker , `` statistical behavior in deterministic quantum systems with few degrees of freedom '' , phys . lett . , * 54 * , 1879 ( 1985 ) ; k. saito , s. takesue , and s. miyashita , `` system - size dependence of statistical behavior in quantum system '' , j. phys . . jpn . * 65 * , 1243 ( 1996 ) ; `` thermal conduction in a quantum system '' , phys . e * 54 * , 2404 ( 1996 ) . a version of dnt ( equilibration as opposed to thermalization ) has been proved in@xcite . here it is shown that the time average of the difference in the instantaneous expectation values of an observable and its ensemble average is very small . ensemble here is the diagonal ensemble ( equation 11 ) . in literature it is called equilibration@xcite . the agreement of instantaneous expectation value of an observable with microcanonical or canonical ensemble is called thermalization which is a much more difficult problem and possibly involves chaos . hal tasaki , `` the approach to thermal equilibrium and ' ' thermodynamic normality``an observation based on the works by goldstein , lebowitz , mastrodonato , tumulka , and zanghi in 2009 , and by von neumann in 1929 '' , arxiv:1003.5424 . r. balian and c. bloch , `` distribution of eigenfrequencies for the wave equation in a finite domain : iii . eigenfrequency density oscillations'',ann . phys.(ny ) * 69 * , 76 ( 1972 ) ; `` solution of the schrdinger equation in terms of classical paths '' , bf 85 , 514 ( 1974 ) .

the weyl tensor and the ricci tensor can be algebraically classified in a lorentzian spacetime of arbitrary dimensions using alignment theory . used in tandem with the boost weight decomposition and curvature operators , the algebraic classification of the weyl tensor and the ricci tensor in higher dimensions can then be refined utilizing their eigenbivector and eigenvalue structure , respectively . in particular , for a tensor of a particular algebraic type , the associated operator will have a restricted eigenvector structure , and this can then be used to determine necessary conditions for a particular algebraic type . in principle , this analysis can be used to study all of the various algebraic types ( and their subclasses ) in more detail . we shall present an analysis of the discriminants of the associated characteristic equation for the eigenvalues of an operator to determine the conditions on ( the associated ) curvature tensor for a given algebraic type . we will describe an algorithm which enables us to completely determine the eigenvalue structure of the curvature operator , up to degeneracies , in terms of a set of discriminants . since the characteristic equation has coefficients which can be expressed in terms of the scalar polynomial curvature invariants of the curvature tensor , we express these conditions ( discriminants ) in terms of these polynomial curvature invariants . in particular , we can use the techniques decribed to study the necessary conditions in arbitrary dimensions for the weyl and ricci curvature operators ( and hence the higher dimensional weyl and ricci tensors ) to be of algebraic type * ii * or * d * , and create syzygies which are necessary for the special algebraic type to be fulfilled . we are consequently able to determine the necessary conditions in terms of simple scalar polynomial curvature invariants in order for the higher dimensional weyl and ricci tensors to be of type * ii * or * d*. we explicitly determine the scalar polynomial curvature invariants for a weyl or ricci tensor to be of type * ii * or * d * in five dimensions . a number of simple examples are presented to illustrate the calculational method and the power of the approach . in particular , we will present a detailed analysis of the important example of a 5 dimensional rotating black ring .

1011.2175

higher dimensional lorentzian spacetimes are of considerable interest in current theoretical physics . lorentzian spacetimes for which all polynomial scalar invariants constructed from the riemann tensor and its covariant derivatives are constant are called @xmath0 spacetimes @xcite . all curvature invariants of all orders vanish in an @xmath1-dimensional lorentzian @xmath2 spacetime @xcite . the higher dimensional @xmath2 and @xmath0 degenerate kundt spacetimes are examples of spacetimes that are of fundamental importance since they are solutions of supergravity or superstring theory , when supported by appropriate bosonic fields @xcite . higher dimensional black hole solutions are also of current interest @xcite . the introduction of _ alignment theory _ @xcite has made it possible to algebraically classify any tensor in a lorentzian spacetime of arbitrary dimensions by boost weight . in particular , the dimension - independent theory of alignment , using the notions of an aligned null direction and alignment order in lorentzian geometry , can be applied to the tensor classification problem for the weyl tensor in higher dimensions @xcite ( thus generalizing the petrov classifications in four dimensions ( 4d or @xmath3 ) ) . indeed , it is possible to categorize algebraically special tensors in terms of their alignment type , with increasing specialization indicated by a higher order of alignment . in practice , a complete tensor classification in terms of alignment type is possible only for simple symmetry types and for small dimensions @xcite . however , partial classification into broader categories is still desirable . we note that alignment type suffices for the classification of 4d weyl tensors , but the situation for higher - dimensional weyl tensors is more complicated ( and different classifications in 4d are not equivalent in higher dimensions ) . in the higher dimensional classification , the secondary alignment type is also of significance . further refinement using bivectors is also useful ( see below ) . the analysis for higher dimensional weyl tensors can also be applied directly to the classification of higher dimensional riemann curvature tensors . in particular the higher - dimensional alignment types give well defined categories for the riemann tensor ( although there are additional constraints coming from the extra non - vanishing components ) . we can also use alignment to classify the second - order symmetric ricci tensor ( which we refer to as ricci type ) . the ricci tensor can also be classified according to its eigenvalue structure . in addition , using alignment theory , the higher dimensional bianchi and ricci identities have been computed @xcite and a higher dimensional generalization of _ newman - penrose formalism _ has been presented @xcite . another classification can be obtained by introducing _ bivectors _ the algebraic classification of the weyl tensor using bivectors is equivalent to the algebraic classification of the weyl tensor by boost weight in 4d ( i.e. , the petrov classification @xcite ) . however , these classifications are different in higher dimensions . in particular , the algebraic classification using alignment theory is rather course , and it may be useful to develop the algebraic classification of the weyl tensor using bivectors to obtain a more refined classification . the bivector formalism in higher dimensional lorentzian spacetimes was developed in @xcite . the weyl bivector operator was defined in a manner consistent with its boost weight decomposition . the weyl tensor can then be algebraically classified ( based , in part , on the eigenbivector problem ) , which gives rise to a refinement in dimensions higher than four of the usual alignment ( boost weight ) classification , in terms of the irreducible representations of the spins . in particular , the classification in five dimensions was discussed in some detail @xcite . a _ scalar polynomial curvature invariant of order @xmath4 _ ( or , in short , a scalar invariant ) is a scalar obtained by contraction from a polynomial in the riemann tensor and its covariant derivatives up to the order @xmath4 . the kretschmann scalar , @xmath5 , is an example of a zeroth order invariant . scalar invariants have been extensively used in the study of @xmath2 and @xmath0 spacetimes @xcite . in @xcite it was proven that a four - dimensional lorentzian spacetime metric is either _ @xmath6-non - degenerate _ , and hence completely locally characterized by its scalar polynomial curvature invariants , or _ degenerate kundt_. in arbitrary dimensions , demanding that all of the zeroth polynomial weyl invariants vanish implies that the weyl type is * iii * , * n * , or * o * ( similarly for the ricci type ) . it would be particularly useful to find necessary conditions in terms of simple scalar invariants in order for the weyl type ( or the ricci type ) to be * ii * or * d*. _ the main goal of this work is the determination of necessary conditions in higher dimensions for algebraic type , and particularly type * ii * ( or * d * ) , using scalar curvature invariants . _ in 4d , demanding that the complex zeroth order quadratic and cubic weyl invariants @xmath7 and @xmath8 vanish ( @xmath9 ) implies that the weyl ( petrov ) type is * iii * , * n * , or * o * @xcite . in addition , the weyl tensor is of type * ii * ( or more special ; e.g. , type * d * ) if @xmath10 . it is useful to express the weyl type * ii * conditions in non - np form . the syzygy @xmath11 is complex , whose real and imaginary parts can be expressed using invariants of weyl not containing duals . the real part is equivalent to : @xmath12 and the imaginary part is equivalent to : @xmath13 where @xmath14 the ricci type * ii * conditions are : @xmath15 where @xmath16 is the trace - free ricci tensor @xmath17 and to that in @xcite , in order to be self - consistent in this paper . ] @xmath18.\label{riccidef}\end{aligned}\ ] ] if a spacetime is riemann type * ii * , then not only do the weyl type * ii * and ricci type * ii * syzygies hold , but there are additional alignment conditions ( e.g. , @xmath19 are of type * ii * ) . the weyl tensor and the ricci tensor can be algebraically classified in a lorentzian spacetime of arbitrary dimensions by boost weight ( using alignment theory ) . a bivector formalism in higher dimensional lorentzian spacetimes has been developed to algebraically classify the weyl bivector operator . used in tandem with the boost weight decomposition , the algebraic classification of the weyl tensor and the ricci tensor ( based on their eigenbivector and eigenvalue structure , respectively ) can consequently be refined . the purpose of this paper is the determination of necessary conditions for the algebraic type of a higher dimensional weyl tensor or ricci tensor , and particularly type * ii * ( or * d * ) , using scalar curvature invariants . for a tensor of a particular algebraic type , the associated operator @xcite will have a restricted eigenvector structure . for a given curvature operator in arbitrary dimensions , we can consider the eigenvalues of this operator to obtain necessary conditions . in principle , the analysis can be used to study the various subclasses in more detail . in particular , requiring the algebraic type to be * ii * or * d * will impose restrictions on the eigenvalues on the operator . in this paper we shall present an analysis of the discriminants of the associated characteristic equation to determine the conditions on a tensor for a given algebraic type . since the characteristic equation has coefficients which can be expressed in terms of the scalar polynomial curvature invariants of the operator , we can consequently give conditions on the eigenvalue structure expressed manifestly in terms of these polynomial curvature invariants . we will describe the algorithm which will enable us to completely determine the eigenvalue structure of the curvature , up to degeneracies , in terms of a set of discriminants @xmath20 . the resulting syzygies ( discriminants ) can then be written as special scalar polynomial invariants . in particular , we use the technique to study the necessary conditions in arbitrary dimensions for the weyl and ricci curvature operators ( and hence the higher dimensional weyl and ricci tensors ) to be of algebraic type * ii * or * d*. we are consequently able to determine the necessary condition(s ) in terms of simple scalar polynomial curvature invariant for the higher dimensional weyl and ricci tensors to be of type * ii * or * d*. we explicitly display the scalar polynomial curvature invariants for a weyl or ricci tensor to be of type * ii * ( or * d * ) in 5d . a number of specific results are obtained , which are summarized at the end of the paper . in addition , a number of simple examples are presented , including einstein spaces , the 5d schwarzschild spacetime , and 5d space with complex hyperbolic sections . we will also present a detailed analysis of the important example of a 5d rotating black ring @xcite which is generally of type @xmath21 , but can also be of type @xmath22 or @xmath23 . this example serves to illustrate the calculational method and the power of the approach . in particular , we shall show that the rotating black ring is of type @xmath22 ( or type @xmath23 ) on the black hole horizon , by showing that a number of discriminants ( the @xmath24 invariants ) vanish there . we briefly discuss the utility of using these methods to study classification problems that also involve differential scalar polynomial curvature invariants constructed from the riemann tensor and its covariant derivatives , and present a simple illustration . we also make some brief comments on possible future work . in the appendices we review the weyl bivector operator ( particularly for type * ii * or * d * ) and present some important discriminants ( or syzygies ) that are used in the paper . we can use an analysis of the discriminants of the associated characteristic equation to determine the conditions on a tensor for a given algebraic type . in particular , we shall seek necessary conditions for a higher dimensional weyl tensor or ricci tensor to be type * ii * or * d*. in principle , the analysis can be used to study the various subclasses in more detail . for a tensor of a particular algebraic type , the associated operator ( acting on a vector space of dimension @xmath25 ) , non - capitalized latin indices run over @xmath26 , @xmath25 is the dimension of the vector space ( or the order of the associated characteristic equation for the eigenvalues ) of the curvature operator ( for the ricci curvature tensor @xmath27 and for the weyl curvature tensor @xmath28 ) , and capitalized latin letters are used to denote bivector indices . ] will have a restricted eigenvector structure . for a given curvature operator , @xmath29 , we can consider the eigenvalues of this operator to obtain necessary conditions . in particular , requiring the algebraic type to be * ii * or * d * ( * ii*/*d * ) will impose restrictions of the eigenvalues on the operator ( e.g. , , the eigenvalue type ( ` segre type' ) will have to be of a particular form ) . crucial in this discussion is the eigenvalue equation or characteristic equation @xcite : ( r-)=0 . this equation is a polynomial equation in @xmath30 and the eigenvalues are the roots of this equation . since the characteristic equation has coefficients which can be expressed in terms of the invariants of @xmath29 , we can give conditions of the eigenvalue structure expressed manifestly in terms of the invariants of @xmath29 . since the invariants of @xmath29 are polynomial curvature invariants of spacetime , these conditions will be referred to syzygies . henceforth , we will describe an algorithm which enables us to completely determine the eigenvalue structure of @xmath29 using the invariants @xmath31 , up to degeneracies which will be explained later . , we do not need to consider bianchi identities or dimensional dependent identities to simplify the resulting polynomial expressions obtained . ] the characteristic equation can be expanded to a polynomial equation : f()=(-r)=a_0^n+a_1^n-1+ a_i^n - i+ +a_n . [ poly]in our case , the coefficients are expressed in terms of invariants of @xmath29 , using newton s identities . however , the algorithm which follows applies to any polynomial equation . defining the polynomial invariants , since this is how it will be presented in maple expressions . we have also omitted any numerical coefficients in the definitions of the @xmath32 for convenience here ( see later ) . ] ( r ) , ( r^2 ) , ( r^3 ) , , we can generally write the coefficients @xmath33 as a determinant of an @xmath34 matrix as follows : a_i= _ r_1 _ & 1 & 0 & & 0 + _ r_2 _ & _ r_1 _ & 2 & & + _ r_3 _ & _ r_2 _ & _ r_1 _ & & 0 + & & & & ( i-1 ) + _ r_i _ & & _ r_3 _ & _ r_2 _ & _ r_1 _ , [ defai]where @xmath35 ( and @xmath36 ) . explicitly , the first six are given by : a_0&= & 1 , + a_1&= & -_r_1 _ , + a_2&=&12_r_1_^2 - 12_r_2 _ , + a_3&= & -16 _ r_1_^3 + 12 _ r_2 r_1 _ -13 _ r_3 _ , + a_4&=&^4 - 14 _ r_2 r_1_^2 + 13 _ r_3 r_1 _ + 18 _ r_2_^2 - 14 _ r_4 _ , + a_5&= & -^5 + 112 _ r_2 r_1_^3 - 16 _ r_3 r_1_^2 - 18 _ r_1 r_2_^2 + 14_r_4 r_1 _ + & & + 16 _ r_2 r_3 _ -15 _ r_5_. also note that the order of @xmath33 is @xmath37 ; i.e. , @xmath38 . , @xmath39 , where @xmath40 and the expressions above simplify ; e.g. , @xmath41 . ] note that if the invariant of highest order , @xmath42 , is zero , i.e. , @xmath43 , then the eigenvalue equation trivially factorises and we have a zero eigenvalue . therefore , it is convenient to first check the existence of zero - eigenvalues . in particular , if @xmath44 then there exists a zero eigenvalue of multiplicity @xmath45 . if this the case then the polynomial factorises trivally and the order can be reduced . the following procedure can then be simplified accordingly . in general the polynomial , eq.([poly ] ) , can be analysed and criteria for the various segre types can be given . the resulting syzygies are special polynomial invariants which can be used to characterise the various eigenvalue cases ; i.e. , they are _ discriminants_. a complete set of discriminats can be algorithmically found and in the following we will give the algorithm which is found in @xcite ( see also @xcite ) . the resulting discriminants will be denoted @xmath20 , @xmath46 , @xmath47 etc . , where @xmath25 denotes order of the polynomial , and @xmath37 is a running index . these discriminants can be given in terms of the coefficients @xmath33 ; however , using newton s identities we can express them explicitely in terms of the polynomial invariants @xmath48 , @xmath49 , etc . given the polynomial @xmath50 we define the @xmath51 discrimination matrix @xmath52 : a_0 & a_1 & a_2 & & a_n & 0 & & 0 & 0 + 0 & na_0 & ( n-1)a_1 & & a_n-1 & 0 & & 0 & 0 + 0 & a_0 & a_1 & & a_n-1&a_n & & 0 & 0 + 0 & 0 & na_0 & & 2a_n-2 & a_n-1 & & 0 & 0 + & & & & & & & & + 0 & 0 & & 0 & na_0 & ( n-1)a_1 & & a_n-1 & 0 + 0 & 0 & & 0 & a_0 & a_1 & & a_n-1 & a_n consider now the principal minor series , @xmath53 defined as the determinants : d_k= disc(f ) + + let @xmath54 , @xmath55 , then the _ discriminant sequence _ of the polynomial @xmath56 is given by \{^nd_1,^nd_2,^nd_3, ... ,^nd_n}. by expressing the @xmath20 in terms of the curvature invariants , @xmath48 , @xmath49 , etc , we can obtain the _ primary syzygies _ @xmath20 for the operator @xmath29 . note that the order of @xmath57 is @xmath58 . , @xmath59 , denote the descriminants of the 5d weyl curvature operator . ] [ [ sign - list . ] ] sign list . + + + + + + + + + + we call @xmath60 $ ] , where @xmath61 the _ sign list _ of the sequence @xmath62 . [ [ revised - sign - list . ] ] revised sign list . + + + + + + + + + + + + + + + + + + given a sign list @xmath63 $ ] . if this contains any `` internal zeros '' , i.e. , if there is a subsequence @xmath64 $ ] , where @xmath65 and @xmath66 , then we replace this subsequence with : @xmath67.\ ] ] the revised sign list will therefore contain no `` internal '' zeros , but may have zeros at the end . the revised sign list will give us the number of distinct real and complex roots . [ [ number - of - real - and - complex - roots . ] ] number of real and complex roots . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + consider the revised sign list of @xmath62 . let : @xmath68 of the revised sign list . then for @xmath56 : * the number of distinct _ pairs of complex conjugate roots _ is @xmath69 ; and * the number of distinct _ real roots _ is @xmath70 . if we are not interested in the multiplicities of the eigenvalues , the discriminant sequence , @xmath62 is sufficient . in some cases , this is enough to determine the eigenvalue structure of @xmath29 , but not always . [ [ example - trace - free - ricci - tensor - in-3d . ] ] example : trace - free ricci tensor in 3d . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + let @xmath29 be the 3-dimensional tracefree ricci tensor ; i.e. , = ( s^ _ ) , s^ _ = r^ _ -13 r^ _ . this implies , @xmath71 consequently , @xmath72 using the procedure above , we get the discriminants : ^3d_1 & = & 3 , + ^3d_2 & = & -6a_2=3s^ _ s^ _ , + ^3d_3 & = & -4a_2 ^ 3 - 27a_3 ^ 2=12(s^ _ s^ _ ) ^3 - 3(s^ _ s^ _ s^ _ ) ^2 . we clearly have @xmath73 . the possible signs of the discriminants @xmath74 and @xmath75 can now be used to determine the number of real / complex eigenvalues of @xmath76 . 1 . @xmath77 , @xmath78 : 3 distinct real eigenvalues . @xmath77 , @xmath79 : 2 pairs of complex eigenvalues , which is impossible . @xmath80 : 1 real and 2 complex eigenvalues . @xmath81 , @xmath78 : 2 real eigenvalues ( one of them must be of muliplicity 2 ) . @xmath81 , @xmath82 : 2 complex eigenvalues ( impossible , since the last eigenvalue must be real and hence , distinct ) . 6 . @xmath81 , @xmath83 : 1 real eigenvalue ( which must be equal to zero since @xmath76 is tracefree ) . note that in terms of the segre type , a ( tracefree ) ricci type * ii*/*d * is of type @xmath84 , @xmath85 , @xmath86 , or simpler . consequently , if the ( tracefree ) ricci is of type * ii*/*d * , or simpler , then @xmath81 . [ [ multiple - factor - sequence . ] ] multiple factor sequence . + + + + + + + + + + + + + + + + + + + + + + + + + we note that for polynomials of order 4 and more , the discriminants @xmath20 may not be sufficient to determine the complete eigenvalue structure . for example , for quartics , if we have 2 distinct real roots , then we can not , using @xmath20 only , distinguish the cases @xmath87 and @xmath88 . therefore , we need to go a step further in order to distinguish these cases . consider the discriminant matrix @xmath52 . define the submatrices : m(k , l ) disc(f ) + + then , construct the polynomials : _ k(f)=_i=0^kx^k - i , for @xmath89 . the sequence @xmath90 is called the multiple factor sequence of @xmath56 due to the following result @xcite : if the number of zeros in the revised list of the discriminant sequence of @xmath56 is @xmath4 , then @xmath91 . the greatest common devisor ( g.c.d . ) of @xmath56 and @xmath92 is thus always in the multiple factor sequence . indeed , the polynomial @xmath93 is the repeated part of @xmath56 , because if @xmath94 has @xmath4 real roots of multiplicities @xmath95 , @xmath96 , ... , @xmath97 , and @xmath98 has @xmath99 distinct real roots , then @xmath98 has @xmath4 real roots of multiplicities @xmath100 , @xmath101 , ... , @xmath102 , and @xmath103 simplie real roots ( similar arguement for complex roots ) . therefore , by considering @xmath104 _ we reduce the multiplicities of all the roots by 1_. we can now consider the discriminants of the polynomial @xmath104 in the same way as we computed the discriminant sequence of @xmath98 . we will call the discriminant sequence of @xmath104 for @xmath105 . we can now use these to determine the sign list of the @xmath106-sequence , etc . we can repeat this procedure and consider @xmath107 , @xmath108 etc . these have the relation : ^j-1(f)&&(x-_1)^n_1 + 1(x-_2)^n_2 + 1 ... (x-_k)^n_k+1(x-_k+1) ... (x-_m ) , + ^j(f)&&(x-_1)^n_1(x-_2)^n_2 ... (x-_k)^n_k , this gives us the following algorithm for determining the root structure ( or eigenvalue structure ) @xcite : [ [ algorithm - for - root - classification ] ] algorithm for root classification + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + 1 . find the discriminant sequence of @xmath56 : @xmath109 and the revised sign list . find the number of distinct roots by counting sign changes and non - zero elements of the revised sign list . if the revised sign list contains no 0 s , stop . 2 . if the revised sign list contains @xmath4 zeros , then compute the @xmath110 by the definition for the multiple factor sequence . then repeat step 1 for @xmath104 . 3 . continue considering the multiple factor sequence @xmath111 , @xmath108 , ... , until for some @xmath112 , the revised sign list of @xmath113 contains no zeros . we now compute the number of real / complex distinct roots of @xmath113 . we can now determine the roots and multiplicities of @xmath114 , which again enables us to determine the roots and multiplicites of @xmath115 etc . at the end of this process , we have a complete root classification for @xmath56 . note that this procedure will provide us with the discriminants ( or syzygies ) which gives us a complete eigenvalue classification of _ any _ operator @xmath29 . as explained , these discriminants can be expressed in terms of polynomials of the invariants , @xmath116 , @xmath117 , @xmath118 etc . , of @xmath29 . in principle , we can use this method to study the necessary conditions on any curvature operator of any specific eigenvalue type . in particular , we can use the technique to study the necessary conditions of the weyl and ricci curvature operators for it to be of algebraic type * ii*/*d*. we note that the condition @xmath119 will signal a double eigenvalue since the number of eigenvalues is maximum @xmath120 . if @xmath121 also , then we have maximum @xmath122 eigenvalues , etc . we can utilise this to create syzygies which are necessary for the special algebraic type to be fulfilled . types * g * and * i * are both of equal generality with respect to their possible ( eigenbivector / eigenvalue ) roots structure . now , for a tensor to be of type * ii * ( or * d * ) then the eigenvalues of the corresponding operator need to be of a special form . since the invariants of a type * ii * are the same as for type * d * , we will assume type * d*. the type * d * case possesses an important symmetry , namely a boost isotropy ( the tensor , not necessarily the complete spacetime ) . this is what we will utilise in the following . this implies specific structure for ( a ) particular discriminant(s ) , which then gives rise to necessary condition(s ) in terms of scalar polynomial curvature invariants . for the ricci tensor , we note that a type * d * tensor is of segre type @xmath123 , or simpler . this implies that the _ ricci operator has at least one eigenvalue of ( at least ) multiplicity 2 . furthermore , all the eigenvalues are real . _ for the weyl tensor in @xmath1 dimensions we can use the bivector operator in @xcite where the canonical form of a weyl type * d * tensor is given ( see also appendix ) in particular , for type * d * , @xmath124 where @xmath125 is a @xmath126 matrix and @xmath127 is a square matrix ( see appendix for the explicit form of this matrix ) . since the eigenvalues of @xmath125 and @xmath128 are the same , we have that the _ weyl operator has at least @xmath129 eigenvalues of ( at least ) multiplicity 2 . _ these observations connect the algebraic types to the eigenvalue structure and enables us to construct the necessary discriminants . applying the condition @xmath130 for the complex three dimensional weyl tensor we obtain the complex syzygy @xmath11 . this is equivalent to the ( 12th order ) real syzygies given by eqns . ( [ weyl1])-([weyl2 ] ) from the six dimensional system with @xmath131 and @xmath132 . applying the condition @xmath133 for the four dimensional trace - free ricci tensor we obtain the ( 12th order ) syzygy given by eqn . ( [ rictypeii2 ] ) . ) - ( [ weyl2 ] ) and ( [ rictypeii2 ] ) are different ; for example , @xmath134 ( i.e. , @xmath135 , etc . , in ( [ weyl2 ] ) ) . ] we can also apply these conditions to the full riemann tensor ( to be of type * ii*/*d * , which implies both the weyl and ricci tensor are of type * ii*/*d * and aligned ) . alternatively , we note that these will also give us syzygies for mixed tensors . for example requiring that riemann tensor is type * ii*/*d * , implies that both ricci and weyl is type * ii*/*d * , but also mixed tensors , like : @xmath136 the type * ii*/*d * condition therefore implies that we have the syzygy @xmath137 for _ all of _ @xmath138 , @xmath139 and @xmath140 . for the trace - free ricci tensor , we note that type * d * has to be of segre type @xmath141 or simpler . this implies that 2 eigenvalues are equal , while the remaining has to be real . therefore , we get the necessary ( 20th order ) syzygy for the trace - free ricci tensor to be of type * ii*/*d * : @xmath142 [ [ result ] ] result : + + + + + + + _ the necessary condition for the trace - free ricci tensor , s , to be of algebraic type * ii * ( or * d * ) in 5d is that the discriminant @xmath143 is zero , so that the related scalar polynomial curvature invariant @xmath144_. is given explicitly in appendix b.3 . ] for the weyl tensor , we consider the bivector operator @xmath145 . since the bivector space is 10-dimensional , we get a condition involving a syzygy of order 90 ! in particular , the type * ii * operator has 3 eigenvalues of ( at least ) multiplicity 2 . therefore , we get the syzygies : @xmath146 since these polynomial invariants are of particular importance , we will denote them by @xmath147 , the @xmath24 weyl invariants . [ [ result-1 ] ] result : + + + + + + + _ the necessary condition for the weyl tensor to be of type * ii * ( or * d * ) in 5d is that the scalar polynomial curvature invariants @xmath148 . _ these are syzygies of order 90 , 72 and 56 , respectively . in principle , these can be computed , in appendix b.4 . ] but its probably more useful to consider specific metrics . a necessary condition can also be found from considering combinations of the weyl tensor ; for example , the operator @xmath149 . this gives again @xmath150 note that @xmath151 is now a 40th order syzygy ( in the weyl tensor ; a 20th order syzygy in the square of the weyl tensor ) . , which perhaps suggests that an appropriate algebraic combination of @xmath152 and @xmath153 either simplifies or factors . we shall return to this in future work . ] we have stressed that the conditions determined are necessary conditions . indeed , these conditions may not be sufficient . the reason for this is that the characteristic equation for different algebraic types may be identical and consequently the scalar invariants are also identical . for example , its not possible to distingush the segre types @xmath141 ( of ricci type * d * ) and @xmath154 ( of ricci type * i * ) . this implies also that the @xmath24 conditions may be fulfilled in spite of the fact that the spacetime is of type * g * or * i*. an explicit example of this is the following result : assume a 5d spacetime has a weyl tensor with @xmath155 isotropy . then it fulfills the @xmath24 syzygies ; i.e. , @xmath148 . this result can be seen from using the bivector operator and imposing the @xmath155-symmetry . one then sees that this forces that there must be 3 pairs of equal eigenvalues , which implies that there are maximum 7 distinct eigenvalues ; hence , the result follows . this _ degeneracy _ in the classification is a fundamental problem when considering scalar invariants only . sometimes these cases can be resolved by considering other invariants ; however , there is no guarantee that this can be achieved . using the invariants only , we can only determine the _ eigenvalue type _ of the operator . for example , if we find the eigenvalue type to be @xmath156 , then this can correspond to three ricci types : @xmath141 , @xmath154 , and @xmath157 because all of these ricci types have the same eigenvalue type . this is a _ fundamental degeneracy _ in the classification of tensors using scalar invariants only and has been discussed in earlier papers @xcite . therefore , we need to keep this in mind when using the discriminants . note that this degeneracy is a _ discrete _ degeneracy , unlike the notion of @xmath6-degenerate metrics @xcite which require a _ continuous _ deformation . in higher dimensions we will obtain similar syzygies for type * ii*/*d * tensors . in @xmath25 dimensions , the ricci and weyl type * ii*/*d * conditions are the corresponding syzygies ( @xmath158 : & & ^nd_n=0 , + & & ^md_m= ^md_m-1= ... =^md_m - n+2=0 . note that the ricci syzygy is of order @xmath159 , while the highest weyl syzygy is of order @xmath160 . [ [ the - mathcalcmathcalhmathcalp - conditions - are - non - trivial ] ] the @xmath24 conditions are non - trivial : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + let us consider a simple metric which shows the @xmath161 conditions are non - trivial conditions . consider the solvmanifold : s^2=-t^2+e^2p_1 tx^2+e^2p_2 ty^2+e^2p_3 tz^2+e^2p_4 tw^2 the computation for this metric is a bit lengthy ( even for maple ) ; however , by choosing randomly some values of the parameters , for example , @xmath162 , @xmath163 , @xmath164 , and @xmath165 , maple quickly calculates the values of the discriminants for the weyl operator , obtaining : ^10d_10>0 , ^10d_9>0 , ^10d_8>0 , ... , ^10d_2>0 , showing that this metric is not type * ii * or * d * ( and that the @xmath24 invariants are not trivial ) . [ [ einstein - spaces ] ] einstein spaces + + + + + + + + + + + + + + + in an einstein space all of @xmath166 , and hence all of the discriminants for the trace - free ricci eigenvalue equation are zero , and the only non - trivial scalar invariant is the ricci scalar @xmath167 . [ [ d - schwarzschild - spacetime ] ] 5d schwarzschild spacetime : + + + + + + + + + + + + + + + + + + + + + + + + + + + for the weyl operator @xmath145 we get @xmath168 this implies that the weyl operator has 3 distinct real eigenvalues which agrees with the results of @xcite . in fact , this spacetime is of type * d*. [ [ d - space - with - complex - hyperbolic - sections . ] ] 5d space with complex hyperbolic sections . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + let us consider the example in @xcite with complex hyperbolic spatial sections : s^2= -t^2+a(t)^2e^-2w(x+12(yz - zy))^2 & & + + e^-w(y^2+z^2)+w^2 . & & [ hc2 ] for the weyl operator @xmath145 we get @xmath168 this again implies that the weyl operator has 3 distinct real eigenvalues , like the 5d schwarzschild spacetime . however , unlike the schwarzschild case , we note that the some of the invariants @xmath33 , defined in eq.([defai ] ) , are zero : @xmath169 as explained , this is a signal that there is a zero - eigenvalue of multiplicity 7 ! hence , since there are 3 distrinct eigenvalues of which one of must be zero with multiplicity 7 , the eigenvalue structure is @xmath170 . in particular , since the weyl operator is trace - free , we explicitly get eigenvalues : @xmath171 , \quad \lambda~[\times 2 ] , \quad -2\lambda,\ ] ] which agrees with the results of @xcite . however , this spacetime is not of type * ii*/*d * ; indeed , it is @xmath6-non - degenerate . this can be seen by computing the operator @xmath149 which is of `` segre '' type @xmath172 . however , due to the fundamental degeneracy in the eigenvalue type , we need to compute differential invariants to delineate this case completely . the 5d rotating black ring @xcite is generally of type @xmath21 , but can also be of type @xmath22 or @xmath23 at different locations and for particular values of the parameters @xmath173 . assuming that the form of the metric is given by eqn . ( 9 ) in @xcite ( in terms of the parameters @xmath173 , where @xmath174 has been set to unity ) , we consider the coordinate ranges @xmath175 and @xmath176 ( and hence @xmath177 and @xmath178 ) , corresponding to the regions @xmath179 in @xcite in order to retain the correct ( lorentzian ) signature . we consider the algebraic type of the 5d weyl tensor . calculating the polynomial invariants @xmath180 and evaluating at the ` target ' point @xmath181 and @xmath182 in the region under consideration , and @xmath183 we obtain flat space in this region , but this case will not be included here . ] all of the @xmath32 and hence all of the resulting discriminants are functions of the parameters @xmath173 only . then , at the ` target ' point , in general the metric is of type @xmath21 , the case @xmath184 corresponds to the myers - perry metric ( type @xmath23 ) , @xmath185 corresponds to the black hole horizon ( @xmath186 , type @xmath22 ) , @xmath187 corresponds to the static subcase , and @xmath188 corresponds to a curvature singularity . let us first consider the trace - free part operator @xmath149 , which gives us the discriminant : _ t^5d_5=f ( , ) where @xmath189 is a polynomial which is generically not zero . on the horizon @xmath190 , we see that @xmath191 , and computing @xmath192 we get @xmath193 except for special values of @xmath30 . this is a signal that the metric is of type * ii * on the horizon . indeed , at the horizon , @xmath190 , the computation simplifies and we can compute the @xmath161 invariants : @xmath194 while : _ w^10d_7 & & + & & ( - 1)^2 ( + 2)^2 ( + 1)^2 ( 4 ^2 - - 6)^2 ( 2 ^2 - - 4)^2 + & & ( 4 ^3 - 4 ^2 - 3 + 6)^2 ( 2 ^4 - 12 ^3 + 9 ^2 + 16 - 24)^2 + & & ( 4 ^4 - 14 ^2 + + 12)^2 ( ^3 - 2 ^2 + + 6)^2 ( 2 + 1)^6 , where a ( postive ) numerical factor has been ignored . since , the @xmath161 syzygies are satisfied , this gives further evidence that the the metric is of type * ii * on the horizon . note that we actually get further contraints from the secondary discriminants , as can be seen from the table in appendix [ dim10 ] . indeed , by calculating the secondary discriminant @xmath195 on the horizon , we get : _ w^10f_3 & & , which we see is non - zero as long as @xmath196 . consequently , as long as @xmath196 , then the eigenvalue type is @xmath197 . this is consitent with type * ii*. another interesting special case is @xmath184 ( myers - perry ) , for which both @xmath198 , and : @xmath199 we can also here compute the @xmath161 invariants , which are all zero . [ [ note ] ] note : + + + + + we note that @xmath151 is a 40th order syzygy in the weyl tensor . therefore , a useful strategy in practical computations ( for example , determining the algebraic type of a 5d weyl tensor ) , as illustrated by this example , might be to test for necessicity using an operator like t , which is relatively simple . if the syzygy is not satisfied we have a definitive result . it is possible that the syzygy can only be satified for certain coordinate values ( or parameter values ) , whence the @xmath161 syzygies can be tested in these simpler particular cases . for a curvature tensor of a particular algebraic type , the associated operator will have a restricted eigenvector structure . for a given curvature operator in arbitrary dimensions , we can thus consider the eigenvalues of this operator to obtain necessary conditions in order for the tensor to be of a particular algebraic type . in principle , this analysis can be used to study all of the various subclasses of particular algebraic types in more detail . in particular , requiring the algebraic type to be * ii * or * d * will impose useful restrictions . in this paper we have used an analysis of the discriminants of the associated characteristic equation to determine the conditions on a tensor for a given algebraic type . since the characteristic equation has coefficients which can be expressed in terms of the scalar polynomial curvature invariants of the operator , we can give conditions , or syzygies , on the eigenvalue structure expressed manifestly in terms of these polynomial scalar curvature invariants . indeed , we have described an algorithm which enables us to completely determine the eigenvalue structure of the curvature , up to degeneracies , in terms of a set of _ discriminants _ @xmath20 , @xmath200 , etc .. the resulting syzygies ( discriminants ) can then be written as special scalar polynomial invariants . in particular , we have used the technique to study the necessary conditions in arbitrary dimensions for the weyl and ricci curvature operators ( and hence the higher dimensional weyl and ricci tensors ) to be of algebraic type * ii*/*d * , and created syzygies which are necessary for the special algebraic type to be fulfilled . we are consequently able to determine the necessary conditions in terms of simple scalar polynomial curvature invariant for the higher dimensional weyl and ricci tensors to be of type * ii * or * d*. we have explicitly determined the scalar polynomial curvature invariants for a weyl or ricci tensor to be of type * ii * ( or * d * ) in 5d . this will be of considerable utility in classifying higher dimensional solutions obtained in supergravity or superstring theory @xcite or higher dimensional black hole solutions @xcite . a number of specific results have been obtained in this work . the necessary condition for the trace - free ricci tensor , @xmath201 , to be of algebraic type * ii * ( or * d * ) in 5d is that the discriminant @xmath143 is zero , so that the related scalar polynomial curvature invariant @xmath144 . the necessary condition for the weyl tensor to be of type * ii * ( or * d * ) in 5d is that the scalar polynomial curvature invariants @xmath148 . in principle , we can repeat a similar analysis for other algebraic types ( and making more use of the secondary discriminants ) . a number of simple examples were presented , including einstein spaces , the 5d schwarzschild spacetime , and 5d space with complex hyperbolic sections . in addition , a simple solvmanifold was considered to show that the @xmath24 conditions are non - trivial . we also presented a detailed analysis of the important example of a 5d rotating black ring @xcite which is generally of type @xmath21 , but can also be of type @xmath22 or @xmath23 for particular values of the parameters . this example serves to illustrate the calculational method and the power of the approach . in particular , we showed that the rotating black ring is of type @xmath22 ( or type @xmath23 ) on the black hole horizon ( @xmath186 ) , by showing that @xmath202 on the horizon ( and studying some of the secondary discriminants ) . the example also illustrates the utility in practical computations of employing rather more simple discriminants like @xmath151 ( which is a 40th order syzygy in the weyl tensor ) . in lorentzian spacetimes , identical metrics are often given in different coordinate systems , which disguises their true equivalence . perhaps the easiest way of distinguishing metrics is through their _ scalar polynomial curvature invariants _ , due to the fact that inequivalent invariants implies inequivalent metrics . in @xcite the notion of an @xmath6-non - degenerate spacetime metric in the class of 4d lorentzian manifolds , which implies that the spacetime metric is locally determined by its scalar polynomial curvature invariants , was introduced . by determining an appropriate set of projection operators from the riemann tensor and its covariant derivatives , it was proven that a 4d lorentzian spacetime metric is either _ @xmath6-non - degenerate _ or _ degenerate kundt _ @xcite . therefore , a metric that is not characterized by its curvature invariants must be of degenerate kundt form . these results were generalized to higher dimensions in @xcite . the @xmath6-non - degenerate theorem contains not only zeroth order invariants but also differential scalar polynomial curvature invariants constructed from the riemann tensor and its covariant derivatives . for example , if the spacetime is of weyl type * n * , then the differential invariants @xmath203 and @xmath204 vanish if the spacetime is degenerate kundt @xcite ( the definitions of the invariants @xmath203 and @xmath205 are given therein ) . similar results follow for weyl type * iii * spacetimes ( in terms of invariants @xmath206 and @xmath207 ) and in the conformally flat ( but non - vacuum ) case ( in terms of similar invariants @xmath203 and @xmath204 constructed from the ricci tensor @xcite . these conditions are _ necessary _ conditions in order for a spacetime not to be @xmath6-non - degenerate @xcite . in the case that @xmath208 ( weyl types * ii * or * d * ) , in @xcite two higher order invariants @xmath209 and @xmath210 were given as _ sufficient _ conditions for @xmath6-non - degeneracy ( if @xmath10 , but @xmath211 or @xmath212 , then the spacetime is @xmath6-non - degenerate ) . -non - degenerate , then essentially we can construct positive boost weight terms in the derivatives of the curvature and determine an appropriate set of differential scalar curvature invariants . ] this analysis in 4d can be repeated using discriminants . let us focus on the ricci tensor for illustrative purposes . the necessary condition for the ricci tensor to be of type * ii * or * d * is given by eqn . ( [ rictypeii2 ] ) which , as noted earlier , follows from a discrimant analysis . now , if we consider the covariant derivatives of the ricci tensor , @xmath213 , then for the spacetime to be @xmath6-non - degenerate each covariant derivative term must also be of type * ii * or * d*. hence we could study the eigenvalue structure of the operators associated with the @xmath213 and apply the type * ii*/*d * necessary conditions in turn . for example , considering the trace - free parts of the tensors @xmath214 , we obtain necessary conditions of the form of eqn . ( [ rictypeii2 ] ) but with the @xmath215 . this can be repeated for the weyl tensor and in higher dimensions @xcite . [ [ example . ] ] example . + + + + + + + + the class of vacuum type * d * spacetimes which are @xmath6-_non - degenerate _ , are invariantly classified by their scalar polynomial curvature invariants @xcite . for example , for the kinnersley class i type * d * vacuum spacetime @xcite ( the other cases work in a similar way ) , there are 4 algebraically independent ( complex ) cartan invariants , which can be written in terms of 4 independent ( complex ) scalar polynomial invariants ( e.g. , @xmath7 , and invariants such as @xmath216 , where @xmath217 which include first and second covariant derivatives ) . the schwarzschild vacuum type * d * spacetime belongs to the kinnersley class i and , as discussed in @xcite , all of the algebraically independent cartan scalars are related to the two functionally independent polynomial scalar curvature invariants @xmath218 and @xmath216 ( which are equal to @xmath219 and @xmath220 , respectively , as functions of the two parameters @xmath221 and @xmath125 in canonical coordinates ) @xcite . the kerr solution belongs to kinnersley class iia ; this spacetime has been invariantly characterized intrinsically @xcite . [ [ kerr - metric . ] ] kerr metric . + + + + + + + + + + + + for illustration let us consider the example of the kerr metric . the kerr metric is of petrov type * d * and we are thus interested in whether the covariant derivative , @xmath222 , is of type * d * or not . as noted above , the methods discussed earlier can also be used for differential invariants . the covariant derivative of the weyl tensor , @xmath223 , has no natural operator associated with it . however , we can , for example , consider the second order operator @xmath224 defined above . if @xmath225 is not of type * d*/*ii * , then @xmath222 can not be of type * d*/*ii * either . for the kerr metric , we obtain the syzygy : _ t^4d_4=f_1 ^ 2f_2 , where g_&=&r^44ar^3 - 6a^2r^2 ^ 24a^3r^3+a^4 ^ 4 , and @xmath226 and @xmath227 are some complicated polynomials . we note that away from the horizon , the ergosphere , and some other special points , this syzygy is non - zero and hence @xmath222 is not of type * d*/*ii * ( generically ) , outside the horizon . the kerr metric is therefore @xmath6-non - degenerate by the results of @xcite . recently , there has been considerable interest in black holes in more than four dimensions @xcite . while the study of black holes in higher dimensions was perhaps originally motivated by supergravity and string theory , now the physical properties of such black holes are of interest in their own right . indeed , studies have shown that even at the classical level gravity in higher dimensions exhibits much richer dynamics than in 4d , and one of the most remarkable features of higher dimensions is the non - uniqueness of black holes @xcite there now exist a number of different higher dimensional black hole solutions @xcite , including the rotating black rings @xcite , that are the subject of ongoing research in classical relativity and string theory . some of these new spacetimes have be classified algebraically @xcite . however , in order to make further progress it is absolutely crucial to be able to develop new techniques for solving the vacuum field equations in higher dimensions and to be able to comprehensively classify such solutions , and the algebraic techniques recently introduced @xcite will be of fundemental importance in this development . however , the algebraic techniques used to date now are rather difficult to apply , and the development of simpler criteria , including the use of necessary conditions in terms of scalar curvature invariants introduced here , will hopefully prove to be of great utility . therefore , the analysis presented in this paper will be of considerable importance for analysing higher dimensional black hole solutions @xcite ( and solutions in supergravity or superstring theory @xcite ) . indeed , the detailed analysis of the 5d rotating black ring @xcite serves to illustrate the power of the approach . in future work we hope to extend the analysis presented in this paper and further generalize it to the study of differential operators . in addition , we intend to discuss a number of other applications , including the algebraic classification of some other known higher dimensional black hole solutions . the main part of this work was done during a visit to dalhousie university april - june 2010 by sh . the work was supported by nserc of canada ( ac ) and by a leiv eirikson mobility grant from the research council of norway , project no : * 200910/v11 * ( sh ) . given a vector basis @xmath228 we can define a set of ( simple ) bivectors @xmath229}={\bf k}^{\mu}\wedge{\bf k}^{\nu},\ ] ] spanning the space of antisymmetric tensors of rank 2 . consider a @xmath230-dimensional lorentzian space with the following null - frame @xmath231 so that the metric is @xmath232 let us consider the following bivector basis : @xmath233 or for short : @xmath234 $ ] , @xmath235 $ ] , @xmath236 $ ] , @xmath237 $ ] . the lorentz metric also induces a metric , @xmath238 , in bivector space . if @xmath239 , then @xmath240 where @xmath241 , and @xmath242 are the unit matrices of size @xmath243 and @xmath244 , respectively , and we have assumed the bivector basis is in the order given above . this metric can then be used to raise and lower bivector indices . let @xmath245 be the vector space of bivectors at a point @xmath246 . then consider an operator @xmath247 . we will assume that it is symmetric in the sense that @xmath248 . with these assumptions , the operator @xmath145 can be written in the following @xmath249-block form @xcite : [ weyloperator ] c= the eigenbivector problem can now be formulated as follows . a bivector @xmath251 is an eigenbivector of @xmath145 if and only if @xmath252 such eigenbivectors can now be determined using standard results from linear algebra . the lorentz transformations ( boosts , spins and null rotations ) in @xmath253-dimensions are explicitely written down in @xcite . ^i _ j = c_0i0j , & & ^i _ j = c_1i1j , + ^i _ b = c_0ijk , & & ^i _ b = c_1ijk , + ^i = c_010i , & & ^i =- c_011i , + m^i _ j = c_1i0j , & & = c_0101 , + a^b = c_01ij , & & |h^b _ c = c_ijkl . the weyl tensor is also traceless and obeys the bianchi identity : @xmath255 these conditions translate into conditions on our block matrices . we can consider each boost weight in turn , and use this to express these matrices into irreducible representations of the spins @xcite . here we have c_0101=c_0i1^i , c_0i1j=-12c_ikj^k+12c_01ij , c_i(jkl)=0 . starting with the latter , this means that the matrix @xmath256 fulfills the reduced bianchi identies . it is also symmetric which means that it has the same symmetries as an @xmath25-dimensional riemann tensor . hence , we can split this into irreducible parts over @xmath257 using the `` weyl tensor '' , `` trace - free ricci '' and `` ricci scalar '' as follows ( @xmath258 ) : m_ij&=&-|r_ij-12 |s_ij-12a_ij + & = & -12|r . this means that the boost weight 0 components can be specified using the irreducible compositions above ( @xmath259 ) . we note that in lower dimensions we have the special cases for the @xmath25-dimensional riemann tensor : ( i ) dim 4 ( @xmath260 ) : @xmath261 , ( ii ) dim 5 ( @xmath262 ) : @xmath263 , ( iii ) dim 6 ( @xmath264 ) : @xmath265 , where @xmath266 and @xmath267 are the self - dual , and the anti - self - dual parts of the weyl tensor , respectively . the same can be done with the antisymmetic tensor @xmath268 . a spin @xmath269 acts as follows on the various matrices : ( m,,a,|h)(gmg^-1,,|ga,|g|h|g^-1 ) . if @xmath270 is the weyl tensor , the type * d * case is therefore completely characterised in terms of a @xmath25-dimensional ricci tensor , a weyl tensor , and an antisymmetric tensor @xmath271 therefore , the spins are first used to diagonalise the `` ricci tensor '' @xmath272 . this matrix can then be described in terms of the segre - like notation corresponding to its eigenvalues . there is a degeneracy in the eigenvalues which occurs when two , or more , eigenvalues are equal . using a segre - like notation , we therefore get the types for @xmath273 : \{1111 .. } , \{(11)11 .. } , \{(11)(11 ) ... } , , \{0111 .. } , \{0(11)1 .. } , \{00(11 ) ... } , , + where a zero indicates a zero - eigenvalue . regarding the antisymmetric matrix @xmath271 , this must be of even rank and can be put into canonical block - diagonal form , and we can characterise an antisymmetric matrix using the rank . the antisymmetric matrix @xmath274 may also possess further degeneracies . finally , characterisation of the `` weyl tensor '' @xmath275 reduces to characterisating the weyl tensor of the corresponding fictitious @xmath25-dimensional riemannian manifold . let us consider the classification in @xcite , and investigate the different algebraic types in turn . in general , there will be algebraically special cases of type * g*. the type * i * , * iii * and * n * s were delineated in @xcite . it is of interest to explicitly review the type * ii*/*d * s here . then there will be algebraic subcases according to whether some of the irreducible components of boost weight 0 are zero or not . a complete characterisation of all such subcases is very involved in its full generality . however , a rough classification in terms of the vanishing the irreducible components under spins can be made : ( a ) type * ii*/*d*(a ) : @xmath277 , ( b ) type * ii*/*d*(b ) : @xmath278 , ( c ) type * ii*/*d*(c ) : @xmath263 . note that we can also have a combination of these ; for example , type * ii*(ac ) , which means that @xmath277 and @xmath279 ( i.e. , further algebraically special subcases can arise ) . in 4d , the weyl operator can always be put into type i form by using a null rotation ( hence , @xmath280 ) . furthermore , the irreducible representations under the spins are : @xmath281 . utilizing the unused freedom of one spin , one boost and two null - rotations , in each of the algebraically special cases we can use these to simplify the weyl tensor even further . ; consequently , the weyl operator @xmath145 has eigenvalues : _ 1,2=_3,4=-14(|r2ia_34 ) , _ 5,6=12(|r2ia_34 ) . we note that this is in agreement with the standard type * d * analysis in 4d ( see @xcite ) . the type * d * case is boost invariant , and also invariant under spins , consequently the isotropy is 2-dimensional . the two subcases @xmath284 and @xmath285 ( type * d*(a ) and * d*(b ) , respectively ) are in 4d referred to the purely `` electric '' and `` magnetic '' cases , respectively . in 4d , there is a duality relation , @xmath286 , which interchanges these two cases ; i.e. , @xmath287 interchanges the electic and magnetic parts . the 5d case is considerably more difficult than the 4d case . the complexity drastically increases and hence the number of special cases also increases . however , the 5d case is still managable and some simplifications occur ( compared to the general case ) . most notably , @xmath263 , and @xmath288 can be written , using a matrix @xmath289 , as follows ( similarly for @xmath290 ) : ^i _ jk=_jkl^li , where the conditions on @xmath291 imply that @xmath292 is symmetric and trace - free . therefore , we can use the spins to diagonalise @xmath292 . thus the general case is @xmath293 ( all eigenvalues different ) , with the special cases @xmath294 , @xmath295 and @xmath296 . furthermore , in the general case , the vector @xmath297 needs not be aligned with the eigenvectors of @xmath292 . there would consequently be special cases where @xmath297 is an eigenvector of @xmath292 . the components in 5d are displayed in table 1 . for type * d * we have that @xmath298 , and hence @xmath299 . the simplest of the three syzygies , the discriminant @xmath153 ( which is of 56th order in terms of @xmath302 , and contains 13377 terms ) is too lengthy to write down explicitly here , is given in @xcite . ] but it has the symbolic form : & = & ( _ s_6s_8_-_s_7_^2)_s_2_^21 + ( ) _ s_2_^20 + + & & + & & + + + + & & + & & + ( ) _ s_10_^4+(10 _ s_4 s_2 _ - 10 _ s_3_^2 - _ s_2_^3 ) _ s_10_^5 . a. coley , s. hervik and n. pelavas , 2009 , class . * 26 * , 125011 [ arxiv:0904.4877 ] ; a. coley , s. hervik and n. pelavas , 2006 , class . * 23 * , 3053 ; a. coley , s. hervik and n. pelavas , 2008 , class . * 25 * , 025008 . a. coley , r. milson , v. pravda and a. pravdova , 2004 , class . * 21 * , l35 [ gr - qc/0401008 ] ; r. milson , a. coley , v. pravda and a. pravdova , 2005 , int . j. geom . * 2 * , 41 ; a. coley , 2008 , class . quant . grav . * 25 * , 033001 [ arxiv:0710.1598 ] . the form of the metric considered here is given in : v. pravda and a. pravdova , 2005 , gen . . grav . * 37 * , 1277 [ gr - qc/0501003 ] ; the original reference is : r. emparan and h. s. reall , 2002 , phys . letts . * 88 * 101101 . l. yang , x.r . hou , and z.b . zeng , 1996 , sci . china series * e * * 39 * , 628 ; l. yang and b. xia , _ explicit criterion to determine the number of positive roots of a polynomial _ , 1997 , mm research preprints , no.15 , 134 .

we give a new proof of the cobham s first theorem using ideas from symbolic dynamics and of the cobham - semenov theorem ( in the primitive case ) using ideas from tiling dynamics .

0712.4337

given a subset @xmath0 of @xmath1 can we find an elementary algorithm which accepts the elements of @xmath0 and rejects those that do not belong to e ? by `` elementary algorithm '' we mean a finite state automaton . this question originates from the work of bchi ( 1960 , @xcite ) . cobham gave two answers to this question . in 1969 he proved that the existence of such an algorithm deeply depends on the numeration base , more precisely : * cobham s first theorem . * _ let @xmath2 and @xmath3 be two multiplicatively independent integers greater or equal to @xmath4 . then , a set @xmath5 is both @xmath2-recognizable and @xmath3-recognizable if and only if @xmath0 is the finite union of arithmetic progressions . _ here `` @xmath2-recognizable '' means that there exists an automaton which accepts exactly the language consisting of the expansions in base @xmath2 of the elements of @xmath0 , and @xmath2 and @xmath3 are * multiplicatively independent * if whenever `` @xmath6 for some non - negative integers @xmath7 and @xmath8 '' then @xmath9 . note that finite sets are finite unions of arithmetic progressions ( of length 1 ) and hence are @xmath2-recognizable for all @xmath2 . for example , we will see that the set @xmath10 is @xmath4-recognizable and , as it is not a finite union of arithmetic progressions , it can not be 3-recognizable . cobham s theorem implies that the set @xmath11 is @xmath2-recognizable for any @xmath12 . but it does not tell us anything about the structure of recognizable sets of integers . cobham gave a second answer which gives a complete description of their structure : * cobham s second theorem . * _ a set @xmath5 is @xmath2-recognizable if and only if its characteristic sequence @xmath13 ( @xmath14 if and only if @xmath15 ) is the image by a letter to letter morphism of a fixed point of a substitution of constant length @xmath2 . _ the original proof of the cobham s first theorem was considered by s. eilenberg in @xcite as `` almost elementary '' but `` highly technical '' , the technicality being concentrated in the direct implication , the reverse implication being much less difficult . in 1974 , eilenberg suggested in @xcite to find some more readable proofs . hansel gave in @xcite some ideas in this matter and pointed out that the proof consists of two steps : 1 ) @xmath0 is syndetic , 2 ) completion of the proof . all the other proofs and all the proofs of the generalizations follow the same dichotomy . later muchnik introduced the key concept of _ definable sets _ and , using the formalism of first order logic , gave a more comprehensible proof of cobham s first theorem . in 1993 , michaux and villemaire @xcite used these definable sets to prove what is known as the cobham - semenov theorem . this theorem was originally proved by semenov in @xcite and is a multidimensional generalization of the cobham s first theorem . other characterizations were given in terms of congruences with finite index @xcite and in terms of algebraic power series @xcite . due to all these characterizations , many generalizations of cobham s first theorem can be stated ( see @xcite for a very nice overview or @xcite ) . in this work we are interested in two other directions : the dynamical systems and the tilings . in section [ definitions ] we give the general background of this paper . we will see in this section that cobham s first theorem can be stated using dynamical systems . this will enable us to give a new proof of this theorem prompted by ideas in @xcite and @xcite . section [ section - frequences ] is devoted to the computation of the frequencies of the substitution language words . it is done in details in order to make the computation in the higher dimensional case easier . this will be used to prove cobham s first theorem ( in section [ section - primitive ] for the primitive case and in section [ section - general ] for the general case ) . the way we treat the primitive case strongly differs from what is done in @xcite ( where the substitutions are not necessarily of constant length ) . in @xcite we use what is called `` derived sequences '' ( see also @xcite ) and here we use the frequencies of words . the interest of these two sections is that in the constant length context the proof is easier to read and to understand , and it makes the generalization to higher dimensions more comprehensible . this is done in section [ section - multi ] . it gives a new proof of the cobham - semenov theorem in the primitive case . we will use ideas coming from tiling dynamical system theory and more specifically some results in @xcite . the main argument we use consists of the description of the frequencies of some patterns that occur in the tilings we consider . let @xmath16 be a finite alphabet . an * automaton over @xmath16 * , @xmath17 , is a directed graph labelled by elements of @xmath16 where @xmath18 is the set of _ states _ , @xmath19 is the set of * initial * states , @xmath20 is the set of * terminal * states and @xmath21 is the set of labelled edges . if @xmath22 , we write @xmath23 . the automaton is * finite * if @xmath18 is finite . a * path * in the automaton is a sequence @xmath24 where @xmath25 for all @xmath26 . we say it is * admissible * if @xmath27 belongs to @xmath28 and @xmath29 belongs to @xmath30 . we call @xmath31 the set of all admissible paths . the * label * of @xmath32 is the word @xmath33 . the set @xmath34 of labels of admissible paths is called the language of @xmath35 . a subset @xmath36 of @xmath37 is said to be * recognizable by a finite automaton * if there exists a finite automaton @xmath35 such that @xmath38 . a * numeration system * is a strictly increasing sequence of integers @xmath39 such that 1 . @xmath40 , 2 . the set @xmath41 is bounded . let @xmath42 be a numeration system and @xmath43 be the upper bound of @xmath41 . let @xmath44 be the alphabet @xmath45 where @xmath46 is the upper integer part of @xmath43 . using the euclidean algorithm we can write every integer @xmath47 in a unique way as follows @xmath48 @xmath49 is the unique integer such that @xmath50 and @xmath51 , @xmath52 , @xmath53 , where @xmath54 is the quotient of the euclidean division of @xmath55 by @xmath56 and @xmath57 the remainder , and @xmath58 . we will say that @xmath59 is the * @xmath60-representation * of @xmath47 and we set @xmath61 we say a set @xmath5 is * @xmath60-recognizable * if the language @xmath62 is recognizable by a finite automaton . we will focus on the standard numeration systems @xmath63 where @xmath2 is a positive integer greater or equal to @xmath4 . for these numeration systems , we say that @xmath0 is @xmath2-recognizable and we set @xmath64 and @xmath65 . let us give some examples to illustrate these notion . let @xmath66 @xmath67 , \rho_2 ( n ) = \epsilon_k \cdots \epsilon_1 \epsilon_0 \right\}.\ ] ] we have @xmath68 , @xmath69 @xmath70 \right\ } .\ ] ] hence these sets are 2-recognizable , respectively , by the following automata . = 1,5truecm = 2truecm = 1,5truecm here an arrow going into a state means that this state is an initial state and an arrow going outside of a state means that this state is a terminal state . the set @xmath71 being an arithmetic progression , cobham s first theorem asserts that @xmath71 is 3-recognizable . we have @xmath72 \}$ ] and the automaton recognizing this set is the following . = 1,5truecm let us now consider the multidimensional case . let @xmath73 . we write @xmath74 in base @xmath2 as a tuple of words , on the alphabet @xmath75 , with the same length by adding leading zeroes if needed . for example , in base @xmath4 we write @xmath76 as @xmath77 . we say @xmath78 is @xmath2-recognizable if @xmath0 written in base @xmath2 is recognizable by a finite automaton . * cobham - semenov theorem . * _ let @xmath2 and @xmath3 be two multiplicatively independent integers greater or equal to @xmath4 . then , a set @xmath79 is both @xmath2-recognizable and @xmath3-recognizable if and only if @xmath0 is semilinear , _ where * semilinear * means that there exist @xmath80 and finite sets @xmath81 , @xmath82 such that @xmath83 we call * alphabet * a finite set of elements called * letters*. let @xmath16 be an alphabet , a * word * on @xmath16 is an element of the free monod generated by @xmath16 , denoted by @xmath84 , i.e. a finite sequence ( possibly empty ) of letters . let @xmath85 be a word , its * length * is @xmath86 and is denoted by @xmath87 . the * empty word * is denoted by @xmath88 , @xmath89 . the set of non - empty words on @xmath16 is denoted by @xmath90 . if @xmath91 $ ] is an interval of @xmath92 then @xmath93 denotes the word @xmath94 and is called a * factor * of @xmath47 . analogous definitions hold for open or semi - open intervals . we say that @xmath93 is a prefix of @xmath47 when @xmath95 and a suffix when @xmath96 . if @xmath97 is a factor of @xmath47 , we call * occurrence * of @xmath97 in @xmath47 every integer @xmath49 such that @xmath98}= u$ ] . let @xmath97 and @xmath99 be two words , we denote by @xmath100 the number of occurrences of @xmath97 in @xmath99 . the elements of @xmath101 are called * sequences*. for a sequence @xmath102 we use the notation @xmath103 and the terms `` occurrence '' and `` factor '' exactly as for a word . the set of factors of length @xmath86 of @xmath104 is written @xmath105 , and the set of factors of @xmath104 , or * language * of @xmath104 , is represented by @xmath106 ; @xmath107 . the sequence @xmath104 is * periodic * if it is the infinite concatenation of a word @xmath99 . a * gap * of a factor @xmath97 of @xmath104 is an integer @xmath108 which is the difference between two successive occurrences of @xmath97 in @xmath104 . we say that @xmath104 is * uniformly recurrent * if each factor has bounded gaps . let @xmath16 , @xmath109 and @xmath110 be three alphabets . a * morphism * @xmath111 is a map from @xmath16 to @xmath112 . such a map induces by concatenation a map from @xmath84 to @xmath112 . if @xmath113 is included in @xmath114 , it induces a map from @xmath101 to @xmath115 . all these maps are written @xmath111 also . to a morphism @xmath111 , from @xmath16 to @xmath112 , is associated the matrix @xmath116 where @xmath117 is the number of occurrences of @xmath49 in the word @xmath118 . to the composition of morphisms corresponds the multiplication of matrices . for example , let @xmath119 , @xmath120 and @xmath121 be three morphisms such that @xmath122 ( we will quite often forget the composition sign ) , then we have the following equality : @xmath123 . in particular , if @xmath124 is a morphism from @xmath16 to @xmath37 we have @xmath125 for all non - negative integers @xmath86 . a * substitution * on the alphabet @xmath16 is a morphism @xmath126 satisfying : 1 . [ cond1 ] there exists @xmath127 such that @xmath128 is the first letter of @xmath129 ; 2 . [ cond2 ] for all @xmath130 , @xmath131 . note that condition is not always required in the literature about substitutions . the language of @xmath132 is the set @xmath133 consisting of all the words having an occurrence in some @xmath134 , @xmath80 and @xmath130 . in some papers ( see @xcite for example ) the condition is not required to be a substitution and our definition corresponds to what pansiot call * growing substitutions * in @xcite . it is classical that @xmath135 converges in @xmath101 to a sequence @xmath104 . the substitution @xmath136 being continuous on @xmath101 this sequence is a * fixed point * of @xmath132 , i.e @xmath137 . whenever the matrix associated to @xmath124 is primitive we say that @xmath111 is a * primitive substitution*. it is equivalent to the fact that there exists @xmath86 such that for all @xmath128 and @xmath138 in @xmath16 , @xmath128 has an occurrence in @xmath134 . note that in this case @xmath139 for all fixed points @xmath104 of @xmath132 . it is a * substitution of constant length * @xmath2 if for all @xmath140 the length of @xmath141 is @xmath2 . we will say a sequence @xmath104 is * generated by a substitution * of constant length @xmath2 if it is the image by a letter to letter morphism of a fixed point of a substitution of constant length @xmath2 . these sequences will be also called * @xmath2-substitutive * in this paper . note that usually the definition of @xmath2-substitutive sequences is more general ( see @xcite for example ) . let @xmath109 be another alphabet , we say that a morphism @xmath142 from @xmath16 to @xmath112 is a * letter to letter morphism * when @xmath143 is a subset of @xmath109 . then the sequence @xmath144 is called * substitutive * , and * primitive substitutive * if @xmath111 is primitive . the matrix of @xmath124 is non - negative and consequently has an eigenvalue @xmath145 such that : if @xmath146 is any other eigenvalue then @xmath147 ( this is a consequence of perron - frobenius theorem , see @xcite ) . we will say that it is the * dominant eigenvalue * or the * perron eigenvalue * of @xmath111 and that @xmath144 is * @xmath148-substitutive*. let us explain how cobham s first theorem can be stated using substitutions . first we need to remark that a set @xmath5 is a finite union of arithmetic progressions if and only if its characteristic sequence is ultimately periodic . hence , due to cobham s second theorem , it is an exercise to prove that cobham s first theorem can be formulated in the following equivalent way : * cobham s first theorem ( substitutive version ) * _ let @xmath2 and @xmath3 be two multiplicatively independent integers greater or equal to 2 . let @xmath16 be a finite alphabet and @xmath149 . then the sequence @xmath104 is generated by both a substitution of constant length @xmath2 and a substitution of constant length @xmath3 if and only if it is ultimately periodic . _ in the substitutive context , the two steps of the proof we mentioned before consist of proving that the letters of @xmath16 appear with bounded gaps in @xmath104 and then to conclude . cobham s second theorem asserts that for each of these sets we can find some substitutions of constant length that generate their characteristic sequences . we labelled the states to construct these substitutions . let the set of states be the alphabet @xmath16 of the substitution . the image of the state @xmath128 is the word @xmath150 where @xmath151 is the state you reach starting from @xmath128 and passing through the arrow labelled by @xmath49 . the substitutions we obtain for @xmath71 ( in base 2 ) , @xmath71 ( in base 3 ) , @xmath152 and @xmath153 are : @xmath154 take one of these substitutions and call @xmath104 its unique fixed point starting with the letter @xmath128 , then , identify to 1 the terminal states and to 0 the other states . we obtain the characteristic sequence @xmath155 of the corresponding set of integers @xmath0 . for @xmath153 we obtain the well - known morse sequence @xmath155 : @xmath156 by a * dynamical system * we mean a pair @xmath157 where @xmath158 is a compact metric space and @xmath159 a continuous map from @xmath158 to itself . we say that it is a * cantor system * if @xmath160 is a cantor space . that is , @xmath160 has a countable basis of its topology which consists of closed and open sets and does not have isolated points . the system @xmath157 is * minimal * whenever @xmath158 and the empty set are the only @xmath159-invariant closed subsets of @xmath158 . we say that a minimal system @xmath157 is * periodic * whenever @xmath158 is finite . we say it is @xmath2-periodic if @xmath161 . let @xmath157 and @xmath162 be two dynamical systems . we say that @xmath162 is a * factor * of @xmath157 if there is a continuous and onto map @xmath163 such that @xmath164 ( @xmath165 is called * factor map * ) . if @xmath142 is one - to - one we say that @xmath165 is an * isomorphism * and that @xmath157 and @xmath162 are * isomorphic*. in this paper we deal with cantor systems called * subshifts*. let @xmath16 be an alphabet . we endow @xmath101 with the topology defined by the metric @xmath166 where @xmath167 and @xmath168 are two elements of @xmath101 . by a * subshift * on @xmath16 we shall mean a pair @xmath169 where @xmath158 is a closed @xmath30-invariant ( @xmath170 ) subset of @xmath101 and @xmath30 is the * shift transformation * [ cols= " < , < , < , < , < " , ] we call language of @xmath158 the set @xmath171 } ; \x\in x , i\leq j\}$ ] . let @xmath97 be a word of @xmath37 . the set @xmath172_x = \ { \x\in x ; \x_{[0,|u|-1 ] } = u \}\ ] ] is called * cylinder*. the family of these sets is a base of the induced topology on @xmath158 . when it will not create confusion we will write @xmath173 $ ] and @xmath30 instead of @xmath173_{x}$ ] and @xmath174 . let @xmath104 be a sequence on @xmath16 and @xmath175 be the set @xmath176 } \in l(\x ) , \forall \ [ i , j]\subset { { \mathbb n}}\}$ ] . it is clear that @xmath177 is a subshift . we say that @xmath177 is the subshift generated by @xmath104 . we remark @xmath178 . let @xmath179 be a subshift on @xmath16 , the following are equivalent : 1 . @xmath179 is minimal . 2 . for all @xmath180 we have @xmath181 . 3 . for all @xmath180 we have @xmath182 . we also have that @xmath177 is minimal if and only if @xmath104 is uniformly recurrent . note that if @xmath162 is another subshift then , @xmath183 if and only if @xmath184 . for primitive substitutions @xmath132 , all the fixed points are uniformly recurrent and generate the same minimal subshift , we call it the * substitution subshift generated by @xmath132 * and we denote it @xmath185 ( for more details see @xcite . ) there is another way to generate subshifts . let @xmath36 be a language on the alphabet @xmath16 and define @xmath186 to be the set of sequences @xmath187 such that each word of @xmath106 appears in a word of @xmath36 . the pair @xmath188 is a subshift and we call it the * subshift generated by @xmath36*. if @xmath132 is a primitive substitution , then @xmath189 where @xmath190 . . it is easy to show that if @xmath104 is an ultimately periodic sequence and @xmath191 is minimal , then @xmath104 is periodic . from all the remarks we made before it is not surprising that the following theorem holds : [ cob : systdyn ] let @xmath132 and @xmath111 be two primitive substitutions with dominant eigenvalues @xmath148 and @xmath146 respectively . suppose @xmath148 and @xmath146 are multiplicatively independent . then , @xmath185 is isomorphic to @xmath192 if and only if they are periodic with the same period . the definition of multiplicative independence of real positive numbers is the same as the one we gave before for positive integers : two positive real numbers @xmath193 are multiplicatively independent if whenever @xmath194 , @xmath195 , this implies @xmath196 . remark that it is equivalent to the fact that @xmath197 is an irrational number . we will prove this theorem later and show why it implies cobham s first theorem . a key notion in this proof is the notion of _ return words_. for the rest of the section @xmath104 is a uniformly recurrent sequence on the alphabet @xmath16 and @xmath179 is the minimal subshift it generates . we recall that all sequences in @xmath158 are uniformly recurrent . let @xmath97 be a non - empty word of @xmath198 . a word @xmath199 on @xmath16 is a * return word to @xmath97 * in @xmath104 if there exist two consecutive occurrences @xmath200 of @xmath97 in @xmath104 such that @xmath201 . the set of return words to @xmath97 is denoted by @xmath202 . it is immediate to check that a word @xmath203 is a return word if and only if : 1 . @xmath204 ( i.e. @xmath205 is a factor of @xmath104 ) ; 2 . @xmath97 is a prefix of @xmath206 ; 3 . the word @xmath207 has only two occurrences of @xmath97 . * remarks . * 1 . as @xmath104 is uniformly recurrent , the difference between two consecutive occurrences of @xmath97 in @xmath104 is bounded , and the set @xmath202 of return words to @xmath97 is finite . the previous statement @xmath208 can not be simplified : it is not equivalent to _ @xmath97 is a prefix of @xmath199_. for example , if @xmath209 is a factor of @xmath104 then the word @xmath128 is a return word to @xmath210 . 3 . from this characterization , it follows that the set of return words to @xmath97 is the same for all @xmath211 , hence we set @xmath212 . if it is clear from the context , we write @xmath213 instead of @xmath214 . in this section we develop the tools we will use to prove theorem [ cob : systdyn ] . we say that a sequence @xmath104 on a finite alphabet is * linearly recurrent ( with the constant @xmath215 ) * if it is recurrent and if , for every word @xmath97 of @xmath104 and all @xmath216 it holds @xmath217 we say @xmath179 is a * linearly recurrent subshift * if it is a minimal subshift that contains a linearly recurrent sequence . [ sublinrec ] all primitive substitutive sequences , and the subshifts they generate , are linearly recurrent . let @xmath124 be a primitive substitution and @xmath104 one of its fixed points . we begin proving that there exists a constant @xmath110 such that for all positive integers @xmath7 @xmath218 for all @xmath7 we choose some letters @xmath219 and @xmath220 such that @xmath221 and @xmath222 . by primitivity there exists @xmath223 such that for all @xmath224 the letter @xmath138 has an occurrence in the word @xmath225 . we set @xmath226 . for @xmath227 we have @xmath228 let @xmath97 be a word of @xmath229 and @xmath199 be a return word to @xmath97 . let @xmath7 be the smallest integer such that @xmath230 . the choice of @xmath7 entails that there exists a word @xmath231 of length 2 such that @xmath97 occurs in @xmath232 . let @xmath233 be the largest difference between two successive occurrences of a word of length 2 of @xmath234 . it follows @xmath235 when the substitution @xmath132 is primitive we will also say that @xmath136 is linearly recurrent with some constant . [ linrec ] let @xmath236 be an non - periodic linearly recurrent sequence for the constant @xmath237 . then : 1 . the number of distinct factors of length @xmath86 of @xmath104 is less or equal to @xmath238 . 2 . @xmath104 is @xmath239-power free ( i.e. @xmath240 if and only if @xmath241 ) . 3 . for all @xmath242 and for all @xmath243 we have @xmath244 . 4 . for all @xmath245 , @xmath246 . we start with a remark . let @xmath86 be a positive integer and @xmath247 a word of length @xmath248 . let @xmath249 be a word of length @xmath86 . the difference between two successive occurrences of @xmath99 is smaller than @xmath238 , consequently @xmath97 has at least one occurrence of @xmath99 . we have proved that : for each @xmath86 , every words of length @xmath86 has at least one occurrence in each word of length @xmath248 . from this remark we deduce ( 1 ) . let @xmath247 be a word such that @xmath240 . each factor of @xmath104 of length @xmath250 occurs in @xmath251 . but in @xmath251 occurs at most @xmath250 distinct factors of length @xmath250 of @xmath104 . this contradicts the non - periodicity of @xmath104 . ( we recall that if for some @xmath86 a sequence @xmath252 has at most @xmath86 different words of length @xmath86 , then it is ultimately periodic , see @xcite . ) assume there exist @xmath247 and @xmath253 such that @xmath254 . the word @xmath199 is a return word to @xmath97 therefore @xmath97 is a prefix of @xmath207 . we deduce that @xmath255 is a prefix of @xmath97 . hence @xmath256 belongs to @xmath229 because @xmath207 belongs to @xmath229 . consequently @xmath257 and ( 3 ) is proved . let @xmath97 be a factor of @xmath104 and @xmath258 be a word of length @xmath259 . each word of length @xmath260 occurs in @xmath99 , hence each return word to @xmath97 occurs in @xmath99 . it follows from ( 3 ) that in @xmath99 will occur at most @xmath261 return words to @xmath97 , which proves ( 4 ) . the following well - known theorem is fundamental to prove the existence and compute the frequencies of the words in fixed points of substitutions . the proof can be found in @xcite or @xcite . [ perron - frob ] let @xmath262 be a @xmath263 primitive matrix . then : 1 . the matrix @xmath262 has a positive eigenvalue @xmath264 which is strictly greater than the modulus of any other eigenvalue ; 2 . the eigenvalue @xmath264 is algebraically simple ; 3 . to this eigenvalue corresponds an eigenvector with positive entries . 4 . there exist @xmath265 and @xmath110 such that for all @xmath266 and all @xmath80 we have + @xmath267 + where @xmath268 and @xmath269 are respectively the unique right and left eigenvectors satisfying @xmath270 let @xmath271 be a primitive substitution , @xmath262 its matrix and @xmath104 one of its fixed points . the eigenvalue @xmath264 of the previous theorem will be called the * perron eigenvalue * of @xmath262 or @xmath132 . these real numbers are called * perron numbers*. we take the notations of the previous theorem . for a word @xmath245 we call * frequency * of @xmath97 in @xmath272 the limit ( when it exists ) @xmath273 } \right\ } .\ ] ] we recall that for all @xmath86 and all @xmath274 in @xmath16 we have @xmath275 . consequently from perron theorem we obtain , for all @xmath80 , @xmath276 we set @xmath277 . we fix @xmath127 . now we prove that @xmath278 exists . let @xmath279 . there exist @xmath86 , and , words @xmath280 ( @xmath281 ) and @xmath151 ( @xmath281 ) such that 1 . @xmath282 and @xmath283 for @xmath82 where @xmath284 ; 2 . @xmath285 is non - empty ; 3 . @xmath286 . moreover from there exists a constant @xmath287 such that @xmath288 for all @xmath289 . hence @xmath290 for some constants @xmath291 , where @xmath292 . this means that @xmath278 exists and is equal to @xmath293 . here @xmath294 is a primitive substitution . in this section we prove the frequency of words exists for primitive substitutions . we use the previous subsection and the following substitutions . let @xmath295 be an integer and @xmath296 be the set of words of length @xmath7 belonging to @xmath297 ; we consider @xmath298 as an alphabet and we define a substitution @xmath299 in the following way : for @xmath300 with @xmath301 and @xmath302 , we put @xmath303})(v_{[2,k+1 ] } ) \cdots ( v_{[p , p+k-1 ] } ) .\ ] ] in other words , @xmath304 consists of the ordered list of the first @xmath305 factors of length @xmath7 of @xmath306 . remark that , for every @xmath307 , @xmath308 is associated to @xmath309 in the same way as @xmath310 is associated to @xmath136 : @xmath311 consists of the ordered list of the first @xmath312 factors of length @xmath7 of @xmath313 . in particular we have : latexmath:[\[\begin{aligned } \label{theta = theta2 } if @xmath86 is large enough , every @xmath315 of length @xmath7 is a factor of @xmath316 for every @xmath127 ; thus , @xmath317 occurs in @xmath311 for every @xmath300 . we proved that @xmath310 is primitive . let @xmath199 be a word of length @xmath318 over the alphabet @xmath296 . from the definition of @xmath297 and @xmath319 it can be checked that : @xmath320 if and only if there exists a word @xmath279 of length @xmath321 such that @xmath322})(v_{[2,k+1]})\dots ( v_{[n , n+k-1]})$ ] . clearly , given @xmath323 , the number of occurrences of the symbol @xmath324 in @xmath199 is equal to the number of occurrences of @xmath97 in @xmath99 . and consequently @xmath325 finally applying the results of the previous subsection to @xmath310 for all @xmath7 we obtain the following result . [ prop - freq ] for all @xmath289 there exist constants @xmath326 , @xmath327 and @xmath328 such that for all @xmath279 of length greater than @xmath250 we have @xmath329 the frequency of @xmath97 exists for all @xmath330 . in @xcite is proved the following theorem which is central in the present paper . they use the recognizability property of primitive substitutions . we will give a different proof not using this property but proposition [ linrec ] . [ mesure - cyl ] let @xmath331 be the perron eigenvalue of @xmath132 . there exists a finite set @xmath332 such that for all @xmath80 there exists @xmath333 satisfying @xmath334 we suppose @xmath132 is linearly recurrent with the constant @xmath237 . let @xmath335 be the perron eigenvalue of @xmath336 . from and perron theorem we deduce that @xmath337 . let @xmath338 be the incidence matrix of @xmath336 . from we know that @xmath339 for all @xmath340 . subsection [ perron - freq ] and theorem [ perron - frob ] imply that @xmath341 is the unique right eigenvector of @xmath338 ( for the eigenvalue @xmath335 ) with @xmath342 . let @xmath110 be the constant defined as in . let @xmath289 be a word of length @xmath86 and @xmath7 the smallest integer such that latexmath:[\[\begin{aligned } let @xmath109 be the set of words @xmath344 such that @xmath97 has an occurrence in @xmath345 . the choice of @xmath7 implies this set is non - empty . let @xmath346 and @xmath347 . from proposition [ linrec ] it follows that @xmath348 moreover from we have that @xmath349 and from proposition [ prop - freq ] , for all @xmath350 , @xmath351 let @xmath352 be the last word of length two of @xmath353 . then , @xmath354 consequently , it suffices to take @xmath355\cap { { \mathbb n } } , ab\in a_2 \right\}\ ] ] which is a finite set . in this section we prove the following theorem . [ cobham - prim ] let @xmath132 and @xmath111 be two primitive substitutions whose dominant eigenvalues are respectively @xmath148 and @xmath146 . suppose that @xmath185 and @xmath192 has a common factor @xmath162 . if @xmath356 is non - periodic then @xmath148 and @xmath146 are multiplicatively dependent . theorem [ cob : systdyn ] is clearly a corollary of theorem [ cobham - prim ] . let @xmath142 be a factor map from the subshift @xmath179 on the alphabet @xmath16 onto the subshift @xmath162 on the alphabet @xmath109 . if there exists a * @xmath357-block map * @xmath358 such that @xmath359})$ ] for all @xmath360 and @xmath361 , we shall say that @xmath362 is a * block map associated to @xmath142 * , that @xmath362 * defines * @xmath165 and that @xmath165 is a * sliding block code*. the theorem of curtis - hedlund - lyndon ( theorem 6.2.9 in @xcite ) asserts that factor maps are * sliding block codes*. if @xmath363 is a word of length @xmath364 we define @xmath365 by @xmath366})$ ] , @xmath367 . let @xmath110 denote the alphabet @xmath368 and @xmath369 } ) ; i\in { { \mathbb n}})\in c^{{{\mathbb n } } } ; ( x_n ; n\in { { \mathbb n } } ) \in x \}$ ] . it is easy to check that the subshift @xmath370 is isomorphic to @xmath179 and that @xmath362 induces a 0-block map from @xmath110 onto @xmath109 which defines a factor map from @xmath370 onto @xmath162 . the next lemma was first proved in @xcite . [ mainlemme ] let @xmath179 be a non - periodic lr subshift ( for the constant @xmath237 ) and @xmath162 be a non - periodic subshift factor of @xmath179 . then @xmath162 is lr . moreover , there exists @xmath371 such that : for all @xmath372 with @xmath373 we have 1 . [ point1 ] @xmath374 for all @xmath375 ; 2 . [ point2 ] @xmath376 . we denote by @xmath16 the alphabet of @xmath158 and by @xmath109 the alphabet of @xmath356 . let @xmath377 be a factor map . let @xmath358 be a block map associated to @xmath142 . let @xmath97 be a word of @xmath378 and @xmath379 be such that @xmath380 . we have @xmath381 . if @xmath199 is a return word to @xmath97 then @xmath382 . then , the subshift @xmath162 is linearly recurrent with the constant @xmath383 . moreover : for all @xmath384 such that @xmath385 , and for all @xmath386 , @xmath387 . to obtain the other inequality it suffices to proceed as in the proof of proposition [ linrec ] . let @xmath388 with @xmath373 and @xmath389 be a word of length @xmath390 . each word of length @xmath391 occurs in @xmath99 , hence each return word to @xmath97 occurs in @xmath99 . it follows from the previous assertion that in @xmath99 occurs at the most @xmath392 return words to @xmath97 . [ bounded - preimages ] let @xmath179 be a non - periodic lr subshift ( for the constant @xmath237 ) . let @xmath393 be a factor map such that @xmath162 is a non - periodic subshift and @xmath394 be a @xmath357-block map defining @xmath142 . then there exists @xmath395 such that for all @xmath396 , with @xmath397 , we have @xmath398 let @xmath371 be the integer given by lemma [ mainlemme ] . we set @xmath399 . let @xmath388 such that @xmath400 . the difference between two distinct occurrences of elements of @xmath401 is greater than @xmath402 . moreover @xmath403 is included in @xmath404 and each word of length @xmath405 has an occurrence of each word of @xmath406 . therefore @xmath407 this completes the proof . let @xmath132 be a primitive substitution with dominant eigenvalue @xmath148 and linearly recurrent constant @xmath237 , and , @xmath162 a non - periodic factor of @xmath185 . let @xmath408 be a factor map and @xmath362 be a @xmath357-block map that defines @xmath142 . from theorem [ mesure - cyl ] we know there exists a finite set @xmath409 such that for all @xmath86 there exists @xmath333 satisfying @xmath410 let @xmath411 , @xmath412 , and set @xmath413 with @xmath414 ( proposition [ bounded - preimages ] ) . let @xmath7 be as in for @xmath415 . let @xmath416 and @xmath417 such that @xmath418 . we remark that @xmath419 } \right\}\ ] ] exists and is equal to @xmath420 } \in \ { v_1 , \dots , v_l \}\right\}\ ] ] we denote it @xmath421 . moreover , @xmath422 where @xmath423 is the finite set @xmath424 . we proved : [ mesure - cyl - fact ] let @xmath331 be the perron eigenvalue of the primitive substitution @xmath132 . there exists a finite set @xmath332 such that for all non - periodic subshift factor @xmath162 of @xmath185 , and all @xmath80 there exists @xmath333 satisfying @xmath425 from theorem [ mesure - cyl - fact ] we know there exist two finite sets @xmath426 and @xmath427 such that for all @xmath80 there exists @xmath428 satisfying @xmath429 using statement ( 1 ) of lemma [ mainlemme ] we have that @xmath430 for all @xmath431 . then , @xmath421 tends to @xmath432 when @xmath250 goes to infinity . consequently , there exist @xmath433 , @xmath434 , @xmath435 , @xmath428 , @xmath436 , and @xmath437 , @xmath438 , such that @xmath439 we obtain that @xmath440 , which ends the proof . in this section we prove the substitutive version of cobham s first theorem . the arguments presented below also work for a wide class of non - constant length substitutions that are called in @xcite `` good substitutions '' . we first prove that the letters appear with bounded gaps . then , we could conclude using the previous section . the sufficient part is easier to establish . the following theorem is a key argument in what follows . [ densite ] let @xmath148 and @xmath146 be two multiplicatively independent positive real numbers . then the set @xmath441 is dense in @xmath442 . see @xcite . let @xmath2 and @xmath3 be two multiplicatively independent perron numbers . let @xmath132 ( resp . @xmath111 ) be substitution of constant length @xmath2 ( resp . @xmath3 ) defined on the alphabet @xmath16 ( resp . @xmath109 ) , with fixed point @xmath155 ( resp . @xmath443 ) . we suppose there exist two letter to letter morphisms @xmath444 and @xmath445 such that @xmath446 . [ etoile ] let @xmath127 be a letter which has infinitely many occurrences in @xmath104 . there exist a positive integer @xmath8 , a word @xmath447 and @xmath448 such that for all @xmath80 the word @xmath449 is a prefix of @xmath155 . let @xmath127 be a letter that has infinitely many occurrences in @xmath104 . we set @xmath450 . there exists @xmath451 which has infinitely many occurrences in @xmath104 and such that @xmath452 has an occurrence in @xmath453 . in this way we can construct a sequence @xmath454 such that @xmath450 and @xmath455 occurs in @xmath456 , for all @xmath360 . there exist @xmath457 with @xmath458 such that @xmath459 . it follows that @xmath128 occurs in @xmath460 and @xmath138 occurs in @xmath461 . hence there exist @xmath462 such that @xmath463 and @xmath464 . we set @xmath465 , @xmath466 and @xmath467 . there exists @xmath468 such that @xmath469 is a prefix of @xmath104 . we remark that for all @xmath80 the word @xmath470 is a prefix of @xmath104 too . we set @xmath471 . we have @xmath472 . consequently for all @xmath80 @xmath473 is a prefix of @xmath474 . then @xmath475 is a prefix of @xmath476 and consequently of @xmath104 , for all @xmath80 . [ bgaps ] the letters of @xmath110 which have infinitely many occurrences in @xmath104 appear in @xmath104 with bounded gaps . let @xmath477 which has infinitely many occurrences in @xmath104 . let @xmath478 and @xmath479 . assume that the letter @xmath43 does not appear with bounded gaps . then there exist @xmath127 with infinitely many occurrences in @xmath155 and a strictly increasing sequence @xmath480 of positive integers such that the letter @xmath43 does not appear in @xmath481 . let @xmath482 such that @xmath483 is a prefix of @xmath155 . of course we can suppose that @xmath97 is non - empty . for all @xmath80 we denote @xmath484 the set of letters appearing in @xmath485 . there exist two distinct integers @xmath486 such that @xmath487 . let @xmath488 be the set of letters appearing in @xmath489 . it is easy to show that @xmath490 . consequently the set of letters appearing in @xmath491 is equal to @xmath488 and for all @xmath333 the set of letters appearing in @xmath492 is equal to @xmath488 . we set @xmath493 and @xmath494 . we remark that the letter @xmath43 does not appear in the word @xmath495 and that @xmath496 for all @xmath333 . let @xmath497 be such that latexmath:[$ being of constant length @xmath2 we have @xmath499 from lemma [ etoile ] applied to @xmath111 we have that there exist @xmath500 and @xmath501 such that for all @xmath80 @xmath502}\right ) = c.\ ] ] put @xmath503 . we have @xmath504 from theorem [ densite ] it follows that there exist two strictly increasing sequences of integers , @xmath505 and @xmath506 , and @xmath507 such that the sequence @xmath508 tends to @xmath509|u| ( 1+\epsilon ) , |ua| ( 1-\epsilon)[. $ ] hence , @xmath510 from ( [ encadre ] ) and ( [ limite ] ) there exists @xmath360 such that @xmath511 belongs to @xmath158 by . this gives a contradiction with ( [ gap ] ) . we obtain the same conclusion for the words : [ wordgap ] the words having infinitely many occurrences in @xmath104 appear in @xmath104 with bounded gaps . let @xmath97 be a word having infinitely many occurrences in @xmath104 . we set @xmath512 . to prove that @xmath97 appears with bounded gaps in @xmath104 it suffices to prove that the letter 1 appears with bounded gaps in the sequence @xmath513 defined by @xmath514 } = u\ ] ] and @xmath432 otherwise . in the sequel we prove that @xmath515 is @xmath2 and @xmath3-substitutive . the sequence @xmath516 is a fixed point of the substitution @xmath517 , where @xmath518 is the alphabet @xmath519 , defined for all @xmath520 in @xmath518 by @xmath521 where @xmath522 ( for more details see section v.4 in @xcite for example ) . let @xmath523 be the letter to letter morphism defined by @xmath524 for all @xmath525 . we have @xmath526 , and then @xmath527 . consequently the dominant eigenvalue of @xmath528 is @xmath2 and @xmath529 is @xmath2-substitutive . let @xmath530 be the letter to letter morphism defined by @xmath531 if @xmath532 and @xmath432 otherwise . it is easy to see that @xmath533 hence @xmath515 is @xmath2-substitutive . in the same way we show that @xmath515 is @xmath3-substitutive and proposition [ bgaps ] concludes the proof . we keep the notations of the previous subsection . [ sous - sub ] there exist @xmath333 and @xmath534 such that @xmath535 is a primitive substitution . the proof is left to the reader . let @xmath7 , @xmath536 and @xmath8 , @xmath537 be given by lemma [ sous - sub ] for respectively @xmath132 and @xmath111 . we remark that @xmath538 is a substitution of constant length @xmath539 and @xmath540 a substitution of constant length @xmath541 . let @xmath542 and @xmath543 be the substitutive dynamical systems they generate . each element of @xmath544 and @xmath545 appears infinitely many times and consequently with bounded gaps in @xmath104 . hence @xmath546 , we call it @xmath36 . let @xmath179 be the dynamical system @xmath36 generates . we remark that @xmath547 and @xmath548 defined factor maps . theorem [ cobham - prim ] implies @xmath179 is periodic . hence , there exists a word @xmath97 such that @xmath549 where @xmath250 is the least period . there exists an integer @xmath550 such that all the words of length @xmath250 appear infinitely many times in @xmath551 . we set @xmath552 and we will prove that @xmath515 is ultimately periodic and consequently @xmath104 will be ultimately periodic . the word @xmath97 appears infinitely many times , consequently it appears with bounded gaps . let @xmath553 be the set of return words to @xmath97 . it is finite . there exists an integer @xmath554 such that all the words @xmath555 appear infinitely many times in @xmath104 . hence these words appear with bounded gaps in @xmath104 . we set @xmath556 and we will prove that @xmath557 is periodic and consequently @xmath104 will be ultimately periodic . we can suppose that @xmath97 is a prefix of @xmath557 . then @xmath557 is a concatenation of return words to @xmath97 . let @xmath199 be a return word to @xmath97 . it appears with bounded gaps hence it appears in some @xmath558 and there exist a suffix @xmath357 and a prefix @xmath559 of @xmath97 , and an integer @xmath49 such that @xmath560 . as @xmath250 is the least period of @xmath36 it follows that @xmath561 . it follows that @xmath562 . [ recip ] let @xmath104 be a sequence on a finite alphabet @xmath16 and @xmath563 an integer . if @xmath104 is ultimately periodic then it is @xmath2-substitutive . suppose @xmath564 . it is easy to see that @xmath565 where @xmath566 for some well chosen @xmath97 and @xmath99 . we set @xmath567 . take the alphabet @xmath568 and define the substitution @xmath569 by @xmath570 we call @xmath155 the fixed point of @xmath132 starting with @xmath452 . now we define a letter to letter morphism @xmath571 by @xmath572 then it can be checked that @xmath573 which achieves the proof . in this section we prove the necessary part of the substitutive version of the cobham - semenov theorem in the primitive case . let @xmath16 be a finite alphabet . we call * array * in @xmath574 any function @xmath575 . the collection of all these arrays is @xmath576 that we endow with the product topology of the discrete topologies . a non - empty set @xmath577 of pairwise distinct elements is called a * configuration*. a @xmath110-*pattern * @xmath32 is a map from @xmath110 to @xmath16 . a @xmath110-pattern @xmath32 of @xmath578 is defined as the function @xmath579 with @xmath580 for all @xmath581 , for some @xmath582 . when @xmath583 we set @xmath584 . the set @xmath110 is the * support * of @xmath32 and is denoted by @xmath585 . we set @xmath586 . an * occurrence * of the @xmath110-pattern @xmath32 in @xmath578 is a vector @xmath582 such that @xmath587 . in this situation we will say that @xmath32 has an occurrence in @xmath578 or @xmath32 appears in @xmath578 . the @xmath110-*language * of @xmath578 is defined to be the set @xmath588 of @xmath110-patterns having an occurrence in @xmath578 : @xmath589 . when @xmath590 is a pattern we define in the same way the notion of occurrence of @xmath32 in @xmath590 and the related notions . we will write the elements of @xmath574 as @xmath591 ; we set @xmath592 . the * cube * located at @xmath593 of side @xmath594 is the set @xmath595 . a * cubic - pattern * @xmath32 is a @xmath110-pattern where @xmath596 for some @xmath597 and @xmath598 . we also say @xmath32 is a @xmath233-cubic - pattern . the @xmath599-cubic - patterns can be viewed as letters of @xmath16 ; when it is clear from the context we will identify them . let @xmath158 be a subset of @xmath600 and @xmath32 be a pattern . we denote by @xmath601_x$ ] the subset of @xmath158 consisting of the elements @xmath602 such that @xmath603 . when @xmath32 is a cubic - pattern with @xmath604 , we say @xmath601_x $ ] is a cubic - cylinder . a subset @xmath605 of @xmath606 is * syndetic * if there exist @xmath233 such that for each cube @xmath110 of side @xmath233 the set @xmath607 is non - empty . we say @xmath578 is * syndetic * if @xmath608 is syndetic for all @xmath127 . we say @xmath578 is * uniformly recurrent * if for every pattern @xmath32 there is @xmath609 such that @xmath32 appears in every @xmath233-cubic - pattern of @xmath578 . it clearly fits the definitions of the one - dimensional case . we say @xmath578 is * periodic * ( resp . * ultimately periodic * ) if there exists @xmath610 such that @xmath611 for all @xmath593 ( resp . for all large enough @xmath598 ) . we will also need another notion of periodicity . we say @xmath612 is @xmath598 * -periodic inside * @xmath613 if for any @xmath614 with @xmath615 we have @xmath616 remark that for @xmath622 , local periodicity is equivalent to the ultimate periodicity . we say @xmath578 is * pseudo - periodic * if for all @xmath127 , @xmath623 is locally periodic and every @xmath624-sections of @xmath623 , @xmath625 , @xmath626 and @xmath80 , is pseudo - periodic ( ultimately periodic when @xmath627 ) . we say @xmath79 is * definable * whenever its characteristic function is pseudo - periodic . the following criterion is due to muchnik ( see @xcite for the proof ) . this is also called the @xmath606-shift ; it shifts the array so that the element that used to be at @xmath598 is now at @xmath633 . a @xmath606-*subshift * is a pair @xmath634 where @xmath158 is a closed subset of @xmath635 which is invariant by all translations under @xmath636 . we say @xmath634 is minimal if all orbits are dense : for all @xmath637 , let @xmath2 be a positive integer and @xmath16 a finite alphabet . a @xmath2-substitution ( or substitution if we do not need to specify @xmath2 ) is a map @xmath639 where @xmath640 . the substitution @xmath159 can be considered as a function from @xmath576 into itself by setting @xmath641 we say @xmath159 is * primitive * whenever @xmath650 is primitive ( i.e. there exists @xmath7 such that @xmath651 ) . this means that for every @xmath127 there exists @xmath130 such that @xmath128 appears in @xmath652 . when @xmath159 is primitive there always exist @xmath653 and @xmath654 such that @xmath655 . we say @xmath578 is * generated by a @xmath2-substitution * if there exist a letter to letter morphism @xmath142 and a fixed point @xmath656 of a @xmath2-substitution such that @xmath657 . let @xmath158 be the subset of @xmath635 consisting of the arrays @xmath578 verifying that for each subblock @xmath109 of @xmath578 there exist @xmath658 and @xmath127 such that @xmath109 is a subblock of @xmath659 . it can be checked that @xmath158 is closed and translation invariant . the action of @xmath606 by translations on @xmath158 is a @xmath574-subshift called * substitution subshift * or * @xmath574-subshift generated by a @xmath2-substitution*. the orbit closure of @xmath656 , @xmath660 , is not necessarily equal to @xmath158 . but it is the case when @xmath159 is primitive . in this subsection we `` translate '' the cobham - semenov theorem in terms of substitutions to obtain an equivalent statement we will prove in the primitive case . in @xcite the authors proved the following theorem which is analogous to cobham s second theorem . * cobham - semenov theorem ( substitutive version ) . * _ let @xmath2 and @xmath3 be two multiplicatively independent integers greater or equal to @xmath4 . then , the array @xmath578 is generated by both a @xmath2-substitution and a @xmath3-substitution if and only if @xmath0 is pseudo - periodic . _ in the sequel @xmath661 is a primitive substitution , with @xmath662 . we set @xmath663 . it is not a restriction to suppose that @xmath159 has a fixed point @xmath656 . let @xmath664 be its associated @xmath606-subshift and @xmath665 be its incidence matrix . [ linrec1 ] let @xmath666 be a minimal substitution subshift . there exists a constant @xmath667 such that for every cubic pattern @xmath668 , any cubic pattern @xmath669 satisfying @xmath670 contains a translated copy of @xmath32 . this property is called * linear recurrence * for @xmath622 in @xcite , * strong repetitivity * in @xcite and * linear repetitivity * in @xcite for @xmath73 . we choose to say linearly recurrent with the constant @xmath237 . we say @xmath656 is linearly recurrent . as in the one - dimensional case , all tilings of @xmath158 have the same property . hence we also say that @xmath671 is linearly recurrent . the following result is an extension of the assertion ( 3 ) of proposition [ linrec ] to the multidimensional case . but the proof is not an extension of the proof of proposition [ linrec ] . in proposition [ linrec ] we use the fact that if for some @xmath86 a sequence @xmath252 has at most @xmath86 different words of length @xmath86 , then it is ultimately periodic ( see @xcite ) . for @xmath672 such a result does not exist . [ linrec2 ] let @xmath673 be a non - periodic linearly recurrent @xmath606-subshift . there exists @xmath674 such that for any @xmath675 and any cubic pattern @xmath32 , if @xmath676 and @xmath677 are two different occurrences of @xmath32 and @xmath678 then @xmath679 . in @xcite the author proves that self - similar tiling systems are uniquely ergodic for the action of @xmath682 by translations . the array @xmath656 can be viewed has a self - similar tiling . then , theorem 3.3 of @xcite implies that , for all patterns @xmath32 , the frequency @xmath683 exists for all @xmath578 and let @xmath691 be the set of 2-cubic patterns such that @xmath32 has an occurrence in @xmath692 . the choice of @xmath7 implies this set is non - empty . let @xmath693 . from proposition [ linrec2 ] it follows that let @xmath671 and @xmath700 be two @xmath606-subshifts defined respectively on the alphabets @xmath16 and @xmath109 . we say @xmath671 is a factor of @xmath700 if there exists a continuous and onto map @xmath701 such that @xmath702 for all @xmath578 and @xmath703 . we say @xmath165 is given by a * @xmath233-sliding block code * if there exists a map @xmath704 such that @xmath705})$ ] . we recall @xmath706}$ ] is the restriction of the array @xmath707 to @xmath708 . [ sliding ] consider @xmath709 a minimal non - periodic substitution subshift and @xmath710 a @xmath606-subshift . let @xmath711 be a factor map . then , there exists @xmath36 and @xmath712 such that for all @xmath233-cubic - pattern @xmath32 , the set @xmath713_x ) $ ] is the union of at most @xmath36 @xmath714-cubic - cylinders . from proposition [ linrec1 ] , @xmath709 is linearly recurrent with some constant @xmath237 . the factor map @xmath165 is given by a @xmath712-sliding block code @xmath715 . the @xmath606-subshift @xmath710 is clearly linearly recurrent with the constant @xmath237 : let @xmath674 be the constant given by proposition [ linrec2 ] . let @xmath716 be a @xmath233-cubic - pattern with @xmath717 . the continuity of @xmath142 together with the cantor structure of @xmath356 imply @xmath713_x ) = \cup_{i=1}^n [ p_i]_y$ ] where the @xmath718 are cubic patterns of @xmath356 . we can consider that the @xmath718 are @xmath719-cubic - patterns ( this is the curtis - lyndon - hedlund theorem ) . take a @xmath720-cubic - pattern @xmath18 of @xmath356 . by definition of the linear recurrence , all @xmath721-cubic - patterns have an occurrence in @xmath18 and a fortiori all @xmath718 . proposition [ linrec2 ] says that the difference @xmath676 between two occurrences of @xmath32 is such that @xmath722 . hence if @xmath703 is the occurrence of some @xmath718 and @xmath723 is the occurrence of some @xmath724 , then @xmath725 ; i.e there is at most one occurrence of some @xmath718 in the @xmath726-cubic - patterns . consequently , the number of occurrences of the @xmath718 in @xmath18 is at most @xmath727 . hence @xmath86 is bounded which concludes the proof . let @xmath728 and @xmath729 be two minimal subshifts generated respectively by a @xmath2-substitution and a @xmath3-substitution . if they have a common non - periodic @xmath606-subshift factor @xmath730 then @xmath2 and @xmath3 are multiplicatively dependent . from theorem [ freq - array ] and proposition [ sliding ] , there exist two finite sets @xmath731 and @xmath732 such that @xmath733_z \hbox { cubic - cylinder of } z \ } \subset \left ( \cup_{n\in { { \mathbb n } } } f_1 r_1^{-n } \right ) \cap \left ( \cup_{n\in { { \mathbb n } } } f_2 r_2^{-n } \right ) .\ ] ] the subshift @xmath734 being non - periodic , the set @xmath735 is infinite . hence , using the pigeon hole principle , there exist @xmath736 , @xmath737 , and four positive integers @xmath738 and @xmath739 with @xmath740 note that other cobham type theorems exist . in @xcite the substitutive version of cobham s theorem is extended to what is called `` good substitutions '' . we strongly believe it can be extended to all substitutions but it remains an open problem . in @xcite cobham s first theorem is generalized to non standard numeration systems like fibonacci numeration system : @xmath742 , @xmath743 , @xmath40 . this certainly could be extended to other numeration systems ( see @xcite for example ) . in a different direction , a cobham s type theorem can be found in @xcite . in this paper they consider @xmath682-translation orbit closure of self similar tilings . ( in our paper we consider @xmath574-translation orbit closure of tilings generated by multidimensional constant length substitutions . ) the goal of their paper is not related to cobham s theorem but the main argument they establish ( proposition 3 of @xcite ) is clearly a cobham s theorem for self similar tilings . a. cerny and j. gruska , _ modular trellises _ , in the book of @xmath36 , g. rozenberg and a. salomaa , springer - verlag ( 1985 ) , 45 - 61 . g. christol , t. kamae , m. mends - france et g. rauzy , _ suites algbriques et substitutions _ , bull . france 10 8 ( 1980 ) , 401 - 419 . c. michaux and r. villemaire , _ presburger arithmetic and recognizability of sets of natural numbers by automata : new proofs of cobham s theorem and semenov s theorem _ , annals of pure and applied logic 77 ( 1996 ) , 251 - 277 .

the influence of random surface inhomogeneities on spectral properties of open microresonators is studied both theoretically and experimentally . to solve the equations governing the dynamics of electromagnetic fields the method of eigen - mode separation is applied previously developed with reference to inhomogeneous systems subject to arbitrary external static potential . we prove theoretically that it is the gradient mechanism of wave - surface scattering which is the highly responsible for non - dissipative loss in the resonator . the influence of side - boundary inhomogeneities on the resonator spectrum is shown to be described in terms of effective renormalization of mode wave numbers jointly with azimuth indices in the characteristic equation . to study experimentally the effect of inhomogeneities on the resonator spectrum , the method of modeling in the millimeter wave range is applied . as a model object we use dielectric disc resonator ( ddr ) fitted with external inhomogeneities randomly arranged at its side boundary . experimental results show good agreement with theoretical predictions as regards the predominance of the gradient scattering mechanism . it is shown theoretically and confirmed in the experiment that tm oscillations in the ddr are less affected by surface inhomogeneities than te oscillations with the same azimuth indices . the ddr model chosen for our study as well as characteristic equations obtained thereupon enable one to calculate both the eigen - frequencies and the _ q_-factors of resonance spectral lines to fairly good accuracy . the results of calculations agree well with obtained experimental data .

0811.4734

nowadays microresonators ( disk- , ring- , and spherical - shaped ) evoke considerable interest because new possibilities have recently opened up to develop these types of resonators in the optical frequency range and to utilize them as oscillation systems for optical lasers @xcite . when used in lasers , such oscillation systems offer a number of serious advantages , among which are low - threshold currents , a high quality of the radiation spectrum , etc . microresonators manufactured as a dielectric disk whose diameter is large as compared to the wave length of the radiation are nothing else but the open quasi - optical dielectric disk resonators which have long been known in the resonator technology for their potential to effectively sustain the electromagnetic ( em ) field inside the resonator volume . the retention of the field is provided due to the total internal reflection ( tir ) from the resonator side boundaries of waves making up resonance oscillations . as a result , em oscillations of whispering gallery ( wg ) type arise . as far as their excitation is not accompanied by the additional dissipative loss , superhigh quality factors are achieved in the ddrs . specifically , in laser systems , in spite of ddr s microscopic dimensions ( the disk diameter is normally about a few @xmath0 m large ) , the quality factors can reach the order of @xmath1 and even more @xcite . the high quality factors of the ddrs , that are often prepared from doped silicon , are generally provided not only due to extremely small loss attainable in this material but also governed by resonator geometry and the perfection of the crystal it is made of . with a certain number of inhomogeneities ( local and/or non - local , random and/or regular ) in the resonator material , the ray picture of em fields in the resonator changes dramatically as against its perfectly homogeneous counterpart . the inhomogeneities give rise to local violation of tir conditions and in this way result in additional energy loss which is evident in the quality factor drop . from the above said there arises the problem of studying the effect produced by random inhomogeneities in the ddrs on their spectral characteristics . in the theoretic analysis of the radiation loss of the ddr two types of inhomogeneities are normally distinguished . to the first type the inhomogeneities of volume nature belong , which are related to regular or random spatial variations of the permittivity in the bulk of the material the ddr is made of . the other class of inhomogeneities includes the so called `` surface '' imperfections normally related to the deviation of the ddr shape from the ideal cylindrical one . the influence of bulk random inhomogeneities on the resonance spectrum was previously studied by the present authors in the particular case of cavity resonators filled with randomly distributed dielectric particles @xcite . it was found that the physical mechanism through which inhomogeneities affect resonator spectrum is basically the intermode scattering . in the case of quasi - optical cavity resonator the peculiar feature of this type of scattering is the selective impact of inhomogeneities on different resonance lines . the most affected lines appear to be those which are the least separated on the frequency axis , whereas solitary lines are subjected to much lesser influence . owing to such a selective effect of random inhomogeneities , the originally dense spectrum of the quasi - optical cavity resonator is considerably rarefied . in refs . @xcite , the investigations were undertaken into the impact of _ surface _ inhomogeneities on spectral properties of open dielectric resonators of cylindrical and spherical shape . the influence of boundary roughness upon the resonance lines was described by means of the simplified quasi - geometric approach where em oscillations scattering due to edge inhomogeneities is taken into account by incorporating into the wave equation the fictitious polarization currents ( pc ) randomly distributed in space @xcite . in particular , within the framework of the model adopted in ref . @xcite the electromagnetic fields close to the resonator side surface were considered as being excited by randomly distributed near - surface current sources whose physical parameters were phenomenologically expressed through statistical characteristics of boundary asperities . it was precisely the radiation produced by these sources that led to the radiation loss of the resonator and , consequently , to the quality factor reduction . such an essentially phenomenological approach to the description of the effect produced by the roughness of open resonator boundaries on its spectral properties can not be reckoned as satisfactory . the volume current method suggested in ref . @xcite , which formed the basis for the pc concept , is , to a large extent , rough and approximate . in this method , the radiation loss is most frequently calculated using green function of the helmholtz equation . the particular form of this function is normally chosen proceeding from the prospective solution of wave equation in the far wave zone . meanwhile , the very notion of the far zone is poorly defined for multiple sources located around the periphery of the quasi - optical ddr we deal with in this particular work . specifically , when deriving characteristic equations for open ddr eigen - frequencies , one needs to join em field components exactly at the boundaries of the system under consideration . at this point a significant uncertainty can arise because the actual external fields subject to _ local _ joining with the internal ones in the presence of surface roughness can deviate considerably from the fields approximated into the boundary vicinity from large distances , i.e. from the far wave zone . to correctly determine the em fields near the random - inhomogeneous resonator surface the theories specially adapted for the description of classic and/or quantum wave scattering by rough interfaces should be applied ( see , e.g. , refs . these theories are generally applicable to the cases where wave scattering by random rough surfaces is in a sense weak . typically , this implies fluctuations of system boundaries to be relatively small in height and sufficiently smooth , so that the applicability of rayleigh hypothesis @xcite was not violated . however , for confined systems like the ddr considered in this work the issues pertinent to wave scattering by surface inhomogeneities appear to be much more complicated . firstly , in practice the inhomogeneities are not always small enough , as well as sufficiently smooth , therefore the conditions for scattering weakness are easily violated . moreover , as the resonance ray trajectories in high-_q _ resonators are periodic , the effect of oscillation scattering caused by the boundary roughness is `` path''-accumulated . scattering can become strong even though the asperities are small - in - height and smooth . this makes the applicability of the above - mentioned theories of rough - surface scattering in the case of high - quality resonators highly questionable . in refs . @xcite , the novel transport theory was developed for waveguide - shaped systems with random rough boundaries . in the framework of this theory it was revealed that the wave scattering resulting from fluctuations of inter - media surfaces can be efficiently described in terms of two physical mechanisms , namely , the _ amplitude _ and the _ gradient _ ones . for the first mechanism it is just the mean - square height of the asperities that serves as a main guiding parameter , whereas for the second one the mean slope of the asperities or , in other words , their sharpness , plays the decisive role . both of these mechanisms contribute additively to the scattering amplitude , but partial probabilities pertaining to them may differ essentially . it was shown in @xcite that , in most cases , the role of the gradient scattering appears to be prevalent . yet there have been no experimental confirmations of this fact in the literature so far . apart from peculiar problems associated with surface nature of scattering in edge - disordered resonators , some more questions arise which need to be resolved when studying the spectra of ddrs subject to surface inhomogeneities . among these questions , particularly , are those related to the vector nature of fields which are excited in these essentially non - integrable systems . it is known that even for ideal cylindrical ddrs the electrical- and magnetic - type oscillations can not be separated in the strict sense . they always remain to the certain extent intermixed @xcite . this fact renders theoretic analysis of such systems spectra rather sophisticated . spectral properties of ddrs without random inhomogeneities were examined in a number of theoretical papers ( see , e.g. , refs . @xcite ) . yet , in deriving characteristic equations some inexact _ a priori _ assumptions where used , which still require both theoretical grounding and experimental corroboration . in our present study , one of the goals is to investigate theoretically the physical mechanisms responsible for widening resonance lines of dielectric microresonators with random surface inhomogeneities . in view of such systems being non - integrable , it is necessary to elaborate an appropriate theoretical model allowing for sufficiently accurate determination of both the frequency spectrum and the quality factors of resonance lines . yet another goal of this study is to corroborate experimentally which of the physical mechanisms does play a dominant role in the scattering of em oscillations excited in microresonators with random rough side boundaries . from the theoretical viewpoint , the spectrum analysis in our study is carried out in terms of scalar potentials , specifically , electrical and magnetic hertz functions @xcite . we formulate the conditions wherein these potentials make independent contributions to em fields in the resonator , which is equivalent to decoupling the oscillations of te and tm polarization . the helmholtz equation for hertz potentials in an irregular - shaped open resonator is equivalent to schrdinger equation for electrons moving in the piecewise continuous space subject to random potential . owing to this one can extend the results obtained in the present study to quantum systems as well , in particular , to open quantum dots having random rough boundaries . to solve wave equations in the rough - bounded resonator , the eigen - mode separation method is used , previously developed with reference to waveguide - like systems subject to arbitrary static potential @xcite ) . using this method , the originally posed statistical problem of determining the fields in three - dimensional ddr with complex , randomly rough , side boundary is rigorously reduced to the set of one - dimensional dynamic equations that contain some _ effective _ random potentials . we show that under the conditions where gradient scattering mechanism is dominant , the boundary roughness effect on the resonator spectrum can be described through certain renormalization of mode wave numbers and azimuth indices of the bessel functions in the characteristic equation . the values of renormalized wave numbers and mode indices decrease as the asperities get sharpened , which is consistent with a decrease in the resonator _ q_-factor . we are thus led to the conclusion that the observed reduction of resonant line quality factors results not from the extra dissipative loss but rather from em field intermode rayleigh scattering induced by random surface inhomogeneities . the imperfection of the resonator shape results in the local violation of the tir conditions . this leads to additional radiation loss of em energy and , hence , to a decrease in the level of localization of em fields inside the resonator . since characteristic dimensions of surface inhomogeneities in actual microresonators are always quite small ( of the order of nanometers ) , to verify our theoretical findings experimentally we have decided upon the method of simulation with macroscopic devices . as a model system we have employed a millimeter wave quasi - optical resonator made of a circular teflon disk . wg oscillations of te and tm types were excited in the disk using a special waveguide antenna . the inhomogeneities of the resonator side boundary were made in the form of teflon bracket - bars randomly attached to the outside cylindrical surface . our experimental results have demonstrated excellent qualitative agreement with the developed theory as regards spectral lines widening caused by the resonator side boundary roughness . furthermore , the relatively simple model of the ddr field distribution , which was adopted in our study , enabled us to calculate both the _ q_-factors and the frequencies of the resonance lines with quite satisfactory accuracy . the calculations appeared to be in fair conformity with our experimental data . consider an open disk resonator as a finite - height cylinder made of the dielectric material with permittivity @xmath2 . plain - parallel end boundaries of the cylinder traverse the central axis ( @xmath3 ) at points @xmath4 ( see fig . [ fig1 ] ) , the side boundary ( @xmath5 ) is formed by the generatrix passing parallel to @xmath3-axis along closed contour @xmath6 [ .9 ] and @xmath7 point out the local normal directions to the rough ( @xmath5 ) and to the smoothed resonator side surface , respectively . [ fig1],title="fig : " ] whose distance from the central axis is given by @xmath8\ , \ ] ] @xmath9 is the radius of the averaged ( i.e. , ideally circular ) contour @xmath10 containing no random bends . the function @xmath11 will be regarded as the gaussian random process with a zero mean value , @xmath12 , and binary correlation function @xmath13 here , @xmath14 is the mean - square height of boundary asperities , @xmath15 is the dimensionless function which has a unit maximal value at zero argument and falls to parametrically small values at angle distance @xmath16 ( the correlation angle ) . the angle brackets in eq . denote statistical averaging over the ensemble of realizations of random function @xmath11 . in what follows , for the estimation purposes , along with angle parameter @xmath17 we will use another correlation parameter , viz . , arc correlation length @xmath18 . both the random function @xmath11 and the regular @xmath15 will be thought of as periodic with period @xmath19 . it is well known that in general the em field of an open dielectric resonator can be given as a superposition of oscillations of electric and magnetic types ( tm and te polarized , respectively ) @xcite . the vector fields of both of these types can be expressed in terms of scalar potentials , @xmath20 and @xmath21 ( the so - called electric and magnetic debye potentials ) , which meet the same helmholtz equation but are subject to different joining conditions at the resonator boundaries . in arbitrarily shaped resonators tm and te oscillations are essentially intermixed , being strictly decoupled only in the case of sufficiently symmetric systems such as , e.g. , an infinite dielectric cylinder @xcite or a dielectric sphere @xcite . below it will be shown that tm and te oscillations can also be decoupled ( with high accuracy , though approximately ) in the case of the cylindrical ddr of arbitrary thickness . this fact opens an opportunity to extend the conclusions of the present work to electronic microresonance systems , e.g. , partially open quantum dots . in studying oscillations in the ddr with randomly rough side boundaries we first reduce the problem of wave scattering at the boundary to the problem of scattering in the bulk of the resonator of ideal circular form . the helmholtz equation for scalar wave field @xmath22 , whose role is played by one of the above - mentioned potentials , after rewriting it in cylindrical coordinates and using the conformal coordinate transformation , @xmath23 is reduced to the form @xmath24 \psi(r,\varphi , z)=0\ ] ] ( the tilde signs over coordinate variables from here on are omitted ) . here , @xmath25 , @xmath26 1&,\qquad ( r , z)\not\in\omega \end{array}\right.\ , \ ] ] @xmath27 is the phenomenological frequency parameter which takes into account dissipative and other uncontrollable losses in the system , except for the radiation loss , @xmath28 is the bulk region occupied by the dielectric . effective potentials @xmath29 and @xmath30 in eq . are the operators whose coordinate representation reads [ potentials ] @xmath31 \left[\beta^2(\varphi)-1\right]\ , \\ \label{slopepot } & \hat{v}^{(s ) } = \left[\frac{\xi'(\varphi)}{r\beta(\varphi ) } \frac{\partial}{\partial\varphi}+\frac{\partial}{\partial\varphi } \frac{\xi'(\varphi)}{r\beta(\varphi)}\right]\frac{1}{r}\frac{\partial}{\partial r } -\left[\frac{\xi'(\varphi)}{r\beta(\varphi)}\right]^2 \frac{1}{r}\frac{\partial}{\partial r}r\frac{\partial}{\partial r}\ , \\[6pt ] \notag & \beta(\varphi)=1+\xi(\varphi)/r\ .\end{aligned}\ ] ] indices `` @xmath32 '' and `` @xmath33 '' specifying potentials indicate that the corresponding potential is mainly governed either by fluctuations in the asperity height ( i.e. by height function @xmath11 ) or by fluctuations in the asperity slope ( i.e. , by slope function @xmath34 ) . such a subdivision of the potentials , with regard to the asymptotic suppression of correlations between functions @xmath11 and @xmath34 over angle intervals that are large as compared to the correlation angle , urges one to consider potentials and as corresponding to different physical wave - surface scattering mechanisms @xcite . we will below refer to these mechanisms as the amplitude ( or the height , _ h _ ) and the gradient ( or the slope , _ s _ ) scattering mechanisms , respectively . subsequently we will examine the boundary asperities which are sufficiently small in hight so as to meet inequality @xmath35 in contrast to the widespread belief ( see , e.g. , ref . @xcite ) , this does not necessarily imply that the surface - roughness - induced scattering be regarded as weak . we note , for example , that the local value of the potential is estimated by the parameter @xmath36 , which may take an arbitrary absolute value provided the condition is met . the true conditions for the scattering to be classified as weak will be provided below , based on the operator technique applied to perform the principal calculations . the potentials @xmath29 and @xmath30 in eq . , that account for inhomogeneity of the resonator side - boundary , are defined not quite conveniently from the viewpoint of subsequent use of perturbation theories . the inconvenience relates to non - zero average value of the potential . by separating this average , we can rewrite eq . as @xmath37 \psi(r,\varphi , z)=0\ , \ ] ] where two different slope potentials are introduced instead of potential eq . , which have zero mean values , viz . [ slopepots - defs ] @xmath38\frac{1}{r}\frac{\partial}{\partial r}\ , \\ \label{slpot-2 } \hat{v}^{(s2 ) } = & -\left[{\xi'}^2(\varphi)/r^2-\xi^2\right ] \frac{1}{r}\frac{\partial}{\partial r}r\frac{\partial}{\partial r}\ .\end{aligned}\ ] ] the parameter @xmath39 in eqs . and , which is defined as @xmath40 ^ 2\big > } } \sim\frac{\sigma^2}{s_c^2}\ , \ ] ] serves as a measure for the mean - square slope of boundary asperities . considering that irregularities of the resonator side surface are aligned paraxially , they can not result in additional scattering along axis @xmath3 except for the partial reflection produced by the initial step change in this direction of the permittivity of the medium . this makes it possible to remove from eq . the dependence on the coordinate @xmath3 with the model method similar to the one we use when analyzing the spectrum of homogeneity - free ddr ( see appendix [ unperturbed_spectrum ] ) . specifically , we assume the dependence of the mode wave function on the coordinate @xmath3 to have the same form used to find @xmath33- and @xmath41-solutions for the ideal - shaped resonator ( for the particular @xmath42-symmetric solution this dependence is given in eqs . and ) . afterwards , equation , being divided by the factor of @xmath43 , is reduced to the following form , @xmath44 \psi(r,\varphi|z_{\scriptscriptstyle\lessgtr})=0\ , \ ] ] where [ renormed_parameters ] @xmath45k^2-k_z^2 & \ \text{,}&\qquad |z|&<h/2 \\ k^2+\varkappa_z^2 & \ \text{,}&\qquad |z|&>h/2 \end{array}\ , \right.\\ \label{v(h)-renorm } \widetilde{{v}}^{(h)}(z_{\scriptscriptstyle\lessgtr})&= -\widetilde{k}_{\perp}^2(r|z_{\scriptscriptstyle\lessgtr})\left[\beta^2(\varphi)-1\right]\ .\end{aligned}\ ] ] the tilde signs over the potentials in eq . denote their renormalization by the factor of @xmath46 . equation actually describes two - dimensional fields because symbol @xmath47 acts not as the current axial coordinate but stands for some axial index specifying in which domain of the @xmath3-axis either @xmath48 or @xmath49 the solution to eq . is sought . in the azimuth mode representation , equation assumes the form @xmath50 \psi_{n}(r|z_{\scriptscriptstyle\lessgtr})-\sum_{m\neq n } \widetilde{\mathcal{u}}_{nm}(r|z_{\scriptscriptstyle\lessgtr})\psi_{m}(r|z_{\scriptscriptstyle\lessgtr})=0\ .\ ] ] here , @xmath51 is the matrix element of the entire potential @xmath52 , which is taken between eigen - functions of the azimuth part of the laplace operator ( see appendix [ unperturbed_spectrum ] ) , @xmath53 , the index @xmath54 changes to @xmath55 . later we will refer to matrix elements @xmath56 and @xmath57 as the intra- and inter - mode potentials , respectively . in the general case it is rather difficult to immediately solve the infinite set of coupled equations . however , the solution can be obtained in terms of the operator technique previously developed by the present authors with reference to waveguide - type random systems of arbitrary dimensionality @xcite and advantageously applied afterwards to the analysis of bulk - disordered cavity resonators @xcite . the above technique , as applied to the problem touched upon in the present paper , is adapted in appendix [ mode_separation ] . the advantage of the technique is that it can be used to derive precise closed equations for wave functions of each of the azimuth modes , namely , @xmath58 \psi_{n}(r|z_{\scriptscriptstyle\lessgtr})=0\ .\ ] ] here , along with local intra - mode potential @xmath56 , the operator potential @xmath59 arises , which rigorously allows for the inter - mode scattering . the structure of this potential , though well - recognized , is quite complicated to be operated with at an arbitrary scattering intensity . yet , the estimations we provide in the next section substantially simplify the _ t_-potential in eq . in different limiting cases , and in this way they allow one to obtain the oscillation spectrum of a randomly rough ddr in almost all physically sensible situations . hereinafter we will refer to the system as weakly rough one if r.m.s . height of its boundary asperities meets inequality . as the potentials in eq . are of the operator nature , we will estimate their strength using standard definition of the operator norm @xcite . for our purposes the formula @xmath60 is the most appropriate , in which the parenthesis symbolize the scalar product on the functional space @xmath61 consisting of the class of solutions to eq . with no random potentials . to estimate the potentials in eq . we first calculate their azimuth matrix elements . for relatively small - height asperities the `` height '' potential @xmath62 equals approximately @xmath63 . mode matrix elements of this potential are @xmath64 @xmath65 is the fourier transform of the function @xmath11 . matrix elements of the `` slope '' potentials @xmath66 and @xmath67 are equal , respectively , [ unm - alt ] @xmath68 \frac{1}{r}\frac{\partial}{\partial r}r\frac{\partial}{\partial r}\ .\end{aligned}\ ] ] we will ignore the role of the potential for the intra - mode scattering proceeding from the following considerations . it is evident that under condition the uniform azimuth mode of this potential can result , at most , in relatively small ( @xmath69 ) renormalization of the unperturbed `` in - plane energy '' @xmath70 . yet , in the case we consider below even this does not occur . we will regard the asperities as not only being small in height but also as small - scaled ones in the sense that the inequality holds @xmath71 in this limit the uniform mode of the random process @xmath11 with parametric accuracy ( @xmath72 ) can be put equal to zero , since the process is nearly ergodic within the interval of the angle variable change . gradient potential is not involved in the intra - mode scattering by its definition , so that only the potential should be taken into account , where one must let @xmath73 . we use as trial functions in eq . the solutions of eq . with random potentials equal to zero , i.e. , bessel function @xmath74 $ ] in the interval @xmath75 and hankel function @xmath76 $ ] in the domain @xmath77 . such a choice is justified provided the intra - mode scattering has a slight impact on the mode energies . then we arrive at the following estimate for the potential @xmath78 norm , @xmath79 ^ 2\big>}}\ .\ ] ] after averaging eq . using correlation equality @xmath80 which immediately stems from eq . ( @xmath81 is the fourier transform of correlation function @xmath15 ) , we obtain @xmath82 it can be readily seen that the factor standing at @xmath83 in the right - hand side of eq . is , in view of eq . , small as compared to unity . this substantiates the above assumption about the intra - mode scattering weakness . the inter - mode scattering rate in our resonator is up to the norm of the operator @xmath84 entering _ t_-matrix . _ the `` amplitude '' inter - mode scattering . _ when estimating the norm of the `` height '' item @xmath85 in the operator @xmath84 one is faced with a need to evaluate the expression @xmath86 ( in order not to overload subsequent formulae we omit axial index @xmath47 , as this can not lead to misunderstanding ; function @xmath87 is introduced in appendix [ mode_separation ] ) . the simultaneous presence in eq . of both the integrals over radial coordinate @xmath88 and the sums over mode indices stems from the definition of the functional space where the operator @xmath84 is effective . in view of eqs . and the correlator in the integrand of eq . is calculated to @xmath89 by substituting this into eq . and assuming the asperity correlation function to have the gaussian form , viz . @xmath90 , we arrive at the following estimate for the height term in the inter - mode scattering operator , @xmath91 the rayleigh parameter @xmath92 , to whose square the right - hand side of eq . is proportional , may take small or large values , depending upon the relationship between mean - square height of the asperities and the wave length of the excited oscillations . according to the value of this parameter we will distinguish between weak ( @xmath93 ) and strong ( @xmath94 ) inter - mode scattering caused by the `` height '' potential . _ 2 . the `` gradient '' inter - mode scattering . _ this type of scattering between different azimuth modes is related to the availability in eq . of the potentials . their correlators needed to estimate the operator norms of items @xmath95 and @xmath96 in the operator @xmath84 , are equal , respectively , [ corrukm1ukm2_(ab ) ] @xmath97 { \ensuremath{\big<\widetilde{\mathcal{u}}_{km_1}^{(s2)}(r_1){\widetilde{\mathcal{u}}_{km_2}^{(s2)*}}(r_2)\big>}}= & \frac{1}{(2\pi)^2(1+\xi^2)^2r^4 } \left(\frac{1}{r_1}\frac{\partial}{\partial r_1}r_1\frac{\partial}{\partial r_1}\right ) \left(\frac{1}{r_2}\frac{\partial}{\partial r_2}r_2\frac{\partial}{\partial r_2}\right ) \notag\\ \label{corrukm1ukm2(s2 ) } & \times \sum_{l_1,l_2=-\infty}^{\infty } ( k - l_1)(l_1-m_1)(k - l_2)(l_2-m_2 ) { \ensuremath{\big<\widetilde{\xi}(k - l_1)\widetilde{\xi}(l_1-m_1 ) \widetilde{\xi}^*(k - l_2)\widetilde{\xi}^*(l_2-m_2)\big > } } \ .\end{aligned}\ ] ] assuming the function @xmath11 to be the gaussian random process , the equality can be partially simplified as the double sum over @xmath98 and @xmath99 is reduced to the single one , @xmath100 then the alternate substitution of correlation functions into eq . in place of the height potential correlator results in the following estimation formulas for operators @xmath95 and @xmath96 , [ r(s1s2)_nrm - estim ] @xmath101 \label{r(s2)_nrm - estim } { \ensuremath{\big<\big\|\hat{\mathsf{r}}^{(s2)}\big\|^2\big > } } & \sim \left(\frac{\xi^2}{1+\xi^2}\right)^2(\widetilde{k}_{\perp}r)^2\ . \end{aligned}\ ] ] the collation of norms and reveals that the height and the gradient scattering mechanisms , given their effect is evaluated as a function of roughness statistical parameters and radial components of the azimuth mode wave vectors , can essentially compete against one another . however , the general statement boils down to the fact that it is the gradient scattering which is prevalent in the most part of the parameter region . the amplitude scattering mechanism can also dominate , but this is the case for quite smooth asperities only and for oscillations of very large azimuth indices . one can conclude from eqs . and that for the amplitude scattering to be dominant it is necessary , first , that the average tangent of the asperity angle of slope obey the condition [ height_dominate ] @xmath102 secondly , the inequality must be fulfilled @xmath103 , which is admissible for oscillations with quite large mode indices , i.e. , @xmath104 if only one of the inequalities fails to hold ( indeed , this is the case for the overwhelming majority of parameters from the above - mentioned set ) , the gradient scattering appears to be dominating . based on the above estimations , equation , that governs the azimuth mode spectrum of the resonance system under study can be substantially simplified if one considers the limiting cases of weak and strong scattering . as is seen from eq . , for small r.m.s . height of the asperities ( in the sense of inequality ) the intra - mode scattering due to potential @xmath56 is weak , thereby making it possible to disregard it in the main approximation . as far as the inter - mode scattering is concerned , it is not straightforward to estimate it in the same way as the intra - mode one . from eqs . and it follows that inter - mode scattering should be classified either as weak or strong one depending upon whether the operator @xmath84 norm is small or large as compared with unity . if the inter - mode scattering resulting from both the amplitude and the gradient potentials is thought of as weak one , which occurs when two inequalities hold simultaneously [ weak_intermode_scatt ] @xmath105 \label{weak_intermode(s1 ) } \frac{\sigma}{s_c } \ll \frac{s_c}{r}\ , \end{gathered}\ ] ] the potential @xmath106 in eq . can be ignored with parametric accuracy . in this instance the resonator spectrum can be obtained from the equation which differs from the initial unperturbed one simply by renormalizing both the mode index ( @xmath107 ) and the radial wave number ( @xmath108 ) . in the parameter region where both of the inequalities are satisfied this renormalization is extremely small ( @xmath109 ) and can be safely disregarded . conditions , that indicate the inter - mode scattering weakness , can be easily violated . by virtue of peculiar technology , when manufacturing micro- and nano - sized quantum resonance systems it is normally difficult to obey , e.g. , inequality which corresponds to extreme smoothness of surface inhomogeneities . moreover , inequality is fulfilled within only limited , i.e. , long - wavelength , part of the resonator bandwidth . if , at least , one of the conditions is violated , the inter - mode scattering fails to be classified as a weak one in the sense that the norm of scattering operator @xmath84 in the potential becomes large as against the unity . nevertheless , in this case one may as well simplify the _ t_-potential by expanding it in a series in the inverse scattering operator , @xmath110 . the expression between the projection operators in eq . is identically transformed as @xmath111 here we use the symbolic notation @xmath112 for the full ( i.e. , including the whole set of azimuth modes ) green operator with no intermode potentials . as a result of `` coating '' eq . with the projection operators @xmath113 the first term in its right - hand side is eliminated , and the operator acting on the wave function in eq . assumes the following approximate form , @xmath114\ .\end{aligned}\ ] ] by comparing the expression in the right - hand side of eq . with the one in its left - hand side , with potentials @xmath115 and @xmath106 set to zero , one can notice that the limiting cases of weak and strong ( in the sense of the operator @xmath84 norm ) inter - mode scattering differ from one another solely by doubling the wave operator in the latter case . clearly , such doubling implies that in strong surface scattering the amplitude of the excited oscillations is twice as small as the one in the weak scattering limit . however , it is evident that the change in the common factor multiplying the wave operator can not reveal itself in dispersion relations . we thus have demonstrated that the formal algebraic structure of the wave operator remains basically unchanged under conditions of weak and strong scattering . in both of these cases the wave operator differs slightly from its unperturbed form . the main difference between dispersion relations for resonators with perfect and rough side boundaries is that in the latter case the initial transverse wave parameter @xmath116 and the mode index @xmath54 are renormalized by the gradient factor of @xmath117 . this enables us to immediately write down the dispersion equations for a ddr with random inhomogeneous side boundary based upon the results given in appendix [ unperturbed_spectrum ] . specifically , for a rough - bounded ddr the oscillation spectrum is governed by the equation @xmath118 \bigg[\frac{1}{\widetilde{k}_{\perp}^{\varepsilon}}\cdot \frac{{j_{\widetilde{n}}}'(\widetilde{k}_{\perp}^{\varepsilon}r ) } { j_{\widetilde{n}}(\widetilde{k}_{\perp}^{\varepsilon}r ) } - \frac{1}{\widetilde{k}_{\perp}}\cdot\frac{{h_{\widetilde{n}}^{(1)}}'(\widetilde{k}_{\perp}r ) } { h_{\widetilde{n}}^{(1)}(\widetilde{k}_{\perp}r)}\bigg]= \frac{\widetilde{n}^2}{r^2}(\varepsilon-1)^2 \frac{k_z^2\widetilde{k}^2}{{(\widetilde{k}_{\perp}^{\varepsilon})}^4\widetilde{k}_{\perp}^4}\ , \end{aligned}\ ] ] which has to be supplemented with additional relationships between @xmath119 and @xmath120 , the latter resulting from joining the fields at the end interfaces . in eq . , the notations are introduced @xmath121 , the wave numbers @xmath122 and @xmath123 are defined in eq . . as the additional connection between wave parameters @xmath119 and @xmath120 in the case of a perfect cylindrical resonator equation is obtained for @xmath42-symmetric oscillations . for @xmath42-antisymmetric oscillations eq . holds . one can easily check that in going from unperturbed wave equation to equation describing the ddr with rough side wall the equations derived through joining em field components at the end boundaries of the dielectric disk remain unchanged . finally , proceeding from the above calculations we are led to conclude that with any roughness of the ddr side wall , which is basically restricted by the smallness condition , to obtain the oscillation spectrum one can make use of the relationship eq . mainly coincident in form with the dispersion equation for an infinitely long dielectric cylinder . as a supplementary condition to interconnect the lengthwise and transverse wave vector components equation or should be applied , depending upon @xmath42-symmetry of the desired solution . the fundamental difference of eq . from its perfect - cylinder counterpart is the renormalization of basic wave parameters , which is governed by geometric properties of the asperities . a special emphasis should be placed on the fact that it is exactly the gradient scattering mechanism , not the amplitude one , that most crucially affects the ddr spectrum . this particular fact makes itself evident in that the renormalization of wave parameters in eq . is dictated not by the height - type parameters , such as , e.g. , rayleigh parameter @xmath92 or the ratio @xmath124 , but is mainly regulated by the mean - square slope of the asperities against the unperturbed resonator boundary , which is specified by parameter @xmath36 . the main goal of experimental studies in this work was to validate our theory as regards microresonator spectra and the effect of random surface inhomogeneities on them . the point is that essential assumption adopted in the theory is that it virtually does not consider electromagnetic fields radiated from the resonator external edges into the corner regions labeled by numbers 4 and 4@xmath125 in fig . [ fig5 ] ( see appendix [ unperturbed_spectrum ] ) . at the same time , without making quite complex calculations one can not make _ a priori _ statements about these fields being small enough to neglect them in calculating the resonator spectrum . yet another goal of the experiment was to examine our theoretical findings concerning the physical mechanism that adequately describes the influence of random surface inhomogeneities on microresonator spectral properties . the studies of the microresonator spectrum were performed through modeling these quite small systems in the millimeter wave band . for this purpose we used a quasioptic dielectric disk resonator . physically , oscillation properties of this macroscopic system are identical with properties of silicon microresonators used as oscillation systems in real optical lasers . our resonator was made of teflon whose permittivity is not significantly far from unity ( @xmath126 ) and whose dielectric loss in the millimeter range are fairly small ( @xmath127 ) . in the model ddr , whispering gallery oscillations were excited with the em field concentrated at the periphery of the disk , in the narrow region close to its side boundary . this enabled us to use the disk core to fix it in the level position without introducing additional dissipative losses to the experiment . the source of wg modes was positioned close to the resonator side boundary . its role was played by the waveguide antenna powered by the microwave generator . the antenna was fabricated as a waveguide tapered along the short wall , whose butt end was positioned near the resonator side surface . the receiving antenna was made identical to the source antenna and was placed at the diametrically opposite point off the disk . thin dielectric bracket - bars were used as the inhomogeneities of the resonator side boundary . they were attached to the ddr side surface . the basic requirement imposed on the inhomogeneities was for them not to cause noticeable additional dissipative losses . to this end , the bracket - bars [ .75 ] the gap between the antenna and the ddr s side surface , which is used to tune the coupling of antenna with the resonator . the diameter of the ddr is 102 mm , the thickness is 7.6 mm . the inhomogeneities dimensions are : the length , the width , and the thickness 7.6 mm , 3 mm , and 2 mm , respectively . [ fig2],title="fig : " ] were made of the same teflon as the resonator body . the inhomogeneities distribution on the resonator side surface was random and varied for each different realization . in fig . [ fig2 ] , the general view of the resonator is shown along with the exciting and the receiving antennas as well as with the attached teflon bracket - bars . in order to excite te or tm oscillations in the resonator , two different configurations of antenna - versus - resonator were used . for te oscillations the antenna magnetic field was directed along the resonator axis , @xmath3 . to achieve this , the wide plate of the waveguide was aligned parallel to this axis . in the case of tm oscillations , along the axis @xmath3 the electric field was directed . for this purpose the wide side of the antenna was oriented transversely . by changing the distance between the exciting waveguide butt end and the resonator side boundary , as well as the angle between them , we were able to adjust the coupling between the antenna and the resonator to optimize it . as the optimal coupling we accepted the one whereby the additional loss caused by the antenna was much less than the eigen - loss in the resonator . at the same time , it was necessary to keep the level of the coupling sufficient for spectral lines to be traceable . the pattern of whispering gallery em fields in the ddr in known to be very sensitive to the frequency variation . therefore the resonator - to - waveguide coupling is different for each of the modes . based upon this , we established the optimal coupling for each of the spectral lines separately while carrying out the measurements in a wide frequency range . spectral measurements with the model resonator were made in the on - pass regime using millimeter waveband standing - wave ratio meter . since the experiments were conducted over a wide range of frequencies and for large number of realizations of surface inhomogeneities , all spectral measurements were rendered automatic . the signal from the ratio meter was sent to the computer and processed by means of the especially designed program to find both the frequencies and the quality factors of resonance lines . the accuracy of the measurements was 0.01% and 5% for the resonance frequencies and the quality factors , respectively . to identify the spectral lines , it was necessary to determine the value of azimuth index for each of the lines in the absence of inhomogeneities . to this end , the miniature rotating probe made of a thin metal plate was used , which we inserted into the region at the resonator disk where the electric field antinode was positioned . in so doing , the source was tuned to the frequency of the particular spectral line . when the probe rotated about the resonator axis , the signal registered by the receiver varied in time at the rate the probe passed across the regions with electric field loops . this enabled us to determine the desired mode index through the measurements of the signal modulation frequency . in fig . [ fig3 ] , the results of numerical calculations of the resonator spectrum are shown along with spectral measurements data . as seen from fig . [ fig3]a , the measured spectra of te and tm modes correlate well with the calculated spectra . the difference between spectral line frequencies found from eq . , including intrinsic . for @xmath128 mode : 1 calculation , 2 experiment ; for @xmath129 mode : 3 calculation , 4 experiment . [ fig3],title="fig : " ] . for @xmath128 mode : 1 calculation , 2 experiment ; for @xmath129 mode : 3 calculation , 4 experiment . [ fig3],title="fig : " ] dissipation loss in the dielectric , and experimentally measured frequencies is no more than 1% . the quality factors obtained in the experiment ( as shown in fig . [ fig3]b ) are smaller than those calculated theoretically . we attribute this to the fact that the measured @xmath130-factor includes not only eigen - loss in the resonator material but also is affected by the loss resulting from coupling with the antenna . from the diagrams in fig . [ fig3 ] one can observe the nearly equidistant character of the spectrum ( the frequency interval @xmath131 ghz ) , which is typical for open resonators with wg - type oscillations @xcite , as well as the exponential dependence on the mode index @xmath54 of the calculated and the measured @xmath130-factors . our calculations suggested , and that was experimentally confirmed , that for tm oscillations the resonator quality factor is significantly larger than that for te oscillations , both of them being taken with the same azimuth index . this suggests that in the case of tm oscillations the ddr has the property to more efficiently retain electromagnetic field in its volume than the resonator with te oscillations does . we have thus established that both the theoretical model of the resonator we have chosen for our calculations and the characteristic equations obtained thereupon have found impressive experimental confirmations . according to the findings of our theory , the main physical mechanism for em field scattering by random inhomogeneities of the resonator side boundaries is the gradient mechanism . the scattering arisen due to fluctuations in the asperity slope can be allowed for by way of modification of the cylindrical functions indices along with the wave numbers in the characteristic equation eq . . the modification reduces to multiplying the parameters @xmath54 , @xmath119 , @xmath123 and @xmath122 by factor of @xmath132 . in order to identify the main scattering mechanism experimentally and thus to check the developed theory it was necessary to estimate the value of the parameter @xmath39 starting from the parameters relevant to a particular experiment . we were governed by the following considerations . setting the correlation length equal to an average distance between the centers of the attached dielectric bracket - bars , i.e. , @xmath133 , where @xmath134 is the total number of the bracket - bars , with gaussian distributed inhomogeneities we obtain @xmath135^{-1/2}\ .\ ] ] it can be easily seen that the parameter @xmath136 tends to decrease with an increasing number of bracket - bars , which leads to a decrease in the effective wave number @xmath137 and the effective mode index @xmath138 . since the dependence of the quality factor on the mode index is nearly exponential , @xmath139 , the availability of surface inhomogeneities should result in the decrease in the resonator quality factor , whose origin is not an additional dissipative loss . according to our theory , with a small number of inhomogeneities the effect of slope - controlled scattering must be fairly slight ( the parameter @xmath39 decreases with lowering @xmath134 ) . therefore the value of quality factor must be weakly dependent on @xmath134 as well . such was indeed the case in our experiments whose results are depicted in fig . [ fig4 ] . [ .58]-factor curves versus the number of inhomogeneities on the resonator side surface . @xmath140 oscillations : 1 theory , 2 experiment ; @xmath141 oscillations ; 3 theory , 4 experiment . calculations were carried out at @xmath142 mm and @xmath143 mm . [ fig4],title="fig : " ] in the same figure , the curves @xmath144 are plotted for comparison , which are calculated from characteristic equation including modification factor @xmath136 and the dissipation loss in the resonator material . the loss was taken into consideration phenomenologically , by adding the imaginary part to permittivity @xmath145 , see eq . . with the statistical nature of our measurements ( the averaging is done over a large number of realizations of the bracket - bar set ) , the agreement between theoretical and experimental results appears to be quite satisfactory . this suggests that , since both in our theory and in the numerical calculations based upon it the slope scattering mechanism alone is taken into account , the qualitative agreement between theory and experiment unambiguously corroborates the dominant role of this particular type of wave - surface scattering . as is also seen from fig . [ fig4 ] , the influence of surface inhomogeneities on the ddr spectrum differs significantly for te- and tm - type oscillations . the quality factors of te oscillations are to a far larger extent subjected to the resonator boundary roughness as against the factors of tm oscillations . with a given number of inhomogeneities , the @xmath130-factor of te oscillations is noticeably smaller than that of tm oscillations . this fact , which is quite important for practical applications of microresonators in laser oscillation systems , was given certain attention in ref . yet in the present work it has been comprehensively substantiated . to summarize , we have investigated spectral properties of a dielectric disk resonator with randomly rough side boundary both theoretically and experimentally . it is shown that the azimuth modes of the resonator oscillations can be rigorously separated at whatever level of surface inhomogeneities . this allowed us to obtain the asymptotically exact dispersion equations which are valid over a wide range of the roughness parameters . the only requirement imposed on the inhomogeneities and effectively utilized in the calculations was that their mean - square height be small as compared with the unperturbed ( i.e. , non - rough ) disk resonator radius . in deriving dispersion relations , we have shown that electromagnetic wave scattering resulting from the boundary roughness can be described in terms of two fundamentally different physical mechanisms , specifically , the amplitude ( the height ) scattering mechanism and the gradient ( the slope ) mechanism . for the first of them , the ratio of the mean height of the asperities to the oscillation wavelength ( the rayleigh parameter ) acts as the main guiding parameter whereas for the gradient mechanism the mean slope of the asperities relative to the unperturbed resonator boundary plays the same role . our estimations have revealed that it is exactly the gradient scattering that is of primary importance for the formation of rough resonator spectrum . the effect of this type of scattering can be described through the approximate dispersion equation that differs from the analogous equation for the ideally circular ddr by the gradient renormalization of the basic wave parameters , i.e , , the mode wave numbers and the azimuth index . our theory aimed at describing the influence of random surface inhomogeneities on microresonator spectral properties analytically is actually of model nature . the point is that it does not incorporate the electromagnetic fields radiated from the resonator into external angle sectors . to neglect these fields theoretically is quite a challenge . therefore , to verify the conclusions of our approximate ( to the certain extent ) theory we have made experimental measurements on the model millimeter wave band dielectric disk resonator . surface inhomogeneities were presented by teflon bracket - bars of relatively small cross - section , which were attached randomly to the resonator side boundary . for the perfect - wall resonator , i.e. the resonator with no inhomogeneities attached , our measurement data have demonstrated excellent agreement with the developed theory concerning both the frequency spectrum and the quality factors of spectral lines . close qualitative agreement is also attained with regard to the effect produced on the resonator spectrum by random surface inhomogeneities placed at the side boundary . first , we have confirmed the theoretical predictions about the leading role of the gradient scattering mechanism in describing the effect of random rough boundaries . second , we ascertained that the system of model dispersion equations we have obtained herein is quite effective in describing the open microresonator spectra . besides , our experiments have revealed that the effect produced by the surface inhomogeneities of the ddr on the te and tm oscillation spectra is fundamentally different . the tm oscillations quality factor exceeds significantly the analogous factor for the te oscillations , being yet less affected by surface inhomogeneities . in our work this particular fact , which is essential for the production of microresonator - based lasers , has been theoretically and experimentally validated . this work was partially supported by the science and technology center of ukraine ( stcu ) , project no . proceeding from the methodology considerations , we outline the technique for deriving dispersion equations for cylindrical finite - size resonator with perfectly smooth bounding surfaces . consider equation without potentials @xmath29 and @xmath30 . upon going to fourier representation over angle variable @xmath146 using the complete set of eigenfunctions @xmath147 , where @xmath148 , the equation for @xmath54-th angular component of the wave function becomes @xmath149\psi_n(r , z)=0\ .\ ] ] the function @xmath150 is sought to be finite as @xmath151 , whereas at @xmath152 and @xmath153 the radiation conditions are meant to be fulfilled . it is difficult to find the explicit solution of eq . in the entire domain of variables @xmath88 and @xmath3 . yet basically there is no need to make use of such a solution . it will suffice to obtain dispersion relations which are not strictly valid but satisfied with good accuracy . we will seek a desired solution in the model form , as was done , e.g. , in refs . @xcite , imposing the pair of basic requirements . one of the requirements is to fulfil fundamental boundary conditions at zero and infinite distances from the resonator center , while the other is to provide correct joining of the em field components at the dielectric disk interfaces . by representing the solution of eq . as a sum of @xmath42-symmetric and @xmath42-antisymmetric summands ( hereinafter we will specify them by indices ( @xmath33 ) and ( @xmath41 ) , respectively ) we will seek the debye potentials in the form given below , [ un(s ) ] @xmath154 [ vn(s ) ] @xmath155 in eqs . and , the subscripts 1 , 2 , 3 [ .7 ] between which the joining of em fields is performed . the shaded region is filled up with the dielectric . [ fig5],title="fig : " ] correspond to the regions ( see fig . [ fig5 ] ) which are labeled with corresponding numbers . the wave parameters @xmath123 , @xmath122 and @xmath156 are given by [ wave_parameters ] @xmath157 by representing em field components in terms of the potentials and and by joining the in- and the out- field components at the resonator side boundary ( i.e. , between regions 1 and 2 ) we obtain the well - known equation that couples together wave parameters @xmath119 and @xmath120 @xcite , @xmath158 \bigg[\frac{1}{k_{\perp}^{\varepsilon } } \frac{{j_n}'(k_{\perp}^{\varepsilon}r)}{j_n(k_{\perp}^{\varepsilon}r)}- \frac{1}{k_{\perp}}\frac{{h_n^{(1)}}'(k_{\perp}r)}{h_n^{(1)}(k_{\perp}r)}\bigg]= \frac{n^2}{r^2}(\varepsilon-1)^2 \frac{k_z^2k^2}{{(k_{\perp}^{\varepsilon})}^4k_{\perp}^4}\ .\ ] ] in order to obtain yet another coupling equation for the same wave parameters we join the tangential components of the field at the end boundary @xmath159 . this leads to the set of equations [ end_matching ] @xmath160 \notag\\ & \hspace{5 cm } + \frac{n}{r}j_n(k_{\perp}^{\varepsilon}r ) \,k\bigg[\sin\frac{k_zh}{2}a_n^{m(s)}- \exp\big(-\varkappa_zh/2\big)c_n^{m(s)}\bigg]=0\ , \\ & k_{\perp}^{\varepsilon}{j_n}'(k_{\perp}^{\varepsilon}r ) \,k\bigg[\sin\frac{k_zh}{2}a_n^{m(s ) } - \exp\big(-\varkappa_zh/2\big)c_n^{m(s)}\bigg]\notag\\ & \hspace{5 cm } + \frac{n}{r}j_n(k_{\perp}^{\varepsilon}r ) \bigg[-k_z\sin\frac{k_zh}{2}a_n^{e(s ) } + \varkappa_z\exp\big(-\varkappa_zh/2\big)c_n^{e(s)}\bigg]=0\ , \\ & k_{\perp}^{\varepsilon}{j_n}'(k_{\perp}^{\varepsilon}r ) \bigg[k_z\cos\frac{k_zh}{2}a_n^{m(s ) } + \varkappa_z\exp\big(-\varkappa_zh/2\big)c_n^{m(s)}\bigg ] \notag\\ & \hspace{5 cm } -\frac{n}{r}j_n(k_{\perp}^{\varepsilon}r ) \,k\bigg[\varepsilon \cos\frac{k_zh}{2}a_n^{e(s ) } - \exp\big(-\varkappa_zh/2\big)c_n^{e(s)}\bigg]=0\ , \\ & k_{\perp}^{\varepsilon}{j_n}'(k_{\perp}^{\varepsilon}r ) \,k\bigg[\varepsilon \cos\frac{k_zh}{2}a_n^{e(s ) } - \exp\big(-\varkappa_zh/2\big)c_n^{e(s)}\bigg ] \notag\\ & \hspace{5 cm } -\frac{n}{r}j_n(k_{\perp}^{\varepsilon}r ) \bigg[k_z\cos\frac{k_zh}{2}a_n^{m(s ) } + \varkappa_z\exp\big(-\varkappa_zh/2\big)c_n^{m(s)}\bigg]=0\ , \end{aligned}\ ] ] where the number of the unknowns ( to the latter we assign not only constant factors included in eqs . and ) but also the functions @xmath161 and @xmath162 ) exceeds the number of the equations thus obtained . the situation can be improved if one adds to the system a pair of equalities resulting from joining the em field normal components at the same end boundary , specifically , [ norm_comp_eh ] @xmath163 using eqs . and we arrive at a set of four coupled equations which can be naturally combined in pairs , viz . [ eesems - eqs ] @xmath164 since the bessel functions do not vanish simultaneously with their derivatives , one can satisfy eqs . in two ways . the first one is to equate the expression in parenthesis of eq . to zero , putting the constant @xmath165 in eqs . . alternatively , the parenthesis in eq . can be set equal to zero concurrently with the coefficient @xmath166 in eqs . . as can be readily seen from eqs . and , the former solution corresponds to the tm - type oscillations whereas the latter is consistent with te polarization . both of these cases can be combined into one equation of the following form @xmath167 the lower curly brackets in eq . indicate the polarization which correlates with vanishing the expression in the corresponding parenthesis . by carrying out the calculations similar to those given above , for @xmath42-antisymmetric solution of eq . we arrive , instead of eq . , at the equality @xmath168 note that the relationships , and were previously obtained in ref . @xcite , although the latter two of them were of somewhat different form . in that paper , however , polarization of the excited em field was not discussed at all . at first glance , from our derivation of eqs . and it may seem that tm- and te - type oscillations in the ddr of finite thickness can be perfectly separated , at least , with no regard for the boundary inhomogeneity . as a matter of fact , this is not the case because one must take into account the inaccuracy , i.e. , the model character of wave solutions and used when deriving dispersion relations . these solutions , as well as the analogous ones for the antisymmetric case , are well adapted for joining the in- and the out- field components at the boundaries between regions 12 and 13 in fig . the fields in region 4 are not taken into consideration , which must inevitably result in some `` overflow '' of the spectrum obtained in such a way . it is a difficult task to correctly evaluate the possibility to neglect the fields in the 4-th region in fig . [ fig5 ] without resorting to rigorous calculations . nevertheless , by now there does not exist a rigorous theory for a finite thickness ddr , even in the seemingly simple case where the boundaries are perfectly smooth . for this reason , in order to test the results we obtained by the model calculations we choose to compare the spectrum resulting from eqs . , , with the one measured in the experiment consider the equation for green function of eq . without rendering both the physical nature of mode potentials and their absolute value concrete , @xmath169 g_{nn'}(r , r'|z)-\sum_{m\neq n}\widetilde{\mathcal{u}}_{nm}(r|z_{\scriptscriptstyle\lessgtr } ) g_{mn'}(r , r'|z_{\scriptscriptstyle\lessgtr})= \frac{1}{r}\delta(r - r')\delta_{nn'}\ .\ ] ] along with the exact green function , which has a matrix structure in variables @xmath54 and @xmath88 , we introduce for each of the azimuth modes the trial mode propagator , which is assumed to obey the closed equation @xmath170 g^{(v)}_{n}(r , r'|z_{\scriptscriptstyle\lessgtr})=\frac{1}{r}\delta(r - r')\ ] ] resulting from eq . providing that the inter - mode scattering is disregarded . we will use the notation @xmath171 for the operator in square brackets of eq . . the initial equation can then be recast as @xmath172 or , in the equivalent integral form , as @xmath173 by setting in eq . index @xmath174 and then re - labelling all mode indices we can write this equation as the equation for solely inter - mode components of the green matrix @xmath175 , the intra - mode ones being thought of as known functions , @xmath176 at this stage we introduce three operators , @xmath177 , @xmath178 and @xmath84 , which are assumed to act in the reduced coordinate - mode space @xmath179 consisting of the half - axis @xmath180 and the entire set of mode indices except for particular index @xmath54 . the operators are specified by their matrix elements equation can now be recast as the matrix equality @xmath182 or , equivalently , in the general operator form as @xmath183 . here , @xmath113 is the projection operator whose action reduces to assigning the value @xmath54 to the nearest mode index of an arbitrary operator standing adjacent to it , either to the left or right . multiplying both sides of the operator equality thus obtained by the operator @xmath184 whose regularity was substantiated in ref . @xcite we arrive at the integral relation between the non - diagonal and diagonal mode matrix elements of the green function , viz . @xmath185 setting in eq . index @xmath186 and substituting the inter - mode propagators in the form we eventually obtain the closed equation for intra - mode green function @xmath187 , @xmath188 g_{nn}(r , r'|z_{\scriptscriptstyle\lessgtr})=\frac{1}{r}\delta(r - r')\ .\ ] ] here , @xmath106 is the operator that accounts for the inter - mode scattering ( it is just the @xmath189-matrix well - known in quantum scattering theory @xcite ) , which has the form @xmath190 trial green function @xmath191 entering operator potential through matrix elements obeys equation and the same boundary conditions just like the desired mode propagator @xmath187 . equation along with eq . determines the entire green function of wave equation and serves as a basis for deriving the system of uncoupled equations for azimuth components of the sought - for wave function .

we call _ monomer _ a b - dna base pair and study , analytically and numerically , electron or hole oscillations in _ monomers _ , _ dimers _ and _ trimers_. we employ two tight binding ( tb ) approaches : ( i ) at the base - pair level , using the on - site energies of the base pairs and the hopping parameters between successive base pairs i.e. _ a wire model _ , and ( ii ) at the single - base level , using the on - site energies of the bases and the hopping parameters between neighbouring bases , specifically between ( a ) two successive bases in the same strand , ( b ) complementary bases that define a base pair , and ( c ) diagonally located bases of successive base pairs , i.e. _ an extended ladder model _ since it also includes the diagonal hoppings ( c ) . for _ monomers _ , with tb ii , we predict periodic carrier oscillations with frequency 50 - 550 thz . for _ dimers _ , with tb i , we predict periodic carrier oscillations with 0.25 - 100 thz . for _ trimers made of identical monomers _ , with tb i , we predict periodic carrier oscillations with 0.5 - 33 thz . in other cases , either with tb i or tb ii , the oscillations may be not strictly periodic , but fourier analysis shows similar frequency content . for dimers and trimers , tb i and tb ii are successfully compared giving complementary aspects of the oscillations . _ keywords _ : dna , charge transfer in biological systems , electronic transport in nanoscale materials and structures +

1609.00180

charge transfer ( ct ) in biological molecules attracts recently considerable interest among the physical , chemical , biological and medical communities . it also attracts a broad spectrum of interdisciplinary scientists and engineers . this is because ct constitutes the basis of many biological processes e.g. in various proteins @xcite including metalloproteins @xcite and enzymes @xcite with medical and bioengineering applications @xcite . ct plays a central role in dna damage and repair @xcite and it , also , might be an indicator to discriminate between pathogenic and non - pathogenic mutations at an early stage @xcite . at least for twenty years , there have been many experimental attempts to recognize the electronic properties of dna cf . e.g. @xcite . today we know that many factors related to environment ( the aqueous solution , the concentration of counterions ) , extraction process , conducts , purity , base - pair sequence , geometry etc influence ct in dna . these factors can be categorized into intrinsic and extrinsic . in this work we focus on maybe the most important of the intrinsic factors , i.e. the effect of alternating the base ( or base - bair ) sequence , which affects the overlaps across the @xmath1-stack . additionally , we have to discriminate between the words transport ( usually implying the use of electrodes ) , transfer , migration ( a transfer over rather long distances ) . the carriers ( electrons or holes ) can be inserted via electrodes , generated by uv irradiation or by reduction and oxidation . the nature of charge transport , transfer , migration along the dna double helix is important for many scientific fields like physics , chemistry , biology , medicine and engineering . although unbiased charge transfer in dna nearly vanishes after 10 to 20 nm @xcite , dna still remains a promising candidate as an electronic component in molecular electronics , e.g. as a short molecular wire @xcite . favouring geometries and base sequences have still to be explored e.g. incorporation of sequences serving as molecular rectifiers , using non - natural bases or using the triplet acceptor anthraquinone for hole injection @xcite . structural fluctuations could be another important factor which influences quantum transport through dna molecular wires @xcite . the contact of dna segments with experimentally involved surfaces and interfaces is another research direction e.g. for bio - sensoric applications . for example , charge transfer on the contact of dna with gold nanoparticles @xcite and polyelectrolyte multilayers @xcite has been recently investigated . during the last decade , the scientific literature has been enriched with works studying carrier oscillations within `` molecular '' systems . real - time time - dependent density functional theory ( rt - tddft ) @xcite simulations predicted oscillations ( @xmath2 0.1 - 10 phz ) within p - nitroaniline and ftc chromophore @xcite , as well as within zinc porphyrin , green fluorescent protein chromophores and the adenine - thymine base pair @xcite . it was shown that in a simplified single - stranded helix of 101 bases , a collinear uniform electric field induces thz bloch oscillations @xcite . single and multiple charge transfer within a typical dna dimer in connection to a bosonic bath has been studied , too @xcite . each base pair was approximated by a single site , as in our tight binding ( tb ) approach at the base - pair level denoted in the present article as tb i [ cf . section [ sec : general ] and appendix a ] . in the subspace of single charge transfer between base pairs , having initially placed the charge at the donor site and having used a `` typical hopping matrix element '' of 0.2 ev , the authors obtained a period slightly greater than 10 fs . let us call @xmath3 the period and @xmath4 the frequency . applying our equation @xcite @xmath5 , with @xmath6 0.2 ev for the `` typical hopping matrix element '' and difference of the on - site energies @xmath7 for identical dimers i.e. as in ref . @xcite , we obtain @xmath8 10.34 fs , which agrees splendidly with the dotted line in fig . 4 of ref . @xcite . recently , we studied b - dna dimers , trimers and polymers with tb i @xcite . this approach allowed us to readily determine the spatiotemporal evolution of holes or electrons along a @xmath9 base - pair dna segment . with tb i , we have already shown @xcite that for all dimers and for trimers made of identical monomers the carrier movement is periodic with frequencies in the mid- and far - infrared i.e. approximately in the thz domain @xcite . this part of the electromagnetic ( em ) spectrum is significant for biological sciences because it can be used to extract complementary to traditional spectroscopic measurements information e.g. on low - frequency bond vibrations , hydrogen bond stretching , bond torsions in liquids and gases etc and because it is relatively non - invasive compared to higher - frequency regions of the em spectrum @xcite . within tb i , we also showed that , generally , increasing the number of monomers above three , periodicity is lost @xcite . even for the simplest tetramer , the carrier movement is not periodic @xcite . for periodic cases , we defined @xcite the maximum transfer percentage @xmath10 , e.g. the maximum probability to find the carrier at the last monomer having placed it initially at the first monomer and the pure maximum transfer rate @xmath11 . for all cases , either periodic or not , the pure mean transfer rate @xmath12 ( cf . eq . [ meantransferraten(i ) ] in appendix a ) and the speed @xmath13 , where @xmath14 3.4 is the charge transfer distance , can be used to characterize the system . within tb i , our analytical calculations and numerical results showed that for dimers @xmath15 and for trimers made of identical monomers @xmath16 . using @xmath12 to evaluate the easiness of charge transfer , we calculated the inverse decay length @xmath17 for exponential fits @xmath18 and the exponent @xmath19 for power law fits @xmath20 . studying b - dna polymers and segments taken from experiments @xcite , we determined the ranges of values of @xmath17 and @xmath19 . our tb i was used @xcite to reproduce theoretical and experimental findings for various b - dna segments @xcite . for monomer - polymers and dimer - polymers , we studied homo and lumo eigenspectra , mean over time probabilities to find the carrier at a particular monomer and mean transfer rates , we illustrated how increasing the number of different parameters involved in tb i , the fall of @xmath18 or @xmath20 becomes steeper and we circumscribed the range covered by @xmath17 and @xmath19 @xcite . finally , both for the time - independent and the time - dependent problem , we analyzed the _ palindromicity _ and the _ degree of eigenspectrum dependence _ of the probabilities to find the carrier at a particular monomer @xcite . as in refs . @xcite , here we call _ monomer _ a b - dna base pair and study carrier oscillations in _ dimers _ and _ trimers_. moreover , for the first time we study _ monomers_. however , here we employ two tight binding ( tb ) approaches : ( i ) at the base - pair level , using the on - site energies of the base pairs and the hopping parameters between successive base pairs which is _ a wire model _ @xcite ( this was the approach employed in our previous works @xcite ) and ( ii ) at the single - base level , using the on - site energies of the bases and the hopping parameters between neighbouring bases , specifically between ( a ) two successive bases in the same strand , ( b ) complementary bases that define a base pair , and ( c ) diagonally located bases of successive base pairs , which is _ an extended ladder model _ since it also includes the diagonal hoppings ( c ) . hence , it is more elaborate than the usual ladder model @xcite . extrinsic effects , such as the aqueousness and the presence of counterions can also be taken into account in tb models which then can be renormalized via a decimation procedure to obtain either wire or ladder models , like the ones presented here @xcite . inclusion of the diagonal hoppings is essential e.g. for dimers made of identical monomers with crosswise purines cf . section 4 . additionally , tb ii allows us to study charge oscillations within monomers , which is not possible for tb i since there a site is a base pair . a few preliminary results with tb ii have been included in ref . @xcite and in ref . we assume that isolation of a few consecutive b - dna base pairs is possible , e.g. by connecting at the boundaries moieties with very small transfer integrals with the segment of interest . the tb parameters that we use can be found in refs . we solve analytically and numerically , with the eigenvalue method , a system of ( i ) @xmath9 or ( ii ) @xmath21 coupled differential equations to determine the spatiotemporal evolution of an extra carrier ( electron or hole ) along a @xmath9 base - pair dna segment . carriers move either between the homos or between the lumos of the relevant sites [ ( i ) base pairs , ( ii ) bases ] . a legitimate critique to our theoretical predictions could be relative to a possible comparison with the experiment . as far as we know , such experiments in so short dna segments ( monomers , dimers , trimers ) do not exist . however , nowadays a variety of experimental techniques can be used to probe ct in biological molecules including optical , electrochemical and scanning probe techniques such as electrochemical stm and conductive afm @xcite . in our case , we can imagine the isolation of an oligomer by putting at its end moieties with very small hopping integral with the ends of the oligomer under investigation and subsequent use of one of these techniques . the techniques must be in the position to probe oscillations with frequency content in the 0.1 to 1000 thz regime i.e. time scales of 10 ps to 1 fs . additionally , our method tb i has already been used to successfully reproduce experimental results with longer dna segments relative to transfer rates and mean occupation probabilities @xcite . in this work we show that for monomers tb ii predicts periodic carrier oscillations with frequency @xmath0 50 - 550 thz ( but with very small transfer percentages ) , while tb i predicts periodic carrier oscillations with @xmath0 0.25 - 100 thz for dimers and with @xmath0 0.5 - 33 thz for trimers made of identical monomers . in other cases , either with tb i or tb ii , oscillations are not strictly periodic , but fourier analysis shows similar frequency content . for dimers and trimers tb i and tb ii give complementary aspects of the oscillations . for dimers made of identical monomers we have large carrier transfer and the occupation probability is equally shared between the two monomers which constitute the dimer . for dimers made of identical monomers , if purines are crosswise to purines , interstrand carrier transfer dominates , i.e. we have significant diagonal transfer uncovered by tb ii which includes _ diagonal hoppings _ ; if purines are on the same strand , intrastrand carrier transfer dominates . for dimers made of different monomers , tb ii basically shows intrastrand carrier transfer ( but in small percentage ) . in this work we show that thz oscillations in dna monomers , dimers and trimers exist and we study the frequency content , the maximum transfer percentages , the transfer rates between sites and the mean probabilities to find the carrier at a site . hence , one could imagine the future built of a source or receiver of em radiation in the range 0.1 thz to 1000 thz made of tiny dna segments . the rest of the article is organized as follows : in sec . [ sec : general ] we outline our tb i and tb ii approaches in general terms skipping technical details , which are presented in appendix a. our results for monomers , dimers and trimers are presented in sections [ sec : resultsmonomers ] , [ sec : resultsdimers ] and [ sec : resultstrimers ] , respectively . in sec . [ sec : conclusion ] we state our conclusions . we begin with our notation . by yx we denote two successive base pairs , according to the convention @xmath22 for the dna strands orientation . we denote by x , x@xmath23 , y , y@xmath23 dna bases , where x@xmath23 ( y@xmath23 ) is the complementary base of x ( y ) . in other words , the notation yx means that the bases y and x of two successive base pairs are located at the same strand in the direction @xmath24 . x - x@xmath23 is the one base pair and y - y@xmath23 is the other base pair , separated and twisted by 3.4 and @xmath25 , respectively , relatively to the first base pair , along the growth axis of the nucleotide chain . for example , the notation gt denotes that one strand contains g and t in the direction @xmath24 and the complementary strand contains c and a in the direction @xmath26 . in the sense explained above , we can talk for equivalent dimers i.e. @xmath27 and expand the notion of equivalency to @xmath9-mers . furthermore , we suppose that an extra hole or electron inserted in a dna segment travels through homos or lumos , respectively . hence , for each base pair or _ monomer _ , the highest occupied molecular orbital ( homo ) and the lowest unoccupied molecular orbital ( lumo ) play a key role . we utilize two tight - binding approximations , the main points of which are explained below . the mathematical details are given in appendix a. here we merely describe them in general terms to make the manuscript more accessible for readers from different areas of the rather wide dna science . in tb i the carrier is located at a base pair and it can move to the next or to the previous base pair . hence , this is a _ wire model _ @xcite . in tb ii the carrier is located at a base and it can move ( 1 ) to its complementary base of the same base pair or ( 2 ) to the next or the previous base of the same strand or ( 3 ) to the diagonally located base of the other strand of the next or the previous base pair in the @xmath28 or in the @xmath29 direction , respectively . if we ignore ( 3 ) , this would be a _ ladder model _ @xcite . here , since we also include diagonal hoppings we call it an _ extended ladder model _ it will become evident below that in some cases ( e.g. in dimers when purines are crosswise to purines ) interstrand carrier transfer dominates , in other words we have significant diagonal transfer , which justifies the inclusion of diagonal hoppings in our model tb ii . the two tb models are explained schematically in fig . [ fig : tbitbii ] . [ h ! ] in sections [ sec : resultsmonomers ] , [ sec : resultsdimers ] , and [ sec : resultstrimers ] we show that thz oscillations in dna monomers , dimers and trimers exist . we study the frequency content of these oscillations ; where necessary , we employ fourier analysis . for periodic cases , we study the maximum carrier transfer percentage @xmath10 from an initial site to a final site , as well as the relevant maximum transfer rate @xmath30 showing not only how fast the transfer is , but also what the maximum carrier transfer percentage is . for all cases , either periodic or not , we use the mean transfer rate @xmath31 , which shows not only how fast the transfer from site @xmath32 to site @xmath33 is , but also what the mean carrier transfer percentage from site @xmath32 to site @xmath33 is ( cf . eq . [ meantransferraten ] ) . tb i can not be used for charge transfer in monomers , since it considers a monomer as a single site . hence , we use tb ii , supposing that initially we place the carrier at one of the bases . we can prove that an extra hole or electron oscillates between the bases of the two possible monomers ( g - c and a - t ) with frequency ( or period ) @xmath34 where @xmath35 is the hopping integral between the complementary bases and @xmath36 is the energy gap between the on - site energies of the complementary bases . our results for a - t and g - c , both for holes and electrons , are shown in fig . [ fig : monomers ] , with parameters from ref . @xcite ( `` hks parametrization '' ) and from ref . @xcite ( `` ma parametrization '' ) . for hks parametrization , @xmath0 50 - 200 thz ( @xmath8 5 - 20 fs ) , for ma parametrization , @xmath0 250 - 550 thz ( @xmath8 2 - 4 fs ) . these ranges correspond to wavelength @xmath37 545 nm - 6000 nm i.e. from visible to near - infrared and mid - infrared @xcite . we can prove that the maximum transfer percentage @xmath10 [ e.g. @xmath38 for initial conditions @xmath39 or vice versa ] , is given by @xmath40 we observe that the carrier is not very likely to be transferred between the monomer bases ( @xmath10 is very small in all cases ) . the pure maximum transfer rate defined as @xmath30 is also here very small in all cases . the pure mean transfer rate @xmath12 is also shown . it can be analytically proven and numerically shown that here @xmath41 . @xmath3 , @xmath4 , @xmath10 , @xmath30 and @xmath12 do not depend on which base the carrier is initially placed at . a snapshot of electron oscillations between g and c in g - c , according to the hks parametrization , is given in fig . [ fig : electronoscillationsg - c ] . [ h ! ] , frequency @xmath4 , maximum transfer percentage @xmath10 , pure maximum transfer rate @xmath30 and pure mean transfer rate @xmath12 . 1st row : tb parameters from ref . @xcite ( hks parametrization ) . 2nd row : tb parameters from ref . @xcite ( ma parametrization ) . , title="fig:",width=294 ] , frequency @xmath4 , maximum transfer percentage @xmath10 , pure maximum transfer rate @xmath30 and pure mean transfer rate @xmath12 . 1st row : tb parameters from ref . @xcite ( hks parametrization ) . 2nd row : tb parameters from ref . @xcite ( ma parametrization ) . , title="fig:",width=294 ] , frequency @xmath4 , maximum transfer percentage @xmath10 , pure maximum transfer rate @xmath30 and pure mean transfer rate @xmath12 . 1st row : tb parameters from ref . @xcite ( hks parametrization ) . 2nd row : tb parameters from ref . @xcite ( ma parametrization ) . , title="fig:",width=294 ] , frequency @xmath4 , maximum transfer percentage @xmath10 , pure maximum transfer rate @xmath30 and pure mean transfer rate @xmath12 . 1st row : tb parameters from ref . @xcite ( hks parametrization ) . 2nd row : tb parameters from ref . @xcite ( ma parametrization ) . , title="fig:",width=294 ] [ h ! ] this is a two - level system of given stationary states ( the two homos or the two lumos ) with a `` perturbation '' represented by the hopping integral , which impels an extra carrier to oscillate between these stationary states . mathematically , the problem is equivalent to a two - level system ( e.g. atom ) under the influence of an electric field , which impels an electron to oscillate between the two eigenstates [ semiclassical approach after rotating wave approximation or the time - dependent problem with a jaynes - cummings hamiltonian in a full quantum mechanical approach ] . these problems are well known in the context of quantum optics @xcite . the same applies to the dimer problem within tb i discussed briefly below . increasing the number of monomers which make up the dna segment , either with tb i or tb ii , we have to solve generalizations of the above mentioned problem . in other words , to determine the spatiotemporal evolution of electrons or holes along a @xmath9 base - pair dna segment , we have to solve a system of ( i ) @xmath9 or ( ii ) @xmath21 coupled differential equations . the relevant matrices @xmath42 are given in appendix b. for @xmath9 base pairs , the system has @xmath9 states within tb i , or @xmath21 states within tb ii . thus , in this work , for tb i we examine systems composed of two ( dimers ) or three ( trimers ) states , while for tb ii we examine systems composed of two ( monomers ) , four ( dimers ) or six ( trimers ) states . the possible dimers result from the combination by two of the four bases , since at the base located in helix @xmath43 corresponds always its complementary base . the number of possible combinations of @xmath44 ( four nitrogenous bases ) elements by @xmath10 ( two ) is @xmath45=16 . however , six of them are equivalent to other six i.e. gg@xmath46cc , aa@xmath46tt , ag@xmath46ct , ac@xmath46gt , tg@xmath46ca , tc@xmath46ga . hence , the possible dimers are 10 . recently , with tb i , we proved that the carrier movement in all dimers is strictly periodic @xcite . the frequencies ( or periods ) are given by eq . [ fandt ] , where , now @xmath35 is the hopping integral between the base pairs and @xmath36 is the energy gap between the on - site energies of the base pairs . using the tb parameters of ref . @xcite , we found @xcite @xmath0 0.25 - 100 thz , i.e. @xmath8 10 - 4000 fs , i.e. wavelength @xmath37 3 - 1200 @xmath47 m , in other words mainly in the mir and the fir range @xcite . we found that the maximum transfer percentage @xmath48 for dimers made of identical monomers , but @xmath49 for dimers made of different monomers and that the values of @xmath50 do not depend on which of the two monomers the carrier is initially placed at ( cf . also eq . [ periodsorfrequencies ] ) . the hks parametrization @xcite results in the same frequency range , although , the predicted frequencies vary slightly due to the different values of the tb parameters ( a summarizing graph is given in fig . [ fig : dimershks ] ) . [ h ! ] , frequency @xmath4 , maximum transfer percentage @xmath10 , pure maximum transfer rate @xmath30 and pure mean transfer rate @xmath12 . it can be analytically proven and numerically shown that here @xmath41.,title="fig:",width=283 ] , frequency @xmath4 , maximum transfer percentage @xmath10 , pure maximum transfer rate @xmath30 and pure mean transfer rate @xmath12 . it can be analytically proven and numerically shown that here @xmath41.,title="fig:",width=283 ] , frequency @xmath4 , maximum transfer percentage @xmath10 , pure maximum transfer rate @xmath30 and pure mean transfer rate @xmath12 . it can be analytically proven and numerically shown that here @xmath41.,title="fig:",width=283 ] , frequency @xmath4 , maximum transfer percentage @xmath10 , pure maximum transfer rate @xmath30 and pure mean transfer rate @xmath12 . it can be analytically proven and numerically shown that here @xmath41.,title="fig:",width=283 ] let us now compare tb approaches i and ii relatively to the frequency content . using tb ii , one can not strictly determine periodicity in the carrier movement between the four bases . hence , @xmath51 can not be defined , but fourier analysis shows similar frequency content in the thz domain , cf . appendix c , where we depict the fourier spectra given by eq . [ fourierspectra ] , within tb ii and hks parametrization @xcite . let us start with gg , a dimer made of identical monomers with purines on purines ( fourier spectra in appendix c , fig . [ fig : fouriergg ] ) . if we initially place the hole at a1(g ) or a2(g ) , we obtain the main fourier amplitude at @xmath0 30 thz . if we initially place the hole at b1 ( c ) or b2 ( c ) , we obtain the main fourier amplitude at @xmath0 32 thz . the rest of the frequencies show up with almost negligible amplitudes . these results are in accordance with tb i and hks parametrization @xcite where for the gg dimer we obtain @xmath0 30 thz . with tb i and the parametrization of refs . @xcite we had obtained @xmath0 48 thz @xcite . the amplitudes at the main frequencies are @xmath2 0.5 , in accordance with eq . [ fourierspectra ] , expressing the fact that for gg the mean probability to find the hole at a base is , approximately , almost exclusively , equally divided between the base the carrier was initially placed at and the other base of the same strand , cf . also fig . [ fig : dimersiihomo ] below . let us now continue with gc , a dimer made of identical monomers with crosswise purines ( fourier spectra in appendix c , fig . [ fig : fouriergc ] ) . if we initially place the hole at a1(g ) or b2(g ) , we obtain the main fourier amplitude at @xmath0 0.3 thz . if we initially place the hole at b1 ( c ) or a2 ( c ) , we obtain the main fourier amplitude peak at @xmath0 1.6 thz . the rest of the frequencies show up with almost negligible amplitudes . these results are in accordance with tb i and hks parametrization @xcite where for the gc dimer we obtain @xmath0 0.5 thz . with tb i and the parametrization of refs . @xcite we had obtained @xmath0 4.8 thz @xcite . the amplitudes at the main frequencies are @xmath2 0.5 , in accordance with eq . [ fourierspectra ] , expressing the fact that for gc the mean probability to find the hole at a base is , approximately , almost exclusively , equally divided between the base the carrier was initially placed at and the diagonally located base at the opposite strand , cf . also fig . [ fig : dimersiihomo ] below . let us finish with ct , a dimer made of different monomers ( fourier spectra in appendix c , fig . [ fig : fourierct ] ) . the main fourier amplitude is at @xmath0 70.75 thz , if the hole is initially placed at c or t , otherwise hole transfer is negligibly small . within tb i and the hks parametrization @xcite , for the ct dimer we obtain @xmath0 72.5 thz . with tb i and the parametrization of refs . @xcite we had obtained @xmath0 74 thz @xcite . the amplitudes at the main frequencies are @xmath2 0.25 , when the hole is initially placed at c or t and tiny when the hole is initially placed at g or a , in accordance with eq . [ fourierspectra ] , expressing the fact that for ct the mean probability to find the hole at a base is approximately 0.75 at the base the carrier was initially placed at and approximately 0.25 at the other base of the same strand when these bases are c and t , but , hole transfer is negligibly small when the hole is initially placed at g or a , cf . also fig . [ fig : dimersiihomo ] below . examples of snapshots of hole oscillations in gg , gc and ct dimers , according to the hks parametrization and tb ii , are given in fig . [ fig : holeoscillationsdimers ] . [ h ! ] hence , generally , both tb approaches , independently of the specific parametrization , predict oscillations in the same thz range . the little differences arise from the specific values used for the on - site energies and the hopping parameters in each parametrization . tb approaches i and ii allow us to determine the mean probability to find the carrier at a site [ base pair for i , base for ii ] . a comparison between the mean probabilities obtained with tb approaches i and ii is shown in fig . [ fig : comparison ] , using the hks parametrization @xcite . comparing the two tb approaches e.g. looking at fig . [ fig : comparison ] , but also focusing on tb ii e.g. looking at figs . [ fig : dimersiihomo]-[fig : dimersiilumo ] below , we reach the following conclusions : ( a ) carrier transfer is large in dimers made of identical monomers : finally , the probability is equally shared between the two monomers which make up the dimer . ( b ) for dimers made of identical monomers , if purines are crosswise to purines , the carrier changes strand ( from strand 1 to strand 2 or vice versa ) , while if purines are on the same strand , the carrier is transferred through the strand it was initially placed at . ( c ) for dimers made of different monomers , the carrier is transferred ( albeit in small percentage ) mainly through the strand it was initially placed at . hence , carrier transfer is very small in dimers made of different monomers . the carrier basically remains in the base it was initially placed at , while a small percentage passes to the other base of the same strand . a careful inspection in figs . [ fig : fouriergg ] , [ fig : fouriergc ] and [ fig : fourierct ] shows that the fourier analysis confirms conclusions ( a),(b),(c ) . for tb i , in refs . @xcite , we used a somehow different set of tb parameters , the results are similar . [ h ! ] at each site of a dna dimer , as determined with ( i ) the base - pair tb approach ( left column ) and ( ii ) the single - base tb approach ( right column ) . for tb i , the carrier is initially placed at the 1st monomer , while , for tb ii , it is initially placed at the base of the 1st monomer that belongs to the 1st strand . we use the hks parametrization @xcite.,title="fig:",width=283 ] at each site of a dna dimer , as determined with ( i ) the base - pair tb approach ( left column ) and ( ii ) the single - base tb approach ( right column ) . for tb i , the carrier is initially placed at the 1st monomer , while , for tb ii , it is initially placed at the base of the 1st monomer that belongs to the 1st strand . we use the hks parametrization @xcite.,title="fig:",width=283 ] at each site of a dna dimer , as determined with ( i ) the base - pair tb approach ( left column ) and ( ii ) the single - base tb approach ( right column ) . for tb i , the carrier is initially placed at the 1st monomer , while , for tb ii , it is initially placed at the base of the 1st monomer that belongs to the 1st strand . we use the hks parametrization @xcite.,title="fig:",width=283 ] at each site of a dna dimer , as determined with ( i ) the base - pair tb approach ( left column ) and ( ii ) the single - base tb approach ( right column ) . for tb i , the carrier is initially placed at the 1st monomer , while , for tb ii , it is initially placed at the base of the 1st monomer that belongs to the 1st strand . we use the hks parametrization @xcite.,title="fig:",width=283 ] [ h ! ] [ h ! ] let us now turn our discussion to carrier mean transfer rates @xmath31 , within tb ii and the hks parametrization @xcite , cf . [ fig : dimershomolumokijall ] . the specific values of @xmath31 depend , of course , on the tb hopping parameters used @xcite . let us call @xmath31 the mean transfer rates of a dimer yx and @xmath52 the mean transfer rates of the equivalent dimer @xmath53 . the following two properties hold : @xmath54 @xmath55 for dimers made of identical monomers with purine on purine ( gg@xmath46cc , aa@xmath46tt ) , hole transfer is almost entirely of intrastrand character i.e. it is along the @xmath56-@xmath57 or @xmath57-@xmath56 directions . moreover , since @xmath58 and @xmath59 satisfy eq . [ k , k relation for equivalent dimers ] , in fig . [ fig : dimershomolumokijall ] we observe a symmetry in the alternation of colors for the couples of equivalent dimers gg@xmath46cc and aa@xmath46tt . for dimers made of identical monomers with crosswise purines ( cg , gc , ta , at ) , there is significant diagonal hole transfer , and furthermore , for cg and ta the stronger hole transfer is along the @xmath57-@xmath57 direction . for dimers made of different monomers ( ag@xmath46ct , ac@xmath46gt , tg@xmath46ca , tc@xmath46ga ) , hole transfer is almost exclusively of intrastrand character i.e. along the @xmath56-@xmath57 or @xmath57-@xmath56 directions ; since @xmath58 and @xmath59 satisfy eq . [ k , k relation for equivalent dimers ] , we observe the same symmetry in the alternation of colors . for the couple tg@xmath46ca , although @xmath58 and @xmath59 are the biggest among all other @xmath31 , they are very small . electron transfer in dimers made of identical monomers with purine on purine ( gg@xmath46cc , aa@xmath46tt ) is qualitatively similar to hole transfer in such dimers . for dimers made of identical monomers with crosswise purines ( gc , cg , at , ta ) electron transfer is slightly different than hole transfer in such dimers in the sense that diagonal channels are important but are not , quantitatively , identically important . electron transfer in dimers made of different monomers has a significant intrastrand character , but there is also intra - base - pair character in some cases . for the same reasons described above , we observe symmetry in color alternation for @xmath31 of equivalent dimers . [ h ! ] between bases @xmath32 and @xmath33 either for homo ( hole transfer , 1st row ) or for lumo ( electron transfer , 2nd row ) , for all dimers , within tb approach ii and the hks parametrization @xcite . the 1st column corresponds to dimers made of identical monomers and the 2nd row to dimers made of different monomers.,title="fig:",width=283 ] between bases @xmath32 and @xmath33 either for homo ( hole transfer , 1st row ) or for lumo ( electron transfer , 2nd row ) , for all dimers , within tb approach ii and the hks parametrization @xcite . the 1st column corresponds to dimers made of identical monomers and the 2nd row to dimers made of different monomers.,title="fig:",width=283 ] between bases @xmath32 and @xmath33 either for homo ( hole transfer , 1st row ) or for lumo ( electron transfer , 2nd row ) , for all dimers , within tb approach ii and the hks parametrization @xcite . the 1st column corresponds to dimers made of identical monomers and the 2nd row to dimers made of different monomers.,title="fig:",width=283 ] between bases @xmath32 and @xmath33 either for homo ( hole transfer , 1st row ) or for lumo ( electron transfer , 2nd row ) , for all dimers , within tb approach ii and the hks parametrization @xcite . the 1st column corresponds to dimers made of identical monomers and the 2nd row to dimers made of different monomers.,title="fig:",width=283 ] let us now compare tb i and tb ii in trimers made of identical monomers . using tb i we proved @xcite that in trimers made of identical monomers an extra carrier oscillates periodically with @xmath60 where @xmath61 are the hopping integrals between the base pairs ( when all purines are on the same strand , @xmath62 ) . with the parametrization of refs . @xcite , we found @xmath0 0.5 - 33 thz ( @xmath8 30 - 2000 fs ) @xcite ; with the parametrization of ref . @xcite , we find @xmath0 0.5 - 21 thz ( @xmath8 48 - 2000 fs ) . in other words , for trimers made of identical monomers , the frequency range is narrower than for dimers . for 0 times crosswise purines , the maximum transfer percentage @xmath63 , while for 1 or 2 times crosswise purines @xmath64 @xcite . tb ii in trimers , generally , does not allow one to strictly determine periodicity ; @xmath65 , and @xmath30 can not be defined . however , fourier analysis shows similar frequency content . ( specific examples are given in appendix c , where we present fourier analysis of hole oscillations in two trimers made of identical monomers : ggg in fig . [ fig : fourierggg ] and gcg in fig . [ fig : fouriergcgandcacandctc ] . ) for hole transfer in ggg , within tb i and the parametrization of refs . @xcite , we found @xmath66 34.2 thz ; now with the hks parametrization @xcite , we find @xmath66 21.2 thz . this is in remarkable agreement with the frequencies obtained by fourier transform , within tb ii and hks parametrization @xcite shown in fig . [ fig : fourierggg ] . specifically , e.g. for initial placement of the hole at base a1(g ) , the main frequencies are around 21.2 thz ( a double peak ) and 42.4 thz ( a single peak ) , while , e.g. for initial placement of the hole at base b1(c ) , the main frequencies are around 22.5 thz ( a double peak ) and 45 thz ( a single peak ) . in other words , for hole transfer in ggg , within the hks parametrization , the period predicted by tb i agrees with the approximate period predicted by tb ii . the mean probabilities to find an extra carrier at a base and the mean transfer rates in ggg and aaa , within tb ii and hks parametrization @xcite , are shown in fig . [ fig : trimersgggaaa ] . we observe that the carrier movement is almost exclusively of intrastrand character , a fact also evident from the fourier analysis in fig . [ fig : fourierggg ] . remarkably , with tb ii , the probabilities to find the carrier at each base pair are either @xmath2 0.375 , 0.25 , 0.375 or @xmath2 0.25 , 0.5 , 0.25 depending on the initial placement of the carrier , in agreement with the rules established in ref . @xcite for tb i. the mean transfer rates @xmath31 confirm the intrastrand character of charge transfer in ggg and aaa . for hole transfer in gcg , within the hks parametrization @xcite , tb i and tb ii give similar results , indicating rather weak transfer . for example , placing the hole initially at the first base pair for tb i or placing the hole initially at the first base for tb ii , the probability to find the hole at the first base pair is @xmath2 0.9990 for tb i and 0.9848 for tb ii and at the last base pair @xmath2 0.0008 for tb i and 0.0006 for tb ii . this is mirrored in the very small fourier amplitudes for gcg ( appendix c , fig . [ fig : fouriergcgandcacandctc ] ) . within tb i , for trimers made of different monomers , carrier movement may be non - periodic @xcite . within tb i , generally , increasing the number of monomers above three , the system becomes more complex and periodicity is lost @xcite ; even in the simplest cases , e.g. tetramers made of identical monomers with all the purines on the same strand , there is no periodicity @xcite . within tb ii , as mentioned before , in trimers , generally , periodicity can not be strictly determined . as a last point , we restrict ourselves in giving ( appendix c , fig . [ fig : fouriergcgandcacandctc ] ) two examples of the frequency content of hole oscillations in trimers made of different monomers , specifically in cac and ctc . [ h ! ] 0.375 , 0.25 , 0.375 or @xmath2 0.25 , 0.5 , 0.25 depending on the initial placement of the carrier , which agrees remarkably with the rules established in ref . @xcite for tb i. _ lower panel : _ the mean transfer rates @xmath31 from base @xmath32 to base @xmath33 . the arrows indicate intrastrand transfer from the strand start to the strand end.,title="fig:",width=283 ] 0.375 , 0.25 , 0.375 or @xmath2 0.25 , 0.5 , 0.25 depending on the initial placement of the carrier , which agrees remarkably with the rules established in ref . @xcite for tb i. _ lower panel : _ the mean transfer rates @xmath31 from base @xmath32 to base @xmath33 . the arrows indicate intrastrand transfer from the strand start to the strand end.,title="fig:",width=283 ] 0.375 , 0.25 , 0.375 or @xmath2 0.25 , 0.5 , 0.25 depending on the initial placement of the carrier , which agrees remarkably with the rules established in ref . @xcite for tb i. _ lower panel : _ the mean transfer rates @xmath31 from base @xmath32 to base @xmath33 . the arrows indicate intrastrand transfer from the strand start to the strand end.,title="fig:",width=566 ] using two tb approaches , a wire model called here tb i , where a site coincides with a dna base pair , and an extended ( i.e. including also diagonal hoppings ) ladder model called here tb ii , where a site coincides with a dna base , we demonstrated that thz oscillations in dna monomers , dimers and trimers exist . we studied various aspects of the effect , e.g. frequency content , maximum transfer percentages and transfer rates between sites , and mean probabilities to find the carrier at a particular site . we also compared successfully the two tb approaches . naturally , tb ii allows for greater detail . for dna monomers , i.e. for adenine - thymine and guanine - cytosine , with tb ii , we predicted electron or hole oscillations in the range @xmath0 50 - 550 thz ( @xmath8 2 - 20 fs ) , i.e. @xmath37 545 nm - 6 @xmath47 m , from visible to near- and mid infrared @xcite . we found that the maximum transfer percentage @xmath10 and the pure maximum transfer rate @xmath30 between the bases are very small . for dna dimers , with tb i , we predicted electron or hole oscillations in the range @xmath0 0.25 - 100 thz ( @xmath67 10 - 4000 fs ) i.e. @xmath37 3 - 1200 @xmath47 m , approximately in the mid- and far - infrared . for dimers made of identical monomers the maximum transfer percentage @xmath63 , but for dimers made of different monomers @xmath64 . with tb ii , the carrier oscillations are not strictly periodic but the frequency content is similar to that predicted with tb i. for the mean probabilities to find the carrier at a particular site , the two approaches give coherent , complementary results . tb ii shows that for dimers made of identical monomers , when purines are crosswise to purines , interstrand carrier transfer dominates , i.e. we have significant diagonal transfer , justifying the inclusion of diagonal hoppings in our model , while if purines are on the same strand , intrastrand carrier transfer dominates . for dimers made of different monomers , we carrier transfer is mainly intrastrand character but the transfer percentage is small . with tb i , for trimers made of identical monomers , the carrier oscillates periodically with @xmath0 0.5 - 33 thz ( @xmath8 30 - 2000 fs ) if we use the parametrization of refs . @xcite . with the hks parametrization @xcite , @xmath0 0.5 - 21 thz . for 0 times crosswise purines @xmath63 , for 1 or 2 times crosswise purines @xmath64 . with tb ii , the carrier oscillations are not strictly periodic but the frequency content is similar to that predicted with tb i. for the mean probabilities to find the carrier at a particular site , the two approaches give coherent , complementary results . finally , we would like to mention that increasing the number of monomers , i.e. constructing an oligomer or a polymer , the frequency spectrum becomes more fragmented and moves towards lower frequencies . a systematic study of longer segments is beyond the scope of the present manuscript . it seems that a source or receiver of electromagnetic radiation made of dna monomers , dimers or trimers , with frequencies from fractions of thz to just below phz , could be envisaged . + * * related work at * + ` http://users.uoa.gr/~csimseri/physics_of_nanostructures_and_biomaterials.html ` + * acknowledgements * a. morphis thanks the state scholarships foundation - iky for a ph.d . research scholarship via `` iky fellowships of excellence '' , hellenic republic - siemens settlement agreement . m. tassi thanks the state scholarships foundation - iky for a post - doctoral research fellowship via `` iky fellowships of excellence '' , hellenic republic - siemens settlement agreement . + [ aa ] according to tb i [ description at the base - pair level ] the homo or lumo wave function of a given dna segment , made of @xmath9 base pairs , @xmath68 , is considered as a linear combination of the base - pair wave functions @xmath69 with time - dependent coefficients @xmath70 @xmath71 gives the probability of finding the carrier ( hole for homo , electron for lumo ) at base pair @xmath47 . using the time - dependent schrdinger equation , @xmath72 we find that the time evolution of the coefficients @xmath73 obeys the system of equations @xcite @xmath74 where @xmath75 are the homo or lumo on - site energies of the base pairs , and @xmath76 are the homo or lumo hopping integrals between two nearest neighbouring base pairs @xmath47 and @xmath77 . according to tb ii [ description at the single - base level ] @xmath68 is derived from the single - base wave functions , according to the expression @xmath78\ ] ] where @xmath79 and @xmath80 , is the wave function of the base at the @xmath47-th base pair and in the @xmath81-th strand . @xmath71 , @xmath82 give the probability to find the carrier at the base of strand 1 and 2 , respectively , of the @xmath47-th base pair . again , using the time - dependent schrdinger equation i.e. eq . [ tdse ] , we find that in this case , the system of equations is @xcite @xmath83 @xmath84 @xmath85 are the homo or lumo on - site energies of the base at the @xmath47-th base pair and in the @xmath81-th strand , and @xmath86 are the homo or lumo hopping parameters between neighbouring bases , i.e. between ( a ) two successive bases in the same strand , ( b ) complementary bases that define a base pair , and ( c ) diagonally located bases of successive base pairs . to determine the temporal and spatial evolution of electrons or holes along a @xmath9 base - pair dna segment , we solve the system of ( i ) @xmath9 or ( ii ) @xmath21 coupled differential equations i.e. eq . [ bpequations ] or eqs . [ bequations - strand1]-[bequations - strand2 ] , respectively , with the eigenvalue method , which is explained below . tb ii allows us to examine the system in higher detail than with tb i , but tb i uses a smaller number of parameters and it is readily implemented . to solve eq . [ bpequations ] for tb i or eqs . [ bequations - strand1]-[bequations - strand2 ] for tb ii , we define the vector matrix @xmath87 \ ; \textrm { for tb i \quad \textit{or } } \quad \vec{x}(t ) = \left [ \begin{array}{c } a_1(t ) \\ b_1(t ) \\ \vdots \\ b_n(t ) \end{array } \right ] \ ; \textrm { for tb ii}\ ] ] and therefore eq . [ bpequations ] for tb i or eqs . [ bequations - strand1]-[bequations - strand2 ] for tb ii become @xmath88 where we can define @xmath89 we solve eq . [ xdotmathcalax ] via the _ eigenvalue method _ , i.e. , we look for solutions of the form @xmath90 hence , eq . [ xdotmathcalax ] reads @xmath91 or @xmath92 @xmath93 in other words , we have to solve an eigenvalue problem . the order of the matrices @xmath94 or @xmath95 is @xmath9 for tb i or @xmath21 for tb ii . the matrices @xmath95 for tb i and tb ii are shown in appendix b. already , we called @xmath47 the base - pair index i.e. @xmath96 and @xmath81 the strand index i.e. @xmath43 or @xmath97 . then , the base index @xmath98 and @xmath99 . schematically , @xmath100 having checked that the normalized eigenvectors @xmath101 corresponding to the eigenvalues @xmath102 of eq . [ eigenvalueproblem2 ] are linearly independent , the solution of our problem is @xmath103 where @xmath104 is the matrix dimension , i.e. , @xmath9 for tb i or @xmath21 for tb ii . if we initially place the carrier at the first base pair ( tb i ) or if we initially place the carrier at the first base pair either at strand 1 or at strand 2 ( tb ii ) , then , the initial conditions would be @xmath105 \ ! = \ ! \left [ \begin{array}{c } 1 \\ 0 \\ \vdots \\ 0 \end{array } \right ] \ ; \textrm{(tb i ) } \quad \textit{or } \quad \vec{x}(0 ) \ ! = \ ! \left [ \begin{array}{c } a_1(0 ) \\ b_1(0 ) \\ \vdots \\ b_n(0 ) \end{array } \right ] \ ! = \ ! \left [ \begin{array}{c } 1 \\ 0 \\ \vdots \\ 0 \\ 0 \end{array } \right ] \textrm{or } \left [ \begin{array}{c } 0 \\ 1 \\ \vdots \\ 0 \\ 0 \end{array } \right ] \ ; \textrm{(tb ii)},\ ] ] respectively . other initial conditions could be similarly defined . from the initial conditions we determine the coefficients @xmath106 . using @xmath107 as a generic index for either @xmath47 in tb i or @xmath17 in tb ii , for initial placement of the extra carrier at a site @xmath32 , we can show analytically that @xmath108 , where @xmath109 is the @xmath32-th component of the eigenvector @xmath101 . generally , the hamiltonians describing equivalent @xmath9-mers are related with a similarity transformation , hence , they have the same eigenvalues and their eigenvectors are connected by @xmath110 . in tb i , the mean over time probability to find the extra carrier at base pair @xmath47 is @xmath111 , while , in tb ii , the mean over time probability to find the extra carrier at the 1st strand base of base pair @xmath47 is @xmath111 and at the 2nd strand base of base pair @xmath47 is @xmath112 ) . using @xmath113 , @xmath107 , as a generic symbol for either @xmath73 in tb i or @xmath73 and @xmath114 in tb ii , we can show analytically that the mean over time probability to find the carrier at a site @xmath32 is given by @xmath115 furthermore , supposing that the eigenvalues are arranged in ascending order , i.e. @xmath116 , the frequencies or periods involved in charge transfer are @xmath117 we can analytically show that the one - sided fourier spectrum for the probability @xmath118 is given by @xmath119 , while all other @xmath120 , @xmath121 . hence , for a polymer consisting of @xmath9 monomers , a _ pure _ mean transfer rate can be defined as @xmath122 where @xmath123 is the first time @xmath124 becomes equal to @xmath125 i.e. `` the mean transfer time '' . an analogous definition can be given within tb ii . suppose that initially we place the carrier at a given base , then , e.g. @xmath126 , while all other @xmath120 , @xmath121 and all @xmath127 , @xmath128 . then , if we are interested in @xmath129 or @xmath130 , we can define @xmath131 where `` the mean transfer time '' @xmath123 is the first time @xmath124 becomes equal to @xmath125 or the first time @xmath132 becomes equal to @xmath133 , respectively . it is possible to give a more general definition either for tb i or tb ii . from an initial base pair in tb i or base in tb ii , @xmath32 , to a final base pair in tb i or base in tb ii , @xmath33 , @xmath134 where @xmath135 and all other @xmath136 . + [ bb ] for tb ( i ) , the matrix @xmath95 is a symmetric tridiagonal matrix , i.e. , @xmath137 . % } \ ] ] for tb ii , the matrix @xmath95 has a different form , i.e. , @xmath138 \textrm { ( for dimers)}\ ] ] @xmath139 \textrm { ( for trimers)}\ ] ] @xmath140 \textrm { ( for tetramers)}\ ] ] and so on . + [ cc ] we present the fourier analysis figures mentioned in the main text . [ h ! ] [ h ! ] [ h ! ] [ h ! ] [ h ! ] 99 a.m. kannan , v. renugopalakrishnan , s. filipek , p. li , g.f . audette , and l. munukutla , bio - batteries and bio - fuel cells : leveraging on electronic charge transfer proteins , j. nanosci . nanotechnol . * 9 * ( 2009 ) 1665 . storm , j. van noort , s. de vries , and c. dekker , insulating behavior for dna molecules between nanoelectrodes at the 100 nm length scale , appl . * 79 * ( 2001 ) 3881 d. porath , a. bezryadin , s. de vries , and c. dekker , direct measurement of electrical transport through dna molecules , nature * 403 * ( 2000 ) 635 . h. cohen , c. nogues , r. naaman , and d. porath , direct measurement of electrical transport through single dna molecules of complex sequence , proc . * 102 * ( 2005 ) 11589 . yoo , d.h . ha , j .- o . lee , j.w . park , j. kim , j.j . kim , h .- y . lee , t. kawai , and h.y . choi , electrical conduction through poly(da)-poly(dt ) and poly(dg)-poly(dc ) dna molecules , phys . * 87 * ( 2001 ) 198102 . fink and c. schnenberger , electrical conduction through dna molecules , nature * 398 * ( 1999 ) 407 . b. xu , p. zhang , x. li , and n. tao , direct conductance measurement of single dna molecules in aqueous solution , nano lett . * 4 * ( 2004 ) 1105 . c. simserides , a systematic study of electron or hole transfer along dna dimers , trimers and polymers , chem . phys . 440 ( 2014 ) 31 . k. lambropoulos , m. chatzieleftheriou , a. morphis , k. kaklamanis , m. theodorakou , and c. simserides , unbiased charge oscillations in b - dna : monomer polymers and dimer polymers , phys . e 92 ( 2015 ) 032725 . r. gutirrez , r. caetano , p.b . woiczikowski , t. kubar , m. elstner and g. cuniberti , structural fluctuations and quantum transport through dna molecular wires : a combined molecular dynamics and model hamiltonian approach , new journal of physics 12 ( 2010 ) 023022 . abouzar , a. poghossian , a.g . cherstvy , a.m. pedraza , s. ingebrandt , and m.j . schning , label - free electrical detection of dna by means of field - effect nanoplate capacitors : experiments and modeling , phys . status solidi a * 209 * , 925 ( 2012 ) . a. poghossian , m. weil , a.g . cherstvy , m.j . schning , electrical monitoring of polyelectrolyte multilayer formation by means of capacitive field - effect devices , anal . * 405 * , 6425 ( 2013 ) . ` http://www.nwchem-sw.org/index.php/release62:rt-tddft ` y. takimoto , f.d . vila and j.j . rehr , real - time time - dependent density functional theory approach for frequency - dependent nonlinear optical response in photonic molecules , j. chem . ( 2007 ) 154114 . k. lopata and n. govind , modeling fast electron dynamics with real - time time - dependent density functional theory : application to small molecules and chromophores , j. chem . theory comput . 7 ( 2011 ) 1344 . x. yin , b.w .- h ng , and d. abbott , chapter 2 : terahertz sources and detectors , in terahertz imaging for biomedical applications : pattern recognition and tomographic reconstruction , springer science+business media , llc 2012 , ` http://www.springer.com/978-1-4614-1820-7 ` , isbn 978 - 1 - 4614 - 1820 - 7 e - isbn 978 - 1 - 4614 - 1821 - 4 , ( springer , new york , dordrecht , heidelberg , london , 2012 ) . k. lambropoulos , charge transfer in small dna segments : description at the base - pair level , diploma thesis , national and kapodistrian university of athens , greece , 2014 . g. cuniberti , e. maci , a. rodrguez , r.a . rmer , chapter 1 , tight - binding modeling of charge migration in dna devices , in the book charge migration in dna : perspectives from physics , chemistry , and biology , edited by t. chakraborty , springer berlin heidelberg new york ( 2007 ) e. maci barber , chapter 6 , the aperiodic crystal of life , in aperiodic structures in condensed matter , fundamentals and applications , crc press , taylor and francis group , llc 2009 , isbn-13 : 978 - 1 - 4200 - 6827 - 6 . k. lambropoulos , k. kaklamanis , g. georgiadis , m. theodorakou , m. chatzieleftheriou , m. tassi , a. morphis , and c. simserides , thz oscillations in dna monomers , dimers and trimers , in proceedings of piers ( 2015 ) ( progress in electromagnetics research symposium ) , prague , czech republic ( the electromagnetics academy , cambridge , ma , 2015 ) , pp . 879 - 883 . l. g. d. hawke , g. kalosakas and c. simserides , electronic parameters for charge transfer along dna , eur . j. e , 32 ( 2010 ) 291 . , erratum to : electronic parameters for charge transfer along dna , 34 ( 2011 ) 118 . albuquerque , u.l . fulco , v.n . freire , e.w.s . caetano , m.l . lyra , f.a.b.f . de moura , dna - based nanobiostructured devices : the role of quasiperiodicity and correlation effects , physics reports * 535 * ( 2014 ) 139 . g. cuniberti , e. maci , a. rodrguez , and r.a . rmer , chapter : tight - binding modeling of charge migration in dna devices , pp . 1 - 20 in t. chakraborty ( editor ) , charge migration in dna perspectives from physics , chemistry , and biology , 2007 , springer - verlag berlin heidelberg . c. simserides , 2016 , quantum optics and lasers . athens : hellenic academic libraries link . ( http://hdl.handle.net/11419/2108 ) license : attribution - non - commercial - no derivatives . uri : http://hdl.handle.net/11419/2108 isbn : 978 - 960 - 603 - 073 - 4 http://repository.kallipos.gr/handle/11419/2108?locale=en

we present results of a chandra observation of a field in ngc 2264 . the observations were taken with the acis - i camera with an exposure time of 48.1 ks . we present a catalog of 263 sources , which includes x - ray luminosity , optical and infrared photometry and x - ray variability information . we found 41 variable sources , 14 of which have a flare - like light curve , and 2 of which have a pattern of a steady increase or decrease over a 10 hour period . the optical and infrared photometry for the stars identified as x - ray sources are consistent with most of these objects being pre - main sequence stars with ages younger than 3 myr .

astro-ph0401533

ngc 2264 is a cluster of young stars , part of the mon ob1 association . it is about 3 myrs old @xcite , and the shape of its initial mass function is very similar to that of orion @xcite . pre - main sequence ( pms ) stars have been identified in ngc 2264 using a variety of methods , including @xmath0 spectroscopy @xcite , @xmath0 narrow band photometry @xcite , irregular variability @xcite , near ir photometry @xcite , proper motions @xcite , and x - ray flux @xcite . furthermore , several techniques have been used to identify circumstellar disk candidates , including excess ultraviolet ( @xmath1 ) emission , excess near ir ( @xmath2 and @xmath3 ) emission , and @xmath0 emission line equivalent widths ( e.g. * ? ? ? @xcite have found a good correlation between disk indicators and report lower limits for the disk fraction ranging from 21% to 56% . they also found a typical value for mass accretion rates of @xmath4 , comparable to the values derived for orion and taurus - auriga . ngc 2264 provides a laboratory for studying the interrelationships of rotation , mass accretion , disk indicators and x - ray luminosity of pms stars . the question of how exactly these things are related is still an open problem in star formation phenomenology . there is a clear relationship between rotation rate ( period ) and x - ray luminosity ( @xmath5 ) found in late - type stars in clusters as old as ngc 2547 ( 15 - 40 myrs , * ? ? ? * ) through the hyades ( @xmath6500 myrs , * ? ? ? the ratio between the x - ray and bolometric luminosity , @xmath5/@xmath7 , increases with increasing rotation rate , until the most rapidly rotating stars reach a maximum x - ray luminosity ( or saturation level ) such that @xmath5/@xmath7@xmath8 ( see * ? ? ? * and references within ) . it is much less clear that rotation and @xmath5/@xmath7 are related in younger clusters ( e.g. * ? ? ? @xcite see no obvious correlation between rotation and @xmath5/@xmath7 for their orion ( @xmath6 1 myr old ; * ? ? ? * ) sample , and conclude that the x - ray generation mechanism for young pms stars must be different from that responsible in young main sequence stars . @xcite analyze data for a number of young associations and clusters ( including orion ) , and agree that there is little correlation between @xmath5/@xmath7 and rotation at very young ages , but conclude that the data are consistent with a single physical mechanism , where the orion - age stars are simply all at or near the saturation level ( and that level and the critical velocity for it are a function of gravity / age ) . we want to study at what age the relationship between rotation and @xmath5/@xmath7 emerges . ngc 2264 ( @xmath6 3 myrs old ; * ? ? ? * ; * ? ? ? * ) , being slightly older than orion , is at an ideal age to probe the relationship between these parameters . chandra observations provide a unique tool to improve the x - ray sample of pms stars in ngc 2264 . the high spatial resolution provided by the acis camera allows us to resolve the source confusion present in previous rosat data samples . the chandra sensitivity extends the x - ray flux limit from the rosat value of log@xmath5(erg / s)@xmath630.1 to log@xmath5(erg / s)@xmath628.5 , allowing us to detect lower mass stars and explore the mass dependence of the rotation and @xmath5 relationship . in the present paper , we present results of chandra observations of ngc 2264 . we discuss source detection , variability and @xmath5 determination , providing a catalog of 263 x - ray sources . in paper ii @xcite , we will discuss in more detail the relationships found here between rotation rate , mass accretion rate , disk indicators , and x - ray luminosity . we rely heavily on data from our earlier papers in this cluster , @xcite and @xcite , providing optical photometry and periods , respectively . ngc 2264 was observed with the advanced ccd imaging spectrometer ( acis ) detector on board the _ chandra x - ray observatory _ @xcite on 2002 february 9 . the results presented here arise from the imaging array ( acis - i ) , which consists of four 1024@xmath91024 front - side illuminated ccds . the array is centered at @xmath10 and covers an area on the sky of about @xmath11 . figure [ dss_field ] shows a @xmath12 image of ngc 2264 from the palomar digital sky survey @xcite , with the chandra acis - i field of view marked as a box . the field was selected to maximize the number of stars in the field of view with known periods from @xcite and with minimal overlap with another contemporaneous chandra ngc2264 observation @xcite . the total exposure time of the acis observations is 48.1 ks . we started data analysis with the level 1 processed event list provided by the pipeline processing at the _ chandra _ x - ray center . we have kept all events , including the ones flagged as cosmic ray afterglows by the pipeline . the presence of spurious sources due to cosmic ray afterglows was eliminated using the light curves of the detected sources ( see sec . [ variability ] ) . the energy and grade of each data event were corrected for charge transfer inefficiency ( cti ) , applying the algorithm code described in @xcite . then , the event file was filtered by asca grades ( keeping grades 0 , 2 , 3 , 4 , and 6 ) ; by time intervals ; and by background flaring due to solar activity . removal of the background flares identified by the latter step reduced the exposure time to 47.4 ks . the filtering process was done using the chandra interactive analysis of observations ( ciao ) package and following the science threads provided by the _ chandra _ x - ray center . finally , the energy range was restricted to 0.3 10 kev . figure [ image ] shows the acis - i image of the filtered observations . the ciao package provides three source detection tools , as part of their @xmath13 software . the @xmath14 tool uses a sliding square cell to search for statistically significant enhancements over the background . this tool has been widely used in einstein and rosat data , and works well in the detection of faint sources outside crowded fields . the @xmath15 tool determines individual densities for every occupied pixel , and analyzes the distribution of densities for significant source enhancements . this tools works well in the detection of low surface brightness features , but combines closely spaced point sources . the @xmath16 tool performs a mexican hat wavelet decomposition and reconstruction of the image and then searches for significant correlations . this tool works well in separating closely spaced point sources . we choose the @xmath16 tool to determine the x - ray sources in our @xmath17 observations , since it is the tool that properly handles crowded fields like ours . the algorithm used by @xmath16 is described in @xcite . we used wavelet scales ranging from 1 to 16 pixels in steps of @xmath18 , and the default source significance threshold of @xmath19 . the _ wavdetect _ tool was run separately in the four acis - i ccds images and it produced an original list of 313 sources ( see sec [ xray_phot ] ) . the positions of the sources obtained by _ wavdetect _ were correlated with @xmath20band positions @xcite from the optical / infrared catalog of ngc 2264 stars compiled for our project ( see description of the catalog in sec . [ catalog ] ) . each x - ray source was manually checked to confirm that the optical / infrared counterpart lay within the radius for count extraction ( defined in sec . [ xray_phot ] ) . we checked the astrometry of the chandra observations using all x - ray sources located within an off - axis angle ( @xmath21 ) less than 5 arcmin that have @xmath20band counterparts from @xcite . a total of 80 sources meet this criteria . we determined a mean offset in r.a . of 0.74 @xmath22 0.20 and a mean offset in dec of 0.14 @xmath22 0.20 between the chandra and @xmath20band coordinates . figure [ offset ] shows the coordinate offsets of the 80 sources within @xmath23 and with @xmath20band counterparts . the mean offset is marked by a cross . the chandra positions were corrected by these mean offsets to have the x - ray sources in the same reference frame as their optical counterparts . x - ray aperture photometry was initially performed on the 313 sources detected by the ciao tool _ wavdetect_. the radius for count extraction , @xmath24 , is different for each source , given the variation of the point spread function ( psf ) across the field . @xcite defined the 95% and 99% encircled energy radii ( @xmath25(95%ee ) and @xmath25(99%ee ) ) , as a function of off - axis angle ( @xmath21 ) , in their footnote number 12 . we set @xmath24 equal to @xmath25(95%ee ) , except for sources with higher than 1000 counts , where we enlarged @xmath24 to be equal to @xmath25(99%ee ) . the x - ray counts ( @xmath26 ) were extracted within a circular region of radius @xmath24 , using the ciao tool _ dmextract_. we also defined an annulus around each source for background determination . the annulus is defined to be between 1.2@xmath25(99%ee ) and 1.5@xmath25(99%ee ) . the background counts were also extracted using the ciao tool _ dmextract_. we computed the background in counts@xmath9arcsec@xmath27 as a function of off - axis angle ( @xmath21 ) , performing a 3@xmath28 rejection fit to avoid background counts from annuli that have other sources within . indeed , 38 annuli contained detected sources . the computed background was constant as a function of @xmath21 , and it had a value of @xmath29=(0.063@xmath220.023)counts@xmath9arcsec@xmath27 . the net count , _ n. c. _ , for each source was computed as : @xmath30 to compute the count rate , _ c. r. _ , we also need to determine the enclosed fraction of the psf , @xmath31 , and the effective exposure time , @xmath32 . we computed @xmath31 , using the ciao tool _ mkpsf_. the effective exposure time is determined from exposure maps , which were generated by the ciao tool _ mkexpmap_. the exposure maps are images of effective area that contain information about instrumental artifacts that are both energy- and position - dependent . the effective exposure time is the exposure time corrected to account for a 1.3% loss due to readout , and scaled to the effective area within @xmath24 with respect to the effective area at the optical axis . the count rate is then computed as : @xmath33 we carefully inspected the light curves of all the sources ( see sec . [ variability ] ) and their appearance in the image of the chandra field of view . a total of 48 ( 15% ) sources were rejected from the original list : 43 sources had light curves consistent with cosmic ray afterglows , 2 sources were detected twice , since _ wavdetect _ was run separately in each ccd , and 3 sources had net counts less than zero . of the 43 light curves having a cosmic ray shape , 40 ( 93% ) contain cosmic ray afterglows flagged by the pipeline . our final list of 263 x - ray sources is listed in table [ tab_xray ] . we selected all sources with more than 500 net counts , extracted their spectra and fit them to measure their x - ray fluxes . a total of 15 sources meeting this criteria are listed in table [ tab_spectra ] . the spectra are extracted within @xmath24 using the ciao tool _ dmextract_. the spectra of these 15 brightest x - ray sources are shown in figure [ spectra ] . the redistribution matrix files ( rmf ) and auxiliary response files ( arf ) files contain instrument response information that are essential for the spectral fitting process . we use the rmf files provided by the cti corrector by @xcite . arf files were created for each source , using the ciao tool _ mkarf_. the computed arfs were later corrected to account for the acis low energy quantum efficiency degradation , using the acisabs absorption model . we used the ciao tool _ sherpa _ for the spectral fitting . we adopted a photoelectric absorption model ( @xmath34 ) , which uses wisconsin cross sections from @xcite . this model has one parameter that is the equivalent hydrogen column density ( @xmath35 ) . the hydrogen column density was fixed to a value of 0.08@xmath36 to match an extinction value of @xmath37 = 0.41 , which is the most likely value of the observed extinction towards ngc 2264 , although the @xmath37 can be as high as 3.5 mag ( see * ? ? ? * for more discussion ) . the value of the column density was derived using the relationship of @xmath38 . if we consider a @xmath39 , which corresponds to three magnitudes more than the assumed @xmath37 , the derived @xmath5 would be only 0.1 dex less . thus , the error in @xmath5 that comes from fixing @xmath35 is negligible . we also adopted a thermal emission model ( @xmath40 ) , which is based on model calculations of @xcite , and @xcite with fe l calculations by @xcite . this model includes line emissions from several elements . the remaining two parameters in the model - the plasma temperature ( kt ) and a normalization factor - were varied in order to fit the spectra . we initially tried one temperature models , but they failed to fit the low energy part of the spectrum . a two temperature model significantly improved the spectral fitting for all the sources . the brightest x - ray source in the sample 196 ( s mon ) has a distinctive soft x - ray spectrum , characteristic of its early spectral type ( see sec . [ ostar ] ) . it also has a high count rate of 155 counts ks@xmath41 , and may be affected by pileup ( at high count rate the observed flux is no longer a linear function of the count rate ) . given the spectral characteristics of s mon and its possible pileup effects , we did not include it in the determination of the conversion factor that leads to the determination of x - ray luminosity for our sample of x - ray sources . the spectral parameters obtained from the two plasma temperature models are listed in table [ tab_spectra ] . the x - ray flux of s mon listed in table [ tab_spectra ] is the integration of the two temperature model between 0.3 and 8.0 kev . the x - ray flux for the rest of the bright sources is determined from the best spectral model derived from the mean model parameters . we computed mean plasma temperatures of ( 0.51@xmath220.06 ) kev and ( 2.5@xmath220.3 ) kev . the mean plasma temperatures are held constant and the integration of the best fit between 0.3 and 8.0 kev provides the x - ray flux . the resulting x - ray fluxes are listed in table [ tab_spectra ] . we use the x - ray fluxes of the 14 bright sources to compute a x - ray flux weighted conversion factor between count rate and x - ray flux . the obtained conversion factor is ( 6.16@xmath220.13)@xmath42 ( erg/@xmath43/s)/(counts / ks ) . figure [ cfactor ] shows the relationship between @xmath44 and x - ray flux for the 14 x - ray sources used to determine the conversion factor . the solid line corresponds to the derived conversion factor . we compiled a list of conversion factors obtained by published studies using acis - i data in young stars ; see table [ tab_compare ] . the values we list for @xcite and @xcite were computed using the count rate and x - ray luminosity they provide in their tables 3 and 1 respectively . our conversion factor is in good agreement with the ones derived by @xcite and @xcite , and it differs from the other values by a factor of less than 2 . the x - ray flux for our catalog of x - ray sources in ngc 2264 , listed in column 10 of table [ tab_xray ] , was computed using the derived conversion factor . the x - ray luminosity , @xmath5 , listed in column 11 of table [ tab_xray ] , is computed assuming a distance to ngc 2264 of 760 pc @xcite . the limiting luminosity in our x - ray observations varies within the field of view , because of the variation of the psf across the field . in figure [ limit ] , we have plotted the net rate of the detected x - ray sources as a function of the off - axis angle , @xmath21 . the faintest source is located at @xmath45 and it has a count rate of 0.08 counts / ks , corresponding to a x - ray luminosity of 28.5 at the distance of ngc 2264 . about 80 % of our sources are located within @xmath46 . at that off - axis angle the limiting count rate has increased to 0.12 counts / ks , corresponding to a x - ray luminosity of log(@xmath5)=28.7 at the distance of ngc 2264 . at @xmath47 , we can not detected x - ray sources fainter than 0.35 counts / ks ( log(@xmath5 ) = 29.2 ) . therefore , we adopt a value of log(@xmath5)=28.5 dex as the limiting x - ray luminosity for our observations , keeping in mind that this value holds for sources located within @xmath45 . @xcite observed a 40@xmath9100 field towards ngc2 2264 using rosat , detecting 169 sources . our chandra field ( 17@xmath917 ) is covered within the spatial extent of the rosat observations but goes 1.5 dex deeper in log(@xmath5 ) ( a factor of @xmath6 30 in @xmath5 ) . there are 34 sources in common between our catalog of x - ray sources in ngc 2264 and the rosat sample of @xcite . we compared our x - ray luminosities with the ones given by @xcite ( see figure [ rosat ] ) . we found a mean difference of 0.1 dex with a standard deviation of 0.3 dex . this difference does not include an offset of + 0.3 dex in the rosat @xmath5 , as described in @xcite . given that many of these pms stars are highly variable ( see sec . [ variability ] ) , we conclude that our luminosities are consistent with respect to those of @xcite . light curves were determined for all 313 sources detected by _ wavdetect _ using the ciao tool _ lightcurves _ with a bin time of 2500 s. the statistics of the light curves of the sources of our x - ray catalog were obtained using the xronos tool _ lcstats_. this provides , among other values , the constant source probability as derived from the chi - square value comparing the data with a constant light curve , @xmath48 . the @xmath48 values are listed in column 12 of table [ tab_xray ] . the light curves of the sources with @xmath49 90% were analyzed further , since they are the most likely to be variable . we obtained the reduced @xmath50 of those sources , using two additional light curves with bin times of 5000 s and 7500 s. if the reduced @xmath50 obtained for both bin times was less than 2.5 , then we consider those sources to have constant light curves . then , we defined a variable source as those having 2500 s bin light curves with @xmath49 90% , and 5000 s and 7500 s bin light curves with reduced @xmath51 2.5 . there are 41 variable sources that meet this criteria . variable sources are marked with a v in column 13 of table [ tab_xray ] . among the variable sources , ten have known periods from @xcite , but we find no obvious dependence of x - ray flux with rotational phase for 9 of these stars ( see bellow the discussion about source 181 ) . there are 14 variable sources which show a flare shape , defined as a rapid increase and a slow decrease in the x - ray flux . variable sources with a flare - like light curve are marked with an additional f in column 13 of table [ tab_xray ] . in figure [ light_curves ] , we show a selection of light curves of sources of comparable luminosity . in the top panels , we show two light curves with a flare shape ; in the middle panels , there are two variable light curves . finally in the bottom panels , two constant light curves are plotted . there are six sources that show a possible flare pattern , described as an increase in x - ray flux happening at the end of our observations or a decrease in x - ray flux occurring at the beginning of our observations . variable sources with a possible flare pattern are marked with an additional p in column 13 of table [ tab_xray ] . @xcite detected 10 flaring objects in orion using rosat observations , over a sample of 389 x - ray sources ( @xmath6 3% of flaring objects ) ; @xcite detected 14 flaring objects in ic 348 using chandra observations over a sample of 215 x - ray sources ( @xmath6 7 % of flaring objects ) ; we detected a comparable fraction , about 8% flaring sources ( 5% excluding possible flares ) . there are two additional variable sources which show a distinctive pattern of variation , either a steady increase or a steady decrease in x - ray flux . the sources that show this pattern are 75 ( r 2752 ) and 181 ( r 3245 ) , they are marked with an additional s in column 13 of table [ tab_xray ] , and their light curves are shown in figure [ light_curves_2 ] . the amount of variation in these two cases is about a factor of 10 in about 10 hours . a similar pattern is observed in the light curves of three x - ray ic 348 sources determined by @xcite using chandra observations . they mention that those light curves may be explained by rotational variability , as described by @xcite , who present a model of rotational modulation of x - ray flares in three t tauri stars and algol . the amplitude of the variations modeled by @xcite vary from a factor of 2.5 to a factor of 20 , depending on the strength of the x - ray flare . although both x - ray sources 181 and 75 have optical counterparts , only 181 ( r 3245 ) has an observed period of 1.21 days ( 29 hrs ) from @xcite . if the lightcurve of 181 can be understood by rotational modulation of x - ray flares , then the observed decay time ( @xmath6 11 hrs ) should be less than half of the stellar period . therefore , the period implied by the x - ray light curve should be greater than 22 hrs , which is consistent with the optical observations . we correlated the x - ray sources with a catalog of stars in ngc2264 ; see table [ tab_optical ] . we constructed the catalog by combining our original @xmath52 survey from @xcite ( `` r '' names ) with other published catalogs . if optical photometry and/or spectral types were not available in @xcite , we took photometry from @xcite , @xcite , or @xcite , in that order , augmented by 15 spectral types from the simbad database . all of the @xmath53 photometry comes from 2mass , with the exception of a handful of stars slightly fainter than the 2mass limit ; those @xmath53 values come from @xcite . we included several other catalogs primarily to keep track of commonly - used nomenclature ; see table [ tab_xids ] . @xcite , updated by @xcite , lists spectral types and is the origin of the `` lkha '' nomenclature ; @xcite provides spectral types , membership probabilities , and the commonly - used walker ( or `` w '' ) names ; @xcite conducted an h@xmath54 survey to identify young stars and provides some spectral types ; and @xcite and @xcite are the origin of the `` vsb '' and `` hbc '' names , respectively . @xcite conducted a rosat survey of this cluster , and we included these data in our catalog as well ( see sec . [ xray_lum ] above ) . @xcite rosat sources are named as `` fx '' . because we are also interested in rotational information ( for our future work ) , we included in our catalog periods from @xcite and @xcite , as well as projected rotational velocities ( @xmath55 ) from @xcite , @xcite and @xcite . we found that 213 ( 81% ) of our 263 x - ray sources have optical and/or infrared counterparts ( 108 ( 41% ) have spectral types and 44 ( 17% ) have known periods ) . those 213 sources and their corresponding optical and/or infrared photometry are listed in table [ tab_optical ] . most , but probably not all , of these 213 stars are likely to be members of ngc2264 . other names of the sources given by the different catalogs used in our compilation are listed in table [ tab_xids ] . there are 747 stars in our optical / infrared catalog with @xmath56 , @xmath57 , and @xmath58 photometry and with positions inside the field of view of the chandra observations . among those 747 stars , 199 ( 27% ) have x - ray chandra counterparts . figure [ histogram ] shows a @xmath56 magnitude histogram of all the 747 stars with @xmath56 , @xmath57 , and @xmath58 photometry and positions inside the chandra field ( solid line ) and the histogram of the 199 x - ray chandra counterparts ( dotted line ) . the completeness limit of the infrared sample is essentially that of 2mass . we can see in figure [ histogram ] that our infrared sample goes deeper than the sample of stars with x - ray counterparts . a similar behavior is seen in the optical sample histogram . this means that all the x - ray sources should have been matched to sources in our optical / infrared catalog if they are associated with stars earlier than m0 at the distance of ngc 2264 . there are 51 x - ray sources that do not have optical nor infrared counterparts . based on the limiting magnitude of our optical catalog ( @xmath59 19 mag ) , the x - ray to optical flux ratio , @xmath60 , of the sources with only x - ray detection is greater than about @xmath61 . this means that these sources could be either m stars or extragalactic objects @xcite . ngc 2264 is located 2 degrees above the galactic plane , roughly towards near the anticenter . therefore , most of the m stars should be foreground , and hence detected by 2mass . considering the presence of a dark cloud behind ngc 2264 , the detection of background m stars in x - rays is unlikely . source counts in the chandra south and north deep field @xcite predict the presence of about 100 agn in the acis field of view at the flux limit of our observations . thus , we believe that the most likely explanation is that the 51 objects detected only in x - rays are active galaxies . in figure [ color_color ] , we have plotted the @xmath62 , @xmath3 color - color diagram of all the infrared sources in the field of view of our chandra observation . most of the x - ray sources with infrared colors are located near the locus of the classical t - tauri ( ctt ) stars @xcite . there are two x - ray sources that do not follow the position of the rest of the sources . they are sources 161 and s mon ( 196 ) and they are marked in the color - color diagram with their x - ray identification numbers . s mon ( see sec . [ ostar ] , is a o star ; its position in figure [ color_color ] is correct in @xmath3 , but its 2mass @xmath56 magnitude is suspect and can be explained by contaminated from a nearby diffraction spike . the second source , 161 , is a field galaxy ( see sec . [ galaxy ] ) . there are 686 stars in our optical / infrared catalog with @xmath63 and @xmath64 photometry and with positions inside the field of view of the chandra observations . among those 686 stars , 201 have x - ray chandra counterparts . in figure [ cmd ] , we have plotted ( @xmath65)@xmath66 color magnitude diagrams of the optical sources in the field of view of our chandra observations . the dereddened @xmath65 color and the absolute magnitude @xmath66 were obtained assuming an average extinction of @xmath37 = 0.41 , and dereddening relationships from @xcite and @xcite . in both panels , we have also plotted isochrones of from @xcite , as a reference . the dashed line corresponds to @xmath66 = 8.75 mag . , which is the lower limit for a low mass star rotating at the saturation level ( @xmath5/@xmath7=3 ) with @xmath5=28.5 ( limit luminosity of our x - ray sample ) . this means that the most slowly rotating stars with optical counterparts should be detected by our x - ray sample . for comparison , the corresponding line from the rosat @xcite study would be at @xmath66 = 5 mag . most of our x - ray sources are younger than 3@xmath67 years at the distance of ngc 2264 , and furthermore the x - ray sources are heavily concentrated towards the youngest isochrones with respect to the general population . all the sources with @xmath66 brighter than 0.1 magnitudes have @xmath5/@xmath7 @xmath68 . most of those sources have spectral types earlier than a0 ( see right panel of figure [ cmd ] ) . there are 5 x - ray sources that appear to be much older than @xmath69 years , of which one of them ( 12 ) is not a member of ngc 2264 , two of them ( 146 & 273 ) lack proper motion membership information , and two of them ( 132 & 259 ) are probably members ( using preliminary membership information as discussed in * ? ? ? unfortunately , the optical and infrared photometry of source 259 is unreliable , because it is contaminated by a saturated column from a nearby bright star . source 132 is classified as a m0 star . there are four sources that are located above the @xmath70 year old isochrone , of which two of them ( 75 & 136 ) are not members of ngc 2264 , one of them ( 60 ) lacks proper motion membership information , but it is an a0 star , and one of them ( 279 ) is probably a member ( using preliminary membership information as discussed in * ? ? ? source 279 is classified as a k5 star . we have dereddened sources 279 and 132 , using the intrinsic @xmath62 , @xmath3 and @xmath71 colors of stars of their respective spectral types . for source 279 , we derive a @xmath37 value of 5.7 magnitudes and for source 132 , @xmath37 is about 3.0 magnitudes . the locations of sources 279 and 132 are indicated along with their dereddened positions in figure [ color_color ] . the dereddened colors of source 279 are consistent with colors of a k star , similar to its spectral type , and the dereddened colors of source 132 fall in the locus of ctt stars . in figure [ cmd3 ] , we have plotted the @xmath72 color magnitude diagram of all the infrared sources with x - ray counterparts , with isochrones from @xcite . the location of sources 279 and 132 is indicated along with their dereddened position . it is possible that these two objects are the youngest and most embedded of our x - ray sources . the unusual optical colors of source 132 could be explained by scattered light from an edge on disk . there are 16 stars for which rotation and spectral type information exist in our optical / infrared catalog , and for which no x - ray counterpart was found . in order to allow us to use these stars in the next paper , we determined upper limits for their x - ray luminosity in the following way . we extracted the counts with a @xmath73 equal to the 95% encircled energy radius . we computed the background counts the same way as our detected sources . we used the bayesian statistics method from @xcite to determine the net count upper limit to a confidence level of 95% . then , we computed @xmath31 and @xmath32 in the same manner as our detected sources , to determine the upper limit for the count rate . the upper limit for the x - ray flux was determined using our derived conversion factor , and the upper limit for the x - ray luminosity was computed assuming a distance to ngc 2264 of 760 pc @xcite . the upper limit results are listed in table [ tab_upp_lim ] . x - ray source 196 is associated with s mon , a o7 v star . the x - ray spectrum of s mon is well fitted by two plasma temperatures of 0.18 kev and 0.57 kev , but nearly 80% of the total x - ray flux comes from the 0.18 kev component . we derive a log(@xmath5/@xmath7)=6.78 , which is in good agreement with the canonical value of log(@xmath5/@xmath7 ) @xmath67 observed in o stars @xcite . @xcite provide a catalog of ob stars observed in the rosat all - sky survey , which includes s mon . they derive ( from hardness ratios ) a x - ray plasma temperature of 0.23 kev and log(@xmath5/@xmath7)=6.63 , which are in good agreement with the values we derive considering the differences in the applied techniques ( spectral fitting vs. hardness ratios ) . the x - ray emission mechanism for early type stars is thought to be very different from the magnetic reconnection flares that generate the x - rays in low mass young stars . models predict that the x - ray emission of hot stars comes from shocks formed within a radiatively driven wind @xcite , producing x - rays of about 0.5 kev . the shocks may be formed by line - forced instabilities @xcite . evidence coming from grating x - ray spectroscopy suggests that magnetically confined hot plasma near the surface of the star may be important in some early type stars that show an unusually hard x - ray spectrum ( plasma temperature higher than 2 kev ) and narrow x - ray emission lines ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? the lack of a hard component in the x - ray spectrum of s mon may suggest that its x - ray emission is likely produced by shocks within the radiatively driven stellar wind . walker 90 is an early - type emission - line star that lies below the zero - age main sequence in the @xmath64 , @xmath74 color magnitude diagram @xcite , suggesting very non selective extinction . @xcite confirmed the existence of anomalous circumstellar dust extinction , which they explained by a graphite - silicate mixture with larger grains than those present in the diffuse interstellar medium . walker 90 is in our chandra field , but no x - ray source was detected at its position . an upper limit for its x - ray luminosity is provided in table [ tab_upp_lim ] . our x - ray source 161 is coincident in position and has similar morphology to the extended infrared source 2masx j06405286 + 0948570 . the @xmath56 , @xmath57 , and @xmath58 magnitudes from the 2mass extended source catalog are listed in table [ tab_optical ] . this source is also a known radio source , with a flux of 6.45 mjy / beam at 3.6 cm @xcite and a flux of 61.9 mjy at 20 cm @xcite . given its radio and x - ray emission , we conclude that source 161 is a radio - quiet agn @xcite , seen though the molecular cloud . we present a catalog of ngc 2264 x - ray sources . the observations were taken with the acis - i on board the chandra x - ray observatory . the catalog , consisting of 263 sources , includes x - ray luminosity , optical and infrared photometry and x - ray variability information . we found 41 variable sources , 14 of which have a flare like light curve , and 2 of which have a pattern of a steady increase or decrease . from the optical and infrared counterparts of the x - ray sources , we have learned that most of the x - ray sources have colors consistent with ctts that are younger than 3@xmath67 years . this catalog of x - ray sources will be used to study the relationship between rotational properties and x - ray characteristics of ngc 2264 stars in paper ii @xcite . we plan to discuss correlations of @xmath5/@xmath7 with rotation rate ( period and @xmath75sin@xmath76 ) , disk indicators ( @xmath77 , @xmath3 , @xmath1 , and @xmath0 ) , and mass accretion rate as derived from @xmath1 excess . we will also compare the @xmath5/@xmath7 values found here with those from other young clusters . svr gratefully thanks jeonghee rho , kenji hamaguchi , ettore flaccomio , peter freeman & scott wolk for the useful correspondence about x - ray and chandra data processing and analysis . svr also thanks august muench , paul eskridge , richard pogge , phil appleton , mark lacy , and dario fadda for interesting discussions . we thank the anonymous referee for her / his careful review of the manuscript . financial support for this work was provided by nasa grant go2 - 3011x . this research has made extensive use of nasa s astrophysics data system abstract service , the simbad database , operated at cds , strasbourg , france , and the nasa / ipac infrared science archive , which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration . the research described in this paper was partially carried out at the jet propulsion laboratory , california institute of technology , under a contract with the national aeronautics and space administration . brandt , w. n. , alexander , d. m. , hornschemeier , a. e. , garmire , g. p. , schneider , d. p. , barger , a. j. , bauer , f. e. , broos , p. s. , cowie , l. l. , townsley , l. k. , burrows , d. n. , chartas , g. , feigelson , e. d. , griffiths , r. e. , nousek , j. a. , & sargent , w. l. w. , 2001 , , 122 , 2810 harnden , f. r. , jr . , adams , n. r. , damiani , f. , drake , j. j. , evans , n. r. , favata , f. , flaccomio , e. , freeman , p. , jeffries , r. d. , kashyap , v. , micela , g. , patten , b. m. , pizzolato , n. , schachter , j. f. , sciortino , s. , stauffer , j. , wolk , s. j. , zombeck , m. v. , 2001 , , 547 , l141 reid , i. n. , brewer , c. , brucato , r. j. , mckinley , w. r. , maury , a. , mendenhall , d. , mould , j. r. , mueller , j. , neugebauer , g. , phinney , j. , sargent , w. l. w. , schombert , j. , & thicksten , r. , 1991 , , 661 rosati , p. , tozzi , p. , giacconi , r. , gilli , r. , hasinger , g. , kewley , l. , mainieri , v. , nonino , m. , norman , c. , szokoly , g. , wang , j. x. , zirm , a. , bergeron , j. , borgani , s. , gilmozzi , r. , grogin , n. , koekemoer , a. , schreier , e. , zheng , w. , 2002 , , 566 , 667 rcllcccccccrc 1 & cxorrs j064009.6 + 095338 & 6 40 9.69 & 9 53 38.355 & 9.85 & 14.07 & 0.99 & 40.5 & 0.37 & 0.228e-14 & 29.20 & 100 & + 3 & cxorrs j064012.8 + 094853 & 6 40 12.84 & 9 48 53.952 & 8.78 & 11.07 & 0.97 & 30.2 & 0.16 & 0.986e-15 & 28.83 & 100 & + 4 & cxorrs j064013.5 + 095313 & 6 40 13.57 & 9 53 13.437 & 8.79 & 11.12 & 0.98 & 42.2 & 0.21 & 0.129e-14 & 28.95 & 99 & + 6 & cxorrs j064018.3 + 094416 & 6 40 18.31 & 9 44 16.054 & 9.72 & 13.73 & 0.99 & 31.0 & 0.94 & 0.579e-14 & 29.60 & 100 & + 8 & cxorrs j064018.5 + 095206 & 6 40 18.57 & 9 52 6.259 & 7.34 & 7.72 & 0.98 & 41.9 & 0.52 & 0.320e-14 & 29.35 & 100 & + 9 & cxorrs j064019.3 + 094829 & 6 40 19.34 & 9 48 29.537 & 7.37 & 7.77 & 0.98 & 41.0 & 3.30 & 0.203e-13 & 30.15 & 65 & v , f + 11 & cxorrs j064020.1 + 094511 & 6 40 20.16 & 9 45 11.315 & 8.78 & 11.12 & 0.99 & 39.4 & 4.95 & 0.305e-13 & 30.32 & 98 & + 12 & cxorrs j064020.9 + 094405 & 6 40 20.91 & 9 44 5.391 & 9.36 & 12.69 & 0.99 & 39.3 & 2.01 & 0.124e-13 & 29.93 & 99 & + 13 & cxorrs j064020.9 + 094821 & 6 40 20.93 & 9 48 21.477 & 7.04 & 7.08 & 0.97 & 40.1 & 0.67 & 0.413e-14 & 29.46 & 100 & + 14 & cxorrs j064020.9 + 095519 & 6 40 20.99 & 9 55 19.184 & 8.05 & 9.30 & 0.98 & 40.9 & 0.65 & 0.400e-14 & 29.44 & 100 & + 15 & cxorrs j064021.8 + 095209 & 6 40 21.86 & 9 52 9.178 & 6.54 & 6.15 & 0.97 & 44.0 & 0.34 & 0.209e-14 & 29.16 & 100 & + 16 & cxorrs j064022.2 + 095428 & 6 40 22.28 & 9 54 28.859 & 7.33 & 7.68 & 0.98 & 39.2 & 14.69 & 0.905e-13 & 30.80 & 0 & v , f + 18 & cxorrs j064022.6 + 094945 & 6 40 22.62 & 9 49 45.912 & 6.28 & 5.71 & 0.97 & 38.5 & 0.90 & 0.554e-14 & 29.58 & 99 & + 19 & cxorrs j064022.9 + 095312 & 6 40 22.99 & 9 53 12.159 & 6.62 & 6.30 & 0.97 & 40.9 & 0.41 & 0.253e-14 & 29.24 & 100 & + 20 & cxorrs j064023.3 + 095454 & 6 40 23.39 & 9 54 54.251 & 7.34 & 7.72 & 0.98 & 39.1 & 0.71 & 0.437e-14 & 29.48 & 99 & + 22 & cxorrs j064023.7 + 095523 & 6 40 23.77 & 9 55 23.330 & 7.55 & 8.17 & 0.98 & 37.8 & 6.18 & 0.381e-13 & 30.42 & 96 & + 23 & cxorrs j064024.8 + 095311 & 6 40 24.89 & 9 53 11.135 & 6.17 & 5.51 & 0.97 & 39.8 & 1.32 & 0.813e-14 & 29.75 & 99 & + 24 & cxorrs j064025.5 + 094825 & 6 40 25.52 & 9 48 25.778 & 5.96 & 5.17 & 0.96 & 44.0 & 3.27 & 0.201e-13 & 30.14 & 99 & + 25 & cxorrs j064026.9 + 095138 & 6 40 26.97 & 9 51 38.500 & 5.24 & 4.13 & 0.96 & 40.1 & 0.38 & 0.234e-14 & 29.21 & 100 & + 26 & cxorrs j064027.3 + 095307 & 6 40 27.36 & 9 53 7.553 & 5.59 & 4.62 & 0.96 & 44.8 & 0.25 & 0.154e-14 & 29.03 & 100 & + 27 & cxorrs j064027.6 + 095349 & 6 40 27.66 & 9 53 49.571 & 5.87 & 5.02 & 0.97 & 44.5 & 0.28 & 0.172e-14 & 29.08 & 100 & + 28 & cxorrs j064027.8 + 094119 & 6 40 27.85 & 9 41 19.667 & 10.61 & 16.48 & 1.00 & 38.9 & 0.42 & 0.259e-14 & 29.25 & 99 & + 29 & cxorrs j064028.8 + 094823 & 6 40 28.85 & 9 48 23.232 & 5.22 & 4.08 & 0.96 & 41.7 & 2.89 & 0.178e-13 & 30.09 & 99 & + 30 & cxorrs j064029.0 + 094217 & 6 40 29.01 & 9 42 17.432 & 9.63 & 13.43 & 0.99 & 40.3 & 5.46 & 0.336e-13 & 30.37 & 50 & v + 32 & cxorrs j064029.3 + 094407 & 6 40 29.34 & 9 44 7.148 & 8.01 & 9.20 & 0.98 & 41.8 & 0.23 & 0.142e-14 & 28.99 & 100 & + 33 & cxorrs j064029.4 + 094736 & 6 40 29.45 & 9 47 36.838 & 5.47 & 4.43 & 0.96 & 41.4 & 0.28 & 0.172e-14 & 29.08 & 100 & + 34 & cxorrs j064029.9 + 095010 & 6 40 29.93 & 9 50 10.049 & 4.44 & 3.15 & 0.95 & 45.4 & 1.75 & 0.108e-13 & 29.87 & 99 & + 36 & cxorrs j064030.6 + 095014 & 6 40 30.61 & 9 50 14.085 & 4.28 & 3.00 & 0.95 & 45.5 & 0.65 & 0.400e-14 & 29.44 & 100 & + 37 & cxorrs j064030.6 + 094610 & 6 40 30.68 & 9 46 10.398 & 6.20 & 5.56 & 0.97 & 40.4 & 3.65 & 0.225e-13 & 30.19 & 27 & v , f + 38 & cxorrs j064031.6 + 094449 & 6 40 31.62 & 9 44 49.861 & 7.11 & 7.23 & 0.98 & 39.4 & 0.35 & 0.216e-14 & 29.17 & 100 & + 39 & cxorrs j064031.6 + 094427 & 6 40 31.68 & 9 44 27.278 & 7.44 & 7.92 & 0.98 & 39.5 & 0.89 & 0.548e-14 & 29.58 & 100 & + 40 & cxorrs j064031.6 + 094823 & 6 40 31.70 & 9 48 23.290 & 4.64 & 3.35 & 0.95 & 40.8 & 0.25 & 0.154e-14 & 29.03 & 100 & + cccccc 196 & r 3295 & 155.8 & 0.18 & 0.57 & 14.10 + 113 & r 2942 & 84.5 & 0.30 & 1.36 & 5.25 + 181 & r 3245 & 54.9 & 0.76 & 3.92 & 3.45 + 75 & r 2752 & 42.1 & 0.80 & 5.16 & 2.48 + 94 & r 2840 & 36.4 & 0.40 & 2.21 & 2.34 + 203 & r 3309 & 27.4 & 0.58 & 2.34 & 1.68 + 142 & r 3091 & 24.1 & 1.00 & 4.46 & 1.58 + 251 & r 3470 & 21.3 & 0.34 & 1.25 & 1.26 + 228 & r 3390 & 18.5 & 0.64 & 2.58 & 1.04 + 60 & r 2633 & 18.0 & 0.30 & 1.66 & 0.79 + 208 & r 3323 & 17.6 & 0.25 & 1.11 & 1.03 + 230 & r 3394 & 17.2 & 0.32 & 1.62 & 0.95 + 268 & r 3555 & 15.9 & 0.68 & 3.28 & 0.93 + 16 & r 2173 & 14.7 & 0.41 & 2.23 & 0.83 + 214 & r 3342 & 13.3 & 0.32 & 2.08 & 0.76 + cllrrrrrrrrr r 1817 & 6 40 10.00 & 9 53 41.19 & 1 & 18.47 & 18.36 & 16.93 & 16.00 & 14.94 & 13.53 & 12.94 & 12.73 + r 2065 & 6 40 18.52 & 9 44 18.73 & 6 & 19.65 & 19.09 & 17.50 & 16.51 & 15.48 & 14.06 & 13.40 & 13.15 + r 2066 & 6 40 18.53 & 9 52 05.73 & 8 & & 20.28 & 18.94 & 17.94 & 16.15 & 14.55 & 13.89 & 13.65 + r 2093 & 6 40 19.37 & 9 48 29.83 & 9 & 18.03 & 17.38 & 15.90 & 15.00 & 13.99 & 12.59 & 11.99 & 11.89 + r 2137 & 6 40 20.87 & 9 44 11.29 & 12 & & 21.94 & 20.51 & & 18.06 & 16.32 & 15.84 & 15.22 + r 2163 & 6 40 21.84 & 9 52 09.11 & 15 & & & 19.92 & & 17.00 & 15.41 & 14.76 & 14.59 + r 2173 & 6 40 22.28 & 9 54 28.65 & 16 & 18.20 & 17.16 & 15.62 & 14.61 & 13.56 & 12.29 & 11.56 & 11.31 + r 2182 & 6 40 22.68 & 9 49 45.92 & 18 & 19.41 & 18.57 & 17.23 & 16.20 & 15.12 & 13.88 & 13.12 & 12.87 + r 2190 & 6 40 23.01 & 9 53 12.08 & 19 & & & 20.50 & 18.79 & 17.16 & 15.49 & 14.78 & 14.34 + r 2204 & 6 40 23.48 & 9 54 55.30 & 20 & & 20.18 & 18.54 & 17.29 & 15.44 & 14.01 & 13.35 & 12.97 + r 2217 & 6 40 23.79 & 9 55 23.42 & 22 & 18.59 & 19.03 & 17.55 & 16.32 & 14.92 & 13.12 & 12.26 & 11.81 + r 2251 & 6 40 24.84 & 9 53 11.45 & 23 & 19.89 & 19.23 & 17.49 & & 15.04 & 13.59 & 12.56 & 12.27 + r 2271 & 6 40 25.52 & 9 48 25.88 & 24 & 17.97 & 16.83 & 15.51 & 14.66 & 13.88 & 13.24 & 12.61 & 12.45 + r 2374 & 6 40 28.66 & 9 48 24.31 & 29 & & & 15.76 & 15.02 & 14.09 & 13.00 & 12.32 & 12.04 + r 2383 & 6 40 29.01 & 9 42 17.18 & 30 & 17.80 & 16.79 & 15.47 & 14.66 & 13.92 & 12.94 & 12.24 & 12.05 + r 2391 & 6 40 29.29 & 9 44 07.42 & 32 & 19.17 & 18.85 & 17.68 & 16.62 & 15.40 & 14.06 & 13.34 & 13.06 + r 2401 & 6 40 29.46 & 9 47 36.88 & 33 & 19.79 & 20.54 & 20.15 & 17.74 & 16.77 & 14.54 & 13.89 & 13.50 + r 2419 & 6 40 29.95 & 9 50 10.25 & 34 & 16.51 & 15.62 & 14.44 & 13.78 & 13.16 & 12.22 & 11.68 & 11.56 + r 2442 & 6 40 30.64 & 9 50 14.38 & 36 & & 18.83 & 17.10 & 16.06 & 14.53 & 12.75 & 11.67 & 11.19 + r 2443 & 6 40 30.66 & 9 46 10.59 & 37 & 10.21 & 16.63 & 15.32 & 14.61 & 13.90 & 12.86 & 12.23 & 12.06 + r 2474 & 6 40 31.70 & 9 48 23.26 & 40 & 17.25 & 16.73 & 15.57 & 14.78 & 14.08 & 13.02 & 12.35 & 12.18 + r 2477 & 6 40 31.73 & 9 53 30.09 & 41 & & & & & 17.31 & 14.46 & 13.11 & 12.49 + r 2476 & 6 40 31.73 & 9 49 59.20 & 42 & & & & & 17.39 & 15.57 & 14.90 & 14.61 + r 2507 & 6 40 32.65 & 9 49 32.90 & 44 & & & 20.02 & 18.19 & 16.37 & 14.94 & 14.28 & 14.05 + r 2503 & 6 40 32.42 & 9 41 45.13 & 45 & 20.48 & 20.03 & 18.51 & 17.25 & 15.81 & 14.29 & 13.59 & 13.34 + r 2511 & 6 40 32.85 & 9 51 28.97 & 46 & 18.12 & 17.06 & 15.68 & 14.78 & 13.87 & 12.68 & 12.00 & 11.79 + r 2518 & 6 40 33.18 & 9 49 54.31 & 48 & & & 20.50 & 18.60 & 17.08 & 14.36 & 12.98 & 12.16 + r 2550 & 6 40 33.85 & 9 48 43.64 & 50 & 19.25 & 18.62 & 17.21 & 16.11 & 14.76 & 13.45 & 12.72 & 12.35 + r 2590 & 6 40 34.87 & 9 45 45.11 & 51 & 20.20 & 19.77 & 18.24 & 17.11 & 15.85 & 14.23 & 13.59 & 13.24 + r 2593 & 6 40 34.94 & 9 54 06.87 & 52 & & 20.10 & 18.23 & 17.05 & 15.35 & 13.74 & 13.05 & 12.81 + r 2597 & 6 40 35.23 & 9 51 56.25 & 53 & 19.00 & 17.76 & 16.20 & 15.17 & 14.01 & 12.61 & 11.83 & 11.47 + r 2614 & 6 40 36.06 & 9 47 35.97 & 56 & & & 19.53 & & 16.24 & 14.29 & 13.94 & 13.51 + r 2631 & 6 40 36.58 & 9 50 45.47 & 57 & 17.89 & 18.13 & 16.93 & 15.91 & 14.59 & 12.93 & 12.14 & 11.71 + r 2635 & 6 40 36.69 & 9 48 22.75 & 58 & 16.94 & 16.72 & 15.61 & 14.85 & 14.06 & 13.04 & 12.33 & 12.21 + r 2636 & 6 40 36.71 & 9 52 02.89 & 59 & 17.23 & 17.51 & 16.27 & 15.33 & 14.35 & 12.73 & 11.76 & 11.10 + r 2633 & 6 40 36.67 & 9 47 22.53 & 60 & 13.12 & 13.20 & 12.43 & & 10.91 & 9.65 & 9.00 & 8.59 + rllll 1 & r 1817 & & 2mass j06400993 + 0953415 & + 6 & r 2065 & sung 1487 & 2mass j06401846 + 0944188 & + 8 & r 2066 & & 2mass j06401847 + 0952060 & + 9 & r 2093 & sung 68 & 2mass j06401930 + 0948299 & + 12 & r 2137 & sung 1526 & 2mass j06402081 + 0944114 & + 15 & r 2163 & sung 1539 & 2mass j06402178 + 0952092 & + 16 & r 2173 & sung 78 & 2mass j06402221 + 0954288 & + 18 & r 2182 & sung 79 & 2mass j06402262 + 0949462 & + 19 & r 2190 & sung 1560 & 2mass j06402295 + 0953125 & + 20 & r 2204 & sung 1567 & 2mass j06402342 + 0954555 & + 22 & r 2217 & sung 1578 & 2mass j06402373 + 0955238 & v594 mon , ogura 74 + 23 & r 2251 & sung 1599 & 2mass j06402484 + 0953114 & + 24 & r 2271 & sung 88 & 2mass j06402547 + 0948259 & + 29 & r 2374 & & & ogura 76 + 30 & r 2383 & & & walker 52 + 32 & r 2391 & sung 1705 & 2mass j06402924 + 0944075 & + 33 & r 2401 & sung 1708 & 2mass j06402941 + 0947369 & + 34 & r 2419 & sung 104 & 2mass j06402989 + 0950104 & walker 54,vsb 40 + 36 & r 2442 & sung 1739 & 2mass j06403059 + 0950147 & + 37 & r 2443 & sung 108 & 2mass j06403061 + 0946106 & ogura 81,fx 15,vsb 170 + 40 & r 2474 & sung 113 & 2mass j06403164 + 0948233 & walker 58,v413 mon + 41 & r 2477 & & 2mass j06403168 + 0953304 & + 42 & r 2476 & & 2mass j06403167 + 0949593 & + 44 & r 2507 & sung 1773 & 2mass j06403258 + 0949332 & + 45 & r 2503 & sung 1767 & 2mass j06403237 + 0941449 & + 46 & r 2511 & sung 118 & 2mass j06403280 + 0951293 & fx 17 + 48 & r 2518 & sung 1781 & 2mass j06403311 + 0949547 & + 50 & r 2550 & sung 1798 & 2mass j06403378 + 0948438 & + 51 & r 2590 & sung 1827 & 2mass j06403482 + 0945452 & + 52 & r 2593 & sung 1828 & 2mass j06403489 + 0954071 & + 53 & r 2597 & sung 128 & 2mass j06403518 + 0951567 & ogura 85 + 56 & r 2614 & sung 1841 & & + 57 & r 2631 & sung 129 & 2mass j06403652 + 0950456 & + 58 & r 2635 & sung 130 & 2mass j06403662 + 0948229 & fx 19 + 59 & r 2636 & sung 131 & 2mass j06403665 + 0952032 & + 60 & r 2633 & sung 132 & 2mass j06403667 + 0947225 & walker 66,v780 mon , vsb 46 + llllccccc r 2772 & sung 153 & 2mass 06404102 + 0947577 & & @xmath78 4.11 & @xmath78 0.0955 & @xmath780.0056 & @xmath78 28.61 + r 2898 & sung 173 & 2mass 06404464 + 0948021 & walker 90 , v590 mon , hbc 219 , vsb 62 & @xmath78 3.00 & @xmath78 0.0703 & @xmath780.0041 & @xmath78 28.48 + r 2982 & sung 2066 & 2mass 06404729 + 0947274 & & @xmath7812.63 & @xmath78 0.3940 & @xmath780.0231 & @xmath78 29.22 + r 2491 & sung 1760 & 2mass 06403200 + 094935 & & @xmath78 3.00 & @xmath78 0.1794 & @xmath780.0105 & @xmath78 28.88 + & & & v590 mon , hbc 219 , herbig 25 & @xmath78 4.11 & @xmath78 0.0962 & @xmath780.0056 & @xmath78 28.61 + r 2792 & sung 159 & 2mass 06404156 + 0955174 & vsb 56 & @xmath78 3.00 & @xmath78 0.1984 & @xmath780.0116 & @xmath78 28.93 + & & & ln mon , hbc 215 , herbig 21 & @xmath78 4.82 & @xmath78 0.1193 & @xmath780.0070 & @xmath78 28.71 + r 3167 & sung 2178 & 2mass 06405413 + 0948434 & & @xmath78 4.11 & @xmath78 0.0944 & @xmath780.0055 & @xmath78 28.60 + r 3216 & sung 2206 & 2mass 06405573 + 0946456 & & @xmath78 3.82 & @xmath78 0.0903 & @xmath780.0053 & @xmath78 28.58 + & & & lr mon , hbc 220 , herbig 27 & @xmath78 4.11 & @xmath78 0.1816 & @xmath780.0106 & @xmath78 28.89 + r 2972 & sung 2056 & 2mass 06404694 + 0955036 & & @xmath78 7.24 & @xmath78 0.1697 & @xmath780.0099 & @xmath78 28.86 + r 3022 & sung 195 & mass 06404888 + 0951444 & walker 100 , hd261841 , vsb 72 & @xmath78 7.40 & @xmath78 0.1677 & @xmath780.0098 & @xmath78 28.85 + r 3041 & sung 196 & 2mass 06404953 + 0953230 & walker 104 , vsb 74 & @xmath78 6.78 & @xmath78 0.1561 & @xmath780.0092 & @xmath78 28.82 + r 3073 & sung 2125 & 2mass 06405033 + 0954158 & & @xmath78 4.11 & @xmath78 0.0958 & @xmath780.0056 & @xmath78 28.61 + r 3110 & sung 209 & 2mass 06405155 + 0951494 & walker 109 , hd261878 , vsb 79 & @xmath78 5.82 & @xmath78 0.1328 & @xmath780.0078 & @xmath78 28.75 + & & & fx 67 , vsb 245 & @xmath78 3.94 & @xmath78 0.0926 & @xmath780.0054 & @xmath78 28.60 +

polynomial interpretations are a useful technique for proving termination of term rewrite systems . they come in various flavors : polynomial interpretations with real , rational and integer coefficients . as to their relationship with respect to termination proving power , lucas managed to prove in 2006 that there are rewrite systems that can be shown polynomially terminating by polynomial interpretations with real ( algebraic ) coefficients , but can not be shown polynomially terminating using polynomials with rational coefficients only . he also proved the corresponding statement regarding the use of rational coefficients versus integer coefficients . in this article we extend these results , thereby giving the full picture of the relationship between the aforementioned variants of polynomial interpretations . in particular , we show that polynomial interpretations with real or rational coefficients do not subsume polynomial interpretations with integer coefficients . our results hold also for incremental termination proofs with polynomial interpretations .

1407.0406

polynomial interpretations are a simple yet useful technique for proving termination of term rewrite systems ( trss , for short ) . while originally conceived in the late seventies by lankford @xcite as a means for establishing direct termination proofs , polynomial interpretations are nowadays often used in the context of the dependency pair ( dp ) framework @xcite . in the classical approach of lankford , one considers polynomials with integer coefficients inducing polynomial algebras over the well - founded domain of the natural numbers . to be precise , every @xmath0-ary function symbol @xmath1 is interpreted by a polynomial @xmath2 in @xmath0 indeterminates with integer coefficients , which induces a mapping or _ interpretation _ from terms to integer numbers in the obvious way . in order to conclude termination of a given trs , three conditions have to be satisfied . first , every polynomial must be _ well - defined _ , i.e. , it must induce a well - defined polynomial function @xmath3 over the natural numbers . in addition , the interpretation functions @xmath4 are required to be _ strictly monotone _ in all arguments . finally , one has to show _ compatibility _ of the interpretation with the given trs . more precisely , for every rewrite rule @xmath5 , the polynomial @xmath6 associated with the left - hand side must be greater than @xmath7 , the corresponding polynomial of the right - hand side , i.e. , @xmath8 for all values of the indeterminates . already back in the seventies , an alternative approach using polynomials with real coefficients instead of integers was proposed by dershowitz @xcite . however , as the real numbers @xmath9 equipped with the standard order @xmath10 are not well - founded , a subterm property is explicitly required to ensure well - foundedness . it was not until 2005 that this limitation was overcome , when lucas @xcite presented a framework for proving polynomial termination over the real numbers , where well - foundedness is basically achieved by replacing @xmath10 with a new ordering @xmath11 requiring comparisons between terms to not be below a given positive real number @xmath12 . moreover , this framework also facilitates polynomial interpretations over the rational numbers . thus , one can distinguish three variants of polynomial interpretations , polynomial interpretations with real , rational and integer coefficients , and the obvious question is : what is their relationship with regard to termination proving power ? for knuth - bendix orders it is known @xcite that extending the range of the underlying weight function from natural numbers to non - negative reals does not result in an increase in termination proving power . in 2006 lucas @xcite proved that there are trss that can be shown polynomially terminating by polynomial interpretations with rational coefficients , but can not be shown polynomially terminating using polynomials with integer coefficients only . likewise , he proved that there are trss that can be handled by polynomial interpretations with real ( algebraic ) coefficients , but can not be handled by polynomial interpretations with rational coefficients . in this article we extend these results and give a complete comparison between the various notions of polynomial termination .. ] in general , the situation turns out to be as depicted in figure [ fig : summary ] , which illustrates both our results and the earlier results of lucas @xcite . ( 0,0 ) node[anchor = south west ] terminating trss rectangle ( 13cm,4 cm ) ; ( 7cm,20 mm ) ellipse ( 36 mm and 13 mm ) ; ( 86mm,25 mm ) ellipse ( 9 mm and 9 mm ) ; ( 75mm,20 mm ) ellipse ( 2 cm and 10 mm ) ; ( 7cm,20 mm ) ellipse ( 36 mm and 13 mm ) ; ( 32mm,12 mm ) rectangle + + ( 7mm,14 mm ) ; ( 36.5mm,20 mm ) node @xmath13 ; ( 52mm,17.5 mm ) rectangle + + ( 5mm,5 mm ) ; ( 54.5mm,20 mm ) node @xmath14 ; ( 92mm,27 mm ) ellipse ( 11 mm and 11 mm ) ; ( 79mm,22 mm ) rectangle + + ( 5mm,5 mm ) ; ( 81.5mm,24.5 mm ) node @xmath15 ; ( 45mm,20 mm ) node @xmath16 ; ( 70mm,20 mm ) node @xmath17 ; ( 98mm,24 mm ) node @xmath18 ; ( 95mm,34 mm ) node @xmath19 ; in particular , we prove that polynomial interpretations with real coefficients subsume polynomial interpretations with rational coefficients . moreover , we show that polynomial interpretations with real or rational coefficients do not subsume polynomial interpretations with integer coefficients by exhibiting the trs @xmath20 in section [ sect : nvsr ] . likewise , we prove that there are trss that can be shown terminating by polynomial interpretations with real coefficients as well as by polynomial interpretations with integer coefficients , but can not be shown terminating using polynomials with rational coefficients only , by exhibiting the trs @xmath21 in section [ sect : nrvsq ] . the trss @xmath22 and @xmath23 can be found in section [ sect : incrementality ] . the remainder of this article is organized as follows . in section [ sect : prelim ] , we introduce some preliminary definitions and terminology concerning polynomials and polynomial interpretations . in section [ sect : qvsr ] , we show that polynomial interpretations with real coefficients subsume polynomial interpretations with rational coefficients . we further show that for polynomial interpretations over the reals , it suffices to consider real algebraic numbers as interpretation domain . section [ sect : nvsr ] is dedicated to showing that polynomial interpretations with real or rational coefficients do not subsume polynomial interpretations with integer coefficients . then , in section [ sect : nrvsq ] , we present a trs that can be handled by a polynomial interpretation with real coefficients as well as by a polynomial interpretation with integer coefficients , but can not be handled using polynomials with rational coefficients . in section [ sect : incrementality ] , we show that the relationships in figure [ fig : summary ] remain true if incremental termination proofs with polynomial interpretations are considered . we conclude in section [ sect : conclusion ] . this paper is an extended version of @xcite , which contained the result of section [ sect : nvsr ] . the results in sections [ sect : qvsr ] , [ sect : nrvsq ] and [ sect : incrementality ] are new . as usual , we denote by @xmath15 , @xmath24 , @xmath14 and @xmath9 the sets of natural , integer , rational and real numbers , respectively . an _ irrational _ number is a real number , which is not in @xmath14 . given some @xmath25 and @xmath26 , @xmath27 denotes the standard order of the respective domain and @xmath28 . a sequence of real numbers @xmath29 _ converges _ to the _ limit _ @xmath30 if for every real number @xmath31 there exists a natural number @xmath32 such that the absolute distance @xmath33 is less than @xmath34 for all @xmath35 ; we denote this by @xmath36 . as convergence in @xmath37 is equivalent to componentwise convergence , we use the same notation also for limits of converging sequences of vectors of real numbers @xmath38 . a real function @xmath39 is _ continuous _ in @xmath37 if for every converging sequence @xmath38 it holds that @xmath40 . finally , as @xmath14 is dense in @xmath9 , every real number is a rational number or the limit of a converging sequence of rational numbers . for any ring @xmath41 ( e.g. @xmath24 , @xmath14 , @xmath9 ) , we denote the associated _ polynomial ring _ in @xmath0 _ indeterminates _ @xmath42 by @xmath43 $ ] , the elements of which are finite sums of products of the form @xmath44 , where the _ coefficient _ @xmath45 is an element of @xmath41 and the exponents @xmath46 in the _ monomial _ @xmath47 are non - negative integers . if @xmath48 , we call a product @xmath44 a _ term_. the _ degree _ of a monomial is just the sum of its exponents , and the degree of a term is the degree of its monomial . an element @xmath49 $ ] is called an _ ( @xmath0-variate ) polynomial with coefficients in @xmath41_. for example , the polynomial @xmath50 is an element of @xmath51 $ ] , the ring of all univariate polynomials with integer coefficients . in the special case @xmath52 , a polynomial @xmath53 $ ] can be written as follows : @xmath54 ( @xmath55 ) . for the largest @xmath56 such that @xmath57 , we call @xmath58 the _ leading term _ of @xmath59 , @xmath60 its _ leading coefficient _ and @xmath56 its _ degree _ , which we denote by @xmath61 . a polynomial @xmath53 $ ] is said to be _ linear _ if @xmath62 , and _ quadratic _ if @xmath63 . we assume familiarity with the basics of term rewriting and polynomial interpretations ( e.g. @xcite ) . the key concept for establishing ( direct ) termination of trss via polynomial interpretations is the notion of well - founded monotone algebras as they induce reduction orders on terms . [ def : ma ] let @xmath64 be a signature , i.e. , a set of function symbols equipped with fixed arities . an @xmath64-_algebra _ @xmath65 consists if a non - empty _ carrier _ set @xmath66 and a collection of interpretation functions @xmath67 for each @xmath0-ary function symbol @xmath68 . the _ evaluation _ or _ interpretation _ @xmath69_\mathcal{a}(t)$ ] of a term @xmath70 with respect to a variable assignment @xmath71 is inductively defined as follows : @xmath72_\mathcal{a}(t ) = \begin{cases } \alpha(t ) & \text{if $ t \in \mathcal{f}$ } \\ f_a([\alpha]_\mathcal{a}(t_1),\dots,[\alpha]_\mathcal{a}(t_n ) ) & \text{if $ t = f({{t_1},\dots,{t_{n}}})$ } \end{cases}\ ] ] let @xmath73 be a binary relation on @xmath66 . for @xmath74 , an interpretation function @xmath75 is _ monotone in its @xmath76-th argument _ with respect to @xmath73 if @xmath77 implies @xmath78 for all @xmath79 . it is said to be _ monotone _ with respect to @xmath73 if it is monotone in all its arguments . we define @xmath80 as @xmath69_\mathcal{a}(s ) \sqsupset [ \alpha]_\mathcal{a}(t)$ ] for all assignments @xmath81 . in order to pave the way for incremental polynomial termination in section [ sect : incrementality ] , the following definition is more general than what is needed for direct termination proofs . [ def : wma ] let @xmath82 be an @xmath64-algebra together with two binary relations @xmath83 and @xmath84 on @xmath66 . we say that @xmath85 and a trs @xmath86 are _ ( weakly ) compatible _ if @xmath87 ( @xmath88 ) for each rewrite rule @xmath89 . an interpretation function @xmath90 is called strictly ( weakly ) monotone if it is monotone with respect to @xmath83 ( @xmath84 ) . the triple @xmath82 ( or just @xmath65 if @xmath83 and @xmath84 are clear from the context ) is a _ weakly ( strictly ) monotone @xmath64-algebra _ if @xmath83 is well - founded , @xmath91 and for each @xmath68 , @xmath90 is _ weakly ( strictly ) monotone_. it is said to be an _ extended monotone @xmath64-algebra _ if it is both weakly monotone and strictly monotone . finally , we call @xmath82 a _ well - founded monotone @xmath65-algebra _ if @xmath83 is a well - founded order on @xmath66 , @xmath84 is its reflexive closure , and each interpretation function is strictly monotone . it is well - known that well - founded monotone algebras provide a complete characterization of termination . a trs is terminating if and only if it is compatible with a well - founded monotone algebra . [ def : polyint_n ] a _ polynomial interpretation over @xmath15 _ for a signature @xmath64 consists of a polynomial @xmath92 $ ] for every @xmath0-ary function symbol @xmath68 such that for all @xmath68 the following two properties are satisfied : 1 . _ well - definedness _ : @xmath93 for all @xmath94 , 2 . _ strict monotonicity _ of @xmath4 in all arguments with respect to @xmath95 , the standard order on @xmath15 . due to well - definedness , each of the polynomials @xmath4 induces a function from @xmath96 to @xmath15 . hence , the pair @xmath97 constitutes an @xmath64-algebra over the carrier @xmath15 . now @xmath98 where @xmath99 is the reflexive closure of @xmath95 constitutes a well - founded monotone algebra , and we say that a polynomial interpretation over @xmath15 is _ compatible _ with a trs @xmath86 if the well - founded monotone algebra @xmath98 is compatible with @xmath86 . finally , a trs is _ polynomially terminating over @xmath15 _ if it admits a compatible polynomial interpretation over @xmath15 . in the sequel , we often identify a polynomial interpretation with its associated @xmath64-algebra . [ rem : isom ] in principle , one could take any set @xmath100 ( or even @xmath101 ) instead of @xmath15 as the carrier for polynomial interpretations . however , it is well - known @xcite that all these sets are order - isomorphic to @xmath15 and hence do not change the class of polynomially terminating trss . in other words , a trs @xmath86 is polynomially terminating over @xmath15 if and only if it is polynomially terminating over @xmath100 . thus , we can restrict to @xmath15 as carrier without loss of generality . the following simple criterion for strict monotonicity of a univariate quadratic polynomial will be used in sections [ sect : nvsr ] and [ sect : nrvsq ] . [ lem : nquadmon ] the quadratic polynomial @xmath102 with @xmath103 is strictly monotone and well - defined if and only if @xmath104 , @xmath105 , and @xmath106 . now if one wants to extend the notion of polynomial interpretations to the rational or real numbers , the main problem one is confronted with is the non - well - foundedness of these domains with respect to the standard orders @xmath107 and @xmath10 . in @xcite , this problem is overcome by replacing these orders with new non - total orders @xmath11 and @xmath108 , the first of which is defined as follows : given some fixed positive real number @xmath12 , @xmath109 analogously , one defines @xmath108 on @xmath14 . thus , @xmath11 ( @xmath108 ) is well - founded on subsets of @xmath9 ( @xmath14 ) that are bounded from below . therefore , any set @xmath110 ( @xmath111 ) could be used as carrier for polynomial interpretations over @xmath9 ( @xmath14 ) . however , without loss of generality we may restrict to @xmath112 ( @xmath113 ) because the main argument of remark [ rem : isom ] also applies to polynomials over @xmath9 ( @xmath14 ) , as is already mentioned in @xcite . [ def : polyint_r ] a _ polynomial interpretation over @xmath9 _ for a signature @xmath64 consists of a polynomial @xmath114 $ ] for every @xmath0-ary function symbol @xmath68 and some positive real number @xmath115 such that @xmath116 is well - defined over @xmath112 , i.e. , @xmath117 for all @xmath118 . analogously , one defines polynomial interpretations over @xmath14 by the obvious adaptation of the definition above . let @xmath119 . as for polynomial interpretations over @xmath15 , the pair @xmath120 constitutes an @xmath64-algebra over the carrier @xmath121 due to the well - definedness of all interpretation functions . together with @xmath122 and @xmath123 , the restrictions of @xmath124 and @xmath125 to @xmath121 , we obtain an algebra @xmath126 , where @xmath122 is well - founded ( on @xmath121 ) and @xmath127 . hence , if for each @xmath68 , @xmath128 is weakly ( strictly ) monotone , that is , monotone with respect to @xmath123 ( @xmath122 ) , then @xmath126 is a weakly ( strictly ) monotone @xmath64-algebra . however , unlike for polynomial interpretations over @xmath15 , strict monotonicity of @xmath126 does not entail weak monotonicity as it can very well be the case that an interpretation function is monotone with respect to @xmath122 but not with respect to @xmath123 . let @xmath119 . a polynomial interpretation over @xmath129 is said to be _ weakly ( strictly ) monotone _ if the algebra @xmath126 is weakly ( strictly ) monotone . similarly , we say that a polynomial interpretation over @xmath129 is _ ( weakly ) compatible _ with a trs @xmath86 if the algebra @xmath126 is ( weakly ) compatible with @xmath86 . finally , a trs @xmath86 is _ polynomially terminating over @xmath129 _ if there exists a polynomial interpretation over @xmath129 that is both compatible with @xmath86 and strictly monotone . we conclude this section with a more useful characterization of monotonicity with respect to the orders @xmath130 and @xmath131 than the one obtained by specializing definition [ def : wma ] . to this end , we note that a function @xmath132 is strictly monotone in its @xmath76-th argument with respect to @xmath130 if and only if @xmath133 for all @xmath134 with @xmath135 . from this and from the analogous characterization of @xmath131-monotonicity , it is easy to derive the following lemmata , which will be used in sections [ sect : nrvsq ] and [ sect : incrementality ] . [ lem : rlinmon ] [ lem : qlinmon ] [ lem : qrlinmon ] for @xmath119 and @xmath136 with @xmath115 , the linear polynomial @xmath137 in @xmath138 $ ] is monotone in all arguments with respect to @xmath122 and well - defined if and only if @xmath139 and @xmath140 for all @xmath74 . [ lem : qrquadmon ] for @xmath119 and @xmath136 with @xmath115 , the quadratic polynomial @xmath141 in @xmath142 $ ] is monotone with respect to @xmath122 and well - defined if and only if @xmath104 , @xmath143 , @xmath144 , and @xmath145 or @xmath146 . in the remainder of this article we will sometimes use the term `` polynomial interpretations with integer coefficients '' as a synonym for polynomial interpretations over @xmath15 . likewise , the term `` polynomial interpretations with real ( rational ) coefficients '' refers to polynomial interpretations over @xmath9 ( @xmath14 ) . [ lucas ] lucas @xcite considers a different definition of polynomial termination over @xmath9 ( @xmath14 ) . he allows an arbitrary subset @xmath147 ( @xmath148 ) as interpretation domain , provided it is bounded from below and unbounded from above . the definition of well - definedness is modified accordingly . according to his definition , polynomial termination over @xmath15 trivially implies polynomial interpretations over @xmath9 ( and @xmath14 ) since one can take @xmath149 and @xmath150 , in which case the induced order @xmath151 is the same as the standard order on @xmath15 . our definitions are based on the understanding that the interpretation domain together with the underlying order determine whether one speaks of polynomial interpretations over the reals , rationals , or integers . as a consequence , several of the new results obtained in this paper do not hold in the setting of @xcite . in this section we show that polynomial termination over @xmath14 implies polynomial termination over @xmath9 . the proof is based upon the fact that polynomials induce continuous functions , whose behavior at irrational points is completely defined by the values they take at rational points . [ lem : continuity ] let @xmath152 be continuous in @xmath37 . if @xmath153 for all @xmath154 , then @xmath153 for all @xmath155 . let @xmath156 and let @xmath157 be a sequence of vectors of non - negative rational numbers @xmath158 whose limit is @xmath159 . such a sequence exists because @xmath160 is dense in @xmath37 . then @xmath161 by continuity of @xmath1 . thus , @xmath162 is the limit of @xmath163 , which is a sequence of non - negative real numbers by assumption . hence , @xmath162 is non - negative , too . [ thm : qvsr ] if a trs is polynomially terminating over @xmath14 , then it is also polynomially terminating over @xmath9 . let @xmath86 be a trs over the signature @xmath64 that is polynomially terminating over @xmath14 . so there exists some polynomial interpretation @xmath164 over @xmath14 consisting of a positive rational number @xmath12 and a polynomial @xmath165 $ ] for every @xmath0-ary function symbol @xmath68 such that : 1 . for all @xmath0-ary @xmath68 , @xmath166 for all @xmath167 , 2 . for all @xmath68 , @xmath168 is strictly monotone with respect to @xmath131 in all arguments , 3 . for every rewrite rule @xmath89 , @xmath169 for all @xmath170 . here @xmath6 ( @xmath7 ) denotes the polynomial associated with @xmath171 ( @xmath172 ) and the variables @xmath173 are those occurring in @xmath5 . next we note that all three conditions are quantified polynomial inequalities of the shape `` @xmath174 for all @xmath154 '' for some polynomial @xmath59 with rational coefficients . this is easy to see for the first and third condition . as to the second condition , the function @xmath168 is strictly monotone in its @xmath76-th argument with respect to @xmath131 if and only if @xmath175 for all @xmath176 with @xmath177 , which is equivalent to @xmath178 for all @xmath176 . from lemma [ lem : continuity ] and the fact that polynomials induce continuous functions we infer that all these polynomial inequalities do not only hold in @xmath113 but also in @xmath112 . hence , the polynomial interpretation @xmath164 proves termination over @xmath9 . [ rem : qvsr ] not only does the result established above show that polynomial termination over @xmath14 implies polynomial termination over @xmath9 , but it even reveals that the same interpretation applies . we conclude this section by showing that for polynomial interpretations over @xmath9 it suffices to consider real _ _ algebraic _ _ numbers as interpretation domain . concerning the use of real algebraic numbers in polynomial interpretations , in ( * ? * section 6 ) it is shown that it suffices to consider polynomials with real algebraic coefficients as interpretations of function symbols . now the obvious question is whether it is also sufficient to consider only the ( non - negative ) real algebraic numbers @xmath179 instead of the entire set @xmath9 of real numbers as interpretation domain . we give an affirmative answer to this question by extending the result of @xcite . a finite trs is polynomially terminating over @xmath9 if and only if it is polynomially terminating over @xmath179 . let @xmath86 be a trs over the signature @xmath64 that is polynomially terminating over @xmath9 . there exists a positive real number @xmath12 and a polynomial @xmath114 $ ] for every @xmath0-ary function symbol @xmath68 such that : 1 . for all @xmath0-ary @xmath68 , @xmath180 for all @xmath118 , 2 . for all @xmath68 , @xmath116 is strictly monotone with respect to @xmath130 in all arguments , 3 . for every rewrite rule @xmath89 , @xmath181 for all @xmath182 . next we treat @xmath12 as a variable and replace all coefficients of the polynomials in @xmath183 by distinct variables @xmath184 . thus , for each @xmath0-ary function symbol @xmath68 , its interpretation function is a parametric polynomial @xmath185 \subseteq { \mathbb{z}}[{{x_1},\dots,{x_{n}}},{{c_1},\dots,{c_{j}}},\delta]$ ] , where all non - zero coefficients are @xmath186 . as a consequence , we claim that all three conditions listed above can be expressed as ( conjunctions of ) quantified polynomial inequalities of the shape @xmath187 for some polynomial @xmath188 $ ] . this is easy to see for the first condition . for the third condition it is a direct consequence of the nature of the interpretation functions and the usual closure properties of polynomials . for the second condition we additionally need the fact that @xmath116 is strictly monotone in its @xmath76-th argument with respect to @xmath130 if and only if @xmath189 for all @xmath134 now any of the quantified inequalities can readily be expressed as a formula in the language of ordered fields with coefficients in @xmath24 , where @xmath184 and @xmath12 are the only free variables . by taking the conjunction of all these formulas , existentially quantifying @xmath12 and adding the conjunct @xmath115 , we obtain a formula @xmath190 in the language of ordered fields with free variables @xmath184 and coefficients in @xmath24 ( as @xmath86 and @xmath64 are assumed to be finite ) . by assumption there are coefficients @xmath191 such that @xmath192 is true in @xmath9 , i.e. , there exists a satisfying assignment for @xmath190 in @xmath9 mapping its free variables @xmath184 to @xmath191 . in order to prove the theorem , we first show that there also exists a satisfying assignment mapping each free variable to a real algebraic number . we reason as follows . because real closed fields admit quantifier elimination ( ( * ? ? ? * theorem 2.77 ) ) , there exists a quantifier - free formula @xmath193 with free variables @xmath184 and coefficients in @xmath24 that is @xmath9-equivalent to @xmath190 , i.e. , for all @xmath194 , @xmath195 is true in @xmath9 if and only if @xmath196 is true in @xmath9 . hence , by assumption , @xmath197 is true in @xmath9 . therefore , the sentence @xmath198 is true in @xmath9 as well . since both @xmath9 and @xmath179 are real closed fields with @xmath199 and all coefficients in this sentence are from @xmath200 , we may apply the tarski - seidenberg transfer principle ( ( * ? ? ? * theorem 2.80 ) ) , from which we infer that this sentence is true in @xmath9 if and only if it is true in @xmath179 . so there exists an assignment for @xmath193 in @xmath179 mapping its free variables @xmath184 to @xmath201 such that @xmath202 is true in @xmath179 , and hence also in @xmath9 as @xmath193 is a boolean combination of atomic formulas in the variables @xmath184 with coefficients in @xmath24 . but then @xmath203 is true in @xmath9 as well because of the @xmath9-equivalence of @xmath190 and @xmath193 . another application of the tarski - seidenberg transfer principle reveals that @xmath203 is true in @xmath179 , and therefore the trs @xmath86 is polynomially terminating over @xmath179 ( whose formal definition is the obvious specialization of definition [ def : polyint_r ] ) . this shows that polynomial termination over @xmath9 implies polynomial termination over @xmath179 . as the reverse implication can be shown to hold by the same technique , we conclude that polynomial termination over @xmath9 is equivalent to polynomial termination over @xmath179 . as far as the relationship of polynomial interpretations with real , rational and integer coefficients with regard to termination proving power is concerned , lucas @xcite managed to prove the following two theorems .. ] [ thm : qvsn ] there are trss that are polynomially terminating over @xmath14 but not over @xmath15 . [ thm : rvsq ] there are trss that are polynomially terminating over @xmath9 but not over @xmath14 or @xmath15 . hence , the extension of the coefficient domain from the integers to the rational numbers entails the possibility to prove some trss polynomially terminating , which could not be proved polynomially terminating otherwise . moreover , a similar statement holds for the extension of the coefficient domain from the rational numbers to the real numbers . based on these results and the fact that we have the strict inclusions @xmath204 , it is tempting to believe that polynomial interpretations with real coefficients properly subsume polynomial interpretations with rational coefficients , which in turn properly subsume polynomial interpretations with integer coefficients . indeed , the former proposition holds according to theorem [ thm : qvsr ] . however , the latter proposition does not hold , as will be shown in this section . in particular , we present a trs that can be proved terminating by a polynomial interpretation with integer coefficients , but can not be proved terminating by a polynomial interpretation over the reals or rationals . in order to motivate the construction of this particular trs , let us first observe that from the viewpoint of number theory there is a fundamental difference between the integers and the real or rational numbers . more precisely , the integers are an example of a discrete domain , whereas both the real and rational numbers are _ _ dense _ _ and @xmath205 , there exists a real ( rational ) number @xmath45 in between . ] domains . in the context of polynomial interpretations , the consequences of this major distinction are best explained by an example . to this end , we consider the polynomial function @xmath206 depicted in figure [ fig : nvsrpoly ] and assume that we want to use it as the interpretation of some unary function symbol . now the point is that this function is permissible in a polynomial interpretation over @xmath15 as it is both non - negative and strictly monotone over the natural numbers . however , viewing it as a function over a real ( rational ) variable , we observe that non - negativity is violated in the open interval @xmath207 ( and monotonicity requires a properly chosen value for @xmath12 ) . hence , the polynomial function @xmath206 is not permissible in any polynomial interpretation over @xmath9 ( @xmath14 ) . ( -0.2,0 ) ( 2.4,0 ) node[right ] @xmath30 ; ( 0,-0.1 ) ( 0,3.7 ) node[above ] ; /in 1/1 , 2/2 , 3/3 , 4/4 , 5/5 , 6/6 , 7/7 ( 2pt,0pt ) ( -2pt,0pt ) node[left ] @xmath208 ; /in 0/0 , 1/1 , 2/2 ( 0pt,2pt ) ( 0pt,-2pt ) node[below ] @xmath209 ; ( 0,0 ) parabola bend ( 0.25,-0.125 ) ( 2.1,6.72 ) ; ( 0,0 ) circle ( 0.05 cm ) ; ( 1,0.5 ) circle ( 0.05 cm ) ; ( 2,3 ) circle ( 0.05 cm ) node[below left ] @xmath210 ; thus , the idea is to design a trs that enforces an interpretation of this shape for some unary function symbol , and the tool that can be used to achieve this is polynomial interpolation . to this end , let us consider the following scenario , which is fundamentally based on the assumption that some unary function symbol @xmath211 is interpreted by a quadratic polynomial @xmath212 with ( unknown ) coefficients @xmath213 , @xmath205 and @xmath45 . then , by polynomial interpolation , these coefficients are uniquely determined by the image of @xmath211 at three pairwise different locations ; in this way the interpolation constraints @xmath214 , @xmath215 and @xmath216 enforce the interpretation @xmath217 . next we encode these constraints in terms of the trs @xmath86 consisting of the following rewrite rules , where @xmath218 abbreviates @xmath219 , 2 ( ) & ( ) + ^2 ( ) & ( ( ) ) & ( ( ) ) & + ^7 ( ) & ( ^2 ( ) ) & ( ^2 ( ) ) & ^5 ( ) and consider the following two cases : polynomial interpretations over @xmath15 on the one hand and polynomial interpretations over @xmath9 on the other hand . in the context of polynomial interpretations over @xmath15 , we observe that if we equip the function symbols @xmath220 and @xmath221 with the ( natural ) interpretations @xmath222 and @xmath223 , then the trs @xmath86 indeed implements the above interpolation constraints . is sufficient for this purpose . ] for example , the constraint @xmath224 is expressed by @xmath225 and @xmath226 . the former encodes @xmath227 , whereas the latter encodes @xmath228 . moreover , the rule @xmath229 encodes @xmath230 , which is equivalent to @xmath231 in the domain of the natural numbers . thus , this interpolation constraint can be expressed by a single rewrite rule , whereas the other two constraints require two rules each . summing up , by virtue of the method of polynomial interpolation , we have reduced the problem of enforcing a specific interpretation for some unary function symbol to the problem of enforcing natural semantics for the symbols @xmath220 and @xmath221 . next we elaborate on the ramifications of considering the trs @xmath86 in the context of polynomial interpretations over @xmath9 . to this end , let us assume that the symbols @xmath220 and @xmath221 are interpreted by @xmath232 and @xmath233 , so that @xmath220 has some kind of _ successor function _ semantics . then the trs @xmath86 translates to the following constraints : 2 s_0 - & _ _ ( 0 ) + 2 s_0 - & _ _ ( s_0 ) & _ ( s_0 ) & _ 0 + + 7 s_0 - & _ _ ( 2 s_0 ) & _ ( 2 s_0 ) & _ 5 s_0 + hence , @xmath234 is confined to the closed interval @xmath235 $ ] , whereas @xmath236 is confined to @xmath237 $ ] and @xmath238 to @xmath239 $ ] . basically , this means that these constraints do not uniquely determine the function @xmath240 . in other words , the method of polynomial interpolation does not readily apply to the case of polynomial interpretations over @xmath9 . however , we can make it work . to this end , we observe that if @xmath241 , then the above system of inequalities actually turns into the following system of equations , which can be viewed as a set of interpolation constraints ( parameterized by @xmath242 ) that uniquely determine @xmath240 : 3 _ ( 0 ) & = 0 & _ ( s_0 ) & = s_0 & _ ( 2 s_0 ) & = 6 s_0 clearly , if @xmath243 , then the symbol @xmath211 is fixed to the interpretation @xmath244 , as was the case in the context of polynomial interpretations over @xmath245 ( note that in the latter case @xmath150 is implicit because of the equivalence @xmath246 ) . hence , we conclude that once we can manage to design a trs that enforces @xmath241 , we can again leverage the method of polynomial interpolation to enforce a specific interpretation for some unary function symbol . moreover , we remark that the actual value of @xmath242 is irrelevant for achieving our goal . that is to say that @xmath242 only serves as a scale factor in the interpolation constraints determining @xmath240 . clearly , if @xmath247 , then @xmath240 is not fixed to the interpretation @xmath244 ; however , it is still fixed to an interpretation of the same ( desired ) shape , as will become clear in the proof of lemma [ lem : mainlemma1 ] . in the previous subsection we have presented the basic method that we use in order to show that polynomial interpretations with real or rational coefficients do not properly subsume polynomial interpretations with integer coefficients . the construction presented there was based on several assumptions , the essential ones of which are : 1 . [ ass1 ] the symbol @xmath220 had to be interpreted by a linear polynomial of the shape @xmath248 . [ ass2 ] the condition @xmath241 was required to hold . [ ass3 ] the function symbol @xmath211 had to be interpreted by a quadratic polynomial . now the point is that one can get rid of all these assumptions by adding suitable rewrite rules to the trs @xmath86 . the resulting trs will be referred to as @xmath20 , and it consists of the rewrite rules given in table [ tab : trs_r1 ] . the rewrite rules and serve the purpose of ensuring the first of the above items . informally , constrains the interpretation of the symbol @xmath220 to a linear polynomial by simple reasoning about the degrees of the left- and right - hand side polynomials , and does the same thing with respect to @xmath249 . because both interpretations are linear , compatibility with can only be achieved if the leading coefficient of the interpretation of @xmath220 is one . concerning item above , we remark that the tricky part is to enforce the upper bound of two on the degree of the polynomial @xmath240 that interprets the symbol @xmath211 . to this end , we make the following observation . if @xmath240 is at most quadratic , then the function @xmath250 is at most linear ; i.e. , there is a linear function @xmath251 such that @xmath252 , or equivalently , @xmath253 , for all values of @xmath30 . this can be encoded in terms of rule as soon as the interpretation of @xmath254 corresponds to addition of two numbers . and this is exactly the purpose of rules , and . more precisely , by linearity of the interpretation of @xmath249 , we infer from that the interpretation of @xmath254 must have the linear shape @xmath255 . furthermore , compatibility with and implies @xmath256 due to item above . hence , the interpretation of @xmath254 is @xmath257 , and it really models addition of two numbers ( modulo adding a constant ) . next we comment on how to enforce the second of the above assumptions . to this end , we remark that the hard part is to enforce the condition @xmath258 . the idea is as follows . first , we consider rule , observing that if @xmath211 is interpreted by a quadratic polynomial @xmath240 and @xmath220 by the linear polynomial @xmath248 , then ( the interpretation of ) its right - hand side will eventually become larger than its left - hand side with growing @xmath242 , thus violating compatibility . in this way , @xmath242 is bounded from above , and the faster the growth of @xmath240 , the lower the bound . the problem with this statement , however , is that it is only true if @xmath240 is fixed ( which is a priori not the case ) ; otherwise , for any given value of @xmath242 , one can always find a quadratic polynomial @xmath240 such that compatibility with is satisfied . the parabolic curve associated with @xmath240 only has to be flat enough . so in order to prevent this , we have to somehow control the growth of @xmath240 . now that is where rule comes into play , which basically expresses that if one increases the argument of @xmath240 by a certain amount ( i.e. , @xmath259 ) , then the value of the function is guaranteed to increase by a certain minimum amount , too . thus , this rule establishes a lower bound on the growth of @xmath240 . and it turns out that if @xmath240 has just the right amount of growth , then we can readily establish the desired upper bound @xmath12 for @xmath242 . finally , having presented all the relevant details of our construction , it remains to formally prove our main claim that the trs @xmath20 is polynomially terminating over @xmath15 but not over @xmath9 or @xmath14 . the trs @xmath20 is polynomially terminating over @xmath15 . we consider the following interpretation : @xmath260 note that the polynomial @xmath244 is a permissible interpretation function as it is both non - negative and strictly monotone over the natural numbers by lemma [ lem : nquadmon ] ( cf . figure [ fig : nvsrpoly ] ) . the rewrite rules of @xmath20 are compatible with this interpretation because the resulting inequalities 2 1 & > _ 0 & 32x^2 + 60x + 28 & > _ 32x^2 - 16x + 20 + 2 & > _ 1 & 4x + 8 & > _ 4x + 6 + 7 & > _ 6 & 4x+4 & > _ 2x + 1 & > _ 0 & x+1 & > _ x + 6 & > _ 5 & x+1 & > _ x + 2x^2 + 7x + 6 & > _ 2x^2 + 7x + 4 & 2x^2 + 3x + 4 & > _ 2x^2 + 3x + 1 are clearly satisfied for all natural numbers @xmath30 . [ lem : mainlemma1 ] the trs @xmath20 is not polynomially terminating over @xmath9 . let us assume that @xmath20 is polynomially terminating over @xmath9 and derive a contradiction . compatibility with rule implies @xmath261 as a consequence , @xmath262 , and because @xmath263 and @xmath264 must be strictly monotone , we conclude @xmath265 . the same reasoning applied to rule yields @xmath266 . hence , the symbols @xmath220 and @xmath249 must be interpreted by linear polynomials . so @xmath267 and @xmath268 with @xmath269 and , due to lemma [ lem : rlinmon ] , @xmath270 and @xmath271 . then the compatibility constraint imposed by rule gives rise to the inequality @xmath272 which must hold for all non - negative real numbers @xmath30 . this implies the following condition on the respective leading coefficients : @xmath273 . because of @xmath270 and @xmath271 , this can only hold if @xmath274 . hence , @xmath275 . this result simplifies to @xmath276 , which implies @xmath277 . from this , we conclude that @xmath278 and @xmath279 . now suppose that the function symbol @xmath211 were also interpreted by a linear polynomial @xmath240 . then we could apply the same reasoning to rule because it is structurally equivalent to , thus inferring @xmath280 . however , this would contradict @xmath279 ; therefore , @xmath240 can not be linear . next we turn our attention to the rewrite rules , and . because @xmath264 is linear , compatibility with constrains the function @xmath281 to be at most linear . this can only be the case if @xmath282 contains no terms of degree two or higher . in other words , @xmath283 , where @xmath284 , @xmath285 and @xmath286 ( cf . lemma [ lem : rlinmon ] ) . because of @xmath275 , compatibility with implies @xmath287 , and compatibility with implies @xmath288 ; thus , @xmath289 . using the obtained information in the compatibility constraint associated with rule , we get @xmath290 this implies that @xmath291 , which simplifies to @xmath292 because @xmath293 . consequently , @xmath240 must be a quadratic polynomial . without loss of generality , let @xmath294 , subject to the constraints : @xmath295 and @xmath296 because of non - negativity ( for all @xmath297 ) , and @xmath298 because @xmath299 due to strict monotonicity of @xmath240 . next we consider the compatibility constraint associated with rule , from which we deduce an important auxiliary result . after unraveling the definitions of @xmath130 and the interpretation functions , this constraint simplifies to @xmath300 which implies the following condition on the respective leading coefficients : @xmath301 ; from this and @xmath279 , we conclude @xmath302 and note that @xmath303 . hence , @xmath304 expresses the change of the slopes of the tangents to @xmath240 at the points @xmath305 and @xmath306 , and thus actually sets a lower bound on the growth of @xmath240 . now let us consider the combined compatibility constraint imposed by the rules and , namely @xmath307 , which implies @xmath308 by definition of @xmath130 . thus , we conclude @xmath309 . in fact , we even have @xmath241 , which can be derived from the compatibility constraint of rule using the conditions @xmath309 , @xmath298 and @xmath310 , the combination of the former two conditions : @xmath311 hence , @xmath312 , or equivalently , @xmath313 . but because of and @xmath309 , this inequality can only be satisfied if : @xmath314 this result has immediate consequences concerning the interpretation of the constant @xmath221 . to this end , we consider the compatibility constraint of rule , which simplifies to @xmath315 . because of and the fact that @xmath316 and @xmath317 must be non - negative , we conclude @xmath318 . moreover , condition is the key to the proof of this lemma . to this end , we consider the compatibility constraints associated with the five rewrite rules : 2 s_0 & > _ _ 0,s_0 _ ( 0 ) + 2 s_0 & > _ _ 0,s_0 _ ( s_0 ) & _ ( s_0 ) & > _ _ 0,s_0 0 + 7 s_0 & > _ _ 0,s_0 _ ( 2 s_0 ) & _ ( 2 s_0 ) & > _ _ 0,s_0 5 s_0 by definition of @xmath319 , these inequalities give rise to the following system of equations : 3 _ ( 0 ) & = 0 & _ ( s_0 ) & = s_0 & _ ( 2 s_0 ) & = 6 s_0 after unraveling the definition of @xmath240 and substituting @xmath320 , we get a system of linear equations in the unknowns @xmath321 , @xmath205 and @xmath45 3 c & = 0 & z + b & = 1 & 4z + 2b & = 6 which has the unique solution @xmath322 , @xmath323 and @xmath324 . hence , @xmath240 must have the shape @xmath325 in every compatible polynomial interpretation over @xmath9 . however , this function is not a permissible interpretation for the function symbol @xmath211 because it is not non - negative for all @xmath297 . in particular , it is negative in the open interval @xmath326 ; e.g. @xmath327 . hence , @xmath20 is not compatible with any polynomial interpretation over @xmath9 . in this proof the interpretation of @xmath211 is fixed to @xmath328 , which violates well - definedness in @xmath112 . however , this function is obviously well - defined in @xmath110 for a properly chosen negative real number @xmath329 . so what happens if we take this @xmath110 instead of @xmath112 as the carrier of a polynomial interpretation ? to this end , we observe that @xmath330 and @xmath331 . now let us consider some negative real number @xmath332 . we have @xmath333 and thus @xmath334 , which means that @xmath240 violates monotonicity with respect to the order @xmath335 . the previous lemma , together with theorem [ thm : qvsr ] , yields the following corollary . the trs @xmath20 is not polynomially terminating over @xmath14 . finally , combining the material presented in this section , we establish the following theorem , the main result of this section . [ thm : nvsr ] there are trss that can be proved polynomially terminating over @xmath15 , but can not be proved polynomially terminating over @xmath9 or @xmath14 . we conclude this section with a remark on the actual choice of the polynomial serving as the interpretation of the function symbol @xmath211 . as explained at the beginning of this section , the trs @xmath20 was designed to enforce an interpretation for @xmath211 , which is permissible in a polynomial interpretation over @xmath15 but not over @xmath9 ( @xmath14 ) . the interpretation of our choice was the polynomial @xmath244 . however , we could have chosen any other polynomial as long as it is well - defined and strictly monotone over @xmath15 but not over @xmath9 ( @xmath14 ) . the methods introduced in this section are general enough to handle any such polynomial . so the actual choice is not that important . this section is devoted to showing that polynomial termination over @xmath15 and @xmath9 does not imply polynomial termination over @xmath14 . the proof is constructive , so we give a concrete trs having the desired properties . in order to motivate the construction underlying this particular system , let us consider the following quantified polynomial inequality @xmath336 where @xmath337 $ ] is a polynomial with integer coefficients , all of whose roots are irrational and which is positive for some non - negative integer value of @xmath213 . to be concrete , let us take @xmath338 and try to satisfy in @xmath15 , @xmath113 and @xmath112 , respectively . first , we observe that @xmath339 is a satisfying assignment in @xmath112 . moreover , is also satisfiable in @xmath15 by assigning @xmath340 , for example , and observing that the polynomial @xmath341 is non - negative for all @xmath342 . however , can not be satisfied in @xmath113 as non - negativity of @xmath341 does not hold for all @xmath343 and @xmath59 has no rational roots . to sum up , is satisfiable in @xmath15 and @xmath112 but not in @xmath113 . thus , the basic idea now is to design a trs containing some rewrite rule whose compatibility constraint reduces to a polynomial inequality similar in nature to . to this end , we rewrite the inequality @xmath344 to @xmath345 because now both the left- and right - hand side can be viewed as a composition of several functions , each of which is strictly monotone and well - defined . in particular , we identify the following constituents : @xmath346 , @xmath347 , @xmath348 and @xmath349 . thus , the above inequality can be written in the form @xmath350 which can easily be modeled as a rewrite rule . ( note that @xmath351 is not strictly necessary as @xmath352 , but it gives rise to a shorter encoding . ) and then we also need rewrite rules that enforce the desired interpretations for the function symbols @xmath254 , @xmath353 , @xmath354 and @xmath355 . for this purpose , we leverage the techniques presented in the previous section , in particular the method of polynomial interpolation . the resulting trs will be referred to as @xmath21 , and it consists of the rewrite rules given in table [ tab : trs_r2 ] . .the trs @xmath21 . [ cols="^,^ " , ] as in section [ sect : nrvsq ] , each block serves a specific purpose . the one made up of enforces the desired semantics for the symbol @xmath353 , that is , a linear function @xmath356 that doubles its input , while the block enforces a linear function @xmath357 for the symbol @xmath358 whose slope @xmath359 is proportional to the square of the interpretation of the constant @xmath360 . finally , encodes the main idea of the construction , as mentioned above . [ lem : trs_r_5_sn ] the trs @xmath361 is incrementally polynomially terminating over @xmath15 and @xmath9 . for incremental polynomial termination over @xmath15 , we start with the interpretation @xmath362 all interpretation functions are well - defined over @xmath15 and strictly monotone ( i.e. , monotone with respect to @xmath95 ) as well as weakly monotone ( i.e. , monotone with respect to @xmath99 ) . moreover , it is easy to verify that this interpretation is weakly compatible with @xmath361 . in particular , the rule gives rise to the constraint @xmath363 which holds for all @xmath342 . after removing the rules from @xmath361 for which ( strict ) compatibility holds , we are left with the rules , , , and , all of which can be handled ( that is , removed at once ) by the following linear interpretation : @xmath364 for incremental polynomial termination over @xmath9 , we consider the interpretation @xmath365 which is both weakly and strictly monotone according to lemmata [ lem : qrquadmon ] and [ lem : qrquadwmon ] . so all interpretation functions are well - defined over @xmath112 and monotone with respect to @xmath130 and @xmath366 . moreover , one easily verifies that this interpretation is weakly compatible with @xmath361 . in particular , the constraint @xmath367 associated with trivially holds . after removing the rules from @xmath361 for which ( strict ) compatibility holds ( i.e. , for which the corresponding constraint remains true after strengthening @xmath366 to @xmath130 ) , we are left with , , , and , all of which can be removed at once by the following linear interpretation : @xmath368 [ lem : trs_r_5 ] the trs @xmath361 is not incrementally polynomially terminating over @xmath14 . let @xmath369 be a strictly monotone polynomial interpretation over @xmath14 that is weakly compatible with @xmath361 . according to lemma [ lem : trs_s ] , the symbols @xmath221 , @xmath220 , @xmath211 , @xmath249 , @xmath254 and @xmath355 are interpreted as follows : @xmath370 where @xmath371 and @xmath372 . as the interpretation of @xmath355 is linear , weak compatibility with the rule implies that the interpretation of @xmath353 is at most linear as well , i.e. , @xmath373 with @xmath374 and @xmath375 . we also have @xmath376 due to weak compatibility with and @xmath377 due to weak compatibility with ; hence , @xmath378 . similarly , by linearity of @xmath379 and weak compatibility with , the interpretation of @xmath358 must have the shape @xmath380 . then weak compatibility with yields @xmath381 ; hence , @xmath382 , @xmath383 . next we note that weak compatibility with and implies that @xmath384 , which evaluates to @xmath385 . from this we infer that @xmath386 as @xmath104 and @xmath387 due to weak compatibility with ; i.e. , @xmath388 . finally , we consider the weak compatibility constraint associated with , which simplifies to @xmath389 however , this inequality is unsatisfiable as the polynomial @xmath390 is negative for some @xmath343 and @xmath391 can not be zero because both @xmath213 and @xmath392 must be rational numbers . so we conclude that there is no strictly monotone polynomial interpretation over @xmath14 that is weakly compatible with the trs @xmath361 . this implies that @xmath361 is not incrementally polynomially terminating over @xmath14 . together , lemma [ lem : trs_r_5_sn ] and lemma [ lem : trs_r_5 ] yield the main result of this subsection . [ cor : nrvsq_inc ] there are trss that are incrementally polynomially terminating over @xmath15 and @xmath9 but not over @xmath14 . in this subsection , we show that there are trss that are incrementally polynomially terminating over @xmath15 but not over @xmath9 . for this purpose , we extend the trs @xmath393 of table [ tab : trs_s ] by the single rewrite rule @xmath394 and call the resulting system @xmath395 . [ lem : trs_r_6_sn ] the trs @xmath395 is incrementally polynomially terminating over @xmath15 . first , we consider the interpretation @xmath396 which is both weakly and strictly monotone as well as weakly compatible with @xmath395 . in particular , the constraint @xmath397 associated with @xmath398 holds for all @xmath342 . removing the rules from @xmath395 for which ( strict ) compatibility holds leaves us with the rules , and @xmath398 , which are easily handled , e.g. by the linear interpretation @xmath399 [ lem : trs_r_6 ] the trs @xmath395 is not incrementally polynomially terminating over @xmath9 or @xmath14 . let @xmath119 , and let @xmath369 be a polynomial interpretation over @xmath129 that is weakly compatible with @xmath395 , and in which the interpretation of the function symbol @xmath211 has the shape @xmath400 with @xmath104 . then the weak compatibility constraint @xmath401 associated with @xmath398 does not hold for all @xmath402 because the polynomial @xmath403 is negative in the open interval @xmath404 . as the above assumption on the interpretation of @xmath211 follows from lemma [ lem : trs_s ] if @xmath369 is strictly monotone , we conclude that there is no strictly monotone polynomial interpretation over @xmath9 or @xmath14 that is weakly compatible with the trs @xmath395 . this implies that @xmath395 is not incrementally polynomially terminating over @xmath9 or @xmath14 . together , lemma [ lem : trs_r_6_sn ] and lemma [ lem : trs_r_6 ] yield the main result of this subsection . [ cor : nvsr_inc ] there are trss that are incrementally polynomially terminating over @xmath15 but not over @xmath9 or @xmath14 . the results presented in this section can be summarized by stating that the relationships expressed in figure [ fig : summary ] remain true for incremental polynomial termination , after replacing @xmath20 by @xmath395 and @xmath21 by @xmath361 . in this article , we investigated the relationship of polynomial interpretations with real , rational and integer coefficients with respect to termination proving power . in particular , we presented three new results , the first of which shows that polynomial interpretations over the reals subsume polynomial interpretations over the rationals , the second of which shows that polynomial interpretations over the reals or rationals do not properly subsume polynomial interpretations over the integers , a result that comes somewhat unexpected , and the third of which shows that there are trss that can be proved terminating by polynomial interpretations over the naturals or the reals but not over the rationals . these results were extended to incremental termination proofs . in @xcite it is shown how to adapt the results to the dependency pair framework @xcite . we conclude this article by reviewing our results in the context of automated termination analysis , where linear polynomial interpretations , i.e. , polynomial interpretations with all interpretation functions being linear , play an important role . this naturally raises the question as to what extent the restriction to linear polynomial interpretations influences the hierarchy depicted in figure [ fig : summary ] , and in what follows we shall see that it changes considerably . more precisely , the areas inhabited by the trss @xmath20 and @xmath21 become empty , such that polynomial termination by a linear polynomial interpretation over @xmath15 implies polynomial termination by a linear polynomial interpretation over @xmath14 , which in turn implies polynomial termination by a linear polynomial interpretation over @xmath9 . the latter follows directly from theorem [ thm : qvsr ] and remark [ rem : qvsr ] , whereas the former is shown below . polynomial termination by a linear polynomial interpretation over @xmath15 implies polynomial termination by a linear polynomial interpretation over @xmath14 . let @xmath86 be a trs that is compatible with a linear polynomial interpretation @xmath164 over @xmath15 , where every @xmath0-ary function symbol @xmath211 is associated with a linear polynomial @xmath405 . we show that the same interpretation also establishes polynomial termination over @xmath14 with the value of @xmath12 set to one . to this end , we note that in order to guarantee strict monotonicity and well - definedness over @xmath15 , the coefficients of the respective interpretation functions have to satisfy the following conditions : @xmath406 and @xmath407 for all @xmath408 . hence , by lemma [ lem : qlinmon ] , we also have well - definedness over @xmath113 and strict monotonicity with respect to the order @xmath409 . ( strict monotonicity also follows from ( * ? ? ? * theorem 2 ) . ) moreover , as @xmath86 is compatible with @xmath164 , each rewrite rule @xmath89 satisfies @xmath410 where @xmath6 ( @xmath7 ) denotes the polynomial associated with @xmath171 ( @xmath172 ) and the variables @xmath173 are those occurring in @xmath5 . since linear functions are closed under composition , the polynomial @xmath411 is a linear polynomial @xmath412 , such that holds if and only if @xmath413 and @xmath414 for all @xmath415 . however , then we also have @xmath416 which shows that @xmath86 is compatible with the linear polynomial interpretation @xmath417 over @xmath14 . hence , linear polynomial interpretations over @xmath9 subsume linear polynomial interpretations over @xmath14 , which in turn subsume linear polynomial interpretations over @xmath15 , and these subsumptions are proper due to the results of @xcite , which were obtained using linear polynomial interpretations . we thank harald zankl for finding the incremental polynomial interpretation given in the proof of lemma [ lem : r1_ipt_q ] . the comments by the reviewers improved the presentation and helped to clarify the contributions of salvador lucas @xcite . d. hofbauer . termination proofs by context - dependent interpretations . in _ proc . 12th international conference on rewriting techniques and applications ( rta 2001 ) _ , volume 2051 of _ lecture notes in computer science _ , pages 108121 , 2001 . http://dx.doi.org/10.1007/3-540-45127-7_10[doi : ] . s. lucas . on the relative power of polynomials with real , rational , and integer coefficients in proofs of termination of rewriting . , 17(1):4973 , 2006 . http://dx.doi.org/10.1007/s00200-005-0189-5[doi : ] . s. lucas . practical use of polynomials over the reals in proofs of termination . in _ proc . 9th international conference on principles and practice of declarative programming ( ppdp 2007 ) _ , pages 3950 . association of the computing machinery , 2007 . http://dx.doi.org/10.1145/1273920.1273927[doi : ] . f. neurauter and a. middeldorp . polynomial interpretations over the reals do not subsume polynomial interpretations over the integers . in _ proc . 21st international conference on rewriting techniques and applications ( rta 2010 ) _ , volume 6 of _ leibniz international proceedings in informatics _ , pages 243258 , 2010 . http://dx.doi.org/10.4230/lipics.rta.2010.243[doi : ] .

a posteriori estimates for mixed finite element discretizations of the navier - stokes equations are derived . we show that the task of estimating the error in the evolutionary navier - stokes equations can be reduced to the estimation of the error in a steady stokes problem . as a consequence , any available procedure to estimate the error in a stokes problem can be used to estimate the error in the nonlinear evolutionary problem . a practical procedure to estimate the error based on the so - called postprocessed approximation is also considered . both the semidiscrete ( in space ) and the fully discrete cases are analyzed . some numerical experiments are provided .

1011.2878

we consider the incompressible navier stokes equations @xmath0 in a bounded domain @xmath1 ( @xmath2 ) with a smooth boundary subject to homogeneous dirichlet boundary conditions @xmath3 on @xmath4 . in ( [ onetwo ] ) , @xmath5 is the velocity field , @xmath6 the pressure , and @xmath7 a given force field . for simplicity in the exposition we assume , as in @xcite , @xcite , @xcite , @xcite , @xcite , that the fluid density and viscosity have been normalized by an adequate change of scale in space and time . let @xmath8 and @xmath9 be the semi - discrete ( in space ) mixed finite element ( mfe ) approximations to the velocity @xmath5 and pressure @xmath6 , respectively , solution of ( [ onetwo ] ) corresponding to a given initial condition @xmath10 we study the a posteriori error estimation of these approximations in the @xmath11 and @xmath12 norm for the velocity and in the @xmath13 norm for the pressure . to do this for a given time @xmath14 , we consider the solution ( @xmath15 , @xmath16 ) of the stokes problem @xmath17 we prove that @xmath15 and @xmath16 are approximations to @xmath5 and @xmath6 whose errors decay by a factor of @xmath18 faster than those of @xmath8 and @xmath9 ( @xmath19 being the mesh size ) . as a consequence , the quantities @xmath20 and @xmath21 , are asymptotically exact indicators of the errors @xmath22 and @xmath23 in the navier - stokes problem ( [ onetwo])([ic ] ) . furthermore , the key observation in the present paper is that ( @xmath24 ) is also the mfe approximation to the solution @xmath25 of the stokes problem ( [ eq : stokes ] ) . consequently , any available procedure to a posteriori estimate the errors in a stokes problem can be used to estimate the errors @xmath26 and @xmath21 which , as mentioned above , coincide asymptotically with the errors @xmath22 and @xmath23 in the evolutionary ns equations . many references address the question of estimating the error in a stokes problem , see for example @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite and the references therein . in this paper we prove that any efficient or asymptotically exact estimator of the error in the mfe approximation @xmath27 to the solution of the _ steady _ stokes problem ( [ eq : stokes ] ) is also an efficient or asymptotically exact estimator , respectively , of the error in the mfe approximation @xmath27 to the solution of the _ evolutionary _ navier - stokes equations ( [ onetwo])([ic ] ) . for the analysis in the present paper we do not assume to have more than second - order spatial derivatives bounded in @xmath28 up to initial time @xmath29 , since demanding further regularity requires the data to satisfy nonlocal compatibility conditions unlikely to be fulfilled in practical situations @xcite , @xcite . the analysis of the errors @xmath30 and @xmath31 follows closely @xcite where mfe approximations to the stokes problem ( [ eq : stokes ] ) ( the so - called postprocessed approximations ) are considered with the aim of getting improved approximations to the solution of ( [ onetwo])([ic ] ) at any fixed time @xmath32 . in this paper we will also refer to ( @xmath15 , @xmath16 ) as postprocessed approximations although they are of course not computable in practice and they are only considered for the analysis of a posteriori error estimators . the postprocessed approximations to the navier - stokes equations were first developed for spectral methods in @xcite , @xcite , @xcite , @xcite and also developed for mfe methods for the navier - stokes equations in @xcite , @xcite , @xcite . for the sake of completeness , in the present paper we also analyze the use of the computable postprocessed approximations of @xcite for a posteriori error estimation . the use of this kind of postprocessing technique to get a posteriori error estimations has been studied in @xcite , @xcite and @xcite for nonlinear parabolic equations excluding the navier - stokes equations . we refer also to @xcite where the so - called stokes reconstruction is used to a posteriori estimate the errors of the semi - discrete in space approximations to a linear time - dependent stokes problem . we remark that the stokes reconstruction of @xcite is exactly the postprocessing approximation ( @xmath33 ) in the particular case of a linear model . in the second part of the paper we consider a posteriori error estimations for the fully discrete mfe approximations @xmath34 and @xmath35 , ( @xmath36 for @xmath37 ) obtained by integrating in time with either the backward euler method or the two - step backward differentiation formula ( bdf ) . for this purpose , we define a stokes problem similar to ( [ eq : stokes ] ) but with the right - hand - side depending now on the fully discrete mfe approximation @xmath38 ( problem ( [ posth0n])([posth1n ] ) in section [ sec:4 ] below ) . we will call time - discrete postprocessed approximation to the solution @xmath39 of this new stokes problem . as before , @xmath40 is not computable in practice and it is only considered for the analysis of a posteriori error estimation . observe that in the fully discrete case ( which is the case in actual computations ) the task of estimating the the error @xmath41 of the mfe approximation becomes more difficult due to the presence of time discretization errors @xmath42 , which are added to the spatial discretization errors @xmath43 . however we show in section [ sec:4 ] that if temporal and spatial errors are not very different in size , the quantity @xmath44 correctly esimates the spatial error because the leading terms of the temporal errors in @xmath45 and @xmath38 get canceled out when subtracting @xmath44 , leaving only the spatial component of the error . this is a very convenient property that allows to use independent procedures for the tasks of estimating the errors of the spatial and temporal discretizations . we remark that the temporal error can be routinely controlled by resorting to well - known ordinary differential equations techniques . analogous results were obtained in @xcite for fully discrete finite element approximations to evolutionary convection - reaction - diffusion equations using the backward euler method . as in the semidiscrete case , a key point in our results is again the fact that the fully discrete mfe approximation @xmath46 to the navier - stokes problem ( [ onetwo])([ic ] ) is also the mfe approximation to the solution @xmath47 of the stokes problem ( [ posth0n])([posth1n ] ) . as a consequence , we can use again any available error estimator for the stokes problem to estimate the spatial error of the fully discrete mfe approximations @xmath46 to the navier - stokes problem ( [ onetwo])([ic ] ) . computable mixed finite element approximations to @xmath47 , the so - called fully discrete postprocessed approximations , were studied and analyzed in @xcite where we proved that the fully discrete postprocessed approximations maintain the increased spatial accuracy of the semi - discrete approximations . the analysis in the second part of the present paper borrows in part from @xmath48 . also , we propose a computable error estimator based on the fully discrete postprocessed approximation of @xcite and show that it also has the excellent property of separating spatial and temporal errors . the rest of the paper is as follows . in section 2 we introduce some preliminaries and notation . in section 3 we study the a posteriori error estimation of semi - discrete in space mfe approximations . in section 4 we study a posteriori error estimates for fully discrete approximations . finally , some numerical experiments are shown in section 5 . we will assume that @xmath49 is a bounded domain in @xmath50 , of class @xmath51 , for @xmath52 . when dealing with linear elements ( @xmath53 below ) @xmath49 may also be a convex polygonal or polyhedral domain . we consider the hilbert spaces @xmath54 endowed with the inner product of @xmath55 and @xmath56 , respectively . for @xmath57 integer and @xmath58 , we consider the standard spaces , @xmath59 , of functions with derivatives up to order @xmath60 in @xmath61 , and @xmath62 . we will denote by @xmath63 the norm in @xmath64 , and @xmath65 will represent the norm of its dual space . we consider also the quotient spaces @xmath66 with norm @xmath67 . we recall the following sobolev s imbeddings @xcite : for @xmath68 , there exists a constant @xmath69 such that @xmath70 for @xmath71 , ( [ sob1 ] ) holds with @xmath72 . the following inf - sup condition is satisfied ( see @xcite ) , there exists a constant @xmath73 such that @xmath74 where , here and in the sequel , @xmath75 denotes the standard inner product in @xmath76 or in @xmath28 . let @xmath77 be the @xmath78 projector onto @xmath79 . we denote by @xmath80 the stokes operator on @xmath81 : @xmath82 applying leray s projector @xmath83 to ( [ onetwo ] ) , the equations can be written in the form @xmath84 where @xmath85 for @xmath5 , @xmath86 in @xmath87 . we shall use the trilinear form @xmath88 defined by @xmath89 it is straightforward to verify that @xmath90 enjoys skew - symmetry : @xmath91 let us observe that @xmath92 for @xmath93 @xmath94 . let us consider for @xmath95 and @xmath96 the operators @xmath97 and @xmath98 , which are defined by means of the spectral properties of @xmath80 ( see , e.g. , @xcite , @xcite ) . notice that @xmath80 is a positive self - adjoint operator with compact resolvent in @xmath79 . an easy calculation shows that @xmath99 where , here and in what follows , @xmath100 when applied to an operator denotes the associated operator norm . we shall assume that the solution @xmath5 of ( [ onetwo])-([ic ] ) satisfies @xmath101 for some constants @xmath102 and @xmath103 . we shall also assume that there exists a constant @xmath104 such that @xmath105 finally , we shall assume that for some @xmath106 @xmath107 so that , according to theorems 2.4 and 2.5 in @xcite , there exist positive constants @xmath108 and @xmath109 such that the following bounds hold : @xmath110 where @xmath111 and @xmath112 for some @xmath113 . observe that for @xmath114 , we can take @xmath115 and @xmath116 . for simplicity , we will take these values of @xmath117 and @xmath118 . let @xmath119 , @xmath120 be a family of partitions of suitable domains @xmath121 , where @xmath19 is the maximum diameter of the elements @xmath122 , and @xmath123 are the mappings of the reference simplex @xmath124 onto @xmath125 . let @xmath126 , we consider the finite - element spaces @xmath127 where @xmath128 denotes the space of polynomials of degree at most @xmath129 on @xmath130 . as it is customary in the analysis of finite - element methods for the navier - stokes equations ( see e. g. , @xcite , @xcite , @xcite , @xcite , @xcite ) we restrict ourselves to quasiuniform and regular meshes @xmath131 , so that as a consequence of ( * ? ? ? * theorem 3.2.6 ) , the following inverse inequality holds for each @xmath132 @xmath133 where @xmath134 , @xmath135 . we shall denote by @xmath136 the so - called hood taylor element @xcite , when @xmath137 , where @xmath138 and the so - called mini - element @xcite when @xmath53 , where @xmath139 , and @xmath140 . here , @xmath141 is spanned by the bubble functions @xmath142 , @xmath143 , defined by @xmath144 , if @xmath145 and 0 elsewhere , where @xmath146 denote the barycentric coordinates of @xmath147 . for these elements a uniform inf - sup condition is satisfied ( see @xcite ) , that is , there exists a constant @xmath73 independent of the mesh grid size @xmath19 such that @xmath148 we remark that our analysis can also be applied to other pairs of lbb - stable mixed finite elements ( see ( * ? ? ? * remark 2.1 ) ) . the approximate velocity belongs to the discrete divergence - free space @xmath149 which is not a subspace of @xmath150 . we shall frequently write @xmath151 instead of @xmath152 whenever the value of @xmath153 plays no particular role . let @xmath154 be the discrete leray s projection defined by @xmath155 we will use the following well - known bounds @xmath156 we will denote by @xmath157 the discrete stokes operator defined by @xmath158 let @xmath159 be the solution of a stokes problem with right - hand side @xmath160 , we will denote by @xmath161 the so - called stokes projection ( see @xcite ) defined as the velocity component of solution of the following stokes problem : find @xmath162 such that @xmath163 the following bound holds for @xmath164 : @xmath165 the proof of ( [ stokespro ] ) for @xmath166 can be found in @xcite . for the general case , @xmath121 must be such that the value of @xmath167 satisfies @xmath168 . this can be achieved if , for example , @xmath169 is piecewise of class @xmath170 , and superparametric approximation at the boundary is used @xcite . under the same conditions , the bound for the pressure is @xcite @xmath171 where the constant @xmath172 depends on the constant @xmath173 in the inf - sup condition ( [ lbbh ] ) . we will assume that the domain @xmath49 is of class @xmath51 , with @xmath174 so that standard bounds for the stokes problem @xcite , @xcite imply that @xmath175 for a domain @xmath49 of class @xmath176 we also have the bound ( see @xcite ) @xmath177 in what follows we will apply the above estimates to the particular case in which @xmath178 is the solution of the navier stokes problem ( [ onetwo])([ic ] ) . in that case @xmath179 is the discrete velocity in problem ( [ stokesnew])([stokesnew2 ] ) with @xmath180 . note that the temporal variable @xmath181 appears here merely as a parameter , and then , taking the time derivative , the error bound ( [ stokespro ] ) can also be applied to the time derivative of @xmath182 changing @xmath5 , @xmath6 by @xmath183 , @xmath184 . since we are assuming that @xmath49 is of class @xmath51 and @xmath52 , from ( [ stokespro ] ) and standard bounds for the stokes problem @xcite , we deduce that @xmath185 we consider the semi - discrete finite - element approximation @xmath27 to @xmath178 , solution of ( [ onetwo])([ic ] ) . that is , given @xmath186 , we compute @xmath187 and @xmath188 , @xmath189 $ ] , satisfying @xmath190 for @xmath191 , provided that ( [ stokespro])([stokespre ] ) hold for @xmath192 , and ( [ eq : u - inf])([eq : u - int ] ) hold for @xmath193 , then we have @xmath194 ( see , e.g. , @xcite ) , and also , @xmath195 where @xmath196 if @xmath197 and @xmath198 if @xmath199 . see ( * ? ? ? * proposition 3.2 ) . let us consider the mfe approximation @xmath27 at any time @xmath200 $ ] to @xmath201 obtained by solving ( [ ten])([ten2 ] ) . we consider the postprocessed approximation @xmath202 in @xmath203 which is the solution of the following stokes problem written in weak form @xmath204 for all @xmath205 and @xmath206 . we remark that the mfe approximation @xmath207 to @xmath201 is also the mfe approximation to the solution @xmath202 of the stokes problem ( [ posth0])([posth1 ] ) . in theorems [ semi_disve ] and [ semi_dispre ] below we prove that the postprocessed approximation @xmath202 is an improved approximation to the solution @xmath178 of the evolutionary navier - stokes equations ( [ onetwo])([ic ] ) at time @xmath208 . although , as it is obvious , @xmath202 is not computable in practice , it is however a useful tool to provide a posteriori error estimates for the mfe approximation @xmath27 at any desired time @xmath14 . in theorem [ semi_disve ] we obtain the error bounds for the velocity and in theorem [ semi_dispre ] the bounds for the pressure . the improvement is achieved in the @xmath209 norm when using the mini - element ( @xmath53 ) and in both the @xmath28 and @xmath209 norms in the cases @xmath210 . in the sequel we will use that for a forcing term satisfying ( [ tildem2 ] ) there exists a constant @xmath211 , depending only on @xmath104 , @xmath212 and @xmath213 , such that the following bound hold for @xmath214 : @xmath215 the following inequalities hold for all @xmath216 and @xmath217 , see ( * ? ? ? * ( 3.7 ) ) : @xmath218 the proof of theorem [ semi_disve ] requires some previous results which we now state and prove . we will use the fact that @xmath219 for @xmath220 , from where it follows that @xmath221 where the constant @xmath222 is independent of @xmath19 . let @xmath178 be the solution of ( [ onetwo])([ic ] ) and fix @xmath113 . then there exists a positive constant @xmath223 such that for @xmath224 satisfying the threshold condition @xmath225 the following inequalities hold for @xmath226 : @xmath227 due to the equivalence ( [ a-1equiv ] ) and and since @xmath228 for @xmath229 it is sufficient to prove @xmath230 for @xmath231 or @xmath232 . we follow the proof ( * ? ? ? * lemma 3.1 ) where a different threshold assumption is assumed . we do this for @xmath231 , since the case @xmath232 is similar but yet simpler . we write @xmath233 where @xmath234 . we first observe that @xmath235 where , in the last inequality , we have used that thanks to sobolev s inequality ( [ sob1 ] ) we have @xmath236 . similarly , @xmath237 the proof of the case @xmath238 in ( [ eq : aux_bat ] ) is finished if we show that for @xmath239 , both @xmath240 and @xmath241 are bounded in terms of @xmath103 and the value @xmath242 in the threshold assumption ( [ threshold ] ) . to do this , we will use the inverse inequality ( [ inv ] ) and the fact that the stokes projection @xmath179 satisfies that @xmath243 for some constant @xmath244 ( see for example the proof of lemma 3.1 in @xcite ) . we have @xmath245 where in the last inequality we have applied ( [ inv ] ) , and , similarly , @xmath246 where we also have used that @xmath247 for @xmath248 . now the threshold assumption ( [ threshold ] ) and ( [ stokespro ] ) show the boundedness of @xmath240 and @xmath241 . finally , the proof of the case @xmath249 in ( [ eq : aux_bat ] ) is , with obvious changes , that of the equivalent result in ( * ? ? ? * lemma 3.1 ) . in the sequel we consider the auxiliary function @xmath250\rightarrow v_h$ ] solution of @xmath251 according to ( * ? ? ? * remark 4.2 ) we have @xmath252 for some constant @xmath253 . the following lemma provides a superconvergence result . [ le : sup1 ] let @xmath178 be the solution of ( [ onetwo])([ic ] ) . then , there exists a positive constant @xmath222 such that the solution @xmath254 of ( [ atenopv ] ) and the galerkin approximation @xmath8 satisfy the following bound , @xmath255.\ ] ] since for @xmath256 we have @xmath257 where @xmath258 , it follows that @xmath259 applying ( [ 4.21 - 4.22 ] ) we have @xmath260 , so that taking into account that @xmath261 it follows that @xmath262 since applying ( * ? ? ? * lemma 4.2 ) we obtain @xmath263 a generalized gronwall lemma @xcite , together with ( [ 4.21 - 4.22 ] ) allow us to conclude @xmath264 then by writing @xmath265 and applying ( [ eq : error - pi_h ] ) and ( [ eq : cota - z - low ] ) , the proof is finished if we check that the threshold condition ( [ threshold ] ) holds for @xmath266 and @xmath267 . in view of ( [ eq : cota - z - low ] ) , ( [ eq : error - pi_h ] ) and the inverse inequality ( [ inv ] ) we have indeed that @xmath268 , for @xmath226 . in the case of @xmath8 the threshold condition holds due to ( [ eq : err_vel(t ) ] ) . [ le : e_t ] let @xmath178 be the solution of ( [ onetwo])([ic ] ) . then there exists a positive constant @xmath222 such that @xmath269 , \label{e_t-1}\ ] ] where @xmath254 and @xmath270 are defined by ( [ atenopv ] ) and ( [ ten])-([ten2 ] ) respectively . the difference @xmath271 satisfies that @xmath272 , so that multiplying by @xmath273 and taking norms , thanks to ( [ 4.21 - 4.22b ] ) , we have @xmath274 now we write @xmath275 so that ( [ eq : error - pi_h ] ) , ( [ eq : cota - z - low ] ) and ( [ supercuad ] ) , allow us to write , @xmath276 then , applying ( [ a-1equiv ] ) the proof is finished . [ le : z_t ] let @xmath178 be the solution of ( [ onetwo])([ic ] ) . then there exists a positive constant @xmath222 such that @xmath277 , \quad r=2,3,4,\ ] ] where @xmath278 when @xmath53 and @xmath279 otherwise . the case @xmath210 is proved in ( * ? ? ? * lemma 5.1 ) . for the case @xmath53 we write @xmath280 a simple duality argument and the fact that @xmath281 , easily show that @xmath282 the bound of the third term on the right - hand side of ( [ demini1 ] ) is given in lemma [ le : e_t ] , so that , thanks to the equivalence ( [ a-1equiv ] ) we are left with estimating @xmath283 we notice that @xmath284 where @xmath285 . thus , @xmath286 recalling ( [ eq : vabien ] ) by means of the change of variables @xmath287 it is easy to show that @xmath288 where @xmath289 is the beta function ( see e. g. , @xcite ) . thus , we have @xmath290 the first term on the right - hand side above is bounded by @xmath291 . for the second one we notice hat @xmath292 so that using ( [ eq : error - pi_h ] ) , ( [ stokespro ] ) and ( [ eq : cota - z - low ] ) it is bounded by @xmath293 . [ semi_disve ] let @xmath178 be the solution of ( [ onetwo])-([ic ] ) . then , there exists a positive constant @xmath222 such that the postprocessed velocity @xmath15 , defined in ( [ posth0])-([posth1 ] ) , satisfies the following bounds : \(i ) if @xmath53 then @xmath294 \(ii ) if @xmath210 then @xmath295 the proof follows the same steps as ( * ? ? ? * theorem 5.2 ) . subtracting ( [ posth0 ] ) from ( [ onetwo ] ) , standard duality arguments show that @xmath296 to bound the second term on the right - hand side above we apply lemma [ le : z_t ] , whereas for the second we apply ( [ eq : aux_bat ] ) to get @xmath297 so that applying ( [ eq : err_vel(t ) ] ) the proof of ( [ utilde_lineal ] ) and the case @xmath249 of ( [ utilde_mas ] ) are finished . we now get the error bounds in the @xmath11 norm . it is easy to see that @xmath298 then , by applying @xmath299 to both sides of the above equations , we obtain @xmath300 as regards the nonlinear term , applying ( * ? ? ? * lemma 4.1 ) we obtain @xmath301 to bound the second term on the right - hand side above we apply ( [ eq : err_vel(t ) ] ) , whereas the first one is bounded in the proof of ( * ? ? ? * theorem 5.2 ) by @xmath302 finally , to bound @xmath303 we apply ( * ? ? ? * lemma 5.1 ) to obtain @xmath304 which concludes the proof . in the following theorem we obtain the error bounds for the pressure @xmath16 . [ semi_dispre ] let @xmath178 be the solution of ( [ onetwo])-([ic ] ) . then , there exists a positive constant @xmath222 such that the postprocessed pressure , @xmath16 , satisfies the following bounds : @xmath305 where @xmath278 if @xmath53 and @xmath279 if @xmath210 . the proof follows the same steps as ( * ? ? ? * theorem 5.3 ) . applying the inf - sup condition ( [ inf - sup ] ) it is easy to see that @xmath306 applying now ( [ utilde_lineal ] ) and ( [ utilde_mas ] ) to bound the first term and reasoning as in the proof of theorem [ semi_disve ] to bound the other two terms we conclude ( [ ptilde_mas ] ) . [ remark31 ] as a consequence of theorems [ semi_disve ] and [ semi_dispre ] we obtain in the proof of theorem [ apos_semi1 ] that @xmath307 is an asymptotically exact estimator of the error @xmath308 while @xmath309 is an asymptotically exact estimator of the error @xmath310 . however , as we have already observed @xmath15 and @xmath16 are not computable in practice . in theorems [ apos_semi1 ] , 4 and [ th_pos_esti ] we present different procedures to get computable error estimators . as we pointed out before the mfe approximations @xmath27 to the velocity and the pressure of the solution @xmath178 of the evolutionary navier - stokes equations ( [ onetwo])-([ic ] ) at any fixed time @xmath208 are also the approximations to the velocity and pressure of the steady stokes problem ( [ posth0])-([posth1 ] ) . in theorem [ apos_semi1 ] we show that any a posteriori error estimator of the error in the steady stokes problem ( [ posth0])-([posth1 ] ) gives us an a posteriori indicator of the error in the approximations to the evolutionary navier - stokes equations . using the notation of @xcite we will denote in the sequel by @xmath311 , @xmath226 , any a posteriori error estimator of the error @xmath312 in the norm of @xmath313 in the approximation to the velocity in the steady stokes problem ( [ posth0])-([posth1 ] ) . we will denote by @xmath314 any error estimator of the quantity @xmath315 . [ apos_semi1 ] let @xmath178 be the solution of ( [ onetwo])-([ic ] ) and fix any positive time @xmath14 . assume that the galerkin approximation @xmath27 satisfies , for @xmath19 small enough and @xmath316 , @xmath317 for some positive constant @xmath318 . \(i ) if @xmath319 , @xmath226 , is an efficient error indicator of the error in the mfe approximation to the steady stokes problem ( [ posth0])-([posth1 ] ) . that is , if there exist positive constants , @xmath320 and @xmath321 , that are independent of the mesh size @xmath19 , such that the following bound holds @xmath322 then @xmath319 , @xmath226 , it is also an efficient error indicator of the error in the mfe approximation to the evolutionary navier - stokes equations , i.e. there exist positive constants @xmath323 and @xmath324 that are independent of the mesh size @xmath19 such that the following bound holds for @xmath19 small enough @xmath325 \(ii ) if @xmath319 , @xmath226 is an asymptotically exact error estimator of the error in the steady stokes problem then it is also an asymptotically exact error estimator of the error in the evolutionary navier - stokes equations . \(iii ) analogous results are obtained in the approximations to the pressure . in the case @xmath53 , the results are valid only in the @xmath12 norm . for simplicity in the exposition we will concentrate on the cases @xmath210 in the approximations to the velocity , the proof for the approximations to the pressure and for the case @xmath53 being the same except for obvious changes . let us first observe that @xmath326 dividing by @xmath327 , using ( [ saturacion ] ) and applying theorem [ semi_disve ] we obtain @xmath328 now , using ( [ efficient_steady ] ) we get @xmath329 taking @xmath19 small enough so that @xmath330 we get @xmath331 now , we use the decomposition @xmath332 reasoning as before we get @xmath333 since @xmath334 we finally reach @xmath335 from ( [ primer ] ) and ( [ segun ] ) we conclude ( [ efficient_evol ] ) with @xmath336 and @xmath337 . let us now assume that @xmath338 is an asymptotically exact error estimator . using again the decomposition ( [ eq : asterisco ] ) we have @xmath339 the last equality being a consequence of theorem ( [ semi_disve ] ) and the saturation hypothesis ( [ saturacion ] ) . as we pointed out before , this limit implies that @xmath307 is an asymptotically exact estimator of the error @xmath308 . then @xmath340 and @xmath338 is also an asymptotically exact estimator of the error in the approximation to the velocity of the evolutionary navier - stokes equations . we remark that with hypothesis ( [ saturacion ] ) we are merely assuming that the term of order @xmath341 is really present in the asymptotic expansion of the galerkin error . let us also notice that the constant @xmath342 in ( [ saturacion ] ) is , in general @xmath343 , so that the ratio @xmath344 in the proof of theorem [ apos_semi1 ] is , in general , @xmath345 . the key point in theorem [ apos_semi1 ] comes from the observation that if we decompose @xmath346 the first term on the right hand side of ( [ decom_error ] ) , @xmath347 is in general smaller , by a factor of size @xmath348 , than the second one , @xmath20 ( theorem [ semi_disve ] ) . then , to estimate the error @xmath22 we can safely omit the term @xmath30 in ( [ decom_error ] ) . comparing with the analysis of @xcite for a nonstationary linear stokes model problem the main difference is that the two terms in ( [ decom_error ] ) are taken into account . in theorem [ the_new ] below we show that this kind of technique can also be applied to the nonlinear navier - stokes equations . the advantage of this point of view is that hypothesis ( [ saturacion ] ) is not required for the proof of theorem [ the_new ] . let us finally observe that @xmath349 are the mfe approximations to the solution @xmath350 of the stokes problem that we obtain deriving respect to the time variable the stokes problem ( [ posth0])-([posth1 ] ) . then , we will denote by @xmath351 , @xmath352 , any a posteriori error estimator of the error @xmath353 in the norm of @xmath313 in the approximation to the velocity of the corresponding steady stokes problem . the proof of the following theorem follows the steps of the proof of ( * ? ? ? * theorem 1 ) . [ the_new ] let @xmath178 be the solution of ( [ onetwo])-([ic ] ) and let @xmath27 be its mfe galerkin approximation . then , the following a posteriori error bound holds for @xmath214 and a constant @xmath222 independent of @xmath19 . @xmath354 let us denote by @xmath355 . from ( [ posth0])-([posth1 ] ) it follows that @xmath356 then @xmath357 satisfies the equation @xmath358 taking into account ( [ lee1 ] ) we get @xmath359 we first observe that for any @xmath360 we have @xmath361 . then , taking into account ( [ eq : aux_bat ] ) we get @xmath362 let us observe that in order to apply ( [ eq : aux_bat ] ) we require @xmath8 to satisfy ( [ threshold ] ) , which holds due to ( [ eq : err_vel(t ) ] ) , and @xmath363 and @xmath364 to be bounded . using ( [ sob1 ] ) both norms are bounded in terms of @xmath365 . applying ( [ eq : may06-nueva ] ) we get @xmath366 finally , using that @xmath367 is uniformly bounded , see ( [ a_h_uh ] ) , and reasoning as in ( [ eq : adelanto0 ] ) to bound the second and forth terms above we conclude @xmath365 is uniformly bounded . then , we arrive at @xmath368 and then @xmath369 a standard application of a generalized gronwall lemma @xcite gives @xmath370 now , using decomposition ( [ decom_error ] ) we conclude the proof . we observe that using the same proof , a similar bound for the @xmath209 norm of the error can be obtained changing only @xmath371 by @xmath372 and @xmath373 by @xmath374 . let us also remark that theorem [ the_new ] allows to a posteriori obtain upper error bounds for the error in the approximation to the nonlinear navier - stokes equations using only upper error bounds for some stokes problems depending only on the data and the computed approximation . however , the estimation of the error at a time @xmath181 requires the estimation of the error of a family of stokes problems with right hand side depending on @xmath117 , for all @xmath375 $ ] . we now propose a simple procedure to estimate the error which is based on computing a mfe approximation to the solution @xmath202 of ( [ posth0])-([posth1 ] ) on a mfe space with better approximation capabilities than @xmath376 in which the galerkin approximation @xmath27 is defined . this procedure was applied to the @xmath6-version of the finite - element method for evolutionary convection - reaction - diffusion equations in @xcite . the main idea here is to use a second approximation of different accuracy than that of the galerkin approximation of @xmath178 and whose computational cost hardly adds to that of the galerkin approximation itself . let us fix any time @xmath377 $ ] and let approximate the solution @xmath378 of the stokes problem ( [ posth0])-([posth1 ] ) by solving the following discrete stokes problem : find @xmath379 satisfying @xmath380 where @xmath381 is either : 1 . the same - order mfe over a finer grid . that is , for @xmath382 , we choose @xmath383 . 2 . a higher - order mfe over the same grid . in this case we choose @xmath384 @xmath385 . we now study the errors @xmath386 and @xmath387 . [ teo4 ] let @xmath178 be the solution of ( [ onetwo])([ic ] ) and for @xmath316 , and let ( [ eq : u - inf])([eq : u - int ] ) hold with @xmath388 then , there exists a positive constant @xmath222 such that the postprocessed mfe approximation to @xmath5 , @xmath389 satisfies the following bounds for @xmath316 and @xmath390 $ ] : \(i ) if the postprocessing element is @xmath391 , then @xmath392 \(ii ) if the postprocessing element is @xmath393 , then @xmath394 for @xmath53 only the case @xmath249 in ( [ postcuad1 ] ) and ( [ postcuad2 ] ) holds . in ( [ postcuad1])([ulti2 ] ) , @xmath278 when @xmath53 and @xmath279 otherwise . the cases @xmath210 have been proven in theorems 5.2 and 5.3 in @xcite . following the same arguments , we now prove the results corresponding to @xmath53 and @xmath391 , the case @xmath395 being similar , yet easier . we decompose the error @xmath396 , where @xmath397 is the solution of @xmath398 that is , @xmath399 is the stokes projection of @xmath5 onto @xmath400 . since in view of ( [ stokespro])([stokespre ] ) we have @xmath401 we only have to estimate @xmath402 and @xmath403 . to do this , we subtract ( [ posth0dis ] ) from ( [ july-24th11 ] ) , and take inner product with @xmath404 to get @xmath405 now applying lemma [ le : z_t ] , ( [ eq : aux_bat ] ) and ( [ eq : err_vel(t ) ] ) the proof of ( [ postcuad1 ] ) is finished . to prove ( [ ulti1 ] ) , again we subtract ( [ posth0dis ] ) from ( [ july-24th11 ] ) , rearrange terms and apply the inf - sup condition ( [ inf - sup ] ) to get @xmath406 and the proof is finished with the same arguments used to prove ( [ postcuad1 ] ) . to estimate the error in @xmath407 we propose to take the difference between the postprocessed and the galerkin approximations : @xmath408 in the following theorem we prove that this error estimator is efficient and asymptotically exact both in the @xmath28 and @xmath209 norms and it has the advantage of providing an improved approximation when added to the galerkin mfe approximation . [ th_pos_esti ] let @xmath178 be the solution of ( [ onetwo])-([ic ] ) and fix any positive time @xmath14 . assume that condition ( [ saturacion ] ) is satisfied . then , there exist positive constants @xmath409 , @xmath410 , and @xmath320 , @xmath321 , @xmath323 and @xmath324 such that , for @xmath411 and @xmath412 , the error estimators @xmath413 @xmath414 satisfy the following bounds when @xmath415 and @xmath416 : @xmath417 furthermore , if @xmath415 , with @xmath418 , @xmath419 , or @xmath420 then @xmath421 for the mini element , the case @xmath238 in ( [ efi_posdis ] ) and ( [ asin_posdis ] ) must be excluded . we will prove the estimates for the velocity in the case @xmath210 , since the estimates for the pressure and the case @xmath53 are obtained by similar arguments but with obvious changes . let us observe that for @xmath226 @xmath422 on the other hand @xmath423 using ( [ saturacion ] ) we get @xmath424 taking @xmath425 and @xmath19 and @xmath426 sufficiently small , the bound ( [ efi_posdis ] ) is readily obtained . the proof of ( [ asin_posdis ] ) follows straightforwardly from ( [ july-24th ] ) , since in the case when @xmath415 with @xmath418 , @xmath419 , the term @xmath427 when @xmath19 tends to zero , and in the case when @xmath420 the term containing the parameter @xmath428 is not present . in practice , it is not possible to compute the mfe approximation exactly , and , instead , some time - stepping procedure must be used to approximate the solution of ( [ ten])-([ten2 ] ) . hence , for some time levels @xmath429 , approximations @xmath430 and @xmath35 are obtained . in this section we assume that the approximations are obtained with the backward euler method or the two - step bdf which we now describe . for simplicity , we consider only constant stepsizes , that is , for @xmath431 integer , we fix @xmath432 , and we denote @xmath433 , @xmath434 . for a sequence @xmath435 we denote @xmath436 given @xmath437 , a sequence @xmath438 of approximations to @xmath439 , @xmath440 , is obtained by means of the following recurrence relation : @xmath441 where @xmath442 in the case of the backward euler method and @xmath443 for the two - step bdf . in this last case , a second starting value @xmath444 is needed . here , we will always assume that @xmath444 is obtained by one step of the backward euler method . also , for both the backward euler and the two - step bdf , we assume that @xmath437 , which is usually the case in practical situations . we now define the time - discrete postprocessed approximation . given an approximation @xmath445 to @xmath446 , the time - discrete postprocessed velocity and pressure @xmath47 are defined as the solution of the following stokes problem : @xmath447 for reasons already analyzed in @xcite and @xcite we define @xmath448 as an adequate approximation to the time derivative @xmath446 . for the analysis of the errors @xmath449 and @xmath450 we follow @xcite , where the mfe approximations to the stokes problem ( [ posth0n])([posth1n ] ) are analyzed . we start by decomposing the errors @xmath449 and @xmath450 as follows , @xmath451 where @xmath452 and @xmath453 are the temporal errors of the time - discrete postprocessed velocity and pressure @xmath454 . the first terms on the right - hand sides of ( [ eq : decomtilde])([eq : decomtildep ] ) are the errors of the postprocessed approximation that were studied in the previous section . let us denote by @xmath42 , the temporal error of the mfe approximation to the velocity , and by @xmath455 , the temporal error of the mfe approximation to the pressure . in the present section we bound @xmath456 and @xmath457 in terms of @xmath458 . the error bounds in the following lemma are similar to those of ( * ? ? ? * proposition 3.1 ) where error estimates for mfe approximations of the stokes problem ( [ posth0n])([posth1n ] ) are obtained . [ lema_need]there exists a positive constant @xmath459 such that @xmath460 let us denote by @xmath461 where @xmath462 . subtracting ( [ posth0n])([posth1n ] ) from ( [ posth0])([posth1 ] ) we have that the temporal errors @xmath463 of the time - discrete postprocessed velocity and pressure are the solution of the following stokes problem @xmath464 on the other hand , subtracting ( [ tend])-([tend2 ] ) from ( [ ten])-([ten2 ] ) and taking into account that , thanks to definition ( [ dtu ] ) @xmath465 , we get that the temporal errors @xmath466 of the fully discrete mfe approximation satisfy @xmath467 and thus @xmath466 is the mfe approximation to the solution @xmath463 of ( [ otro_stokes])([otro_stokes2 ] ) . using then ( [ stokespro+1 ] ) we get @xmath468 for the pressure we apply ( [ stokespre ] ) and ( [ prehr ] ) to obtain @xmath469 then , to conclude , it only remains to bound @xmath470 . from the definition of @xmath471 it is easy to see that @xmath472 so that @xmath473 now , by writing @xmath160 as @xmath474 and using ( [ eq : adelanto0])([eq : adelanto1 ] ) we get @xmath475 from which we finally conclude ( [ error_post_fully ] ) and ( [ error_post_fully_pre ] ) . let us consider the quantities @xmath44 and @xmath476 as a posteriori indicators of the error in the fully discrete approximations to the velocity and pressure respectively . then , we obtain the following result : [ th_esti_time_d ] let @xmath178 be the solution of ( [ onetwo])([ic ] ) and let ( [ tildem2 ] ) hold . assume that the fully discrete mfe approximations @xmath438 , @xmath477 are obtained by the backward euler method or the two - step bdf ( [ tend])([tend2 ] ) , and let @xmath478 be the solution of ( [ posth0n])([posth1n ] ) . then , for @xmath479 , @xmath480 where @xmath481 is the constant in ( [ pri_fully])([pri_fully_pres ] ) , @xmath482 for the backward euler method and @xmath483 for the two - step bdf . in ( * ? ? ? * theorems 5.4 and 5.7 ) we prove that if ( [ tildem2 ] ) and the case @xmath484 in ( [ stokespro ] ) hold , the errors @xmath458 of these two time integration procedures satisfy for @xmath485 small enough that @xmath486 for a certain constants @xmath487 and @xmath321 , where @xmath482 for the backward euler method and @xmath483 for the two - step bdf . since @xmath488 , and then @xmath489 , from ( [ eq : orden1 ] ) and ( [ error_post_fully])-([error_post_fully_pre ] ) we finally reach that for @xmath485 small enough @xmath490 where @xmath481 is a positive constant . let us decompose the estimators as follows : @xmath491 which implies @xmath492 thus , in view of ( [ pri_fully])([pri_fully_pres ] ) we obtain ( [ decom_prin2 ] ) and ( [ decom_prin2p ] ) let us comment on the practical implications of this theorem . observe that from ( [ decom_prin ] ) and ( [ decom_prinp ] ) the fully discrete estimators @xmath44 and @xmath493 can be both decomposed as the sum of two terms . the first one is the semi - discrete a posteriori error estimator we have studied in the previous section ( see remark [ remark31 ] ) and which we showed it is an asymptotically exact estimator of the spatial error of @xmath38 and @xmath494 respectively . on the other hand , as shown in ( [ pri_fully])([pri_fully_pres ] ) , the size of the second term is in asymptotically smaller than the temporal error of @xmath38 and @xmath494 respectively . we conclude that , as long as the spatial an temporal errors are not too unbalanced ( i.e. , they are not of very different sizes ) , the first term in ( [ decom_prin ] ) and ( [ decom_prinp ] ) is dominant and then the quantities @xmath495 and @xmath493 are a posteriori error estimators of the spatial error of the fully discrete approximations to the velocity and pressure respectively . the control of the temporal error can be then accomplished by standard and well - stablished techniques in the field of numerical integration of ordinary differential equations . now , we remark that @xmath47 are obviously not computable . however , we observe that the fully discrete approximation @xmath438 of the evolutionary navier - stokes equation is also the approximation to the stokes problem ( [ posth0n])-([posth1n ] ) whose solution is @xmath496 . then , one can use any of the available error estimators for a steady stokes problem to estimate the quantities @xmath497 and @xmath498 , which , as we have already proved , are error indicators of the spatial errors of the fully discrete approximations to the velocity and pressure , respectively . to conclude , we show a procedure to get computable estimates of the error in the fully discrete approximations . we define the fully discrete postprocessed approximation @xmath499 as the solution of the following stokes problem ( see @xcite ) : @xmath500 where @xmath381 is as in ( [ posth0dis])-([posth1dis ] ) . let us denote by @xmath501 and @xmath502 the temporal errors of the fully discrete postprocessed approximation @xmath503 ( observe that the semi - discrete postprocessed approximation @xmath504 is defined in ( [ posth0dis])-([posth1dis ] ) ) . let us denote , as before , by @xmath458 the temporal error of the mfe approximation to the velocity . then , we have the following bounds . [ prop : err_post_fully ] there exists a positive constant @xmath505 such that for @xmath506 the following bounds hold @xmath507 the bound ( [ error_post_fully_vel ] ) is proved in ( * ? ? ? * proposition 3.1 ) . to prove ( [ duda2pp ] ) we decompose @xmath508 the second term above is bounded in ( [ error_post_fully_pre ] ) of lemma [ lema_need ] . for the first we observe that @xmath509 is the mfe approximation in @xmath510 to the pressure @xmath511 in ( [ otro_stokes])-([otro_stokes2 ] ) so that the same reasoning used in the proof of lemma [ lema_need ] allow us to obtain @xmath512 using ( [ eq : orden1 ] ) as before , we get the analogous to ( [ pri_fully ] ) and ( [ pri_fully_pres ] ) , i.e. , for @xmath485 small enough the following bound holds @xmath513 where @xmath481 is a positive constant . similarly to ( [ decom_prin])([decom_prinp ] ) we write @xmath514 and @xmath515 , so that in view of ( [ pri_fully_dis])([pri_fully_dis_pre ] ) we have the following result . [ th_esti_fully_d ] let @xmath178 be the solution of ( [ onetwo])([ic ] ) and let ( [ tildem2 ] ) hold . assume that the fully discrete mfe approximations @xmath438 , @xmath477 are obtained by the backward euler method or the two - step bdf ( [ tend])([tend2 ] ) , and let @xmath516 be the solution of ( [ posth0ndis])([posth1ndis ] ) . then , for @xmath479 , @xmath517 where @xmath481 is the constant in ( [ pri_fully_dis])([pri_fully_dis_pre ] ) , @xmath482 for the backward euler method and @xmath483 for the two - step bdf . the practical implications of this result are similar to those of theorem [ th_esti_time_d ] , that is , the first term on the right - hand side of ( [ decom_prin2_dis ] ) is an error indicator of the spatial error ( see theorem [ th_pos_esti ] ) while the second one is asymptotically smaller than the temporal error . as a consequence , the quantity @xmath518 is a computable estimator of the spatial error of the fully discrete velocity @xmath38 whenever the temporal and spatial errors of @xmath38 are more or less of the same size . as before , similar arguments apply for the pressure . we remark that having balanced spatial and temporal errors in the fully discrete approximation is the more common case in practical computations since one usually looks for a final solution with small total error . as in the semi - discrete case , the advantage of these error estimators is that they produce enhanced ( in space ) approximations when they are added to the galerkin mfe approximations . we consider the equations @xmath519 in the domain @xmath520\times[0,1]$ ] subject to homogeneous dirichlet boundary conditions . for the numerical experiments of this section we approximate the equations using the mini - element @xcite over a regular triangulation of @xmath49 induced by the set of nodes @xmath521 , @xmath522 , where @xmath523 is an integer . for the time integration we use the two - step bdf method with fixed time step . for the first step we apply the backward euler method . in the first numerical experiment we study the semi - discrete in space case . to this end in the numerical experiments we integrate in time with a time - step small enough in order to have negligible temporal errors . we take the forcing term @xmath524 such that the solution of ( [ onetwo_nu ] ) with @xmath525 is @xmath526 we chose @xmath527 in the first numerical experiment . when using the mini - element it has been observed and reported in the literature ( see for instance @xcite , @xcite , @xcite @xcite , @xcite and @xcite ) that the linear part of the approximation to the velocity , @xmath528 , is a better approximation to the solution @xmath5 than @xmath8 itself . the bubble part of the approximation is only introduced for stability reasons and does not improve the approximation to the velocity and pressure terms . for this reason in the numerical experiments of this section we only consider the errors in the linear approximation to the velocity . also , following @xcite , we postprocess only the linear approximation to the velocity , i.e. , we solve the stokes problem ( [ posth0dis])-([posth1dis ] ) with @xmath528 and @xmath529 on the right - hand - side instead of @xmath8 and @xmath530 . the finite element space at the postprocessed step is the same mini - element defined over a refined mesh of size @xmath428 . we show the galerkin errors and the a posteriori error estimates obtained at time @xmath531 by taking the difference between the postprocessed and the standard approximations to the velocity and the pressure . in figure [ semi ] , we ( asterisks ) and @xmath12 ( circles ) for @xmath532 , @xmath533 , @xmath534 , @xmath535 and @xmath536 and @xmath537 , @xmath538 , @xmath539 , @xmath540 and @xmath541 respectively . on the left , error estimations for the first component of the velocity . on the right , error estimations for the pressure.,title="fig:",width=226 ] ( asterisks ) and @xmath12 ( circles ) for @xmath532 , @xmath533 , @xmath534 , @xmath535 and @xmath536 and @xmath537 , @xmath538 , @xmath539 , @xmath540 and @xmath541 respectively . on the left , error estimations for the first component of the velocity . on the right , error estimations for the pressure.,title="fig:",width=226 ] have represented the errors in the first component of the velocity of the galerkin approximation in the @xmath11 and @xmath12 norms and the errors for the pressure in the @xmath11 norm using solid lines . we have used dashed lines to represent the error estimations . the results for the second component of the velocity are completely analogous and they are not reported here . the @xmath11 errors of the pressure , on the right of figure [ semi ] , are approximately twice as those of the @xmath12 errors of the velocity , on the left of figure [ semi ] , in this example . we can observe that with the procedure we propose in this paper we get very accurate estimations of the errors , specially in the @xmath12 norm of the velocity . the difference between the behavior of the error estimations in the @xmath11 and @xmath12 norms of the velocity are due to the fact that for first order approximations the postprocessed procedure increases the rate of convergence of the standard method only in the @xmath12 norm for the velocity and the @xmath11 norm for the pressure . however , since the postprocessed method produces smaller errors than the galerkin method also in the @xmath11 norm it can also be used to estimate the errors in this norm , as it can be checked in the experiment . on the right of figure [ semi ] we can clearly observe the asymptotically exact behavior of the estimator in the @xmath11 errors in the pressure in agreement with ( [ asin_posdis ] ) of theorem [ th_pos_esti ] . let us denote by @xmath542 the efficiency indexes for the first component of the velocity and for the pressure . in table [ tabla1 ] we have represented the values of the @xmath11 and @xmath12 norms of the velocity index and the @xmath13 norm of the pressure index for the experiments in figure [ semi ] . we deduce again from the values of the efficiency indexes that the a posteriori error estimates are very accurate , all the values are remarkably close to @xmath543 , which is the optimal value for the efficiency index . more precisely , we can observe that the values of the efficiency index in the @xmath11 norm for the velocity in this experiment belong to the interval @xmath544 $ ] . the values in the @xmath12 norm for the velocity lie on the interval @xmath545 $ ] and , finally , the values for the pressure are in the interval @xmath546 $ ] . to conclude , we show a numerical experiment to check the behavior of the estimators in the fully discrete case . we choose the forcing term @xmath7 such that the solution of ( [ onetwo_nu ] ) is ( [ solu_fix ] ) with @xmath547 . the value of @xmath525 and the final time @xmath531 are the same as before . ( asterisks ) and @xmath12 ( circles ) for @xmath548 . on the left : euler ; on the right : two - step bdf for @xmath549 to @xmath550.,title="fig:",width=226 ] ( asterisks ) and @xmath12 ( circles ) for @xmath548 . on the left : euler ; on the right : two - step bdf for @xmath549 to @xmath550.,title="fig:",width=226 ] in figure [ ulti ] , on the left , we have represented the errors obtained using the implicit euler method as a time integrator for different values of the fixed time step @xmath485 ranging from @xmath549 to @xmath550 halving each time the value of @xmath485 . for the spacial discretization we use the mini - element with always the same value of @xmath548 . we use solid lines for the errors in the galerkin method and dashed lines for the estimations , as before . the @xmath11 norm errors are marked with asterisks while the @xmath12 norm errors are marked with circles . we estimate the errors using the postprocessed method computed with the same mini - element over a refined mesh of size @xmath551 . we observe that the galerkin errors decrease as @xmath485 decreases until a value that corresponds to the spatial error of the approximation . on the contrary , the error estimations lie on an almost horizontal line , both for the velocity in the @xmath11 and @xmath12 norms and for the pressure . this means , as we stated in section 4.2 , that the error estimations we propose are a measure of the spatial errors , even when the errors in the galerkin method are polluted by errors coming from the temporal discretization . in this experiment the error estimations are very accurate for the spatial errors of the velocity in the @xmath12 norm and for the errors in the pressure . as commented above , the fact that postprocessing linear elements does not increase the convergence rate in the @xmath11 norm is reflected in the precision of the error estimations in the @xmath11 norm . on the right of figure [ ulti ] we have represented the errors obtained when we integrate in time with the two - step bdf and fixed time step . the only remarkable difference is that , as we expected from the second order rate of convergence of the method in time , the temporal errors are smaller for the same values of the fixed time step @xmath485 . again , the estimations lie on a horizontal line being essentially the same as in the experiment on the left . , _ gamma function and related functions in handbook of mathematical functions with formulas , graphs , and mathematical tables _ , edited by milton abramowitz and irene a. stegun . reprint of the 1972 edition . dover publications , inc . , new york , 1992 . j. g. heywood and r. rannacher , _ finite element approximation of the nonstationary navier stokes problem_. i. _ regularity of solutions and second - order error estimates for spatial discretization _ , siam j. numer . , 19 ( 1982 ) , pp . 275311 . , _ finite element approximation of the nonstationary navier stokes problem . _ iii : _ smoothing property and higher order error estimates for spatial discretization _ , siam j. numer . , 25 ( 1988 ) , pp . 489512 .

many radioactive nuclei relevant for gamma - ray astrophysics are synthesized during explosive events , such as classical novae and supernovae . a review of recent results of explosive nucleosynthesis in these scenarios will be presented , with a special emphasis on the ensuing gamma - ray emission from individual nova and supernova explosions . the influence of the dynamic properties of the ejecta on the gamma - ray emission features , as well as the still remaining uncertainties in nova and supernova modelling will also be reviewed .

astro-ph0103418

in this paper i will review the sites of explosive nucleosynthesis relevant for gamma - ray astronomy . two main types of explosion are responsible for the emission of gamma - rays in the galaxy : supernovae , both thermonuclear and gravitational ( core - collapse ) and classical novae . thermonuclear supernovae , or supernovae of type ia , are exploding white dwarfs in close binary systems , which do not leave a remnant after the explosion . core collapse supernovae ( supernovae of type ii , ib / c ) are exploding massive stars ( m@xmath0 ) , which leave as remnant either a black hole or a neutron star . typical velocities of supernovae ejecta are some @xmath1 km s@xmath2 , energies involved @xmath3 erg and ejected masses some m@xmath4 . classical novae are the result of the explosion of the external h - rich accreted shells of a white dwarf in a binary system . these explosions are recurrent phenomena ( contrary to supernovae ) , since an explosion is expected every time the critical accreted mass on top of the white dwarf is reached . typical velocities of novae ejecta are between some @xmath5 and @xmath6 km s@xmath2 , energies involved @xmath7 erg and ejected masses between @xmath8 and @xmath9 m@xmath4 . the radioactive isotopes synthesized during explosive nucleosynthesis , either in novae or in supernovae , are summarized in table 1 . three types of decay chains can occur : electron captures ( @xmath10ni @xmath11 @xmath10co , @xmath12ni @xmath11 @xmath12co @xmath11 @xmath12fe , @xmath13ti @xmath11 @xmath13sc and @xmath14be @xmath11 @xmath14li ) , @xmath15 decays ( @xmath10co @xmath11 @xmath10fe , @xmath13sc @xmath11 @xmath13ca , @xmath16al @xmath11 @xmath16 mg and @xmath17na @xmath11 @xmath17ne ) and @xmath18 decays ( @xmath19fe @xmath11 @xmath19co @xmath11 @xmath19ni ) . the first six isotopes in the table ( @xmath10ni , @xmath10co , @xmath12ni , @xmath13ti , @xmath16al and @xmath19fe ) are produced in supernova explosions ( although not exclusively , at least in the case of @xmath16al ) , whereas the last two are synthesized in classical novae . in sections 2 and 3 below , i will discuss how the synthesis of these radioactive nuclei proceeds in these scenarios . in the case of core - collapse supernovae , it is important to distinguish the nucleosynthesis during the pre - explosive stage of the massive star evolution from that in the explosive phases ; it is crucial as well to know which part of the star will finally be ejected , since this quantity will determine the final enrichment of the galaxy in radioactive ( and other ) elements . i wo nt discuss the synthesis of radioactive isotopes in other ( non - explosive ) sites , like the agb stars ( see the contribution by mowlavi , these proceedings ) or the wolf - rayet stars ( which can eject radioactive nuclei by strong stellar winds , during their hydrostatic evolution ; see arnould and meynet 1997 and meynet and arnould , 1999 for recent reviews ) . two types of isotopes can be distinguished , depending on their lifetime ( see diehl and timmes , 1998 , for a recent review ) . short - lived isotopes , such as @xmath10ni , @xmath12ni ( and their daughters @xmath10co and @xmath12co ) , @xmath13ti and @xmath19co , have lifetimes short enough ( see table 1 ) to make them detectable in individual objects . @xmath10ni and @xmath12ni are produced in all types of supernovae ; @xmath13ti is mainly produced in core - collapse supernovae , but it can also be synthesized in thermonuclear supernovae of the sub - chandrasekhar type ( provided that they exist ; see discussion of sneia types below ) . @xmath19co is produced directly and from @xmath19fe decay , with @xmath19fe belonging to the long - lived isotopes group . long - lived radioactive isotopes , such as @xmath16al and @xmath19fe , have lifetimes long enough to make them undetectable in individual sources , because the nuclei can be quite far away from their source and mixed with those coming from other explosions ( since the lifetime is longer than the typical period between two succesive explosions in the galaxy ) . for these isotopes , only the accumulated emission in the galaxy can be observed and used as diagnostic of models and of the galactic distribution of the sources . the same classification scheme applies to isotopes synthesized in novae ; in this case , @xmath14be belongs to the short - lived group , whereas @xmath17na belongs to both of them ( see section 3 below ) . a very recent and interesting compilation of papers about astronomy with radioactivities can be found in diehl and hartmann ( 1999 ) . gamma - ray astronomy provides an unique opportunity to detect radioactive isotopes in individual objects , giving a proof of ongoing nucleosynthesis in them . in the case of supernovae , another tool for the determination of the amount of radioactive nuclei in the ejecta is the bolometric light curve ( uvoir , from _ ultraviolet , optical _ and _ infrared _ ) . in addition to the well known fact that @xmath10co ( daughter of @xmath10ni ) powers the early evolution of the light curve ( @xmath10ni mass can be determined from luminosity at maximum ) , @xmath12co is responsible for powering the light curve from around day 1000 after maximum . later on @xmath13ti will provide a floor to the bolometric light curve and @xmath17na and @xmath19co could also play a minor role , depending on the supernova type and the specific yields of these radioactivities ( see , e.g. , woosley , pinto and hartmann , 1989 , timmes et al . 1996 , diehl and timmes 1998 ) . these two types of observational approaches to the radioactive content in the ejecta had been possible for only one object so far : the supernova 1987a , which exploded 13 years ago in the lmc , only 55 kpc away from us . this was a type ii supernova , which is not the most favorable case to look for gamma - ray emission , since its is much more opaque and has a smaller content of the most relevant radioactivites than snia ( see below ) ; however , its very short distance allowed for detection of its gamma - ray emission and also for a follow - up until very late times of its light curve ( through photometry in the uvbrijhk bands ) . gamma - ray lines from @xmath10co decay , at 847 and 1238 kev , were detected in sn 1987a with the grs instrument of the smm satellite ( matz et al . 1988 ) and confirmed by several ballon - borne instruments ( i.e. , teegarden et al . 1989 , mahoney et al . 1988 , sandie et al . 1988 , cook et al . 1988 , rester et al . one surprising fact was the appearence of these lines only 200 days after the explosion ( matz et al . , 1988 ) , much earlier than expected . this has been interpreted as a sign of some early extra mixing of @xmath10co in order to transport this isotope into regions of low gamma - ray optical depth ( see , e.g. , pinto and woosley , 1988 , leising 1988 , bussard , burrows and the , 1989 , leising and share , 1990 ) . the line profiles observed with gris ( teegarden et al . 1989 , tueller et al . 1990 ) have also put constraints on theoretical models of supernova explosions , since sphericity and homogeneity of the ejecta were incompatible with the observed fluxes and widths of the @xmath10co lines . another crucial gamma - ray observation of short - lived isotopes in sn 1987a was the detection of gamma - ray radiation from @xmath12co decay ( between 50 and 136 kev ) , with osse on the cgro ( kurfess et al . the deduced @xmath12co content ( for models with low gamma - ray optical depth , see kurfess et al . 1992 ) was such that the original ratio @xmath12ni/@xmath10ni produced in the explosion should be about 1.5 times the solar @xmath12fe/@xmath10fe ratio . observations up to now show the change of slope related to the sequence of @xmath10co-@xmath12co decays ( see figure 3 in timmes et al . future uvoir observations would possibly be able to show the light curve powering from @xmath13ti . the spi instrument onboard integral has some possibilities to detect the gamma - ray emission from this @xmath13ti , which would provide a unique proof of the nucleosynthesis in core collapse supernovae and an important link between the uvoir and the gamma - ray observations . coming back to @xmath12co it was first thought that the sn 1987a @xmath12co content deduced from osse observations was not enough to power the available bolometric light curve , since 5 times solar @xmath12fe/@xmath10fe ratio was required ( suntzeff et al . 1992 ) and alternative mechanisms to power the bolometric light curve were suggested ( clayton et al . 1992 ) . however , more recent observations seem to require a smaller amount of @xmath12co - decay to power the light curve , in agreement with the @xmath12fe/@xmath10fe ratio deduced from osse observations ( see figure 9 in diehl and timmes 1998 ) . the excitement induced by the above mentioned gamma - ray observations ( together with many other observations at other wavelength ranges ) has led the theorists to suggest different possibilities for mixing both during the explosion and the ejection phases ( to quote only a few of the early works , see e.g. , arnett , fryxell and mller 1989 , benz and thielemann 1990 , fryxell , arnett and mller 1991 , herant and benz 1992 , and also the general reviews of sn 1987a from arnett et al . 1989 and mccray 1993 and references therein ) . another detection ( tentative ) of gamma - ray emission from a supernova , with comptel on cgro , was that of sn 1991 t ( which was an overluminous snia ) , in ngc 4527 at around 17 mpc distance . in that case , a marginal detection of the 847 kev line was reported ( morris et al . 1995 , 1997 ) , leading to a prediction of a quite large @xmath10ni mass , implying that all the white dwarf mass should have been incinerated to @xmath10ni ( in contradiction with current theoretical models ) . more recently , upper limits to the fluxes of the 847 and 1238 kev lines from @xmath12co - decay in the type ia supernova 1998bu , in ngc 3368 at around 8 mpc distance , have been deduced from comptel observations ( georgii et al . 2000 ) . although no detection has been obtained , these limits are restrictive enough to constrain some of the available models of sneia nucleosynthesis ( like a sub - chandrasekhar mass model from nomoto et al . 1997 ) . there is another important observation of gamma - ray lines related to short and medium - lived radioactivities in supernovae : the discovery of @xmath13ti emission at 1157 kev in the cas a supernova remnant ( iyudin et al . 1994 ; see reviews from diehl and timmes 1998 and kndlseder , these proceedings ) . again the observations in gamma - rays are in some way puzzling , because the amount of @xmath13ti deduced from observations implies a @xmath10ni content ( according to theoretical models of supernovae nucleosynthesis ) which should have originated a very bright supernova , in contrast with the absence of historical records ( timmes et al . 1996 ) . observations in gamma - rays push forward the theoretical models , in order to balance all the available possibilities and to consider new ones . a different kind of information is obtained from the observations of long - lived radioactivities ( @xmath16al and @xmath19fe ) . in this case , what is seen is not the ongoing nucleosynthesis in a particular object , but the integrated nucleosynthesis in the galaxy . up to now , this has been possible for the 1809 kev @xmath16al emission . the @xmath16al map obtained with the comptel instrument onboard the compton gamma - ray observatory cgro ( diehl et al . 1995 , 1997 , oberlack et al . 1996 , kndlseder 1997 , 1999 , plschke et al . these proceedings ) has posed interesting questions about the origin of the galactic @xmath16al ( see , e.g. , review from prantzos and diehl , 1996 ) . it provides a direct and unique insight on the integrated nucleosynthesis during the last @xmath20 years . some regions of enhanced emission have been discovered ( cygnus , carina , vela ) , indicating the presumable link between @xmath16al emission and massive star formation , as well as the relationship with spiral structure of the galaxy ( diehl et al . 1996 , kndleseder et al . 1996a , b , diehl et al . 1999 , kndlseder , these proceedings ) . for @xmath19fe , a similar map should be observed by integral , because the sources of this isotope are the same as those of @xmath16al , being the yields smaller by some factor ( timmes et al . 1995 , diehl et al . , 1997 ) . it is worth mentioning that the integrated nucleosynthesis of @xmath13ti and @xmath17na may also be seen , if instruments are sensitive enough . the future @xmath13ti and @xmath17na maps will provide a precious information about their sources , i.e. , supernovae ( mainly core collapse ones ) and novae , respectively . the defining characteristic of sneia is the lack of hydrogen in their spectra , as well as the presence of a p cygni feature related to siii , @xmath21 , at maximum light ( wheeler and harkness 1990 ) ; in general , intermediate - mass elements ( o , mg , si , s , ar , ca ) appear in the spectrum near maximum light with high velocities ( 8000 - 30000 km s@xmath2 ) . sneia are quite homogeneous from the observational point of view ( i.e , @xmath2290% of all sneia have similar spectra , light curves and peak absolute magnitudes ) , although some differences exist ( i.e. , subluminous explosions , like sn1991bg and sn1992k , and overluminous ones , like sn1991 t ) . sneia appear in both elliptical and spiral galaxies and , therefore , their progenitors should be long - lived . these facts all together suggest that the thermonuclear disruption of mass - accreting carbon - oxygen ( co ) white dwarfs is responsible for these explosions . already in the sixties , hoyle and fowler ( 1960 ) suggested that thermonuclear burning in an electron - degenerate stellar core might be responsible for type i supernova ( there was no subclassification at the epoch ) explosions , with the explosion energy coming from the thermonucler burning of co into higher mass elements ( see also , e.g. , the pioneering works by arnett 1969 , hansen and wheeler 1969 ) . it was also suggested at the epoch that the early supernova luminosity might have its origin on the radioactive decay of @xmath10ni ( colgate and mckee 1969 ) , which was already known to be a product of supernova nucleosynthesis , and that gamma - ray lines should be emitted from those explosions ( clayton , colgate and fishman , 1969 ) . but the particular _ scenario _ where the explosion occurs ( see , e.g. , livio 1999 ) and the _ physics of the flame _ itself ( see , e.g. , hillebrandt and niemeyer 2000 ) are far from being understood . two types of progenitors have been suggested so far , concerning the mass of the exploding co white dwarf : chandrasekhar and sub - chandrasekhar mass models . in the chandrasekhar mass models , a co white dwarf explodes when reaching that mass , with central carbon ignition propagating outwards being responsible for the explosion . the main problems related to this model are the uncertainties concerning burning propagation ( deflagration , detonation , delayed detonation , see below ) , but also the scenario is unclear . either a double degenerate ( merging of two co white dwarfs ) or a single degenerate scenario is possible . in all cases , the growth to the chandrasekhar mass is problematic , because both mass loss ( through nova episodes , for instance ) and accretion induced collapse ( if the initial mass is high enough and/or the white dwarf is made of oxygen and neon ) should be avoided ( see , e.g. , canal , isern and labay 1990 , 1992 , canal et al . 1990 , isern , canal and labay 1991 , nomoto and kondo 1991 , bravo and garca - senz 1999 ) . in addition , for the double degenerate scenario there is a problem of statistics : there are not enough double white dwarf systems with sufficiently short period and total mass in excess of the chandrasekhar mass able to explode in less than the hubble time and to explain the galactic sneia rate . in fact there was nt any observed system fulfilling these conditions until the very recent discovery of kpd 1939 + 2752 ( maxted , marsh and north , 2000 ) , which is the first snia progenitor candidate observed . in the sub - chandrasekhar mass models , a co white dwarf of low - mass ( 0.6 - 0.8 m@xmath4 ) accretes helium ( @xmath23 m@xmath4 ) , reaching a final mass smaller than the chandrasekhar mass . provided the accretion rate is moderate ( around @xmath24m@xmath25 ) , there is he ignition on the top of the co core . this ignition causes an outward propagating he - detonation wave ( basically transforming he into ni at high velocity ) and an inward propagating pressure wave . the last one finally provoques a carbon ignition ( central or off - center ) , which leads to an outward carbon - detonation incinerating all the white dwarf , and synthesizing intermediate - mass elements , in addition to ni ( see , e.g. , livne 1990 , livne and glasner 1991 , woosley and weaver 1994 ) . therefore , in this model ( called `` indirect double detonation '' , idd , or `` edge lit detonation '' , eld ) there is an outer layer of high - velocity ni and he above the intermediate - mass elements , which does not exist in the chandrasekhar - mass models . sub - chandrasekar mass models are not considered as good sneia progenitors nowadays , because of both observational and theoretical problems ; observational : the high velocity ni above intermediate mass elements is not seen in the spectra ; theoretical : the he - driven carbon detonation is very model dependent ( see for instance the 3d models from garca - senz , bravo and woosley 1999 ) . but it is still a possibility that sub - chandrasekar mass models explain some subluminous sneia , like sn1991bg ( see , e.g. , ruiz - lapuente , canal and burkert 1997 ) . in summary , the bulk of normal sneia are assumed to be exploding chandrasekhar - mass co white dwarfs , but there is still room for the sub - chandrasekhar mass models to explain some peculiar objects . therefore , whether sneia come from single or double - degenerate scenarios and whether they come from carbon or helium plus carbon ignition are not closed issues ( see , e.g. , the recent paper from branch 2000 ) . the main problems still remaining on the modeling of sneia affect the ignition process and the flame propagation . different possibilities exist : deflagration ( subsonic flame speed ) , detonation ( supersonic ) and a combination of both ( delayed detonation ) . a detonation with densities larger than @xmath26 g @xmath27 is not a viable mechanism , since all the star would be incinerated to ni , without synthesis of intermediate - mass elements . on the contrary , if the density is lower , intermediate - mass elements are synthesized , in agreement with the observations . concerning deflagrations , they produce nucleosynthesis at velocities in general agreement with the observed spectra , but some neutronized isotopes ( such as @xmath28fe , @xmath28cr and @xmath29ni ) are overproduced in amounts incompatible with the chemical evolution of the galaxy . to overcome this problem , delayed detonations were suggested ( khokhlov 1991a ) . there are two situations in which a deflagration to detonation transition ( ddt ) could occur in supernovae ( see , e.g. , khokhlov , oran and wheeler , 1997 ) : ddt could occur directly or as a result of a previous expansion . for instance , in the pulsation delayed detonation , a first slow deflagration is quenched because of the expansion of the white dwarf , which subsequently pulses and recontracts , causing a detonation upon recollapse ( khokhlov 1991b ) . the propagation of the detonation wave through the pre - expanded star produces the required intermediate mass elements in the outer layers at densities lower than @xmath26 g @xmath27 ( which are not synthesized in detonations at larger densities ) . in these models , the problem of overproduction of highly - neutronized nuclei is alleviated but not solved ( khokhlov 1991a , b ) . models of delayed detonations in 2d , both of the first deflagration phase and of the subsequent detonation phase , have been performed by arnett and livne ( 1994a , b ) ; they show that the first slow deflagration is insufficient to unbind the star , that a pulsation of large amplitude is generated and that reignition occurs after the first contraction phase . in summary , there is a general consensus about the fact that , in order to explain spectroscopic observations , burning should proceed subsonically ( deflagration ) in the inner core ( where densities are large , i.e. , @xmath30 g @xmath27 ) , whereas burning becomes supersonic ( detonation ) in the outer lower density zones ( see examples of models in bravo et al . 1993 , hflich and khokhlov 1996 , bravo et al . 1996 , woosley 1997 ) . but the way in which the deflagration to detonation transition ( ddt ) occurs is not yet clear , ( see , e.g. , discussions in recent papers by niemeyer and woosley 1997 , niemeyer 1999 , lisewski , hillebrandt and woosley 2000 , and in the review by hillebrandt and niemeyer 2000 ) . there is also ample debate about the way in which the initial burning occurs : flame instabilities , flame - turbulence interactions ( see review about turbulence and thermonuclear burning by hillebrandt and niemeyer 1997 , and references therein ) . all 1d models ( which were the unique ones available up to the nineties and still are the only ones to include _ complete nucleosynthesis _ ) rely on prescriptions based on some parametrization of the flame speed and , in the case of delayed detonations , of the deflagration - detonation transition -ddt- densities . different groups work in models of thermonuclear snia and their nucleosynthesis , including the radioactivities . it is out of the scope of this paper to mention even a small fraction of them , but a small sample can be useful to show the main results and the main caveats still remaining ( see the recent books _ thermonuclear supernovae _ , edited by ruiz - lapuente , canal and isern , 1997 , and _ type ia supernovae : theory and cosmology _ , edited by niemeyer and truran , 2000 ) . nomoto and coworkers have computed detailed nucleosynthesis in carbon deflagration supernovae ( nomoto , thielemann and yokoi , 1984 , thielemann , nomoto and yokoi , 1986 ) , and also in other types of explosive carbon burning ( such as delayed detonations , with parametrized ignition densities and deflagration - detonation transition -ddt- densities , see iwamoto et al . the yields of radioactive isotopes are mainly affected by the ddt density ( i.e. , synthesized mass of @xmath10ni ranges from 0.55 to 0.77 m@xmath4 , and @xmath12ni from 9.6x@xmath8 to 1.98x@xmath31 m@xmath4 , in iwamoto et al.s models ) . these yields are larger than those from core collapse supernovae ( see below ) and distributed in less opaque zones , since there is nt much mass above them . this makes type ia supernovae better targets for integral than sneii ( but see section 2.1 for observational results ) . in the context of gamma - ray astronomy , it is important to stress that sub - chandrasekhar mass models synthesize larger amounts of @xmath13ti than chandrasekhar mass ones ( see , e.g. , woosley and weaver , 1994 ) . gamma - ray spectra of sneia for the different models provide important signatures of the explosion mechanism , although unfortunately there is nt much observational data to compare with ( see previous section ) . prospects for sneia explosion mechanism identification with gamma - rays have been analyzed recently by gmez - gomar et al . ( 1998a ) , with a special emphasis on detectability with the instruments that will be onboard integral ( see also burrows and the 1990 , hflich , khokhlov and mller 1994 , kumagai and nomoto 1997 , hflich , wheeler and khokhlov 1998 ) . lines from @xmath10ni - decay ( 158 , 750 , 812 kev ) are prominent during the first days after the explosion , but they disappear very fast , because of the short @xmath10ni - lifetime . lines from @xmath10co ( 847 , 1238 kev ) and @xmath12co ( 122 , 136 kev ) appear later and have longer durations . the most intense lines are those at 847 , 1238 , 812 and 158 kev , in addition to the annihilation line at 511 kev . the strongest line is always the 847 kev one ( detectable up to 11 - 16 mpc with spi on integral ) , whereas the 158 kev line ( from @xmath10ni - decay ) is the most interesting to discriminate between models . the 158 kev line is narrower and , therefore , detectable at longer distances with spi , than another @xmath10ni - line ( at 812 kev ) , despite being fainter . it is almost undetectable in pure deflagration models , whereas it is even stronger than the 1238 line in detonation models . another interesting signature of the models is the ratio between the 847 kev and the 158 kev line fluxes ( 200 days after maximum and at maximum , respectively ) , because it provides information about the ratio between total @xmath10ni in the ejecta and @xmath10ni in the external layers : the late emission at 847 kev comes from @xmath10co - decay ( coming from @xmath10ni - decay ) , while only the @xmath10ni present in the outermost shells is responsible for the 158 kev line flux ( see gmez - gomar et al . 1998a for details ) . finally , line profiles will also provide important information allowing for discrimination between the models , for explosions at distances short enough ( see again gmez - gomar et al . 1998a ) . all supernova types except type ia s ( i.e. , type ii , ib / c ) are explained by the explosion of massive stars . stars with initial masses ( m@xmath32 m@xmath4 ) do nt end their lives as white dwarfs . succesive phases of thermonuclear burning ( c , ne , o , si ) give as a result a star with an `` onion - skin '' structure , where a central iron core is surrounded by shells made of elements of progressively lower atomic mass . the chemical composition along the star is the following ( see , e.g. , figure 10.8 , corresponding to a 25 m@xmath4 star , in arnett 1996 ) : fe core , _ si - burning _ zone ( made mainly of elements from si to ni , without o ) , _ o - burning _ zone ( o , si - ca ) , _ ne - burning zone _ ( ne , mg and o , no c ) , _ c - burning _ zone ( c and o , ne and mg ) , radiative he - burning zone ( he , c and o ) , convective he zone , inert part of old he core ( interior to h - burning shell ) , material above the h - burning shell ( plus some inert zones associated with the si , o , ne and c - burning zones and just outside them ) . once the fe - core reaches the chandrasekhar mass , it becomes unstable and collapses to form a neutron star . the gravitational energy released ( @xmath33 erg ) during core collapse is responsible for the ensuing supernova explosion , but it is not yet completely understood how the conversion of this potential energy into kinetic energy proceeds ( only 0.1% of the available potential energy is needed ) . baade and zwicky ( 1934 ) were the first to suggest that the gravitational energy released during the formation of a neutron star could produce a supernova explosion . colgate and white ( 1966 ) built a supernova model , considering that the transfer of energy takes place by the emission and deposition of neutrinos ; wilson ( 1971 ) showed that the electron capture neutrino burst was not strong enough to eject material . the weinberg - salam model of electroweak interactions opened new possibilities of neutrino interactions with matter ( neutral currents ) . in 1974 , freedman noticed the importance of neutral currents in the physics of core collapse supernovae ; as a result of the increased cross section of core material to neutrinos , these particles are trapped during the collapse . it was shown by bethe et al . in 1979 that one of the consequences of neutrino trapping is that the entropy of the core changes little during collapse ( it remains low ) , leaving the collapse continue up to nuclear densities . further compression is prevented by the repulsive component of the strong interaction ( stiffness of nuclear matter ) , leading to the core bounce . a shock wave is generated at its boundary and propagates outwards . but it has been shown that the energy of this shock is mainly invested in the photodisintegration of heavy nuclei and in neutrino losses ; therefore , the shock stalls and the explosion via the so called `` prompt mechanism '' is unsuccessful . in the `` delayed mechanism '' , there is a revival of the stalled shock because of neutrino heating behind the shock ( bethe and wilson 1985 ) . however , the explosion energy does not reach easily the necessary @xmath3 erg . further works introduced the effect of convective instabilities , caused by a negative entropy gradient , in order to deliver energy to the shock ( see , e.g. , bethe , 1990 , herant , benz and colgate 1992 , herant et al . 1994 , bethe 1995 , janka and mller , 1995 , burrows , hayes and fryxell , 1995 , to quote only a few of the papers dealing with this topic ) . convection aids the explosion because it increases the efficiency at which neutrino energy is deposited ( material that rises cools and converts energy from neutrino deposition into kinetic energy , instead of re - radiating it as neutrinos ) and also reduces the energy required to launch the explosion ( by reducing the pressure at the accretion shock ) ( see recent reviews by fryer , 2000 , burrows 2000 , and references therein ) . the handling of this process is very model dependent : treatement of neutrino transport , multidimensional aspects . also the structure of the stellar core before collapse ( i.e. , the presupernova model ) are important for the final outcome of the explosion . fortunately , nucleosynthesis during core collapse supernova explosions can be computed without a complete knowledge of the explosion mechanism itself . as in the case of thermonuclear supernovae , all the details of the physics involved in the explosion are not required to have an approximate , but quite good when compared with the observations , idea of which are the main nucleosynthetic yields of core collapse supernovae . two steps are needed to compute snii ( and ib / c ) yields : nucleosynthesis during the massive star evolution ( i.e. , pre - supernova phase ) and explosive burning when a shock wave crosses the mantle surrounding the collapsing core . there are different ways to simulate the explosion artificially . one is by means of a `` thermal bomb '' , i.e. , injecting thermal energy inside the fe core , in a way such that the ejecta attains the desired kinetic energy , @xmath34erg ( see , e.g. , thielemann , nomoto and hashimoto , 1996 ) . another alternative is the injection of momentum , through a piston , inward - moving during the infall previous to the explosion , and outward - moving during the explosion , with a velocity such that the desired kinetic energy of the ejecta is obtained ( see , e.g. , woosley and weaver , 1995 ) . the _ mass cut _ between the collapsing core and the ejecta determines the amount of mass ejected ( and that of @xmath10ni and other radioactive isotopes , in particular ) . in the `` thermal bomb '' method , they adjust it taking into account the relationship between supernova progenitor masses and @xmath10ni masses ejected deduced from some observations . in the piston approach , the mass cut is obtained from the choice of piston position and energy ; a mass cut located outside the piston is often obtained ( for a discussion of the differences between both models , including an analysis of the influence of the nuclear reaction rates , see hoffman et al . , 1999 ) . in summary , both groups have performed calculations of detailed nucleosynthesis by inducing the core - collapse supernova explosion on massive stars ( previously evolved following all the nucleosynthesis phases ) . other groups have performed studies of massive star evolution , but there is no room in this short review to mention all of them . the masses studied by thielemann et al . ( 1996 ) range between 13 and 25 m@xmath4 , with initial metallicities , z , equal to solar ( see nakamura et al . 1999 for the effect of low z ) . woosley and weaver ( 1995 ) studied the range 11 - 40 m@xmath4 , for z=0 and z between @xmath35 and z@xmath4 . si , o , ne and c explosive burning occur when the shock wave crosses the corresponding zones in the pre - supernova ( see above for the description of its structure ) . a brief description of the results concerning the synthesis of radioactive isotopes follows . @xmath10ni and @xmath12ni are produced when either oxygen or silicon - rich layers with low neutron excess ( y@xmath36 ) are heated to temperatures above 4x@xmath37 k ( explosive o- and si - burning ) . they are produced whether the material ejected is alpha - rich or not , although @xmath12ni synthesis is favored in alpha - rich freeze - out ; this happens when material , initially in nuclear statistical equilibrium ( nse ) at relatively low density , is cooled so rapidly that the free alpha particles do not have time to merge via the 3@xmath38 reaction and , therefore , matter cools down in the presence of a large concentration of @xmath38-particles , which modify the final composition ( with respect to the normal freeze - out ) . @xmath13ti is also produced during @xmath38-rich freeze - out from nse in the hottest and deepest layers ejected during the explosion . therefore , the yields of these radioactive isotopes are very sensitive to the mass - cut location ( woosley and hoffman 1991 , hoffman et al . 1995 , woosley and weaver 1995 , timmes et al . for example , stars with masses larger than 30 m@xmath4 do nt eject any @xmath10ni ( nor @xmath12ni and @xmath13ti ) if the kinetic energy ( at infinity ) is around 1.2x@xmath3 erg . if this energy is enhanced , the mass - cut is lowered and some @xmath10ni ( and @xmath12ni and @xmath13ti ) are ejected . ejected masses of @xmath10ni are around 0.1 m@xmath4 and those of @xmath13ti between @xmath39 and @xmath35 m@xmath4 . similar results are obtained by thielemann et al . ( 1996 ) , except for the larger amounts of @xmath13ti , probably because of the different way of simulating the explosion , which possibly injects larger entropy in the inner shells and favors a larger ejected mass and an enhanced @xmath38-rich freeze - out ( see , e.g. , aufderheide , baron and thielemann 1991 and hoffman et al . 1999 ) . @xmath16al is another important radioactive isotope which is produced in core collapse supernovae ( and in other scenarios ) through the @xmath40mg(p,@xmath41 ) reaction . @xmath16al yields depend on pre - supernova evolution ( h- and o - ne burning shells ) and on the explosion . two factors enhance @xmath16al production during the explosion : explosive burning in o - ne shells and @xmath42-spallation reactions on @xmath43ne , @xmath44o , @xmath45na , @xmath46 mg , which liberate protons that are captured by @xmath40 mg . it is important to stress that another important long - lived radioisotope , @xmath19fe , is coproduced with @xmath16al in the same regions within snii ( this isotope is synthesized by neutron captures on @xmath47fe in the o - ne burning shell and in the base of the he - burning shell , both pre - explosively and explosively ) . therefore , these nuclei should have similar spatial distributions in the ejecta ( timmes et al . the @xmath16al/@xmath19fe ratio depends on the mass of the pre supernova : @xmath16al/@xmath19fe is larger than 1 for m larger than 25 m@xmath4 and similar to 1 for smaller masses . the typical yields of @xmath16al are @xmath35 m@xmath4 and those of @xmath19fe 4x@xmath9 m@xmath4 . the final yields depend on three aspects : presupernova evolution , explosion energy and details of the explosion mechanism ( see diehl and timmes , 1998 and thielemann 1999 for recent analyses ) . the main issues concerning presupernova evolution are those affecting general stellar evolution of low - mass stars , plus some specific ones relative to massive stars . for instance , the treatement of convection affects nucleosynthesis ; in particular , convective burning in the o - shell of massive stars . models with m=20 m@xmath4 have deserved a particular attention for the theorists , since they are crucial to understand the mixing of radioactive ( and other ) isotopes , like @xmath10ni , in supernova ejecta , which has been deduced from the observations of sn1987a ( see section 2.1 above ) . 2d models of o - burning ( bazan and arnett 1994 ) obtain significant mixing beyond the boundaries defined by mixing - length convection . what they obtain are perturbations in density in the oxygen shell that are sufficiently large to `` seed '' hydrodynamic instabilities , which will mix the `` onion - skin '' composition of the presupernova ( bazan and arnett 1998 ) . this occurs in precisely the region in which @xmath10ni is explosively produced by oxygen burning behind the explosion shock . this result poses some problems to the models of explosive nucleosynthesis based on 1d presupernova evolution . rotation can also have some effect ( see , e.g. , works by meynet and maeder 1997 , heger , langer and woosley 2000 ) , as well as mass - loss during the presupernova evolution , in the final yields of radioactive elements . concerning the energy of the explosion and the details of the explosion mechanism , one of the main problems is the location of the mass - cut , which determines how much mass falls back into the collapsing core ( and therefore whether it will be a neutron star or a black hole ) and how much mass is ejected and with wich composition ( the profile of some isotopes is steep around the mass - cut location and , therefore , the yield is affected by it ) . therefore , the mass - cut determines crucially the final yields of radioactive elements , specially for those produced in the inner regions of the supernova ( @xmath10ni , @xmath12ni and @xmath13ti ) . as mentioned above , the explosion energy , and the corresponding entropy in the inner shells , crucially affect the degree of @xmath38-rich freeze - out and , therefore , the yields of the fe - group nuclei and of @xmath13ti . classical novae explosions are the most common explosions in the galaxy . the cause of the explosion is a thermonuclear runaway ( tnr ) on top of a white dwarf , ensuing the degenerate burning of the accreted hydrogen ( starrfield 1989 , hernanz & jos 2000 ) . the synthesis of radioactive isotopes in classical novae is important for two reasons . first , some of the isotopes produced are crucial for the explosion mechanism itself ( i.e. , @xmath48o , @xmath49o , @xmath50f with lifetimes 102 , 176 and 93s , respectively ) . the reason is that these isotopes are transported by convection to the outer layers of the envelope , during the runaway , where they subsequently decay ( @xmath51 ) and cause the expansion of the envelope and the increase in visual luminosity . second , the decay of the unstable nuclei originates gamma - ray emission , because of either direct emission of gamma - ray photons or positrons ( for @xmath52-unstable nuclei ) , which annihilate with electrons . the photons emitted ( 511 kev , positronium continuum , 478 and 1275 kev , see table 1 ) experience comptonization in the nova expanding envelope . therefore , the emission from novae consists of lines plus a continuum ( see gmez - gomar et al . 1998b , hernanz et al . 1999 , hernanz et al . these proceedings and references therein ) . the potential role of classical novae as gamma - ray emitters had been alredy pointed out many years ago ( clayton and hoyle 1974 , clayton 1981 , leising and clayton 1987 ) . the first available hydrodynamic models of nova explosions ( starrfield et al . 1978 and prialnik et al . 1978 ) realized that there was a need of an initial enrichment in cno isotopes both to power the explosion and to explain some observed abundances . two and three dimensional simulations of the thermonuclear runaway of a co white dwarf , valid when the accreted envelope has been already built up , are the only available up to now ( glasner et al . , 1997 , kercek et al 1998 , 1999 ) . they predict that enrichment proceeds too slowly if the accreted gas has nearly solar cno abundances at the onset of the thermonuclear runaway , and conclude that fast nova outbursts require huge enrichments of c and o. the mechanism which leads to such enhancements must operate prior to the outburst and has not been modeled up to now . therefore , it is known that some mixing with core material ( either co or one ) during the accretion phase prior to the runaway should occur , but this process has not been modeled yet in a self - consistent way . another approach to the problem of initial enrichment comes from the multicycle 1d models ( of co novae only , up to now ) , from prialnik and kovetz ( prialnik and kovetz 1995 , kovetz and prialnik , 1997 ) ; diffusion is responsible for the enrichment , which becomes larger after a number of flashes . however , the large metallicities and neon abundances observed in some novae are not well reproduced . another approach is based on 1d models with an initial ( parametrized ) enrichment , such that the general properties of observed novae ( mainly abundances ) are well modeled ( see for instance starrfield et al . 1998 , jos and hernanz 1998 ) . classical novae synthesize many radioactive isotopes , which vary depending on the nova type ( which in turn depends on the type -co or one- of the underlying white dwarf ) . co novae produce mainly @xmath14be , whereas one produce @xmath17na and @xmath16al . other radioactivities with shorter lifetimes , such as @xmath53n and @xmath54f ( @xmath55s and 158min , respectively ) are produced in similar amounts in co and one novae ( see jos and hernanz , 1998 , jos , coc and hernanz , 1999 , hernanz et al . 1999 and hernanz et al . , these proceedings for details ) . typically , @xmath56m@xmath4 of @xmath53n , @xmath57m@xmath4 of @xmath54f are ejected in both co and one explosions . co novae also eject @xmath58m@xmath4 of @xmath14be , and one novae @xmath57m@xmath4 of @xmath17na and @xmath24m@xmath4 of @xmath16al . the reason of the different explosive nucleosynthesis results in co and one nova types is that some seed nuclei ( such as @xmath43ne , @xmath17ne , @xmath59 mg ) are necessary to synthesize @xmath17na and @xmath16al . that s because temperatures attained at the peak of the nova outburst are not high enough to break the cno cycle towards nena - mgal cycles . the two short - lived isotopes @xmath53n and @xmath54f are crucial for the prompt gamma - ray emission of novae , which is the most intense emission ( @xmath8 phot @xmath60 s@xmath2 ) , but of very short duration ( a few days ) and appearing before optical detection . the medium - lived isotopes ( @xmath14be and @xmath17na ) produce fluxes of around @xmath61 and @xmath9 phot @xmath60 s@xmath2 , for distances of 1 kpc . the prospects for detectability with the spi instrument onboard integral are analyzed in hernanz et al . ( these proceedings , and references therein ) . it is important to remind that , in addition to the @xmath14be and @xmath17na emission from individual novae , the cumulative emission from all the galactic novae can give important information about the distribution of the sources , specially if there is only one dominant source for that particular isotope . for @xmath17na , novae are the main individual contributors . therefore , the detection of galactic @xmath17na emission , and the corresponding 1275 kev emission map ( see jean et al . , 1999 , 2000 ) , would be a very valuable tool to study the distribution of novae in the galaxy , which is very poorly known from optical - uv and ir observations because of interstellar extinction . finally , @xmath16al is produced in one novae in such an amount that makes it quite improbable that novae contribute largely to the @xmath16al content of the galaxy , as observed through its emission at 1809 kev . research partially supported by the cicyt - p.n.i.e . ( esp98 - 1348 ) , by the dges ( pb98 - 1183-c03 - 02 and pb98 - 1183-c03 - 03 ) and by the aihf1999 - 0140 aufderheide m.b . , baron e. , thielemann f .- k . , 1991 , apj 370 , 630 arnett d. , 1969 , astrophys . space sci . 5 , 180 arnett d. , 1996 , supernovae and nucleosynthesis , princeton university press arnett d. , livne e. , 1994a , apj 427 , 315 arnett d. , livne e. , 1994b , apj 427 , 330 arnett d. , bahcall j.n . , kirshner r.p . , woosley s.e . , 1989 , 27 , 629 arnett d. , fryxell b. , mller e. , 1989 , apj 341 , l63 arnett d. , truran j.w . , woosley s.e . , 1971 , apj 165 , 87 arnould m. , meynet g. , 1997 , in 2@xmath62 integral workshop `` the transparent universe '' . esa sp-382 , noordwijk , p. 33 baade w. , zwicky f. , 1934 , phys . 45 , 138 bazan g. , arnett d. , 1994 , apj 433 , l41 bazan g. , arnett d. , 1998 , apj 496 , 316 benz w. , thielemann f .- k . , 1990 , apj 348 , l17 bethe h.a . , 1990 , rev . 62 , 801 bethe h.a . , 1995 , apj 449 , 714 bethe h.a . , wilson j.r . , 1985 , apj 295 , 14 bethe h.a . , brown g.e . , applegate j. , lattimer j. , 1979 , nucl . a324 , 487 branch d. , 2000 , pasp , in press ( astro - ph/0009337 ) bravo e. , garca - senz d. , 1999 , mnras 307 , 984 bravo e. , tornamb a.,domnguez i. , isern j. , 1996 , a&a 306 , 811 bravo e. , domnguez i. , isern j. , canal r. , hflich p. , labay j. , 1993 , a&a 269 , 187 burrows a. , 2000 , nature 403 , 727 burrows a. , the l.s . , 1990 , apj 360 , 626 burrows a. , hayes j. , fryxell b.a . , 1995 , apj 450 , 830 bussard r.w . , burrows a. , the l.s . , 1989 , apj 341 , 401 canal r. , isern j. , labay j. , 1990 , ann . astrophys . , 28 , 183 canal r. , isern j. , labay j. , 1992 , apj 398 , l49 canal r. , garca d. , isern j. , labay j. , 1990 , apj 356 , l51 clayton d.d . , 1981 , apj 244 , l97 clayton d.d . , hoyle f. , 1974 , apj 187 , l101 clayton d.d . , colgate s.a . , fishman g.j . , 1969 , apj 155 , 75 clayton d.d . , leising m.d . , the l.s : , johnson w.n . , kurfess , j.d . , 1992 , apj 399 , l141 colgate s.a . , mckee c. , 1969 , apj 157 , 623 colgate s.a . , white r.h . , 1966 , apj 143 , 626 cook w. r. , palmer d. m. , prince t. a. , schindler s. m. , starr c. h. , stone e.c . , 1988 , apj 334 , l87 diehl r. , hartmann d. , eds . , 1999 , astronomy with radioactivities , mpe report 274 diehl r. , timmes f.x . , 1998 , pasp 110 , 637 diehl et al . 1996 , a&a suppl . 120 , 321 diehl et al . 1995 , a&a 298 , 445 diehl et al . 1997 , 4th compton symp . , diehl et al . 1999 , astroph . lett . & commun . , 38 , 357 freedman d.z . , 1974 , phys . d 9 , 1389 fryer c.l . , in in cosmic explosions , eds . s.s.holt and w.w . zhang , aip conf . 522 ( new york ) , p. 113 fryxell b. , arnett d. , mller e. , 1991 , apj 367 , 619 garca - senz d. , bravo e. , woosley s.e . , 1999 , a&a 349 , 177 georgii r. , et al . 2000 , in 5@xmath63compton symposium , eds . mcconnell m.l . & ryan j.m . , aip 510 , new york , p. 49 glasner s.a . , livne e. , truran j.w . , 1997 , apj 475 , 754 gmez - gomar j. , isern j. , jean p. , 1998a , mnras 295 , 1 gmez - gomar j. , hernanz m. , jos j. , isern j. , 1998b , mnras 296 , 913 hansen c.j , wheeler j.c . , 1969 , space sci . 3 , 464 heger a. , langer n. , woosley s.e . , 2000 , apj 528 , 368 herant m. , benz w. , 1992 , apj 387 , 294 herant m. , benz w. , colgate s. , 1992 , apj 395 , 642 herant m. , benz w. , hix w.r . , fryer c.l . , colgate s.a . , 1992 , apj 435 , 339 hernanz m. , jos j. , coc a. , gmez gomar j. , isern j. , 1999 , apj 526 , l97 hernanz m. , jos j. , 2000 , in cosmic explosions , eds . s.s . holt and w.w . zhang , aip conf . 522 ( new york ) , p.339 hillebrandt w. , niemeyer j.c , 1997 , in thermonuclear supernovae , nato asi series , kluwer ( dordrecht ) , p. 337 hillebrandt w. , niemeyer j.c , 2000 , ann . , in press ( astro - ph/0006305 ) hflich p. , khokhlov a.m. , 1996 , apj 457 , 500 hflich p. , khokhlov a.m. , mller e. , 1994 , apjs 92 , 501 hflich p. , wheeler j.c . , khokhlov a.m. , 1998 , apj 492 , 228 hoffman r.d . , woosley s.e . , weaver t.a . , rauscher t. , thielemann , f .- k . , 1999 , apj 521 , 735 hoffman r.d . , woosley s.e . , weaver t.a . , timmes f.x . , eastman r. , hartmann d.h . , 1995 , in the gamma - ray sky with compton gro and sigma , eds . m. signore , p. salati and g. vedrenne , kluwer ( dordretch ) , p. 267 hoyle , fowler , 1960 apj 132 , 565 isern j. , canal r. , labay j. , 1991 , apj 372 , l83 iwamoto k. , brachwitz f , nomoto k. , kishimoto n. , umeda h. , hix w.r . , thielemann , f .- k . , 1999 , apj 125 , 439 iyudin a. , et al . , 1994 , a&a 284 , l1 jean p. , hernanz m. , gmez - gomar j. , jos j. , 1999 , astrophys . lett . & comm . , 38 , 433 jean p. , hernanz m. , gmez - gomar j. , jos j. , 2000 , mnras , in press janka h .- , mller e. , 1995 , apj 448 , l109 jos j. , hernanz m. , 1998 , apj 494 , 680 jos j. , coc a. , hernanz m. , 1999 , apj 520 , 347 kercek a. , hillebrandt w. , truran j.w . , 1998 , a&a 337 , 339 kercek a. , hillebrandt w. , truran j.w . , 1999 , a&a 345 , 831 khokhlov a.m. , 1991a , a&a 245 , 114 khokhlov a.m. , 1991b , a&a 245 , l25 khokhlov a.m. , oran e.s . , wheeler j.c . , 1997 , apj 478 , 678 kndlseder j. et al . , 1996a , a&a suppl . 120 , 327 kndlseder j. et al . , 1996b , a&a suppl . 120 , 335 kndlseder j. 1997 , phd thesis kndlseder j. 1999 , a&a 345 , 813 kovetz a. , prialnik d. , 1997 , apj 477 , 356 kumagai s. , nomoto k. , 1997 , in thermonuclear supernovae , nato asi series , kluwer ( dordrecht ) , p. 515 kurfess j.d . , et al . , 1992 , apj 399 , l140 leising m.d . , 1988 , nature 332 , 516 leising m.d . , clayton d.d . , 1987 , apj 323 , 157 leising m.d . , share g.h . , 1990 , apj 357 , 638 lisewski a.m. , hillebrandt w. , woosley s.e . , 2000 , apj 538 , 831 livio m. , 1999 , in type ia supernovae : theory and cosmology , cambride univ . press ( astro - ph/9903264 ) livne e. , 1990 , apj 354 , l53 livne e. , glasner a.s . , 1991 , apj 370 , 272 mahoney w.a . , varnell l.s . , jacobson a.s . , ling j.c . , radocinski r.g . , wheaton w.a . , 1988 , apj 334 , l81 matz s.m . , share g.h . , leising m.d . , chupp e.l . , vestrand w.t . , 1988 , nature 331 , 416 maxted p.f.l . , marsh t.l . , north r.c . , 2000 , mnras , in press ( astro - ph/0007257 ) mccray r. , 1993 , ann . 31 , 175 meynet g. , arnould m. , 1999 , astronomy with radioactivities , mpe report 274 , p. 85 meynet g. , maeder a. , 1997 , a&a 321 , 465 morris d.j . , et al . , 1995 , in 17@xmath63 texas symposium , eds . bhringer et al . , new york , p. 397 morris d.j . , et al . , 1997 , in 4@xmath63compton symposium , eds . dermer c.d . , strickman m.s . and kurfess j.d . , aip 410 , new york , p. 1084 nakamura t. , umeda h. , nomoto k. , thielemann f .- k . , burrows a. , 1999 , apj 517 , 193 niemeyer j.c . , 1999 , apj 523 , l57 niemeyer j.c . , truran , j.w . , 2000 , type ia supernovae : theory and cosmology , cambride univ . press niemeyer j.c . , woosley s.e . , 1997 , apj 475 , 740 nomoto k. , kondo y. , 1991 , apj 367 , l19 nomoto k. , thielemann f .- k . , yokoi k. , 1984 , apj 286 , 644 nomoto k. et al . , 1997 , in thermonuclear supernovae , nato asi series , kluwer ( dordrecht ) , p. 349 oberlack et al . 1996 , a&a suppl . 120 , 311 pinto p.a . , woosley s.e . , 1988 , nature 333 , 534 prantzos n. , diehl r. , 1996 , phys . rep . 267 , 1 prialnik d. , shara m.m . , shaviv g. , 1978 , a&a 62 , 339 prialnik d. , kovetz a. , 1995 , apj 445 , 789 rester a. c. , coldwell r.l . , dunnam f.e . , eichhorn g. , trombka j.i . , starr r. , lasche g.p . , 1989 , apj 342 , l71 ruiz - lapuente p. , canal r. , burkert a. , 1997 , in thermonuclear supernovae , nato asi series , kluwer ( dordrecht ) , p. 205 ruiz - lapuente p. , canal r. , isern j. , eds . , 1997 , thermonuclear supernovae , nato asi series , kluwer ( dordrecht ) sandie w. g. , nakano g. h. , chase l. f. jr . , fishman g.j . , meegan c. a. , wilson r. b. , paciesas w. s. , lasche g. p. , 1988 , apj 334 , l91 starrfield s. , 1989 , in classical novae , eds . bode and a. evans , john wiley ( chichester ) , p. 39 starrfield s. , truran j.w . , sparks w.m . , 1978 , apj 226 , 186 . starrfield s. , truran j.w . , wiescher m. , sparks w.m . , 1998 , mnras 296 , 502 suntzeff n.b . , phillips m.m . , elias j.h . , depoy d.l . , walker a.r . , 1992 , apj 384 , l33 teegarden b.j . , barthelmy s.d . , gehrels n. , tueller j. , leventhal m. , 1989 , nature 339 , 122 thielemann f .- k . , 1999 , in astronomy with radioactivities , mpe report 274 , p. 123 thielemann f .- k . , nomoto k. , hashimoto m. , 1996 , apj 460 , 408 thielemann f .- k . , nomoto k. , yokoi k. , 1986 , a&a 158 , 17 timmes f.x . , woosley s.e . , hartmann d.h . , hoffman r.d . , weaver t.a . , matteucci f. , 1995 , apj 449 , 204 timmes f.x . , woosley s.e . , hartmann d.h . , hoffman r.d . , 1996 , apj 464 , 332 tueller j. , barthelmy s.d . , gehrels n. , teegarden b.j . , leventhal m. , maccallum c.j . , 1990 , apj 351 , l41 wheeler j.c , harkness r.p . , 1990 , rep . phys . , 53 , 1467 wilson j.r . , 1971 , apj 163 , 209 woosley s.e . , 1997 , in thermonuclear supernovae , nato asi series , kluwer ( dordrecht ) , p. 313 woosley s.e . , hoffman r.d . , 1991 , apj 368 , l31 woosley s.e . , pinto p.a . , hartmann d. , 1989 , apj 346 , 395 woosley s.e . , weaver t.a . , 1994 , apj 423 , 371 woosley s.e . , weaver t.a . , 1994 , apjs 101 , 181

non - dominated sorting arranges a set of points in-dimensional euclidean space into layers by repeatedly removing the coordinatewise minimal elements . it was recently shown that nondominated sorting of random points has a hamilton - jacobi equation continuum limit . the obvious numerical scheme for this pde has a slow convergence rate of . in this paper , we introduce two new numerical schemes that have formal rates of and we prove the usual theoretical rates . we also present the results of numerical simulations illustrating the difference between the formal and theoretical rates .

1508.01557

in this paper , we introduce new finite difference schemes for the hamilton - jacobi equation @xmath4 and prove rates of convergence . the hamilton - jacobi equation appeared recently as the continuum limit of nondominated sorting , which is widely used in scientific and engineering contexts @xcite . let us briefly describe the connection . let @xmath5 be _ i.i.d . _ random variables on @xmath6 with continuous density @xmath7 . let @xmath8 denote the elements in @xmath9 that are coordinatewise minimal . the set @xmath8 is called the _ first pareto front _ of @xmath10 , and the elements of @xmath8 are called _ pareto optimal _ or _ nondominated_. the _ second pareto front _ , denoted @xmath11 , is the set of minimal elements from @xmath12 , and the @xmath13 _ pareto front _ is defined as @xmath14 the process of sorting the set @xmath10 into pareto fronts , or nondominated layers , is called _ nondominated sorting _ , and is widely used in multi - objective optimization ( see @xcite and references therein ) , with recent applications to machine learning @xcite . it turns out that nondominated sorting is equivalent to finding the longest chain in a partially ordered set , which has a long history in probability and combinatorics @xcite . it was shown in @xcite that the pareto fronts converge almost surely in the limit as @xmath15 to the level sets of the unique nondecreasing is _ nondecreasing _ if @xmath16 is nondecreasing for all @xmath17 . ] viscosity solution of . figure [ fig : demo ] gives an illustration of this continuum limit . in @xcite , a fast algorithm called _ pde - based ranking _ was proposed for approximate nondominated sorting of large datasets . the basic idea is to estimate the density @xmath7 from a ( relatively small ) subset of @xmath5 , and then use the numerical solution of the hamilton - jacobi equation as an approximation of nondominated sorting . it was shown in @xcite that pde - based ranking is substantially faster than nondominated sorting in relatively low dimensions , while maintaining high levels of sorting accuracy . the numerical scheme for used in @xcite is based on backward finite differences and can be solved very efficiently in a single pass . due to the fact that information flows along coordinate axes in the definition of nondominated sorting , this scheme is upwind ( or monotone ) , and convergence of the scheme was established in @xcite . although the scheme used in @xcite is simple and efficient , it suffers from poor accuracy , with formal rates of convergence on the order of @xmath1 for a grid with spacing @xmath18 in dimension @xmath0 . in this paper , we propose two new and highly efficient finite difference schemes for solving . both schemes have a formal accuracy of @xmath2 when the solution is smooth , and we prove the usual @xmath3 rates in the context of non - smooth viscosity solutions . these schemes can be used to increase the accuracy of pde - based ranking @xcite _ without _ increasing computational complexity . alternatively , with these highly accurate schemes we can afford to use a coarser grid resolution , and thus we can reduce the computational complexity of pde - based ranking while maintaining high accuracy . this is particularly important in applications of nondominated sorting @xcite , which will benefit from highly accurate and efficient algorithms for sorting massive datasets . we detail the new schemes and our main results in the next section . we pose the hamilton - jacobi equation on a compact domain as follows : @xmath19^n\\ u & = 0 & & \text{on } \gamma , \end{aligned}\right\}\ ] ] where @xmath20^n \setminus ( 0,1]^n$ ] and @xmath21 . notice we have modified by taking the positive parts of the partial derivatives . this is necessary to obtain existence of a viscosity solution of , and is a well - known issue with viscosity solutions on boundaries of domains ( see @xcite ) . we elaborate on this briefly in section [ sec : comparison ] . we should mention that there is no loss of generality in considering the domain @xmath22^n$ ] in . indeed , we can make a simple scaling argument to transform the domain of into @xmath23 $ ] for any @xmath24 . let @xmath18 , @xmath25 , and @xmath26 . for @xmath27 set @xmath28 . we recall the numerical scheme for from @xcite : @xmath29^n_h\\ u_h(x ) & = 0 & & \text{if } x \in \gamma_h , \end{aligned}\right\}\ ] ] where @xmath30^n_h \to { \mathbb{r}}$ ] is the numerical solution and @xmath31 the solution @xmath32 of can be solved efficiently in a single pass , which is reminiscent of the fast marching @xcite and fast sweeping @xcite algorithms . in dimension @xmath33 , the scheme is quadratic and can be solved in closed form @xmath34 in dimensions @xmath35 , the scheme can be solved via any iteration method , such as a bisection search . convergence of to the viscosity solution of was established in @xcite . while is optimal in terms of computational complexity on a fixed grid , its accuracy is at best @xmath1 . to see this , consider the special case of @xmath36 and @xmath37 . the solution @xmath38 is smooth on @xmath39^n$ ] , and has a gradient singularity on the boundary @xmath40 . we can use the comparison principle @xcite for to show that in general @xmath41^n } f^\frac{1}{n } \leq u(x ) \leq n(x_1\cdots x_n)^\frac{1}{n } \sup_{[0,1]^n } f^\frac{1}{n}.\ ] ] therefore , the gradient singularity on @xmath40 is typical for solutions of whenever @xmath42^n } f > 0 $ ] . let @xmath43 . by the concavity of @xmath44 @xmath45 on the other hand , if @xmath46 then @xmath47 therefore , for any @xmath48^n_h$ ] such that @xmath46 for some @xmath17 we have @xmath49 setting @xmath50 and invoking the comparison principle for yields @xmath51 letting @xmath52 we find that @xmath53 therefore , the scheme makes an error on the order of @xmath1 in the immediate vicinity of the boundary @xmath40 . since @xmath38 is generally not smooth , the best theoretical rate that one can prove in the context of viscosity solutions is typically strictly worse than the formal rate ( see @xcite ) . in section [ sec : numerics ] , we show numerical results indicating that the @xmath54 convergence rate of @xmath1 is typically observed in practice . since the slow convergence rate is caused by a singularity in the gradient of @xmath38 on @xmath40 , it is natural to look for a transformation of @xmath38 that removes this singularity . with this in mind , we set @xmath55 where @xmath38 is the nondecreasing viscosity solution of . when @xmath56 we have @xmath57 , which is lipschitz continuous ( in fact smooth ) on @xmath22^n$ ] . in general , we prove in lemma [ lem : vexun ] that @xmath58^n)$ ] whenever @xmath59^n)$ ] . furthermore , @xmath60 is a viscosity solution of @xmath61^n\\ v & = 0 & & \text{on } \gamma . \end{aligned}\right\}\ ] ] since @xmath60 is lipschitz continuous , it is reasonable to suspect that a numerical scheme for would have a better convergence rate than . we therefore propose the following finite difference scheme for : @xmath62^n_h\\ v_h(x ) & = 0 & & \text{if } x \in \gamma_h , \end{aligned}\right\}\ ] ] where @xmath63^n_h \to { \mathbb{r}}$ ] . we take @xmath64 to be the largest solution of at each @xmath48^n_h$ ] . it is interesting to note that when @xmath7 is constant , the solution of the scheme is @xmath65 , which is the _ exact _ solution of . the scheme can be solved efficiently in a single pass , similar to , and in dimension @xmath33 we have the closed form expression @xmath66 where @xmath67 in theorem [ thm : convpv ] , we prove convergence of when @xmath7 is continuous and nonnegative . our main result is the following convergence rate . [ thm : ratev ] suppose @xmath68^n)$ ] and @xmath69 . let @xmath70 be the solution of and let @xmath38 be the nondecreasing viscosity solution of . then @xmath71 and @xmath72_{1;[0,1]^n},\inf_{[0,1]^n}f\big)$ ] . notice that has a zeroth order term with the wrong sign for comparison principle arguments to hold . we can see this by observing that the method of vanishing viscosity takes the form @xmath73 since the standard proof of convergence rates for numerical approximations to viscosity solutions is based on the proof of the comparison principle @xcite , we can not directly apply these techniques to . our proof of theorem [ thm : ratev ] passes through an auxiliary problem , which actually suggests another numerical scheme for solving . based on our observation that @xmath37 is the viscosity solution of corresponding to @xmath36 , it is natural in more general settings to make the ansatz @xmath74 for some function @xmath75^n \to [ 0,\infty)$ ] . if this ansatz is correct , then @xmath76 would be a viscosity solution of @xmath77^n.\ ] ] it turns out that is well - posed within the class of bounded viscosity solutions _ without _ imposing a boundary condition . the boundary condition is actually encoded into the pde due to the fact that the term @xmath78 vanishes on @xmath79 . this suggests , for example , that we should expect @xmath80 . due to the degeneracy of the terms @xmath81 , there are in general infinitely many unbounded viscosity solutions of . we characterize @xmath76 from as the maximal bounded viscosity solution of , and we show in lemma [ lem : wlip ] that @xmath82^n)$ ] whenever @xmath59^n)$ ] . we propose the following numerical scheme for : @xmath83^n_h,\ ] ] where @xmath84^n_h \to [ 0,\infty)$ ] . here , we take @xmath85 to be the largest solution of at each @xmath86^n_h$ ] . we note that for @xmath86^n_h$ ] such that @xmath87 , the quantity @xmath88 is undefined . since this term appears in the form @xmath89 , its value is not used in the scheme . it is interesting to note that when @xmath7 is constant , the scheme gives the _ exact _ solution of . this scheme can be solved efficiently in a single pass , and in dimension @xmath33 the scheme can be solved in closed form @xmath90 where @xmath91 and @xmath92 since the zeroth order term in has the correct sign , we can use a modification of the standard convergence proof to establish the following convergence rate . [ thm : rate1 ] suppose that @xmath68^n)$ ] and @xmath93 . let @xmath94 be the solution of , and let @xmath76 be the maximal bounded viscosity solution of . then latexmath:[\[\label{eq : rate1 } where @xmath72_{1;[0,1]^n } , \inf_{[0,1]^n}f\big)$ ] . our proof of theorem [ thm : ratev ] proceeds by first showing ( in lemma [ lem : changeofvar ] ) that @xmath96 and then invoking theorem [ thm : rate1 ] . although both and have the same provable convergence rates , our numerical results presented in section [ sec : numerics ] suggest that in general has a better experimental convergence rate than . this can be explained by observing that @xmath38 can only have gradient singularities when transitioning from zero to a positive value . the transformation @xmath97 regularizes these gradient singularities _ anywhere _ in the domain @xmath22^n$ ] , and not just on the boundary @xmath40 . on the other hand , the transformation @xmath98 is designed only to capture singularities on the boundary @xmath40 . in section [ sec : numerics ] , we give an example of a discontinuous function @xmath7 for which exhibits a better convergence rate than for the reason outlined above . this paper is organized as follows . in section [ sec : comparison ] , we prove comparison principles for viscosity solutions of and . in section [ sec : conv ] , we use the barles - souganidis framework @xcite to prove convergence of the schemes and under the assumption that @xmath7 is continuous and nonnegative . in section [ sec : rates ] we prove theorems [ thm : ratev ] and [ thm : rate1 ] establishing rates of convergence for and when @xmath7 is positive and lipschitz . the proofs of the convergence rates require lipschitz estimates for the viscosity solutions of and . these are obtained in section [ sec : s3lip ] . in section [ sec : numerics ] , we show the results of numerical simulations comparing all three schemes . we first prove comparison principles for that will be utilized later in the convergence proofs . let us first briefly comment on the differences between and . as we mentioned in section [ sec : main ] , it is necessary to modify by taking the positive parts of @xmath99 when posing the pde on compact domains . to see why this is necessary , consider with @xmath36 in dimension @xmath33 without this modification : @xmath100 ^ 2\\ u & = 0 & & \text{on } \gamma . \end{aligned}\right\}\ ] ] this hamilton - jacobi equation of course has a classical solution @xmath101 that is smooth on @xmath39 ^ 2 $ ] . however , @xmath38 is not a viscosity solution of . to see this , let @xmath102 . then @xmath103 has a local maximum at @xmath104 relative to @xmath39 ^ 2 $ ] for every @xmath105 . since @xmath106 for @xmath107 , the viscosity subsolution property fails to hold at @xmath104 this is a well - known issue with viscosity solutions on boundaries of domains ( see @xcite ) . notice , however , that @xmath108 taking the positive parts of @xmath99 in gives the pde a useful monotonicity property that we will exploit in this paper . since it is useful to abstract this property , we make the following definition . [ def : directed ] let @xmath109 . we say @xmath110^n\times { \mathbb{r}}\times { \mathbb{r}}^n \to { \mathbb{r}}$ ] is _ directed _ if for all @xmath111^n\times{\mathbb{r}}$ ] @xmath112 if @xmath113 is directed , it is simple to construct a monotone ( or upwind ) numerical scheme using backward difference quotients . indeed , let us consider the scheme @xmath114^n_h,\ ] ] where @xmath115^n_h \to { \mathbb{r}}$ ] and @xmath116 to see that is monotone , suppose that @xmath117 and @xmath118 for @xmath119^n_h$ ] . then @xmath120 for all @xmath121 , and since @xmath113 is directed @xmath122 the following theorem is a direct consequence of this monotonicity and ( * ? ? ? * theorem 2.1 ) . [ thm : bs - conv ] suppose that @xmath113 is continuous and directed . let @xmath123 be solutions of satisfying @xmath124^n_h } } |u_h(x)| < \infty.\ ] ] then @xmath125 is a viscosity subsolution ( resp . supersolution ) of @xmath126 in @xmath127^n$ ] . when @xmath128 , there is a unique nondecreasing viscosity solution @xmath38 of @xcite . we now show that the restriction of @xmath38 to @xmath22^n$ ] is the unique nondecreasing viscosity solution of . this establishes the equivalence of and . [ lem : comp ] suppose @xmath129^n)$ ] is nonnegative . let @xmath130^n)$ ] be a viscosity subsolution of , and let @xmath131^n)$ ] be a nondecreasing viscosity supersolution of . then @xmath132 on @xmath22^n$ ] . let @xmath133 and set @xmath134 . since @xmath60 is nondecreasing , @xmath135 is a viscosity solution of @xmath136^n.\ ] ] the standard comparison argument based on doubling the variables ( see @xcite ) shows that @xmath137 on @xmath22^n$ ] . sending @xmath138 completes the proof . [ lem : equivalence ] let @xmath128 be nonnegative and let @xmath38 be the nondecreasing viscosity solution of . then the restriction of @xmath38 to @xmath22^n$ ] is the nondecreasing viscosity solution of . by lemma ( * ? * lemma 3.3 ) , the numerical solutions @xmath32 of satisfy the estimate @xmath139 for all @xmath140^n_h$ ] , where @xmath141^n}f)$ ] . combined with theorem [ thm : bs - conv ] , this shows that @xmath32 converges uniformly on @xmath22^n$ ] to the unique nondecreasing viscosity solution of . by ( * theorem 3.4 ) , we also have @xmath142 uniformly on @xmath22^n$ ] , which completes the proof . in this section , we establish a comparison principle for . when @xmath129^n)$ ] , the function @xmath55 is a nondecreasing viscosity solution of ( see lemma [ lem : nonuniq ] ) . we will call @xmath97 the _ maximal _ viscosity solution of ( see lemma [ lem : vmaximal ] ) . since has a zeroth order term of the wrong sign for comparison to hold directly , we find that actually has infinitely many nondecreasing viscosity solutions . [ lem : nonuniq ] let @xmath129^n)$ ] be nonnegative , let @xmath143^n$ ] , and let @xmath38 be the nondecreasing viscosity solution of @xmath144\\ u & = 0 & & \text{on } \prod_{i=1}^n [ y_i,1]\setminus \prod_{i=1}^n ( y_i,1 ] , \end{aligned}\right\}\ ] ] and extend @xmath38 to @xmath22^n$ ] by setting @xmath145 for @xmath146^n \setminus\prod_{i=1}^n [ y_i,1]$ ] . then @xmath55 is a nondecreasing viscosity solution of . let @xmath147^n$ ] and let @xmath148^n)$ ] such that @xmath149 has a local maximum at @xmath150 . we also assume that @xmath151 . if @xmath152 then since @xmath38 is nondecreasing , we see that @xmath153 for all @xmath154 that are coordinatewise less than @xmath150 . it follows that @xmath155 for all @xmath17 , and the subsolution property is trivially satisfied . if @xmath156 , then @xmath157 and therefore @xmath158 $ ] . setting @xmath159 we find that @xmath160 has a local maximum at @xmath150 . therefore @xmath161 since @xmath162 , this becomes @xmath163 which verifies the subsolution property . the proof of the supersolution property is similar . the lack of uniqueness of nondecreasing viscosity solutions of indicates that we can not expect a comparison principle to hold for arbitrary sub- and supersolutions of . however , we show in the following lemma that every subsolution is bounded above by @xmath97 . this turns out to be sufficient to prove convergence of . [ lem : vmaximal ] let @xmath129^n)$ ] be nonnegative and let @xmath164^n)$ ] be a nonnegative viscosity subsolution of . then @xmath165 on @xmath22^n$ ] , where @xmath38 is the nondecreasing viscosity solution of . we define @xmath166 . let @xmath147^n$ ] and @xmath148^n)$ ] such that @xmath167 has a local maximum at @xmath150 and @xmath168 . letting @xmath169 , we see that @xmath170 has a local maximum at @xmath150 and therefore @xmath171 since @xmath172 , we have @xmath173 if @xmath156 then @xmath174 if @xmath152 , then since @xmath60 is nonnegative , @xmath175 for all @xmath17 , which verifies . by lemma [ lem : comp ] , @xmath176 , where @xmath38 is the unique nondecreasing viscosity solution of . before establishing a comparison principle for , let us comment on the properties of solutions of . let @xmath129^n)$ ] be nonnegative and let @xmath38 be the nondecreasing viscosity solution of . by lemma [ lem : comp ] we have @xmath177^n}f^\frac{1}{n } \ \ \text { for all } x \in ( 0,1]^n.\ ] ] setting @xmath178^n)$ ] we have @xmath179^n}f^\frac{1}{n } \ \ \text { for all } x \in ( 0,1]^n.\ ] ] we also have that @xmath76 is a viscosity solution of that satisfies @xmath180^n\ ] ] in the viscosity sense for all @xmath17 . to see this , let @xmath147^n$ ] and @xmath181^n)$ ] such that @xmath182 has a local minimum at @xmath150 . we can also assume @xmath183 , so that @xmath184 in a neighborhood of @xmath150 . setting @xmath185 it follows that @xmath186 in a neighborhood of @xmath150 and @xmath187 . therefore @xmath188 has a local minimum at @xmath150 . since @xmath38 is nondecreasing , @xmath189 for all @xmath17 and @xmath190 a simple computation shows that @xmath191 and @xmath192 the subsolution property is verified similarly . we will call @xmath76 the _ maximal bounded _ viscosity solution of . notice the boundary condition @xmath193 on @xmath40 is only used to show that @xmath76 is bounded . indeed , it is clear that the argument above holds when @xmath38 is _ any _ viscosity solution of @xmath194^n.\ ] ] this yields an infinite number of unbounded viscosity solutions of . for instance , when @xmath36 the function @xmath195 for any @xmath196 is a viscosity solution of . taking @xmath197 gives the bounded viscosity solution of interest from . the following theorem characterizes this solution as the unique bounded viscosity solution of satisfying . [ thm : comparison - w ] assume that @xmath129^n)$ ] is nonnegative . let @xmath198^n)$ ] and @xmath199^n)$ ] be bounded viscosity sub- and supersolutions of , respectively , and suppose that @xmath200 satisfies in the viscosity sense for all @xmath17 . then @xmath201 on @xmath39^n$ ] . for @xmath202 , we define @xmath203^n\\ 0 , & \text{if } x \in \gamma.\end{cases}\ ] ] since @xmath198^n)$ ] is bounded , @xmath204^n)$ ] . similarly , @xmath205^n)$ ] . we first show that @xmath206 is a nondecreasing viscosity supersolution of . let @xmath207^n$ ] and @xmath181^n)$ ] such that @xmath208 has a local minimum at @xmath209 and @xmath210 . we define @xmath211^n.\ ] ] then it follows that @xmath212 has a local minimum at @xmath209 . since @xmath200 is a viscosity supersolution of we have @xmath213 since @xmath214 and @xmath200 satisfies we have @xmath215 therefore @xmath216 for all @xmath17 and @xmath217 this establishes that @xmath206 is a nondecreasing ( e.g. , see ( * ? ? ? * lemma 5.17 ) ) viscosity supersolution of . we can similarly show that @xmath218 is a viscosity subsolution of . the proof is completed by invoking lemma [ lem : comp ] . in this section we prove convergence of the schemes and under the assumption that @xmath7 is continuous and nonnegative . let @xmath18 and @xmath48^n_h$ ] . given values for @xmath219 , we define @xmath64 to be the largest solution of . if we let @xmath220 for @xmath221 and @xmath222 , then this is equivalent to finding the largest solution @xmath223 of @xmath224 we define @xmath225 since @xmath226 is always a solution of , it is easy to see that @xmath227 is a nonnegative real number for all @xmath228 . with these definitions , the solution @xmath70 of satisfies @xmath229 for @xmath230 and @xmath231^n_h.\ ] ] we shall refer to @xmath70 as the _ maximal _ solution of . we now establish some important properties of @xmath10 . [ lem : scheme ] let @xmath228 . then 1 . @xmath232 and @xmath233 if and only if @xmath234 , 2 . @xmath235 for all @xmath236 , 3 . if @xmath237 then @xmath238 whenever @xmath239 , and 4 . @xmath240 is nondecreasing in all variables . by symmetry , we may assume that @xmath241 . for ( i ) , we simply note that @xmath242 and @xmath243 . therefore there exists @xmath244 such that @xmath245 . it follows that @xmath246 . if @xmath247 , then clearly @xmath248 . conversely , suppose that @xmath237 . if @xmath249 then @xmath250 . if @xmath251 , then @xmath252 and hence @xmath253 . for ( ii ) if @xmath234 , then @xmath248 , and it is clear that @xmath254 for all @xmath255 . if @xmath237 , then by ( i ) , @xmath253 and @xmath256 . for any @xmath257 such that @xmath245 we have @xmath258 it follows that @xmath235 for all @xmath255 . this establishes ( ii ) . for ( iii ) , suppose first that @xmath249 . then @xmath259 and @xmath260 , from which ( iii ) immediately follows . if @xmath261 , then since @xmath237 we have @xmath262 for all @xmath263 . therefore there exists @xmath264 such that @xmath262 for @xmath265 . define @xmath266 clearly @xmath267 and @xmath268 . for any @xmath269 satisfying @xmath245 , we have by that @xmath270 . it follows that @xmath271 , which establishes ( iii ) . for ( iv ) we set @xmath272 by ( i ) , @xmath253 for every @xmath273 . therefore @xmath274 for all @xmath273 , and it follows from the implicit function theorem that the restriction of @xmath10 to @xmath275 is smooth . since @xmath256 we have @xmath276 differentiating in @xmath277 for @xmath278 we obtain @xmath279 since @xmath280 for all @xmath281 , we have @xmath282 it follows that @xmath283 . differentiating in @xmath284 we have @xmath285 as before , it follows that @xmath286 . therefore @xmath10 is strictly increasing on @xmath275 . if @xmath234 , then @xmath233 is nondecreasing . if @xmath287 , then @xmath260 is again nondecreasing . the continuity of @xmath10 establishes ( iv ) . [ rem : lips ] in the proof of lemma [ lem : scheme ] ( iv ) , we can use the inequality @xmath288 to find that @xmath289 for all @xmath273 and @xmath278 . since @xmath290 , we have the bound @xmath291 using the properties of @xmath10 from lemma [ lem : scheme ] we can establish a comparison principle for the scheme . [ thm : svcomp ] let @xmath18 and suppose @xmath292 . let @xmath293 be a subsolution of and let @xmath294 be a supersolution of satisfying @xmath295^n_h.\ ] ] then @xmath296 on @xmath22^n_h$ ] . we will prove the result by induction . we have @xmath297 for @xmath298 by definition . now let @xmath48^n_h$ ] and suppose that @xmath299 since @xmath293 is a subsolution of we have @xmath300 it follows from lemma [ lem : scheme ] ( ii ) that @xmath301 recalling and lemma [ lem : scheme ] ( iv ) we have @xmath302 the proof is completed by induction . [ rem : poscomp ] if @xmath7 is positive and @xmath60 is a supersolution of that is positive on @xmath39^n_h$ ] , then it follows from lemma [ lem : scheme ] ( iii ) that @xmath303^n_h.\ ] ] [ rem : s2exact ] notice that @xmath304^n}f \ \ \text { and } \ \ \underline{v}(x ) = ( x_1\cdots x_n ) \inf_{[0,1]^n}f\ ] ] are super- and subsolutions of , respectively . by theorem [ thm : svcomp ] and remark [ rem : poscomp ] @xmath305^n } f \leq v_h(x ) \leq ( x_1\cdots x_n ) \sup_{[0,1]^n}f \ \ \text { for all } x\in [ 0,1]_h^n.\ ] ] as a consequence , if @xmath306 then @xmath307 , which is _ exactly _ equal to the maximal viscosity solution of . we can now establish convergence of the scheme . [ thm : convpv ] let @xmath129^n)$ ] be nonnegative , and for each @xmath18 let @xmath70 denote the maximal solution of . then @xmath308 uniformly on @xmath22^n$ ] as @xmath309 , where @xmath60 is the maximal viscosity solution of . let @xmath310 by theorem [ thm : bs - conv ] and remark [ rem : s2exact ] , @xmath311^n)$ ] is a viscosity subsolution of . by lemma [ lem : vmaximal ] , @xmath312 . let @xmath32 be the solution of , and let @xmath313 . since @xmath314 is convex for @xmath105 and @xmath32 is nondecreasing , we have @xmath315 therefore @xmath316^n_h.\ ] ] by theorem [ thm : svcomp ] , @xmath317 on @xmath22^n_h$ ] . as in the proof of lemma [ lem : equivalence ] , we have that @xmath318 uniformly on @xmath22^n$ ] as @xmath319 . it follows that @xmath320 , which completes the proof . [ rem : onesided ] notice in the proof of theorem [ thm : convpv ] we showed that @xmath321 , where @xmath32 is the numerical solution of , and @xmath70 is the solution of . under the assumptions of theorem [ thm : ratev ] , this gives a one - sided convergence rate for of the form @xmath322 where @xmath38 is the nondecreasing viscosity solution of . when @xmath38 is concave , we can actually prove that @xmath323 ( see ( * ? ? ? * lemma 3.5 ) ) . recall that we defined the solution @xmath84^n_h\to { \mathbb{r}}$ ] of inductively by taking @xmath85 to be the largest solution of the polynomial equation defining at each @xmath154 . it is easy to see that @xmath324^n_h \ \text { and } \ i \in \{1,\dots , n\}.\ ] ] the following lemma shows that admits a comparison principle whenever the supersolution satisfies . [ lem : s3comp ] suppose @xmath7 is nonnegative . let @xmath325 and @xmath200 be sub- and supersolutions of , respectively , and suppose @xmath200 satisfies . then @xmath326 on @xmath22^n_h$ ] . let @xmath86^n_h$ ] be a maximum of @xmath327 , and suppose to the contrary that @xmath328 . then we have @xmath329 recalling we have @xmath330 which is a contradiction . we now prove convergence of . [ thm : convs3 ] let @xmath129^n)$ ] be nonnegative , and for @xmath18 let @xmath94 be the maximal solution of . then @xmath331 uniformly on @xmath39^n$ ] as @xmath309 , where @xmath76 is the maximal bounded viscosity solution of . let @xmath332 we can use the comparison principle from lemma [ lem : s3comp ] to show that @xmath333^n}f^\frac{1}{n } \leq w_h \leq \sup_{[0,1]^n}f^\frac{1}{n}.\ ] ] therefore @xmath333^n}f^\frac{1}{n } \leq \underline{w } \leq { { \overline{w } } } \leq \sup_{[0,1]^n}f^\frac{1}{n}.\ ] ] by theorem [ thm : bs - conv ] , @xmath334^n)$ ] is a bounded viscosity subsolution of and @xmath335^n)$ ] is a bounded viscosity supersolution of . furthermore , since @xmath94 satisfies another application of theorem [ thm : bs - conv ] shows that @xmath336 is a viscosity solution of for all @xmath17 . by theorem [ thm : comparison - w ] , @xmath337 , where @xmath76 is the maximal bounded viscosity solution of in order to obtain a rate of convergence , it is necessary to show that the either the numerical solution or the viscosity solution of the continuum equation is lipschitz continuous up to the boundary @xmath40 . since none of the hamiltonians considered in this paper are coercive , the standard textbook estimates @xcite do not apply . in dimension @xmath33 , we can prove the required lipschitz estimate for @xmath60 directly by differentiating and using a comparison principle . let us briefly sketch the argument , which is made rigorous at the level of the numerical scheme in theorem [ thm : vlipd2 ] . formally differentiating in the variable @xmath338 we have @xmath339 setting @xmath340 and noting that @xmath341 ^ 2 } f$ ] we have @xmath342_{1;[0,1]^2}\sup_{[0,1]^2 } f.\ ] ] consider using a comparison function of the form @xmath343 . since @xmath344 , the first term in can be ignored . furthermore , for @xmath345 ^ 2}f$ ] , the sign of the zeroth order term in is positive , which suggests that a comparison principle should hold , allowing us to show that @xmath346 for a possibly larger constant @xmath347 . of course , @xmath60 is not in general smooth enough to use this argument directly , so we need to regularize . one obvious approach would be the method of vanishing viscosity . however , there are some technical challenges with this approach , due to the sign of the zeroth order term in and the ( non - obvious ) fact that for positive viscosity , @xmath348 can fail to be nondecreasing . our approach is to apply the argument above at the level of the numerical scheme . since we showed convergence of in theorem [ thm : convpv ] , a lipschitz estimate on @xmath70 directly carries over to the maximal viscosity solution @xmath60 of . heading in this direction , we now recall some basic properties of finite differences . [ prop : finite - diff - properties ] for all @xmath349 we have 1 . @xmath350 and @xmath351 , 2 . @xmath352 , 3 . @xmath353 , we now give a preliminary regularity result for @xmath70 . [ thm : vlipd2 ] let @xmath33 , let @xmath68 ^ 2)$ ] be nonnegative , and let @xmath70 be any nondecreasing solution of . then for @xmath354 we have @xmath355_{1;[0,1]^2 } + 3\sup_{[0,1]^2}f \ \ \text { for all } x \in ( 0,1]^2_h.\ ] ] let us prove the result for @xmath356 ; the case of @xmath357 follows by symmetry . we extend @xmath70 to a function on @xmath358 ^ 2_h$ ] by setting @xmath229 for @xmath359 ^ 2_h\setminus [ 0,1]^2_h$ ] . with these definitions , we have @xmath360 for @xmath298 . since @xmath70 is nondecreasing , we therefore have @xmath361 ^ 2_h.\ ] ] we now apply the operator @xmath362 to the equation above and use proposition [ prop : finite - diff - properties ] ( ii ) to obtain @xmath363 for all @xmath48 ^ 2_h$ ] . setting @xmath364 and using proposition [ prop : finite - diff - properties ] ( i ) we have @xmath365 applying theorem [ thm : svcomp ] and remark [ rem : s2exact ] @xmath366_{1;[0,1]^2}\sup_{[0,1]^2}f=:m\ ] ] for all @xmath48 ^ 2_h$ ] , and @xmath367 for @xmath368 . we can assume @xmath369 , otherwise @xmath370 and is trivial . let @xmath371 ^ 2}f\right)(1 + x_2)$ ] and note that since @xmath70 is nondecreasing we have @xmath372 ^ 2}f.\ ] ] we claim that @xmath373 . assume by way of contradiction that @xmath374 ^ 2_h}(\psi - { { \overline{\psi } } } ) > 0 $ ] . since @xmath375 on @xmath376 , @xmath377 must attain its positive maximum at some @xmath48 ^ 2_h$ ] . it follows that @xmath378 for @xmath202 and @xmath379 . since @xmath380 and @xmath70 is nondecreasing , we can combine and to find that @xmath381 ^ 2}f \leq m,\ ] ] which is a contradiction to the positivity of @xmath382 . by cauchy s inequality @xmath383 ^ 2}f\right)(1 + x_2 ) \leq [ f]_{1;[0,1]^2 } + 3\sup_{[0,1]^2}f,\ ] ] for all @xmath48 ^ 2_h$ ] . since @xmath308 uniformly ( theorem [ thm : convpv ] ) , we can extend the lipschitz estimate in theorem [ thm : vlipd2 ] to . [ cor : vlipd2 ] let @xmath33 , let @xmath68 ^ 2)$ ] be nonnegative , and let @xmath60 be the maximal viscosity solution of . then there exists @xmath347 such that @xmath384_{1;[0,1]^2 } \leq c \|f\|_{c^{0,1}([0,1]^n)}.\ ] ] the proof of theorem [ thm : vlipd2 ] does not work in dimensions @xmath35 due to the additional nonlinearity of , which increases with @xmath0 . in section [ sec : second ] we study the auxiliary problem , which allows us to prove lipschitz regularity of the maximal viscosity solution of in arbitrary dimension ( see lemma [ lem : vexun ] ) , provided the strictly stronger condition @xmath59^n)$ ] holds . we expect theorem [ thm : vlipd2 ] to hold in dimensions @xmath35 under the natural condition @xmath68^n)$ ] , but we currently do not know how to prove this . in this section we prove lipschitz regularity and rates of convergence for the scheme . this establishes one of our main results ( theorem [ thm : rate1 ] ) . these results are used in section [ sec : ratess2 ] to prove similar rates of convergence for . throughout this section we set @xmath385 for @xmath386 . let us first give a formal argument suggesting a lipschitz estimate on the maximal bounded viscosity solution @xmath76 of . assuming @xmath76 is smooth and @xmath93 , we can take the logarithm of to obtain @xmath387 differentiating both sides in the variable @xmath277 and setting @xmath388 we have @xmath389 where @xmath390 if @xmath391 and @xmath392 if @xmath393 . notice that at a positive maximum of @xmath44 we have @xmath394 for all @xmath395 and hence @xmath396 where we employed the inequality of arithmetic and geometric means in the first inequality . this suggests that @xmath397^n } \frac{1}{n } f^{1-\frac{1}{n}}f_{x_i } = \sup_{[0,1]^n } \partial_{x_i } ( f^\frac{1}{n}).\ ] ] in theorem [ thm : lip ] and lemma [ lem : wlip ] , we make this argument rigorous by applying it to the numerical scheme . one obvious gap in the argument above is the existence of a positive maximum for @xmath398 . indeed , when @xmath76 is any of the infinitely many unbounded viscosity solutions of , @xmath398 either has no maximum value , or no minimum value , on @xmath39^n$ ] . this is a reflection of the fact that has no boundary condition . this issue is resolved by showing ( in proposition [ prop : extension ] ) that the numerical solution @xmath94 can be extended by projection to a solution of on the domain @xmath399^n_h$ ] . for @xmath400 , let us set @xmath401 [ prop : extension ] suppose @xmath7 is nonnegative , and let @xmath76 be a sub- ( resp . super- ) solution of . then @xmath402 is a sub- ( resp . super- ) solution of @xmath403^n_h,\ ] ] where @xmath404 . we claim that @xmath405^n_h , \ , i \in \{1,\dots , n\}.\ ] ] if @xmath406 the result is trivial . hence we may assume that @xmath407 , so that @xmath408 . then @xmath409 which establishes the claim . for @xmath410^n_h\setminus [ 0,1]^n_h$ ] , we have @xmath411^n_h$ ] and @xmath412 since @xmath413 , the result immediately follows . we also note that we can extend the product rule for finite differences in proposition [ prop : finite - diff - properties ] ( iii ) by induction as follows : [ lem : prod - rule ] for all @xmath414 and @xmath415 we have @xmath416 where the final summation is over all @xmath417 with @xmath418 . the proof of lemma [ lem : prod - rule ] is deferred to the appendix . we now establish a lipschitz estimate on the solution @xmath94 of . [ thm : lip ] suppose that @xmath68^n)$ ] is positive on @xmath22^n$ ] , and let @xmath94 be the maximal solution of . then @xmath419 for all @xmath420^n_h \ , : \ , x_k \leq 1-h\}$ ] . in particular @xmath421_{1;[0,1]^n}\sup_{[0,1]^n}f^{\frac{1}{n}-1}.\ ] ] for notational convenience , we will write @xmath76 in place of @xmath94 throughout the proof . by proposition , we can extend @xmath76 and @xmath7 to functions on @xmath399^n_h$ ] by setting @xmath422 and @xmath423 , and we have @xmath424^n_h.\ ] ] let @xmath425 for all @xmath410^n_h$ ] with @xmath426 . the proof is split into two steps . we first show that @xmath427 we may assume @xmath44 is positive somewhere , otherwise trivially holds . for any @xmath154 such that @xmath428 , we have @xmath429 . this implies that @xmath430 . likewise , whenever @xmath431 , @xmath432 and we have @xmath433 . it follows that @xmath44 attains its positive maximum value at some @xmath434 . therefore @xmath435 since @xmath436 , @xmath437 where @xmath438 by lemma [ lem : prod - rule ] @xmath439 using proposition [ prop : finite - diff - properties ] ( ii ) we have @xmath440 for @xmath86^n_h$ ] . invoking proposition [ prop : finite - diff - properties ] ( i ) and noting that @xmath441 , we deduce @xmath442 inserting into yields @xmath443 for all @xmath395 . combining this with and recalling that @xmath444 gives @xmath445 using the inequality of geometric and arithmetic means yields @xmath446 combining and establishes . we now show that @xmath447 we may assume that @xmath44 assumes negative values , otherwise is trivial . as in the first part of the proof , @xmath44 attains its negative minimum at some @xmath448 . therefore @xmath449 by , @xmath450 for all @xmath395 . combining this with lemma [ lem : prod - rule ] yields @xmath451 this completes the proof . [ rem : wlip ] since @xmath331 uniformly on @xmath22^n$ ] , it is an easy consequence of theorem [ thm : lip ] that @xmath82^n)$ ] and @xmath452_{1;[0,1]^n } \leq \frac{1}{\sqrt{n } } [ f]_{1;[0,1]^n}\sup_{[0,1]^n } f^{\frac{1-n}{n}},\ ] ] whenever @xmath68^n)$ ] is positive . a sharper result is contained in the following lemma . [ lem : wlip ] let @xmath7 be a nonnegative function for which @xmath59^n)$ ] , and let @xmath76 be the maximal bounded viscosity solution of . then @xmath82^n)$ ] and @xmath453_{1;[0,1]^n } \leq \sqrt{n}[f^\frac{1}{n}]_{1;[0,1]^n}.\ ] ] we first assume that @xmath454^n)$ ] and @xmath93 . for @xmath18 , let @xmath94 be the maximal solution of . since @xmath454^n)$ ] , there exists a constant @xmath347 such that @xmath455 for all @xmath17 and all @xmath456^n_h \ , : \ , x_i \leq 1-h\}$ ] . by theorem [ thm : lip ] @xmath457_{1;[0,1]^n } + ch.\end{aligned}\ ] ] the opposite inequality is obtained similarly , and we find that @xmath458_{1;[0,1]^n } + ch,\ ] ] for all @xmath17 and all @xmath459 . the estimate follows from the uniform convergence @xmath331 as @xmath319 . we now suppose that @xmath7 is nonnegative and @xmath59^n)$ ] . extend @xmath7 to a function on @xmath460 by setting @xmath461 for @xmath462^n$ ] , where @xmath463^n$ ] is the closest point projection . since @xmath464 is 1-lipschitz , @xmath465_{1,{\mathbb{r}}^n } = [ f^\frac{1}{n}]_{1,[0,1]^n}$ ] let @xmath466 denote the standard mollifier , and define @xmath467 it is easy to verify that @xmath468^n)$ ] , @xmath469 uniformly on @xmath22^n$ ] as @xmath470 , and @xmath471_{1;[0,1]^n } \leq [ f^\frac{1}{n}]_{1;[0,1]^n}.\ ] ] let @xmath472 denote the maximal bounded viscosity solution of corresponding to @xmath473 . since @xmath473 is smooth and positive , the argument above yields @xmath474_{1;[0,1]^n } \leq \sqrt{n}[(f^{\varepsilon})^\frac{1}{n}]_{1;[0,1]^n } \stackrel{\eqref{eq : moll}}{\leq}\sqrt{n } [ f^\frac{1}{n}]_{1;[0,1]^n}.\ ] ] by the comparison principle for ( theorem [ thm : comparison - w ] ) and standard results on viscosity solutions @xcite , we have @xmath475 uniformly on @xmath22^n$ ] as @xmath470 . therefore @xmath476_{1;[0,1]^n } \leq \sqrt{n } [ f^\frac{1}{n}]_{1;[0,1]^n},\ ] ] which completes the proof . [ rem : tight ] the regularity result in lemma [ lem : wlip ] is tight , and can not be significantly generalized . for intance , if @xmath477 , then @xmath7 is smooth , but @xmath478 is not lipschitz on @xmath22^n$ ] . the corresponding solution of is @xmath479 , which also fails to be lipschitz on @xmath22^n$ ] . note that the solution of is @xmath480 , which is smooth on @xmath22^n$ ] . we can now extend lipschitz regularity to viscosity solutions of . [ lem : vexun ] suppose @xmath7 is a nonnegative function for which @xmath59^n)$ ] , and let @xmath60 be the maximal viscosity solution of . then @xmath58^n)$ ] and @xmath481_{1;[0,1]^n } \leq c \|f^\frac{n-1}{n}\|_{l^\infty([0,1]^n)}\|f^\frac{1}{n}\|_{c^{0,1}([0,1]^n)}.\ ] ] notice that we can write @xmath482^n,\ ] ] where @xmath76 is the maximal bounded viscosity solution of . therefore @xmath483_{1;[0,1]^n } \leq [ w^n]_{1;[0,1]^n } + [ x_1\cdots x_n]_{1;[0,1]^n}\sup_{[0,1]^n } f.\ ] ] the estimate follows from lemma [ lem : wlip ] and the inequality @xmath484_{1;[0,1]^n } \leq n\|w\|^{n-1}_{l^\infty([0,1]^n)}[w]_{1;[0,1]^n}.\qedhere\ ] ] we expect lemma [ lem : vexun ] to hold under the weaker condition that @xmath68^n)$ ] , but we do not currently know how to prove this . since we know the scheme converges , we can prove a result analogous to proposition [ prop : extension ] regarding extensions of viscosity solutions of . [ prop : extension2 ] suppose @xmath129^n)$ ] is nonnegative and let @xmath76 be the maximal bounded viscosity solution of . then @xmath485 is a viscosity solution of @xmath486^n,\ ] ] where @xmath413 . let @xmath84^n_h \to{\mathbb{r}}_+$ ] denote the solution of for @xmath18 . by theorem [ thm : convs3 ] , @xmath331 uniformly on @xmath22^n$ ] . set @xmath487 for @xmath410^n_h$ ] . by proposition [ prop : extension ] , @xmath488 is a solution of the discrete scheme @xmath489^n_h.\ ] ] since @xmath490 uniformly on @xmath399^n$ ] , theorem [ thm : bs - conv ] shows that @xmath491 is a viscosity solution of . before proving a convergence rate for , we need another preliminary proposition . [ prop : h1 ] for @xmath492 such that @xmath493 @xmath494 since @xmath493 we can write @xmath495 in a sufficiently small neighborhood of @xmath496 . in this neighborhood @xmath497 where we used the inequality of arithmetic and geometric means . we now give the proof of theorem [ thm : rate1 ] . we extend @xmath7 to a function on @xmath399^n$ ] by setting @xmath423 . by proposition [ prop : extension ] we can extend @xmath94 to a function on @xmath399^n_h$ ] satisfying @xmath498^n_h,\ ] ] by setting @xmath499 . similarly , by proposition [ prop : extension2 ] we can extend @xmath76 to a function on @xmath399^n$ ] by setting @xmath422 , and @xmath76 is a viscosity solution of @xmath500^n.\ ] ] we will show that @xmath501 the opposite inequality is proved similarly . we can assume that @xmath502^n_h } ( w_h - w)>0 $ ] . for @xmath503 , @xmath410^n_h$ ] and @xmath504^n$ ] , set @xmath505 since @xmath499 , @xmath506 and @xmath507 , there exists @xmath508^n_h$ ] and @xmath509^n$ ] such that @xmath510^n_h \times ( -\infty,1]^n } \phi.\ ] ] let @xmath511 . then @xmath512 attains its maximum at @xmath513 with respect to the grid @xmath399^n_h$ ] . therefore @xmath514 for all @xmath17 . since @xmath508^n_h$ ] and @xmath113 is directed , @xmath515 since @xmath516 , we see that @xmath517 where we invoked remark [ rem : wlip ] in the last inequality . since @xmath518 we have @xmath519 where @xmath520 . by , @xmath521 . therefore , we can invoke the local lipschitzness of @xmath113 in all variables to obtain @xmath522 combining this with yields @xmath523 by we also have @xmath524 subtracting these equations yields @xmath525 since @xmath526^n_h } ( w_h - w)>0,\ ] ] we have @xmath527 . since @xmath528 , we can invoke proposition [ prop : h1 ] to find that @xmath529 choosing @xmath530 we have @xmath531^n_h } ( w_h - w ) \leq w_h(x_\alpha ) - w(y_\alpha ) \leq c\sqrt{h}. \qedhere\ ] ] we now prove a convergence rate for . this follows directly from theorem [ thm : rate1 ] and the following result . [ lem : changeofvar ] suppose @xmath68^n)$ ] is positive . let @xmath94 be the solution of , and let @xmath70 be the maximal solution of . then @xmath532_{1;[0,1]^n } , \inf_{[0,1]^n}f\big).\ ] ] set @xmath533 for @xmath86^n_h$ ] . by lemma [ lem : prod - rule ] , remark [ rem : wlip ] , and the inequality @xmath534 , we have @xmath535 here , we use the notation @xmath2 to denote function of @xmath154 that is bounded uniformly by @xmath536 with @xmath537_{1;[0,1]^n},\inf_{[0,1]^n}f\big).\ ] ] since @xmath94 satisfies we have @xmath538 it follows that @xmath539 therefore , there exists a constant @xmath540 such that @xmath541 let @xmath542 and set @xmath543 . then @xmath544 since @xmath545 and @xmath93 , we can use theorem [ thm : svcomp ] and remark [ rem : poscomp ] to conclude that @xmath546 on @xmath22^n_h$ ] . likewise , there exists a constant @xmath540 such that @xmath547 let @xmath548 , and take @xmath18 small enough so that @xmath549 . define @xmath550 . then @xmath551 since @xmath552 on @xmath39^n_h$ ] we have @xmath553 on @xmath22^n_h$ ] . we now have the proof of theorem [ thm : ratev ] . let @xmath94 be the solution of and let @xmath76 be the maximal bounded viscosity solution of . by theorem [ thm : rate1 ] , @xmath554^n_h.\ ] ] by lemma [ lem : changeofvar ] we have @xmath555 for all @xmath86^n_h$ ] and @xmath18 sufficiently small . the inequality follows from the lower bounds @xmath556^n } f \ \text { for all } x \in [ \delta,1]^n_h.\qedhere\ ] ] in this section , we discuss the bisection search method for solving , and we show the results of numerical simulations comparing the three schemes . in dimension @xmath33 , all three schemes have a closed form solution ( see , , and ) , and can therefore be efficiently solved up to machine precision . in dimensions @xmath35 , it is necessary to use an iterative method to approximate the solution of the scheme . due to the monotonicity properties of all three schemes , it is natural and efficient to use the bisection method , or slight variations thereof . the bisection method requires an initial interval to start the search , and a specified tolerance for terminating the bisections . the main idea is to set the tolerance to match the truncation error of each scheme , so that the error from the bisection method is no greater than the error incurred by using finite differences . we describe the details for each scheme below . for the scheme ( s1 ) , we use the bisection search method at each @xmath48^n_h$ ] to construct an approximation @xmath557 of the solution @xmath32 of . for the initial interval , we use @xmath558,\ ] ] and with a slight modification of the bisection method , we can ensure that the solution @xmath557 satisfies @xmath559^n_h,\ ] ] and @xmath560 for @xmath298 . by comparison for we have @xmath561^n_h,\ ] ] where @xmath32 is the exact solution of . since this error is no worse than the truncation error from using first order finite differences , it is unnecessary to continue the bisection search once is satisfied . for the scheme ( s2 ) , we use the same bisection search technique to find @xmath562 satisfying @xmath563^n_h,\ ] ] with @xmath564 for @xmath298 . by remark [ rem : lips ] , we may take the initial interval to be @xmath565.\ ] ] letting @xmath70 denote the exact solution of we have by theorem [ thm : svcomp ] and remark [ rem : s2exact ] that @xmath566 as before , this error is smaller than the truncation error introduced by discretizing , which justifies our choice of stopping condition for the bisection search . finally , for the scheme , we use the bisection method to obtain @xmath567 satisfying @xmath568^n_h.\ ] ] we take the initial interval for the search to be @xmath569,\ ] ] where @xmath570 by lemma [ lem : s3comp ] we have @xmath571 where @xmath94 is the exact solution of . we show here the results of some numerical simulations comparing the three schemes , , and . we first consider @xmath572 the solution of is given by @xmath573 the level sets of @xmath218 are illustrated in figure [ fig : demo1 ] . tables [ f1n2 ] , [ f1n3 ] and [ f1n4 ] show the @xmath54 errors and orders of convergence for @xmath574 in dimensions @xmath33 , @xmath575 , and @xmath576 , respectively . for each scheme , the @xmath54 errors are measured by how well the schemes approximate the viscosity solution @xmath218 of . .rates of convergence for @xmath574 in dimension @xmath33 . [ cols="^,^,^,^,^,^,^ " , ] the author is grateful to selim esedoglu and lawrence c. evans for valuable discussions about this work . we give the proof of lemma [ lem : prod - rule ] here . we proceed by induction . the base case of @xmath577 is exactly proposition [ prop : finite - diff - properties ] ( ii ) . assume holds for some @xmath415 . then by the inductive hypothesis and base case @xmath578 notice that @xmath579 where in the final summation , @xmath580 . we also have @xmath581 combining this with and we have @xmath582 this verifies for @xmath583 . the proof is completed by mathematical induction .

we present the historic photographic light curves of three little known blazars ( two bl lacs and one fsrq ) , bzb j1058 + 5628 , bzq j1148 + 5254 and bzb j1209 + 4119 spanning a time interval of about 50 years , mostly built using the asiago plate archive . all objects show evident long - term variability , over which short - term variations are superposed . one source , bzb j1058 + 5628 , showed a marked quasi - periodic variability of 1 mag on time scale of about 6.3 years , making it one of the few bl lac objects with a quasi- periodic behavior .

1004.0239

long - term ( @xmath0 20 years ) optical light curves are still available only for a relatively small number of active galactic nuclei ( agn ) and blazars . pioneering work in this field was made by several authors ( pica et al . 1988 ; webb et al . 1988 ) on relatively large samples of sources over a time window of about 20 years . for some sources these curves show long - term trends occurring over time scales of decades , while others show only short - term variability . a few sources had dedicated papers to their long - term optical variability . among the best studied sources we recall : oq 530 , whose optical brightness faded at 0.035 mag / yr for about one century ( massaro et al . 2004 ) ; s5 0716 + 71 , showing a monotonic brightening trend of 0.11 mag / yr of its mean luminosity over the last 40 years ( nesci et al . 2005 ) ; on 231 , with a long term decreasing trend of 0.023 mag / yr followed by an increasing one of 0.07 mag / yr ( massaro et al . 2001 ) ; wga 0447.9 - 0322 , with a long monotonic trend of 0.11 mag / yr , similar to s5 0716 + 71 ( nesci et al . 2007 ) ; 5c 3.178 declining at 0.03 mag / yr over 30 years ( sharov 1995 ) . in some cases a periodicity of the outbursts has also been found , the best case by far being oj 287 ( sillampaa et al 1988 , valtonen et al . 2009 ) . other such sources are ao 0235 + 164 , with a possible 5 years period of strong outbursts ( raiteri et al . 2008 and references therein ) and s5 0716 + 71 ( raiteri et al . 2003 ) with recurrence of about 3 years . also in the radio band a few sources have shown periodic outbursts ( e.g. 3c 454.3 , ciaramella et al . 2004 , qian et al . 2009 ) . the origin of the fast variations in blazars is generally explained by the relativistic boosting of perturbations moving down a jet pointing close to the line of sight . the relevant quantity for the boosting is the beaming factor @xmath1 , where @xmath2 and @xmath3 are the lorentz factor and the velocity ( in units of the speed of light ) of the perturbations bulk motion and @xmath4 is the angle between the jet and the line of sight . on the other hand the nature of secular variations is unclear . a suggestive possibility is that they can be associated with changes in the structure and/or direction of the inner jet ( see e.g. kadler et al . it is difficult , however , to obtain a clear evidence of such changes because it requires long and accurate multifrequency campaigns with vlbi angular resolution on a sample of several sources : such a study has been done e.g. by nesci et al . ( 2005 ) in the case of s5 0716 + 71 , suggesting that precession of the jet may indeed explain its observed light curve behavior , or massaro et al . ( 2004 ) in the case of oq 530 . with the successful launch of the fermi satellite for gamma ray astronomy , an all - sky monitoring of the blazars emission has started , which will probably bring to the attention of astronomers a number of poorly known sources . lists of potential gamma ray blazars have been prepared in the framework of the glast project by massaro et al . ( 2009 ) , containing about 2800 objects , and sowards - emmerd et al . ( 2005 ) , listing about 770 northern sky sources . for most of these sources very few data exist , basically those allowing their detection and the classification as an agn . a better knowledge of the properties of these sources will be useful for the interpretation of the gamma - ray data now available from the fermi mission . to determine the historic light curve of agns , an effective way is to use survey plates taken with wide angle instruments , like the schmidt telescopes , in fields covered over a large time span for patrol of other targets , like supernovae or variable stars . a good mine of such material is the archive of the asiago observatory ( http://dipastro.pd.astro.it/asiago/ ) , with its two schmidt instruments , the 67/92 cm , operative between 1965 and 1998 , and the 40/50 cm , operative between 1958 and 1992 . we selected therefore , from the massaro et al . ( 2009 ) catalogue , those sources for which no historic optical light curve is still published , nominally bright enough to be well measurable on the plates of the 67/92 cm telescope ( b@xmath517.5 ) , without an obvious strong host galaxy around , and for which a large ( @xmath650 ) number of plates is available spread over a long ( @xmath610 years ) time interval , so that a meaningful historic light curve can be derived . unfortunately , a very small number of sources matched these conditions , mainly due to their optical faintness . in this paper we report the results of our photometric measurements of these plates and the first optical historic light curve for three sources , bzbj1058 + 5611 , bzqj1148 + 5254 and bzbj1209 + 4119 . plates were digitized at the asiago observatory with an epson 1680 plus scanner , a sampling step of 16 micron ( 1600 dpi ) in grayscale / transparency mode and 16 bit resolution . plate scanning included also the unexposed borders to measure the plate fog level ( @xmath7 ) . the transformation of the recorded plate transparency @xmath8 of each pixel into a relative intensity @xmath9 was obtained applying the simple relation @xmath10 , where @xmath11 is the instrumental zero level . the best way to determine this level is to use the value of the darkest pixel in the center of the most overexposed star . most of the plates were with 103ao emulsion and gg13 filter , similar to the iiiaj+gg385 emulsion+filter combination used in the second palomar sky survey ( poss ii ) : therefore we established , for each field , a photometric comparison sequence selecting about 20 stars and taking their nominal magnitudes from the gsc2.3 catalogue , using the `` j '' photometric band , which is based on the poss ii . instrumental magnitudes for each plate were obtained using iraf / apphot tasks with a fixed aperture of 3.0 pixels for the 67 cm schmidt plates , corresponding to the fwhm of the average stellar profiles . we checked that the selected stars were not variable . the sky value was taken from a concentric annulus of 10 pixels inner radius and 10 pixels width , using the `` mode '' algorithm for the computation . a calibration curve for each plate was then obtained using the gsc2.3 values of the comparison stars . these curves were reasonably tight but showed a slight departure from a simple linear relation , so we fitted them with a 2nd order polynomial law . the most discrepant star was then automatically excluded and a new , final fit , recomputed . having a large number of plates for each field , we performed an intercalibration of our reference stars magnitudes . the differences with respect with the gsc2 values were nearly always smaller than 0.1 mag , indicating that this catalogue is a very good starting point to derive historical photographic light curves . the intercalibrated magnitudes ( which we will call b / asiago ) were then used to rebuild the calibration curve for each plate , which were again well fitted by a 2nd order polynomial relation with an appreciably smaller scatter . the @xmath12 magnitude of the agn was then derived from the instrumental one using the calibration curve . the scatter of the points around the fitting curve depends on the plate quality : typical rms errors are 0.1 mag . these errors were adopted as estimates of our photometric accuracy . a number of plates were taken without filter , so that a proper systematic correction had to be derived . indeed an agn is typically bluer than the average field stars and should appear therefore brighter when the uv part of the spectrum is not cut by the blue filter . unfortunately we did not find plates taken with the two filter / emulsion combinations in the same night ( or nearby nights ) for our sources . to find the systematic correction between the two combinations , we decided therefore to plot the light curves derived separately from the plates with and without filter and empirically look for a systematic shift between the two . this trick was successful in the case of 1148 + 52 , for which there was a large overlap for the two light curves : the correction was found to be small , about 0.2 mag , and was adopted for all three sources . one source ( 1058 + 56 ) was sufficiently bright that could be well detected also on the films taken with the smaller 40/50 cm schmidt telescope . this allowed to extend the light curve back to the years 1962 - 65 . these films were generally unfiltered and had the panchro royal emulsion ; very wisely , however , the observatory policy for the patrol fields taken with the 40/50 schmidt was to secure sometimes a couple of images taken in the same night with panchro royal and 103ao ( unfiltered ) emulsion for comparison . we tried to use these couples of films , where the agn was well exposed , to determine an empirical transformation from panchro royal into 103ao magnitudes , but the intrinsic photometric accuracy of the stars was rather low ( 0.2 mag ) and the number of pairs too small ( 5 ) to derive a statistically reliable correction ( -0.4 @xmath130.2 mag ) . also in this case therefore we overplotted the uncorrected light curves obtained from the 103ao+gg13 plates and from the unfiltered panchro royal films in the years 1965 - 1968 : the systematic shift was derived comparing the average values of the source in the two data sets . the result was found to be in good agreement with that obtained from the 5 plate pairs seen above . finally the correction from 103ao ( unfiltered ) into 103ao+gg13 derived from the 67 cm plates was applied to derive the final @xmath12 magnitudes . for each source a finding chart is given in fig . 1 , 2 and 3 , derived from the asiago plates , where the comparison stars are marked with letters and a is the agn . table 1 to 3 give the star flag ( column 1 ) , ra and dec ( j2000 ) derived from the gsc2.3 ( column 2 and 3 ) , adopted b@xmath14 magnitude ( column 4 ) and its uncertainty ( column 5 ) derived from the rms scatter in our dataset . the photometric data of the three agns are reported respectively in tables 5 , 6 and 7 . we report here just a few lines of table 5 as example . column 1 is the plate identification , column 2 is the filter / emulsion , column 3 the jd , column 4 is the rms deviation of the calibration curve , column 5 the b@xmath14 magnitude . the meaning of column 2 is : none=103ao unfiltered ; panc= panchro - royal emulsion without filter ; gg13= 103ao+gg13 . the resulting light curves are shown in fig . 4 , 5 and 6 . two additional points were obtained from the poss i and poss ii digitized blue plates available on - line ( e.g. at the eso website ) , using the same comparison stars used for the asiago plates and iraf / apphot for aperture photometry . the two poss plates have different emulsion / filter combination ( respectively 103ao unfiltered and iiiaj+gg395 ) so the poss i magnitude was corrected like the asiago 103ao unfiltered ones . another recent literature point can be derived from the sloan digital sky survey ( sdss , collinge et al . 2005 ) , if the source is not saturated , converting the g magnitude into the gsc2.3 b@xmath14 scale . a linear fit on a scatter plot of b@xmath14 _ vs. _ g of the comparison stars for each of our fields gave a slope @xmath151 and a small shift in the zero point . we expected however the presence of a small color term , due to the redder passband of the g filter with respect to the iiiaj+gg395 combination : given that the agns have an ( u - g ) color definitely bluer than the average field stars , this term should be appreciable . to this purpose , scatter plots of the ( u - g ) _ vs. _ ( g-@xmath12 ) magnitudes of the comparison stars for each field were made : the correlation however was not very tight , probably due to the lower accuracy of the gsc2.3 magnitudes , and in any case all the stars were substantially redder than the agn so that a large extrapolation of the linear fit would be required to evaluate the color correction ; eventually we preferred to ignore the color effect ( which is in any case about 0.1 - 0.2 magnitudes ) and just apply the zero - point correction . we made furthermore in the summer 2009 a final photometric point for each source using our vallinfreda 50 cm newtonian telescope and ccd camera with a marconi 47 - 10 back - illuminated chip and standard johnson - cousins filters . all these additional photometric points are also reported in the respective tables . bzbj1058 + 5628 the first optical spectrum of this source was published by marcha et al . ( 1996 ) giving a redshift z=0.41 form a strong [ o iii ] emission line . subsequent spectroscopic observations did not show any more these lines , but only the spectrum of the host galaxy with absorption lines at z=0.144 ( laurent - muehleisen et al . 1988 , bade et al . 1988 ) . a more recent spectrum can be found in the sdss ( adelman - mccarthy et al . 2008 ) , which provides the same redshift value ( z=0.143 ) from several absorption lines of the host galaxy and the faint h@xmath16 + n[ii ] emission lines . the equivalent width of this blend is about 5 , giving a classification of bl lac to this source . the host galaxy is well evident in the image of the sdss , but absolutely invisible in the asiago plates . to evaluate the galaxy contribution in the b band for our source we used the equivalent width of the g - band absorption line . the catalogue of absorption lines of elliptical galaxies ( trager et al . 1998 ) gives an average value 5.5 @xmath17 for this feature . the sdss spectrum of our source has a g - band e.w.=1.0 , so that about 80% of the flux is actually due to the blazar component at 4300 rest - frame . due to the redshift , the observed b band is centered on the caii h and k lines , where the galaxy contribution is still lower , so that the flux variation amplitude is little affected by the stable underlying galaxy contribution . we found 47 plates with 103ao+gg13 and 25 plates without filter taken with the 67/92 schmidt , covering the time interval 28 - 11 - 1965 to 06 - 01 - 1992 . their distribution in time however was not well suited to evaluate the systematic difference between the two combinations . given the brightness of the source , it was also measurable in several ( 41 panchro royal , 9 103ao ) films of the small schmidt , allowing us to derive its light curve in the years 1962 - 65 . films taken during the year 1969 were used to determine the systematic difference between the two filter / emulsion combinations . the asiago light curve covers about 11000 days and shows a definite monotonic decreasing trend of 2 magnitudes ( 0.053 mag / year ) . superposed on this trend there are wide oscillations with peak - to - peak amplitude of about 1 mag , with a time scale of about 2500 days ( 7 years ) . the fourier spectrum is dominated by three low frequencies ( 0.00039 , 0.00066 and 0.00082 , corresponding to 2550 , 1490 and 1220 days ) with similar amplitude , but a lot of other frequencies are present with non negligible amplitude . making a fit of the light curve keeping only the three low frequencies , the amplitude of the 2550 days period is the largest ( 0.27 mag ) followed by 1220 ( 0.14 ) and 1490 ( 0.11 ) , with a rms deviation from the actual light curve of 0.22 mag . the deviations are mainly concentrated in some segments of the light curve , and can be modelled only including several much higher frequencies : these deviations can be interpreted as flaring episodes in the jet , overimposed to the basic oscillating behavior of @xmath15 2500 days . such random variability is typical of bl lac objects , so it is expected to be present . using only the lowest frequency ( which is the strongest ) to fit the light curve , the best fit is achieved with a period is of 2590 days and the rms deviation is 0.25 , not dramatically larger than the 3-frequencies case . a formal monte carlo estimate of the period accuracy , made using the tools of period04 , gives 2590+/-30 days . a sinusoidal curve with period 2590 days and a decreasing trend of 0.053 mag / yr is reported in fig . 4 to guide the eye . we measured the source also on the poss i digitized plate ( 1953 - 03 - 18 , 103ao unfiltered , jd 34454 ) using our comparison sequence : the fit showed a rather large scatter ( 0.3 mag ) and the source was at 16.05 , which translate into b@xmath14 16.25 on our scale ( see below the discussion of bzqj1148 + 52 ) . the same procedure on the poss ii plate ( 1991 - 02 - 09 , iiiaj+gg395 , jd 48296 ) gave a much better fit ( rms 0.09 ) and the source was at @xmath12=16.19 , similar to the gsc2.3 catalogue value of 16.04 . the poss ii observation is fully consistent with the asiago light curve ( see fig . 4 ) , while the poss i observation is well below the extrapolation of the long term decreasing trend between jd 37,500 and jd 47,000 : this is suggestive that this trend may be just part of an oscillating behavior with longer timescale . further indications that this is actually the case come from the last points of the asiago light curve ( after jd 47000 ) and by more recent observations . one is from the sdss ( jd 52263 , g=16.50 ) which transforms into b@xmath14=16.20 in the gsc2.3 scale . another is from our own photometry using the vallinfreda 50 cm telescope ( jd 54995 ) , which gives b=15.33 @xmath130.16 and therefore tells that the source has returned to the very high state of the early 1960 s ( see fig . 4 ) . bzqj1148 + 5254 the spectrum of this source in the sdss shows strong emission lines and a redshift z=1.632 : it is therefore classified as a flat spectrum radio quasar in the roma bzcat . the image is point - like in the sdss , as expected for a high - redshift agn . we found 67 plates with 103ao+gg13 and 19 plates with 103ao without filter taken with the 67/92 schmidt , covering the time interval 24 - 01 - 1966 to 09 - 05 - 1994 . the time distribution of the plates for this source is well suited to evaluate the systematic difference between the two dataset : the light curves derived separately were indeed rather similar , with the unfiltered points systematically brighter , as expected . to derived the systematic correction for the unfiltered values we averaged the magnitudes of the agn in the time interval jd 44000 - 46500 , ( 12 points without filter and 17 with gg13 ) , obtaining @xmath18m=0.20 this value was adopted also for the other two sources . the asiago light curve covers about 9000 days and shows a monotonic decreasing trend of about 1.8 mag ( 0.073 mag / year ) with a slope quite similar to that of 1058 + 56 . the data are scattered around this trend , with no definite shorter time - scales . the photometric accuracy is however lower than in the case of 1058 + 56 , due to the lower flux level of this source . as expected from the visual inspection , the dft analysis did not show any remarkable frequency in the data besides the seasonal one year time scale of the observations . a couple of points definitely below the average ( jd 43311 and 46478 ) are probably underestimated due to the lower quality of the plates . on the poss i plate ( 103ao unfiltered , 1950 - 03 - 20 , jd 33360.85 ) we measured the source at 15.98 @xmath130.10 , which becomes b@xmath14=16.18 in the 103ao+gg13 system . the poss i point is therefore brighter than the brightest one in the asiago light curve , but definitely fainter than the straight extrapolation backwards of the observed trend . on the poss ii plate ( iiiaj+gg395 , jd 49456.78 ) we measured the source at b@xmath14 17.20 @xmath130.07 . this observation is about 500 days after the last asiago photographic point , and is roughly consistent with the general trend . on the sdss image ( jd 52288 ) the source was at g=16.93 , which becomes b@xmath14=17.15 on our scale . this is suggestive that the monotonic decrease has continued up to the year 2002 . our ccd photometry from vallinfreda ( jd 54995 ) , however , puts the agn at b@xmath14=16.6 , indicating that the source has considerably brightened and is now back at the levels of 1960 s . bzbj1209 + 4119 the source has an sdss nearly featureless spectrum , and can be classified therefore as a bl lac . in the sdss database it is given a redshift z=0.377 and the image is point - like . we found 32 useful plates with 103ao+gg13 and 18 plates without filter in the 67/92 schmidt archive . most of the plates were centered on the star 2 cvn and some on 67 uma . the dataset of two filter / emulsion combinations are not well mixed , so it is not easy to determine from them the systematic effect on the agn magnitude . we used therefore the correction obtained from 1148 + 52 as discussed above . the source was too faint to be measurable on the small schmidt films . the asiago light curve covers about 9500 days . the source is rather faint for the 67/92 schmidt so the photometric accuracy is worse than for the other two sources . in this case a slow brightening trend ( 0.04 mag / year ) seems to be present , with a large dimming episode around jd 45000 . the rms deviations of the comparison stars of comparable faintness ( s , g , t , q , see table 3 ) from their average values are anyway much smaller than that of the agn , so that at least in a statistical sense the overall source variation is real . in the final part of the light curve the flux variation of the agn is smaller , suggesting that part of the variations observed when the source was fainter may be due to its low signal / noise ratio ; however , the faint comparison stars were always detected on the asiago plates , so that we are confident that the low state of the agn around jd 45000 , recorded on several plates , is not just an artifact of photometric uncertainty . due to the low s / n ratio of the data , the low sampling frequency ( on average 1 point every six months ) and the presence of a single strong dimming feature in the light curve we do not think than a periodicity search is reliable . on the digitized poss i plate ( 1955 - 03 - 24 , jd 35190 ) we measured the source at 17.32 @xmath130.10 , which converts into b@xmath14=17.52 in our system . on the poss ii plate ( 1996 - 03 - 19 , jd 50161 ) we measured the source at b@xmath14=17.57 @xmath130.04 . on the sdss image ( jd 52731 ) the source is at g=17.98 , which becomes b@xmath14=18.28 on our scale . we observed the source from vallinfreda ( jd 55040 ) in the r band , because it was too faint to be detected in b ; in this case we used the sdss r magnitudes for our comparison sequence stars and the source was at about 17.5 , very near to the level of the sdss observation ( r=17.62 ) , suggesting that the source did not change very much in the last years . the slow historic trend shown by the asiago plates does not extend outside the time window sampled by the asiago observations . indeed if we extrapolate the trend to the poss i and poss ii epochs we would expect 18.1 and 16.8 , while the source was at 17.5 in both plates . the two more recent observations of the sdss and our own found the agn in a much fainter state , at about 18.3 , definitely not on the extrapolated tend ( 16.6 ) but rather at a flux level similar to the historic minimum recorded by six asiago plates . only a long time monitoring will tell us if such dimmings are a real characteristic of this source . pica et al . ( 1988 ) and webb et al . ( 1988 ) studied the long - term behavior of several tens of agns with photographic plates for 15 to 20 years . for 61 sources they had enough data to morphologically classify their light curve in four types : class i flickering without long term trend ; class ii long term trend with small flickering ; class iii long term and short term variability of comparable amplitude ; class iv rare outburst with stable flux level . in the blazar sample of 22 sources of webb et al . ( 1988 ) there is no significant difference in the frequency of the light curve classes between fsrq and bl lac objects , while a marked difference between blazars and qso exists in the pica et al . ( 1988 ) sample of 39 sources , which includes a good number of radio steep spectrum quasars and radio quiet qso . to have some physical insight of our sources we report in table 7 some basic data : column 1 is the name , column 2 the log(power ) at 1.4 ghz , column 3 the absolute r magnitude computed from the usno b1 catalogue magnitude and literature redshift , column 4 the radio / optical flux ratio , column 5 the nir spectral slope from the jhk magnitudes in the 2mass catalogue , column 6 the optical spectral slope from the sdss data . we remark that , at variance with the other two sources , the spectral slopes computed for bzqj1148 + 52 have a very poor @xmath19 and are therefore marked with `` : '' . actually the spectrum of this source can not be well fitted with a power law in neither of the two explored ranges , probably due to the strong emission lines ( e.g. c iv equivalent width is 77 ) and of the uv bump which falls in the optical due to the source redshift . it is apparent from this table that our three sources are flat - spectrum radio - loud objects ( radio / optical flux ratio @xmath610 ) but have different absolute luminosities and show substantially different behaviors in their optical light curve . the strong - lined object is the brightest both in the radio and optical bands . only bzbj1058 + 56 was detected in gamma - rays by fermi - lat ( abdo et al . 2009 ) and possibly also by egret ( bloom et al . 2000 ) . from the point of view of the overall spectral energy distribution , a much used tool for the classification of blazars is the @xmath20 diagram ( padovani and giommi 1995 ) . on this diagram the blazars mainly occupy two areas : a horizontal branch and a diagonal branch ; a diagonal line of negative slope -1 is a line of constant radio / x - ray flux ratio . the line at @xmath21=0.75 is the formal border between hbl and lbl sources ( padovani and giommi 1995 ) . extreme hbl sources are located at the left side of the horizontal branch , extreme lbl at the upper side of the diagonal branch . for a synchrotron self compton emission model , as the peak of the synchrotron emission of a blazar moves from lower ( @xmath22hz ) to higher ( @xmath23hz ) frequencies the location of the source on this diagram moves from the upper left corner to the lower right one along the diagonal line and then back to the left along the horizontal branch . we report in fig . 7 this diagram for the sources in the roma bzcat , with the positions of our three objects marked . bzbj1209 + 41 and bzqj1148 + 52 have @xmath21 larger than 0.75 , and are located in the diagonal branch , with bzbj1209 being the most radio loud . bzbj1058 + 56 is already on the horizontal branch and is an hbl , as discussed by donato et al . ( 2005 ) also on the basis of bepposax x - ray spectra . none of them is however an extreme case . bzbj1058 + 56 showed regular oscillations of about 1 mag amplitude , with timescale of @xmath152300 days , over a monotonic decreasing trend of 0.07 mag / year ; it can be classified as class iii and is the source with the larger variability . its historic light curve contains 5 outbursts sampled by the asiago plate archive ; this source seems therefore an interesting case of quasi - periodic bl lac . further multiwavelength monitoring of this source , should be performed . bzqj1148 + 52 showed a monotonic decreasing trend with a slope similar to 1058 + 56 but without the oscillating behavior : the detected short - term variability is comparable to our photometric uncertainty , so no firm conclusions can be derived on their time scale . it can be put into class ii . it showed a substantially smaller variability than the other two bl lacs , both on short and long time scales . bzbj1209 + 41 showed a slight increasing trend ( 0.04 mag / year ) with large dips in its light curve : it is therefore quite unusual ( the opposite of class i ) and deserves further monitoring . unfortunately it is not bright enough to be easily followed with small telescopes . when a time interval of about 50 years in considered , all the long term trends detected in the time window of 27 years sampled by the asiago plates do not seem to hold , so that they might be considered just as part of longer variability trends . which processes can be behind these secular trends ? both physical processes and geometrical effects can be at work : in the first case one can imagine a monotonic variation of the number of radiating electrons , or of the average ambient magnetic field ; in the second case , a change in the doppler boosting factor along our line of sight due to the jet precession . the latter possibility can be considered as an indicator of a massive black hole binary system in the nuclear region ( see e.g. romero et al . 2003 ) . if we interpret the long term trends of our three sources as due to a slow precession of the jet , as in was supposed to be in the case of s5 0716 + 71 ( nesci et al . 2005 ) , then the periodicity should be of several 10@xmath24 days and therefore comparable to ( or even larger than ) the human lifetime . this poses a strong challenge because observations must be accumulated for several tens of years before any firm conclusion can be reached . a further difficulty for the data interpretations , if the monitoring is not dense enough , could be the occurrence of fast and/or large occasional outbursts / dips , which can mask the long - term trends . a strong support to confirm the precession model could come from imaging at high radio frequencies with vlbi techniques , which could detect monotonic variations in the jet direction and/or at lower frequencies showing residuals of radio emission in regions involved by the crossing of the jet in the past ( see e.g. massaro et al . 2004 ) . finally , we remark that the detection of the quasi - periodicity of bzb j1058 + 56 from the asiago plates suggests that further discoveries could be made using other , still unexplored , photographic plate archives . we thank the asiago observatory and the department of astronomy of padova university for hospitality for the plate archive search , and alfredo segafredo for scanning part of the plates . the guide star catalogue - ii is a joint project of the space telescope science institute and the osservatorio astronomico di torino ( oato / inaf ) . this research made use of the cds ( strasbourg ) , ned ( nasa /ipac extragalactic database ) and sdss ( sloan digital sky survey ) databases . abdo , a. a. et al . 2009 , , 700 , 597 adelman - mccarthy et al . 2008 , , 175 , 297 . the sixth data release of the sloan digital sky survey bade , n. , beckmann , v. , douglas , n. g. , barthel , p. d. , engels , d. , cordis , l. , nass , p. , & voges , w. 1998 , , 334 , 459 bloom , s. d. , hallum , j. , terasranta , h. , & tornikoski , m. 2000 529 , 675 ciaramella , a. , bongardo , c. , aller , h.d . , aller , m.f . , de zotti , g. , et al . 2004 , 419 , 485 collinge m. j. ; strauss , m. a. ; hall , p.b . 2005 , , 129 , 2542 donato , d. , sambruna , r. m. , & gliozzi , m. 2005 , , 433 , 1163 giveon , u. , maoz , d. , kaspi , s. , netzer , h. , & smith , p.s . 1999 , , 306 , 637 kadler , m. , hughes , p.a . , ros , e. , aller , m.f . , & aller , h.d . 2006 , , 456 , l1 laurent - muehleisen , s. a. , kollgaard , r. i. , ciardullo , r. , feigelson , e. d. , brinkmann , w. , & seibert , j. 1998 , , 118 , 127 lenz , p. , breger , m. , 2005 , coast 146 , 53 marcha m. j. m. , browne , i. w. a.,impey , c. d. , & smith , p. s. 1996 , , 281 , 425 massaro , e. , mantovani , f. , fanti , r. , nesci , r. , tosti , g. , & venturi , t. 2001 , , 374 , 435 massaro , e. , mantovani , f. , fanti , r. et al . 2004 , , 423 , 935 massaro , e. , giommi , p. , leto , c. , et al . 2009 , , 495 , 691 nesci , r. , massaro , e. , rossi , c.,sclavi , s. , maesano , m. , & montagni , f. 2005 , , 130 , 1466 nesci , r. , mandalari , m , & gaudenzi , s. , 2007 , , 133 , 965 padovani , p. , & giommi , p. , 1995 , , 444 , 567 pica , a.j . , smith , a.g . , webb , j.r . , leacock , r.j . , clements , s. , & gombola , p.p . 1998 , , 96 , 1215 qian , s.j . , kudryavtseva , n.a . , britzen , s. , krichbaum , t.p . , witzel , a , et al . 2007 chjaa , 7 , 364 raiteri , c. m. , villata , m. , chen , w. p. , et al . 2008 , , 485 , l17 raiteri , c.m . , villata , m. , tosti , g. , nesci , r. , et al . 2003 , 402 , 151 romero , g. e. , fan , jun - hui , & nuza , s. e. 2003 , chjaa , 3 , 516 sharov , a. , s. , 1995 , astronomy letters 21 , 447 . sillanp , a. , haarala , s. , valtonen , m. j. , sundelius , b. , & byrd , g. g. 1988 , , 325 , 628 sowards - emmerd , d. , romani , r.w . , michelson , p.f . , healey , s.e . , & nolan , p.l . 2005 , , 626 , 95 trager , s. c. , worthey , g. , faber , s. m. , burstein , d. , & gonzalez , j. j. 1998 , , 116,1 valtonen , m. j. , nilsson , k. , villforth , c. , lehto , h. j. , takalo , l. o. , et al . 2009 , , 698 , 781 lindfors , e. ; sillanp , a. ; hentunen , v .- p . ; mikkola , s. ; zola , s. ; webb , j.r . , smith , a.g . , leacock , r.j . , fitzgibbons , g.l . , gombola , p.p . , & shepherd , d.w . 1998 , , 95 , 374 cccccc + 164.27350 & 56.53899 & b & 16.55 & 16.62 & 0.08 + 164.32579 & 56.43882 & c & 17.72 & 17.66 & 0.08 + 164.34494 & 56.44861 & e & 17.68 & 17.74 & 0.09 + 164.35300 & 56.38940 & d & 16.09 & 16.00 & 0.09 + 164.38134 & 56.46707 & f & 17.22 & 17.29 & 0.09 + 164.47731 & 56.45302 & i & 15.61 & 15.80 & 0.07 + 164.47977 & 56.58929 & j & 16.16 & 16.38 & 0.08 + 164.60784 & 56.44601 & k & 16.58 & 16.65 & 0.08 + 164.61842 & 56.54285 & l & 17.01 & 16.98&0.09 + 164.62266 & 56.43455 & m & 15.58 & 15.57 & 0.05 + 164.65048 & 56.34049 & n & 15.19 & 15.33 & 0.06 + 164.65070 & 56.53979 & o & 15.94 & 15.92 & 0.06 + 164.67445 & 56.50317 & p & 17.73 & 17.67 & 0.08 + 164.69448 & 56.39540 & q & 15.43 & 15.31 & 0.06 + 164.75473 & 56.33177 & r & 17.12 & 17.03 & 0.07 + 164.80188 & 56.49624 & s & 17.71 & 17.60 & 0.08 + 164.82489 & 56.33905 & t & 15.50 & 15.43 & 0.09 + 164.84320 & 56.37278 & u & 17.18 & 17.32 & 0.13 + 164.97340 & 56.44394 & y & 16.18 & 16.02 & 0.16 + 164.97473 & 56.48899 & w & 17.43 & 17.35 & 0.14 + cccccc + 177.21235 & 52.93091 & b & 16.14 & 16.20 & 0.09 + 177.28202 & 52.84490 & e & 17.34 & 17.27 & 0.12 + 177.43974 & 52.97246 & f & 17.87 & 17.77 & 0.14 + 177.43928 & 52.98250 & g & 17.77 & 17.78 & 0.14 + 177.49464 & 52.97587 & h & 16.29 & 16.22 & 0.11 + 177.50878 & 52.98667 & i & 17.40 & 17.48 & 0.15 + 177.44770 & 53.04665 & l & 16.67 & 16.90 & 0.12 + 177.02230 & 52.99680 & m & 17.80 & 17.83 & 0.13 + 177.09700 & 53.01454 & n & 17.81 & 17.70 & 0.13 + 177.10528 & 52.90447 & p & 15.66 & 15.48 & 0.05 + 177.13476 & 52.89412 & q & 15.92 & 16.05 & 0.13 + 177.19421 & 53.00136 & s & 15.98 & 15.95 & 0.09 + 177.25626 & 53.02330 & t & 17.35 & 17.27 & 0.14 + 176.96426 & 52.95703 & u & 15.29 & 15.43 & 0.08 + 177.15185 & 52.84147 & v & 16.94 & 16.92 & 0.12 + cccccc + 182.48757 & 41.35025 & b & 17.02 & 17.09 & 0.09 + 182.38187 & 41.43152 & d & 17.00 & 17.06 & 0.08 + 182.41075 & 41.24259 & e & 16.23 & 16.30 & 0.07 + 182.23561 & 41.42514 & f & 16.98 & 17.09 & 0.09 + 182.39451 & 41.35070 & g & 18.34 & 18.24 & 0.16 + 182.38566 & 41.35420 & h & 16.07 & 16.20 & 0.12 + 182.19538 & 41.40745 & i & 16.59 & 16.67 & 0.07 + 182.20402 & 41.23872 & j & 14.97 & 14.97 & 0.11 + 182.33838 & 41.27927 & m & 15.36 & 15.33 & 0.07 + 182.29515 & 41.21913 & o & 17.49 & 17.40 & 0.09 + 182.35833 & 41.18281 & p & 15.34 & 15.29 & 0.06 + 182.29730 & 41.24304 & q & 18.63 & 18.42 & 0.16 + 182.37593 & 41.18086 & r & 17.83 & 17.81 & 0.13 + 182.37925 & 41.37451 & s & 18.39 & 18.02 & 0.12 + 182.44092 & 41.37890 & t & 18.03 & 18.39 & 0.21 + 182.21330 & 41.44316 & w & 18.01 & 17.91 & 0.16 + 182.20821 & 41.44291 & z & 17.83 & 17.86 & 0.12 + lcccc osc o712 & none & 34454.7917 & 16.25 & 0.35 + s40 02499&panc & 37694.2986 & 14.92 & 0.06 + s40 02529&panc & 37696.4341 & 14.83 & 0.08 + s40 02552&panc & 37698.4230 & 15.55 & 0.10 + s40 02605&panc & 37702.3035 & 14.73 & 0.08 + s40 02615&panc & 37707.5452 & 14.84 & 0.09 + s40 02634&panc & 37719.3348 & 15.00 & 0.10 + s40 02659&panc & 37732.3153 & 14.88 & 0.08 + s40 02691&panc & 37753.4709 & 14.93 & 0.08 + s40 02708&panc & 37760.3709 & 14.92 & 0.07 + s40 02732&panc & 37763.3938 & 14.83 & 0.11 + s40 02755&panc & 37780.3271 & 15.06 & 0.09 + s40 02781&panc & 37786.3313 & 14.74 & 0.07 + s40 02831&panc & 37817.4757 & 14.67 & 0.05 + s40 02845&panc & 37821.4848 & 14.92 & 0.21 + s40 02861&panc & 37824.5278 & 15.10 & 0.12 + s40 02896&panc & 37843.4882 & 14.74 & 0.16 + s40 03291&panc & 37972.6174 & 15.43 & 0.09 + s40 03324&panc & 37992.5924 & 15.22 & 0.15 + s40 03360&panc & 37998.5827 & 15.54 & 0.12 + s40 03400&panc & 38001.6591 & 15.27 & 0.10 + s40 03435&panc & 38005.6243 & 15.35 & 0.08 + s40 03504&panc & 38020.5355 & 15.47 & 0.09 + s40 03600&panc & 38049.5528 & 15.41 & 0.09 + s40 03675&panc & 38055.5382 & 15.41 & 0.06 + s40 03741&panc & 38081.4813 & 15.34 & 0.08 + s40 03817&panc & 38089.6466 & 15.20 & 0.10 + s40 03864&panc & 38106.3757 & 15.20 & 0.08 + s40 03908&panc & 38139.3243 & 15.35 & 0.14 + s40 03946&panc & 38166.4146 & 15.51 & 0.10 + s40 04007&panc & 38193.3931 & 15.06 & 0.06 + s40 04055&panc & 38207.5049 & 15.40 & 0.08 + s40 04138&none & 38316.6216 & 15.62 & 0.12 + s40 04235&panc & 38346.6146 & 15.40 & 0.07 + s90 00060&none & 39093.6000 & 15.41 & 0.04 + s90 00182&none & 39168.5667 & 15.06 & 0.04 + s90 00228&none & 39199.4076 & 15.64 & 0.11 + s90 00473&none & 39475.6201 & 15.28 & 0.06 + s90 00573&none & 39506.6292 & 15.54 & 0.10 + s90 00635&none & 39587.3840 & 15.35 & 0.08 + s90 00651&none & 39609.3285 & 15.34 & 0.06 + s90 00690&none & 39622.4472 & 15.18 & 0.04 + s90 01295&none & 39864.5653 & 15.48 & 0.09 + s90 01466&gg13 & 39914.4542 & 15.71 & 0.14 + s90 01571&gg13 & 39942.4750 & 15.56 & 0.07 + s90 01587&gg13 & 39947.3979 & 15.68 & 0.09 + s90 01623&gg13 & 39972.3958 & 15.62 & 0.06 + s40 06837&panc & 40150.6035 & 15.34 & 0.11 + s40 06838&none & 40150.6146 & 15.14 & 0.34 + s90 02203&gg13 & 40260.4757 & 15.39 & 0.05 + s90 02225&gg13 & 40263.4660 & 15.30 & 0.12 + s90 02230&gg13 & 40267.5819 & 15.12 & 0.08 + s90 02246&gg13 & 40271.4701 & 15.31 & 0.04 + s90 02276&gg13 & 40290.3903 & 15.81 & 0.04 + s40 07158&panc & 40319.4063 & 15.20 & 0.13 + s90 02289&gg13 & 40319.4861 & 15.30 & 0.05 + s90 02301&gg13 & 40321.5708 & 15.33 & 0.03 + s90 02315&gg13 & 40322.4868 & 15.30 & 0.04 + s40 07220&panc & 40327.3757 & 15.46 & 0.13 + s90 02350&gg13 & 40329.5382 & 15.51 & 0.03 + s40 07301&panc & 40392.4237 & 15.49 & 0.11 + s40 07302&panc & 40392.4333 & 15.49 & 0.12 + s40 07321&panc & 40412.4341 & 15.60 & 0.12 + s40 07494&panc & 40505.6098 & 15.68 & 0.13 + s90 04272&none & 40979.4882 & 16.34 & 0.06 + s90 04212&none & 40999.5375 & 16.10 & 0.09 + s90 04213&none & 40999.5521 & 16.06 & 0.06 + s90 05376&none & 41446.3486 & 15.82 & 0.19 + s90 07799&gg13 & 42425.4889 & 15.53 & 0.05 + s90 07880&gg13 & 42459.5257 & 15.46 & 0.05 + s90 07895&gg13 & 42514.3826 & 15.59 & 0.06 + s90 08415&gg13 & 42833.4562 & 15.81 & 0.10 + s90 08453&gg13 & 42863.4681 & 15.84 & 0.03 + s90 08925&gg13 & 43187.3549 & 16.36 & 0.15 + s90 08949&gg13 & 43226.4403 & 16.19 & 0.10 + s90 09446&gg13 & 43540.3833 & 16.30 & 0.11 + s90 09479&gg13 & 43569.3611 & 16.49 & 0.09 + s90 09818&gg13 & 43843.5243 & 16.16 & 0.09 + s90 09991&gg13 & 43966.5528 & 16.45 & 0.05 + s90 10300&gg13 & 44203.5389 & 16.38 & 0.07 + s90 10318&gg13 & 44220.4896 & 16.57 & 0.04 + s90 10368&none & 44250.3910 & 16.67 & 0.13 + s90 10393&none & 44273.3826 & 16.35 & 0.11 + s90 10467&none & 44344.4153 & 16.18 & 0.07 + s90 10780&none & 44611.5354 & 15.73 & 0.06 + s90 10816&none & 44630.3521 & 15.97 & 0.07 + s90 10864&none & 44636.4799 & 15.88 & 0.08 + s90 11238&none & 44931.5403 & 15.62 & 0.09 + s90 11273&none & 44941.4799 & 15.61 & 0.11 + s90 11328&gg13 & 44988.4722 & 16.11 & 0.06 + s90 11416&gg13 & 45026.5333 & 15.85 & 0.05 + s90 11462&gg13 & 45053.4306 & 15.91 & 0.06 + s90 11535&gg13 & 45105.4257 & 15.68 & 0.07 + s90 11847&none & 45343.4743 & 16.01 & 0.06 + s90 11934&none & 45367.4444 & 16.16 & 0.06 + s90 11981&none & 45398.4174 & 16.12 & 0.08 + s90 12029&none & 45407.4458 & 16.06 & 0.05 + s90 12057&none & 45440.3674 & 16.24 & 0.09 + s90 12307&gg13 & 45647.5139 & 17.05 & 0.05 + s90 12453&gg13 & 45766.5125 & 16.88 & 0.05 + s90 12483&gg13 & 45814.3646 & 17.20 & 0.05 + s90 12669&gg13 & 46027.5111 & 16.67 & 0.04 + s90 12748&gg13 & 46062.4549 & 16.71 & 0.09 + s90 13784&gg13 & 46909.4167 & 16.77 & 0.05 + s90 13998&gg13 & 47184.5174 & 16.77 & 0.07 + s90 14019&gg13 & 47205.4882 & 16.56 & 0.04 + s90 14052&gg13 & 47212.5042 & 16.54 & 0.04 + s90 14066&gg13 & 47231.3722 & 16.62 & 0.03 + s90 14413&gg13 & 47556.4083 & 16.44 & 0.04 + s90 14427&gg13 & 47558.4354 & 16.40 & 0.05 + s90 14984&gg13 & 48275.5528 & 16.30 & 0.09 + osc sj03821 & gg395 & 48296.8333 & 16.19 & 0.07 + s90 15014&gg13 & 48301.4118 & 16.34 & 0.12 + s90 15031&gg13 & 48329.3840 & 16.46 & 0.05 + s90 15136&gg13 & 48628.5125 & 15.98 & 0.05 + sdss ccd & g & 52263.0000 & 16.20 & 0.05 + vallin ccd & b & 54995.3625 & 15.33 & 0.16 + rrrrr osc o59 & none & 33360.8611 & 16.18 & 0.10 + s90 00201&none & 39181.4630 & 16.32 & 0.06 + s90 00235&none & 39201.4450 & 16.34 & 0.08 + s90 00484&none & 39477.5120 & 16.45 & 0.07 + s90 00561&none & 39503.4980 & 16.29 & 0.06 + s90 00625&gg13 & 39565.6040 & 16.40 & 0.10 + s90 00654&none & 39610.3470 & 16.51 & 0.06 + s90 01482&gg13 & 39938.3600 & 16.43 & 0.05 + s90 02382&gg13 & 40362.5150 & 16.48 & 0.08 + s90 02389&gg13 & 40364.4920 & 16.38 & 0.10 + s90 02403&gg13 & 40375.3970 & 16.63 & 0.11 + s90 02404&gg13 & 40386.3900 & 16.53 & 0.09 + s90 02405&gg13 & 40387.3900 & 16.55 & 0.10 + s90 02420&gg13 & 40406.4180 & 16.88 & 0.10 + s90 04273&gg13 & 41040.4990 & 16.74 & 0.06 + s90 04301&gg13 & 41057.4030 & 16.79 & 0.04 + s90 05319&none & 41393.4840 & 16.91 & 0.12 + s90 07151&gg13 & 42134.3530 & 16.76 & 0.08 + s90 07155&gg13 & 42149.4560 & 16.82 & 0.07 + s90 07870&gg13 & 42453.5000 & 16.91 & 0.05 + s90 07893&gg13 & 42511.5220 & 16.82 & 0.04 + s90 07898&gg13 & 42514.4610 & 16.85 & 0.05 + s90 08416&gg13 & 42835.4810 & 16.96 & 0.03 + s90 08532&gg13 & 42889.4340 & 16.73 & 0.10 + s90 08854&gg13 & 43140.5700 & 16.99 & 0.04 + s90 08928&gg13 & 43191.5140 & 16.91 & 0.05 + s90 08951&gg13 & 43226.4930 & 16.89 & 0.05 + s90 09444&gg13 & 43311.3970 & 16.73 & 0.11 + s90 09472&gg13 & 43553.5480 & 16.78 & 0.09 + s90 09486&gg13 & 43631.3660 & 16.95 & 0.05 + s90 09838&gg13 & 43846.5620 & 16.75 & 0.09 + s90 09910&gg13 & 43905.5020 & 16.81 & 0.04 + s90 09918&gg13 & 43926.4350 & 17.04 & 0.08 + s90 09943&gg13 & 43936.4310 & 16.96 & 0.05 + s90 09961&gg13 & 43960.3160 & 16.83 & 0.06 + s90 09967&gg13 & 43962.3930 & 16.79 & 0.06 + s90 09969&gg13 & 43963.3750 & 16.85 & 0.05 + s90 09984&gg13 & 43966.4140 & 16.95 & 0.11 + s90 10000&gg13 & 43986.4000 & 16.89 & 0.07 + s90 10028&gg13 & 44013.4060 & 16.86 & 0.10 + s90 10061&gg13 & 44017.4480 & 16.76 & 0.09 + s90 10309&gg13 & 44219.4630 & 16.92 & 0.07 + s90 10372&none & 44252.3540 & 17.13 & 0.18 + s90 10391&none & 44275.3410 & 17.09 & 0.11 + s90 10817&none & 44630.3770 & 17.03 & 0.10 + s90 10887&none & 44640.4960 & 16.79 & 0.09 + s90 11240&none & 44931.6040 & 16.87 & 0.05 + s90 11305&gg13 & 44972.6130 & 16.88 & 0.06 + s90 11329&gg13 & 44988.4990 & 16.99 & 0.05 + s90 11452&gg13 & 45048.4390 & 17.02 & 0.10 + s90 11481&gg13 & 45055.4430 & 16.90 & 0.04 + s90 11502&gg13 & 45086.4250 & 17.00 & 0.06 + s90 11855&none & 45344.4910 & 17.05 & 0.07 + s90 11942&none & 45378.4680 & 16.92 & 0.05 + s90 12028&none & 45407.4260 & 17.07 & 0.06 + s90 12059&none & 45440.4140 & 16.97 & 0.09 + s90 12455&gg13 & 45766.5700 & 17.10 & 0.08 + s90 12482&gg13 & 45813.4320 & 16.87 & 0.11 + s90 12485&gg13 & 45814.4090 & 17.02 & 0.10 + s90 12700&none & 46035.5630 & 17.00 & 0.07 + s90 12711&gg13 & 46049.3910 & 16.77 & 0.12 + s90 12746&gg13 & 46061.4730 & 17.02 & 0.06 + s90 12829&gg13 & 46171.3960 & 17.09 & 0.10 + s90 13211&none & 46419.5460 & 17.12 & 0.09 + s90 13231&gg13 & 46442.5120 & 17.16 & 0.06 + s90 13292&gg13 & 46478.4550 & 17.40 & 0.07 + s90 13303&gg13 & 46495.4500 & 17.18 & 0.10 + s90 13335&none & 46563.3790 & 17.06 & 0.08 + s90 13602&gg13 & 46770.5420 & 17.09 & 0.04 + s90 13756&gg13 & 46879.4530 & 17.01 & 0.05 + s90 13779&gg13 & 46908.3580 & 17.07 & 0.04 + s90 13796&none & 46937.3680 & 17.23 & 0.12 + s90 14025&gg13 & 47208.3990 & 17.24 & 0.04 + s90 14069&gg13 & 47231.4520 & 17.02 & 0.03 + s90 14422&gg13 & 47560.5290 & 17.17 & 0.03 + s90 14487&gg13 & 47586.4520 & 17.05 & 0.05 + s90 14516&gg13 & 47593.3880 & 17.19 & 0.05 + s90 14535&gg13 & 47651.3990 & 17.11 & 0.07 + s90 15374&gg13 & 48987.4730 & 17.31 & 0.10 + osc sj05730 & gg395 & 49456.7896&17.20 & 0.05 + sdss ccd & b & 52288.0000&17.17 & 0.03 + vallin ccd & b & 54995.3500 & 16.61 & 0.09 + rrrrr osc o1367 & none & 35190.8611 & 17.50 & 0.10 + s90 00184 & none & 39171.5993 & 18.10 & 0.08 + s90 00237 & none & 39201.4681 & 18.15 & 0.13 + s90 00297 & none & 39287.4389 & 17.78 & 0.17 + s90 00631 & none & 39585.4861 & 18.00 & 0.09 + s90 00648 & none & 39593.3966 & 18.15 & 0.12 + s90 00692 & none & 39622.4952 & 17.76 & 0.13 + s90 01519 & gg13 & 39942.5063 & 17.54 & 0.08 + s90 05377 & none & 41449.3452 & 17.54 & 0.08 + s90 06896 & gg13 & 42016.6196 & 17.62 & 0.21 + s90 07667 & none & 42391.5167 & 17.73 & 0.05 + s90 07748 & gg13 & 42402.5980 & 17.77 & 0.10 + s90 07851 & gg13 & 42451.5091 & 18.10 & 0.06 + s90 07905 & gg13 & 42515.4299 & 18.02 & 0.07 + s90 07942 & gg13 & 42541.4966 & 17.66 & 0.07 + s90 07976 & gg13 & 42569.3959 & 17.80 & 0.07 + s90 08368 & gg13 & 42817.5660 & 17.87 & 0.04 + s90 08461 & gg13 & 42866.4771 & 17.70 & 0.08 + s90 08523 & gg13 & 42872.5396 & 17.49 & 0.08 + s90 08530 & gg13 & 42889.3917 & 17.61 & 0.13 + s90 08567 & gg13 & 42904.5466 & 17.81 & 0.12 + s90 08907 & gg13 & 43165.4924 & 17.65 & 0.13 + s90 08922 & gg13 & 43189.5119 & 17.34 & 0.05 + s90 08954 & gg13 & 43226.5675 & 17.68 & 0.07 + s90 09013 & gg13 & 43284.4786 & 17.41 & 0.08 + s90 09034 & gg13 & 43307.4015 & 17.60 & 0.15 + s90 09499 & gg13 & 43657.4376 & 18.00 & 0.11 + s90 09928 & gg13 & 43927.5189 & 18.21 & 0.04 + s90 10052 & gg13 & 44016.4473 & 17.86 & 0.12 + s90 10346 & none & 44226.5035 & 17.98 & 0.13 + s90 10906 & none & 44642.5404 & 17.54 & 0.09 + s90 10947 & none & 44691.4446 & 17.42 & 0.14 + s90 11337 & gg13 & 44989.4911 & 18.61 & 0.08 + s90 11453 & gg13 & 45048.4675 & 18.15 & 0.09 + s90 11482 & gg13 & 45055.4723 & 18.36 & 0.09 + s90 11518 & gg13 & 45088.5147 & 18.48 & 0.10 + s90 11877 & none & 45347.4869 & 18.69 & 0.08 + s90 11885 & none & 45350.4841 & 18.87 & 0.12 + s90 11989 & none & 45399.4445 & 17.98 & 0.05 + s90 12040 & none & 45412.4348 & 17.65 & 0.10 + s90 12055 & none & 45439.5035 & 17.56 & 0.09 + s90 12441 & gg13 & 45738.4689 & 17.34 & 0.11 + s90 12493 & gg13 & 45816.5279 & 17.93 & 0.10 + s90 14075 & gg13 & 47232.4258 & 17.14 & 0.07 + s90 14382 & gg13 & 47535.5251 & 17.05 & 0.09 + s90 14442 & gg13 & 47564.4377 & 17.34 & 0.14 + s90 14994 & gg13 & 48277.5331 & 17.14 & 0.15 + s90 15022 & gg13 & 48306.5585 & 16.88 & 0.09 + s90 15043 & gg13 & 48331.4518 & 17.27 & 0.11 + osc sj06694&gg395 & 50161.9111 & 17.50 & 0.10 + sdss ccd & b & 52731.0000 & 18.30 & 0.05 + vallin ccd & b & 55040.3560 & 18.30 & 0.20 + band of bzbj1058 + 56 . from march 1953 to june 2009 . filled circles are data from the asiago 67/92 cm schmidt plates ; crosses from the asiago 40/50 cm schmidt corrected for the zero point shift ; open squares other data . error bars are omitted for clarity . a sinusoidal line with a period of 2590 days and a monotonic decreasing trend is reported for comparison . [ lc1058 ] ] band of bzbj1148 + 52 . from march 1950 to june 2009 . filled circles are data from the asiago 67/92 cm schmidt plates with gg13 filter ; crosses are unfiltered data from the same telescope corrected for the zero point shift ; open squares are other data . error bars are omitted for clarity . [ lc1148 ] ] band of bzbj1209 + 41 . from march 1955 to july 2009 . filled circles are data from the asiago 67/92 cm schmidt plates with gg13 filter ; crosses are unfiltered data from the same telescope corrected for the zero point shift ; open squares are other data . error bars are omitted for clarity . [ lc1209 ] ]

we study dephasing in electron transport through a large quantum dot ( a fabry - perot interferometer ) in the fractional quantum hall regime with filling factor . in the regime of sequential tunneling , dephasing occurs due to electron fractionalization into counterpropagating charge and neutral edge modes on the dot . in particular , when the charge mode moves much faster than the neutral mode , and at temperatures higher than the level spacing of the dot , electron fractionalization combined with the fractional statistics of the charge mode leads to the dephasing selectively suppressing aharonov - bohm oscillations but not oscillations , resulting in oscillation - period halving .

1501.06030

a fractional quantum hall ( qh ) system of filling fraction @xmath3 has edge channels that support fractional charges obeying fractional braiding statistics @xcite . at @xmath4 , the edge states are decomposed into a @xmath5 charge mode and a counterpropagating neutral mode @xcite . they originate from renormalization of two counterpropagating charge modes @xcite , @xmath6 and @xmath7 , and stabilize at low temperature under strong disorder . neutral modes have attracted much attention , as they are charge neutral and carry energy . they have been recently detected through shot noise measurements @xcite , and their properties such as energy and decay length have been extensively studied @xcite . electron interaction is a dominant source of dephasing at low temperature @xcite . it leads to electron fractionalization @xcite in quantum wires ; an electron , injected into a wire , splits into constituents ( spin - charge separation , charge fractionalization ) , showing reduction of interference visibility or dephasing @xcite . interestingly , when the wire is finite , the constituents recombine after bouncing at wire ends , resulting in coherence revival @xcite . fractionalization was detected @xcite in a non - chiral wire , and studied in the integer qh edge @xcite . coherent transport , as well as dephasing , can be tackled through the study of low energy dynamics at the edge . this is particularly important in the context of the fractional qh regime . the present study implies that the presence of neutral modes could be a dominant source of dephasing . note that neutral modes have been observed in almost all fractional qh systems @xcite . at the same time there is no uncontested observation of anyonic interference oscillations in the pure aharonov - bohm regime of a fractional qh interferometer . the present study of the @xmath4 qh regime emphasizes two dephasing mechanisms by fractionalization of an electron into charge and neutral components , _ plasmonic dephasing _ and _ topological dephasing_. concerning the plasmonic dephasing mechanism , the overlap between the plasmonic parts of the charge and neutral components decreases with time , as the two components propagate with different velocities in the opposite directions . the resulting dephasing is similar to the plasmonic dephasing that takes place in a quantum wire or in integer qh edges . on the other hand , the topological dephasing is a new mechanism unnoticed so far . it occurs because the zero - mode parts of the components , satisfying fractional statistics , may braid with thermally excited anyons . thermal average of the resulting braiding phase leads to dephasing that occurs only in the interfering processes characterized by particular values of topological winding numbers . , coupled to lead edge states of @xmath4 ( black solid lines ) through quantum point contacts ( qpcs ) at @xmath8 . electron ( rather than fractional quasiparticle ) tunneling occurs through the qpcs ( dotted lines ) . following the tunneling , each electron ( and the hole left behind in the lead edge ) fractionalizes into a charge component propagating at velocity @xmath9 ( solid blue arrow ) and a neutral component counterpropagating at velocity @xmath10 ( dashed red ) . the magnetic flux in the dot area is @xmath11 . ] our analysis addresses the ab oscillation of differential conductance @xmath12 through a quantum dot ( qd ) in the @xmath4 qh regime . we focus on linear response of electron sequential tunneling into the qd . @xmath12 is decomposed into the harmonics of the ab flux @xmath11 in the qd , @xmath13 where @xmath14 is a flux quantum ; see fig . [ setup1 ] . semiclassically , @xmath15 represents the relative winding number of a fractionalized charge component , around the circumference @xmath16 of the qd , between two interfering paths : an electron , after tunneling into the qd , fractionalizes into charge and neutral components ; see fig . [ setup1 ] . the charge ( neutral ) component has propagation velocity @xmath17 , spatial width @xmath18 at temperature @xmath19 , level spacing @xmath20 , and scaling dimension @xmath21 ( @xmath22 ) in the electron tunneling operator at low temperatures . @xmath23 is determined by the overlaps of the components of the same kind between two interfering paths of relative charge winding @xmath15 . we find two mechanisms suppressing @xmath24 , the plasmonic dephasing and the topological dephasing ; the former ( latter ) involves plasmon ( zero - mode ) parts of the components . in the plasmonic dephasing , @xmath23 is contributed from the two interfering paths whose charge components overlap maximally between the paths . but , their neutral components overlap only partially between the interfering paths , reducing @xmath23 ; similar dephasing occurs in other fractionalizations @xcite . the topological dephasing additionally occurs , but depending on @xmath15 , in contrast to the other known mechanisms . when @xmath25 @xcite , the first harmonics @xmath26 is suppressed at @xmath27 ( namely , @xmath28 ) . it is because the charge component gains thermally fluctuating fractional braiding phase of @xmath29 ( leading to @xmath30 ) , while it winds once ( @xmath31 ) around @xmath32 electronic or anyonic thermal excitations on the qd edge or in the bulk . by contrast , the second harmonics @xmath33 is not affected by the topological dephasing ( as braiding phase @xmath34 and @xmath35 are trivial ) and dominates @xmath12 , resulting in @xmath2 ab oscillations . these above findings occur in both the regimes of strong disorder and weak disorder in the edge of the qd . note that the topological dephasing does not occur in the coulomb dominated regime @xcite where coulomb interactions between the bulk and edge of the qd is strong , as discussed later . the @xmath36 qd is coupled to two lead edges via quantum point contacts ( qpcs ) @xcite ; see fig . [ setup1 ] . the hamiltonian is @xmath37 . @xmath38 describes the edge of the qd , while @xmath39 the @xmath4 left ( right ) lead edge . each edge consists of the bosonic mode @xmath40 ( @xmath6 ) and the counterpropagating @xmath41 ( @xmath7 ) . @xmath42 supports charge @xmath43 and satisfies @xmath44 = i\pi \nu_i^{\textrm{edge } } \text{sgn}(x - x')\delta_{ii'}$ ] at positions @xmath45 . introducing the charge mode @xmath46 ( supporting charge @xmath47 ) and the neutral mode @xmath48 , one writes @xcite @xmath49 \nonumber \\ & + & \int^l_0 dx [ \xi ( x ) \exp(i\sqrt{2}\phi_n ) + \text{h.c.}].\end{aligned}\ ] ] disorder - induced tunneling amplitude @xmath50 between @xmath40 and @xmath41 is modeled by a gaussian random variable with mean zero and variance @xmath51 . for a finite range of bare parameters , @xmath52 and @xmath53 decouple @xcite at low temperatures , rendering @xmath54 irrelevant . @xmath55 is written similarly to @xmath56 , except @xmath57 in eq . . note that we ignore the coulomb interaction between the bulk and edge of the qd , considering that the qd size is large enough @xcite . the qpcs are almost closed , so electron tunneling is facilitated . renormalization group analysis @xcite indicates four equally most relevant electron tunneling operators between the electron field operators , @xmath58 at @xmath59 on the qd , and @xmath60 on lead edge @xmath61 ; @xmath62 is an ultra - violet cutoff and @xmath63 has the same form as @xmath64 . so the tunneling hamiltonian is @xmath65 $ ] , where @xmath66 is the tunneling strength . we show that at @xmath4 , fractionalization and fractional statistics cause the topological dephasing . we address the number operator @xmath67 of charge ( neutral ) mode at the qd edge , @xmath68 defined through the zero - mode parts of @xmath69 ( see appendix [ appennumber ] ) . the number operator @xmath70 of @xmath71 is an integer since @xmath72 and @xmath73 are the elementary charges of @xmath69 ; @xmath32 is an integer measuring charge excitations in the units of @xmath74 ( @xmath75 for a quasiparticle of charge @xmath74 ; @xmath76 for an electron ) . a quasiparticle of charge @xmath74 at position @xmath77 on the qd edge is written as @xmath78 @xcite . consider clockwise exchange of two such quasiparticles . since @xmath79 = i\pi \text{sgn}(x - x')$ ] , the exchange of the two charge components results in statistical phase @xmath80 , @xmath81 so , after the charge component of the electron operator @xmath82 winds once clockwise around @xmath32 charge - mode excitations on the edge , a phase @xmath83 is gained @xcite . here , 3 means the number of charge components forming @xmath84 , and @xmath85 refers to braiding ( double exchanges ) . similarly , the exchange of the neutral components of the two quasiparticles leads to exchange phase @xmath86 . so the neutral component of @xmath84 gains @xmath87 , after winding once around @xmath88 neutral - mode excitations ; the number of the neutral components of @xmath82 is @xmath89 . this has implications on the dynamics of an electron which enters into the qd and then fractionalizes . when @xmath90 , there is a process where the charge component of the electron winds once around the qd , while the neutral component moves very little . in terms of the winding numbers of the charge and neutral components , @xmath91 and @xmath92 , this process is denoted by @xmath93 . this process interferes with that of no winding @xmath94 , contributing to the @xmath1 harmonics @xmath95 ; see fig . [ interferencepaths2 ] . the relative winding numbers between the two interfering paths are @xmath96 , and the net braiding phase gained from that winding around @xmath32 charge and @xmath88 neutral excitations on the edge is @xmath97 . since @xmath32 is an integer , thermal fluctuations of quasiparticle ( or electron ) excitations on the edge give rise to fluctuations of the braiding phase factor @xmath98 [ @xmath99 ( @xmath100 ) for even ( odd ) @xmath32 ] , suppressing the @xmath1 harmonics . this topological braiding - induced dephasing also occurs due to thermal quasiparticle or electron fluctuations in the bulk ; see appendix [ appendix_quasiflucbulk ] for quasiparticle fluctuations in the bulk . note that this topological dephasing is utterly different from a dephasing mechanism at zero temperature , arising when quasiparticles travelling along an edge change internal degrees of freedom within the bulk ( e.g. , ref . @xcite ) . by contrast , the main contribution to the @xmath2 harmonics @xmath33 comes from @xmath102 = ( 2,0 ) . in this case , the braiding phase factor @xmath103 , regardless of @xmath32 being even or odd . hence , @xmath33 is immune to the topological dephasing . in general , such dephasing occurs only with odd @xmath104 , since the fluctuating @xmath105 is always even ; see eq . . when @xmath106 , the topological dephasing does not occur , since @xmath107 and @xmath105 is even . the above arguments hold for the pure ab regime ( or for the intermediate regime between pure ab and coulomb - dominated regimes ) . an apt question is to what extent this analysis holds for the coulomb dominated regime . when an electron of a given energy enters the qd by the process of sequential tunneling , it occupies a certain orbital state of the qd edge , satisfying energy conservation . in the pure ab regime , the area enclosed by the orbit ( hence , the ab phase assigned to the orbit ) is not modified when the number of quasiparticles or electrons in the bulk of the qd fluctuates thermally : edge - bulk interactions are negligible . hence , such thermal fluctuations affect only the braiding phase gained by the electron , leading to topological dephasing . by contrast , in the coulomb - dominated regime , the fluctuations are fully screened by the edge , reflecting the effect of edge - bulk interaction . this screening leads to modification of the area of the orbit , hence it modifies the ab phase of the orbit . this change of the ab phase exactly cancels out the change of the braiding phase caused by the thermal fluctuations . it follows that topological dephasing disappears in the coulomb - dominated regime . ) and neutral ( @xmath92 ) components . ( a ) @xmath108 : an electron injected at @xmath109 fractionalizes into charge ( moving along blue solid arrows ) and neutral ( red dashed ) components . ( b ) @xmath110 . for @xmath111 , the charge ( neutral ) component arrives at @xmath109 , after winding once around the qd , @xmath112 ( almost once , @xmath113 ) . the interference of relative winding numbers @xmath114 between ( a ) and ( b ) contributes to @xmath95 . reduced overlap between the neutral components of ( a ) and ( b ) leads to plasmonic dephasing . for @xmath90 , the dynamics is depicted for ( c ) @xmath93 and ( d ) @xmath115 . the charge component winds once in ( c ) and twice in ( d ) , while the neutral component moves little by @xmath116 in ( c ) and @xmath117 in ( d ) ( hence mainly @xmath118 ) . the interference between ( a ) and ( c ) suffers from topological dephasing with odd @xmath104 . the interference between ( a ) and ( d ) is immune to topological dephasing . ] we first sketch the topological dephasing in the case of @xmath90 and no disorder , and derive it , by expanding the zero - mode contribution to @xmath12 in harmonics of the winding numbers . the discussion is valid even with strong disorder . as in the main text , we consider the interference between the processes in figs . 2(a ) and 2(c ) . in the semiclassical regime of @xmath301 , counting the winding numbers @xmath91 and @xmath92 of the _ center _ of the spatial distributions of the charge and neutral components , we find that this interference contributes mainly to @xmath162 , and gains the net phase of @xmath302 from those windings that braid with @xmath32 charge excitations and @xmath88 neutral excitations . at lower temperature of @xmath303 , the tails ( namely the spatial width @xmath304 ) of the spatial distributions of the components are non - negligible , and imply that the two processes can also contribute to quantum mechanical net windings @xmath305 that can differ from @xmath102 . to see the contribution to different windings , we expand the average of @xmath306 over the thermal fluctuations of @xmath32 and @xmath88 in the harmonics of @xmath305 , @xmath307 where we introduce a fictitious `` neutral flux '' @xmath308 in the mathematical analogy of @xmath309 in order to have the expansion ; @xmath199 is now generalized to @xmath310 , and we put @xmath311 at the end . the thermal fluctuations of @xmath32 and @xmath88 are governed by the qd - energy @xmath312 ( cf . notice @xmath313 and @xmath314 , meaning that @xmath199 and @xmath315 are restored by changing @xmath32 and @xmath88 by 2 , when each flux shifts by one as @xmath316 and @xmath317 . we decompose the amplitude @xmath318 into the average @xmath319 over ( even @xmath32 , even @xmath88 ) and that @xmath320 over ( odd @xmath32 , odd @xmath88 ) ; @xmath32 and @xmath88 should have the same parity , according to eq . ( 3 ) in the main text . we find a useful relation of @xmath321 , obtained from the fact that the thermal average of @xmath306 over odd @xmath32 and odd @xmath88 at @xmath322 is identical to that over even @xmath32 and even @xmath88 at @xmath323 , according to @xmath315 . this relation leads to @xmath324 for odd @xmath325 , describing the topological dephasing . for the interference between those in figs . 2(a ) and 2(c ) , the contribution from the semiclassical winding of @xmath326 vanishes ( independent of temperature @xmath19 ! ) , while @xmath95 is contributed dominantly from the quantum mechanical winding of @xmath327 of the tail . we confirm the above discussion mathematically . the flux dependence of @xmath12 in eq . comes from the zero - mode part of the electron field operator @xmath82 , hence , from the green s function @xmath328 in eq . . using eq . , @xmath329 \exp(-i\pi v_{n}t n_{n } / l ) \rangle$ ] is computed as @xmath330 \nonumber \\ & \big / \big [ \sum\limits_{n_c , n_n = -\infty}^{\infty } \big \ { \exp \big ( -\frac{\frac{e_c}{12 } \big(2n_c + 1 -\frac{2\phi}{\phi_0 } + 1\big)^2 + \frac{e_n}{4 } \big(2n_n + 1\big)^2}{k_b t } \big ) + \exp \big ( - \frac{\frac{e_c}{12 } \big(2n_c -\frac{2\phi}{\phi_0 } + 1\big)^2 + \frac{e_n}{4 } ( 2n_n)^2}{k_b t } \big ) \big \ } \big ] . \label{zeromodeaverage_sup } \end{aligned}\ ] ] the first term of eq . comes from odd integers @xmath331 and @xmath332 , and the second from even integers @xmath333 and @xmath334 . utilizing the poisson summation formula of @xmath335 \exp [ 2bi(n+\delta ) ] = \sum_{p ' = -\infty}^{\infty } \exp(-2\pi i p ' \delta ) \exp [ -(\pi p ' + b)^2/a]$ ] with real constants , @xmath62 , @xmath336 and @xmath337 , we obtain @xmath338 \exp \big[-\frac{\pi^2 k_b t}{e_n}\big ( \frac{v_{n } t}{l } + \delta q'\big)^2 \big ] } { \sum\limits_{\delta p '' , \delta q '' = -\infty}^{\infty } ( -1)^{\delta p '' } ( 1+(-1)^{\delta p''+\delta q''})\exp ( 2\pi i \delta p '' \phi/ \phi_0)\exp \big(-\frac{3 \pi^2 k_b t}{e_c}{\delta p''}^2 \big ) \exp \big ( -\frac{\pi^2 k_b t}{e_n}{\delta q''}^2 \big ) } \nonumber \\ & = \frac{\sum\limits_{\delta p'+ \delta q ' \in 2 \mathbb{z } } ( -1)^{\delta p ' } \exp(2\pi i \delta p'\phi/ \phi_0 ) \exp \big [ -\frac{3 \pi^2 k_b t}{e_c}\big(\frac{v_{c } t}{l } - \delta p ' \big)^2 \big ] \exp \big[-\frac{\pi^2 k_b t}{e_n}\big ( \frac{v_{n } t}{l } + \delta q'\big)^2 \big ] } { \sum\limits_{\delta p''+ \delta q''\in 2 \mathbb{z } } ( -1)^{\delta p '' } \exp ( 2\pi i \delta p '' \phi/ \phi_0)\exp \big(-\frac{3 \pi^2 k_b t}{e_c}{\delta p''}^2 \big ) \exp \big ( -\frac{\pi^2 k_b t}{e_n}{\delta q''}^2 \big ) } . \label{zeromodeaverage2_sup } \end{aligned}\ ] ] here @xmath339 is the set of even integers . the denominator is approximated as 1 at high temperature @xmath340 ; this condition of @xmath340 is chosen for simplicity , and it is not a condition for the topological dephasing . then @xmath294 is written by the harmonics of winding numbers @xmath341 and @xmath342 , @xmath343 \nonumber \\ & \times \exp \big[-\frac{\pi^2 k_b t}{e_n}\big ( \frac{v_{n } t}{l } + \delta q_{\textrm{qm}}\big)^2 \big ] . \label{g0}\end{aligned}\ ] ] notice that the windings of odd @xmath344 do not contribute to @xmath294 , as mentioned in the main text . we compute @xmath12 in eq . to the order of sequential tunneling , @xmath119 where @xmath120 , @xmath121 is the ( renormalized ) electron tunneling rate between the qd and lead edge @xmath122 , @xmath123 \big\rangle$ ] is the green function describing the time ( @xmath124 ) evolution of the fractionalized components of an injected electron described by @xmath125 , and @xmath126 is a constant . the start and end positions of the green function @xmath127 coincide , since the green function describes the sequential tunneling . the injection leaves a hole behind on the lead edge . @xmath128 accounts for the fractionalization of the hole . for the detailed derivation of eq . , see appendix [ secappendix_seq ] . @xmath23 comes from the interference between two processes of relative charge winding number @xmath15 . at @xmath129 , the charge component has spatial width @xmath130 . then , @xmath131 contributes to @xmath23 mainly around the times @xmath132 , at which the charge component arrives at the initial injection point @xmath133 after winding @xmath15 times around @xmath16 ; the neutral component winding times @xmath134 are much less important , because of the scaling dimensions @xmath135 . we focus on @xmath136 and @xmath137 , as they involve the shorter times of @xmath138 , are more robust against the dephasing discussed below , hence are much larger than @xmath139 at @xmath140 . at @xmath140 , we compute @xmath136 and @xmath137 analytically in the absence of disorder and interaction ( @xmath141 , @xmath142 ) , and also in the strong disorder regime , based on a finite - size bosonization @xcite and a semiclassical approximation ( see appendix [ appendix_semi ] ) . we first deal with the regime of @xmath141 and @xmath142 [ see eq . ] and then discuss the regime of weak disorder and weak intermode interaction . we treat various contributions to dephasing quantitatively for the two cases of @xmath111 and @xmath90 . when @xmath111 and @xmath143 , only the plasmonic dephasing is important . the dominant contribution to @xmath12 comes from the @xmath1 harmonics . with the additional condition of @xmath144 , we obtain @xmath145 where @xmath146 . we explain two processes , whose interference dominates @xmath136 . in one process [ fig . [ interferencepaths2](a ) ] , an electron tunnels from lead edge @xmath147 into the qd and fractionalizes at @xmath148 at time @xmath149 , while at @xmath150 in the other [ fig . [ interferencepaths2](b ) ] . the charge components of the two processes interfere at @xmath148 at @xmath151 , contributing to @xmath136 , after respective windings @xmath152 and @xmath153 . at that time , the distance between the neutral components is @xmath154 , leading to partial overlap , hence , to the third factor @xmath155 of @xmath136 . the tunneling leaves a hole behind on @xmath147 , which also fractionalizes into charge and neutral components ( not shown in fig . [ interferencepaths2 ] ) . the partial overlap at @xmath151 between the two charge components from the holes created at @xmath156 and @xmath157 , and that between the two neutral components , lead to the first two exponential factors of @xmath136 in eq . , respectively . in the other limit of @xmath90 , both plasmonic and topological dephasings are crucial . there are two interfering processes for @xmath136 shown in figs . [ interferencepaths2](a ) and [ interferencepaths2](c ) , and for @xmath137 in figs . [ interferencepaths2](a ) and [ interferencepaths2](d ) . when @xmath143 , we obtain @xmath158 \label{conductancevcggvn1 } \\ \mathcal{g}_{\delta p = 2 } & \propto & \tilde{\gamma } l ( k_bt)^3 \exp[-2(\frac{l}{l_{t , c}}+\frac{\delta l}{l_{t , n}}+\frac{\delta l}{l_{t , n } } ) ] . \label{conductancevcggvn2}\end{aligned}\ ] ] the first three exponential factors of @xmath136 and @xmath137 result from plasmonic dephasing , as those of eq . . the third factor has a form different from that of eq . of @xmath111 , as the interfering neutral components in the qd are now @xmath116 apart in space . the factor @xmath85 in the arguments of @xmath137 arises from the double winding . another exponential factor @xmath159 $ ] of @xmath136 comes from the plasmonic part of the neutral component ; it is cancelled out with zero - mode contributions in eqs . and and also in other cases @xcite . the last suppression factor in eq . , @xmath160 $ ] , represents the topological dephasing , arising from the zero - mode parts of @xmath161 . the process in fig . [ interferencepaths2](c ) ( where the center of the neutral component hardly moves , while the charge component winds once around @xmath16 ) interferes with that of fig . [ interferencepaths2](a ) , contributing to @xmath162 . as discussed around eq . , this interference with @xmath163 is suppressed by the thermally fluctuating braiding phase factor of @xmath164 . the suppression factor is interestingly determined by the spatial tail ( or finite @xmath165 ) of the zero - mode part of the neutral component . the tail indicates that the neutral component can quantum mechanically wind once more than the semiclassical number @xmath92 of the center ; the quantum mechanical winding is well defined by the poisson formula ( see appendix [ appendix_top ] ) . hence , from the processes in figs . [ interferencepaths2](a ) and [ interferencepaths2](c ) , interference with the total relative winding number of @xmath166 can occur . as @xmath167 is even , this interference avoids the topological dephasing , dominantly contributing to @xmath26 , but it is reduced by the separation @xmath154 of the neutral components of @xmath168 relative windings . this explains the factor @xmath160 $ ] ; the exponent is quadratic in @xmath169 , since it originates from the zero - mode part @xcite . we point out that the topological dephasing occurs when @xmath170 ( @xmath171 ) , as seen in the last exponential factor in eq . . in contrast , the plasmonic dephasing occurs when @xmath143 . note that we choose the condition of @xmath143 for the derivation of eq . , to show both of the plasmonic dephasing and the topological dephasing simultaneously . because of the topological dephasing , @xmath137 is much larger than @xmath136 when @xmath90 ; @xmath172 is much smaller than the other factors . as a result , @xmath12 shows @xmath2 ab oscillations . in fig . [ conductance3 ] , we numerically compute @xmath12 for both @xmath111 and @xmath90 without employing the semiclassical approximation . the result for @xmath90 demonstrates the topological dephasing and consequent period halving even at @xmath173 . for ( a ) @xmath174 ( period @xmath175 ) and for ( b ) @xmath176 ( period @xmath177 ) at @xmath178 ( blue curve ) . @xmath12 is measured in units of @xmath179 and @xmath180 . ] so far , we have discussed the regime of no intermode interaction ( @xmath181 ) and no disorder ( @xmath141 ) . the argument of the regime holds also in the regime of weak intermode interaction and weak disorder , with slight modifications . in this regime , the plasmonic part of the neutral component decays , together with the diffusive spreading of the plasmonic part of the charge component @xcite . these slightly modify the plasmonic dephasing ( the first three dephasing factors in eqs . @xmath100 , but do not affect the topological dephasing . note that the weak disorder regime is realized when the renormalization of @xmath182 stops by temperature @xmath19 or qd size @xmath16 before going to the strong disorder regime , and a weak intermode interaction occurs in a dot , when the coulomb interaction between the charge modes is larger than the confining potential ( see appendix [ appen_coulomb ] and refs . @xcite ) . in recent experiments @xcite , neutral modes are measured with qds of size @xmath183m@xmath184 , implying that the intermode interaction is sufficiently weak in the qds . we show that eqs . @xmath100 hold in the strong disorder regime of a qd edge without any modification . in this regime , the neutral component is totally decoupled with the charge component ( @xmath142 in eq . ) @xcite . we start with the diagonalized form of @xmath38 @xmath185 $ ] . this form is obtained from eq . , where the effect of disorders is included . here , @xmath186 , the unitary matrix @xmath187 $ ] represents random disorder scattering , @xmath188 is a two - component fermionic operator , @xmath189 is an auxiliary bosonic field , and @xmath190 , @xmath191 and @xmath192 are the pauli matrices . the equal - position correlator @xmath193 \rangle$ ] is replaced by @xmath194 \rangle$ ] when we choose the global gauge transformation making @xmath195 . then , it is readily computed because the hamiltonian is free in the basis of @xmath196 , and is the same as @xmath197 in eq . that is obtained in the absence of interaction ( @xmath142 ) between the charge and neutral components and disorder ( @xmath141 ) . hence , eqs . @xmath100 can be also applied to the strong disorder regime . we have studied electron dephasing at @xmath4 . electron fractionalization into charge and neutral components leads to plasmonic dephasing . when @xmath90 ( which is likely @xcite ) and at @xmath198 , a new type of dephasing additionally arises . this dephasing is topological , resulting from the fractionalization and the fractional braiding statistics of the components , and occurs depending on the topological sectors characterized by the winding numbers @xmath102 of the components ; its dependence on the even - odd parity of @xmath104 is mathematical reminiscent of the parity ( integer versus half - integer spin ) dependent role of the topological @xmath199 term in antiferromagnetic spin chains @xcite . it leads to period halving of the ab oscillations . we emphasize that the topological dephasing occurs in both the regimes of strong and weak disorder , when bulk - edge interactions are not strong . in the case of weak disorder , which may be realized in high temperatures , weak intermode interaction causes the decay of the plasmonic part of the neutral component , accompanied by the diffusive spreading of the plasmonic part of the charge component @xcite . these do not affect the topological dephasing , hence the emergence of the @xmath200 oscillations . on the other hand , in the case of strong disorder at low temperatures , @xmath54 renormalizes towards zero @xcite , and then does not change the physics of the topological dephasing . note that bulk - edge coulomb interactions become weaker in qds of larger area ; the pure ab regime ( or the intermediate regime between the pure ab and the coulomb - dominated regimes ) could be achieved when the edge - to - bulk capacitance is smaller than other capacitances even when strong backscattering occurs at qpcs . we also note that the qh edges at @xmath4 may undergo more complex edge reconstruction at about @xmath201 mk @xcite . at lower temperature our analysis is applicable , while at higher temperature different topological dephasing may occur . assuming @xmath202 m / s , @xmath203 m / s , and @xmath204 , we expect that the @xmath2 oscillation will appear at temperature @xmath205 mk . in this case , the oscillation will be suppressed at @xmath206 mk , due to the plasmonic dephasing . detection of the period halving supports the topological dephasing , thus , the fractional statistics of the charge component at @xmath4 . the plasmonic dephasing and the topological dephasing will occur , with modifications , in other anyon interferometers or at other @xmath3 s . it should be mentioned that the known mechanisms yielding @xmath2 oscillations in other mesoscopic systems do not apply to our setup . the altshuler - aronov - spivak mechanism @xcite employs disorder averaging in multi - channel geometries , which is not present in our setup . another mechanism for @xmath200 oscillations @xcite relies on integer qh edge modes in an antidot at temperatures much below the charging energy of the antidot . moreover , our setup does not have superconducting fluctuations that support such periodicity . we thank g. rafael for a helpful discussion . hss was supported by korea nrf ( grant no . nrf-2010 - 00491 and grant no . nrf-2013r1a2a2a01007327 ) . yg was supported by dfg ( grant no . ro 2247/8 - 1 ) . we derive the hamiltonian @xmath38 ( cf . eq . in the main text ) @xmath38 is written in terms of the mode @xmath40 of filling factor @xmath207 and the counterpropagating mode @xmath41 of @xmath208 , as @xmath209 \nonumber \\ & + \int_{0}^{l } dx [ \xi ( x ) \exp(i\phi_1 + 3i\phi_2)+\text{h.c.}].\end{aligned}\ ] ] @xmath210 is the velocity of @xmath71 ( renormalized by the intra - mode interactions ) and @xmath211 describes the inter - mode interaction . @xmath69 satisfies @xmath212=i\pi \nu_i^{\textrm{edge } } \text{sgn}(x - x')\delta_{ii'}$ ] . each field @xmath213 is decomposed through @xmath214 into a plasmonic mode @xmath215 , satisfying the periodic boundary condition of @xmath216 , and a zero mode @xmath217 , @xmath218 the number operator @xmath219 counts the excess number of quasiparticles of charge @xmath220 . its canonical conjugate @xmath221 satisfies @xmath222 = i \delta_{i i'}$ ] , and @xmath223 changes @xmath219 by @xmath89 , acting as a klein factor . this ensures @xmath224 = 2 i \delta_{ii ' } \pi \nu_i^{\textrm{edge } } ( x - x')/l $ ] . combined with the commutation rule of the plasmonic part @xmath225 = i\pi \nu_i^{\textrm{edge } } ( \text{sgn}(x - x')-2 ( x - x')/l)\delta_{ii'}$ ] , this leads to @xmath212=i\pi \nu_i^{\textrm{edge } } \text{sgn}(x - x')\delta_{ii'}$ ] . the term @xmath226 in the bracket of eq . is introduced , to impose the boundary condition of electron operators , @xmath227 . the magnetic flux @xmath11 enclosed by the qd edge states causes the shift of @xmath228 in eq . ; as @xmath11 increases ( decreases ) by @xmath175 , the edge state with filling factor @xmath229 is energetically stabilized by removing ( adding ) its own quasiparticle of charge @xmath230 . combining @xmath231 s , one introduces the charge mode @xmath232 and the neutral mode @xmath233 , satisfying @xmath234=\pm i\pi \text{sgn}(x - x')$ ] and @xmath235=0 $ ] . putting this into eq . , we derives eq . we derive eq . in the absence of disorder ( @xmath141 ) . the charge ( neutral ) mode is decomposed into the zero - mode part @xmath236 ( @xmath237 ) and the plasmonic part @xmath238 ( @xmath239 ) . the latter describes edge plasmonic excitations , while the former anyon number excitations . the zero - mode parts are determined from eq . as @xmath240 we impose @xmath241 = i$ ] , @xmath242=0 $ ] , and @xmath243=\pm i \pi \text{sgn}(x - x ' ) \mp 2\pi i ( x - x')/l$ ] . these ensure @xmath244=\pm 2\pi i ( x - x')/l$ ] and @xmath234=\pm i\pi \text{sgn}(x - x')$ ] . comparing eqs . with , one gets eq . . in this appendix , we derive @xmath12 in eq . . the electron current @xmath245 along the right lead edge is given by @xmath245 , @xmath246 \rangle \nonumber \\ & = -\frac{ie}{\hbar } \sum_{i , j=\pm } \bigg \langle \textrm{t}_t \exp \bigg(\frac{i}{\hbar } \int_{-\infty}^{0 } dt\hat{h}_{\textrm{t } } ( t)\bigg ) \big [ t_{\mathcal{r}ij } \hat{\psi}_{\mathcal{r},i}^{\dagger}(0,t'=0 ) \hat{\psi}_{j}(x_{\mathcal{r}},t'=0)-\text{h.c.}\big ] \textrm{t}_t \exp \bigg(-\frac{i}{\hbar } \int_{-\infty}^{0 } dt \hat{h}_{\textrm{t } } ( t)\bigg ) \bigg \rangle , \end{aligned}\ ] ] @xmath247 counts electron number in the right lead edge , @xmath248 is the time ordering , the operators with ( without ) caret are in the interaction ( heisenberg ) picture , and @xmath249 is the thermal average . we set @xmath250 at which @xmath245 is measured . to second order in the tunneling strengths , @xmath251 is calculated as @xmath252 \big\rangle \nonumber \\ & = \frac{e}{\hbar^2 } \sum_{i , j=\pm } |t_{\mathcal{r}ij}|^2 \int_{-\infty}^{0 } dt\text{re } \big [ \exp \big(-\frac{i}{\hbar}(\mu_\mathcal{r}-\mu_\textrm{d})t\big ) \big ( g_{\textrm{d } , j}^k(-t ) ( g_{\mathcal{r } , i}^r - g_{\mathcal{r } , i}^a)(t)- ( g_{\textrm{d } , j}^r - g_{\textrm{d } , j}^a)(-t ) g_{\mathcal{r } , i}^k(t)\big ) \big ] . \end{aligned}\ ] ] @xmath253 , @xmath254 and @xmath255 are the retarded , advanced , and keldysh green s functions of lead edge @xmath256 ( qd ) , @xmath257\rangle,\nonumber\\ ( g_{\textrm{d } , \pm}^r - g_{\textrm{d } , \pm}^a)(t)&\equiv -i\langle \ { \hat{\psi}_{\pm}(x_{\alpha},t ) , \hat{\psi}_{\pm}^{\dagger}(x_{\alpha},0)\}\rangle , \,\,\,\,\,\,\,\,\,\,\,\,\,\ , g_{\textrm{d } , \pm}^k(t)\equiv -i\langle [ \hat{\psi}_{\pm}(x_{\alpha},t ) , \hat{\psi}_{\pm}^{\dagger}(x_{\alpha},0)]\rangle , \end{aligned}\ ] ] and @xmath258 ( @xmath259 ) is the chemical potential for lead edge @xmath61 ( the qd ) . @xmath259 is assumed to be uniform over the entire region of the qd , which is valid in the linear response regime . the expression @xmath260 of electron current in the left lead edge is similar to that of @xmath261 . to second order in the tunneling strengths , the current @xmath262 through the qd is written as @xmath263 . applying the current conservation condition of @xmath264 , we write the symmetrized form of @xmath262 as @xmath265\nonumber \\ & = \frac{e}{4\hbar } a v_c^{3/4 } v_n^{1/4 } \frac { \gamma_{\mathcal{l}}\gamma_{\mathcal{r } } } { \gamma_{\mathcal{l } } + \gamma_{\mathcal{r } } } \sum_{i , j=\pm}\int_{-\infty}^{0 } dt \nonumber \\ & \times \text{re } \big [ \big ( e^{-\frac{i}{\hbar}(\mu_{\mathcal{r}}-\mu_{\textrm{d}})t}-e^{-\frac{i}{\hbar}(\mu_{\mathcal{l}}-\mu_{\textrm{d}})t } \big ) \big ( g_{\textrm{d } , j}^k(-t ) ( g_{\mathcal{r } , i}^r - g_{\mathcal{r } , i}^a)(t)- ( g_{\textrm{d } , j}^r - g_{\textrm{d } , j}^a)(-t ) g_{\mathcal{r } , i}^k(t)\big ) \big].\end{aligned}\ ] ] in the second equality , we used the simplification that the left and right lead edges are symmetric ( @xmath266 ) , namely @xmath267 , @xmath268 , and @xmath269 . we also used the fact that the green s functions are independent of the index @xmath270 of the electron field operators in @xmath271 . the differential conductance @xmath272 is written as @xmath273 , \nonumber\end{aligned}\ ] ] where @xmath274 . the green s functions for the lead edges are computed as @xmath275 , \nonumber \\ g_{\alpha , i}^k(t ) & = \frac{1}{\pi a } \text{im } \bigg [ \bigg ( \frac{\sinh(\frac{i\pi a k_b t}{\hbar v_{c}})}{\sinh \big(\frac{\pi k_b t}{\hbar v_{c}}(ia - v_{c}t)\big ) } \bigg ) ^{\frac{3}{2 } } \bigg ( \frac{\sinh(\frac{i\pi a k_b t}{\hbar v_{n}})}{\sinh \big(\frac{\pi k_b t}{\hbar v_{n}}(ia - v_{n}t)\big ) } \bigg ) ^{\frac{1}{2 } } \bigg].\end{aligned}\ ] ] the green s functions are independent of the index @xmath276 , because of the imposed symmetry between the left and right lead edges . since @xmath277 at @xmath278 , and since the processes of @xmath279 do not contribute to the interference in @xmath12 , we can ignore @xmath280 in the expression of @xmath12 . then , a simplified form of @xmath12 is obtained as @xmath281 here , @xmath120 , @xmath282 , the weight factor @xmath283 , and @xmath123 \big\rangle$ ] represents a green s function of the qd ; its starting position @xmath133 is the same with the ending one in the sequential tunneling regime . the weight factor @xmath284 decays rapidly as @xmath285 for @xmath286 , describing the plasmonic dephasing ( by partial overlap due to different positions of the neutral components between the interfering paths ) occuring at lead edge @xmath61 . the above is the derivation of eq . in the main text . below , we further compute @xmath12 in the case of @xmath142 . we note that in the absence of disorders ( @xmath141 ) and interaction ( @xmath142 ) between the two modes , the hamiltonian @xmath287 of the zero - mode parts is obtained @xmath288 here we define energy scales @xmath289 . then , the green s function @xmath290 is decomposed into the charge and neutral components . then @xmath12 is simplified as @xmath291 , \end{aligned}\ ] ] where the plasmonic parts @xmath292 and @xmath293 of the charge and neutral modes and the zero - mode part @xmath294 are @xmath295 the elliptic - theta function of the first kind is @xmath296 , and @xmath297 @xcite . in the derivation of eq . , we used the relations of @xmath298 , @xmath299 , and @xmath300 . the zero - mode part @xmath294 will be calculated in appendix [ appendix_top ] . in this appendix , we compute the analytic expression of @xmath23 for the two cases @xmath111 ( eq . ) and @xmath90 ( eqs . and ) , utilizing a semiclassical approximation such that the time @xmath124 in the integrand of eq . is replaced by @xmath345 except for the time argument @xmath124 in @xmath346 . the approximation is based on the fact that the dominant contribution to the integrand of eq . comes from a peak structure of @xmath346 near @xmath347 ; the other peaks from @xmath293 near @xmath348 is more monotonous and less important because the scaling dimension ( @xmath22 ) of the neutral component in the electron tunneling operator is smaller than that ( @xmath21 ) of the charge component . this approximation is applicable when ( i ) the spatial distance between two interfering neutral components in the qd at @xmath349 is much larger than the width @xmath350 of the neutral components ( then @xmath293 is sufficiently monotonous ) , and ( ii ) @xmath351 ( then @xmath352 , coming from the lead edge , is sufficientlly monotonous ) . we focus on the contribution from near @xmath353 and near @xmath354 , since that from larger times @xmath355 of @xmath356 is much more smaller due to more dephasing . we first compute @xmath136 in the @xmath111 case with the semiclassical conditions of @xmath357 and @xmath358 , where @xmath359 . in eq . , the main contribution occurs at @xmath360 . near @xmath361 , we use the following approximations : ( 1 ) for the portion of @xmath362 in the zero - mode part , @xmath363 \simeq - \cos(2\pi \phi/ \phi_0 ) \exp ( - ( l - \delta l)^2 / ( l l_{t , n } ) ) $ ] ; cf . ( 2 ) for the lead edge part , @xmath364 / ( \hbar v_c ) = ( 8l / 3l_{t , c } ) \exp(-l / l_{t , c } ) \exp(-\delta{l}/l_{t , n})$ ] . ( 3 ) for the qd plasmon part , @xmath365 we have used @xmath366\exp(-\pi^2/4\gamma)\sinh(\pi ( u - n\pi)/\gamma)$ ] for @xmath367 . we compute eq . , merging together ( 1)-(3 ) , to get eq . in the main text , @xmath368 here @xmath369 is a constant , @xmath370 is the gamma function , and we have used the integral formula of @xmath371 notice that the factor @xmath372 of @xmath373 $ ] ( zero - mode part ) exactly cancels out @xmath374 $ ] from @xmath293 ( the plasmonic part ) . this fact was found in a luttinger liquid with finite size @xcite . we next move to @xmath95 in the @xmath25 case . the main contribution to @xmath95 also occurs near @xmath360 . we observe the followings . ( 1 ) because the portion of @xmath375 in @xmath363 $ ] fully vanishes , the dominant contribution comes from the portion of @xmath362 , which leads to @xmath376 \simeq - \cos(2\pi \phi/ \phi_0 ) \exp ( - ( l - \delta l ) ^2 / ( l l_{t , n } ) ) $ ] . ( 2 ) @xmath377 . ( 3 ) for the qd plasmon part , the same expression is obtained as eq . , except that @xmath154 is replaced by @xmath116 . merging ( 1)@xmath100(3 ) and using eq . , we obtain eq . in the main text , @xmath378 \nonumber \\ & = -g_0 \tilde{\gamma } l ( k_b t)^3 \exp(-\frac{l}{l_{t , n}})\exp(-\frac{l}{l_{t , c}}).\end{aligned}\ ] ] in the same way , we obtain eq . , we consider quasiparticle fluctuations in the bulk , differentiating the qd bulk from the edge . in the limit of bulk charging energy smaller than temperature , we show that the zero - mode part @xmath328 has the same expression as eq . in the presence of quasiparticle fluctuations in the bulk , hence , that the period halving ( @xmath200 oscillation ) takes place . we argue that the zero - mode part of the qd states is characterized by four numbers ( @xmath380 , @xmath381 , @xmath382 , @xmath383 ) , when we additionally consider the bulk degrees of freedom . @xmath384 and @xmath381 are necessary to describe excess electrons in the qd . for an additional excess electron , the number @xmath384 of excess electrons increases by 1 . this electron is decomposed into 3 additional charge components ( @xmath385 ; the charge of the charge component is @xmath74 ) and @xmath89 additional neutral component ( @xmath386 ) , as discussed in the main text . @xmath381 counts the number of the neutral components by excess electrons in the qd , as in the main text . on the other hand , @xmath382 and @xmath383 are introduced to describe quasiparticle excitations in the qd bulk . when an additional quasiparticle excites in the bulk , the number @xmath382 of quasiparticles increases by 1 . this quasiparticle is decomposed into 1 charge component and @xmath89 neutral component ( @xmath387 ) . @xmath383 counts the number of the neutral components by bulk quasiparticle excitations . taking into account quasiparticle fluctuations in the bulk and electron fluctuations inside the qd , the zero - mode part of the qd hamiltonian @xmath287 is expressed as @xmath388 where @xmath389 is the bulk charging energy . the first ( second ) term describes the interaction between the charge ( neutral ) components on the edge while the third term describes the interaction between the charge components in the bulk . the flux dependence in the bulk - charging energy term describes charge accumulation in the bulk as the magnetic flux increases . we ignore the interactions between the neutral components in the bulk because they are of dipole type hence weaker than the interaction terms of the total charge . we also assume that it is in the aharonov - bohm regime neglecting electrostatic coupling between quasiparticles in the bulk and on the edge . we consider the case that the relaxation time from the qd edge to the bulk is much longer than the qd dwell time of an electron injected from a lead edge , hence , that the electron enters only into the qd edge . then , the zero - mode part @xmath390 of the electron operators @xmath391 is evolved by eq . as @xmath392 and , the zero - mode part @xmath393 of the green s function @xmath394 is written as @xmath395 \exp(\pm\frac { i\pi v_{n}t } { l } ( n_{n } - n_{qp , n } ) ) \big\rangle \nonumber \\ & = \bigg \ { \sum\limits_{\delta p_{\textrm{qm } } , \delta p ' , \delta q_{\textrm{qm } } = -\infty}^{\infty } e^{-i\pi \delta p ' /3 } ( 1+(-1)^{\delta p_{\textrm{qm}}+\delta q_{\textrm{qm}}})(1+(-1)^{\delta p'+\delta q_{\textrm{qm}}})\exp(2\pi i \delta p_{\textrm{qm } } \phi/ \phi_0 ) \nonumber \\ & \times \exp \big [ -\frac{3 \pi^2 k_b t } { e_c}\big(\frac{v_{c } t}{l } + \frac{\delta p'}{3 } \big)^2 \big ] \exp \big[-\frac{\pi^2 k_b t}{e_n}\big ( \delta q_{\textrm{qm } } \pm \frac{v_{n } t}{l } \big)^2 \big ] \exp \big[-\frac{\pi^2 k_b t}{4e_{\textrm{bc}}}\big ( \delta p_{\textrm{qm } } - \frac{\delta p'}{3 } \big)^2 \big ] \bigg \ } \nonumber \\ & \bigg/ \bigg \ { \sum\limits_{\delta p_{\textrm{qm } } , \delta p ' , \delta q_{\textrm{qm } } = -\infty}^{\infty } e^{-i\pi \delta p ' /3 } ( 1+(-1)^{\delta p_{\textrm{qm}}+\delta q_{\textrm{qm}}})(1+(-1)^{\delta p'+\delta q_{\textrm{qm}}})\exp(2\pi i \delta p_{\textrm{qm } } \phi/ \phi_0 ) \nonumber \\ & \times \exp \big [ -\frac{3 \pi^2 k_b t } { e_c}\big(\frac{\delta p'}{3}\big)^2 \big ] \exp \big[-\frac{\pi^2 k_b t\delta q_{\textrm{qm}}^2 } { e_n}\big ] \exp \big[-\frac{\pi^2 k_b t}{4e_{\textrm{bc}}}\big ( \delta p_{\textrm{qm } } - \frac{\delta p'}{3 } \big)^2 \big ] \bigg \}. \label{zeromodeaveragebulk_sup } \end{aligned}\ ] ] we applied the poisson summation formula as in eq . . at temperature much higher than the bulk - charging energy , the portion of @xmath396 survives , resulting in the same expression as eq . . when the edge potential is smooth enough @xcite , we below show that @xmath9 is much larger than @xmath10 and @xmath54 ; see eq . in the main text . in the absense of coulomb interaction , the edge hamiltonian at @xmath36 has the form of @xmath397 $ ] in terms of the velocities of the original edge modes , @xmath398 and @xmath399 , which are solely determined by the edge confining potential . when the coulomb interaction is tuned on , it leads to an interaction hamiltonian @xmath400 , which counts the interaction by the total charge density @xmath401 ; here , we assume that @xmath402 is indepedent of momentum , ( which is valid when the coulomb interaction is short ranged due to the screening by gates ) . then , the total hamiltonian is @xmath403 $ ] , where @xmath404 , @xmath405 , and @xmath406 . this shows that @xmath54 and @xmath10 are much smaller than @xmath9 , if the edge potential is smooth enough such that @xmath407 , @xmath399 . a. e. hansen , a. kristensen , s. pedersen , c. b. sorensen , and p. e. lindelof , phys . b * 64 * , 045327 ( 2001 ) ; g. seelig and m. buttiker , phys . rev . b * 64 * , 245313 ( 2001 ) ; i. neder , m. heiblum , y. levinson , d. mahalu , and v. umansky , phys . * 96 * , 016804 ( 2006 ) ; s .- c . youn , h .- w . lee , and h .- s . sim , phys . 100 * , 196807 ( 2008 ) ; i. neder and e. ginossar , phys . lett . * 100 * , 196806 ( 2008 ) ; d. l. kovrizhin and j. t. chalker , phys . b * 81 * , 155318 ( 2010 ) ; d. b. gutman , yuval gefen , and a. d. mirlin , phys . * 101 * , 126802 ( 2008 ) ; a. p. dmitriev , i. v. gornyi , v. yu . kachorovskii , and d. g. polyakov , phys . * 105 * , 036402 ( 2010 ) . j. u. kim , w .- r . lee , h .- w . lee , and h .- s . sim , phys . lett . * 102 * , 076401 ( 2009 ) . h. steinberg , g. barak , a. yacoby , l. n. pfeiffer , k. w. west , b. i. halperin , and k. le hur , nature phys . * 4 * , 116 ( 2008 ) . b. rosenow and b. i. halperin , phys . lett . * 98 * , 106801 ( 2007 ) ; y. zhang , d. t. mcclure , e. m. levenson - falk , c. m. marcus , l. n. pfeiffer , and k. w. west phys . b * 79 * , 241304(r ) ( 2009 ) ; b. i. halperin , a. stern , i. neder , and b. rosenow , phys . b * 83 * , 155440 ( 2011 ) ; the braiding phase of a fractionalized component accumulated by winding around quasiparticle excitations has not been discussed earlier , although that of an elementary quasiparticle was mentioned in other papers ; d. arovas , j. r. schrieffer , and f. wilczek , phys . rev * 53 * , 722 ( 1984 ) ; s. kivelson , phys . rev . lett . * 65 * , 3369 ( 1990 ) ; c. de c. chamon , d. e. freed , s. a. kivelson , s. l. sondhi , and x. g. wen , phys . b * 55 * , 2331 ( 1997 ) . a. stern and b. i. halperin , phys . * 96 * , 016802 ( 2006 ) . b. l. altshuler , a. g. aronov , and b. z. spivak , _ pisma zh . eksp . teor , fiz . _ * 33 * , 101 ( 1981 ) [ jetp lett . * 33 * , 94 ( 1981 ) ] ; d. y. sharvin and y. v. sharvin , _ pisma zh . eksp . teor , fiz . _ * 34 * , 285 ( 1981 ) [ jetp lett . * 34 * , 272 ( 1982 ) ] ; a. g. aronov and y. v. sharvin , rev . mod . phys . * 59 * , 755 ( 1987 ) .

the propagation of a surface acoustic wave ( saw ) on gaas / algaas heterostructures is studied in the case where the two - dimensional electron gas ( 2deg ) is subject to a strong magnetic field and a smooth random potential with correlation length and amplitude . the electron wave functions are described in a quasiclassical picture using results of percolation theory for two - dimensional systems . in accordance with the experimental situation , is assumed to be much smaller than the sound wavelength . this restricts the absorption of surface phonons at a filling factor to electrons occupying extended trajectories of fractal structure . both piezoelectric and deformation potential interactions of surface acoustic phonons with electrons are considered and the corresponding interaction vertices are derived . these vertices are found to differ from those valid for three - dimensional bulk phonon systems with respect to the phonon wave vector dependence . we derive the appropriate dielectric function to describe the effect of screening on the electron - phonon coupling . in the low temperature , high frequency regime , where is the saw frequency and is the electron drift velocity , both the attenuation coefficient and are independent of temperature . the classical percolation indices give . the width of the region where a strong absorption of the saw occurs is found to be given by the scaling law . the dependence of the electron - phonon coupling and the screening due to the 2deg on the filling factor leads to a double - peak structure for .

cond-mat9604137

surface acoustic waves ( saw s ) @xcite provide a useful tool for experimental studies of the two - dimensional electron gas ( 2deg ) in gaas / algaas - heterostructures . in particular , saw s have been used in recent years in investigations of the integer @xcite and the fractional @xcite quantum hall regimes . due to the quantum hall effect , the interaction of the saw with the charge carriers can lead to strong oscillations in the attenuation and the velocity of the sound waves as function of the applied magnetic field . quantum oscillations have also been reported for the sound - induced currents and voltages @xcite . previous theoretical descriptions of these experiments have been based essentially on classical models for the propagation of saw s @xcite . according to these models , which are originally derived for systems in the absence of an applied magnetic field , the sound attenuation is expressed in terms of the electrical dc conductivity . this relation is derived under the assumptions that @xmath12 and ( ` local regime ' ) , where and @xmath6 are the wave vector and the frequency of the sound , respectively , and @xmath13 and @xmath14 are the mean free path and the scattering time of the conduction electrons , respectively . if a ( classical ) magnetic field is applied , the first condition has to be replaced by @xmath15 , where @xmath16 is the cyclotron radius@xcite . it is much more difficult , however , to determine under which conditions the above mentioned theories are valid when the 2deg is subject to a quantizing magnetic field . in this case , the electron system is characterized by ( at least ) two more length scales , namely the magnetic length @xmath17 and the localization length @xcite @xmath18 . while @xmath19 is always fulfilled under typical experimental conditions , the localization length can be much larger than the surface acoustic wavelength @xmath2 . a series of experiments @xcite has shown a reasonable agreement with the predictions of the classical models in a wide range of frequencies @xmath6 and magnetic field strengths . on the other hand , some deviations have also been detected . for example , deviations of the saw attenuation from the classically predicted behavior with increasing frequency have been reported@xcite . these were attributed to nonlocal effects of the interaction between the saw and the 2deg which should occur when the sound wavelength becomes of the order of or smaller than a characteristic length scale of the electron gas . in the fractional quantum hall regime , an anomaly in the absorption coefficient for filling factor @xmath20 was found @xcite . this anomaly was discussed in the framework of the composite fermion model of ref . . according to this approach , the electrons are replaced by composite fermions moving in an effective magnetic field of zero average ( at @xmath20 ) . then , the sound absorption due to these particles is described by classical formulae except that the dc conductivity is replaced by the wave vector dependent nonlocal conductivity which , however , represents a very important difference ( see ref . for details ) . in this paper we study the propagation of saw s in the integer quantum hall regime . the calculation of the saw attenuation is carried out for filling factors near 1/2 and is based on a percolation approach to the electronic states in a very strong magnetic field . from this point of view we may anticipate a nonlocal behavior of the attenuation arising from the large characteristic length scales ( e. g. , the size of a percolation cluster @xmath21 ) inherent in that framework . the effect of electron - electron interaction is taken into account by the screening of the electron - phonon coupling . the same problem has also been studied in ref . . these authors calculated the nonlocal conductivity due to variable - range hopping between pairs of localized states . then , in the spirit of the classical description of sound absorption , this conductivity is related to the saw attenuation coefficient . a comparison with our results will be given in sec . the system considered is a 2deg , subject to a very strong magnetic field @xmath22 and a smooth random potential @xmath23 . the potential can be characterized by its amplitude @xmath1 and its correlation length @xmath0 . the amplitude also determines the width of the landau levels . the correlation length is of the order of the spacer layer that separates the 2deg from the dopant layers . under the assumption that @xmath24 a quasiclassical description of the electron states can be applied . that is , one considers the drift motion of the guiding center of an electron on the equipotential lines ( el s ) of @xmath23 separately from the rapid motion relative to it @xcite . the drift velocity is given by @xmath25 . using @xmath26 , the drift velocity can be estimated to be @xmath27 . depending on the ratio between the correlation length @xmath0 and the sound wavelength , two regimes can be distinguished . for @xmath28 , the electron - phonon interaction can be considered locally neglecting the global structure of the el s @xcite . [ for single - phonon absorption and emission processes to occur , one has also to require that the sound velocity @xmath29 is smaller than the drift velocity . this is usually referred to as @xmath30erenkov condition@xcite . ] this regime is valid for , e. g. , thermal phonons@xcite . however , saw s have a much larger wavelength , and hence @xmath31 is typically fulfilled . in this case , the local absorption and emission of phonons is exponentially small , and the el as a whole has to be considered . it becomes important that the motion of the guiding center on an extended el [ with a length @xmath32 resembles a random walk , with a diffusion coefficient @xmath33 . since the ratio @xmath34 , for real systems , is not very different from unity , one deals with the limit @xmath35 of this diffusion process . [ to be precise@xcite , the parameter @xmath34 has to lie in the range @xmath36 . ] indeed , for @xmath37 t ( @xmath38 nm ) , @xmath39mev , @xmath40 nm , @xmath41@xmath42 , and @xmath43cms@xmath44 we find @xmath45 and @xmath46 . it is this particular ( diffusive ) regime which will be addressed in this paper . in the same regime , the electron life - time and the energy relaxation time due to interaction with 3d bulk phonons have been calculated in ref . . the quantum mechanical calculation of the attenuation coefficient ( as well as of other quantities associated with saw s ) requires the knowledge of the hamiltonians which describe the interaction of electrons with acoustic surface phonons . as far as we know these interaction hamiltonians have not yet been derived . instead , many theoretical investigations have addressed the interaction of 2d electrons with 3d ( bulk ) or 2d phonon systems . the latter one , a single layer of vibrating atoms , represents merely a theoretical construction . three - dimensional phonons do not provide an appropriate approach when the 2deg is located near a free crystal surface @xcite . this implies that it is by no means clear that the interaction of a saw with the 2deg is described well by the formulae which are valid in the case of bulk phonons . in fact , we find that the interaction vertices appearing in the general electron - phonon interaction hamiltonian [ see eq . ( [ hdefdet ] ) ] differ from those for 3d phonons not only by numerical constants but also in the phonon wave vector dependence and the relative phase between the deformation potential and the piezoelectric interactions . the paper is organized as follows . the interaction vertices are derived and discussed in the next section . in sec . iii , we describe the quasiclassical electron states of a 2deg in a strong magnetic field and a random potential @xmath23 . we show that the absorption of the saw and the dielectric function depend crucially on the occupation and the properties of electron states which correspond to very long el s . the structure of these el s is deduced from the 2d percolation theory . the matrix elements for transitions between different electron states are given in sec . the screening of the electron - phonon interaction due to the 2deg is accounted for by a dielectric function @xmath47 which is calculated in sec . v. based on these results , the saw attenuation coefficient is obtained in sec . its dependence on the filling factor ( or the fermi energy ) , the saw frequency and the temperature are discussed . a short summary is given in sec . to simplify the calculations we use the following assumptions . since the saw wavelength @xmath48 is much longer than the lattice constant , the crystal can be approximated by a continuous medium . its elastic properties are assumed to be isotropic . furthermore , we disregard the fact that the gaas - substrate is coated with layers which differ slightly in their elastic properties . the overall thickness of these layers @xcite @xmath49 nm is much smaller than the wavelength of sound . it has been shown @xcite that for @xmath50 the deviations of the wave propagation resulting from a thin overlayer coating an homogeneous substrate can be accounted for by a systematic expansion in this small parameter . in our case @xmath51 , i. e. , these corrections are indeed negligible . thus , we end up with the standard problem of sound waves which are propagated in an isotropic medium bounded by a plane @xcite . ( effects resulting from the anisotropy of the lattice become important for @xmath52 , see ref . . ) let the surface be in the @xmath53-@xmath54-plane and the medium in the half space @xmath55 . the longitudinal and transversal components of the displacement field @xmath56 , @xmath57 , obey the wave equations @xmath58where @xmath59 are the corresponding sound velocities . by definition , curl@xmath60 and div@xmath61 . surface waves are composed of particular solutions of eqs . ( [ waveeq ] ) that decay exponentially with increasing distance from the surface . in addition , these solutions have to satisfy the boundary conditions at the free surface @xmath62 , namely , the normal components of the stress tensor should vanish there . it turns out that these boundary conditions can only be fulfilled by a linear combination of @xmath63 and @xmath64 , i. e. , pure longitudinal or transversal surface waves do not exist @xcite . the full displacement field for a mode with a two - dimensional wave vector can be written as [ uq ] @xmath65with @xmath66 that is , the displacement @xmath67 is polarized in the sagittal plane which is spanned by the propagation direction @xmath68 and the surface normal @xmath69 . the decay of the displacements into the interior of the medium is described by @xmath70where @xmath71 and @xmath18 is a root of an algebraic equation of sixth order containing the parameter @xmath72 only ( see p. 104 in ref . ) . @xmath18 enters the dispersion relation of the surface waves in the form @xmath73 where @xmath29 is the saw velocity . finally , the factor @xmath74 is given by @xmath75 in order to quantize the displacement field ( [ uq ] ) , the normalization constant @xmath76 of each individual mode has first to be fixed in an appropriate way . that is , the energy associated with the mode @xmath77 in the normalization volume has to coincide with the energy @xmath78 of the corresponding phonon . since the wave is propagated freely along the surface , the energy is normalized with respect to a large but finite square of area @xmath79 in the @xmath53-@xmath54-plane . on the contrary , no such restriction is necessary with respect to the @xmath80-coordinate because @xmath67 decays exponentially with increasing distance from the surface . thus , the normalization volume can be extended from @xmath62 to @xmath81 under the chosen surface area . adding a kinetic energy term to the potential energy @xcite associated with a displacement field @xmath82 , the total energy can be written as @xmath83 , \]]where @xmath84 is the mass density of the medium and @xmath85is the strain tensor . inserting @xmath67 , eq . ( [ uq ] ) , into this formula and imposing the condition @xmath86 determines the normalization as [ norm ] @xmath87 with a numerical factor @xmath88 \ , . \nonumber\end{aligned}\ ] ] equations ( [ norm ] ) show that the normalization leads merely to a constant prefactor , i. e. , in contrast to the case of bulk phonons , @xmath89 does not introduce a further dependence on the wave number @xmath90 . we are now in a position to quantize the displacement field @xmath67 [ eqs . ( [ uq ] ) ] of a saw . according to the familiar rules , we define the phonon annihilation and creation operators @xmath91 and @xmath92 and find for the complete wave field the expression @xmath93 .\ ] ] the deformation potential is proportional to the change in volume , @xmath94 , which an infinitesimal volume element undergoes due to the wave @xcite . introducing an electron - phonon interaction constant @xmath95 , the hamiltonian of the deformation potential can be written as @xmath96the spread of the transversal component of the electron wave function as well as the distance @xmath97 of the 2deg from the surface are small compared to @xmath98 . thus , in evaluating eq . ( [ hdefgen ] ) , one can set all exponentials in @xmath99 , eq . ( [ uq2 ] ) , equal to 1 . conveniently , the electron - phonon interaction hamiltonian can be written in the general form @xmath100for a deformation potential interaction we derive from eqs . ( [ uqcomplete ] ) and ( [ hdefgen ] ) the interaction vertex @xmath101following a notation introduced in ref . , the electron - phonon interaction constant @xmath95 can be replaced by a nominal scattering time @xmath102 . this gives @xmath103where @xmath104 and @xmath105 . @xmath106 is the frequency of longitudinal optical phonons and @xmath107 is the effective mass of the electrons . along with the deformation potential interaction , the piezoelectric electron - phonon interaction appears in crystals which lack a center of symmetry , cf . for example ref . . in this case , an elastic wave leads to a polarization of the lattice , @xmath108where @xmath109 is the tensor of the piezoelectric moduli . the corresponding interaction hamiltonian follows from the electric potential @xmath110 associated with the polarization and reads @xmath111the polarization and the electric potential are related to one another via poisson s equation @xmath112where @xmath113 is the dielectric displacement and @xmath114 is the dielectric constant of gaas . in the case of interest here , the general expression ( [ pj ] ) is simplified because the gaas - samples used in experiments are cubic crystals and a crystal cut is chosen [ the ( 100 ) surface ] where the surface is spanned by two lattice axes @xcite . then , the tensor @xmath109 has only components in which all three indices @xmath115 , @xmath116 , @xmath13 differ from each other and all components are equal to @xmath117 . hence , eq . ( [ pj ] ) reduces to @xmath118substituting the displacement field ( [ uqcomplete ] ) into eq . ( [ strain ] ) for the strain tensor yields the polarization ( [ pola ] ) . then the poisson equation ( [ poisson ] ) for @xmath119 can be solved most easily by a fourier transform in the @xmath53-@xmath54-plane , leading to @xmath120 \varphi(z , t ) = \beta c q_x q_y e^{-i\omega_q t } \left [ -3\kappa_l e^{-\kappa_l qz } + f(1 + 2 \kappa_t^2 ) e^{-\kappa_t qz } \right ] + c.c . \ , .\]]the solution of this inhomogeneous differential equation can be constructed in the usual way . discarding the exponentially increasing term @xmath121 , one obtains that every mode with a wave vector is associated with an electric potential @xmath122where @xmath123 and @xmath124 . we note that for the geometry under consideration the total number of decay lengths for the elastic displacements and the electric potential is three , cf . ref . for comments on the general case . @xmath125 is a constant of integration which is determined by the boundary conditions at the surface @xmath62 . in view of the experiments , we assume that the surface of the crystal is an ` electrically free ' boundary @xcite to vacuum . that is , the normal component of the dielectric displacement , eq . ( [ poisson ] ) , and the parallel components of the electric field are continuous at the surface , [ bcond ] @xmath126@xmath127 note that the boundary conditions ( [ bcond ] ) differ from the requirement @xmath128 for a sample which is covered with a thin metallic film . an appropriate ansatz for the electric potential outside of the crystal ( @xmath129 ) is @xmath130 . substituting this ansatz and the general solution ( [ phiq ] ) for @xmath131 into the boundary conditions ( [ bcond ] ) yields that the constant of integration is @xmath132 , \]]where @xmath133 is the average of the dielectric constants of gaas and the space above the sample surface ( vacuum ) , respectively . for large values of @xmath134 , @xmath135 , this result coincides with the one which follows from the approximate boundary condition @xmath136 , cf.eq . ( [ bcond1 ] ) . the electric potential ( [ phiq ] ) associated with a single displacement mode is now completely determined . assigning the amplitudes @xmath91 and @xmath92 to the first and second term in eq . ( [ phiq ] ) , respectively , summing over all wave vectors , and introducing the result into the hamiltonian ( [ hpiezo ] ) , the piezoelectric vertex [ see eq . ( [ hdefdet ] ) ] becomes @xmath137 , \]]where we set @xmath62 in @xmath138 , eq . ( [ phiq ] ) . obviously , the strongest piezoelectric interaction occurs when the saw is propagated along a diagonal direction ( @xmath139 ) . in the experiments just this piezoelectric active direction [ @xmath140 $ ] ] is chosen . in terms of a nominal time @xmath141 [ cf . ( [ tauda ] ) ] the interaction vertex reads @xmath142where all the numerical quantities are absorbed in the prefactor @xmath143 . let us compare the results for the interaction vertices @xmath144 in the hamiltonian ( [ hdefdet ] ) with those valid for 3d bulk phonons ( or a fictitious 2d phonon system ) . there are two significant differences . first , the interaction vertices for saw s have a different dependence on the wave vector : @xmath145 exhibits an additional factor @xmath90 compared to @xmath146 . this applies to both the deformation potential and the piezoelectric interaction . consequently , the use of the saw interaction vertices in calculations of various physical quantities can give rise to results which deviate from those which are based on the assumption of two- or three - dimensional phonon systems . second , for surface phonons , the deformation potential and the piezoelectric interaction are _ in _ phase . this is in contrast to the case of bulk phonons where these vertices are _ out _ of phase , i. e. they contribute additively to latexmath:[$|\gamma_{bulk}|^2= ref . . the absolute value squared of @xmath144 is the relevant quantity which determines the total electron - phonon interaction . clearly , the ` out of phase ' or the ` in phase ' property is of importance only when the interaction vertices for the deformation potential and the piezoelectric interaction are of the same order of magnitude . this depends on the wavelength of the saw because @xmath148 , eq . ( [ vertpi ] ) , does not depend on the magnitude of @xmath90 whereas @xmath149 , eq.([vertdp ] ) , increases linearly with @xmath90 . in the case of gaas , the relative strength of these two interaction mechanisms is thus @xmath150 where we have used the numerical values given below . thus , for the range of wavelengths used in recent experiments on the attenuation of a saw in gaas - samples ( see , for example , refs . and ) , the deformation potential scattering can be neglected in comparison with the piezoelectric interaction , except for propagation along @xmath151 directions . this result corroborates very well with the fact that the experimental findings could be explained in terms of the piezoelectric electron - phonon coupling alone @xcite . for easy reference , we summarize the numerical values for the parameters appearing in the interaction vertices @xmath144 . for gaas , @xmath152cm / s , @xmath153cm / s and , hence , @xmath154 . the corresponding solution of the algebraic equation @xcite for @xmath18 is @xmath155 . substituting these values in eq . ( [ norm ] ) yields @xmath156 . using @xmath157ps , @xmath158ps , @xmath159k ( this corresponds to @xmath160ev , @xmath161ev / cm ) , and @xmath162g/@xmath163 ( see ref . ) , we obtain @xmath164ev@xmath165 and @xmath166ev cm . the above calculations are restricted , with respect to the piezoelectric interaction , to cubic crystals and a particular crystal cut . it is straightforward , however , to perform calculations for different crystals or surfaces along the same lines . consider a 2deg in a strong magnetic field @xmath22 perpendicular to the plane of the 2deg and in a smooth potential @xmath167 ( see , for instance , the paper by trugman@xcite and references therein ) . the potential @xmath167 is assumed to vary slowly on the scale of the magnetic length @xmath168 . electron - electron interactions are neglected . the wave function @xmath169 of a state with energy @xmath170 is appreciable only in the vicinity of an equipotential line ( el ) of the potential @xmath171 . the width of the wave functions perpendicular to the el is @xmath172 . explicitly , the electron states of the @xmath173-th landau level ( ll ) can be approximated , in the limit @xmath174 , by @xmath175^{-1/2 } \chi_n(v ) e^{i\varphi(u , v ) } \ , .\]](we have omitted the part @xmath176 of the wave function which corresponds to the lowest occupied subband perpendicular to the plane of the 2deg . ) the orthogonal variables @xmath177 and @xmath178 parametrize the distances along and perpendicular to the el , respectively . the function @xmath179 is the @xmath173th harmonic oscillator function . below , we shall restrict ourselves to the lowest landau level , i. e. @xmath180 . in eq . ( [ psi ] ) , @xmath181 is a gauge - dependent phase , and @xmath182 is the period associated with one revolution around the el of an electron moving with the drift velocity @xmath7 . that is , @xmath183for the wave function ( [ psi ] ) to be single - valued , @xmath181 has to change by an integral multiple of @xmath184 around an el . this condition leads to the quantization of the allowed constant - energy lines . in other words , only a discrete sequence of el s corresponds to the electron eigenstates . two adjacent eigenstates enclose an area @xmath185 and are , on the average , a distance @xmath186 apart , where @xmath187 is the length of one of the el s . an important quantity is the difference @xmath188 of the corresponding eigenenergies , where the frequency @xmath189 is determined by @xmath190the quasi - classical description of the electron states that we have outlined above is a valid approximation when@xcite @xmath191where @xmath192 is the local radius of curvature of the el and @xmath193 is the effective electron mass . the first condition is related to the smoothness of the potential , while the second one prevents the mixing of different ll s . additionally , one should keep in mind that quantum tunneling@xcite between classical el s is important when @xmath194 is smaller than @xmath195 . in what follows @xmath167 is a smooth _ random _ potential . the potential is assumed to be gaussian , with @xmath196where @xmath197 defines its amplitude and @xmath0 its correlation length . ( @xmath1 determines also the broadening of the ll . ) the zero of energy is chosen such that @xmath198 , i. e. , the energy @xmath170 is measured from the center of the lowest ll . using @xmath1 and @xmath0 , we can rewrite the conditions ( [ cond ] ) in the form @xmath199where @xmath200 is the cyclotron frequency . these conditions are fulfilled , for example , for the following experimental values : @xmath37 t ( @xmath38 nm , @xmath201thz ) , @xmath39mev and @xmath202 nm . the separation between two consecutive ll s is much larger than their broadening , @xmath203 . as discussed in ref . , most el s with @xmath170 in the tail of the ll , i. e. @xmath204 , have diameters @xmath205 which are small compared to @xmath0 and their length @xmath187 is of the order of @xmath205 . when the energy approaches the center of the ll , the size of the el s grows @xcite . in particular , for @xmath206 , @xmath207 holds for most of the el s . such el s will be denoted as ` standard ' . a further reduction of @xmath194 does not lead to an increase in the size of almost all el s , i. e. most of them remain standard ones . however , a minority of the el s merges and forms large extended el s with diameters @xmath208 . the structure of these extended el s is described by percolation theory @xcite when the energies @xmath194 are near the percolation threshold @xmath209 ( or @xmath210 ) . the subsequent calculations show that just this range of energies is of interest justifying the use of the percolation picture . an extended el has a fractal structure which is reflected in the relation between its length and diameter @xcite @xmath211 where @xmath212 is the scaling exponent . ( for the definition of the perimeter of discrete percolation clusters and the transition to continuum percolation see , e. g. , refs . and . ) an extended el can be viewed as a self - avoiding random walk path with steps of length @xmath0 . indeed , @xmath213 is close to the value 2 which applies to a simple random walk . the exponent is less than 2 due to the self - avoiding nature of the el . the distribution of extended el s of a given energy @xmath170 is described by one scale , the so - called critical diameter @xmath214 which is considered to be the localization length in the semiclassical theory . the scaling index is @xmath215 . el s with diameters @xmath216 are exponentially rare , while the probability to find an extended el with a diameter @xmath217 is proportional to @xmath218 , where @xmath219 . an electron that moves on an extended el ( @xmath220 ) experiences different regions of the random potential . during one revolution , the drift velocity @xmath221 [ see eq . ( [ periodt ] ) ] follows the varying slope of the potential @xmath167 and takes on many different values . in other words , the motion on an extended el corresponds to an averaging process with respect to @xmath221 . it is therefore reasonable to introduce an average drift velocity@xcite @xmath222 , defined by @xmath223 , that is assumed to be independent of the length of the el under consideration . the dependence on the energy @xmath170 can be generally excluded since @xmath167 and @xmath224 are statistically independent for a gaussian distribution @xcite . consequently , the energy level spacing ( [ omtdef ] ) associated with the extended el s is a function of @xmath187 alone and eq . ( [ omtdef ] ) can be written conveniently in the form @xmath225where @xmath226the frequency @xmath227 gives by order of magnitude the level spacing for standard el s since it is associated with the revolution around an el with @xmath228 . the corresponding frequencies for the extended el s are lower . the lowest frequencies belong to the longest el s which have the critical length @xmath229 corresponding to the critical diameter @xmath230 ( [ dcrit ] ) . from eqs . ( [ ldrela ] ) and ( [ dcrit ] ) @xmath231 below , we shall use the distribution of the el s with respect to their lengths . let @xmath232 be the number of el s with energy @xmath170 and a length between @xmath187 and @xmath233 . the normalization of this distribution can be found by equating the total average length of the el s in an area of size @xmath79 to the result given in the literature ( see sec . iii.a in ref . or ref . ) @xmath234^{1/2 } \exp[-{\epsilon^2\over 2\delta^2 } ] \ , , \ ] ] where @xmath235 is defined in eq . ( [ correlator ] ) . while the distribution of the standard el s is not really known , percolation theory gives the following ansatz @xcite for the distribution of extended el s @xmath236 } g\left({{\cal l}\over{\cal l}_{c}(\epsilon)}\right)d{\cal l } \ , , \qquad { \cal l}\gg \lambda \ , , \ ] ] where @xmath237 is the normalization constant and @xmath238 is a function which is exponentially small for @xmath239 and of order unity for @xmath240 . hence , @xmath241 yields a smooth cut - off of the distribution for @xmath242 , where @xmath229 is defined in eq . ( [ areac ] ) . an additional , energy - independent cut - off appears in a finite sample . here , the size @xmath243 of the system restricts the critical diameter @xmath230 ( [ dcrit ] ) to values such that @xmath244 . this translates into @xmath245 , using eq . ( [ ldrela ] ) . thus , in a finite system , the critical length @xmath246 in eq . ( [ feext ] ) should be replaced by min@xmath247 . to find the normalization constant @xmath237 let us decompose the normalization integral ( [ normfe ] ) into @xmath248 and estimate both terms of this decomposition . the second integral can be estimated from the distribution ( [ feext ] ) of extended el s . with the value @xmath249 , this integral is determined by its lower limit @xmath0 and is of the order @xmath250 . using a reasonable ansatz for the distribution of standard and short el s ( for example @xmath251 ) , the first integral is determined by its upper limit @xmath0 and is again of the order @xmath250 . thus , the total normalization integral is also of this order . up to a factor of order unity , the normalization constant @xmath237 is then given by @xmath252 . the numerical factor can be absorbed in @xmath241 leading to the following distribution function for extended el s @xmath253 } g\left({{\cal l}\over{\cal l}_{c}(\epsilon)}\right)d{\cal l } \ , .\]]we note that the above estimates confirm that the majority of el s belongs to the standard ones with @xmath228 , since these el s are relevant in the normalization integral . emission and absorption of phonons are associated with electronic transitions with energy transfer @xmath78 . we have seen in the previous section that the separation in energy between two consecutive el s is given by @xmath254 [ eq . ( [ omtdef ] ) ] . real _ transitions are generally restricted to el s for which @xmath255 . for the parameters used above for conditions ( [ condr ] ) , the frequency @xmath227 , eq . ( [ omega ] ) , is about @xmath256ghz , whereas the frequencies of the saw s used in experiments vary typically in the range @xmath257mhz @xmath258 1ghz . we therefore conclude that only extended el s for which @xmath259 [ see eq . ( [ omegat ] ) ] contribute to the sound absorption . thus , the matrix elements of the interaction hamiltonian ( [ hdefdet ] ) @xmath260where @xmath261 and @xmath262 denote the initial and final wave functions of the form ( [ psi ] ) , have to be calculated for extended trajectories . this calculation has been performed in ref . . the matrix element , averaged over all trajectories with the same period @xmath182 and the same energy @xmath170 , reads @xmath263where @xmath264 is a numerical factor of order unity . the matrix element is valid under the assumptions @xmath36 , where the exponent @xmath265 follows from @xmath266 with @xmath212 , cf . ( [ ldrela ] ) . clearly , these inequalities imply @xmath31 . for @xmath267 , the matrix element @xmath268 is exponentially small . this implies that transitions occur only within the lowest ll and that transitions to other ll s can be neglected ( @xmath269 ) . it is also assumed that the initial and final states are close to one another in real space : in order that @xmath270 and @xmath271 will overlap , the separation in real space , @xmath272 , should satisfy @xmath273 . the condition @xmath274 is even more restrictive . this can be seen in the following way . as mentioned above , the mean distance in real space between two adjacent el s is given by @xmath275 . hence , the distance between the two states @xmath276 and @xmath74 is of order @xmath277 . the maximum of this expression is found for the largest allowed energy difference @xmath278 . using the definition of @xmath279 in eq . ( [ omegat ] ) and the estimate for @xmath227 in eq . ( [ omega ] ) , the corresponding maximum distance in real space is found to be @xmath280 , cf . the inequalities ( [ condr ] ) . while the sound absorption is due to transitions between extended states , the calculation of the dielectric function ( see the next section ) necessitates also matrix elements between standard el s . ( transitions between a standard el and an extended el are exponentially rare due to their large separation in space . ) since the majority of the el s belongs to the standard ones , one might even expect that the standard el s dominate the dielectric function . this is not the case , as is shown below . noting that for typical phonon wave vectors ( e. g. @xmath281@xmath42 ) , one has @xmath282 , the matrix element for standard el s with @xmath228 can be approximated by @xmath283where the zero - order term in the expansion of the exponential function disappears due to the orthogonality of the two states @xmath276 and @xmath74 . the vector @xmath284 denotes some point in the vicinity of the @xmath276th el . the matrix element on the right - hand - side of eq . ( [ meshort1 ] ) is of order @xmath0 . hence , for transitions between two standard el s @xmath285where it is understood that the el s @xmath276 and @xmath74 are very close in real space and in energy ; otherwise the matrix element is exponentially small . the first condition guarantees the overlap of the wave functions @xmath286 , see eq . ( [ psi ] ) . the second one is necessary to ensure that the integrand of the @xmath177-integration along the perimeter of the el s is not a fast oscillating function . result ( [ meshort2 ] ) agrees essentially with eq . ( [ me1 ] ) replacing there the energy difference @xmath287 and the level spacing @xmath254 by the value @xmath288 appropriate for standard el s . the matrix element ( [ mael ] ) includes the screening of the electron - phonon interaction due to the lattice [ eq . ( [ vertpi ] ) ] . the screening arising from the 2deg can be accounted for by renormalizing the matrix element @xmath289where @xmath4 is the dielectric function of the 2deg . for a nearly half - filled ll , the dielectric function can be calculated assuming linear screening@xcite . that is , the change in the electron density resulting from a small applied potential is proportional to the strength of the perturbing potential . indeed , one can estimate that for the saw intensities used in experiments the electron density oscillates only weakly around its average value , see for example ref . . the assumption of linear screening leads to the general expression @xmath290where @xmath291and @xmath292 is defined in eq . ( [ coninte ] ) . to evaluate @xmath293 explicitly , we transform the sum @xmath294 in eq . ( [ dencor ] ) into @xmath295 , where @xmath296 means @xmath297 . let us first focus on the case of zero temperature , i. e.all levels below the fermi energy @xmath298 are occupied , @xmath299 , whereas all levels above @xmath298 are empty , @xmath300 . then @xmath301 in order to yield an appreciable matrix element , the el s corresponding to @xmath302 and @xmath303 must be close ( in real space and in energy ) to an el with @xmath304 . this ` fermi ' el ( fel ) need not to be an electron state . we can therefore represent the summation over states in eq . ( [ pitzero ] ) as a sum over el s near a certain fel and then sum over all fel s . in the first sum the states are distributed nearly equidistantly with an energy spacing @xmath305 , cf . below . in the summation over fel s we may first sum over the fel s with the same period @xmath182 . since these el s are situated in different regions of the random potential , this summation is equivalent to an average of the matrix element over fel s with the same period . thus , the averaged matrix elements ( [ me1 ] ) and ( [ meshort2 ] ) for extended and standard el s , respectively , can be substituted in eq . ( [ pitzero ] ) . we begin with the contribution of the extended el s to @xmath293 . as we shall see , this is the dominant contribution . it is easy to see that the energy spacing for the relevant states near a fixed fel is given by the value of @xmath254 at the fermi energy . to this end , we have to calculate the change in @xmath279 arising from a change in the energy of the el by at most @xmath306 [ see eq . ( [ me1 ] ) ] . since the frequency @xmath279 for the extended el s is merely a function of @xmath187 , one has @xmath307 , where @xmath308 is the area enclosed by the el . to get the second equality , we have used @xmath309 , with@xcite @xmath310 . the change in the enclosed area is given by @xmath311 and thus @xmath312 . consequently , the sum over el s which are near a given fel can be simplified by introducing an explicit representation for the energies @xmath313the integers @xmath314 and @xmath173 are subject to the restrictions @xmath315 [ see eq . ( [ me1 ] ) ] and @xmath316 . using the representation ( [ energyeps ] ) and the matrix element ( [ me1 ] ) , the double sum over states near one fel in eq . ( [ pitzero ] ) can be reduced to a sum over @xmath317 and one obtains @xmath318where @xmath319 and @xmath320 we have replaced the upper limit in the sum by infinity , since the relevant @xmath321 are of order @xmath322 and the above mentioned restriction for @xmath323 can be neglected . in other words , the el s which contribute significantly are separated in energy by @xmath324 . for extended states , using eq . ( [ omegat ] ) , @xmath325 , and hence the contribution to @xmath326 from a fel is a function of its length alone . as a result the total @xmath326 can be written as a sum over all lengths . using the distribution function @xmath327 given in eq . ( [ fel ] ) , we find @xmath328 where @xmath329 in a finite system of size @xmath243 , @xmath330 has to be replaced by the maximum of @xmath330 and @xmath331 , see the discussion following eq . ( [ feext ] ) . using the explicit form for the distribution function @xmath74 with @xmath219 yields @xmath332 the quantity @xmath333 has a very intuitive interpretation : it is the energy at which the energy level spacing @xmath334 for an el with the critical length is equal to the phonon energy @xmath324 of the saw . in other words , @xmath333 determines the absorption threshold in the sense that real transitions occur only for @xmath335 . this is reflected in the imaginary part of @xmath293 , calculated below . the real and imaginary parts of the function @xmath336 are given by [ hf ] @xmath337@xmath338 where @xmath339 denotes the principal part of the integral . the behavior of @xmath340 in an infinite system is shown in fig . 1 . the @xmath54-axis in fig . 1 has been scaled in terms of @xmath341 corresponding to the dependence of @xmath336 on the fermi energy , see eq . ( [ yfdef ] ) . the limiting behaviors for large and small arguments are given by @xmath342 where @xmath343 . in order to discuss the analytic expressions of @xmath340 we note that the sum in eq . ( [ hfimag ] ) and the integral in eq . ( [ hfreal ] ) are truncated at @xmath321 or @xmath53 of order @xmath344 , as implied by @xmath241 , eq . ( [ feext ] ) . the imaginary part ( [ hfimag ] ) is therefore exponentially small @xmath345 for @xmath346 and of order unity in the opposite case . thus , the sum over @xmath321 increases with decreasing @xmath54 and approaches its maximum as @xmath54 goes to zero . in a finite system , @xmath54 is restricted from below by @xmath347 imposing an upper limit @xmath348 on the sum . this leads to a smaller maximum value of im@xmath336 . as for the real part of @xmath340 , we are able to study the limiting cases . for both @xmath349 and @xmath350 , re@xmath340 goes to zero according to a power law . in the intermediate region @xmath351 but not @xmath350 , re@xmath336 is slowly varying and of order unity . in a finite system , the real part approaches a small but finite value as @xmath54 goes to @xmath347 . up to this point , the evaluation of @xmath293 , eq . ( [ dencor ] ) , has been performed for zero temperature , @xmath352 . the calculation of @xmath293 for finite temperatures such that @xmath353 can be done along the same lines . therefore , we give only a brief description . the calculations for @xmath352 have shown that the real and the imaginary parts of @xmath336 reach a value of order unity when the fermi energy becomes of order @xmath333 . this energy range corresponds to the contribution of el s with a length @xmath354 and an energy level spacing of order @xmath324 to the transition processes . therefore these extended el s give the dominant contributions to @xmath293 , eq . ( [ dencor ] ) . this remains true for finite temperatures . however , now the states @xmath276 and @xmath74 need not be in the immediate vicinity of the fel s . instead , we can fix some initial state @xmath276 and consider transitions to final states above , @xmath355 , and below , @xmath356 , the chosen one ( essentially in a range @xmath357 around @xmath298 ) . to do this , we use the representation ( [ energyeps ] ) where @xmath279 now refers to the energy @xmath302 . expanding the fermi distribution @xmath358 in eq.([dencor ] ) around @xmath302 leads to @xmath359 ( for the relevant transitions @xmath287 is much smaller than @xmath357 ) . the sum over the energies @xmath303 is reduced to twice the expression ( [ sums ] ) . hence @xmath360the sum over the initial states @xmath276 comprises a summation over all el s with the same energy @xmath302 but with different lengths @xmath187 , and a summation over @xmath302 . the first sum can be again replaced by an integral using the distribution function @xmath327 , eq . ( [ fel ] ) . then , the order of summation and integration is inverted to obtain @xmath361the condition @xmath362 ensures that only those initial states which possess a critical length @xmath229 equal to or larger than @xmath363 are included . here @xmath364 , see eq . ( [ areac ] ) . however , the dominant contributions to @xmath365 follow from a particular group of el s , rendering this condition unnecessary . taking also into account that the relevant el s have an energy level spacing which is small compared to the thermal energy @xmath366 , and that the number of states per energy interval is given by @xmath367 , the sum over @xmath302 can be replaced by the integral given on the right - hand - side of eq . ( [ tnonzero ] ) . the limits of integration have been extended with negligible error . the resulting integral over @xmath187 coincides with the right - hand - side of eq . ( [ pzero2 ] ) except for the value of the energy : instead of the fixed fermi energy @xmath298 there appears now the variable @xmath302 . thus , we finally arrive at @xmath368where we have dropped the index @xmath276 of @xmath302 . this equation shows that a finite temperature leads to an average of the @xmath352 result over energies within an interval of order @xmath357 around the fermi level . since the function @xmath369 varies on the scale @xmath333 , finite temperature effects are negligible if @xmath370 . this is the condition for eq . ( [ pzero2 ] ) to hold . for @xmath371 , the width of @xmath372 as function of the fermi energy increases with temperature , i. e. the behavior of @xmath293 deviates substantially from the @xmath352 result . in the following we assume @xmath373 . substituting the @xmath352 result for @xmath372 [ eq . ( [ pzero2 ] ) , which is based on transitions between extended el s ] in eq . ( [ dfdef ] ) yields for the dielectric function @xmath374 and , for the renormalization of the matrix element in eq.([renor ] ) , @xmath375 the contribution to the dielectric function resulting from standard el s is derived below , see eq . ( [ chishort ] ) . the comparison of that result with eqs . ( [ dfgen ] ) and ( [ dfspec ] ) shows that the dielectric function is essentially given by the contribution due to extended el s , whereas the influence of transitions between standard el s can be neglected . we consider therefore eq . ( [ dfspec ] ) as the final result for @xmath376 . the dielectric function ( [ dfspec ] ) renormalizes the matrix element ( [ renor ] ) via the expression @xmath377 . the dependence of @xmath378 on the fermi energy is given by @xmath379 , eqs . ( [ hf ] ) . the latter is of order unity for @xmath335 and decreases for larger values of the fermi energy according to the power law @xmath380 , @xmath381 . thus , the magnitude of @xmath382 , eq . ( [ dfspec ] ) , is determined by the large dimensionless parameter @xmath383 ( @xmath384 for gaas ) . this is the ratio of the electrostatic energy of two electrons a distance @xmath98 apart and the energy of a surface phonon , @xmath385 . let us now show that the contribution of the standard el s to the dielectric function is negligible . we start afresh from the expression ( [ pitzero ] ) for @xmath293 , substituting the matrix element ( [ meshort2 ] ) for transitions between standard el s . the energy difference between two standard el s is of order @xmath306 , eq . ( [ omegat ] ) , i. e. much larger than @xmath324 . the latter can thus be neglected in comparison with @xmath386 in eq.([pitzero ] ) . then the imaginary term @xmath387 can be dropped , as no real transitions can occur . the sum over all final states @xmath74 leads merely to a factor of order one , since only a few el s in the immediate vicinity of the initial el contribute to the matrix element ( [ meshort2 ] ) . the remaining sum over the initial states counts the standard el s which are just below the fermi level . the number of these el s is essentially the number of all fel s , because the number of very short ( @xmath388 ) and very long ( @xmath220 ) el s is negligibly small for fermi energies near the center of the ll . the required quantity may therefore be deduced from eq . ( [ normfe ] ) which states that the total length of all el s is , up to a numerical factor , equal to @xmath389 . since the mean length of all el s is of order @xmath0 , the number of fel s in a system of size @xmath243 is of order @xmath390 . collecting these results , we obtain the contribution due to standard el s @xmath391it can be shown that this estimate is valid independent of the ratio @xmath392 as long as max@xmath393 . the comparison of eq . ( [ chishort ] ) with eq . ( [ dfgen ] ) shows that the contribution of the standard el s to the dielectric function is @xmath394 times smaller than the term resulting from the extended el s . it is instructive to consider briefly an alternative derivation of the dielectric function which reproduces correctly the order of magnitude @xmath395 . this derivation relies on the fact that the motion of an electron on a fractal trajectory can be considered as a self - avoiding random walk with single steps of length @xmath0 . in fact , the relation between the diameter and the length of an extended el is similar to what one would expect for a simple random walk , cf . ( [ ldrela ] ) . for the diffusive regime , the density correlator @xmath293 , eq . ( [ dencor ] ) , is given by @xmath396where @xmath397 is the diffusion constant and @xmath398 the density of states at the fermi level . in our case , we can assume @xmath399 . the density of states of the ll is given by @xmath400 for @xmath401 . substituting these quantities into eq . ( [ dfdef ] ) yields @xmath402where the term @xmath403 has been neglected compared to @xmath404 in the denominator of the density correlator . interestingly , this approach predicts an essentially imaginary result for @xmath405 which agrees with the behavior of eq . ( [ dfspec ] ) for @xmath406 , i. e. in the case when some the el s become arbitrarily long . the intensity of the saw decreases due to absorption of phonons by the 2deg with the distance @xmath53 as exp@xmath407 . the attenuation coefficient @xmath8 can be expressed in terms of the sound velocity @xmath29 [ cf . ( [ omega ] ) ] and the life time @xmath408 as @xmath409 , where @xmath408 is defined by the rate equation @xmath410here @xmath411 is the phonon occupation number . the net change in @xmath411 is given by @xmath412 , \end{aligned}\]]where @xmath413 is the fermi distribution function and @xmath414 are the unscreened matrix elements ( [ mael ] ) for emission or absorption of a phonon with wave vector @xmath415 . for a saw induced by interdigital transducers , the phonon occupation number @xmath411 is macroscopically large . the difference between @xmath416 and @xmath411 is therefore negligible . combining eqs . ( [ rate1 ] ) and ( [ rate2 ] ) yields @xmath417replacing the @xmath418-function by the imaginary part of @xmath419^{-1}$ ] , we find @xmath420where @xmath293 is defined by eq . ( [ dencor ] ) . using the zero - temperature results for @xmath293 and the dielectric function , eqs . ( [ pzero2 ] ) and ( [ dfspec ] ) , respectively , as well as the relation between the life time @xmath408 and the attenuation coefficient , we find @xmath421with @xmath422where the term 1 in the expression ( [ dfspec ] ) for @xmath382 has been neglected . the function @xmath423 is defined such that @xmath424 [ since re@xmath425 , cf . ( [ limith ] ) ] , i. e. , @xmath426 coincides with the attenuation coefficient at the center of the ll , @xmath427 . we begin with the discussion of the magnitude of @xmath426 and consider the function @xmath423 afterwards . substituting the expressions ( [ tauda ] ) and ( [ taupa ] ) for the interaction vertices in eqs . ( [ gq ] ) yields [ gmag ] @xmath428@xmath429 that is , despite the fractal structure of the extended el s on which these results are based , the frequency dependence of the magnitude of @xmath8 is simple and is not characterized by scaling exponents . moreover , @xmath426 is independent of the magnetic field and the parameters @xmath0 and @xmath1 of the random potential . the numerical values on the right - hand - side of eqs . ( [ gmag ] ) have been calculated replacing the parameters @xmath430 etc.by theirs values given in sec . ii and assuming @xmath431 . for a finite system , im@xmath432 [ see eqs . ( [ limith ] ) ] has to be replaced by im@xmath433 , see the discussion following eq . ( [ yfdef ] ) , leading to an increase of the attenuation coefficient at the center of the ll . the function @xmath434 in eq . ( [ gscr ] ) accounts for the dependence of @xmath8 on the fermi energy ( or the filling factor @xmath435 or the magnetic field @xmath22 ) . this dependence is determined by the ratio of @xmath436 and the energy @xmath437 as follows . the absorption of the saw is very small when the fermi energy is far from the center of the ll @xmath438 , i. e. @xmath439 . a strong increase of @xmath440 and , hence , of @xmath8 occurs when @xmath436 is reduced to @xmath441 . in this region the number of occupied extended el s with an energy level spacing @xmath442 , eq . ( [ omegat ] ) , undergoes the change from an exponentially small quantity to some power - law function of @xmath443 . ( nevertheless , the number of these states is negligible compared to the majority of el s with @xmath228 . ) a further rise of the absorption is prevented by the enhanced screening @xmath444 at @xmath445 which even reduces @xmath8 as the fermi energy goes to zero . this results in a shallow double - peak structure with a cusp at the center of the ll . in fact , if we use the limiting forms of @xmath446 , eqs . ( [ limith ] ) , we find @xmath447 . we believe therefore that the double - peak structure of @xmath448 is independent of the function @xmath449 used to describe the exponential cut - off of the extended el s , see the discussion following eq . ( [ feext ] ) . the maxima of @xmath448 are located near @xmath450 , see fig . it is clear , however , that such a particular feature as the cusp has to be considered with caution , for it is exclusively based on the quasiclassical model for the electron states . quantum tunneling between critical trajectories may modify this result . it is worth to note that the tunneling band of width @xcite @xmath227 [ eq . ( [ omega ] ) ] around @xmath209 [ cf . the discussion after eq . ( [ cond ] ) ] is narrow compared to the characteristic energy range @xmath333 ; indeed , @xmath451 . one may speculate that the absorption coefficient is only weakly affected by quantum tunneling . indeed , most of the el s contributing to @xmath8 can not be connected by low saddle points with transmission coefficients of order unity , and so quantum tunneling in - between them is insignificant . to simplify the estimates , we rewrite @xmath333 in the form @xmath452as discussed in sec . v , the zero - temperature result ( [ pzero2 ] ) for @xmath293 remains valid for finite temperatures such that @xmath370 . this is also the condition for eq . ( [ gscr ] ) to hold . using the values for @xmath1 and @xmath0 given above and @xmath453mhz , we obtain @xmath454k . for temperatures of the order of or larger than @xmath333 , the attenuation coefficient is found from eq . ( [ rate4 ] ) using expression ( [ pzerot ] ) in the calculation of the dielectric function and im@xmath293 . two results for @xmath8 at finite temperatures are shown in fig.2 . with increasing temperature the minimum of the attenuation near the center of the ll is reduced and the absorption peak becomes broader . the increasing magnitude of @xmath8 results from the significant broadening of the imaginary part of @xmath293 and the reduced screening , see eq . ( [ rate4 ] ) . for @xmath371 , the magnitude of @xmath8 and the width of the absorption region are strongly temperature dependent . the dependence of @xmath8 on the saw frequency @xmath6 is shown in fig . the curve is calculated for the low temperature regime @xmath370 and the piezoelectric electron - phonon interaction . the attenuation coefficient has been written in the form @xmath455 , with @xmath456 , @xmath457 . ( note that @xmath458 does not depend on frequency . ) the fermi energy is fixed to some value @xmath401 and defines the smallest level spacing @xmath459 for extended el s . consequently , @xmath460 marks the onset of strong saw attenuation . for high frequencies @xmath461 , the attenuation coefficient increases linearly with frequency . this is just the behavior predicted by the classical description of sound absorption for piezoelectric interaction , see eq . ( [ gsig ] ) below . for @xmath462 , the width of the absorption region is determined by @xmath463 . this is merely the condition for real transitions to occur and is neither associated with the interaction vertices @xmath144 nor is a consequence of the matrix element ( [ me1 ] ) whose derivation is based on the particular assumption @xmath3 . we believe therefore that this result applies to other half - integer filling factors @xmath464 as well . to express the relation @xmath463 in terms of the filling factor @xmath465 , we write the electron density @xmath173 as an integral over the gaussian density of states , @xmath466and the fermi distribution function @xmath74 , @xmath467and expand around the center of the ll with respect to @xmath468 . this gives for @xmath469 @xmath470then , the width of the absorption region is obtained as @xmath471the exponent is given by @xmath472 . this value agrees with the exponent @xmath473 which determines the shrinking of the peaks in the longitudinal conductivity @xcite @xmath474 as the temperature @xmath357 goes to zero , @xmath475 . in our case the broadening of the absorption peak arises from the frequency @xmath6 . in this sense , @xmath476 may be considered as an effective temperature which replaces the real temperature @xmath357 . frequency scaling in the integer quantum hall regime has been observed in microwave experiments @xcite . for spin - split ll s , the width of the peaks in re@xmath474 corresponding to different ll s was found to scale as @xmath477 , with @xmath478 . due to the drift velocity @xmath479 , the width ( [ dnu ] ) depends weakly on the absolute value of the magnetic field . in terms of the filling factor , eq . ( [ dnu ] ) can be written in the form @xmath480 . thus , @xmath481 is smaller for higher ( half - integer ) filling factors @xmath464 . the width of the absorption region scales with the phonon wave vector as @xmath482 . in contrast , in the fractional quantum hall regime , the width increases linearly with @xmath90 for @xmath20 . this linear dependence is derived within the composite fermion model@xcite and is well confirmed experimentally @xcite . the absorption of saw s in the integer quantum hall regime has also been studied in ref . . these authors determine first the ac - conductivity of the 2deg which is then related to the attenuation coefficient using the equation @xmath483where @xmath484 represents the effective piezoelectric coupling constant ( @xmath485 for gaas@xcite ) and @xmath486 , @xmath487 and @xmath488 . [ note that eq . ( [ gsig ] ) can be obtained from eq.([gscr ] ) writing the dielectric function of the 2deg in the form @xmath489 . ] assuming that @xmath490 for sufficiently high frequencies@xcite , eq . ( [ gsig ] ) reduces to @xmath491 . that is , in this case , the sound absorption and the longitudinal conductivity are related such that the width and the shape of their peaks as function of @xmath464 are identical . the calculation of the ac - conductivity in ref . is based on the concept of variable - range hopping between pairs of localized states . for @xmath492 , the absorption of saw s is due to resonant phononless transitions of the electrons from one site of a pair to the other . this mechanism is strongly affected by the electron - electron interaction @xcite . the width of the absorption peak at half - integer filling factors was found to be @xmath493 is the scaling exponent of the localization length @xmath494and @xmath495 is assumed to be of the order of the magnetic length . [ note the differences between the last equation and the semiclassical definition of the localization length in eq . ( [ dcrit ] ) . ] the result of ref . , eq . ( [ nuas ] ) , agrees with our result , eq . ( [ dnu ] ) , in both the numerical value of the exponent and the dependence on @xmath90 . however , the width @xmath481 in eq . ( [ nuas ] ) exhibits a different dependence on the magnetic field , namely @xmath496 in contrast to @xmath497 predicted by eq.([dnu ] ) . the authors of ref . did not give a definite description of the shape of the absorption peak but rather suggested two scenarios which eventually lead to a flat peak with a broad maximum or a double - peak , respectively . our results support the latter one , see fig . 2 . we have calculated the dielectric function @xmath4 and the attenuation coefficient @xmath8 of a surface acoustic wave for a 2deg in a smooth random potential ( with amplitude @xmath1 and correlation length @xmath0 ) and a strong magnetic field corresponding to a filling factor @xmath435 close to @xmath498 . both quantities become independent of temperature as the temperature is reduced below a frequency - dependent value @xmath499 , where @xmath9 , @xmath500 and @xmath222 is the average drift velocity of the electrons on the equipotential lines of the random potential . in this low temperature , high frequency regime ( e. g. @xmath501k for @xmath502mhz ) , im@xmath4 and @xmath8 are only appreciable when @xmath298 is within a narrow region around the center of the landau level , and re@xmath4 decreases according to a power law with increasing distance from the center . in particular , the attenuation of the saw is exponentially small except for a region whose width @xmath503 . this scaling is non - universal because @xmath504 depends on the absolute value of the magnetic field , see eq . ( [ dnu ] ) . the dependence of @xmath8 on the fermi energy ( or the filling factor ) yields a double - peak which is centered at the filling factor @xmath505 , cf . fig . 2 . the minimum of the absorption at @xmath505 results from the enhanced screening due to the 2deg , i. e. , from the large magnitude of the dielectric function @xmath506 , where @xmath292 is the average of the dielectric constants of gaas and vacuum , and @xmath29 is the sound velocity . the double - peak in @xmath8 is most pronounced for an infinite system where the critical diameter @xmath230 , eq . ( [ dcrit ] ) , of the equipotential lines of the random potential is allowed to take on arbitrarily large values . a real system of size @xmath243 restricts the diameter to @xmath244 resulting in an increase of the attenuation coefficient near the center of the landau level . while this effect is weak for a macroscopic sample size , a similar but more pronounced effect may arise from a non - uniform electron density associated with a spatially varying filling factor . in the high temperature , low frequency regime , the dielectric function decreases with rising temperature leading to an increase of the magnitude of the attenuation coefficient and a significant increase of the width of the absorption region around @xmath505 . financial support by the german - israeli foundation is gratefully acknowledged . we thank j. hajdu and d. polyakov for valuable discussions and comments . one of us ( a. k. ) thanks the deutsche forschungsgemeinschaft for financial support and b. zingermann for a discussion of some properties of gaussian distributions .

a well - known consequence of the ergodic decomposition theorem is that the space of invariant probability measures of a topological dynamical system , endowed with the weak topology , is a non - empty metrizable choquet simplex . we show that every non - empty metrizable choquet simplex arises as the space of invariant probability measures on the post - critical set of a logistic map . here , the post - critical set of a logistic map is the-limit set of its unique critical point . in fact we show the logistic map can be taken in such a way that its post - critical set is a cantor set where is minimal , and such that each invariant probability measure on this set has zero lyapunov exponent , and is an equilibrium state for the potential . logistic map , post - critical set , invariant measures , choquet simplices , minimal cantor system , generalized odometer 37e05 , 37a99 , 37b10 , 54h20

0901.3835

a well - known consequence of the ergodic decomposition theorem is that the space of invariant probability measures of a topological dynamical system , endowed with the weak@xmath0 topology , is a metrizable choquet simplex .. ] the purpose of this paper is to show that every non - empty metrizable choquet simplex arises in this way within the _ logistic family _ of maps @xmath4}$ ] , where for each parameter @xmath5 in @xmath6 $ ] the _ logistic map _ @xmath7 \to [ 0 , 1]$ ] is defined by , @xmath8 to make a more precise statement note that @xmath9 is the unique point in @xmath10 $ ] at which the derivative of @xmath11 vanishes . we call @xmath9 the _ critical point _ of @xmath12 , and its @xmath1-limit set is called the _ post - critical set _ of @xmath11 . it is a compact set that is forward invariant by @xmath11 . the following is our main result . recall that for a compact topological space @xmath13 a continuous map @xmath14 is _ minimal _ , if every forward orbit of @xmath15 is dense in @xmath13 . * main theorem . * _ for each non - empty metrizable choquet simplex @xmath16 there is a parameter @xmath17 $ ] such that the post - critical set of the logistic map @xmath11 is a cantor set , the restriction of @xmath11 to this set is minimal , and such that the space of invariant probability measures supported by this set , endowed with weak@xmath0 topology , is affine homeomorphic to @xmath16 . _ the first result of this kind was shown by downarowicz in @xcite , who showed that every non - empty metrizable choquet simplex arises , up to an affine homeomorphism , as the space of invariant probability measures of a `` minimal cantor system '' ; that is , a dynamical system generated by a minimal homeomorphism of a cantor set . in fact he showed that the minimal cantor system can be taken as a `` 0 - 1 toeplitz flow '' : a special type of subshift of @xmath18 . see also @xcite for a different approach to this result , and see @xcite for an analogous result in the case of actions of @xmath19 . the following corollary is a direct consequence of the main theorem and of the fact that for each non - empty polish space @xmath20 there is a metrizable choquet simplex whose set of extreme points is homeomorphic to @xmath20 , see for example @xcite . [ c : extreme realization ] for each non - empty polish space @xmath20 there is a parameter @xmath17 $ ] such that the post - critical set of the logistic map @xmath11 is a cantor set , the restriction of @xmath11 to this set is minimal , and such that the space of ergodic and invariant probability measures supported by this set , endowed with weak@xmath0 topology , is homeomorphic to @xmath20 . the special case where the polish space @xmath20 is compact and totally disconnected is precisely ( * ? * main theorem ) . the first result in this direction was shown by bruin , who gave an example of a parameter @xmath17 $ ] such that the post - critical set of @xmath12 is a cantor set where @xmath12 is minimal , but not uniquely ergodic ( * ? ? ? * theorem 4 ) . the proof of the main theorem is based on the tools developed by bruin in @xcite , and by bruin , keller and st . pierre in @xcite . one of the interesting features of the main theorem , in constrast with the other realization results mentioned above , is that the systems we consider have a natural differentiable structure . it turns out that , for the parameters @xmath17 $ ] given by ( the proof of ) the main theorem , the invariant measures supported by the post - critical set of @xmath12 correspond precisely to those invariant measures @xmath21 of @xmath12 whose lyapunov exponent @xmath22 vanishes ( * ? ? ? * lemma 21 ) . it also turns out that every invariant probability measure supported on the post - critical set of @xmath12 is an `` equilibrium state of @xmath12 for the potential @xmath23 '' . that is , if for each invariant measure @xmath21 we denote by @xmath24 its measure theoretic entropy , then the supremum @xmath25 is attained at each invariant probability measure supported by the post - critical set of @xmath12 , see ( * ? ? ? * lemma 21 ) . we thus obtain the following corollary of the main theorem . [ c : ergodic theory interval ] for each non - empty metrizable choquet simplex @xmath16 there is a parameter @xmath17 $ ] verifying the conclusions of the main theorem , and such that in addition the space of invariant probability measures of @xmath12 ( resp . equilibrium states of @xmath12 for the potential @xmath26 ) that are of zero lyapunov exponent , endowed with the weak@xmath0 topology , is affine homeomorphic to @xmath16 . this result is in sharp contrast with the fact that for a logistic map there can be at most one ergodic equilibrium state whose lyapunov exponent is strictly positive . for future reference we state an holomorphic version of corollary [ c : ergodic theory interval ] , shown in appendix [ a : ergodic theory ] . for a complex parameter @xmath27 denote by @xmath28 the quadratic polynomial defined by @xmath29 viewed as a dynamical system acting on @xmath30 . [ c : ergodic theory holomorphic ] for each non - empty metrizable choquet simplex @xmath16 there is a parameter @xmath17 $ ] verifying the conclusions of the main theorem , and such that in addition , if we denote by @xmath31 the hausdorff dimension of the julia set of @xmath28 , then the space of invariant probability measures of @xmath28 ( resp . equilibrium states of @xmath28 for the potential @xmath32 ) that are of zero lyapunov exponent , endowed with the weak@xmath0 topology , is affine homeomorphic to @xmath16 . we end this introduction by stating some questions that arise naturally from the main theorem . to do this , for each @xmath17 $ ] we will denote by @xmath33 the post - critical set of @xmath12 . given a compact metrizable topological space @xmath13 and a continuous map @xmath14 , the main theorem implies that there is a parameter @xmath17 $ ] and an affine homeomorphism @xmath34 between the space of invariant probability measures supported on @xmath33 . it is thus natural to ask whether the parameter @xmath17 $ ] and @xmath34 can be chosen in such a way that @xmath34 is induced by a continuous map between @xmath33 and @xmath13 . more precisely , the question is if @xmath17 $ ] and @xmath34 can be chosen in such a way that there is a continuous map @xmath35 such that for each invariant probability measure @xmath21 supported by @xmath33 we have @xmath36 . this type of problem is very well understood in the setting of minimal cantor systems : giordano , putnam , and skau have shown in @xcite that for two minimal cantor systems @xmath37 and @xmath38 there exists a homeomorphism @xmath39 that induces an affine homeomorphism between the corresponding spaces of invariant probability measures , if , and only if , @xmath37 and @xmath40 are `` orbit equivalent '' : there is a homeomorphism between @xmath13 and @xmath41 mapping each orbit of @xmath15 to an orbit of @xmath42 . furthermore , to each minimal cantor system @xmath37 one can associate a dimension group that is a complete invariant for the orbit equivalence relation @xcite : two minimal cantor systems are orbit equivalent if , and only if , the corresponding dimension groups are isomorphic as ordered groups with unit . and taking values in @xmath43 , by the subgroup of functions whose integral with respect to each invariant measure vanishes ; the positive cone is the set of those classes containing a function taking values in @xmath44 , and the unit is the class of the constant function equal to 1 . see for example ( * ? ? ? * theorem 1.13 ) . ] it is thus natural to look for a special class of minimal cantor systems realizing all of the orbit equivalent classes . since the dimension group associated to each toeplitz flow contains the dimension group of an odometer as a subgroup ( * ? ? ? * section 4.1 ) , the class of toeplitz flows is not sufficient to realize all orbit equivalence classes , in spite of the fact that this class realizes all the non - empty metrizable choquet simplices as sets of invariant probability measures @xcite . so the question remains whether minimal post - critical sets of logistic maps realize all orbit equivalence classes . in order to formulate a precise question we will consider natural extensions to stay in the class of minimal cantor systems , and use the generalized odometer associated to a kneading map , see [ ss : unimodal ] , [ ss : kneading odometer ] for definitions . does every orbit equivalence class contain the natural extension of a generalized odometer associated to a kneading map ? does every uniquely ergodic orbit equivalence class contain the natural extension of a generalized odometer associated to a kneading map ? it is well - known that every odometer can be realized , up to a homeomorphism , as the post - critical set of an infinitely renormalizable logistic map , see also @xcite . in [ ss : example ] we give an example of a uniquely ergodic generalized odometer associated to a kneading map , whose natural extension is not orbit equivalent to an odometer , nor to a toeplitz flow . although corollary [ c : extreme realization ] is stronger than ( * ? ? * main theorem ) , we use this last result in the proof of the main theorem to deal with case of finite dimensional choquet simplices . we have stated the main theorem and corollary [ c : extreme realization ] for the logistic family for simplicity . we show that an analogous statement holds for each full family of unimodal maps , as well as for the family of symmetric tent maps . see [ ss : unimodal ] for definitions . in fact , for each infinitely dimensional metrizable choquet simplex we construct kneading map @xmath45 such that the conclusions of the main theorem hold for each unimodal map whose kneading map is @xmath45 , see [ s : doubly resonant kneading ] . furthermore the kneading map satisfies for every @xmath46 the inequality @xmath47 ( part 1 of lemma [ l : doubly resonant kneading ] ) , and therefore every full family of unimodal maps , as well as the family of symmetric tent maps , contains a unimodal map whose kneading map is @xmath45 . for the case of finite dimensional choquet simplices see ( * ? ? ? * remark 1 ) . similarly , it follows from ( * ? ? ? * lemma 21 ) that corollary [ c : ergodic theory interval ] holds for every full family of -unimodal maps . see @xcite for a survey on realization results concerning toeplitz flows . see @xcite for the realization of some concrete simplices as the space of invariant measures of minimal cantor systems . in this section we explain the strategy of the proof of the main theorem and simultaneously describe the organization of the paper . we only deal with infinitely dimensional choquet simplices , the finite dimensional case being covered by ( * ? ? ? * main theorem ) . we use a result of lazar and lindenstrauss that characterizes infinite dimensional metrizable choquet simplices as inverse limits of stochastic matrices , see theorem [ t : choquet as inverse limit ] in [ ss : choquet simplices ] . we describe the logistic maps in the main theorem through their associated `` kneading map '' , see [ ss : unimodal ] for the definition of kneading map and further background on unimodal maps . in fact , the conclusions of the main theorem are valid for each unimodal map having the same kneading map as @xmath12 . to ensure that the post - critical set is a cantor set where the unimodal map is minimal , it is enough to require that the kneading map diverges to @xmath48 ( proposition [ p : continuous unimodal ] ) . in [ ss : doubly resonant kneading ] we introduce a class of kneading maps that diverge to @xmath49 . in [ ss : proof of main theorem ] we state a result describing the space of invariant measures supported on the post - critical set of a unimodal map with a kneading map in this class ( theorem [ t : invariant of doubly resonant ] ) . in [ ss : proof of main theorem ] we also give a proof of the main theorem assuming theorem [ t : invariant of doubly resonant ] . in [ ss : kneading odometer ] we recall the definition and some properties of the generalized odometer associated to a kneading map , that was introduced in @xcite . in [ ss : measure isomorphism ] we show that for a unimodal map whose kneading map @xmath45 is as in [ ss : doubly resonant kneading ] , the space of invariant probability measures supported by the post - critical set is affine homeomorphic to that of the generalized odometer associated to @xmath45 . in turn , this space is affine homeomorphic to the corresponding space of the bratteli - vershik system associated to @xmath45 , introduced in @xcite ; see ( * ? ? ? * proposition 2 ) or theorem [ t : reduction to bratteli ] in [ ss : kneading bratteli - vershik ] . the advantage of this last space is that it can be described explicitly as an inverse limit of some `` transition matrices '' , see [ ss : bratteli - vershik ] . we calculate the transition matrices and some of their products in [ ss : transition matrices ] , and give the proof of theorem [ t : invariant of doubly resonant ] in [ ss : proof of invariant of doubly resonant ] . in [ ss : example ] we give an example of a uniquely ergodic generalized odometer associated to a kneading map whose natural extension is not orbit equivalent to a toeplitz flow . in appendix [ a : ergodic theory ] we give the proof of corollary [ c : ergodic theory holomorphic ] . we are grateful with christian skau for his help with dimension groups of toeplitz flows , and an anonymous referee for his / her suggestions to improve the presentation of the paper . after fixing some notation in [ ss : linear algebra ] , we review some concepts and results about choquet simplices ( [ ss : choquet simplices ] ) , and unimodal maps ( [ ss : unimodal ] ) . throughout the rest of this article @xmath50 denotes the ring of strictly positive integers and @xmath51 . we will use the interval notation for subsets of @xmath44 : for @xmath52 such that @xmath53 , we put @xmath54 { : = } \ { k \in { \mathbb{n}}_0 \mid n \le k \le n ' \},\ ] ] and we put @xmath55 { : = } \emptyset$ ] when @xmath56 . given a non - empty finite set @xmath57 , for each @xmath58 we denote by @xmath59 the vector having all of its coordinates equal to @xmath60 , except for the coordinate corresponding to @xmath61 that is equal to @xmath62 . notice in particular that @xmath63 is a base of @xmath64 . furthermore we will denote by @xmath65 the _ unit simplex _ in @xmath64 , which is defined as the ( closed ) convex hull of @xmath63 in @xmath64 , and by @xmath66 the norm on @xmath64 defined by @xmath67 . observe that @xmath66 is constant equal to 1 on @xmath68 . given non - empty finite sets @xmath69 denote by @xmath70 the group of matrices whose entries are real and indexed by @xmath71 . for a matrix @xmath72 we denote by @xmath73 the transpose of @xmath74 , and for @xmath75 we denote by @xmath76 the corresponding entry of @xmath74 , and by @xmath77 the corresponding column vector of @xmath74 . given column vectors @xmath78 in @xmath64 we denote by @xmath79 the matrix in @xmath70 whose column vector corresponding to the coordinate @xmath80 is equal to @xmath81 . we say that a matrix @xmath74 is ( left ) _ stochastic _ if all of its entries are non - negative and if the sum of all the entries in each column is equal to 1 . observe that a stochastic matrix in @xmath70 maps @xmath82 into @xmath68 , and that the product of stochastic matrices is stochastic . [ l : non expanding ] let @xmath69 be non - empty finite sets and let @xmath72 be a stochastic matrix . then for each @xmath83 we have @xmath84 putting @xmath85 and @xmath86 , we have @xmath87 a compact , convex , and metrizable subset @xmath16 of a locally convex real vector space is said to be a ( metrizable ) _ choquet simplex _ , if for each @xmath88 there is a unique probability measure @xmath21 that is supported on the set of extreme points of @xmath16 , and such that @xmath89 . see for example @xcite for several characterizations of choquet simplices . in the proof of the main theorem we will make use of the following characterization of infinite dimensional metrizable choquet simplices . [ t : choquet as inverse limit ] given an infinite dimensional choquet simplex @xmath16 , for each @xmath90 there is a surjective affine map @xmath91 } \to \delta_{[0 , n]}$ ] such that @xmath92 } , a_n)$ ] is affine homeomorphic to @xmath16 . the following lemma is a simple consequence of the previous theorem . we will say that a sequence @xmath93 as in the theorem is _ normalized _ , if for every @xmath90 and @xmath94 $ ] we have @xmath95 . [ l : normalized choquet ] for every infinite dimensional metrizable choquet simplex @xmath16 there is a normalized sequence of linear maps @xmath93 such that @xmath96 } , a_n)$ ] is affine homeomorphic to @xmath16 . let @xmath97 be a sequence of affine maps given by theorem [ t : choquet as inverse limit ] . for each @xmath90 define a permutation @xmath98 of @xmath99 $ ] by induction as follows . let @xmath100 be the identity , and suppose that for some @xmath90 the permutation @xmath98 is already defined . since @xmath101 maps @xmath102}$ ] surjectively onto @xmath103}$ ] , for each @xmath94 $ ] there is @xmath104 $ ] such that @xmath105 . equivalently , there is a map @xmath106 \to [ 0 , n + 1]$ ] such that for each @xmath94 $ ] we have @xmath107 . let @xmath108 be the unique permutation of @xmath109 $ ] such that for each @xmath94 $ ] we have @xmath110 . for each @xmath111 let @xmath112 } \to { \mathbb{r}}^{[0 , n]}$ ] be the linear map so that for each @xmath94 $ ] we have @xmath113 . then , by the definition of @xmath114 it follows that for each @xmath90 the linear map @xmath115 maps @xmath102}$ ] surjectively onto @xmath103}$ ] , and that for every @xmath94 $ ] we have @xmath116 . therefore the sequence of linear maps @xmath93 is normalized , and @xmath117 induces a linear homeomorphism between @xmath92 } , { \widetilde{a}}_n)$ ] and @xmath92 } , a_n)$ ] . we end this section with the following general lemma . [ l : equivalence ] for each @xmath90 let @xmath118 } \to { \mathbb{r}}^{[0 , n]}$ ] be stochastic matrices such that @xmath119 } \ } < + \infty.\ ] ] then the inverse limits @xmath120 } , a_n)$ ] and @xmath120 } , b_n)$ ] are affine homeomorphic . in particular , if both @xmath93 and @xmath121 are normalized and @xmath122 then the inverse limits @xmath120 } , a_n)$ ] and @xmath120 } , b_n)$ ] are affine homeomorphic . let @xmath123},a_n).$ ] for each @xmath124 such that @xmath125 define @xmath126}.\ ] ] in particular we have @xmath127 . when @xmath128 we have @xmath129 from lemma [ l : non expanding ] we get @xmath130 and since @xmath131 } : \|a_nv - b_nv\|_1\},\ ] ] we deduce , after an induction argument , that @xmath132 } \}.\end{aligned}\ ] ] by hypothesis and from equation ( [ eq3 ] ) , we deduce that for a fixed @xmath90 , sequence @xmath133 is a cauchy sequence in @xmath134}$ ] . we denote by @xmath135 its limit . observe that for each @xmath90 , the sequence @xmath136 converges to both @xmath137 and @xmath135 . this implies that @xmath138},b_n).$ ] thus the transformation @xmath139},a_n ) \to \varprojlim_{n}(\delta_{[0,n+1]},b_n)\ ] ] given by @xmath140 is well defined . this map is clearly affine . we will show that @xmath34 is a homeomorphism . in order to verify that @xmath34 is continuous , we just need to show that for each @xmath141 the map @xmath142 is continuous . as @xmath92 } , a_n)$ ] is metrizable we just need to show that @xmath142 is sequentially continuous . to do this , fix @xmath143 and consider a sequence @xmath144 in @xmath145},a_n)$ ] that converges to @xmath146 . for each @xmath147 such that @xmath148 we have @xmath149 we choose @xmath150 sufficiently large so that @xmath151 and @xmath111 such that @xmath152 . we get @xmath153 this shows the continuity of @xmath142 , and hence that of @xmath34 . to show that @xmath34 is a homeomorphism we define in a similar way , @xmath154},b_n ) \to \varprojlim_{n}(\delta_{[0,n+1]},a_n).\ ] ] observe that by lemma [ l : non expanding ] we get @xmath155 by definition of @xmath156 , we have @xmath157 and from we have @xmath158 . thus we conclude that @xmath156 and @xmath34 are inverse of each other . a continuous map @xmath159\to [ 0,1]$ ] is _ unimodal _ if @xmath160 , and if there exists a point @xmath161 $ ] such that @xmath2 is strictly increasing on @xmath162 $ ] , and strictly decreasing on @xmath163 $ ] . the point @xmath164 is called the _ turning _ or _ critical point _ of @xmath2 . for each @xmath17 $ ] the logistic map @xmath12 is a unimodal with critical point @xmath9 . let @xmath2 be a unimodal map with critical point @xmath164 . the @xmath1-limit of @xmath164 will be called the _ post - critical set _ of @xmath2 . when either @xmath165 or @xmath166 , it is easy to see that the post - critical set of @xmath2 reduces to a single point . we will thus ( implicitly ) assume from now on that for each unimodal map @xmath2 that we consider we have @xmath167 . to describe the dynamics of a unimodal map @xmath2 on its post - critical set , we will make the following definitions . let @xmath164 be the critical point of @xmath2 and for each @xmath168 put @xmath169 . define the sequence of compact intervals @xmath170 inductively by @xmath171 $ ] , and for each @xmath172 , by @xmath173 & \text{otherwise}. \end{cases}\ ] ] an integer @xmath168 will be called a _ cutting time _ if @xmath174 . we will denote by @xmath175 the sequence of all cutting times . from our assumption that @xmath167 it follows that @xmath176 and @xmath177 . it can be shown that if @xmath178 and @xmath179 are consecutive cutting times , then @xmath180 is again a cutting time , and that this cutting time is less than or equal to @xmath178 when @xmath2 has no periodic attractors , see for example @xcite . that is , if @xmath2 has no periodic attractors then for each @xmath181 there is a non - negative integer @xmath182 , such that @xmath183 , and @xmath184 putting @xmath185 , the function @xmath186 so defined is called the _ kneading map _ of @xmath2 . it follows from the recursion formula above , and from @xmath176 , that the sequence @xmath175 of cutting times is determined by @xmath45 . we will say that a function @xmath186 is a _ kneading map _ if there is a unimodal map @xmath2 with critical point @xmath164 , such that @xmath167 , such that @xmath2 has no periodic attractors and such that the kneading map of @xmath2 is equal to @xmath45 . if we denote by @xmath187 the lexicographical ordering in @xmath188 , then a function @xmath186 is a kneading map if and only if @xmath185 , for each @xmath181 we have @xmath183 , and if for each @xmath181 we have @xmath189 see @xcite . notice in particular that , if @xmath186 is non - decreasing , @xmath185 and for each @xmath181 we have @xmath183 , then @xmath45 is a kneading map . we will need the following well known facts , see for example the proof of ( * ? ? ? * proposition 4 ) for precise references . [ p : continuous unimodal ] let @xmath2 be a unimodal map whose kneading map diverges to @xmath48 . then the post - critical set of @xmath2 is a cantor set , and the restriction of @xmath2 to this set is minimal and has zero topological entropy . furthermore , if @xmath190 is a unimodal map having the same kneading map as @xmath2 , then the space of invariant probability measures of @xmath190 supported on the post - critical set of @xmath190 is affine homeomorphic to that of @xmath2 . the purpose of this section is to prove the main theorem assuming a result we state as theorem [ t : invariant of doubly resonant ] . we start introducing a class of kneading maps in [ ss : doubly resonant kneading ] . in [ ss : proof of main theorem ] we state a result describing the space of invariant probability measures supported on the post - critical set of a unimodal map with a kneading map in this class ( theorem [ t : invariant of doubly resonant ] ) . in [ ss : proof of main theorem ] we also give a proof of the main theorem assuming theorem [ t : invariant of doubly resonant ] . throughout the rest of this paper we denote by @xmath191 the sequence of integers defined by @xmath192 . note that for each @xmath90 we have @xmath193 . for each @xmath90 let @xmath194}$ ] be given and let @xmath195 be an increasing sequence of integers such that @xmath196 , and such that for each @xmath90 we have , @xmath197 if we put @xmath198 , then we will define a kneading map @xmath199 as follows . for each @xmath200 put @xmath201 \text { and } { j}_n = [ q_{r_{n } } + 1 , q_{r_{n + 1 } - 1}],\ ] ] and note that @xmath202 is a partition of @xmath50 . furthermore , for each @xmath90 and @xmath203 $ ] define @xmath204\ ] ] and @xmath205.\ ] ] by the collection @xmath206 \}$ ] is a partition of @xmath207 . since for each @xmath90 we have @xmath208 , the collection @xmath209 \}$ ] is a partition of @xmath210 . with these notations we put @xmath211 note that @xmath212 $ ] , @xmath213 and that for each @xmath90 we have @xmath214 [ l : doubly resonant kneading ] for @xmath198 , @xmath215 and @xmath216 as above , the following properties hold . 1 . for each @xmath46 we have @xmath217 . the function @xmath216 is a kneading map and for every @xmath218 we have @xmath219 put @xmath220 . * * for @xmath221 $ ] we have @xmath222 so the inequality is satisfied in this case . let @xmath223 be such that @xmath224 , so there is @xmath90 such that @xmath225 $ ] . thus @xmath226 , so when @xmath227 we have @xmath228 . finally observe that @xmath229 * 2 . * let @xmath223 be such that @xmath230 , so there is @xmath172 such that @xmath231 . then @xmath232 , @xmath233 and therefore we have @xmath234 and @xmath235 in view of part 1 and the previous inequality , to show that @xmath45 is admissible we just need to show that for each @xmath236 $ ] and @xmath237 $ ] we have @xmath238 with strict inequality when @xmath239 . in fact , for each @xmath236 $ ] we have @xmath240 , so for each @xmath237 $ ] , we have @xmath241 and is satisfied . when @xmath239 we have @xmath242 , so inequality is strict in this case . the purpose of this section is to give a proof of the main theorem assuming the following one . [ t : invariant of doubly resonant ] for each @xmath90 let @xmath243}$ ] be given and put @xmath198 . furthermore , let @xmath244 be a strictly increasing sequence of integers such that @xmath196 and such that for each @xmath90 we have @xmath245 and let @xmath216 be the corresponding kneading map . define @xmath246 recursively by @xmath176 and @xmath247 , and assume that @xmath248 moreover , for each @xmath90 let @xmath249 } \to { \mathbb{r}}^{[0 , n]}$ ] be the stochastic matrix such that for each @xmath203 $ ] we have @xmath250 and such that @xmath251 then for each unimodal map @xmath2 whose kneading map is equal to @xmath216 , the post - critical set of @xmath2 is a cantor set , @xmath2 is minimal on this set , and the space of invariant probability measures of @xmath2 supported on this set is affine homeomorphic to @xmath92 } , \xi_n)$ ] . to prove the main theorem , we first remark that the case where the metrizable choquet simplex @xmath16 is finite dimensional is given by ( * ? ? ? * main theorem ) . so from now on we assume that @xmath16 is infinite dimensional . then by lemma [ l : normalized choquet ] there is a sequence of normalized affine maps @xmath93 such that @xmath92 } , a_n)$ ] is affine homeomorphic to @xmath16 . in view of lemma [ l : equivalence ] , we just need to find @xmath252 and @xmath253 as in the statement of theorem [ t : invariant of doubly resonant ] , for which is satisfied and such that @xmath254 this is shown in the following lemma , thus completing the proof of the main theorem . [ l : approximation ] for each @xmath90 let @xmath255}$ ] be given . then there are @xmath252 and @xmath253 as in the statement of theorem [ t : invariant of doubly resonant ] for which is satisfied , and such that @xmath256 given @xmath90 and a non - zero vector @xmath257}$ ] with non - negative coordinates , we will denote by @xmath258 the unique vector in @xmath103}$ ] proportional to @xmath259 . we will define @xmath198 and @xmath244 by induction as follows . put @xmath196 , fix @xmath260 , and assume that for some @xmath90 the numbers @xmath261 and the vectors @xmath262 are already defined , in such a way that for each @xmath263 $ ] we have @xmath264 for each @xmath265 $ ] let @xmath266 be defined in such a way that @xmath267 note that these choices determine @xmath268 . * we will show now that for each @xmath265 $ ] we have @xmath269 using the recursion formula @xmath270 and @xmath271 , we get by induction that for every @xmath272 $ ] such that @xmath273 , we have @xmath274 . in particular for every @xmath275 $ ] we have @xmath276 since @xmath277 , it follows by induction that for every @xmath275 $ ] we have @xmath278 hence for each @xmath265 $ ] we have @xmath279 as wanted . * we will show that we can choose @xmath194}$ ] is such a way that @xmath280 for @xmath281 we denote by @xmath282 $ ] integer part of @xmath283 . put @xmath284 , @xmath285 , @xmath286 , for each @xmath94 $ ] put @xmath287 + 1 ) , \text { and put } \zeta = \zeta_1 + \cdots + \zeta_n.\ ] ] since for each @xmath94 $ ] we have @xmath288 , we obtain @xmath289 . so for each @xmath290 $ ] we have @xmath291 this shows that the vector @xmath292}$ ] satisfies @xmath293 . thus , if for each @xmath294 $ ] we put @xmath295 , and if we put @xmath296 , then is satisfies for this choice of @xmath297 . it remains to show that each of the coordinates of @xmath297 belongs to @xmath50 . by definition for each @xmath94 $ ] the integer @xmath298 is a strictly positive multiple of @xmath299 , so the coordinates of @xmath297 are integers and for each @xmath294 $ ] we have @xmath300 . finally observe that by with @xmath301 we have , @xmath302 let @xmath297 be given by part 2 and put @xmath303 . this completes the inductive definition of @xmath253 and @xmath252 . to finish the proof of the lemma just observe that the inequalities imply , and the inequalities imply . the purpose of this section is to recall the definition of the generalized odometer and the bratteli - vershik system associated to a kneading map , that were introduced in @xcite and @xcite , respectively . we start recalling the definition of the generalized odometer in [ ss : kneading odometer ] . after briefly recalling the concepts of bratteli diagram ( [ ss : diagram ] ) and bratteli - vershik system ( [ ss : bratteli - vershik ] ) , we define the bratteli - vershik system associated to a kneading map in [ ss : kneading bratteli - vershik ] . see for example @xcite for background on generalized odometers , and @xcite and references therein for background and further properties of bratteli - vershik systems . let @xmath186 be a kneading map and put @xmath304 if we denote by @xmath175 the strictly increasing sequence of positive integers defined recursively by @xmath176 and @xmath305 , it can be shown that for each non - negative integer @xmath111 there is a unique sequence @xmath306 in @xmath307 , that has at most finitely many @xmath62 s , and such that @xmath308 . the sequence @xmath309 is also characterized as the unique sequence in @xmath310 with finitely many @xmath62 s such that @xmath308 , and that it is minimal with this property with respect to the lexicographical order in @xmath310 . when @xmath45 diverges to @xmath48 the map defined on the subset @xmath311 of @xmath307 by @xmath312 , extends continuously to a map @xmath313 which is onto , minimal , and such that @xmath314 is well defined on @xmath315 ; see ( * ? ? ? * lemma 2 ) . we call @xmath316 the _ _ generalized odometer _ _ is such that for each @xmath46 we have @xmath317 . ] associated to @xmath45 . given @xmath318 and an integer @xmath319 , put @xmath320 . observe that @xmath321 is non - decreasing with @xmath111 , and when @xmath283 has infinitely many @xmath62 s , @xmath322 as @xmath323 . on the other hand , if @xmath283 has at most a finite number of @xmath62 s , then @xmath324 is finite and @xmath325 . for @xmath326 different from @xmath327 we denote by @xmath328 the least integer such that @xmath329 . in ( * theorem 1 ) it is shown that if @xmath17 $ ] is a parameter such that the kneading map of the logistic map @xmath12 is equal to @xmath45 , then for each @xmath330 with infinitely many @xmath62 s the sequence of intervals @xmath331 is nested and that @xmath332 is reduced to a point belonging to the post - critical set @xmath333 of @xmath12 . furthermore , if we denote this point by @xmath334 and for @xmath319 we put @xmath335 , then the map @xmath336 so defined is continuous and conjugates the action of @xmath337 on @xmath307 , to the action of @xmath12 on @xmath333 . a _ bratteli diagram _ is an infinite directed graph @xmath338 , such that the vertex set @xmath57 and the edge set @xmath339 can be partitioned into finite sets @xmath340 with the following properties : * @xmath341 is a singleton . * for every @xmath342 , each edge in @xmath343 starts in a vertex in @xmath344 and arrives to a vertex in @xmath345 . * all vertices in @xmath57 have at least one edge starting from it , and all vertices except @xmath346 have at least one edge arriving to it . for a vertex @xmath347 we will denote by @xmath348 the vertex where @xmath349 starts and by @xmath350 the vertex to which @xmath349 arrives . a _ path _ in @xmath351 is by definition a finite ( resp . infinite ) sequence @xmath352 ( resp . @xmath353 ) such that for each @xmath354 ( resp . @xmath355 ) we have @xmath356 . note that for each vertex @xmath61 distinct from @xmath346 there is at least one path starting at @xmath346 and arriving to @xmath61 . an _ ordered bratteli diagram _ @xmath357 is a bratteli diagram @xmath338 together with a partial order @xmath358 on @xmath339 , so that two edges are comparable if and only if they arrive at the same vertex . for each @xmath342 and @xmath359 the partial order @xmath360 induces an order on the set of paths from @xmath346 to @xmath57 as follows : @xmath361 if and only there exists @xmath362 such that @xmath363 and such that for each @xmath364 we have @xmath365 . we will say that an edge @xmath349 is _ maximal _ ( resp . _ minimal _ ) if it is maximal ( resp . minimal ) with respect to the order @xmath360 on the set of all edges in @xmath339 arriving at @xmath350 . note that for each vertex @xmath61 distinct from @xmath346 there is precisely one path starting at @xmath346 and arriving to @xmath61 that is maximal ( resp . minimal ) with respect to the order @xmath360 . it is characterized as the unique path starting at @xmath346 and arriving at @xmath61 consisting of maximal ( resp . minimal ) edges . fix an ordered bratteli diagram @xmath366 . we denote by @xmath367 set of all infinite paths in @xmath368 starting at @xmath346 . for a finite path @xmath369 starting at @xmath346 we denote by @xmath370 the subset of @xmath367 of all infinite paths @xmath371 such that for all @xmath372 we have @xmath373 . we endow @xmath367 with the topology generated by the sets @xmath370 . then each of this sets is clopen , so @xmath367 becomes a compact hausdorff space with a countable basis of clopen sets . we will denote by @xmath374 ( resp . @xmath375 ) the set of all elements @xmath376 of @xmath367 so that for each @xmath342 the edge @xmath377 is a maximal ( resp . minimal ) . it is easy to see that each of these sets is non - empty . from now on we assume that the set @xmath375 is reduced to a unique point , that we will denote by @xmath378 . we will then define the transformation @xmath379 as follows : * @xmath380 . * given @xmath381 , let @xmath342 be the smallest integer such that @xmath377 is not maximal . then we denote by @xmath382 the successor of @xmath377 and by @xmath383 the unique minimal path starting at @xmath346 and arriving to @xmath384 . then we put , @xmath385 the map @xmath386 is continuous , onto and invertible except at @xmath378 . for @xmath342 and @xmath359 we denote by @xmath387 the number of paths starting at @xmath346 and arriving to @xmath61 , and put @xmath388 . let @xmath389 be the matrix such that for each @xmath390 and @xmath391 the entry @xmath392 is equal to the number of edges starting at @xmath61 and arriving to @xmath80 . observe that @xmath393 , so if we put @xmath394 and for each @xmath342 we denote by @xmath395 the diagonal matrix defined by @xmath396 , then the matrix @xmath397 is stochastic . the following result is well - known , see ( * ? ? ? * lemma 14 ) for a proof in the precise setting considered here . recall that for a finite set @xmath57 we denote by @xmath68 the unit simplex in @xmath64 . [ l : medidas bratteli ] the space of probability measures on @xmath367 that are invariant by @xmath386 , endowed with the weak@xmath0 topology , is affine homeomorphic to @xmath398 . given a kneading map @xmath45 we will now define an ordered bratteli diagram @xmath399 that was introduced by bruin in @xcite . we start defining the bratteli diagram @xmath351 : * @xmath400 , @xmath401 and for @xmath402 , @xmath403 * for @xmath342 , @xmath404 note that for every @xmath402 , each vertex in @xmath405 different from @xmath406 has at most one edge arriving at it . besides @xmath407 , the only edge that can arrive to @xmath408 is @xmath409 , that only exists when @xmath410 . so to define the partial order @xmath360 , we just have to define it , for each @xmath402 , between @xmath411 and @xmath412 when both exist : we put @xmath413 . the rest of the edges are maximal and minimal at the same time . note that for @xmath181 the set @xmath414 is reduced to a point if and only if @xmath415 . so , if for each large @xmath181 we have @xmath415 , then the set @xmath416 is finite . otherwise , it follows that the set @xmath416 is a cantor set . it is straight forward to check that the infinite path @xmath417 is the unique minimal path in @xmath418 . therefore there is a well defined map @xmath419 , see [ ss : bratteli - vershik ] . the following is ( * ? ? * proposition 2 ) , and the last statement follows from ( * ? ? ? * lemma 2 ) . [ t : reduction to bratteli ] let @xmath45 be a kneading map that diverges to @xmath48 , and consider the corresponding bratteli - vershik system @xmath420 and generalized odometer @xmath316 . then there is a homeomorphism between @xmath416 and @xmath307 that conjugates the action of @xmath421 on @xmath416 to the action of @xmath337 on @xmath307 . in particular @xmath420 is minimal . we will also need the following lemma . [ l : recurrence on paths ] let @xmath45 be a kneading map such that for every @xmath46 we have @xmath47 , and such that @xmath422 as @xmath423 . let @xmath424 be the sequence defined recursively by @xmath176 and @xmath305 . then for every @xmath425 we have @xmath426 , @xmath427 , and for every @xmath428 we have @xmath429 . that @xmath430 is a direct consequence of the definition of @xmath405 and the hypothesis that @xmath431 . when @xmath432 , we have @xmath433 and for all @xmath434 we have @xmath435 . so the assertions are satisfied in this case . suppose by induction that the assertions of the lemma hold for some @xmath342 . then by we have , @xmath436 on the other hand , for each @xmath437 contained in @xmath405 we have @xmath438 . finally , for each @xmath439 not in @xmath405 we have @xmath440 and @xmath441 this section is devoted to the proof of theorem [ t : invariant of doubly resonant ] . we start by showing that for each unimodal map whose kneading map is as in [ ss : doubly resonant kneading ] , the space of invariant probability measures supported on its post - critical set is affine homeomorphic to the space of invariant probability measures of the corresponding generalized odometer . in order to describe this space we calculate the transition matrices associated to the corresponding bratteli - vershik system . the key calculation of a suitable product of these transition matrices is stated as proposition [ p : full block ] in [ ss : transition matrices ] . the proof of theorem [ t : invariant of doubly resonant ] is given in [ ss : proof of invariant of doubly resonant ] . the purpose of this section is to prove the following proposition , whose proof is similar to that of ( * ? ? ? * theorem b ) . [ p : measure isomorphism ] let @xmath216 be a kneading map defined as in [ ss : doubly resonant kneading ] , and let @xmath442 be the corresponding generalized odometer . let @xmath2 be a unimodal map whose kneading map is equal to @xmath216 , and denote by @xmath443 its post - critical set . then the space of invariant probability measures of @xmath444 is affine homeomorphic to that of @xmath442 . the following lemma is similar to ( * ? ? ? * lemma 11 ) . observe that , since @xmath314 is well defined on @xmath445 , if we denote by @xmath446 the grand orbit of @xmath327 , then @xmath447 and all negative iterates of @xmath337 are well defined on @xmath448 . [ l : criterion satisfied ] let @xmath449 be a kneading map as in [ ss : doubly resonant kneading ] . let @xmath450 be the corresponding generalized odometer . then for each constant @xmath451 , and for every pair of distinct points @xmath452 in @xmath453 that are not in the grand orbit of @xmath327 , there is an integer @xmath454 satisfying @xmath455 let @xmath456 . * 1 . * as in the proof of ( * ? ? ? * lemma 10 ) , it can be shown that there is an integer @xmath457 such that @xmath458 * 2 . * let @xmath457 be the integer given by part 1 , and put @xmath459 and @xmath460 . assume without loss of generality that @xmath461 , so that @xmath462 . if @xmath463 then take @xmath464 . so we assume that @xmath465 . since @xmath466 we have @xmath467 , so there is @xmath90 such that @xmath468 and @xmath469 belong to @xmath470 . the definition of @xmath471 and of @xmath453 imply that for each @xmath472 $ ] we have @xmath473 . indeed , suppose by contradiction that for such a @xmath150 we have @xmath474 . then the definition of @xmath453 implies that for every @xmath475 $ ] we have @xmath476 . since @xmath477 , we get @xmath478 , which is a contradiction . therefore @xmath479 is such that for all @xmath480 $ ] we have @xmath481 . since @xmath482 is not in the grand orbit of @xmath327 this implies that @xmath483 . thus , from the definition of @xmath471 we have @xmath484 on the other hand , since @xmath461 , we have @xmath485 , so @xmath486 this shows that the integer @xmath487 satisfies the desired properties . since the logistic family is full there is a parameter @xmath17 $ ] such that the kneading map of the logistic map @xmath12 is @xmath216 . denote by @xmath488 the post - critical set of @xmath12 . by proposition [ p : continuous unimodal ] the spaces of invariant measures of @xmath489 and @xmath490 are affine homeomorphic . so , without loss of generality we assume that @xmath2 is a logistic map . this ensures the existence of the factor map @xmath491 defined above . since for every sufficiently large integer @xmath150 inequality is satisfied , ( * ? ? ? * lemma 11 ) implies there is a constant @xmath492 such that for every pair of distinct points @xmath493 in @xmath494 that are not in the grand orbit of @xmath327 and that satisfy @xmath495 we have @xmath496 . thus , from lemma [ l : criterion satisfied ] we deduce that @xmath497 is injective on @xmath498 . the rest of the proof follows as the proof of ( * ? ? ? * proposition 9 ) . for each @xmath90 let @xmath243}$ ] be given , and put @xmath198 . furthermore , let @xmath244 be a strictly increasing sequence of integers such that @xmath196 and such that for each @xmath90 we have @xmath245 and let @xmath220 be the corresponding kneading map defined in [ ss : doubly resonant kneading ] . let @xmath499 be the ordered bratteli - vershik diagram associated to the kneading map @xmath45 . from the definition of @xmath418 it follows that @xmath500 $ ] , and that for each @xmath501 $ ] we have @xmath502 $ ] . furthermore , for each @xmath200 , @xmath203 $ ] , and @xmath503 $ ] we have @xmath504 \cup \left ( 1 + \bigcup_{i = 0}^m \left ( { i}_{n + 1 , i } \cup { j}_{n + 1 , i } \right ) \right),\ ] ] and that when @xmath505 we have @xmath506 , and for @xmath507 $ ] we have @xmath508.\ ] ] note in particular that for every @xmath90 we have @xmath509 \setminus \left ( 1 + ( { i}_{n , n } \cup { j}_{n , n } ) \right),\ ] ] @xmath510.\ ] ] [ p : full block ] given @xmath90 , for each @xmath511 $ ] put @xmath512 then the columns of the matrix @xmath513 , [ q_{r_{n + 1 } } + 1 , q_{r_{n + 2 } - 1 } + 1]),\ ] ] are given by , @xmath514 and for every @xmath515 $ ] and @xmath516 such that @xmath517 , by @xmath518 in particular the rank of the matrix @xmath519 is equal to @xmath520 . the proof of this proposition depends on the following lemma . [ l : basic blocks ] fix @xmath90 and let @xmath515 $ ] . if @xmath203 $ ] then the set @xmath521 is equal to the disjoint union of @xmath522 $ ] and of @xmath523 , and we have @xmath524 ; \\ \vec{e}_{q_{r_n + m } + 1 } & \text{if } \ell - 1 \in q^{-1}(q_{r_n + m } ) . \end{cases}\end{gathered}\ ] ] when @xmath505 we have @xmath525 , @xmath526 $ ] , and @xmath527 it follows from the definition of @xmath418 that for each @xmath528 the set @xmath529 is equal to the disjoint union of @xmath530 and @xmath531 , and that @xmath532 on the other hand , for each @xmath533 $ ] we have @xmath534 and @xmath535 a direct computation using the fact that for every @xmath536 $ ] the set @xmath345 is the disjoint union of @xmath537 and @xmath538 $ ] , shows that @xmath539 . \end{cases}\end{gathered}\ ] ] fix @xmath90 and let @xmath515 $ ] . when @xmath203 $ ] the assertion of the lemma about @xmath521 follows from . on the other hand , by if @xmath540 and by if @xmath541 $ ] , we have @xmath542 \subset v_{q_{r_n + m } + 1}.\ ] ] then in this case the assertion of the lemma follows easily from with @xmath543 . suppose now that @xmath505 . by the definition of the sequence @xmath544 we have @xmath506 . the assertion about @xmath545 is given by with @xmath111 replaced by @xmath546 . since by we have @xmath547 \cap v_{q_{r_{n + 1 } - 1 } + 1 } = 1 + ( { i}_{n + 1 } \setminus { i}_{n + 1 , n + 1 } ) = \\ = [ q_{r_{n + 1 } -1 } + 2 , q_{r_{n + 1 } } + 1 - a_{n + 1 , n + 1}]\end{gathered}\ ] ] and since @xmath548 , we conclude from with @xmath549 , that @xmath550 on the other hand , for @xmath551 \cap v_{q_{r_{n + 1 } - 1 } + 1 } = 1 + \left ( { j}_{n + 1 } \setminus { j}_{n + 1 , n + 1 } \right),\ ] ] we have @xmath552 and that for @xmath551 \cap \left ( 1 + q^{-1}(q_{r_{n + 1 } - 1 } ) \right ) = 1 + { j}_{n + 1 , n + 1},\ ] ] we have @xmath553 this completes the proof of the lemma . fix @xmath90 . * we will show by induction that for each @xmath554 $ ] we have @xmath555 ; \\ \sum_{k = q_{r_n}}^{q_{r_n + m } } \vec{e}_{k + 1 } & \text{if } \ell - 1 \in { i}_{n + 1 , m } \cup { j}_{n + 1 , m } \text { and } m \in [ 0 , m_0 - 1 ] . \end{cases}\end{gathered}\ ] ] the case @xmath556 is given by with @xmath557 . suppose that this holds for some @xmath558 $ ] . observe that by the set @xmath559 is the disjoint union of @xmath560\ ] ] and @xmath561 in view of with @xmath562 we obtain @xmath563 for each @xmath564 $ ] we have @xmath565 \\ \sum_{k = q_{r_n}}^{q_{r_n + m } } \vec{e}_{k + 1 } & \text{if } \ell - 1 \in { i}_{n + 1 , m } \cup { j}_{n + 1 , m } \text { and } m \in [ 0 , m_0 - 1 ] . \end{cases}\end{gathered}\ ] ] finally , for each @xmath566 such that @xmath567 we have @xmath568 this completes the proof of the induction step . a direct computation using with @xmath569 and , gives @xmath570 and for every @xmath515 $ ] and @xmath516 such that @xmath571 , @xmath572 the assertion of the proposition is then a direct consequence of the definition of the matrices @xmath573 and lemma [ l : recurrence on paths ] . let @xmath2 be a unimodal having @xmath45 as kneading map . that the post - critical set of @xmath2 is a cantor set and that @xmath2 is minimal on this set is given by proposition [ p : continuous unimodal ] . in view of proposition [ p : measure isomorphism ] and theorem [ t : reduction to bratteli ] , it is enough to prove that the space of invariant probability measures of the bratteli - vershik system @xmath420 is affine homeomorphic to @xmath574 } , \xi_n)$ ] . for each @xmath90 let @xmath575}$ ] be the stochastic matrix defined by @xmath576 using the definition of @xmath577 in the statement of proposition [ p : full block ] , for each @xmath578 $ ] we put @xmath579 furthermore , when @xmath172 , we denote by @xmath580 } \to { \mathbb{r}}^{[0 , n]}$ ] the stochastic matrix defined for @xmath203 $ ] by @xmath581 and by @xmath582 a direct computation shows that for each @xmath172 we have @xmath583 therefore the sequence of maps @xmath584 define a continuous linear map @xmath585 } , a_n),\ ] ] mapping @xmath586 onto @xmath120 } , a_n)$ ] . by proposition [ p : full block ] the rank of the matrix @xmath587 is equal to @xmath546 , so @xmath588 is a homeomorphism and the inverse limits @xmath586 and @xmath120 } , a_n)$ ] are affine homeomorphic . in view of lemma [ l : equivalence ] , the following lemma together with the hypothesis @xmath589 imply that the inverse limit @xmath120 } , a_n)$ ] is affine homeomorphic to @xmath120 } , \xi_n)$ ] . this completes the proof of theorem [ t : invariant of doubly resonant ] . for each @xmath90 and @xmath590 $ ] we have @xmath591 furthermore @xmath592 , and @xmath593 by definition we have @xmath594 , and for each @xmath590 $ ] we have @xmath595 on the other hand , @xmath596 so the final assertion follows from . given @xmath597 put @xmath598 for each such @xmath599 we will construct a kneading map @xmath45 such that the dimension group associated to the generalized odometer @xmath316 , and hence to its natural extension , is isomorphic to @xmath600 . thus we deduce that every simple dimension group which is free of rank @xmath601 is isomorphic as ordered group to the dimension group associated to ( the natural extension of ) a generalized odometer associated to a kneading map . since the rational subdimension group of @xmath600 is @xmath602 , it follows from @xcite that the dimension group @xmath600 is not isomorphic to the dimension group associated to a toeplitz flow , nor to that of an odometer . before defining the kneading map @xmath45 , note that the dimension groups @xmath600 , @xmath603 , and @xmath604 are isomorphic to each other . so we can restrict to the case where @xmath605 . let @xmath606 be the integer determined by @xmath607 , and let @xmath608 $ ] be the continued fraction expansion of @xmath609 \setminus { \mathbb{q}}$ ] . consider the function @xmath186 defined by @xmath610 ; \\ k-1 & \mbox { if } l \in [ k + 1 , k + a_1 ] ; \\ k-1+\sum_{i=1}^na_i & \mbox { if } l \in \left[k+1 + \sum_{i=1}^na_i , k+\sum_{i=1}^{n+1}a_i \right ] . \\ \end{cases}\ ] ] it is non - decreasing and such that for every @xmath611 we have @xmath271 . so @xmath45 is a kneading map . if @xmath612 is the corresponding sequence of transition matrices , then it is easy to see that for each @xmath90 we have @xmath613 by considering a bratteli diagram isomorphic to @xmath418 , that only differs with it in the first @xmath614 levels , we obtain @xmath615 so the dimension group associated to @xmath316 is isomorphic to direct limit @xmath616 see for example ( * ? ? ? * theorem 3.7 ) , which by ( * ? ? ? * theorem 4.8 ) is isomorphic to @xmath617 . the purpose of this appendix is to prove corollary [ c : ergodic theory holomorphic ] . as the parameters given by ( the proof of ) the main theorem are such that the kneading map of corresponding logistic map diverges to @xmath49 , this result is a direct consequence of the following lemma . [ l : ergodic theory holomorphic ] let @xmath17 $ ] be a parameter such that the kneading map of @xmath12 diverges to @xmath48 . consider the corresponding quadratic polynomial @xmath28 , and denote by @xmath31 the hausdorff dimension of the julia set of @xmath28 . then , for an invariant probability measure of @xmath28 that is supported on the julia set of @xmath28 , the following properties are equivalent . as the restriction of @xmath28 to @xmath10 $ ] is the logistic map @xmath12 , the implication @xmath618 is given by ( * ? ? ? * lemma 21 ) . the proof of the implication @xmath619 is analogous to the corresponding implication of the same lemma . the implication @xmath620 being trivial we just need to prove the implication @xmath621 to complete the proof of the lemma . we will show that for each invariant measure @xmath21 that is supported on the julia set of @xmath28 , we have @xmath622 with equality when the lyapunov exponent of @xmath21 is zero . for such a measure we have @xmath623 by @xcite , so there are two cases . if @xmath624 , then @xmath21 is supported on the post - critical set of @xmath28 and therefore we have @xmath625 by the variational principle and the fact that the topological entropy of @xmath28 restricted to its post - critical set is zero ( proposition [ p : continuous unimodal ] ) . so we have @xmath626 in this case . suppose now that @xmath627 . then by @xcite it follows that , if we denote by @xmath628 the hausdorff dimension of @xmath21 , then @xmath629 . as @xmath630 , we obtain @xmath622 in this case . h. bruin . combinatorics of the kneading map . in _ thirty years after sharkovski s theorem : new perspectives ( murcia , 1994 ) _ , volume 8 of _ world sci . nonlinear sci . theme issues proc . _ , pages 7787 . world sci . publ . , river edge , nj , 1995 . reprint of the paper reviewed in mr1361922 ( 96k:58070 ) . ricardo ma . the hausdorff dimension of invariant probabilities of rational maps . in _ dynamical systems , valparaiso 1986 _ , volume 1331 of _ lecture notes in math . _ , pages 86117 . springer , berlin , 1988 .

relying on infrared surface brightness fluctuactions to trace agb properties in a sample of elliptical galaxies in the virgo and fornax clusters , we assess the puzzling origin of the `` uv - upturn '' phenomenon , recently traced down to the presence of a hot horizontal branch ( hb ) stellar component . we find that the uv - upturn actually signals a profound change in the c - m diagram of stellar populations in elliptical galaxies , involving both the hot stellar component and red - giant evolution . first , we encounter that the strengthening of the uv rising branch is always seen to correspond to a shortening in agb deployment ; this trend can be readily interpreted as an age effect , perhaps mildly modulated by metal abundance . a comparison between galaxy magnitudes and population synthesis models confirms that , all the way , brightest stars in ellipticals are genuine agb members , reaching the thermal - pulsing phase , and with the agb tip exceeding the rgb tip by some 0.5 - 1.5 mag . the inferred core mass of these stars is found to be m among giant ellipticals . coupled with the recognized severe deficiency of planetary nebulae in these galaxies , this result strongly calls for an even more critical blocking effect due to a lengthy transition time needed by the post - agb stellar core to become a hard uv emitter and eventually `` fire up '' the nebula . the combined study of galaxy color and integrated h index points , as an explanation for the uv - upturn phenomenon , to a composite hb with a bimodal temperature distribution , i.e. with both a red clump and an extremely blue component , in a relative proportion of roughly [ n(rhb):n(bhb ) ] [ 80:20 ] . as far as metallicity of the bhb stellar population is concerned , we find that [ fe / h ] values of either dex or may provide the optimum ranges to feed the needed low - mass stars ( m m ) , that at some stage begin to join the standard red - clump stars .

0806.2658

the so - called `` uv - upturn '' phenomenon @xcite , i.e. , the rising ultraviolet emission shortward of 2000 , sometimes featuring in the spectral energy distribution ( sed ) of elliptical galaxies and the bulges of spirals , has been for long a puzzling problem for old galaxy environments dominated by stars of mass comparable to the sun . in fact , the implied existence of an important contribution of ( long - lived ) b stars , hotter than @xmath9 k and providing up to about 2% of the galaxy bolometric luminosity @xcite , has been alternately identified with different evolutionary stages . such stages include binaries @xcite , blue stragglers @xcite , blue horizontal - branch ( hb ) stars @xcite , asymptotic giant branch ( agb ) _ manqu _ stars @xcite , and post - agb nuclei of planetary nebulae ( pne ) @xcite ( see * ? ? ? * , for an exhaustive review , and a more recent update by @xcite ) resolved color magnitude ( c - m ) diagrams of stellar populations in m32 @xcite have definitely shown that even its relatively poor uv emission almost entirely arises from a fraction of hot hb stars , further complemented by a minority contribution from post - agb pn nuclei . still , facing the established interpretative scenario , one is left with at least three important issues that need to be assessed to understand the real nature of the uv - upturn phenomenon . _ ( i ) _ the canonical evolutionary framework experienced in galactic globular clusters naturally predicts a blue hb morphology only for old , metal - poor stellar populations @xcite . if this is the case for ellipticals too , then uv stars should represent the @xmath10 tail of a ( supposedly ) broad metallicity distribution seen to peak at much higher values , around solar abundance . clearly , a more composite picture might be envisaged once one admits non - standard models ( i.e. , including the effects of stellar rotation , helium mixing , differential mass loss , etc . ) to account , in particular , for the well known `` second parameter '' dilemma @xcite . however , this unconventional approach still suffers from a somewhat arbitrary fine - tuning of the key physical assumptions . _ ( ii ) _ hot hb stars might , nonetheless , also be naturally predicted among super metal - rich stellar models , as far as metal abundance ( and the linked helium content ) exceeds some critical threshold . presumably , in this case mass - loss allows stars to reach the hb phase with a conveniently low external envelope , compared to the helium core mass @xcite . such `` extreme hb '' stars ( ehb ) have actually been observed , for example , in @xmath11 cen @xcite , ngc 6388 and ngc 6441 @xcite , and in some old galactic open clusters as well , like ngc 6791 @xcite ; they clearly remain the favorite candidates to explain the evolutionary framework of uv - enhanced elliptical galaxies @xcite . this hypothesis implies , however , a direct relationship between chemical abundance ( modulating the helium core mass at the hb onset ) and mass - loss efficiency ( to suitably `` peel off '' the stellar envelope along the rgb ) . as a consequence , one has to expect the uv - to - optical color to be , eventually , one of the most quickly evolving features in the sed of elliptical galaxies @xcite . in theory , the uv - upturn can fade by several magnitudes as the lookback time increases by a few gyr , although the effect is still detectable at intermediate redshift ( @xmath12 ) @xcite . unfortunately , the evolutionary details are extremely model - dependent , and a strong uv excess could be triggered at ages as early as @xmath13 gyr @xcite or as late as @xmath1415 gyr @xcite . _ ( iii ) _ an established correlation seems to be in place between pn luminosity - specific rate and @xmath15 color for elliptical galaxies in the virgo and fornax clusters , and in the leo group @xcite . the sense is that reddest metal - rich systems display , at the same time , a stronger uv - upturn @xcite _ and _ a poorer pn population per unit galaxy luminosity @xcite . if the pn event is the final fate for agb stars at the end of their thermal pulsing phase @xcite , then the relative deficiency of nebulae might be evidence of an incomplete ( or fully inhibited ) agb evolution of low - mass stars under special environment conditions of the parent galaxy . as a central issue in this discussion , _ it is clear therefore that a preeminent connection should exist between uv excess and agb distinctive properties of stellar populations in early - type galaxies . _ on account of the fuel consumption theorem @xcite , a 1 m@xmath2 star of solar metallicity enters its core he - burning phase with , at most , the equivalent of 0.43 m@xmath2 of h to be spent as nuclear fuel . star with solar abundance @xmath16 starts its hb evolution with a total he amount of roughly @xmath17 m@xmath2 , of which @xmath18 m@xmath2 are confined in the core @xcite and @xmath19 m@xmath2 reside in the envelope . metals amount to roughly @xmath20m@xmath2 and , accordingly , fresh h is 0.37 m@xmath2 . taking into account the nuclear rates ( e.g. , * ? ? ? * ) , the h+he fuel provides at most the equivalent of @xmath21 m@xmath2 of hydrogen . ] this means that , under quite general conditions , post - rgb evolution alone could easily account , in principle , for up to 3/4 of the total bolometric luminosity of a galaxy stellar population @xcite . whether this energy is eventually reduced ( if stars loose their fuel before they burn it ) , or whether it is finally released in the form of ultraviolet or infrared photons , crucially depends on mass - loss and its impact along the entire red - giant evolution . hence , it is of special pertinence to constrain the relevant physical conditions that affect agb evolution in favor of an earlier transition of hb stars towards high temperature and enhanced ultraviolet emission . in this paper we would like to draw the reader s attention to a possibly new and powerful approach to the problem , that can find straightforward applications even to distant galaxies . as explained in sec . 2 , the method relies on surface - brightness fluctuation theory to safely tie infrared effective magnitudes ( that _ can _ be determined for unresolved stellar populations ) , to stellar luminosity at the agb tip ( that _ can not _ be directly observed in distant galaxies ) . we will show , in sec . 3 , that these results tightly correlate with the ultraviolet properties of `` uv upturn '' elliptical galaxies , allowing a self - consistent physical picture and a quite accurate diagnostic of the post - rgb evolution of their underlying stellar populations , including hb morphology and agb deployment . our results will be finally summarized and discussed in sec . @xcite and @xcite first realized the potentially useful information about the composing stars hidden in the surface brightness fluctuations ( sbfs ) of galaxies with unresolved stellar populations . the problem has since received a more complete theoretical assessment by @xcite , and @xcite . briefly , the basic relationship upon which the theory relies is @xmath22 the l.h . side of the equation links an observable quantity ( namely , the relative variance of the galaxy surface brightness ) with the theoretical second - order statistical moment of the composing stars . this quantity , in turn , traces the distribution of stellar luminosity ( @xmath23 ) for the whole population ; it is also a natural output of any population synthesis code , and can easily be computed for different photometric bands and distinctive evolutionary phases of simple stellar populations ( ssps ; see , e.g. , the current theoretical databases of * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? the derived `` effective '' stellar luminosity , @xmath24 , in eq . ( [ eq : fluc ] ) has some important properties . _ ( a ) _ it can be derived from a fully observational procedure , without any supplementary theoretical assumptions . _ ( b ) _ it is an intrinsic distinctive parameter of the stellar aggregate , and its empirical measurement does not require any individual star to be ( fully or partially ) resolved . _ ( c ) _ it identifies an `` effective '' magnitude ( i.e. @xmath25 ) in a completely similar way and with the same photometric zero points as `` standard '' magnitudes , and it scales accordingly with distance and foreground screen reddening . [ cmd ] [ kk ] being statistically representative of the stellar system as a whole , the effective magnitude can not be physically associated to any specific star or stellar groups along the c - m diagram of a stellar aggregate ; nonetheless , it can be instructive to identify stars with @xmath26 luminosity as the `` prevailing '' group tracing the whole population in the different photometric bands . this is shown in fig . 1 for the illustrative case of a 15 gyr ssp with solar metallicity and salpeter imf . because of the quadratic @xmath23 dependence of the summation in the numerator of eq . ( [ eq : fluc ] ) , the effective magnitude at any optical / infrared band is highly sensitive to the giant stars and only marginally responds to a shift in the ssp `` mass leverage '' caused by a change in the imf slope @xcite . likewise , as probe of the brightest stars in a population at a given wavelength , the effective magnitude is relatively insensitive to any underlying older component in the case of composite stellar systems . it is evident from fig . 1 that near - infrared ( near - ir ) magnitudes closely trace the bright - end tail of the ssp luminosity function . in particular , one can notice that @xmath0 is potentially the best tracer of the ssp tip stellar luminosity ( @xmath27 ) , as _ both quantities are expected to depend in quite the same way on the overall distinctive parameters of the stellar population , including age , metallicity , imf , and mass - loss . _ `` effective contributors '' ( in the statistical definition of @xcite ) to the @xmath28-band sbf for a galaxy of @xmath29 m@xmath2 . this number further increases to about @xmath30 for @xmath31-band contributors , and to @xmath32 for the @xmath33 band sbf . ] this is shown in fig . 2 , where we compare the @xmath0 vs. @xmath34 relationship for a full collection of ssp models from the @xcite and the updated charlot & bruzual ( cb07 * ? ? ? * ) synthesis codes . band is a variant of the standard @xmath28 filter , sometimes preferred for its reduced background noise ( see * ? ? ? * for specific details ) . both filters have similar effective wavelength and photometric zero points , so that @xmath35 within a typical @xmath36 mag uncertainty , as we directly verified in our tests.]@xmath37 ) from the original isochrone set , while @xmath34 is just read as the absolute magnitude of the brightest stellar type in the agb+rgb phases . ] the correlation appears quite robust , in spite of the extremely wide range explored for the population parameters , and the different input physics adopted by the two sets of theoretical models . formal fits to the whole sets of models yield : @xmath38 a comparison with real stellar populations extending across the widest range of evolutionary parameters would clearly be the ideal test for the envisaged theoretical picture . the magellanic clouds ( mcs ) stand out as striking candidates , in this regard , with the age of their star clusters spanning over four orders of magnitude , from a few myr up to @xmath39 yr . in addition , their relatively close distance allows _ both _ fluctuation luminosity _ and _ agb / rgb luminosity tip to be directly measured on resolved c - m diagrams . near - ir sbfs for a sample of 191 mc star clusters have been obtained by @xcite using @xmath40-band data from the two micron all sky survey ( 2mass , * ? ? ? * ) . to reduce stochastic effects due to small - number stellar statistics along fast evolutionary phases , in their study these authors assembled seven `` super - clusters '' , by homogeneously co - adding several objects for each age class according to the ( * ? * , swb ) classification scheme . the procedure to obtain the sbfs for these data has been extensively described before @xcite ; briefly , direct summation of bright stars provided the numerator of the r.h . fraction of eq . ( [ eq : fluc ] ) , while total luminosity ( at the denominator of the equation ) was safely estimated from integrated photometry of each supercluster mosaic , to include the contribution of faint , unresolved stars . following @xcite , for the present work we have determined the tip luminosity of the agb+rgb by convolving the luminosity function of each supercluster with a sobel edge - detection filter ( the zero - sum kernel [ -2,0,+2 ] ) . the tip luminosity was identified with the peak of the filter response function , after checking that there is indeed an important count discontinuity at that location in the luminosity function ; the error in the measurement equals the width of the histogram bin . we list in table [ tab_mc ] the @xmath40-band absolute fluctuation magnitude , @xmath40 absolute tip magnitude , age , and metallicity of the mc superclusters . lcccc pre & 7.70@xmath410.40&8.7 @xmath41 0.2 & 6.4 @xmath41 0.3 & 0.3 @xmath41 0.2 + i & 8.85@xmath410.12&9.8 @xmath41 0.2 & 7.0 @xmath41 0.3 & 0.3 @xmath41 0.2 + ii & 7.84@xmath410.28&8.9 @xmath41 0.2 & 7.5 @xmath41 0.3 & 0.3 @xmath41 0.2 + iii & 7.45@xmath410.24&8.3 @xmath41 0.2 & 8.0 @xmath41 0.3 & 0.1 @xmath41 0.2 + iv & 7.51@xmath410.18&8.4 @xmath41 0.2 & 8.5 @xmath41 0.3 & 0.8 @xmath41 0.2 + v & 6.69@xmath410.20&8.0 @xmath41 0.2 & 9.0 @xmath41 0.3 & 0.6 @xmath41 0.2 + vi & 6.21@xmath410.24&7.6 @xmath41 0.2 & 9.5 @xmath41 0.3 & 1.0 @xmath41 0.2 + vii & 4.92@xmath410.38&7.2 @xmath41 0.2 & 9.9 @xmath41 0.3 & 1.4 @xmath41 0.2 + [ tab_mc ] [ mcfig ] again , a tight relationship between @xmath34 and effective @xmath0 magnitudes is in place , as shown in fig . 3 , along the entire age range ( and the corresponding metallicity drift , see column 5 in table [ tab_mc ] ) . a fit to the mc data ( dashed line in the figure ) provides : @xmath42 with a data scatter @xmath43 mag across the fitting line . note that this perfectly compares , within the statistical uncertainty of the observations , with both eqs . ( [ eq : ddb ] ) ; this is true even beyond the nominal age range of the theoretical relations since , for instance , the pre - swb and the swb type i mc superclusters are too young to even sport any standard agb or rgb phase ! the link between @xmath0 and @xmath34 can be regarded as a _ much more deeply intrinsic property of ssps _ , not exclusively related to age ( or metallicity ) evolution . of course , our method can not remove , by itself , any age / metallicity degeneracy , that makes integrated photometric properties of young metal - rich stellar systems closely resemble those of older metal - poor ones ( e.g. * ? ? ? * ; * ? ? ? * ) . however , as far as old stellar populations are concerned , the unbiased and model - independent estimate of the tip stellar luminosity , even in unresolved stellar populations , provides an additional and powerful tool to quantitatively size up important effects , like the mass - loss impact on red - giant evolution . [ mloss ] relying on the @xcite standard @xmath44 parameterization , fig . 4 gives an instructive example of how mass - loss can modulate both @xmath34 and @xmath0 . in the figure we explore the behavior of an old ( 15 gyr ) ssp reference model ( big dots in the plot ) , along a full variety of mass - loss scenarios , spanning from a virtually vanishing stellar wind ( i.e. @xmath45 ) , up to a violent mass - loss rate ( @xmath46 ) capable of fully wiping out the external envelope of stars while on the rgb phase . similarly to fig . 3 , we still report the same clean relationship between @xmath34 and @xmath0 , with old agb - enhanced models ( @xmath47 ) closely resembling , all the way , young ( 2 - 5 gyr ) standard ( @xmath48 ) populations in the upper right corner of the panel . again , according to the trend of the 15 gyr model sequence vs. @xmath44 we can propose the following parametric equations : @xmath49 rcccccccc 221 & 1.90 @xmath41 0.03 & 0.198 @xmath41 0.007 & 0.88 & 6.77@xmath50 & 4.50 @xmath41 0.17 & 5.28 @xmath41 0.10 & 5.95 @xmath41 0.14 & 2.30 @xmath41 0.05 + 224 & 2.27 @xmath41 0.03 & 0.324 @xmath41 0.007 & 0.95 & 6.94@xmath51 & 3.51 @xmath41 0.17 & 4.46 @xmath41 0.09 & 5.69 @xmath41 0.14 & 1.66 @xmath41 0.07 + 1316 & 2.38 @xmath41 0.08 & 0.265 @xmath41 0.023 & 0.87 & 7.50@xmath52 & 5.0 @xmath41 0.2 & 5.39 @xmath41 0.19 & ... & 2.20 @xmath41 0.07 + 1379 & 2.11 @xmath41 0.13 & 0.269 @xmath41 0.006 & 0.88 & ... & 3.01 @xmath41 0.10 & 5.11 @xmath41 0.19 & 5.75 @xmath41 0.12 & 1.70 @xmath41 0.09 + 1387 & ... & ... & 0.98 & ... & 2.16 @xmath41 0.05 & 5.4 @xmath41 0.8 & 5.77 @xmath41 0.10 & ... + 1389 & 2.12 @xmath41 0.06 & 0.236 @xmath41 0.002 & 0.90 & ... & 3.38 @xmath41 0.09 & 5.16 @xmath41 0.20 & 6.35 @xmath41 0.13 & ... + 1399 & 2.52 @xmath41 0.03 & 0.357 @xmath41 0.007 & 0.95 & 7.30@xmath52 & 2.05 @xmath41 0.17 & 4.55 @xmath41 0.16 & 5.28 @xmath41 0.15 & 1.41 @xmath41 0.08 + 1404 & 2.39 @xmath41 0.03 & 0.344 @xmath41 0.007 & 0.95 & ... & 3.30 @xmath41 0.17 & 4.76 @xmath41 0.21 & 5.53 @xmath41 0.10 & 1.58 @xmath41 0.08 + 3379 & 2.33 @xmath41 0.03 & 0.329 @xmath41 0.007 & 0.94 & 6.77@xmath53 & 3.86 @xmath41 0.17 & 4.70 @xmath41 0.14 & 5.43 @xmath41 0.17 & 1.46 @xmath41 0.16 + 3384 & 2.20 @xmath41 0.11 & 0.296 @xmath41 0.014 & 0.91 & 6.42@xmath54 & 3.9 @xmath41 0.2 & 4.82 @xmath41 0.22 & ... & 2.05 @xmath41 0.11 + 4278 & 2.45 @xmath41 0.03 & 0.293 @xmath41 0.007 & 0.90 & ... & 2.88 @xmath41 0.17 & 4.49 @xmath41 0.22 & ... & 1.37 @xmath41 0.03 + 4374 & 2.48 @xmath41 0.03 & 0.323 @xmath41 0.007 & 0.94 & 6.77@xmath54 & 3.55 @xmath41 0.17 & ... & 5.71 @xmath41 0.25 & 1.70 @xmath41 0.04 + 4406 & 2.42 @xmath41 0.03 & 0.330 @xmath41 0.007 & 0.90 & 6.89@xmath54 & 3.72 @xmath41 0.17 & ... & 5.74 @xmath41 0.12 & 1.61@xmath41 0.16 + 4472 & 2.49 @xmath41 0.03 & 0.331 @xmath41 0.007 & 0.95 & 7.16@xmath54 & 3.42 @xmath41 0.17 & 4.64 @xmath41 0.11 & 5.64 @xmath41 0.13 & 1.52 @xmath41 0.14 + 4486 & 2.60 @xmath41 0.03 & 0.303 @xmath41 0.007 & 0.93 & 7.10@xmath55 & 2.04 @xmath41 0.17 & ... & ... & 1.38 @xmath41 0.04 + 4552 & 2.44 @xmath41 0.03 & 0.346 @xmath41 0.007 & 0.94 & ... & 2.35 @xmath41 0.17 & ... & 5.86 @xmath41 0.13 & 1.65@xmath41 0.04 + 4621 & 2.41 @xmath41 0.03 & 0.355 @xmath41 0.007 & 0.92 & ... & 3.19 @xmath41 0.17 & ... & 5.59 @xmath41 0.21 & 1.43 @xmath41 0.11 + 4649 & 2.56 @xmath41 0.03 & 0.360 @xmath41 0.007 & 0.95 & 7.22@xmath55 & 2.24 @xmath41 0.17 & ... & ... & 1.26 @xmath41 0.04 + [ tab_tot ] curiously enough , note from the equation set that @xmath34 seems slightly less sensitive to @xmath44 than @xmath0 . as we will see in a moment , such a smaller variation range for @xmath34 stems from the `` bottoming out '' effect of the rgb tip , that eventually replaces the agb in providing the brightest stars in a ssp when mass - loss increases to @xmath56 . the now established importance of mass - loss in old stellar populations dominated by stars with @xmath57 m@xmath2 was first emphasized in a series of important theoretical contributions in the early seventies @xcite ; this was urged by the intervening observational evidence that a mass spread was needed to reproduce hb and agb morphology in galactic globular clusters @xcite . to consistently match the observed agb tip in local globular clusters , for instance , @xcite suggested a fine - tuning value of @xmath58 . this calibration for old pop ii stars might however not so straightforwardly be extended , _ a priori _ , to more metal - rich environments , as one could realistically expect mass - loss efficiency to depend on ( increase with ? ) metallicity . depending on mass - loss strength , standard stellar evolution theory basically features three characteristic scenarios that constrain post - rgb evolution of low - mass stars ( see fig . 5 ) . under different physical conditions and to a different extent , each one of these cases eventually leads to the formation of hot stars , thus potentially supplying an important contribution to the integrated ultraviolet luminosity of a galaxy . _ ( a ) _ for @xmath59 , models tell us that at the end of rgb evolution , low - mass stars ignite helium in a degenerate core ( the so - called `` helium flash '' ) . a bright agb evolution has to be expected , reaching or even surpassing the rgb tip luminosity . if thermal pulses are set on ( @xmath60 ) , stars end up loosing the external envelope and generating a planetary nebula @xcite . in this case , hot post - agb nuclei of nebulae easily exceed an effective temperature of 30000 k @xcite . [ sketch ] _ ( b ) _ if @xmath61 , the pn event is aborted @xcite , and agb evolution is partially or fully inhibited @xcite . in this case , the brightest stars ( both in bolometric luminosity and in the infrared ) belong to the rgb tip , and they settle on the hb with a he core mass close or slightly exceeding 0.47 m@xmath2 ( e.g. , * ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) . for slightly increasing mass - loss efficiency , the stellar envelope becomes thinner and thinner , and the hot internal core is unveiled . if this is the case , stars likely settle at high effective temperature ( @xmath62 k ) and originate a blue hb morphology . _ ( c ) _ as a final case , for an even stronger mass - loss ( @xmath63 ) , the stellar envelope might be wiped out well before the full completion of the rgb phase . as a consequence , standard hb and agb evolution would be fully inhibited ; stars would be left `` midway '' along the rgb , with their rapidly exhausting h - shell emerging at ever shallower stellar layers . such `` hot flasher '' stars would therefore postpone the he ignition by moving straight to the high - temperature region of the c - m diagram , as is likely observed , indeed , in the cluster ngc 2808 @xcite . [ ell2 ] in terms of uv contribution , cases _ ( a ) , ( b ) _ , and _ ( c ) _ represent a sequence of increasing energy budget , as a prevailing fraction of stellar nuclear fuel is more efficiently burnt at high effective temperature . the outlined scenario has , of course , to be regarded as a general scheme , where the @xmath44 thresholds discriminating among the different evolutionary regimes slightly depend on ssp age ( through the ms turn off mass ) . while the illustrative case of fig . 5 refers to a 15 gyr ssp , one could easily verify that the overall trend for @xmath28 luminosity is maintained , for instance for a 10 gyr population , by just `` shrinking '' the @xmath44-scale of the plot by a factor of @xmath64 ; this means , for example , that _ agb - manqu _ stars are produced in a 10 gyr scenario for @xmath65 . in order to assess the relevance of the previous theoretical framework to the appearence and strength of the `` uv - upturn '' phenomenon in ellipticals , it is convenient to parameterize our analysis in terms of the @xmath66 color , that is a measure of the galaxy emission around the 1550 region vs. the johnson visual band , as originally defined by @xcite . to this end , the @xcite iue galaxy sample has been taken as a reference ; we supply in table [ tab_tot ] , for each object , the infrared effective magnitudes from @xcite , @xcite , @xcite , and @xcite and , when available , further relevant pieces of information , like stellar velocity dispersion , integrated @xmath15 color , h@xmath4 and mg@xmath67 lick indices , and the specific pn rate per unit galaxy luminosity ( from * ? ? ? * ) . as a first relevant clue in our analysis , fig . 6 is suggestive of a trend among most ellipticals , with uv - enhanced galaxies being about one mag fainter in @xmath40 effective magnitude ( some 0.7 mag in the inferred @xmath34 ) , compared to uv - poor systems .- band effective luminosities based on the @xmath68 photometry only assuming , empirically from table 2 , @xmath69 , with a conservative error bar of @xmath70 mag . the extrapolated @xmath0-mag data are displayed with a different marker on the plots.[nota ] ] on the plot one should report , however , a couple of outliers , displaying a brighter @xmath0 magnitude ( or , alternatively , a `` bluer '' @xmath66 color ) . these include ngc 1387 and ngc 1389 in the fornax cluster , and ngc 4552 in virgo . we will be back to these objects in the next section , for a brief discussion . in addition , we have to mention that the `` merger '' radio - galaxy ngc 1316 ( fornax a ) exhibits visible dust lanes that , when accounted for , may lead to an intrinsically `` bluer '' ( 1550 - @xmath71 ) color , although with possibly negligible effects on the @xmath28 sbf magnitude . after excluding these controversial objects , the data tend to display a fairly good correlation ( @xmath72 ) that , also recalling eq . ( [ eq : eta ] ) , we can approximate with @xmath73 if we entirely ascribe the dimming in red - giant tip luminosity suggested by fig . 6 to a mass - loss effect , then eq . ( [ eq : eta ] ) implies a spread @xmath74 among the galaxy population , with uv - enhanced ellipticals requiring a reimers parameter @xmath48 - 0.4 , very similar to what @xcite derived , empirically , for the galactic globular clusters . according to fig . 5 , this confirms that , overall , the agb is deployed in luminosity slightly above the rgb tip , so that the brightest stars in ellipticals should indeed always be genuine agb stars . [ fig201 ] [ fig200 ] there is an obvious caveat , though , in this simplified picture . if a standard monolithic scenario is assumed for early - type galaxy formation , then gravitational collapse of the primeval gas clouds should have proceeded over a free - fall timescale @xmath75 ( e.g. , * ? ? ? * ) , leading to a mass age metallicity relation for early - type galaxies , where more massive galaxies are both older and more metal - rich than less massive ones ( e.g. , * ? ? ? the combined action of age and metallicity plays an important role to modulate the red - giant luminosity tip , as the agb is `` naturally '' brighter among young or metal - rich ssps ( see , again , fig . the complex interplay of the different galaxy physical properties is well depicted in figs . 7 and 8 . for one , as @xcite had already shown , the uv upturn seems to appear among the most massive and metal rich ellipticals ( i.e. , there is a positive correlation with @xmath76 and the mg@xmath67 index , respectively ) . figure 7 indicates for these galaxies also a less extended agb , as for a lower stellar core - mass distribution . this feature is likewise expected to directly affect the pn rate per unit galaxy luminosity ( the so - called `` @xmath77 '' index , see fig . 8) ; indeed , there is a scantier pn stellar population among old / metal - rich ( uv - enhanced ) ellipticals @xcite . [ hbeta ] observations of local star samples show that the h@xmath4 index reaches its maximum strength for a - type stars in the temperature range t@xmath78 k @xcite . given this selected sensitivity , synthesis models of old ssps @xcite predict the integrated index to be strongly enhanced ( by roughly 0.8 or more ) in the presence of a broad color - extended hb , as observed for most metal - poor galactic globular clusters . this important piece of information could therefore usefully complement the more extreme @xmath66 color in the analysis of hb morphology for unresolved stellar populations in distant galaxies . the situation is summarized in fig . 9 , where elliptical galaxy data are compared with three illustrative cases of 15 gyr ssps with slightly supersolar metallicity ( [ fe / h]=+0.3 ) , and different ranges of hb temperature distribution . in particular , in the figure we account for _ ( i ) _ a red hb ( rhb ) morphology , that is , a clump of red stars very close to the rgb location , mimicking the real case of metal - rich galactic globular clusters , like 47 tuc ; _ ( ii ) _ an intermediate hb ( ihb ) morphology , with a broad ( roughly bell - shaped ) temperature distribution peaked about the a - stars temperature range and extending up to @xmath79 k ; _ ( iii ) _ a blue hb ( bhb ) morphology , peaked at @xmath80 k and with a tail of hot sdb stars , up to a temperature of @xmath81 k ( see * ? ? ? * , for further details ) . interestingly enough , the comparison with the observations shows that `` uv - upturn '' ellipticals _ can not _ be compatible with any bhb morphology alone . too many hot stars would make the 1550 galaxy luminosity exceedingly bright compared to the energy bulk released at the @xmath71 band . on the other hand , neither an ihb is viable , given the catastrophic impact of its a - star component on the integrated @xmath82 index of the population ( e.g. * ? ? ? definitely , an optimum match to these systems requires a mixed contribution , where the bulk of the galaxy luminosity ( over 80% ) is provided by `` standard '' ( metal - rich ) stellar populations with a red hb , upon which a 20% , or less , in terms of total optical contribution , bhb component is superposed . figure 10 provides an illustrative picture of this case , showing the synthetic c - m diagram and the integrated sed for the composite stellar population corresponding to the big triangle in fig . [ cmdhb ] so , in full agreement with @xcite conclusions , a somewhat dichotomic scenario seems to emerge for giant ellipticals , where the occurrence of the `` uv - upturn '' phenomenon is intimately related to the appearence , among the galaxy hb population , of a hot and very low - mass stellar component ( m@xmath8 m@xmath2 , e.g. * ? ? ? * ) , that at some stage begins to join the standard red - clump stars . it could be useful to note , in this regard , that two metallicity ranges seem to favor the formation of ehb stars , through an optimum combination of msto stellar masses and subsequent mass - loss via stellar winds along the rgb . as displayed in fig . 11 , [ fe / h ] values of either @xmath6 dex or @xmath7 provide in principle the needed low mass at the onset of hb evolution in a standard reimers mass - loss framework with @xmath83 . to add even further to this challenging situation , however , one also has to report that high - resolution uv spectroscopy of a sample of six uv - strong ellipticals ( namely , ngc 1399 , 3115 , 3379 , 4472 , 4552 , and 4649 ) indicates for the claimed bhb component a moderately sub - solar ( @xmath84 ) metallicity @xcite . if confirmed , this result might lead to identify the _ metal - poor _ stellar component of giant ellipticals as the one responsible for the uv excess . whatever the exact mechanism , this will cause hb morphology to readily assume a bimodal distribution , with a relative lack of warm stars at the intermedate @xmath85 range pertinent to the a spectral type ( see fig . in this paper we have carried out a synoptic analysis of the different observational features that have to do with the uv luminosity excess in elliptical galaxies . as far as the canonical picture is assumed , with old stellar populations dominating early - type galaxy luminosity , the appearence of the `` uv upturn '' should readily call for a profound change in the c - m diagram of galaxy stellar populations , not only involving the hot stellar component of the galaxy but also reverberating on red - giant evolution at the low - temperature regime . as we mainly deal with distant , unresolved stellar populations , our analysis has to rely on a combined approach , matching infrared and ultraviolet diagnostic tools in order to probe the main features of the stellar c - m diagram , starting with integrated galaxy photometry . theory of surface - brightness fluctuations provides , in this sense , a natural and quite powerful way to go deep inside the problem and , as far as the infrared wavelength interval is considered , we have demonstrated theoretically that a straight and very clean relationship is in place between a macroscopic measure , such as the galaxy fluctuation magnitude , and the corresponding individual magnitude of the brightest stars in turn at the tip of the red - giant ( agb+rgb ) phases ( fig . 2 ) . played in the @xmath28 band , this correlation leads , from a measurement of @xmath0 , to a value for @xmath34 : @xmath86 with a @xmath87 mag internal uncertainty . as we showed in sec . 2 , our ssp theoretical predictions find full support from the observations , and a direct check on the mc star clusters confirms the @xmath0 vs. @xmath34 relationship to be a much more general and deeply intrinsic property of stellar populations , virtually independent from any assumption about age , metallicity , imf , and mass loss parameters . given its nature , this relationship can not , by itself , help disentangle the problem of age / metallicity degeneracy ; however , quite fruitfully , it provides us with a very direct probe of agb properties , in a number of relevant details that directly deal with the mass - loss impact and the mass of dying stars ( fig . our effort toward exploring the infrared side of galaxy seds has a twofold aim since , as a consequence of the basic principle of energy conservation , any gram of stellar fuel spent to feed ultraviolet luminosity can not ( and will not ) be spent at longer wavelengths . this has led to the key issue of this paper , summarized in fig . 6 , that the _ strengthening of the uv rising branch is always seen to correspond to a weakening in the agb luminosity extension _ , as traced by galaxy @xmath28 fluctuation magnitude . this `` shortening '' in agb deployment is mainly recognized among giant ellipticals ( @xmath0 becomes fainter with increasing galaxy velocity dispersion , @xmath76 , see fig . 7 ) , and could mainly be ascribed to an age effect , as the agb tip naturally fades in luminosity with increasing age of the system ( fig . 3 ) , and high - mass galaxies are recognized to be older than systems of lower mass ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? however , the relationship in place likely calls for a more elaborated physical scenario , once the full range of observing evidences is added to our analysis . vs. lick mg@xmath67 index , for the galaxy sample in table [ tab_tot ] . the lick index is assumed to trace galaxy metallicity according to the @xcite calibration , as reported on the top axis of the plot . the observed decrease in @xmath34 with increasing @xmath88 $ ] can be mostly explained if more metal - rich galaxies are also older , as in a standard monolythic scenario for galaxy formation . the dashed line on the plot marks the minimum luminosity required for stars to experience the thermal pulsing phase along their agb evolution , and thus end their evolution as pne . ] _ ( a ) _ besides being old , `` uv upturn '' galaxies are also metal rich ( i.e. , stronger in mg@xmath67 lick index ) . disregarding any change in mass - loss rate , stellar tracks predict slightly more massive stars to evolve off the ms at a fixed age , with increasing metallicity ( e.g. , * * ; * ? ? ? this leads to correspondingly more massive agb stars and a brighter agb luminosity . facing the observed trend in galaxy distribution , as summarized in fig . 12 , metallicity effects evidently enter by mitigating the dimming action of age on @xmath34 with increasing galaxy mass . in any case , the interplay between age and metal abundance actually makes the derived range for the reimers mass - loss parameter ( i.e. @xmath74 , as discussed in sec . 3 ) a safe upper limit . in fact , it suggests that metal abundance does _ not _ modulate by orders of magnitude mass - loss efficiency via stellar winds .. conversely , also from data of single stars , @xcite argues that @xmath89 is metallicity - independent . from a theoretical point of view , @xcite models suggest that , if mass - loss is really proportional to metallicity , its effect to dim near - ir effective luminosities on average almost exactly offsets the brightening effect of metallicity itself . ] _ ( b ) _ a match of galaxy @xmath0 data with the calibration of fig . 5 confirms that , all the way , brightest stars in ellipticals are genuine agb members , reaching the thermal - pulsing phase ( see also fig . 12 ) , and with the agb tip exceeding the rgb tip by some 0.5 - 1.5 mag . in the temperature range of m giant stars , a major fraction of bolometric luminosity is emitted through the @xmath28 band , and bolometric correction is a nearly constant quantity that we can estimate from @xmath90 mag @xcite ; we can therefore straightforwardly translate the galaxy fluctuation magnitude into an estimate of the bolometric tip luminosity , @xmath91 , and therefrom of the corresponding stellar core mass . from our previous calibration ( eq . [ eq : calib ] ) , we can write then @xmath92\ ] ] ( where the sun has magnitude m@xmath93 ) . following @xcite , from the assumed core mass - luminosity relation for the solar metallicity range , this leads to @xmath94 . \label{mctip}\ ] ] figure 13 reports the inferred core - mass distribution , at the pn onset , for our galaxy sample . note that this is the _ maximum _ actual mass allowed to luminous stars in each galaxy environment , and demonstrates that the _ mass of dying stars tends to decrease with increasing uv - upturn strength _ , being in general @xmath95 m@xmath2 among giant ellipticals . for this mass range , pn lifetime is the largest possible , but the timescale for the nebula to be visible is critically constrained by the transition time ( @xmath96 ) needed by the post - agb stellar core to be hot enough to `` fire up '' the ejected envelope and become a hard uv emitter @xcite . the evident drop of @xmath77 among strong uv - upturn galaxies ( fig . 8) might be a direct consequence , therefore , of an increasing blocking effect of @xmath96 along the inferred @xmath97 range ( i.e. , the stellar core takes longer to heat - up than the shell to evaporate ; * ? ? ? * ) , to which one has to further add a size cut in the overall pn population , as a result of the ehb progenitors ( m@xmath98 m@xmath2 ) evolving as _ agb - manqu _ stars and therefore skipping the nebula event . _ ( c ) _ we remarked , in sec . 3 , the importance of the integrated @xmath82 index as a fairly selective tracer of the warm ( t@xmath99 k ) stellar component in the galaxy stellar population . a proper assessment of the photometric contribution from this range of temperatures is of paramount importance in the framework of early - type galaxy evolution , in order to single out any signature of recent ( i.e. , in the last few gyr or so ) star formation or , conversely , of intervening evolution of the hb morphology among old ssps , as in a more standard canonical scenario . as far as uv - upturn galaxies are concerned , the study of @xmath82 distribution clearly points to _ a substantial lack of a - type stars _ in the galaxy mix ( see fig . while , on one hand , this definitely secures the `` quiescent '' nature of these galaxies , it also poses , on the other hand , a stringent constraint on hb morphology in their old - age context . in fact , a bimodal temperature distribution is required for the hb to assure , at a time , both an enhanced uv emission _ and _ a conveniently low @xmath82 feature . thus , in these systems the expected prevailing bulk of red hb stars should be accompanied , at some point , by a residual population of blue ( metal - poor ? ) hb objects ( coincident with ehb stars in the current empirical classification scheme ) , peaked at about 20000 - 40000 k , in a proportion of , roughly , @xmath100 $ ] . on the other hand , to complete the picture , one can not neglect the masking effects of age distribution , facing a recognized evidence for low - mass ellipticals to display a more silent but also more continuous star formation along their entire galaxy life , that naturally feeds the a - star contribution , thanks to the bluer ms turn - off point ( msto ) exhibited by ssps in the @xmath101 gyr age range , and leads to a younger average " age , compared to high - mass systems . this is what we observe , for instance , among the resolved stellar populations of the local group dwarf spheroidals @xcite ( see , in this regard , the illustrative location of m32 in fig . 9 , and also consider the discussion by * ? ? ( d ) _ along with our discussion of the @xmath0 vs. @xmath66 relation , we noticed , in fig . 6 , the presence of a few outliers about 0.7 mag brighter at infrared magnitudes , or alternatively @xmath102 mag `` bluer '' in the @xmath66 color , than the main galaxy population . in order to further investigate this issue , we tracked the relevant objects also in other figures , whenever possible . if galaxy mass ( _ alias _ @xmath103 ) is considered as the leading physical parameter to compare outlier location with respect to the bulk of the galaxy distribution ( see , for instance fig . 7 ) , one must conclude that both ngc 4552 and ngc 1389 seem to have a brighter agb tip rather than a bluer @xmath66 color . with regard to ngc 1389 , @xcite find that the @xmath104 sbfs of ngc 1389 are also too bright compared to its ( @xmath105 ) color , a fact that would be consistent with either a higher than average metallicity given the age of its most recent burst of star formation or a longer lifetime of its tp - agb stars @xcite . in the case of ngc 4552 , however , @xcite contribute an interesting piece of information . these authors measure near - ir sbfs for several galaxies in virgo , and their results are systematically fainter than those obtained by @xcite . unfortunately , the two groups use slightly different filters ( jensen et al . employ @xmath106 , vs. @xmath40 of pahre & mould ) , but the discrepancy is larger than can be ascribed to the effect of the filters . in particular , jensen et al . find @xmath107 for ngc 4552 ; assuming @xmath108 , ngc 4552 would no longer be deviant in the @xmath104 vs. ( @xmath109 ) plane . concerning ngc 1387 , the dearth of data for this galaxy in the literature ( see table 2 ) makes it hard to propose an origin for its departure from the @xmath104 vs. ( @xmath109 ) sequence . vs. ( @xmath105 ) correlation determined by @xcite . ] on the other hand , ngc 1387 and ngc 1389 are lenticular galaxies , like the merger remnant ngc 1316 , but so are ngc 3384 and ngc 4406 , both of which do not deviate from the correlation . at any rate , ngc 1387 and ngc 1389 constitute privileged candidates for any future `` in - depth '' investigation . we would like to thank gustavo bruzual for providing us with his latest ssp models , in advance of publication , and livia origlia , for useful discussions . the anonymous referee is also acknowledged for his / her competent suggestions , that greatly helped refine the main focus of the paper . partial financial support is acknowledged from the italian miur , under grant inaf - prin05 1.06.08.03 , and mexican conacyt , under grant no . 48589-f , and dgapa , under grant in111007 . bailyn , c. d. 1995 , , 33 , 133 blakeslee , j. p. , vazdekis , a. , & ajhar , e. a. 2001 , , 320 , 193 boothroyd , a. i. , & sackmann , i .- 1988 , , 328 , 641 bressan , a. , chiosi , c. , & fagotto , f. 1994 , , 94 , 63 bressan , a. , chiosi , c. , & tantalo , r. 1996 , , 311 , 425 brodie , j. p. , & huchra , j. p. 1990 , , 362 , 503 brown , t. m. 2004 , astrophys . space sci . , 291 , 215 brown , t. m. , ferguson , h. c. , davidsen , a. f. , dorman , b. 1997 , 482 , 685 brown , t. m. , ferguson , h. c. , stanford , s. a. , deharveng , j .- m . 1998 , , 504 , 113 brown , t. m. , bowers , c. w. , kimble , r. a. , sweigart , a. v. , & ferguson , h. c. 2000 , , 532 , 308 brown , t. m. , ferguson , h. c. , smith , e. , bowers , c. w. , kimble , r. a. , renzini , a. , & rich , r. m. 2003 , , 584 , l69 brown , d. , yi , s. , han , z. , & yoon , s .- j . 2006 , baltic astronomy , 15 , 13 bruzual , g. , 2007 , in proc . of the iau symp . 241 `` stellar populations as building blocks of galaxies '' , eds . a. vazdekis and r. peletier ( cambridge : cambridge univ . press ) , p.125 ( see also astro - ph/0703052 ) bruzual , g. , & charlot , s. 2003 , , 344 , 1000 buonanno , r. , corsi , c. , bellazzini , m. , ferraro , f. r. , & pecci , f. f. 1997 , , 113 , 706 burstein , d. , bertola , f. , buson , l. m. , faber , s. m. , & lauer , t. r. 1988 , , 328 , 440 buson , l. m. , bertone , e. , buzzoni , a. , & carraro , g. 2006 , baltic astronomy , 15 , 49 busso , g. , moehler , s. , zoccali , m. , heber , u. , & yi , s. k. 2005 , , 633 , l29 buzzoni , a. 1989 , , 71 , 817 buzzoni , a. 1993 , , 275 , 433 buzzoni , a. 1995 , , 98 , 69 buzzoni , a. 1998 in t. zanzu , v. testa & m. bellazzini eds . , evolving evolution , oss . di cagliari , cagliari , p. 13 ( see also astro - ph/9811382 ) buzzoni , a. 2008 , in d. valls - gabaud & m. chavez eds . , `` resolved stellar populations '' , asp conf . ( asp : san francisco ) , in press ( see also astro - ph/0509602 ) buzzoni , a. , gariboldi , g. , & mantegazza , l. 1992 , , 103 , 1814 buzzoni , a. , mantegazza , l. , & gariboldi , g. 1994 , , 107 , 513 buzzoni , a. , arnaboldi , m. , & corradi , r. l. m. 2006 , , 368 , 877 cantiello , m. , raimondo , g. , brocato , e. , & capaccioli , m. 2003 , , 125 , 2783 castellani , v. , & renzini , a. 1968 , , 2 , 310 castellani m. , & tornamb a. 1991 , apj , 381 , 393 castellani , v. , giannone , p. , & renzini , a. 1970 , , 9 , 418 castellani m. , limongi m. , tornamb a. 1992 , , 389 , 227 castellani , v. , iannicola , g. , bono , g. , zoccali , m. , cassisi , s. , & buonanno , r. 2006 , , 446 , 569 catelan , m. , ferraro , f. r. , & rood , r. t. 2001 , , 560 , 970 cervio , m. , & luridiana , v. 2006 , , 451 , 475 cervio , m. , luridiana , v. , & castander , f. j. 2000 , , 360 , l5 cervio , m. , valls - gabaud , d. , luridiana , v. , & mas - hesse , j. m. 2002 , , 381 , 51 chabrier , g. 2003 , , 115 , 763 charbonnel , c. , meynet , g. , maeder , a. , & schaerer , d. 1996 , , 115 , 339 chiosi , c. , bertelli , g. , bressan , a. 1992 , , 30,235 code , a. d. , & welch , g. a. 1979 , , 228 , 95 cohen j. g. 1982 , , 258 , 143 colina , l. , bohlin , r. c. , & castelli , f. 1996 , instrument science report cal / scs-008 ( baltimore , stsci ) cox , j.p . & giuli , r.t . 1968 principles of stellar structure , gordon & breach sci . ltd , new york . dcruz , n. l. , dorman , b. , rood , r. t. , & oconnell , r. w. 1996 , , 466 , 359 dcruz , n. l. , morgan , s. m. , bhm - vitense , e. 2000 , , 120 , 990 demleitner , m. , accomazzi , a. , eichhorn , g. , grant , c. s. , kurtz , m. j. , & murray , s. s. 2001 , in astronomical data analysis software and systems x , asp conf . 238 , eds . f. r. harnden , jr . , francis a. primini , and harry e. payne ( san francisco : asp ) , p.321 de vaucouleurs , g. , de vaucouleurs , a. , corwin , h.g . , buta , r.j . , paturel , g. , fouque , p. 1991 , third reference catalog of bright galaxies . springer verlag , heidelberg dorman , b. , rood , r.t . , oconnell , r.w . 1993 , , 419 , 596 dorman , b. , oconnell , r. w. , & rood , r. t. 1995 , , 442 , 105 demarque , p. , mengel , j. g. , & sweigart , a. v. 1972 , , 173 , l27 elson , r. a. w. , & fall , s. m. 1985 , , 299 , 211 elson , r. a. , & fall , s. m. 1988 , , 96 , 1383 ferguson , h.c . , & davidsen , a.f . 1993 , , 408 , 92 frogel , j.a . , mould , j. , & blanco , v.m . 1990 , , 352 , 96 fusi pecci , & f. , renzini , a. 1976 , , 46 , 447 gonzlez , r. a. , liu , m. c. , & bruzual a. , g. 2004 , , 611 , 270 height0.4pt depth0.0pt width1.0 in .2005 , , 621 , 557 gonzlez - lpezlira , r. a. , albarrn , m. y. , mouhcine , m. , liu , m . c. , bruzual - a . , g. , & de batz , b. 2005 , , 363 , 1279 greggio , l. , & renzini , a. 1990 , , 364 , 35 groenewegen , m. a. t. , smith , c. h. , wood , p. r. , omont , a. , & fujiyoshi , t. 1995 , , 449 , l119 hui x. , ford h.c . , ciardullo r. , jacoby g.h . 1993 , apj , 414 , 463 iben , i. j. , & rood , r. t. 1970 , , 161 , 587 iben , i. , jr . , & renzini , a. 1983 , , 21 , 271 jensen , j. b. , luppino , g. a. , & tonry , j. l. 1996 , , 468 , 519 jensen , j. b. , tonry , j. l. , & luppino , g. a. 1998 , , 505 , 111 jensen , j. b. , tonry , j. l. , barris , b. j. , thompson , r. i. , liu , m. c. , rieke , m. j. , ajhar , e. a. , & blakeslee , j. p. 2003 , , 583 , 712 johnson , h. l. 1966 , , 4 , 193 kaluzny , j. , & udalski , a. 1992 , acta astronomica , 42 , 29 kobayashi , c. , & arimoto , n. 1999 , , 527 , 573 kuntschner , h. , lucey , j. r. , smith , r. j. , hudson , m. j. , & davies , r. l. 2001 , , 323 , 615 larson , r.b . 1974 , , 166 , 585 lee , m. g. , freedman , w. l. , & madore , b. f. 1993 , , 417 , 553 lee , y .- w . , ree , c. h. , rich , r.m . , et al . 2005 , , 619 , l103 liu , m. c. , graham , j. r. , & charlot , s. 2002 , , 564 , 216 maraston , c. , greggio , l. , & thomas , d. 2001 , , 276 , 893 marigo , p. , girardi , l. , weiss , a. , groenewegen , m. a. t. , & chiosi , c. 2004 , , 423 , 995 marigo , p. , & girardi , l. 2007 , , 469 , 239 mateo , m. l. 1998 , , 36 , 435 mouhcine , m. , gonzlez , r. a. , & liu , m. c. 2005 , , 362 , 1208 oconnell , r. w. 1980 , , 236 , 430 oconnell , r. w. 1999 , , 37 , 603 paczyski , b. 1970 , acta astronomica , 20 , 47 pahre , m. a. , de carvalho , r. r. , & djorgovski , s. g. 1998 , , 116 , 1606 pahre , m. a. , & mould , j. r. 1994 , , 433 , 567 park , j .- h . , & lee , y .- w . 1997 , , 476 , 28 paturel , g. , petit , c. , prugniel , p. , theureau , g. , rousseau , j. , brouty , m. , dubois , p. , & cambrsy , l. 2003 , , 412 , 45 peimbert m. 1990 , rev . , 20 , 119 persson , s. e. , murphy , d. c. , krzeminski , w. , roth , m. , & rieke , m. j. 1998 , , 116 , 2475 piersanti , l. , tornamb , a. , & castellani , v. 2004 , , 353 , 243 raimondo , g. , brocato , e. , cantiello , m. , & capaccioli , m. 2005 , , 130 , 2625 recio - blanco , a. , aparicio , a. , piotto , g. , de angeli , f. , & djorgovski , s. g. 2006 , , 452 , 875 ree , c. h. , lee , y .- w . , sohn , y .- j . , yi , s. k. , yoon , s .- j . , & the galex science team 2006 , baltic astronomy , 15 , 5 reimers , d. 1975 , mem . soc . lige , 6th ser . , 8 , 87 renzini , a. 1977 , in advanced stages in stellar evolution i. , eds . iben jr . , a. renzini , d.n . schramm ( heidelberg : springer verlag ) . renzini , a. 2006 , , 44 , 141 renzini , a. , & buzzoni , a. 1986 , in spectral evolution of galaxies , eds . c. chiosi & a. renzini ( dordrecht : reidel ) , p. 195 renzini , a. , & fusi pecci , f. 1988 , , 1988 , 26 , 199 rich , r. m. , et al . 1997 , , 484 , l25 rose , j.a . , & deng , s. 1999 , , 117 , 2213 schnberner , d. 1983 , , 272 , 708 schiavon , r.p . , caldwell , n. , & rose , j.a . 2004 , , 127,1513 searle , l. , wilkinson , a. , & bagnuolo , w. g. 1980 , , 239 , 803 seidel , e. , da costa , g. s. , & demarque , p. 1987 , , 313 , 192 skrutskie , m. f. et al . 1997 , assl vol . 210 : the impact of large scale near ir sky surveys , 25 stanghellini , l. , & renzini , a. 2000 , , 542 , 308 sweigart , a. v. 1997 , , 474 , l23 sweigart , a. v. , & gross , p. g. 1976 , , 32 , 367 sweigart , a. v. , brown , t. m. , lanz , t. , landsman , w. b. , hubeny , i. 2002 in omega centauri , a unique window into astrophysics . asp conf . , vol . 265 . f. van leeuwen , j. d. hughes , & g. piotto ( san francisco : asp ) p.261 tantalo , r. , chiosi , c. , bressan , a. , & fagotto , f. 1996 , , 311 , 361 tonry , j. l. 1991 , , 373 , l1 tonry , j. , & schneider , d. p. 1988 , , 96 , 807 unglaub , k. , bues , i. 2001 , , 374 , 570 van den bergh , s. 1962 , , 67 , 486 van loon , j. t. 2000 , , 354 , 125 worthey , g. 1993 , , 409 , 530 worthey , g. 1994 , , 95 , 107 yi , s. k. , & yoon , s .- j . 2004 , , 291 , 205 yi s. , demarque p. , oemler a. j. 1998 , , 492 , 480

the radially outward flow of fluid into a porous medium occurs in many practical problems , from transport across vascular walls to the pressurisation of boreholes . as the driving pressure becomes non - negligible relative to the stiffness of the solid structure , the poromechanical coupling between the fluid and the solid has an increasingly strong impact on the flow . for very large pressures or very soft materials , as is the case for hydraulic fracturing and arterial flows , this coupling can lead to large deformations and , hence , to strong deviations from a classical , linear - poroelastic response . here , we study this problem by analysing the steady - state response of a poroelastic cylinder to fluid injection . we consider the qualitative and quantitative impacts of kinematic and constitutive nonlinearity , highlighting the strong impact of deformation - dependent permeability . we show that the wall thickness ( thick _ vs_. thin ) and the outer boundary condition ( free _ vs_. constrained ) play a central role in controlling the mechanics .

1609.09457

the radially outward flow of fluid into a porous medium is central to many practical problems in , for example , geomechanics , biophysics , and filtration . in geomechanics , pile driving involves the mechanical expansion of a cylindrical cavity in a fluid - saturated soil , generating large pore pressures in the surrounding medium that gradually relax through consolidation ( _ e.g. _ , * ? ? ? similarly , fluid injection into boreholes involves the pressurisation of a cylindrical cavity in a soil or rock , driving flow radially outward into the surrounding medium ( _ e.g _ , * ? ? ? * ; * ? ? ? * ; * ? ? ? biophysical applications include injection into subcutaneous tissue ( _ e.g _ , * ? ? ? * ) and blood flow through arteries and vascular networks , which have permeable walls ( _ e.g _ , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? radially outward flow is also relevant to the design of cylindrical filters ( _ e.g _ , * ? ? ? in many of these cases , the driving pressure is sufficiently large relative to the stiffness of the solid structure that the poromechanical coupling between the fluid and the solid has an important impact on the flow . classically , this coupling is described by the iconic theory of linear poroelasticity ( _ e.g _ , * ? ? ? * ; * ? ? ? * ) , which combines darcy s law with linear elasticity in a linearised kinematic framework and is valid for infinitesimal deformations of the solid . however , soft materials such as biological tissues , weak materials such as soils , thin structures such as vasculature , and scenarios involving large injection pressures such as hydraulic fracturing may result in substantial deformations that violate this linear theory . large deformations are inherently nonlinear from the perspective of kinematics , and typically also result in nonlinear constitutive behaviour such as nonlinear elasticity and deformation - dependent permeability . the role of nonlinearity has been considered in detail in the context of uniaxial deformations , both mathematically and experimentally ( _ e.g. , _ * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? the uniaxial problem is important for a variety of practical applications ; for example , many composite manufacturing processes involve the uniaxial injection of a resin gel or metal melt into a deformable porous matrix @xcite . mathematically , the uniaxial problem is inherently simple since the flow and deformation fields are strictly one - dimensional and the exact relationship between displacement and porosity is linear @xcite . radial deformations are more challenging despite the fact that the velocity and displacement fields remain one - dimensional , since the stress and strain fields become biaxial and the exact relationship between the porosity and displacement becomes nonlinear . radial poroelastic deformations have been studied using linear poroelasticity in the context of both fluid injection or extraction from boreholes ( _ e.g _ , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) and arterial blood flow ( _ e.g. _ , * ? ? ? * ; * ? ? ? nonlinear effects have attracted interest primarily in the latter case , specifically in the context of fluid flow through artery walls . for example , @xcite introduced deformation - dependent permeability within a linear poroelastic framework . @xcite and @xcite accounted partially for the nonlinear kinematics of large deformations while retaining linear elasticity . in a different context , @xcite developed a rigorous and fully nonlinear model , but for a strictly volumetric constitutive law and assuming constant permeability . none of these previous works explicitly defined or explored the general parameter space for axisymmetric deformations , nor did they systematically assess the relative importance of nonlinear kinematics , nonlinear elasticity , and deformation - dependent permeability . here , we consider the axisymmetric deformation of a poroelastic cylinder driven by radially outward fluid flow using a rigorous , fully nonlinear model . we focus , in particular , on the qualitative and quantitative implications of the simplifications of linear poroelasticity , the separate roles of nonlinear kinematics , nonlinear elasticity , and deformation - dependent permeability , and the nontrivial coupling of these with the geometry and boundary conditions . we show , in particular , that the wall thickness and the outer boundary condition play crucial roles in controlling the mechanics of the problem . we consider the radially outward injection of fluid from the centre of a porous cylinder of inner radius @xmath0 and outer radius @xmath1 . we assume axisymmetry and model the 2d annular cross - section , assuming that the material is constrained in the axial direction and is therefore in plane strain . we assume that the inner boundary is mechanically free so that the inner radius @xmath2 expands in response to injection . we assume that the outer boundary is either subject to a constant effective stress @xmath3 , in which case the outer radius @xmath4 also expands in response to injection ( fig . [ fig : rad ] , left ) , or that the outer boundary is constrained such that the outer radius @xmath5 is fixed ( fig . [ fig : rad ] , right ) . the latter situation is useful for comparison to numerical simulations and experiments ( _ e.g. _ , * ? ? ? * ) . and initial outer radius @xmath6 . the inner radius is free to expand , while the outer boundary is either subject to a constant radial effective stress @xmath3 ( left ) or fixed in place ( right ) . note that we assume plane strain and adopt the convention of tension being positive . [ fig : rad ] , width=480 ] large - deformation poroelasticity is a continuum approach to modelling the interactions of two superposed phases , a porous solid skeleton and an interstitial fluid ( _ e.g. _ , * ? ? ? we next summarise this theory in the context of axisymmetric flow and deformation . the fluid velocity @xmath7 , solid displacement @xmath8 , and solid velocity @xmath9 each have only one component , @xmath10 where the subscripts @xmath11 and @xmath12 denote quantities related to the solid and to the fluid , respectively , @xmath13 is the radial coordinate @xmath14 , @xmath15 is time , and @xmath16 is the radial unit vector . we work in an eulerian ( spatial ) reference frame , such that the displacement is given by @xmath17 where @xmath18 denotes the reference position of the material that is located at position @xmath13 at time @xmath15 . without loss of generality , we take @xmath19 such that @xmath20that is , we adopt the initial configuration as the reference configuration . the deformation is fully characterised by the deformation gradient tensor @xmath21 where @xmath22 denotes the identity tensor and @xmath23 the inverse . for an axisymmetric deformation , this can be written @xmath24 where @xmath25 , @xmath26 , and @xmath27 are the three principal stretch ratios . are the eigenvalues of @xmath28 . ] for plane strain , these are given by @xmath29 since only @xmath25 and @xmath26 are distinct and nontrivial , this is a two - dimensional state of strain . the jacobian determinant @xmath30 det@xmath31 measures the local volume change , @xmath32 we assume that the solid and fluid phases are individually incompressible , such that deformation occurs only through rearrangement of the solid skeleton with corresponding changes in the local porosity or fluid fraction , @xmath33 . this then requires that @xmath34 where @xmath35 is the reference ( initial ) porosity , which we take to be uniform . combining eqs . ( [ stretches][jac2 ] ) , we obtain an explicit nonlinear expression for porosity in terms of displacement , @xmath36 conservation of mass for the fluid - solid mixture is given by @xmath37=0,\ ] ] where @xmath38 is the local solid fraction . conservation of solid volume requires that @xmath39 and it can be shown that eq . is identically satisfied by eq . ( [ eq : phi_to_u ] ) , subject to the kinematic boundary conditions @xmath40 and @xmath41 , where @xmath42 and @xmath43 denote the initial inner and outer radii , respectively . we assume that fluid flows relative to the solid skeleton according to darcy s law . in the absence of gravity and other body forces , this can be written @xmath44 where @xmath45 is the dynamic viscosity of the fluid , @xmath46 is the fluid ( pore ) pressure , and @xmath47 is the permeability , which we take to be an isotropic function of porosity ( see 2[ss : permeability ] ) . we model injection as a line source at the origin with flow rate per unit length @xmath48 . thus , eqs . ( [ cons ] ) can be summed and integrated to give @xmath49 = \hat{q}(t).\ ] ] combining eq . ( [ q ] ) with eqs . ( [ cons ] ) and ( [ darcy ] ) , we eliminate @xmath50 , @xmath51 , and @xmath52 to obtain @xmath53 where along the way we obtain expressions for @xmath52 and @xmath51 , @xmath54 we next link the fluid pressure to the stress in the solid . mechanical equilibrium requires that @xmath55 where @xmath56 is the total stress supported by the fluid - solid mixture , and we neglect inertia as well as the effect of gravity and other body forces . the total stress can be decomposed as @xmath57 where terzaghi s effective stress @xmath58 is the portion of the stress supported through deformation of the solid skeleton , and where we adopt the convention of tension being positive . equation ( [ stressful2 ] ) provides mechanical coupling between the fluid and the solid . combining eqs . ( [ mecheqb ] ) and ( [ stressful2 ] ) leads , for an axisymmetric deformation , to @xmath59 where @xmath60 and @xmath61 are the radial and azimuthal ( hoop ) components of the effective stress , respectively . we have now considered kinematics , darcy s law , terzaghi s effective stress , and mechanical equilibrium . the model thus far is exact , assuming only that the fluid and solid constituents are individually incompressible . the common assumption of infinitesimal deformations leads to classical linear poroelasticity @xcite . this corresponds here to the assumptions that @xmath62 and @xmath63 . note that this will clearly be a bad assumption near the inner radius if @xmath64 becomes comparable to @xmath65 . linearising eqs . ( [ eq : phi_to_u ] ) and ( [ governalso ] ) leads to @xmath66 respectively . note that eq . ( [ eq : vol_rad ] ) is not identically satisfied by the kinematic expression in eq . ( [ eq : phi_to_ulin ] ) , implying that the linearised model is not rigorously mass conservative . we next consider the constitutive behaviour of the solid . the relationships between stress and strain and between strain and displacement are constitutive laws for the solid skeleton . we assume that the solid deforms elastically , meaning that these relationships are quasi - static ( _ i.e. _ , rate independent ) and reversible ( _ i.e. _ , history independent ) . we investigate the impact of this relationship on the results by considering both linear and nonlinear elasticity laws . hencky elasticity is a simple , nonlinear , hyperelastic model that provides good agreement with experiments for moderate deformations @xcite . hencky elasticity has several advantageous properties @xcite , including the fact that it uses the same elastic parameters as linear elasticity @xcite . here , we use lam s first parameter @xmath67 and the @xmath46-wave or oedometric modulus @xmath68 . for the displacement field given in eq . , the hencky strain tensor is @xmath69,\ ] ] which again has two nontrivial components since axisymmetric displacement leads to both radial and azimuthal ( `` hoop '' ) strains . the associated cauchy effective stress for hencky elasticity is @xmath70.\ ] ] on substitution of eq . ( [ henckconst ] ) into eq . ( [ govern ] ) , we arrive at @xmath71 the right - hand side of eq . is a function of @xmath64 only . in combination with eqs . ( [ eq : phi_to_u ] ) and ( [ governalso ] ) , this then provides a nonlinear partial differential equation ( pde ) for @xmath64 . linear elasticity combines a linear relationship between strain and displacement with a linear relationship between stress and strain . the linear ( small or infinitesimal ) strain tensor is @xmath72\ ] ] with the associated linear stress tensor @xmath73.\ ] ] on substitution of eq . ( [ eq : sigma_small_rad ] ) into eq . ( [ govern ] ) , we obtain @xmath74 + \frac{\mathcal{m}-\lambda}{r}\left(\frac{\partial{u_s}}{\partial{r}}-\frac{u_s}{r}\right)=\mathcal{m}\,\frac{\partial}{\partial{r } } \left[\frac{1}{r}\frac{\partial}{\partial{r}}\left(ru_s\right)\right].\ ] ] linear elasticity is in some sense an idealised constitutive behaviour that most materials will approximately follow for infinitesimal deformations , and from which most materials will deviate as deformations become finite . for example , hencky elasticity reduces to linear elasticity for infinitesimal deformations ; that is , eqs . ( [ eq : h_rad ] ) and ( [ henckconst ] ) reduce to eqs . ( [ eq : eps_small_rad ] ) and eq . ( [ eq : sigma_small_rad ] ) , respectively , for @xmath75 and @xmath76 . alternatively , linear elasticity can instead be viewed as an exact constitutive law for an idealised material , for which it would be valid for arbitrarily large deformations . equation ( [ eq : p_to_u_small_rad ] ) can be combined with eqs . ( [ eq : phi_to_u ] ) and ( [ governalso ] ) to provide a pde for @xmath64 . in what follows , we use `` hencky elasticity '' to refer to eqs . ( [ eq : h_rad][eq : gradp_to_u_rad_large ] ) and `` linear elasticity '' to refer to eqs . ( [ eq : eps_small_rad][eq : p_to_u_small_rad ] ) . we now combine linearised kinematics ( 2(a)([lin ] ) ) with linear elasticity ( 2(b)([eq : linearelasticity ] ) ) . this then allows us to write eq . ( [ eq : p_to_u_small_rad ] ) directly in terms of @xmath33 using eq . ( [ eq : phi_to_ulin ] ) , @xmath77 equation ( [ eq : phi_to_ulin ] ) can then be rewritten as a linear second - order parabolic pde for @xmath33 . the solid skeleton deforms through rearrangement of the pore structure , leading to changes in the porosity . this is then likely to alter the permeability of the material . for infinitesimal deformations , this effect is second - order in the deformation , and is therefore typically neglected . we consider the impact of this simplification by comparing results for constant permeability with results for deformation - dependent permeability . as in @xcite , we adopt a normalised kozeny - carman formula , @xmath78 where @xmath79 is the reference permeability . although not quantitatively appropriate for all materials , this relation captures the important qualitative behaviour that @xmath47 vanishes as @xmath33 vanishes and @xmath47 diverges as @xmath33 tends to one . note that many materials have a naturally anisotropic permeability . in addition , anisotropic deformations may lead to the emergence of anisotropic permeability . for example , fluid flow through the walls of a porous cylinder leads to compression in the radial direction and stretching in the azimuthal direction , which might be expected to reduce the azimuthal permeability while enhancing the radial permeability . we neglect natural anisotropy here for simplicity , and induced anisotropy is irrelevant under the requirement of axisymmetry . before injection , the porosity is uniform , @xmath80 , the fluid and the solid are at rest , @xmath81 , and the material is relaxed , @xmath82 . we take this initial state to be the reference state , such that @xmath19 . for @xmath83 , we assume that fluid is injected from the origin either at an imposed constant volume flow rate per unit length @xmath84 or via an imposed constant pressure drop @xmath85 . it is straightforward to enforce the former condition since @xmath84 appears explicitly in the pde . enforcing the latter condition is less straightforward ( see appendix ) . the inner boundary is mechanically free , thus the normal effective stress must vanish . the inner boundary is also a material boundary . hence , the appropriate mechanical and kinematic conditions are @xmath86 we consider two distinct sets of conditions at the outer boundary . in both cases , we assume without loss of generality that the fluid pressure vanishes at the outer boundary , @xmath87 if the outer boundary is subject to an applied effective stress , then this is a moving boundary . the appropriate mechanical and kinematic conditions are @xmath88 three conditions are required because the outer radius @xmath89 is unknown , and must be determined as part of the solution . alternatively , if the outer boundary is constrained such that its position is fixed , then the appropriate conditions are @xmath90 this scenario requires only two conditions because the outer radius @xmath1 is fixed and known . the normal component of the effective stress at the outer boundary @xmath91 is unknown , but does not need to be determined as part of the solution . conditions ( [ outerbcb ] ) are convenient for comparison with experiments and numerical simulations ( _ e.g. _ , @xcite ) , and are relevant to industrial applications such as filtration . conditions ( [ outerbca ] ) are likely to be more relevant to biomedical and geotechnical applications . for the kinematically rigorous models , conditions at the inner and outer boundaries ( eqs . [ innerbc][outerbcb ] ) are applied at @xmath92 and @xmath89 , respectively . for the kinematically linearised models , these are instead applied at @xmath65 and @xmath6 , respectively ( _ e.g. _ , @xmath93 ) . to proceed , we non - dimensionalise via the scaling @xmath94 where @xmath95 is the characteristic poroelastic timescale . we can then rewrite eq . ( [ governalso ] ) in dimensionless form , @xmath96 where @xmath97 . injection is characterised either by a fixed dimensionless flow rate @xmath98 or by a fixed dimensionless pressure drop @xmath99 , @xmath100 where , in the latter case , @xmath101 must be calculated from @xmath99 as part of the solution . both of these quantities compare the characteristic pressure due to injection with the characteristic elastic stiffness of the material . the model is additionally characterised by the value of @xmath35 and three other dimensionless parameters : @xmath102 where @xmath103 compares the bulk modulus to the shear modulus ( @xmath104 $ ] , where @xmath105 corresponds to an incompressible material ) . we work in dimensionless quantities from here onwards ; hence , we drop the tildes for convenience . thus far , we have developed several different models for the response of a poroelastic cylinder to radially outward flow by considering two different representations of the kinematics ( linearised and rigorous ) , two different elasticity laws ( linear and hencky ) , and two different permeability laws ( constant and kozeny - carman ) . we categorise these models as linear `` l '' ( linearised kinematics with linear elasticity ) , quasi - linear `` q '' ( rigorous kinematics with linear elasticity ) , and nonlinear `` n '' ( rigorous kinematics with hencky elasticity ) . for each of these , we consider both constant `` @xmath106 '' and kozeny - carman `` @xmath107 '' permeability . we then have six combinations : l-@xmath106 , l-@xmath107 , q-@xmath106 , q-@xmath107 , n-@xmath106 , and n-@xmath107 . note that l-@xmath106 is classical linear poroelasticity and n-@xmath107 is fully nonlinear poroelasticity ; the other four models are intermediate between these two extremes . note also that we do not combine linearised kinematics with hencky elasticity because this is asymptotically inconsistent ; linearising the kinematics requires that @xmath108 and @xmath109 , under which assumptions hencky elasticity reduces to linear elasticity . we now seek solutions to the above models at steady state , for which the fluid velocity is steady ( @xmath110 ) and the solid is stationary ( @xmath111 ) . combining eqs . ( [ cons ] ) , ( [ darcy ] ) , and ( [ govern ] ) , we have @xmath112 where @xmath113 $ ] . combining this with an elasticity law , a permeability law , and a kinematic relationship between @xmath64 and @xmath33 then leads to a second - order ode in @xmath64 for all models . for linear elasticity ( l and q models ) , we combine eq . ( [ cauchy ] ) with eq . ( [ eq : sigma_small_rad ] ) to arrive at @xmath114}.\ ] ] for hencky elasticity ( n models ) , we combine eq . ( [ cauchy ] ) with eq . ( [ henckconst ] ) to arrive at @xmath115+(1-\gamma)\ln\left(\lambda_\theta/\lambda_r\right)-q\lambda_r\lambda_\theta / k[\phi_f(u_s)]}{\lambda_r r \{1-[\ln(\lambda_r)+\gamma\ln(\lambda_\theta)]\}},\ ] ] where the stretches are defined in eq . ( [ stretches ] ) . note that eqs . ( [ a ] ) and ( [ odeelastreg ] ) are valid for any permeability law , boundary conditions , and treatment of kinematics . thus , we have a boundary value problem ( bvp ) comprising a second - order ode ( eq . [ a ] or eq . [ odeelastreg ] ) with two constraints at the inner boundary ( eqs . [ innerbc ] ) and either three or four constraints at the outer boundary , depending on whether the outer boundary is fixed ( eqs . [ p ] and [ outerbcb ] ) or not ( eqs . [ p ] and [ outerbca ] ) . for the l-@xmath106 and q-@xmath106 models , the ode ( eq . [ a ] ) can be solved analytically ( see appendix ) . for the l-@xmath106 model , this provides the full solution to the problem . for the q-@xmath106 model , it remains to solve an implicit algebraic system for @xmath0 and , depending on the outer boundary condition , for @xmath1 . this can be implemented with standard numerical root - finding techniques . for the other four models , the ode can not be solved analytically and we instead solve it numerically using a chebyshev spectral collocation method , as described in 3[pseudo ] . an imposed flow rate @xmath98 will lead to a steady - state pressure drop @xmath116 . the latter is not needed as part of the solution , but can be calculated readily via the integration of eq . ( [ eq : vs ] ) , giving @xmath117 in contrast , an imposed pressure drop @xmath116 will lead to a steady - state flow rate @xmath98 that must be calculated as part of the solution by rearranging eq . ( [ eq : q_to_p ] ) . for constant permeability , this relationship becomes @xmath118 everything else being fixed , the same steady state can therefore be achieved by imposing either @xmath98 or @xmath116 . clearly , the geometry and boundary conditions will have a strong impact on the relationship between @xmath98 and @xmath116 . we explore this relationship in the next section . when the ode can not be solved analytically , it must instead be integrated using standard numerical methods for bvps , such as direct finite differences or a shooting method . for a shooting method , one must guess the locations of the free boundaries , solve the ode as an initial value problem ( ivp ) subject to two of the constraints , and then iterate on the guesses until the remaining constraints are satisfied . for direct finite differences , two approaches are possible . one may follow the same approach as for a shooting method , but solving the bvp directly using finite differences and root finding ( _ e.g. _ , newton s method ) rather than solving it as an ivp . alternatively , one may solve the bvp and all constraints simultaneously using finite differences and root finding . although straightforward to implement , these approaches are unreliable in the present context because the iteration process can easily lead to a nonphysical state that prohibits further iteration . to mitigate these difficulties , we instead use a direct method based on chebyshev spectral collocation ( _ i.e. _ , a chebyshev pseudospectral method ) ( _ e.g. _ , * ? ? ? * ; * ? ? ? that is , we solve the bvp and all constraints simultaneously as described above , but replacing the sparse finite - difference differentiation matrix with a dense chebyshev - pseudospectral differentiation matrix . this approach still requires newton iteration , but is more robust than finite differences because the density of the pseudospectral differentiation matrix directly couples the solution at each discrete point to the solution at every discrete point . this approach also allows for the straightforward incorporation of additional unknowns and constraints , such as solving the problem for an imposed pressure drop @xmath116 rather than for an imposed flow rate @xmath98 . we illustrate the overall structure of the method in figure [ fig : flowchart ] . note that , for purposes of newton iteration , we calculate the jacobian analytically for the l and q models and numerically for the n models . , width=480 ] spectral collocation methods involve discretising the solution domain into a set of @xmath119 points ( collocation points ) , defining a global function ( the interpolant ) that interpolates the solution at these collocation points , and then approximating the derivatives of the solution as the derivatives of the interpolant . in a chebyshev spectral collocation method , the collocation points are the @xmath119 chebyshev points @xmath120 $ ] , which can be defined as @xcite @xmath121 the basis functions from which the interpolant is composed are then a set of @xmath119 polynomials of degree @xmath122 satisfying the criterion that each is nonzero at exactly one distinct collocation point . note that other definitions of the chebyshev points are also commonly used ( _ e.g. _ , * ? ? ? * ) . for the definition given in eq . ( [ eq : chebpoints ] ) , @xcite provide a suite of ` matlab ` functions that generate the chebyshev points and differentiation matrices , and that perform interpolation . we have now developed steady - state solutions for six different models , each for two distinct outer boundary conditions a fixed outer boundary ( `` constrained '' ) and an applied effective stress @xmath3 at the outer boundary ( see 2[summary ] ) . as described in 2[nondim ] , these models are characterised by five dimensionless parameters : @xmath103 , a ratio of elastic constants ; @xmath35 , the initial porosity ; @xmath65 , the ratio of the initial inner radius to the initial outer radius ; @xmath3 , the applied effective stress ; and either @xmath98 , the flow rate , or @xmath116 , the pressure drop . to focus on the impact of model choice , boundary conditions , and geometry , we adopt constant fixed values of @xmath123 and @xmath124 throughout the rest of the paper . varying these two parameters across a moderate range of typical values does not lead to dramatic qualitative differences in the resulting behaviour . similarly , we fix @xmath125 ( `` unconstrained '' ) for simplicity . in this section , we compare and contrast the six models for the two different boundary conditions ( unconstrained and constrained ) in the context of two end - member geometries : a thick - walled cylinder ( fig . [ fig : graph3 ] ) and a thin - walled cylinder ( fig . [ fig : graph4 ] ) . this gives us a preliminary sense for how the geometry impacts the mechanics , which is in turn the focus of 4[ql ] . in fig . [ fig : graph3 ] , we consider a thick - walled cylinder for flow driven by an imposed pressure drop of @xmath126 . for the unconstrained cylinder ( left column ) , the predictions of all models are qualitatively similar . the porosity @xmath33 ( top row ) , azimuthal effective stress @xmath127 ( fourth row ) , and pressure @xmath46 ( last row ) all have maxima at the inner boundary and decrease monotonically from left to right . the porosity remains everywhere greater than @xmath35 , the azimuthal effective stress is strictly tensile , and the pressure drops from @xmath128 to @xmath129 by construction . additionally , the pressure profile is strongly nonlinear for the @xmath107 models , but closer to classical linear poroelasticity ( l-@xmath106 ) for the @xmath106 models . in contrast to the behaviour of these quantities , the displacement @xmath64 ( second row ) and the radial effective stress @xmath130 ( third row ) are non - monotonic . the displacement has an interior maximum that is located in roughly the same place for all models . the radial effective stress vanishes at the inner and outer boundaries by construction . between these limits , it is purely tensile with an interior maximum , with the location of this maximum depending strongly on model choice . ) . we consider an unconstrained cylinder ( left column ) and a constrained cylinder ( right column ) , both for flow driven by an imposed pressure drop @xmath126 . for clarity , we plot the results against the lagrangian coordinate @xmath131 and on a logarithmic horizontal scale . the unconstrained and constrained cylinders exhibit strikingly similar behaviour , implying that the distinction between these two outer boundary conditions becomes unimportant when the walls are very thick ( _ i.e. _ , for small @xmath65 ) . additionally , note that in this case the permeability law has a noticeably greater impact on the results than the elasticity law or the treatment of the kinematics . [ fig : graph3 ] , width=480 ] for the same pressure drop , the constrained cylinder ( right column ) exhibits a strikingly similar behaviour to that of the unconstrained cylinder . the maximum in porosity at the inner boundary is lower than for the unconstrained cylinder , and the porosity now drops slightly below @xmath35 at the outer boundary where the material is slightly compressed . the displacement is qualitatively similar , but a factor of 23 smaller than in the unconstrained case . the radial and azimuthal effective stresses are now both slightly compressive at the outer boundary . this comparison between the unconstrained and constrained cylinders supports the intuition that the difference between these two cases becomes unimportant for thick walls ( _ i.e. , _ @xmath132 ) . in all of the cases shown in fig . [ fig : graph3 ] , the flow is driven by the same imposed pressure drop of @xmath126 . in addition to the above differences between the six models and the two boundary conditions , each of these twelve cases will result in a different flow rate model , for which @xmath98 is independent of the boundary condition . ] @xmath98 ( see legend , bottom of fig . [ fig : graph3 ] ) . in all cases , @xmath98 is lower for the constrained cylinder than for the unconstrained cylinder ( again , except for the l-@xmath106 model ) . this is because the inner radius of the constrained cylinder always expands less than that of the unconstrained cylinder , and @xmath98 is very sensitive to the inner radius ( eq . [ eq : q_to_p ] ) ; the constrained cylinder is also slightly compressed against the outer boundary , which reduces its permeability in the @xmath107 models , amplifying the reduction in @xmath98 . all of the @xmath106 models produce quantitatively similar values of @xmath98 . for each , @xmath98 differs by only a few percent between the two boundary conditions ; between the @xmath106 models for the same boundary condition , @xmath98 differs by about 1020% . by far the largest difference is between the corresponding @xmath106 and @xmath107 models , where the @xmath107 model produces a value of @xmath98 that is roughly 24 times larger than the corresponding @xmath106 model . the permeability law makes a great difference since large deformations of a thick - walled cylinder lead to large and nonuniform changes in porosity . this substantial change in porosity leads to a substantial change in permeability for the @xmath107 models , but has no impact on the @xmath106 models . this effect leads to higher values of @xmath98 for the @xmath107 models because the average porosity is in all cases larger than @xmath35 , so the permeability increases . comparing the n models to the q models , and the q models to the l models , reveals that both rigorous kinematics and nonlinear elasticity also lead to higher values of @xmath98 relative to their linearised counterparts . however , these effects are noticeably weaker than the impact of changing the permeability law . given that the values of @xmath98 vary so widely , it is surprising that the behaviour illustrated in fig . [ fig : graph3 ] is otherwise so similar across the models and boundary conditions . we now consider the other extreme geometry , a thin - walled cylinder , for a driving pressure drop of @xmath133 ( fig . [ fig : graph4 ] ) . note that this value of @xmath116 is more than one order of magnitude less than the value used for the thick - walled cylinder ( fig . [ fig : graph3 ] ) . despite this much smaller value of @xmath116 , @xmath127 here is comparable in magnitude to the thick - walled case while @xmath64 is much larger . we discuss these points in more detail in 4[ql ] . for the unconstrained cylinder ( left column ) , @xmath33 ( first row ) is almost uniform across the domain , with a weak and roughly linear decrease from left to right . this behaviour is mirrored in @xmath64 ( second row ) and @xmath127 ( fourth row ) . the pressure also decreases roughly linearly from left to right , from @xmath134 to @xmath129 , following classical linear poroelasticity for all models . unlike for the thick - walled case , the permeability law is relatively unimportant for these quantities , whereas the kinematics and the elasticity law play much more prominent roles . note that the kinematics consistently account for most of the difference between the l models and the n models ( _ i.e. _ , the q models are closer to the n models than they are to the l models ) . unlike these other quantities , @xmath130 does show a strong dependance on the permeability law . this suggests that the most direct impact of the permeability law is on @xmath130 , and this propagates to all other quantities when @xmath130 is mechanically important ( _ e.g. _ , fig . [ fig : graph3 ] ) . for the unconstrained thin - walled cylinder , @xmath130 vanishes at the boundaries and has an intermediate tensile maximum of order @xmath135 , whereas @xmath127 is uniformly of order @xmath136 . as a result , the stark differences in @xmath130 between the @xmath106 and @xmath107 models are ultimately unimportant . ) . we again consider an unconstrained cylinder ( left column ) and a constrained cylinder ( right column ) , now for flow driven by an imposed pressure drop @xmath133 . for clarity , we plot the results against the lagrangian coordinate @xmath131 on a linear horizontal scale . unlike for the thick - walled cylinder ( fig . [ fig : graph3 ] ) , the two different boundary conditions in this case result in strikingly different behaviour . for this geometry , it appears that the permeability law is relatively unimportant . for the unconstrained cylinder , the most important factors are the elasticity law and the treatment of the kinematics . for the constrained cylinder , the six models exhibit nearly identical behaviour . [ fig : graph4 ] , width=480 ] for the same pressure drop , the constrained cylinder exhibits strikingly different behaviour to the unconstrained cylinder . whereas the unconstrained cylinder expands almost uniformly by 2070% , the constrained cylinder is prevented from doing so . this results in much smaller displacements , with a maximum of order @xmath135 , making model choice essentially unimportant all models approach their asymptotic limit of classical linear poroelasticity ( l-@xmath106 ) . note also that most of the material is in compression , with the porosity decreasing roughly linearly from a value just above @xmath35 at the inner boundary to a value noticeably below @xmath35 at the outer boundary . the displacement is weakly nonlinear , decreasing monotonically from left to right . with regard to the flow rate @xmath98 , we first note that the values of @xmath98 in this case are substantially larger than the corresponding values for the thick - walled cylinder despite the fact that @xmath116 is much smaller . to rationalise this , note that the relationship between @xmath98 and @xmath65 for a given @xmath116 is strongly nonlinear even for a rigid cylinder ( _ i.e. _ , eq . ( [ q ] ) with @xmath137 and @xmath5 ) . the same is also true for classical linear poroelasticity , where the same expression also applies . in other words , this difference in @xmath98 is due in large part to the fact that @xmath65 is much larger . for the constrained thin - walled cylinder , @xmath98 is considerably smaller than for the unconstrained thin - walled cylinder ( except for the l-@xmath106 case , where @xmath98 is independent of the boundary conditions ) . for the @xmath106 cases , this is because the cylinder expands substantially and almost uniformly , which decreases the ratio of @xmath1 to @xmath0 and increases the flow rate ( see eq . ( [ q ] ) ) . this is true to a much lesser extent for the constrained cylinder since the displacements are much smaller . for the @xmath107 models , this increase in @xmath98 is substantially enhanced for the unconstrained cylinder by the noticeable increase in porosity and therefore permeability . the reverse occurs for the constrained cylinder , where the porosity decreases , leading a lower @xmath98 for the @xmath107 models than for the @xmath106 models . as for the thick - walled cylinder , both rigorous kinematics and nonlinear elasticity also lead to higher values of @xmath98 relative to their linearised counterparts . for the unconstrained cylinder , these effects are substantial ; for the constrained cylinder , these effects are noticeably weaker than the impact of changing the permeability law . there is relatively little difference in the @xmath98 values across the six different models for the constrained cylinder , again because the displacements are necessarily small . in this section , we have considered the implications of model choice in the context of two end - member geometries ( thick - walled and thin - walled ) . we have shown that the error associated with linearisation depends strongly on factors such as geometry and boundary conditions . in the next section , we study the mechanics of the problem over the full transition from @xmath132 to @xmath138 . we now investigate in more detail the impact of the geometry . although the n-@xmath107 model is arguably the most ` correct ' of those presented above , it is computationally more expensive and less robust than the other models . for simplicity , we restrict ourselves to the q-@xmath107 model , which offers a good compromise between accuracy and computational efficiency . we showed in the previous section that this model demonstrates the same qualitative behaviour as the n-@xmath107 model for both end - member geometries and for both boundary conditions , and it is also quantitatively similar . in fig . [ fig : graph5 ] we consider the evolution of several key quantities as the initial inner radius @xmath65 varies continuously from @xmath139 ( thick walls ) to @xmath140 ( thin walls ) , and for both the unconstrained ( left column ) and constrained ( right column ) cylinders . we plot the results for several different values of @xmath116 . model at steady state for a continuously varying geometry and for six distinct values of @xmath116 ( @xmath141 $ ] , blue to yellow ) for the unconfined ( left ) and confined ( right ) cylinders . we show the changes in the inner and outer radii , @xmath142 ( solid lines ) and @xmath143 ( dashed lines ) , respectively ( first row ) ; change in wall thickness @xmath144 ( second row ) ; minimum and maximum porosity , @xmath145 ( dashed lines ) and @xmath146 ( solid lines ) , respectively ( third row ) ; maximum absolute azimuthal and radial effective stresses , @xmath147 ( solid lines ) and @xmath148 ( dashed lines ) , respectively ( fourth row ) ; and the normalised flow rate @xmath149 ( last row ) . [ fig : graph5 ] , width=480 ] for the unconstrained cylinder ( left column ) the most striking feature is that all quantities except for @xmath148 diverge at a certain critical value of @xmath65 for each @xmath116 , and this critical value decreases as @xmath116 increases . this divergence indicates that there exists a maximum allowable value of @xmath65 for a given @xmath116 ( or a maximum allowable value of @xmath116 for a given @xmath65 ) above which the material can no longer physically support the load at steady state ; that is , no steady deformation field exists that is compatible with the driving pressure . this maximum allowable @xmath116 is finite and positive for @xmath150 since it must diverge as @xmath65 tends to zero and vanish as @xmath65 tends to one . corresponds to a line source in an infinite domain , for which no steady state exists , whereas @xmath151 corresponds to vanishingly thin walls , which can support no load . ] approaching the maximum allowable @xmath65 for a fixed @xmath116 , all quantities except for @xmath148 increase monotonically in magnitude . approaching the maximum allowable @xmath116 for a fixed @xmath65 , every quantity increases monotonically in magnitude . for all values of @xmath65 and @xmath116 , flow drives the material outward ; thus , both radii increase ( _ i.e. _ , @xmath152 and @xmath153 , first row ) . for most values of @xmath65 , @xmath142 is greater than @xmath143 and therefore the wall thickness decreases ( _ i.e. _ , @xmath154 , second row ) . however , this trend reverses for @xmath155 , where @xmath143 exceeds @xmath142 and the wall thickness increases ( _ i.e. _ , @xmath156 ) . for @xmath157 , the increase in mean radius overwhelms the decrease in wall thickness ( _ i.e. _ , the cross - sectional area increases ) , and thus the porosity of the entire material increases : the minimum porosity @xmath145 and maximum porosity @xmath146 ( third row ; solid and dashed lines , respectively ) both exceed @xmath35 for all values of @xmath65 and @xmath116 . as @xmath116 increases , the difference between these extreme values also increases . the maximum absolute azimuthal effective stress @xmath147 and the maximum absolute radial effective stress @xmath148 ( fourth row ; solid and dashed lines , respectively ) are of interest in the context of material failure . unlike the other quantities , @xmath148 decreases monotonically in magnitude with @xmath65 at a fixed @xmath116 , approaching a finite , nonzero value as other quantities diverge . we consider the flow rate @xmath98 in comparison to the flow rate @xmath158 that would occur for a rigid cylinder with the same initial geometry . plotting @xmath149 ( last row ) then scales out the impact of the initial geometry , allowing us to focus on the impact of deformation . note that @xmath149 is strictly greater than one , meaning that deformation always increases the flow rate for an unconfined cylinder . the increase in @xmath98 over @xmath159 is also amplified by the increase in permeability that results from the increase in porosity , leading to a greater flow rate @xmath98 for the same fixed pressure drop @xmath116 . the constrained cylinder ( right column ) exhibits qualitatively different behaviour . most quantities still increase monotonically with @xmath116 for a fixed @xmath65 , but no quantities diverge . note that we expect the unconstrained and constrained cylinders to approach the same limiting behaviour for @xmath139 , as noted above in the context of fig . [ fig : graph3 ] . the change in inner radius @xmath142 is strictly positive , tending to zero for both small @xmath65 and large @xmath65 . in the latter limit , this is because the outer radius is fixed and there is simply nowhere for the material to go ( _ i.e. _ , @xmath160 as @xmath161 ) . in the former limit , this is because the absolute change in @xmath65 for a given @xmath116 will decrease with @xmath65 even if the relative change in @xmath65 is non - negligible . since the outer radius is fixed , the change in wall thickness is equal and opposite to the change in inner radius ( _ i.e. _ , @xmath162 ) , and is therefore strictly negative . the minimum porosity @xmath145 is always less than @xmath35 and decreases monotonically with @xmath65 and with @xmath116 . the maximum porosity @xmath146 mirrors this behaviour ; it is always greater than @xmath35 , and increases monotonically with @xmath65 and with @xmath116 . in contrast with the unconstrained cylinder , the maximum absolute azimuthal effective stress here decreases monotonically with @xmath65 , tending to zero as @xmath65 tends to one , while the maximum absolute radial effective stress increases as @xmath65 tends to one . both stresses increase monotonically with @xmath116 . the normalised flow rate @xmath149 decreases monotonically with @xmath65 , decreasing from a maximum greater than one at the inner boundary to a minimum less than one at the outer boundary . in other words , deformation increases the flow rate for constrained cylinders with thick walls , but decreases the flow rate for constrained cylinders with thin walls . however , this effect is relatively modest , with @xmath98 decreasing from a few tens of percent above @xmath159 to a few tens of percent below @xmath159 over the full range of @xmath65 . for the unconstrained cylinder , in contrast , deformation dominates the flow rate near the point of divergence . in all of these cases , the flow forces the material radially outward ; the net force due to fluid or pore - pressure loading @xmath163 must be supported through a combination of forces due to internal azimuthal stress @xmath164 and due to external radial traction @xmath165 . we obtain expressions for these forces by performing a macroscopic force balance on the cylinder ( see diagrams , top of fig . [ fig : graph6 ] ) , giving @xmath166 where force balance then requires that @xmath167 . we consider these quantities in fig . [ fig : graph6 ] for both the unconstrained cylinder ( left column ) and the constrained cylinder ( right column ) . we drive the flow by imposing either a fixed pressure drop @xmath116 ( top row ) or a fixed flow rate @xmath98 ( bottom row ) for the full range of @xmath65 , calculating the other quantity ( @xmath98 or @xmath116 , respectively ) as part of the solution . and @xmath116 and calculate @xmath65 ; we do not consider this case here . ] ( solid lines ) that must be supported by a combination of force due to internal azimuthal stress @xmath164 ( dashed lines ) and force due to external radial traction @xmath165 ( dot - dashed lines ) . we plot these force components here for fixed @xmath168 $ ] ( top row , light to dark ) and for fixed @xmath169 $ ] ( bottom row , light to dark ) for the unconfined and confined cylinders ( left and right , respectively ) . for the unconfined cylinder ( left ) , note that @xmath170 and therefore @xmath171 . [ fig : graph6 ] , width=480 ] as should be expected , all three force components increase as @xmath116 ( or @xmath98 ) increases . for the unconstrained cylinder , the radial force @xmath165 is identically equal to zero by construction ( @xmath170 ) ; thus , the azimuthal force @xmath164 must balance the entire pressure force @xmath163 ( @xmath171 ) . for fixed @xmath116 ( top left ) , @xmath164 and @xmath163 increase monotonically with @xmath65 until they diverge . the pressure component increases and ultimately diverges because the first term is proportional to @xmath0 ; the second term in @xmath163 vanishes as the wall thickness vanishes since @xmath172 . the azimuthal component must then diverge to balance @xmath163 , which requires that @xmath173 diverge strongly as the wall thickness vanishes . for fixed @xmath98 ( bottom left ) , @xmath163 and @xmath164 decrease approximately linearly with @xmath65 , vanishing as @xmath65 tends to one ; this reflects the fact that the @xmath116 associated with a given @xmath98 decreases monotonically with @xmath65 . for the constrained cylinder , it is no longer the case that @xmath170 ; instead , @xmath165 will be determined implicitly to satisfy the condition that @xmath174 . for fixed @xmath116 , both @xmath163 and @xmath165 increase monotonically with @xmath65 . for @xmath175 , @xmath164 is similar in magnitude to @xmath165 and increases with @xmath65 ; however , for @xmath157 , @xmath164 decreases rapidly with @xmath65 and becomes almost negligible in the overall force balance . in other words , the outer boundary supports most of the loading for a cylinder with moderate to thick walls . note that @xmath176 for @xmath177 since @xmath178 , but @xmath179 for @xmath180 since @xmath181 . for fixed @xmath98 , the relative balance of @xmath163 , @xmath165 , and @xmath164 is similar to the fixed @xmath116 case except that all vanish as @xmath65 tends to one . , pressure drop @xmath116 , and flow rate @xmath98 define a three - parameter space in which fixing any two determines the third . for the unconstrained cylinder , we explore this by plotting @xmath98 _ vs. _ @xmath116 for several values of @xmath182 $ ] ( left ) . for the constrained cylinder , we instead plot @xmath116 _ vs. _ @xmath65 for several values of @xmath183 $ ] ( right ) . we include the predictions of classical linear poroelasticity on both plots ( eq . [ q ] with @xmath137 and @xmath184 ; light grey ) . [ fig : graph7 ] , width=480 ] finally , we investigate the three - parameter space of @xmath65 , @xmath116 , and @xmath98 fixing any two of these determines the third , although not necessarily uniquely . we explore this space in fig . [ fig : graph7 ] , comparing the results from the q-@xmath107 model ( colours ) with the predictions of classical linear poroelasticity ( grey ) . the former exhibits rich behaviour that is absent from the latter , and that can therefore be attributed entirely to nonlinearity . for the unconstrained cylinder ( left ) , we plot @xmath98 against @xmath116 for several values of @xmath65 . the general trend is a steep increase toward divergence at the maximum allowable @xmath116 for each value of @xmath65 ( _ cf . _ , fig . [ fig : graph5 ] ) . however , for @xmath132 ( _ i.e. _ , cylinders with very thick walls ) , two distinct values of @xmath98 can lead to the same @xmath116 . these correspond to a less - deformed state with a smaller @xmath98 and a more - deformed state with a larger @xmath98 . in other words , the steady - state behaviour becomes non - unique as @xmath65 becomes sufficiently small . for the constrained cylinder ( right ) , we plot @xmath116 _ vs. _ @xmath65 for several values of @xmath98 . for moderate @xmath98 , @xmath116 diverges as @xmath65 tends to zero and decreases monotonically toward zero as @xmath65 tends to one . however , @xmath116 becomes non - monotonic in @xmath65 for sufficiently large @xmath98 , such that two distinct values of @xmath65 can lead to the same value of @xmath116 for a given value of @xmath98 . these correspond to a less - deformed state for a smaller @xmath65 and a more - deformed state for a larger @xmath65 ( _ cf . _ , lower right panel of fig . [ fig : graph6 ] ) . note that , for the same value of @xmath65 , a given value of @xmath98 always leads to a much higher value of @xmath116 in the constrained case . despite being a classical topic in geomechanics and in biophysics , radial poroelastic deformation has not previously been systematically explored , particularly in the context of large deformations . to assess the qualitative and quantitative impacts of large deformations , we considered six different models in the context of two end - member geometries ( thick - walled and thin - walled ) and two different outer boundary conditions ( unconstrained and constrained ) . we showed that the impacts of nonlinear kinematics , nonlinear elasticity , and deformation - dependent permeability depend very strongly on geometry and boundary conditions , as does the relative importance of these facets of nonlinearity . for example , the mechanical response of an unconstrained thin - walled cylinder to an imposed pressure drop is dominated by nonlinear kinematics and elasticity , although the permeability law exerts a strong control on the resulting flow rate through the material . for the same pressure drop , a constrained thin - walled cylinder is limited to much smaller deformations and exhibits what is essentially a linear - poroelastic response ( fig . [ fig : graph4 ] ) . in contrast , the mechanical response of a thick - walled cylinder is much less sensitive to constraint , although the outer boundary condition has a strong impact on the flow rate when the permeability is deformation - dependent ( fig . [ fig : graph3 ] ) . to explore the importance of geometry and constraint in more detail , we then focused on a model that includes rigorous nonlinear kinematics and deformation - dependent permeability , but with the simplification of linear elasticity ( q-@xmath107 ) . this model captures the qualitative and quantitative impacts of large deformations , but is more computationally convenient than a fully nonlinear model . we showed that , for an unconstrained cylinder , a given initial inner radius can conduct an arbitrarily large flow rate but can only support a finite maximum pressure drop , and this maximum allowable pressure drop increases with the thickness of the walls ( figs . [ fig : graph5][fig : graph7 ] ) . for a constrained cylinder , in contrast , the pressure drop increases monotonically with the flow rate but is strongly non - monotonic with the initial inner radius for flow rates above a certain threshold . these behaviours are mirrored in the corresponding force balances ( figs . [ fig : graph6 ] ) . we have assumed here that the constitutive response of the solid skeleton remains elastic for arbitrarily large deformations . this is directly relevant to biomedical applications such as fluid permeation through artery walls , and to the design of radial filters . in geomechanical applications , however , large deformations are typically the result of material failure through plasticity or fracture , which will lead to a fundamentally different constitutive behaviour in the solid . these behaviours will be the subject of future work . the authors are grateful to epsrc for support in the form of a doctoral training award to lca . the authors also thank simon mathias for helpful discussions and advice related to the chebyshev pseudospectral method . 28ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop * * ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) in @noop _ _ ( ) pp . @noop * * , ( ) @noop _ _ ( , ) @noop * * , ( ) , to enforce a constant pressure drop @xmath116 , we must derive an associated expression for the evolving flow rate @xmath185 . to do so , we rearrange and integrate the expression for @xmath186 from eq . to obtain @xmath187 where @xmath188 for linearised kinematics , we replace the latter expression with @xmath189 the above expressions also apply at steady state , for which @xmath190 . assuming linear elasticity , we solve eq . ( [ a ] ) for constant permeability ( @xmath191 \equiv 1 $ ] ) to arrive at a general expression for the displacement , [ genund ] @xmath192 where @xmath193 and @xmath194 are determined by the boundary conditions . this result is solely mechanical and constitutive , and is therefore valid for both the l-@xmath106 and q-@xmath106 models . the general expressions for the effective stresses are then @xmath195 from these expressions , we arrive at four distinct solutions by combining the two different treatments of the kinematics ( rigorous q and linearised l ) with the two different sets of outer boundary conditions ( an applied stress at the outer boundary ( eq . [ outerbca ] ) and a fixed outer boundary ( eq . [ outerbcb ] ) ) . the two l-@xmath106 solutions are classical solutions from linear poroelasticity @xcite . an approximate version of the q-@xmath106 solutions was derived by @xcite and @xcite , who applied boundary conditions at the moving boundary but linearised the relationship between @xmath33 and @xmath64 . for an applied effective stress at the outer boundary , we derive expressions for @xmath193 and @xmath194 by applying the appropriate inner and outer boundary conditions ( eqs . [ innerbc ] and [ outerbca ] , respectively ) . we linearise the kinematics by applying these at @xmath196 ( rather than at @xmath0 ) and at @xmath197 ( rather than at @xmath1 ) , respectively . we obtain @xmath198}{2(1-a_0 ^ 2)}.\ ] ] similarly , for a fixed outer boundary , we apply the appropriate inner and outer boundary conditions ( eqs . [ innerbc ] and [ outerbcb ] , respectively ) at @xmath196 and at @xmath197 , respectively , to obtain @xmath199}{2[a_0 ^ 2(1+\gamma)+(1-\gamma)]}\right\}+\frac{(1+\gamma)}{2}q\ln a_0\ ] ] and @xmath200}{2[a_0 ^ 2(1+\gamma)+(1-\gamma)]}\right\}.\ ] ] all other quantities can be derived from the expressions for @xmath64 . thus , we have complete explicit solutions following classical linear poroelasticity for the two different sets of outer boundary conditions . note that , for linearised kinematics , @xmath33 should be calculated from @xmath64 according to eq . ( [ eq : phi_to_ulin ] ) . for an applied effective stress at the outer boundary , we now apply eqs . ( [ innerbc ] ) and ( [ outerbcb ] ) at @xmath201 and @xmath202 , respectively , to the general elastic solution ( eq . [ genund ] ) . this leads to @xmath203}{2(b^2-a^2)}+\frac{1+\gamma}{2}q\ln(a ) , \quad % b_2 = \frac{a^2b^2[2\sigma_r^\star+(1+\gamma ) q\ln(b / a)]}{2(b^2-a^2)}.\ ] ] this solution is not explicit because the inner radius @xmath0 and outer radius @xmath1 are now determined by the two kinematic conditions ( see eqs . [ innerbc ] and [ outerbca ] ) , leading to two coupled , implicit expressions for @xmath0 and @xmath1 . we solve these expressions numerically using a root - finding technique . for a fixed outer boundary , we now obtain [ b ] @xmath204}{2[a^2(1+\gamma)+(1-\gamma)]}\right\ } + \frac{(1+\gamma)}{2}q\ln a\ ] ] and @xmath205}{2[a^2(1+\gamma)+(1-\gamma)]}\right\}.\ ] ] the problem is closed by applying the kinematic condition at the inner boundary ( see eq . [ innerbc ] ) , leading to an implicit expression for @xmath0 . we again solve this numerically using a root - finding technique . as above , all other quantities can then be derived from the expressions for @xmath64 . note that , for rigorous kinematics , @xmath33 should be calculated from @xmath64 according to eq . ( [ eq : phi_to_u ] ) . we now derive an approximate solution to the q-@xmath106 model in the limit of vanishing wall thickness , starting from eq . with @xmath191\equiv1 $ ] . we do this for the case of an applied effective stress at the outer boundary since the case of no displacement at the outer boundary is ultimately limited to small displacements and thus is well - captured by linear poroelasticity . we begin by defining a new radial coordinate @xmath206 such that @xmath207 $ ] where @xmath208 is the wall thickness . we then rewrite eq . in terms of @xmath209 and seek a solution under the assumption that @xmath210 . from these assumptions , and writing @xmath211 , we obtain at leading order @xmath212 note that these assumptions require for asymptotic consistency that @xmath213 . equation is a linear , second - order ode with solution @xmath214+a_2\exp\left[\left(\frac{\sqrt{5 } -1}{2a}\right)\epsilon\right ] + aq.\ ] ] we now apply the relevant boundary conditions ( eqs . and with @xmath215 ) , which results in four equations for four unknowns : the two integration constant , @xmath216 and @xmath217 , and the inner and outer radii , @xmath0 and @xmath218 , respectively . we use the two conditions at the inner boundary to derive expressions for @xmath216 and @xmath217 in terms of @xmath0 , @xmath219-\gamma(a - a_0)}{\sqrt{5}}.\ ] ] the two conditions at the outer boundary then give the pressure field is given by @xmath223 this then leads to @xmath224 and , since @xmath213 , we have that @xmath225 . this implies that a small pressure drop will drive a large flow rate when the walls are sufficiently thin . model at steady state an unconfined cylinder ( @xmath226 ) in the thin - walled limit for @xmath227 $ ] ( blue to yellow ) . we plot the same quantities as in fig . [ fig : graph5 ] , and the qualitative behaviour agrees with the results shown in fig . [ fig : graph5 ] for @xmath138 . note that that the dashed lines are not visible here because @xmath228 ( first row ) , @xmath229 ( third row ) , and @xmath230 ( fourth row ) . note also that we restrict the range of @xmath65 for each value of @xmath116 to ensure that @xmath225 . [ fig : graph8 ] , width=480 ]

for any graph and positive integer , the _ exact distance- graph _ }}$ ] is the graph with vertex set , which has an edge between vertices and if and only if and have distance in . for odd , neetil and ossona de mendez proved that for any fixed graph class with _ bounded expansion _ , the chromatic number of }}$ ] is bounded by an absolute constant . using the notion of _ generalised colouring numbers _ , we give a much simpler proof for the result of neetil and ossona de mendez , which at the same time gives significantly better bounds . in particular , we show that for any graph and odd positive integer , the chromatic number of }}$ ] is bounded by the weak-colouring number of . for even , we prove that}})$ ] is at most the weak-colouring number times the maximum degree . for odd , the existing lower bound on the number of colours needed to colour}}$ ] when is planar is improved . similar lower bounds are given for-minor free graphs . key words : _ bounded expansion , chromatic number , exact distance graphs , generalised colouring numbers , planar graphs _

1612.02160

all graphs in this paper are assumed to be finite , undirected , simple and without loops . for a graph @xmath11 ( or just @xmath12 if the graph under consideration is clear ) and vertices @xmath13 , let @xmath14 denote the distance between @xmath4 and @xmath5 in @xmath6 , that is , the number of edges contained in a shortest path between @xmath4 and @xmath5 . for a positive integer @xmath1 , the _ @xmath1-th power graph @xmath15 of @xmath6 _ is the graph with @xmath3 as its vertex set and @xmath16 contains the edge @xmath17 if and only if @xmath18 . problems related to the chromatic number @xmath19 of power graphs @xmath20 were first considered by kramer and kramer @xcite in 1969 and have enjoyed significant attention ever since . it is clear that for @xmath21 any power of a star is a clique , and hence there are not many classes of graphs for which @xmath19 can be bounded by a constant . an easy argument shows that for a graph @xmath6 with maximum degree @xmath22 we have @xmath23 however , there are many classes of graphs for which it is possible to find much better upper bounds . recall that a graph @xmath6 is _ @xmath24-degenerate _ if every subgraph of @xmath6 contains a vertex of degree at most @xmath24 . [ largepow ] + let @xmath24 and @xmath1 be positive integers . there exists a constant @xmath25 such that for every @xmath24-degenerate graph @xmath6 we have @xmath26 . in this result , the exponent on @xmath27 is best possible ( see below ) . in particular , @xmath28 is at most linear in @xmath27 for planar graphs @xmath6 . wegner @xcite conjectured that every planar graph @xmath6 with @xmath29 satisfies @xmath30 , and gave examples that show this bound would be tight . the conjecture has attracted considerable attention since it was stated in 1977 . for more on this conjecture we refer the reader to @xcite . in ( * ? ? ? * section 11.9 ) , neetil and ossona de mendez define the notion of _ exact power graph_. let @xmath0 be a graph and @xmath1 a positive integer . the _ exact @xmath1-power graph @xmath31 _ has @xmath3 as its vertex set , and @xmath17 is an edge in @xmath31 if and only if there is in @xmath6 a path of length @xmath1 ( i.e. with @xmath1 edges ) between the vertices @xmath4 and @xmath5 ( the path need not be induced , nor a shortest path ) . similarly , they define the _ exact distance-@xmath1 graph @xmath2}}$ ] _ as the graph with @xmath3 as its vertex set , and @xmath17 as an edge if and only if @xmath32 . since obviously @xmath33}})\subseteq e({g^{\natural p}})\subseteq e(g^p)$ ] , we have @xmath8}})\le\chi({g^{\natural p}})\le\chi(g^p)$ ] . for planar graphs @xmath6 , theorem [ largepow ] gives that the exact @xmath1-power graphs @xmath31 satisfy @xmath34 . this result is best possible , even for outerplanar graphs , as the following examples show . for @xmath35 and @xmath36 , let @xmath37 be the @xmath24-regular tree of radius @xmath38 with root @xmath39 . we say that a vertex @xmath5 is at level @xmath40 if @xmath41 . for every edge @xmath42 between vertices at levels @xmath40 and @xmath43 for some @xmath44 , we do the following : if @xmath1 is even , then add a path of length @xmath43 between @xmath45 and @xmath39 ; if @xmath1 is odd , then add paths of length @xmath43 and @xmath46 between @xmath45 and @xmath39 . call the resulting graph @xmath47 . it is straightforward to check that @xmath48 for even @xmath1 , that @xmath49 for odd @xmath1 , and that there is a path of length @xmath1 between any two vertices at level @xmath38 . since there are @xmath50 vertices at level @xmath38 , this immediately means that @xmath51 . surprisingly , for exact distance graphs , the situation is quite different . [ thm2 ] 1.5to1.5let @xmath1 be an odd positive integer . then there exists a constant @xmath52 such that for every planar graph @xmath6 we have @xmath8}})\le c$ ] . 1.5to1.5let @xmath1 be an even positive integer . then there exists a constant @xmath53 such that for every planar graph @xmath6 we have @xmath8}})\le c'\cdot\delta(g)$ ] . the results in theorem [ thm2 ] are actually special cases of the following more general results . we will recall the concept of a _ graph class with bounded expansion _ in the next subsection . [ thm3 ] + let @xmath54 be a class of graphs with bounded expansion . 1.5to1.5let @xmath1 be an odd positive integer . then there exists a constant @xmath55 such that for every graph @xmath56 we have @xmath8}})\le c$ ] . 1.5to1.5let @xmath1 be an even positive integer . then there exists a constant @xmath57 such that for every graph @xmath56 we have @xmath8}})\le c'\cdot\delta(g)$ ] . we will give two proofs of part(a ) . the two proofs give incomparable bounds . also , both proofs are considerably shorter and provide better bounds than the original proof of part(a ) of neetil and ossona de mendez ( * ? ? ? * theorem 11.8 ) . theorem [ thm3](b ) is new , as far as we are aware . as we showed above , if we consider exact powers instead of exact distance graphs , then we need to use bounds involving @xmath27 if we want to bound @xmath58 , even for odd @xmath1 and if @xmath6 is planar . however , by adding the condition that @xmath6 has sufficiently large _ odd girth _ ( length of a shortest odd cycle ) , @xmath58 can be bounded without reference to @xmath27 , for odd @xmath1 . it follows from theorem [ thm3](a ) that this is possible if the odd girth is at least @xmath59 . this is because odd girth at least @xmath59 guarantees that if there is a path of length @xmath1 between @xmath45 and @xmath39 , then any shortest @xmath42-path has odd length . with some more care we can reprove the following . [ thm4 ] + let @xmath54 be a class of graphs with bounded expansion and let @xmath1 be an odd positive integer . then there exists a constant @xmath60 such that for every graph @xmath56 with odd girth at least @xmath61 we have @xmath62 . theorem [ thm2](a ) and its general version theorem [ thm3](a ) are quite surprising , since the exact distance graphs @xmath2}}$ ] of a planar graph @xmath6 can be very dense . to see this , for @xmath63 let @xmath64 be obtained from the complete graph @xmath65 by subdividing each edge @xmath66 times ( i.e.by replacing each edge by a path of length @xmath67 ) . for @xmath68 , form @xmath69 by adding four sets of @xmath24 new vertices to @xmath64 and joining all @xmath24 vertices in the same set to one of the vertices of degree three in @xmath64 . see figure [ four ] for a sketch of @xmath70 . such that @xmath71}}$ ] has edge density approximately 3/4.,width=200 ] it is easy to check that @xmath69 is a planar graph with @xmath72 vertices , while @xmath73}}$ ] has @xmath74 edges . so for fixed @xmath67 and large @xmath24 , the graph @xmath73}}$ ] has approximately @xmath75 times the number of edges of the complete graph on the same number of vertices . it is interesting to see what actual upper and lower bounds we can get for the chromatic numbers of @xmath2}}$ ] for @xmath6 from some specific classes of graphs and for specific values of ( odd ) @xmath1 . using the proof in @xcite , it follows that for @xmath76 and for planar graphs @xmath6 we can get the upper bound @xmath77}})\le5\cdot2^{20,971,522}$ ] ( see also subsection [ ssec3.0 ] ) . on the other hand , ( * ? ? ? * exercise 11.4 ) gives an example of a planar graph @xmath6 with @xmath77}})=6 $ ] . our new proof of theorem [ thm3](a ) already gives a much smaller upper bound for @xmath77}})$ ] for planar graph @xmath6 . by a more careful analysis , we can reduce that upper bound even further , giving the bound in the following result . we also managed to increase the lower bound , although by one only . details can be found in section [ sec4 ] . [ thm5 ] 1.5to1.5for every planar graph @xmath6 we have @xmath77}})\le143 $ ] . 1.5to1.5there exists a planar graph @xmath78 such that @xmath79}})=7 $ ] . for outerplanar graphs @xmath6 we have that @xmath77}})\le13 $ ] , while there exists an outerplanar graph @xmath80 such that @xmath81}})=5 $ ] ( see the results in sections [ sec3 ] and [ sec4 ] ) . when solving an optimisation problem it is often useful to preorder the input so as to minimise some parameter . one such parameter is the _ colouring number _ @xmath82 of a graph @xmath6 . this is the minimum integer @xmath24 such that there is a linear ordering @xmath83 of @xmath3 such that every vertex @xmath5 has at most @xmath84 neighbours @xmath4 with @xmath85 . ( so the colouring number is one more than the _ degeneracy _ of a graph . ) it is well - known and easy to see that the chromatic number @xmath86 of a graph @xmath6 satisfies @xmath87 . although this bound is far from being tight in many cases , it is often used to show that a specific class of graphs has bounded chromatic number . different generalisations of the colouring number can be found in the literature . chen and schelp @xcite proved that the class of planar graphs has linear ramsey number by also controlling , for all vertices @xmath39 , the number of smaller vertices that can be reached by a path of length two , whose middle vertex is larger than @xmath39 . various versions of their idea were applied by kierstead and trotter @xcite , kierstead @xcite , and zhu @xcite to problems concerning the game chromatic number of graphs and gave rise to the 2-colouring number defined below . in their study of oriented game chromatic number of graphs , kierstead and trotter @xcite considered paths of length four with different configurations of `` large '' internal vertices , which later motivated the notions of 4-colouring number and weak 4-colouring number . kierstead and kostochka @xcite applied 2-colouring numbers to a ( non - game ) packing problem . all of these notions are encompassed in the concepts of the _ @xmath24-colouring number _ and the _ weak @xmath24-colouring number _ of a graph , both of which were first introduced by kierstead and yang @xcite . let @xmath0 be a graph , @xmath83 a linear ordering of @xmath3 , and @xmath24 a positive integer . we say that a vertex @xmath88 is _ @xmath24-accessible _ from @xmath89 if @xmath85 and there exists an @xmath90-path @xmath91 of length at most @xmath24 such that @xmath92 for all internal vertices @xmath93 of @xmath91 . similarly , if all internal vertices @xmath93 of @xmath91 satisfy the less restrictive condition that @xmath94 , then we say that @xmath4 is _ weakly _ from @xmath5 . let @xmath95 be the set of vertices that are @xmath24-accessible from @xmath5 , and @xmath96 the set of vertices that are weakly @xmath24-accessible from @xmath5 . @xmath24-colouring number @xmath97 _ and _ weak @xmath24-colouring number @xmath98 _ of a graph @xmath6 are defined as follows : @xmath99 if we allow paths of any length ( but still have restrictions on the position of the internal vertices ) , we get @xmath100 , @xmath101 , the _ @xmath102-colouring number @xmath103 _ and the _ weak @xmath102-colouring number @xmath104_. we now state the main results of this paper . [ thm10 ] 1.5to1.5for every odd positive integer @xmath1 and every graph @xmath6 we have @xmath8}})\le{\mathrm{wcol}}_{2p-1}(g)$ ] . 1.5to1.5for every even positive integer @xmath1 and every graph @xmath6 we have @xmath8}})\le{\mathrm{wcol}}_{2p}(g)\cdot\delta(g)$ ] . [ thm11 ] + let @xmath1 be an odd positive integer and @xmath6 a graph . set @xmath105 . 1.5to1.5we have @xmath8}})\le \bigl(\bigl\lfloor\frac12p\bigr\rfloor+2\bigr)^q$ ] . 1.5to1.5if @xmath6 has odd girth at least @xmath61 , then @xmath106 . an interesting aspect of generalised colouring numbers is that these invariants can also be seen as gradations between the colouring number @xmath82 and two important minor monotone invariants , namely the _ tree - width @xmath107 _ and the _ tree - depth @xmath108 _ ( which is the minimum height of a depth - first search tree for a supergraph of @xmath6 , see @xcite ) . more explicitly , for every graph @xmath6 we have the following relations . [ pro : coltwtd ] 1.5to1.5@xmath109 ; 1.5to1.5@xmath110 . the equality @xmath111 was first proved in @xcite . the equality @xmath112 is ( * ? ? ? * lemma 6.5 ) . relations between the two sets of numbers exist as well . clearly , @xmath113 and @xmath114 . for the converse , kierstead and yang @xcite proved that @xmath115 . note that this means that if one of the generalised colouring numbers is bounded for a class of graphs ( for some @xmath24 ) , then so is the other one . shortly after neetil and ossona de mendez @xcite introduced the notion of _ classes with bounded expansion _ , zhu provided , in @xcite , a way of characterising these classes in terms of the weak @xmath24-colouring numbers . we will use this characterisation as a definition . [ def7 ] + a class of graphs @xmath54 has _ bounded expansion _ if and only if there exist constants @xmath116 , @xmath117 such that @xmath118 for all @xmath24 and all @xmath56 . by this definition , theorem [ thm3](a ) follows directly from both theorems [ thm10](a ) and [ thm11](a ) . we give the proofs of theorems [ thm10 ] and [ thm11 ] in the next section . the proof of theorem [ thm11 ] actually proves a stronger result . for two graphs @xmath0 and @xmath119 on the same vertex set , define @xmath120 . then the upper bound in both parts of theorem [ thm11 ] holds for @xmath121}}\cup{g^{[\natural 3]}}\cup\cdots\cup{g^{[\natural p]}})$ ] and @xmath122 , respectively . a natural question is if for even @xmath1 we can generalise the bound in theorem [ thm10](b ) by a similar bound @xmath123 , where @xmath124 depends on the generalised colouring numbers . but this is not possible . let @xmath125 be the @xmath126-regular tree of radius 2 . then we have @xmath127 and @xmath128 for all @xmath35 . it is easy to check that @xmath129 , but @xmath130 . these examples generalise to larger distances . the results in theorem [ thm11 ] are best possible in the sense that they give upper bounds of @xmath8}})$ ] and @xmath58 that depend on @xmath131 only , whereas no such results are possible that depend on @xmath98 with @xmath132 . to see this , for @xmath133 let @xmath134 be the @xmath135-subdivision of the complete graph @xmath136 ( that is , the graph formed by replacing the edges of @xmath136 by paths of length @xmath1 ) . then we obviously have @xmath137}})=n$ ] . on the other hand we have @xmath138 . to verify this , order the vertices of @xmath134 as follows . first order the branch vertices ( the vertices in the original clique ) , and then order the subdivision vertices in any way . clearly , each branch vertex will not weakly @xmath135-access any other vertex . an internal vertex of a subdivided edge can only weakly @xmath135-access the other @xmath1 vertices on the path that replaced the edge ( including the two end - vertices of the path ) . so for fixed odd @xmath139 we can not bound @xmath137}})$ ] by an expression that involves @xmath140 only . the bound on the odd girth in theorem [ thm11](b ) is also best possible . to show this , for @xmath141 let @xmath142 be formed by taking the path @xmath143 of length @xmath144 , and adding @xmath24 new vertices that are adjacent to both end - vertices of @xmath143 only . it is clear that if @xmath1 is odd , then @xmath142 has odd girth @xmath1 . since between any of the @xmath24 extra vertices there is a path of length @xmath1 , we have @xmath145 . the ordering obtained by taking the two end - vertices of @xmath143 first , and then ordering the other vertices in any way , shows that @xmath146 . so for fixed odd @xmath139 we can not bound @xmath147 by an expression that involves @xmath148 only . neetil and ossona de mendez ( * ? ? ? * section 11.9.3 ) give examples that even if we replace `` there exists a path of length @xmath1 between @xmath4 and @xmath5 '' by `` there exists an _ induced _ path of length @xmath1 between @xmath4 and @xmath5 '' in the definition of @xmath31 , it is not possible to reduce the bound on the odd girth in theorem [ thm11](b ) . the upper bounds obtained by neetil and ossona de mendez in their proof of theorem [ thm3](a ) are very large , even for @xmath76 . their proof relies on the concept of _ @xmath1-centred colourings _ of graphs . a ( proper ) colouring of a graph @xmath6 is a _ @xmath1-centred colouring _ if for each connected induced subgraph @xmath149 of @xmath6 , either one colour appears exactly once on @xmath149 or @xmath149 gets at least @xmath1 colours . this is what is proved in @xcite . [ realprev ] + let @xmath1 be an odd positive integer . if a graph @xmath6 has a @xmath1-centred colouring that uses at most @xmath150 colours , then @xmath8}})\le n2^{n2^n}$ ] . given a graph @xmath6 , the _ star chromatic number _ @xmath151 is the smallest number of colours needed to properly colour @xmath6 such that every two colours induce a star forest ( a forest where every component is isomorphic to a _ star @xmath152 _ ) . it is easy to see that a colouring of a graph is 3-centred if and only if every two colours induce a star forest . albertson et al . @xcite showed that the star chromatic number of planar graphs is at most 20 , and there exist planar graphs with star chromatic number equal to 10 . this means that the best known upper bound on @xmath77}})$ ] for planar graphs given by theorem [ realprev ] is @xmath153 , while the best possible upper bound for planar graphs that can be found using that theorem directly is @xmath154 . an alternative bound can be obtained from theorem [ realprev ] using the following result . + every graph @xmath6 has a @xmath1-centred colouring that uses at most @xmath155 colours . [ first ] + let @xmath1 be an odd positive integer and @xmath6 a graph . setting @xmath156 we have @xmath8}})\le w2^{w2^w}$ ] . as far as we are aware , the best known upper bound for @xmath157 for planar is given by the following result . [ binomcol ] + for every positive integer @xmath24 and planar graph @xmath6 we have @xmath158 . so we have @xmath159 for planar graphs , which , when combined with corollary [ first ] , unfortunately gives a worse bound for @xmath77}})$ ] for planar graphs than the one obtained earlier . on the other hand , combining theorems [ thm10](a ) and [ binomcol ] already gives the significantly better upper bound @xmath77}})\le231 $ ] for planar graphs . in section [ sec3 ] we will show that this bound can be lowered further to 143 . the remainder of this paper is organised as follows . in the next section we prove our main results , theorems [ thm10 ] and [ thm11 ] . we use the results from that section in section [ sec3 ] to find explicit upper bounds for the chromatic number of exact distance graphs for some specific classes of graphs , including graphs with bounded genus , graphs with bounded tree - width , and graphs without a specified complete minor . in section [ sec4 ] we describe the graph promised in theorem [ thm5](b ) . we close with a number of open problems and directions for further study . we need a few more definitions . for a positive integer @xmath24 , we denote @xmath160=\ { 1,2,\dots , k\}$ ] . for a vertex @xmath161 , we will denote by @xmath162 the _ @xmath24-th neighbourhood _ of @xmath5 , that is , the set of vertices different from @xmath39 with distance at most @xmath24 from @xmath39 ; and we set @xmath163=n^k(v)\cup\{v\}$ ] . as is standard , we write @xmath164 for @xmath165 . for later use , we actually prove a slightly stronger result , which involves a more technical variant of the generalised colouring numbers . let @xmath0 be a graph , @xmath83 a linear ordering of @xmath3 , and @xmath24 a positive integer . for a vertex @xmath89 , let @xmath166 be the set of vertices @xmath4 such that there is a @xmath90-path @xmath167 , with @xmath168 , @xmath169 , of length @xmath170 , such that @xmath4 is the minimum vertex in @xmath171 with respect to @xmath83 , and such that @xmath172 for @xmath173 . we define the _ distance-@xmath24-colouring number @xmath174 _ of a graph @xmath6 as follows : @xmath175 since @xmath176 for every ordering @xmath83 , distance @xmath24 and vertex @xmath5 , we obtain @xmath177 . on the other hand , we also have @xmath178 , which implies that @xmath179 . we will prove the following sharpening of theorem [ thm10 ] . [ thm10d ] 1.5to1.5for every odd positive integer @xmath1 and every graph @xmath6 we have @xmath8}})\le{\mathrm{dcol}}_{2p-1}(g)$ ] . 1.5to1.5for every even positive integer @xmath1 and every graph @xmath6 we have @xmath8}})\le{\mathrm{dcol}}_{2p}(g)\cdot\delta(g)$ ] . ( a)for an odd positive integer @xmath1 and graph @xmath0 , set @xmath180 and let @xmath83 be an ordering of @xmath3 that witnesses @xmath181 . moving along the ordering @xmath83 we assign to each vertex @xmath89 a colour @xmath182 $ ] that is different from @xmath183 for all @xmath184 . next , define @xmath185 as the minimum vertex with respect to @xmath83 of the vertices in @xmath186 $ ] , and define @xmath187 $ ] by @xmath188 . we claim that @xmath189 is a @xmath190-colouring of @xmath2}}$ ] . consider an edge @xmath191}})$ ] . so there exists a path @xmath192 with @xmath193 and @xmath194 . clearly , @xmath195\cap n^{\lfloor p/2 \rfloor}[v]=\varnothing$ ] , and hence @xmath196 . without loss of generality , assume @xmath197 . since @xmath198 $ ] , there exists a path @xmath199 between @xmath200 and @xmath201 of length at most @xmath202 such that @xmath203 $ ] . similarly , there exists a path @xmath204 between @xmath205 and @xmath206 of length at most @xmath207 such that @xmath208 $ ] . since @xmath195\cap n^{\lfloor p/2 \rfloor}[v]=\varnothing$ ] and @xmath209 , we can combine these paths to a path @xmath210 between @xmath200 and @xmath206 of length at most @xmath211 . note that if we write @xmath212 with @xmath213 and @xmath214 , then the vertices @xmath215 for @xmath216 all lie on @xmath204 , hence are in @xmath217 $ ] . since @xmath206 is the minimum vertex in @xmath217 $ ] , we have @xmath218 for those @xmath215 . thus @xmath210 witnesses that @xmath219 . we conclude that @xmath220 , as required . ( b)for an even positive integer @xmath1 and graph @xmath0 , set @xmath221 and let @xmath83 be an ordering of @xmath3 that witnesses @xmath222 . moving along the ordering @xmath83 we assign to each vertex @xmath89 a colour @xmath182 $ ] that is different from @xmath183 for all @xmath223 . additionally , for each vertex @xmath5 , choose an injective function @xmath224 $ ] . next , define @xmath185 as the minimum vertex with respect to @xmath83 of the vertices in @xmath225 $ ] . we also choose an arbitrary vertex in @xmath226 ; call it @xmath227 . to each vertex @xmath5 we assign as its colour the pair @xmath228 . it is clear that this colouring uses at most @xmath229 colours , and we claim that it is a proper colouring of @xmath2}}$ ] . consider an edge @xmath191}})$ ] . first suppose that @xmath230 . then we can follow the proof of part ( a ) to conclude that @xmath231 , and hence the colours of @xmath45 and @xmath39 differ in the first coordinate . so we are left with the case @xmath232 . since @xmath233 , we have that @xmath234 , while @xmath235 . this means that @xmath236 . together with the fact that @xmath237 , we obtain that @xmath238 . this gives that the colours of @xmath45 and @xmath39 differ in the second coordinate , which completes the proof . in the proof of theorem [ thm11 ] we use the following lemmas . [ twoweak ] + let @xmath6 be a graph and @xmath83 a linear ordering of @xmath3 . let @xmath239 be distinct vertices in @xmath6 . if @xmath4 is weakly @xmath24-accessible from @xmath5 , and @xmath93 is weakly @xmath40-accessible from @xmath5 , then @xmath4 is weakly @xmath240-accessible from @xmath93 or @xmath93 is weakly @xmath240-accessible from @xmath4 . since @xmath4 is weakly @xmath24-accessible from @xmath5 , there is a path @xmath241 of length @xmath242 for which all internal vertices @xmath243 satisfy @xmath244 . also , since @xmath93 is weakly @xmath40-accessible from @xmath5 , there is a path @xmath245 of length @xmath246 for which all internal vertices @xmath247 satisfy @xmath248 . then , if @xmath94 , there is an @xmath249-path of length at most @xmath250 with all internal vertices greater than @xmath4 in @xmath83 ; hence , @xmath4 is weakly @xmath240-accessible from @xmath93 . similarly , if @xmath251 , then @xmath93 is weakly @xmath240-accessible from @xmath4 . [ lem2 ] + let @xmath1 be a positive integer and @xmath6 a graph with odd girth at least @xmath61 1.5to1.5 every closed walk of odd length has length at least @xmath61 . 1.5to1.5 let @xmath90 be different vertices and @xmath252 a walk between @xmath4 and @xmath5 of length @xmath253 . then there exists a path between @xmath4 and @xmath5 of length @xmath254 such that @xmath255 and @xmath256 have the same parity . the proof of ( a ) is straightforward , since a closed walk of odd length contains a cycle of odd length . for ( b ) , let @xmath257 , with @xmath258 and @xmath259 . if @xmath252 itself is not a path , then some vertex @xmath93 appears more than once in @xmath252 . the part of @xmath252 between the first and last appearances of @xmath93 is a closed walk @xmath260 of length @xmath261 . using ( a ) we obtain that @xmath262 must be even . hence , if we remove @xmath260 from @xmath252 , we get a shorter walk between @xmath4 and @xmath5 of length @xmath263 . additionally , the resulting walk has fewer vertices that appear more than once than @xmath252 does . hence , if we do not immediately obtain a path , we can repeat this procedure inductively until we obtain an @xmath17-path with the desired property . + for both parts of the theorem we use the same colouring . let @xmath83 be an ordering of @xmath3 such that @xmath264 . we first create an auxiliary colouring @xmath182 $ ] by moving along the ordering @xmath83 , and assigning to each vertex @xmath89 a colour @xmath182 $ ] that is different from @xmath183 for all @xmath265 . next , for a vertex @xmath266 , let @xmath267 be the minimum integer @xmath24 such that @xmath4 is weakly from @xmath5 , and set @xmath268 . define the function @xmath269\rightarrow \bigl[\bigl\lfloor\frac12p\bigr\rfloor\bigr]\cup\{-1,0\}$ ] as follows . for a colour @xmath270 $ ] , let @xmath271 -1 , & \text{otherwise}. \end{array}\right.\ ] ] by lemma [ twoweak ] and the definition of @xmath183 , we see that if @xmath272 satisfies @xmath273 , then @xmath4 is the only vertex in @xmath274 with colour @xmath275 . that implies that @xmath276 is well defined . the number of possible functions @xmath269\rightarrow \bigl[\bigl\lfloor\frac12p\bigr\rfloor\bigr]\cup\{-1,0\}$ ] is @xmath277 . we will prove that labelling each vertex @xmath89 with @xmath276 gives a proper colouring for the graphs and situations described in parts ( a ) and ( b ) of the theorem . it is more convenient to do part ( b ) first . ( b)consider two vertices @xmath278 for which there exists a path of length @xmath1 between @xmath45 and @xmath39 . without loss of generality we assume @xmath279 . if @xmath45 is weakly @xmath1-accessible from @xmath39 in @xmath83 , then we know that @xmath280 , and hence @xmath281 . so we are left with the case in which @xmath45 is not weakly @xmath1-accessible from @xmath39 in @xmath83 . let @xmath24 be the length of the shortest odd - length path between @xmath45 and @xmath39 . we obviously have @xmath282 . because @xmath45 is not weakly @xmath1-accessible from @xmath39 in @xmath83 , we also have @xmath283 , hence @xmath284 . let @xmath285 be a path of length @xmath24 between @xmath286 an @xmath287 . let @xmath288 be the vertex of @xmath91 that is minimum with respect to the ordering @xmath83 . since @xmath279 , we get that @xmath289 , and , since @xmath45 is not weakly @xmath1-accessible from @xmath39 , we see that @xmath290 . therefore , @xmath288 is weakly @xmath291-accessible from @xmath45 and weakly @xmath292-accessible from @xmath39 . first consider the case that @xmath293 . then @xmath294 . we want to prove that @xmath295 . for this , assume that @xmath296 . hence there is a path @xmath297 between @xmath45 and @xmath288 of length @xmath298 . if @xmath40 and @xmath298 have different parity , then the union of @xmath297 and the path @xmath299 gives a closed walk of odd length @xmath300 , which contradicts lemma [ lem2](a ) . so @xmath298 and @xmath40 have the same parity . now if we replace in the path @xmath91 the part @xmath299 with @xmath297 , we get a walk between @xmath45 and @xmath39 of length @xmath301 , hence with odd length . by lemma [ lem2](b ) , this walk contains a path between @xmath45 and @xmath39 of odd length at most @xmath301 , which contradicts the choice of @xmath91 . so we know that @xmath295 . notice that since there is a path of length @xmath302 between @xmath288 and @xmath39 , we have that @xmath303 . since @xmath304 , we have that @xmath305 , and hence @xmath306 . now consider a vertex @xmath307 with @xmath308 . we first prove that @xmath309 . for suppose this is not the case , then there is a path from @xmath39 to @xmath288 of length @xmath40 . together with the part of @xmath310 from the path @xmath91 , this gives a closed walk of length @xmath282 . since @xmath24 is odd , this contradicts lemma [ lem2](a ) . since @xmath311 , @xmath312 and @xmath309 , by lemma [ twoweak ] we get that @xmath4 is weakly @xmath1-accessible from @xmath288 or @xmath288 is weakly @xmath1-accessible from @xmath4 . this gives @xmath313 , which implies , by choice of @xmath4 , that @xmath314 . if @xmath315 , we can prove in a similar way that @xmath316 , which completes the proof of part ( b ) of the theorem . ( a)this time we consider two vertices @xmath278 that have distance @xmath24 in @xmath6 , for some odd integer @xmath282 . ( to prove the statement , it would be enough to prove the case @xmath317 , but we prefer to give the proof of a more general statement . ) we can more or less follow the proof of part ( b ) above , working with a shortest path @xmath285 between @xmath286 and @xmath287 . since @xmath91 is a shortest path , we immediately get that @xmath318 and @xmath319 . this also means that @xmath309 , since @xmath320 . for the remainder , the proofs are exactly the same . the proofs of theorem [ thm11](a ) and ( b ) above give results that are stronger than the statements in the theorem . we already discussed in subsection [ ssec1.2 ] that in fact we prove upper bounds on @xmath121}}\cup{g^{[\natural 3]}}\cup\cdots\cup{g^{[\natural p]}})$ ] and @xmath122 . additionally , in part ( a ) we could replace the condition that we add an edge @xmath42 to @xmath2}}$ ] if @xmath233 , i.e. `` there is a shortest path of length @xmath1 between @xmath45 and @xmath39 '' , by the weaker condition `` there is a path @xmath91 of length @xmath1 between @xmath45 and @xmath39 such that any shorter path between those vertices is internally disjoint from @xmath91 '' . in this section we use theorem [ thm10d](a ) to find explicit upper bounds for the chromatic number of exact distance graphs for certain types of graphs , including planar graphs , graphs with bounded tree - width , and graphs without a complete minor . obtaining these bounds involves finding upper bounds for the distance-@xmath24-colouring numbers @xmath174 . more explicitly , we will prove the following results . [ dcolplan ] + let @xmath24 be a positive integer . 1.5to1.5for every planar graph @xmath6 we have @xmath321 . 1.5to1.5for every graph @xmath6 with genus @xmath322 we have @xmath323 . [ dcoltw ] + let @xmath24 and @xmath262 be positive integers . for every graph @xmath6 with tree - width at most @xmath262 we have @xmath324 . [ minor ] + let @xmath24 and @xmath262 be positive integers with @xmath325 . for every @xmath10-minor free graph @xmath6 we have @xmath326 . since outerplanar graphs @xmath6 have tree - width at most 2 , combining theorems [ thm10d](a ) and [ dcoltw ] gives @xmath77}})\le13 $ ] . similarly , from theorem [ dcolplan ] we see that for planar graphs @xmath6 we have @xmath77}})\le143 $ ] , while for graphs @xmath6 embeddable on the torus we have @xmath77}})\le165 $ ] . we will prove those theorems in the remainder of this section . they are based on the methods developed in van den heuvel et al . @xcite to obtain bounds for the generalised colouring numbers . proposition [ pro : coltwtd ] tells us that @xmath111 . in @xcite , grohe et al.provided a sharp upper bound for the weak colouring numbers @xmath98 of a graph @xmath6 in terms of its tree - width . the following result is implicit in the proof of ( * ? ? ? * theorem 4.2 ) . [ tw ] + let @xmath6 be a graph and @xmath83 a linear ordering of @xmath327 with @xmath328 . for every positive integer @xmath24 and vertex @xmath329 we have @xmath330 . since @xmath6 has tree - width at most @xmath262 , we have @xmath331 , and so there is an ordering @xmath83 of @xmath327 such that @xmath328 . by lemma [ tw ] we get that for every positive @xmath24 and @xmath329 we have @xmath332 . consider @xmath333 , and for each @xmath334 fix a path @xmath335 with @xmath170 , @xmath336 , @xmath286 , such that @xmath45 is minimum on @xmath337 , and such that @xmath338 , which denotes the greatest index with @xmath339 , satisfies @xmath340 . such a path exists by the definition of @xmath341 . it is clear that subpaths of @xmath337 give that @xmath342 is @xmath343-accessible from @xmath39 and @xmath45 is weakly @xmath338-accessible from @xmath342 . since @xmath170 and @xmath340 , we obtain that @xmath342 is @xmath24-accessible from @xmath39 and @xmath45 is weakly @xmath344-accessible from @xmath342 . together with our choice of @xmath83 , this gives us the following estimates : @xmath345 the result follows immediately . in order to provide upper bounds for the generalised colouring numbers for graphs that exclude a fixed minor , van den heuvel et al . @xcite constructed vertex partitions where each part has neighbours in a bounded number of other parts only , and where the intersection of each part with the @xmath24-neighbourhood of any vertex is also bounded . we will make use of these decompositions for our proofs as well . a _ decomposition _ of a graph @xmath6 is a sequence @xmath346 of non - empty subgraphs of @xmath6 such that the vertex sets @xmath347 partition @xmath327 . the decomposition @xmath348 is _ connected _ if each @xmath349 is connected . let @xmath346 be a decomposition of a graph @xmath6 , @xmath67 a positive integer , and @xmath124 a component of @xmath350 . we define the _ @xmath67-th separating number of @xmath124 _ as @xmath351\mid e(c , h_j)\ne\varnothing\}|$ ] , where @xmath352 is the set of edges with one end - vertex in @xmath124 and the other end - vertex in @xmath353 . let @xmath354 , where the maximum is taken over all components @xmath124 of @xmath355 . we define the _ width _ of @xmath356 as @xmath357 . let @xmath6 be a graph , let @xmath358 be a _ connected _ subgraph of @xmath6 , and let @xmath359 be a function . we say that @xmath149 _ @xmath360-spreads on @xmath6 _ if , for every @xmath361 and @xmath333 , we have @xmath362\cap v(h)|\le f(k).\ ] ] we say a decomposition @xmath356 is _ @xmath360-flat _ if each @xmath349 @xmath360-spreads on @xmath363 . we say @xmath356 is a _ flat decomposition _ if @xmath356 is an @xmath360-flat decomposition for some function @xmath364 . van den heuvel et al . @xcite related the width of a connected decomposition to the tree - width of the minor obtained by contracting each part . [ width ] + let @xmath6 be a graph , and let @xmath365 be a connected decomposition of @xmath6 of width at most @xmath262 . by contracting each ( connected ) subgraph @xmath349 to a single vertex , we obtain a graph @xmath149 with @xmath40 vertices and tree - width at most @xmath262 . the proof of the lemma in @xcite shows the power of generalised colouring numbers . it actually gives a short argument that the contracted graph @xmath149 satisfies @xmath366 . the bound on the tree - width then follows by proposition [ pro : coltwtd ] . moreover , the proof shows that the ordering @xmath83 of @xmath367 obtained by setting @xmath368 if @xmath369 satisfies @xmath370 . using this property we can prove that if the decomposition from which @xmath149 was obtained is @xmath360-flat , then we can find an upper bound on @xmath174 in terms of @xmath371 . [ flatbound ] + let @xmath364 and let @xmath372 be positive integers . for every graph @xmath6 that admits a connected @xmath360-flat decomposition of width at most @xmath262 we have @xmath373 . the proof of this lemma is similar to that of ( * ? ? ? * lemma 3.5 ) . let @xmath365 be a connected @xmath360-flat decomposition of @xmath6 of width @xmath262 . since @xmath356 is connected , we know , by lemma [ width ] , that contracting the subgraphs in @xmath356 leads to a graph @xmath149 with tree - width at most @xmath262 . we identify the vertices of @xmath149 with the subgraphs @xmath349 , and define a linear ordering @xmath83 on @xmath367 by setting @xmath368 if @xmath369 . by the proof of ( * lemma 3.1 ) we get that @xmath83 satisfies @xmath370 . using lemma [ tw ] this implies that @xmath374 for any vertex @xmath375 . arguing as in the proof of theorem [ dcoltw ] , this means that for every @xmath375 we have @xmath376 . from @xmath83 we define an ordering @xmath377 on @xmath327 in the following way . for @xmath378 and @xmath379 with @xmath380 , we let @xmath381 if @xmath369 . then , for every @xmath382 , we order the vertices of @xmath349 in any order . it is easy to see that any vertex @xmath383 satisfies @xmath384\cap\bigl(h_i\cup \{\,h_j\mid h_j\in d_{l , k}(h_i)\,\}\bigr).\ ] ] hence , we have that there are at most @xmath385 subgraphs among @xmath386 in @xmath6 that contain vertices from @xmath387 . since @xmath356 is @xmath360-flat , we know that the intersection of each of these subgraphs with @xmath163 $ ] is at most @xmath371 . finally , since @xmath387 is a proper subset of @xmath163 $ ] ( as @xmath388 ) , the result follows . also in @xcite , it was proved that graphs that do not contain a complete graph as a minor have flat decompositions of small width . [ flatslim ] + let @xmath389 and let @xmath364 be the function @xmath390 . for every @xmath10-minor free graph @xmath6 we have that there is a connected @xmath360-flat decomposition of @xmath6 with width at most @xmath391 . combining lemmas [ flatbound ] and [ flatslim ] immediately gives theorem [ minor ] . flat decompositions of even smaller width were found in @xcite for planar graphs . [ maxp ] + let @xmath364 be the function @xmath392 . for every maximal planar graph @xmath6 we have that there is a connected @xmath360-flat decomposition of @xmath6 with width at most 2 . this lemma allows us , through lemma [ flatbound ] , to prove theorem [ dcolplan](a ) . by combining lemmas [ flatbound ] and [ maxp ] , we get the bound on @xmath174 for maximal planar graphs . since @xmath174 can not decrease when edges are added , we conclude that any planar graph satisfies the same inequality . we say a path is _ optimal _ if it is a shortest path between its end - points . to prove theorem [ dcolplan](b ) , we use the following easy result which states that a decomposition in which all subgraphs are optimal paths is @xmath360-flat for @xmath392 . [ short ] + let @xmath6 be a graph , @xmath5 be a vertex of @xmath6 , and @xmath91 be an optimal path in @xmath6 . then @xmath91 contains at most @xmath393 vertices of the closed @xmath24-neighbourhood @xmath394 $ ] of @xmath5 . the proof of this statement is similar to the proof of ( * ? ? ? * theorem 1.5(a ) ) . we assume @xmath395 , as otherwise the result holds by theorem [ dcolplan](a ) . it is well known ( see e.g. ( * ? ? ? * lemma 4.2.4 ) or @xcite ) that a graph of genus @xmath395 contains a non - separating cycle @xmath124 that consists of two optimal paths and such that @xmath396 has genus @xmath397 . we construct a linear order @xmath83 of @xmath3 in the following way . the first vertices in @xmath83 will be the vertices in such a cycle @xmath124 . if after removing that cycle the genus of the resulting graph is greater than 0 , then we choose another such cycle , make its vertices the next ones in the ordering , and remove the cycle . we repeat this process inductively until the resulting graph is a planar graph @xmath398 . the vertices of @xmath398 are placed at the end of @xmath83 , ordered in the way the gives the bound on @xmath399 from theorem [ dcolplan](a ) . lemma [ short ] tells us that for any vertex @xmath5 and optimal path @xmath91 we have @xmath400\cap v(p)|\le 2k+1 $ ] for every @xmath24 . hence @xmath401 for every vertex @xmath5 and optimal path @xmath91 . it follows that for any vertex @xmath5 in @xmath6 , the set @xmath402 can have at most @xmath403 vertices on the removed cycles . ( each of the two optimal paths that form a cycle is optimal after the earlier cycles are removed , and vertices can not belong to @xmath402 through vertices in older cycles . ) only a vertex @xmath4 in the planar graph @xmath398 can have other vertices of @xmath398 in @xmath404 and theorem [ dcolplan](a ) gives us a bound on the number of such vertices . hence , we obtain that every @xmath5 in @xmath6 satisfies @xmath405 the result follows . with @xmath81}})=5$].,height=2 ] in ( * ? ? ? * exercise 11.4 ) a planar graph @xmath6 such that @xmath77}})=6 $ ] is given ( see also @xcite ) . as we will prove below , the outerplanar graph @xmath80 in figure [ five ] satisfies @xmath81}})=5 $ ] . we will use that graph to construct a planar graph @xmath78 such that @xmath79}})=7 $ ] . + there is a planar graph @xmath78 such that @xmath79}})=7 $ ] . we will prove first that @xmath81}})=5 $ ] , using the vertex labelling provided in figure [ five ] . consider a proper colouring of @xmath406}}$ ] . note that @xmath407 and @xmath408 form disjoint 5-cycles @xmath406}}$ ] . hence , the vertices in @xmath409 need at least 3 colours . given that @xmath410 in @xmath406}}$ ] , if we use more than 3 colours on @xmath409 , then we already use at least 5 colours . so assume that the vertices in @xmath409 are coloured with 3 colours only . since @xmath411 in @xmath406}}$ ] for @xmath412 , and @xmath413}})$ ] , we need at least 2 extra colours . so we always use at least 5 colours in a proper colouring of @xmath406}}$ ] . figure [ five ] gives a colouring of @xmath80 with 5 colours which is a proper colouring of @xmath406}}$ ] . this shows that @xmath81}})=5 $ ] . now let @xmath414 and @xmath415 be two disjoint copies of @xmath80 . let @xmath149 be a path on 5 vertices , disjoint from @xmath414 and @xmath415 , with vertices @xmath416 in that order , together with the edge @xmath417 . ( this is exactly the graph formed by the vertices @xmath418 in figure [ five ] . ) the graph @xmath419 has vertex set and edge set : @xmath420 finally , the graph @xmath78 is obtained from @xmath419 by subdividing once all the edges of the form @xmath421 and @xmath422 ( replacing each edge by a path of length 2 ) . since @xmath80 is outerplanar , it is easy to check that @xmath78 is planar . if @xmath423 and @xmath91 is an @xmath278-path in @xmath78 but @xmath424 , then @xmath425 . thus the length of @xmath91 is at least 4 . we conclude that if two vertices @xmath278 have distance 3 in @xmath78 , then any shortest @xmath278-path has all its vertices in @xmath426 . therefore , the number of colours needed to colour the vertices of @xmath414 in @xmath427}}$ ] is 5 , and the same applies to @xmath415 . we now can argue as in the proof of @xmath81}})=5 $ ] above to reach the conclusion @xmath79}})=7 $ ] . since the graph @xmath80 in figure [ five ] is outerplanar , it does not have @xmath65 as a minor . also , the graph @xmath78 we constructed above is planar , so does not have @xmath428 as a minor . we can iterate the construction to obtain graphs @xmath429 that are @xmath10-minor free , for @xmath325 , and for which @xmath430}})\ge2(t-2)+1 $ ] . to obtain @xmath431 from @xmath429 , we take two copies of @xmath429 , one copy of the graph @xmath149 from above , and add paths of length 2 between all vertices in the first copy of @xmath429 and @xmath432 , and between all vertices in the second copy of @xmath429 and @xmath433 . it is straightforward to check that if @xmath429 is @xmath10-minor free , then @xmath431 is @xmath434-minor free , and that @xmath435}}$ ] needs at least 2 more colours than @xmath436}}$ ] does . the property that for @xmath437 there exists graphs @xmath6 that are @xmath10-minor free and satisfy @xmath77}})\ge2(t-2)+1 $ ] does not extend to @xmath438 . to see this , note that the only graphs that are @xmath439-minor free are acyclic graphs ( i.e. forests ) , which implies they are bipartite . and for bipartite graphs @xmath6 we have that @xmath440}}$ ] is bipartite as well ( in fact , even the exact @xmath1-power graph @xmath31 is bipartite for every odd @xmath1 ) , hence @xmath77}})\le2 $ ] . notice that one can construct the graph @xmath80 of figure [ five ] ( and the graphs @xmath429 for @xmath325 ) by using operations similar to those of used in the hajs construction @xcite . consider the graph @xmath210 induced on @xmath80 by @xmath441 . the main connected component of the graph @xmath442}}$ ] consists of a cycle and two apex vertices , @xmath93 and @xmath443 , that are adjacent to all the vertices in the cycle . one can obtain @xmath80 by taking two copies of @xmath210 , identifying the two vertices that correspond to @xmath93 , and adding an edge between the two vertices that correspond to @xmath444 . in the exact distance-3 graph , we see that one of the apex vertices has been identified , while those that correspond to @xmath443 have been joined by an edge . however , the operation of deletion , used in the hajs construction , is not used in our construction . this is mainly because we want to obtain a graph with chromatic number strictly larger than that of the parts it is formed of . in this paper we give bounds on the chromatic number of exact distance graphs for some classes of graphs . in general , the difference between the best lower and upper bounds are still quite large , so we ca nt really claim we have an insight of what the correct best possible bounds are . a second feature of our upper bounds is that they are expressed in terms of generalised colouring numbers , and hence are increasing with the distance . however , for none of the classes of graphs do we have examples that show that @xmath8}})$ ] grows when the odd number @xmath1 increases . this feature was noticed earlier for planar graphs . the following problem , which is attributed to van den heuvel and naserasr , appears in @xcite ( see also @xcite ) . of course , the same question can be asked for graph classes with bounded tree - width and graphs that exclude a complete minor . it is known that for all those classes the weak number is not bounded when @xmath1 increases . grohe et al . @xcite proved that for every pair of positive integers @xmath445 there is a graph @xmath446 of tree - width @xmath255 such that @xmath447 . since graphs of tree - width @xmath255 are @xmath448-minor free , for every @xmath449 the graph @xmath450 is @xmath10-minor free and satisfies @xmath451 . note that this also means that @xmath174 is unbounded for those graphs , since , as remarked earlier , @xmath452 for any graph @xmath6 . the one , trivial , example for which we can answer these questions is for graphs @xmath6 without @xmath439-minor . as noted at the end of the previous section , those graphs are bipartite , and hence @xmath8}})\le\chi({g^{\natural p}})\le2 $ ] for every odd @xmath1 . as we mentioned in section [ sec1 ] , the proof of theorem [ thm11 ] actually gives that for a class of graphs @xmath54 with bounded expansion we can find a constant @xmath453 such that @xmath121}}\cup{g^{[\natural 3]}}\cup\cdots\cup{g^{[\natural p]}})\le n$ ] . there are constructions that show that this constant must grow with @xmath1 , even if @xmath54 is the class of outerplanar graphs . one such construction appears in @xcite . a very simple one , which we sketch in figure [ largek ] , can be found in @xcite . for a graph @xmath6 , a natural generalisation of @xmath456}}\cup { g^{[\natural 3]}}\cup \dots \cup{g^{[\natural p]}}$ ] is the graph @xmath457 , which has the same vertex set as @xmath6 , and @xmath17 is an edge in @xmath457 if and only if @xmath4 and @xmath5 have odd distance . both constructions in the previous paragraph tell us that for outerplanar graphs @xmath6 the chromatic number of @xmath457 can be arbitrarily large because the clique number @xmath454 can be arbitrarily large . this motivates the following open problem of thomass , which appears in @xcite ( see also @xcite ) . another area that is ripe for further research is the chromatic number of exact distance graphs with even distance , for specific classes of graphs . theorem [ thm10](b ) gives a first result for even distances . there is very little we know about the dependencies between @xmath8}})$ ] and @xmath131 for even @xmath1 . it is well - known , and easy to prove ( see , e.g. , @xcite ) , that for every graph @xmath6 we have @xmath459 , hence certainly @xmath460}})\le(2{\mathrm{col}}(g)-3)\cdot\delta(g)$ ] . this suggests that there might exist a function @xmath461 such that @xmath8}})\le\varphi({\mathrm{wcol}}_{p-1}(g))\cdot\delta(g)$ ] , or even @xmath8}})\le\varphi({\mathrm{wcol}}_{p/2}(g))\cdot\delta(g)$ ] . we have not been able to prove such a result . neither do we know what the best value of @xmath462 should be such that a result of the form @xmath8}})\le\varphi({\mathrm{wcol}}_{r(p)}(g))\cdot\delta(g)$ ] is possible for even @xmath1 . krumke , m.v . marathe , and s.s . ravi , _ approximation algorithms for channel assignment in radio networks_. in : _ dial m for mobility ; 2nd international workshop on discrete algorithms and methods for mobile computing and communications_. dallas , texas , 1998 .

we report x - ray observations of the nearby , powerful radio galaxy pictor a with the chandra observatory and optical and near uv observations of its western radio hot spot with the hubble space telescope . x - ray emission is detected from the nucleus , a 19 ( 110 kpc ) long jet to the west of the nucleus , the western radio hot spot some 42 ( 240 kpc ) from the nucleus , and the eastern radio lobe . the morphology of the western hot spot is remarkably similar to that seen at radio and optical wavelengths , where the emission is known to be synchrotron radiation . the x - ray spectrum of the hot spot is well described by an absorbed power law with photon index = 2.07 0.11 . the x - ray jet coincides with a weak radio jet and is laterally extended by 20 ( 1.9 kpc ) . the observed jet is up to 15 times brighter in x - rays than any counter jet , a difference ascribed to relativistic boosting as the western radio lobe is probably the closer . the jet s spectrum is well modelled by an absorbed power law with = 1.94 and poorly fitted by a raymond - smith thermal plasma model . the emission processes responsible for the x - rays are discussed in detail . the radio - optical spectrum of the hot spot breaks or turns down at 10 hz , and its x - ray spectrum is not a simple extension of the radio - optical spectrum to higher frequencies . thermal models for the hot spot s x - ray emission are ruled out . synchrotron self - compton models involving scattering from the _ known _ population of electrons give the wrong spectral index for the hot spot s x - ray emission and are also excluded . a composite synchrotron plus synchrotron self - compton model can match the x - ray observations but requires similar contributions from the two components in the chandra band . we show that the hot spot s x - ray emission could be synchrotron self - compton emission from a hitherto unobserved population of electrons emitting at low radio frequencies , but do not favor this model in view of the very weak magnetic field required . an inverse compton model of the jet , in which it scatters microwave background photons but moves non - relativistically , requires a magnetic field a factor of 30 below equipartition , and ad hoc conditions to explain why the radio lobes are fainter than the jet in x - rays but brighter in the radio . these problems are alleviated if the jet moves relativistically , but models with an equipartition field require an implausibly small angle ( ) between the jet and the line of sight . a model with 23 and a field a factor of 6 below equipartition seems viable . synchrotron radiation is an alternative process for the x - ray emission . the expected synchrotron spectrum from relativistic electrons accelerated by strong shocks and subject to synchrotron radiation losses is in very good agreement with that observed for both the hot spot and jet . the possibility that the relativistic electrons result via photo - pion production by high energy protons accelerated in shocks ( a ` proton induced cascade ' ) is briefly discussed .

astro-ph0008467

x - ray observations of radio galaxies are of value for several reasons . low brightness radio lobes may be thermally confined by the hot , extended atmospheres in elliptical galaxies or clusters of galaxies and x - ray observations can provide the density , temperature and pressure of this gas , thus constraining the pressures of the relativistic gas and magnetic field in a lobe ( e.g. worrall & birkinshaw 2000 ) . high brightness regions , such as the hot spots in fanaroff - riley class ii ( frii ) radio galaxies , can not be statically confined by thermal pressure and must be expanding and/or advancing away from the nucleus . if the motion is supersonic , such a feature must be preceded by a bow shock in the surrounding medium . the effect of bow shocks on the ambient gas may be identifiable in thermal x - ray emission , depending on the run of temperature and density in the post - shock gas ( e.g. clarke , harris & carilli 1997 ) . thermal x - rays may also be detectable from radio jets and lobes themselves if they entrain and shock surrounding gas ( e.g. de young 1986 ) . the relativistic particles responsible for the synchrotron radio emission must also radiate through inverse compton scattering of ambient photons ( such as the microwave background and the starlight of the galaxy ) or the synchrotron photons themselves . this inverse compton radiation is expected to be spread over a wide range of frequencies , including the x - ray band . detection of both synchrotron and inverse compton radiation provides a direct measurement of the magnetic field in the emitting region ( e.g. harris , carilli & perley 1994 ) , allowing a check on the common assumption of equipartition of energy between cosmic rays and magnetic fields . lastly , synchrotron x - ray emission may be observable if electrons and/or positrons of sufficiently high energy are present , a finding which would provide important clues about the origin of the relativistic particles in these objects . for these reasons , we have begun a program of imaging and spectroscopy of radio galaxies with the chandra x - ray observatory . pictor a is the seventh brightest extragalactic radio source in the sky at 408 mhz ( robertson 1973 ) , and the most powerful radio galaxy ( p@xmath7 = 6.7 @xmath8 10@xmath9 w hz@xmath10 for z = 0.035 , q@xmath11 = 0 and h@xmath11 = 50 km s@xmath10 mpc@xmath10 ) with z @xmath12 0.04 . the best radio maps ( perley , rser & meisenheimer 1997 , hereafter prm ) reveal two round , diffuse lobes with very much brighter radio hot spots on the sides away from the nucleus , corresponding to an frii classification . the total angular diameter of the radio emission is about 76 ( 430 kpc ) . the western hot spot , some 42 ( 240 kpc ) from the nucleus , is a remarkable object , being amongst the brightest of radio hot spots in radio galaxies and quasars . prm have found a very faint radio jet extending from the compact , flat spectrum ( cf . jones , mcadam & reynolds 1994 ) nuclear core to the western hot spot . recently , vlbi observations ( tingay et al . 2000 ) have revealed a milli arc second- ( pc- ) scale nuclear radio jet which is approximately aligned with the large - scale radio jet . simkin et al . ( 1999 ) have reported an optical continuum extension @xmath2 01 ( 95 pc ) to the w of the nucleus and the direction to this extension is also roughly aligned with the larger scale radio jet . rser & meisenheimer ( 1987 ) discovered optical emission from the western hot spot . this emission has a featureless continuum , is strongly polarized and thus of synchrotron origin . these authors also found a marginal detection of the hot spot at x - ray wavelengths in an einstein ipc observation . further observations of the hot spot in the infrared have been reported by meisenheimer , yates & rser ( 1997 ) . the overall spectrum of the western hot spot conforms to a power law with index @xmath13 = 0.74 ( s @xmath14 @xmath15 ) from 327 mhz to @xmath16 10@xmath17 hz , where the spectrum turns down ( see fig . 2i of meisenheimer , yates & rser 1997 ) . the morphologies of the hot spot are very similar at radio and optical wavelengths , with a bright , compact , leading `` core '' and a fainter , following `` filament '' , which extends @xmath2 15@xmath18 ( 14 kpc ) more or less perpendicular to a line joining the hot spot to the nucleus ( prm ) . fainter radio emission associated with the hot spot is also seen to the e and ne of the optical and radio filament , the overall angular extent of the radio hot spot being @xmath2 25@xmath18 ( 24 kpc ) . the radio to optical spectrum of the `` filament '' is somewhat steeper ( @xmath19 = 0.98 ) than that of the bright , compact region ( @xmath19 = 0.87 ) . optical imaging and polarimetry of the bright part of the hot spot with the hubble space telescope ( thomson , crane & mackay 1995 ) confirms the high polarization ( @xmath20 50% ) and resolves the hot spot into highly polarized `` wisps '' elongated nearly perpendicular to the nucleus - hot spot line . the nucleus of pictor a has long ( schmidt 1965 ; danziger , fosbury & penston 1977 ) been known for its strong emission - line spectrum , with broad wings to both forbidden and permitted lines . double - peaked , very broad balmer lines were discovered in 1993/1994 ( halpern & eracleous 1994 ; sulentic et al . 1995 ) ; these double - peaked lines were not present in 1983 . such line profiles are commonly interpreted as emission from an accretion disk ( e.g. storchi - bergmann et al . the nucleus of pictor a is a strong x - ray source ; the spatially integrated emission observed with asca has a power - law continuous spectrum and no evidence for an fe k@xmath13 line ( eracleous & halpern 1998 ) . further observations of the x - ray spectrum of pictor a have been reported recently by padovani et al . ( 1999 ) and eracleous , sambruna & mushotzky ( 2000 ) . we selected pictor a for observation with chandra because of the x - ray emission from the western hot spot , tentatively detected with the einstein ipc and confirmed by us through an inspection of an archival rosat pspc observation . further , the optical synchrotron emission from the hot spot is of sufficient extent ( @xmath16 15@xmath18 ) to be well resolved by chandra , making this object a prime candidate for the study of high energy processes in the hot spots of powerful radio galaxies . in addition to the chandra observations , we also report analysis of archival hubble space telescope ( hst ) optical and near uv imaging observations of the western hot spot . section 2 describes the observations and their reduction while section 3 presents the results . in section 4 , we discuss the orientation of the radio source and the processes responsible for the x - ray emission of the western hot spot and jet . concluding remarks are given in section 5 . pictor a was observed by the chandra x - ray observatory on jan 18 2000 ( sequence number 700018 , obsid 346 ) using the advanced ccd imaging spectrometer ( acis ) spectroscopic array , which provides excellent spatial resolution ( @xmath21 1@xmath18 ) with medium spectroscopic resolution ( @xmath2 130 ev fwhm at 1 kev ) at the aim point on chip s3 ( back illuminated ) . in order to obtain good spatial resolution on the western hot spot while retaining excellent resolution on the nucleus , the latter was placed about 1@xmath22 from the aim point in the direction towards the s3/s2 chip boundary ( the y direction ) . the nucleus was thus about 1@xmath22 from the s3/s2 boundary . the western hot spot was @xmath2 35 from the aim point and also on chip s3 . the spatial resolution at the nucleus is barely different from the optimal , while that at the hot spot is 12 @xmath8 19 fwhm . the eastern radio lobe is partly on s3 and partly on s2 ( front illuminated ) . the total integration time ( `` good time intervals '' ) was 26.177 ksecs , taken with the default frame time of 3.2 sec . there were no flares in the background count rate , which remained constant at 1.4 counts per second over the s3 chip . the data were processed using version r4cu4upd4.2 of the pipeline software , and we have used the updated gain file acisd1999-09-16gainn0003.fits . the data extraction and analysis have been performed using version 1.1 of the ciao software and version 11.0 of xspec . in order to create an x - ray image covering the whole radio source ( shown as fig . 1 ) , it was necessary to remove ccd artifacts that take the form of broad , linear features along the readout direction in chips s2 and s3 . this was accomplished by first copying a spatial subsection of the level 2 events file and rotating it by -37.6@xmath6 , so that the axis of the s array ran horizontally . the image was then smoothed with a gaussian with @xmath23 = 10 and the average of the rows computed , including rejection of those pixels in each column which were more than 2@xmath23 from the mean for that column . the average of the rows was then subtracted from each row . finally , the image was rotated by + 37.6@xmath6 , so the rows and columns corresponded to the cardinal directions . this process not only removed most of the linear artifacts , but also the ` streak ' from the very strong nuclear source that results from its readout and much of the differences in background level between the two chips . to derive the x - ray spectrum of the hot spot , counts were extracted from a rectangular region @xmath24 in extent which includes all of the x - ray emission from the hot spot . background counts were taken from an area of identical shape offset to the west . other extraction regions were tried for the background but no significant changes were seen . we obtain a total of 3413 counts from the hot spot and 125 from the background . response and auxiliary response files were created using the standard ciao tools . the source counts were grouped so that each channel has at least 25 counts so that the @xmath25 fitting in xspec is meaningful . we followed a similar procedure to obtain the x - ray spectrum of the jet . counts were taken from a rectangular region 103@xmath18 @xmath8 24 that contains the whole jet and minimises background ( the @xmath16 10@xmath18 gap between the nucleus and the eastern end of the jet was excluded ) . the background was taken from adjacent regions , and no significant difference was found for the jet s spectrum when using different regions for the background . we obtain a total of 481 counts from the jet and 240 from the background . the x - ray response of the ccds depends on the location of the source on the chip . to account for this , the back - illuminated s3 chip has 1024 individual calibration measurements ( 256 per node ) arranged in a @xmath26 grid covering the face of the chip . this oversamples the variation in the response of the ccd , and a single calibration measurement may be used for a source extending up to @xmath27 pixels ( approximately 32@xmath18 @xmath8 32@xmath18 ) . the x - ray jet extends 228 pixels ( @xmath28 ) from the nucleus , so we divided it into four sections , each of which has a separate response . we also note that the jet crosses a ` node boundary ' ( the division between sections of the chip that have different read - out amplifiers ) , and that the response may vary sharply across this boundary . to avoid any problem with this , we also divided the jet at the node boundary so as not to mix events collected by different nodes . each of the four spectra were grouped to give each channel at least 20 counts so that @xmath25 fitting is valid . these spectra were fitted simultaneously in xspec with each spectrum having its own background , response and auxiliary response file . the spectral properties are assumed to be constant along the length of the jet so the parameters in the fit are tied together with the exception of their normalizations . one drawback of treating the jet in this way is that the individual spectra will have extremely low count rates , and fitting may only be valid over a limited range in energy . it is also difficult to produce clear figures of the spectral fits since there are four almost coincident data sets . to overcome these problems the entire jet spectrum may be treated as a single data set , using an ` averaged ' response function that is the photon weighted average of the four individual responses . we use pi spectra , which are gain corrected , to minimize the effect of collecting photons from two different nodes . the results we obtain using this technique are equivalent , within the errors , to those obtained through fitting the spectra individually . six images of the western hot spot were obtained on august 19 , 1995 as part of go program # 5931 ( pi meisenheimer ) , three each in the f300w and f622w broadband filters . the images in each filter were dithered by several arc seconds in order to improve the effective spatial resolution of the images on the pc chip . however , the lack of duplicate exposures at each pointing , i.e. , no cr - split , and the paucity of sources in the field greatly complicated attempts at alignment of the images . instead , a number of cosmic rays were cleaned from each image by hand , and then the total number of sky - subtracted counts in the hot spot was determined for each frame . after averaging the results for each filter , the counts were converted to units of flux using the synphot package in iraf . the derived values , as plotted in fig . 6 , are 30 @xmath29jy and 104 @xmath29jy for the f300w and f622w filters , respectively . the errors in these values have been estimated at 10% . a low resolution x - ray image , covering all of the radio emitting regions and with selected vla radio contours ( from prm ) overlaid , is shown in fig . 1 , while a full resolution x - ray image of the nucleus and structure to the west of the nucleus is shown in fig . the brightest x - ray source is associated with the nucleus of the galaxy . the position of this source , as measured with the chandra aspect system , is within 2@xmath18 of the radio and optical nuclear positions given by prm . the nucleus is so bright in x - rays that it suffers from strong `` pile up '' , so no useful spatial or spectroscopic information can be obtained from this exposure ; a chandra observation with a shorter frame time is scheduled to obtain a reliable nuclear spectrum . a linear feature , visible in both figures , extends 19 ( 110 kpc ) westward from the nucleus and is spatially coincident with the faint radio jet ( prm ) ; we shall refer to this feature as the `` x - ray jet '' . the x - ray jet `` points '' at the western hot spot , which is 42 ( 240 kpc ) from the nucleus . when the nuclear x - ray source is shifted slightly to coincide with the nuclear radio source , the peak of the x - ray emission from the hot spot is within 1@xmath18 of its radio peak . there is also evidence for faint , diffuse x - ray emission around the nucleus and extending @xmath16 1@xmath22 from it , but more sensitive observations are needed to confirm this component . faint x - ray emission extends eastwards from the nucleus to the two eastern radio hot spots , 31 and 34 from the nucleus ( fig . this faint x - ray emission is seen to brighten somewhat within @xmath16 1@xmath22 of these hot spots . there are also four compact x - ray sources within the brightest radio contour around the eastern hot spots ; none of these x - ray sources coincides with either radio hot spot . lastly , we point out a faint ( @xmath30 counts above the background ) , compact x - ray source 323@xmath18 e of the nucleus at @xmath13 = 05@xmath31 20@xmath32 20@xmath33.0 , @xmath34 = 45@xmath6 47@xmath22 44@xmath18 ( j2000 ) . this x - ray source is beyond the eastern radio lobe , but it lies in p.a . 101@xmath6 from the nucleus , a direction which is precisely opposite to the direction to the western hot spot ( p.a . 281@xmath6 ) . examination of the digital sky survey reveals a faint point source within @xmath16 1@xmath18 of this position , almost at the limit of the serc - j survey ( we estimate @xmath35 ) , and marginally detected in the second epoch southern ( uk schmidt ) survey . it is unclear whether this compact x - ray source is related to pictor a. images of the western hot spot at radio ( prm ) , optical ( rser 1989 ; prm ) and x - ray wavelengths with similar resolutions are shown in fig . 3 . as described in the introduction , the radio and optical hot spots have a very similar morphology . in a higher resolution ( 045 @xmath8 009 ) radio image ( fig . 19 of prm ) , the bright `` core '' is found to have a sharp leading ( i.e. away from the nucleus ) edge , with a slower decline on the side towards the nucleus . behind this bright region is a linear feature elongated in p.a . 32@xmath6 ( fig . 3 ) , referred to as the `` filament '' by rser & meisenheimer ( 1987 ) and the `` plateau '' by prm . the overall x - ray morphology is remarkably similar to the radio and optical , with both the `` core '' and the `` filament '' detected . much of the apparent elongation of the brighter part of the `` core '' of the x - ray hot spot ( fig . 3 ) reflects the shape of the point spread function ( psf ) . 4 compares profiles along the major and minor axes of the hot spot through both the hot spot and the model psf at the location of the hot spot . at half maximum , the hot spot is somewhat wider than the model , but the model psf does not include broadening by imperfections in the aspect solution . in order to assess the reliability of the model psf , we have compared a compact x - ray source ( visible in fig . 1 ) some 152@xmath18 from the nucleus in p.a . 40@xmath6 with the chandra model psf at this location . the observed profile of this source ( which is at a similar distance from the aim point as the hot spot ) is in good agreement with the model psf , but broader by 02 at the half power point . this number is similar to the excess broadening of the hot spot over the model psf along its minor axis ( fig . 4 ) , suggesting that the brighter part of the hot spot is unresolved in this direction . the excess fwhm of the hot spot over the psf along its major axis ( fig . 4 ) is somewhat larger . nevertheless , we feel that caution is necessary , since other effects ( e.g. the motion of the source on the detector during an observation ) could lead to further broadening compared with the model psf . the conservative conclusion is that the brightest part of the `` core '' is marginally resolved or unresolved in x - rays . there are , however , clear indications of structure in the hot spot at lower brightness levels . as in the radio , the x - ray core has a sharper edge away from the nucleus than towards it . a narrow `` ridge '' extends @xmath2 35 ( 3.4 kpc ) in p.a . 108@xmath6 ( the general direction towards the nucleus ) from the hot spot peak . a weaker feature some 6@xmath18 ( 5.7 kpc ) ssw of the core and elongated towards p.a . 279@xmath6 , is seen in both the radio and x - ray images . in conclusion , when the small differences in resolution of the x - ray and radio images ( fig . 3 ) are taken into account , the x - ray and radio morphologies are remarkably similar . this morphological similarity indicates that the x - ray , optical and radio - emitting regions are physically co - spatial and argues that the x - ray emission does not originate from a bow shock ahead of the radio hot spot . the x - ray spectrum of the hot spot has been modelled in various ways . a hot plasma with solar abundances and in collisional equilibrium ( `` raymond - smith '' model ) provides a poor fit to the data ( table 1 ) and can be ruled out . if the abundances of all elements heavier than helium are allowed to vary , we find that an abundance of @xmath36 10 per cent solar is needed for this thermal model to provide an acceptable description of the data . however , this model ( table 1 ) requires a hydrogen column well below the galactic value of n@xmath37 = 4.2 @xmath8 10@xmath38 atoms @xmath39 towards pictor a ( heiles & cleary 1979 ) , and is thus implausible . fixing the column at the galactic value while allowing the abundances to vary leads to a poor fit ( table 1 ) . further , we shall argue in section 3.2.3 that a thermal model can be excluded on grounds that the required gas density is so high that the radio emission would be faraday depolarized , contrary to observation . lastly , a power law of photon index @xmath0 = 2.07@xmath40 absorbed by a column density of solar abundance gas n@xmath37 = ( 7.1@xmath41 ) @xmath8 10@xmath38 atoms @xmath39 provides an excellent description of the spectrum ( table 1 , fig . 5 ) this column density is only slightly greater than the galactic column density towards pictor a , suggesting that the interstellar medium of our galaxy is responsible for most of the observed absorption . the unabsorbed flux and luminosity of the western hot spot in the 2 - 10 kev band are 3.1 @xmath8 10@xmath42 erg @xmath39 s@xmath10 and 1.7 @xmath8 10@xmath43 erg s@xmath10 , respectively . we first evaluate the thermal model . if we assume a representative intrinsic dimension of 1.9 @xmath8 0.9 @xmath8 0.9 kpc ( i.e. 2@xmath18 @xmath8 1@xmath18 @xmath8 1@xmath18 ) , as suggested for the bright `` core '' by the x - ray image ( figs 3 and 4 ) , the required gas density for the best fitting raymond - smith model is n@xmath44 @xmath2 3 @xmath45 . because the x - rays come from the same general region of space as the radio and optical emission ( fig . 3 ) , we may estimate the magnetic field in this putative thermal gas from the synchrotron emission . for an equipartition magnetic field of 4.7 @xmath8 10@xmath46 gauss ( obtained assuming a total cosmic ray energy equal to twice that of the electrons , @xmath13 = 0.74 [ meisenheimer , yates & rser 1997 ] , lower and upper cut - off frequencies 10@xmath47 and 10@xmath17 hz , respectively , and an emitting volume corresponding to the size of the radio core - 760 @xmath8 190 @xmath8 190 pc [ 08 @xmath8 02 @xmath8 02 ] ) , the rotation measure through the hot spot is expected to be @xmath2 6 @xmath8 10@xmath48 rad m@xmath49 . if the larger volume adopted above for the x - ray source is used to calculate the equipartition field , a value of 1.4 @xmath8 10@xmath46 gauss and a rotation measure through the hot spot of @xmath2 2 @xmath8 10@xmath48 rad m@xmath49 are found . in contrast , the high polarization observed at 6 cm implies a rotation measure internal to the hot spot @xmath12 900 rad m@xmath49 ( n@xmath44 @xmath12 0.014 @xmath45 ) , a factor of @xmath2 200 lower than required for thermal emission . if the 20 cm polarization is used , the upper limit on rotation measure and density are an order of magnitude smaller , but the hot spot is not well resolved at this wavelength . we thus conclude that our upper limit to the gas density renders a thermal model untenable . this conclusion could be wrong if either a ) the magnetic field has a preferred direction but many reversals ; such a field structure can provide the high observed synchrotron polarization but gives little faraday rotation , or b ) the thermal gas is actually in small , dense clumps with a small covering factor . we favor neither of these : the field structure of a ) is physically implausible , and it is very improbable that thermal gas of density 3 @xmath45 , let alone the higher density required if the gas is clumped , exists 240 kpc from the pictor a galaxy . another argument for a low gas density in the hot spot comes from the photoelectrically absorbing column . most or all of this column comes from gas in our galaxy ; any contribution from other regions is @xmath12 3 @xmath8 10@xmath38 atoms @xmath39 ( table 1 ) . if spread uniformly throughout the hot spot `` core '' of average diameter @xmath16 1 kpc , the density is @xmath12 0.1 @xmath45 ( as long as electrons are not stripped from the k shells of the relevant elements ) , a factor of 30 below the density needed for thermal emission . in summary , our finding ( section 3.2.2 ) that a thermal model provides a poor description of the x - ray spectrum , and the upper limits to the gas density from the limits on faraday depolarization and the absorbing column , rule out a thermal model of the x - rays from the hot spot we conclude that the x - rays are non - thermal in origin and discuss relevant models in section 4.2.2 . the x - ray jet extends 19 ( 110 kpc ) westward from the nucleus in p.a . 281@xmath6 ( fig . this direction is only 1@xmath6 away from that quoted for the radio jet out to 3@xmath22 ( 170 kpc ) from the nucleus ( prm ) . such a difference in p.a . is within the errors of measurement , so we can conclude that the x - ray and radio jets are coincident . the profile of x - ray emission along the jet , shown in fig . 6 , consists of a number of ` knots ' , the brightest of which is @xmath50 from the nucleus . in section 4.1 , we shall argue that the western radio lobe is the nearer . the ratio of the x - ray fluxes of the western and eastern jets , using the brightest part of the western jet , is more than a factor of 10 ( fig . 6 ) , suggesting relativistic boosting , as discussed in section 4.3.2 . we have compared the transverse profiles across the jet with the expected psf . after deconvolution from the psf , the jet is found to be transversely extended . the width of the jet varies somewhat along its length , but its typical fwhm = 20 ( 1.9 kpc ) . we are confident of this transverse extent because : a ) the profile of the compact source 152@xmath18 from the nucleus in p.a . 40@xmath6 agrees well with the expected psf at its location ( section 3.2.1 ) , and b ) the linear feature that results from the ccd readout of the strong nuclear source gives a measure of the 1d psf at the nucleus ; this feature is much narrower than the jet . these arguments show that the measured jet width is not a result of poor aspect solution or focus . the jet is so faint in the radio that it is difficult to measure its width . prm state `` no accurate estimate of the jet s width is possible , but it is clearly not greatly resolved by the 75 beam '' . thus the upper limit to the radio width is consistent with the x - ray width . given the low count rate , the background begins to dominate above 2.5 kev . the same models as used for the hot spot were tried for the jet ( table 2 ) . the raymond - smith thermal models provide a poor fit to the spectrum , even when the metal abundance is a free parameter . an absorbed power - law model provides an excellent description of the spectrum of the jet ( fig . the absorbing column density ( @xmath51 = 5.8@xmath52 @xmath8 10@xmath38 atoms @xmath39 ) and photon index ( @xmath0 = 1.94@xmath3 ) are similar to those found for the hot spot . again we conclude that most of the absorption occurs in the interstellar medium of our galaxy . the unabsorbed 2 - 10 kev flux and luminosity of the jet are 2.5 @xmath8 10@xmath53 erg @xmath39 s@xmath10 and 1.4 @xmath8 10@xmath54 erg s@xmath10 , respectively . even if the metal abundance is left as a free parameter , a raymond - smith thermal plasma model provides a poor fit to the jet s spectrum ( section 3.3.2 , table 2 ) . since there is no information on the radio polarization of the jet , we can not argue against a thermal model on grounds of internal faraday depolarization , as we did for the western hot spot . the x - ray absorbing column within the jet must be @xmath36 8.4 @xmath8 10@xmath38 atoms @xmath39 ( the maximum allowed column [ table 2 ] minus the galactic column ) , which implies a density n@xmath55 @xmath12 0.15 @xmath45 . for the observed x - ray emission to be thermal , the average density of hot gas in the jet ( taken to be a cylinder of radius 1@xmath18 and length 19 ) is n@xmath55 @xmath2 0.05 @xmath45 , compatible with the limits from the absorbing column . thus the only argument against thermal emission is the poor description of the spectrum by a raymond - smith model . we will discuss non - thermal models for the jet in section 4.3.2 . the projected distance , d@xmath56 , of the w radio hot spot of pictor a from the nucleus is larger than that , d@xmath55 , of the e radio hot spots . this observation is consistent with the w radio lobe being the nearer one , since the extra light travel time to the more distant lobe implies that it is seen at an earlier phase in its motion away from the nucleus and thus appears closer to the nucleus . further , the east lobe is more depolarized than the west at long radio wavelengths ( prm ) , again suggesting that the w lobe is the nearer ( laing 1988 ) . both the radio and x - ray jets of pictor a are seen on the w side of the nucleus . these considerations strongly suggest that the jet sidedness is a result of relativistic boosting . the angle of the source axis to the line of sight is , of course , unknown , but can be estimated from d@xmath56 and d@xmath55 if the source is assumed to be intrinsically symmetric , since @xmath57 where @xmath58 = v / c , v is the velocity with which the hot spots are receding from the nucleus and @xmath5 is the angle between the radio axis and our line of sight . a recent study ( arshakian & longair 2000 ) has found the mean value of @xmath58 in a sample of frii radio galaxies to be 0.11 @xmath1 0.013 . for pictor a , we may then write @xmath59 so @xmath5 = 23@xmath6 if v = 0.11 c. in view of its sensitivity to v and the assumption of symmetry , this value of @xmath5 should not be taken too seriously . the radio lobes are remarkably round and are well separated in the vla images ( prm ) , although joined by a ` waist ' near the nucleus . the overall radio extent in the plane of the sky is 430 kpc and the outward motions of the vlbi jet components ( tingay et al . 2000 ) are subluminal . all these properties indicate that @xmath5 is not small . a large value of @xmath5 is suggested by the fact that the hot spots project further away from the nucleus than the outer edges of the lobes themselves . tingay et al . ( 2000 ) have estimated that the vlbi - observed jet has @xmath5 @xmath12 51@xmath6 based on a jet - deflection model . 8 shows the broad band spectrum of the western hot spot . all of the radio points and most of the optical and near infrared ones were taken from meisenheimer , yates & rser ( 1997 ) . to these we added the flux densities at @xmath2 2900 and @xmath2 6200 obtained from our analysis of archival hst data ( section 2.2 ) . these hst values may underestimate the true flux densities as some of the emission is resolved out . also included is the original estimate of the x - ray flux density from the einstein observatory ( rser & meisenheimer 1987 ) and the chandra - measured intrinsic spectrum . the einstein point is a factor of @xmath16 2 higher than the chandra measurement ; this may be a result of partial blending of the hot spot with the much stronger nuclear source in the low spatial resolution einstein observation . the radio spectrum of the hot spot is well described by a power law with @xmath13 = 0.740 @xmath1 0.015 ( meisenheimer , yates & rser 1997 ) , but there must be a break or turnover in the spectrum at 10@xmath4 hz to accommodate the near infrared and optical measurements . it is also apparent that the x - ray spectrum is not a simple extension of the radio and optical measurements to higher frequencies . the luminosities of the radio to optical and the x - ray components are given in table 3 . it has long been realised that the outward motion of hot spots in frii radio galaxies is sub - relativistic . the most recent work ( arshakian & longair 2000 ) obtains a mean outward velocity of 0.11c @xmath1 0.013c . while the velocity of a hot spot in any individual object is uncertain , we shall neglect boosting or diminution of the hot spot fluxes by bulk relativistic motion . the broad - band spectrum ( fig . 8) shows that the x - ray emission of the western hot spot can not be synchrotron radiation from a high energy extension of the population of electrons responsible for the radio to optical synchrotron radiation . it is , therefore , natural to consider inverse compton scattering for the x - ray emission . in the bright core of the hot spot , the radiant energy density ( @xmath60 ) of the synchrotron radiation dominates that of the microwave background radiation and the galaxy starlight ( table 3 ) , so it is appropriate to consider a synchrotron self - compton model . when the scattering electrons follow a power - law distribution in @xmath61 (= e / mc@xmath62 ) , n(@xmath61 ) = n@xmath63 over all @xmath61 , then the inverse compton spectrum is a power law with a spectral index @xmath64 , the same index as the synchrotron spectrum . , where @xmath65 is the frequency of the synchrotron photon and @xmath66 is the rest mass energy of the electron . ] the fact that the radio ( @xmath67 = 0.740 @xmath1 0.015 ) and x - ray ( @xmath68 = 1.07@xmath40 ) spectral indices are different then suggests difficulties with an inverse compton model . however , in reality , the electron energy spectrum is a power law over only a certain range of @xmath61 ( from @xmath69 to @xmath70 ) , and the synchrotron spectrum is a power law over only a certain range of @xmath71 ( from @xmath72 to @xmath73 ) . these limited ranges yield `` end effects '' in synchrotron self - compton spectra : the inverse compton scattered spectra must turn down below @xmath74 and above @xmath75 ( e.g. blumenthal & gould 1970 ; rybicki & lightman 1979 ) . further , the synchrotron self - compton spectrum is no longer an exact power law between these limits . in order to provide a more realistic evaluation of inverse compton and synchrotron self - compton models , we have performed numerical calculations of spectra in spherical geometries using the computer code of band & grindlay ( 1985 , 1986 ) , which was kindly provided by dan harris . given the observed radio optical spectrum ( fig . 8) , we have assumed a power law electron spectrum with @xmath76 = 2.48 from @xmath69 to @xmath70 . the magnetic field is treated as a variable . we first computed a synchrotron self - compton spectrum which passes through the chandra - measured flux ( model 1 ) . this was achieved for a magnetic field strength of @xmath77 gauss , a factor of 14 below the equipartition field in the radio `` core '' of the hot spot ( table 4 ) . however , the predicted spectrum is similar to that of the radio source and does not match the measured x - ray spectrum ( fig . 8) . as an alternative to reducing the field strength , the power radiated in inverse compton radiation could be increased by increasing the radiation density . the nucleus of pictor a might emit a narrow , collimated beam of radiation ( like those inferred to be present in bl lac objects ) along the axis of the jet . this beam could illuminate the hot spot , but be invisible to us . however , an ( isotropic ) nuclear luminosity of 1.6 @xmath8 10@xmath78 erg s@xmath10 is needed merely to _ equal _ the radiant energy density in synchrotron radiation in the core of the hot spot . the ( isotropic ) nuclear luminosity would have to be @xmath2 1.5 @xmath8 10@xmath79 erg s@xmath10 to provide sufficient radiation for the hot spot to radiate the observed x - ray flux by inverse compton scattering if the field has its equipartition value . this luminosity is unreasonably large . an alternative is that the x - rays are a combination of synchrotron and inverse compton radiation . in model 2 , we assumed that the turnover in the synchrotron spectrum at 10@xmath4 hz is an effect of synchrotron losses on a continuously injected electron spectrum with energy index @xmath76 = 2.48 . the assumed electron spectrum is thus a broken power law ( kardashev 1962 ) : @xmath80 the synchrotron spectrum then has an index @xmath13 = 0.74 well below the break and @xmath13 = 0.74 + 0.5 = 1.24 is expected well above it ( dotted line in fig . 8) . by adding this spectrum to the predicted synchrotron self - compton component for a magnetic field of @xmath81 gauss ( dot - dashed line in fig . 8) , which is again well below equipartition , we obtain the solid line in fig . 8 , which is a good description of the chandra spectrum . the parameters of model 2 are given in table 4 . because the energy density in magnetic field is @xmath20 an order of magnitude larger than the synchrotron radiation density , the power in the first order synchrotron self - compton component is less than that in the synchrotron radiation ( e. g. rees 1967 ) further , the second order scattered component is even weaker ( 3 orders of magnitude below the first order radiation in the chandra band ) . assuming that the electrons are continuously accelerated in the hot spot , we may interpret the turnover frequency as the frequency at which the `` half - life '' to synchrotron losses of the radiating electrons is equal to their escape time from the hot spot . for the field of @xmath81 gauss needed to match the x - ray spectrum , electrons radiating at 10@xmath17 hz have a synchrotron `` half - life '' of @xmath2 10@xmath82 yrs . for a hot spot radius of @xmath2 250 pc , the corresponding streaming velocity of the relativistic electrons is @xmath2 0.1c . the problem with model 2 is the fine tuning needed : synchrotron and synchrotron self - compton emission must be present in the chandra band with comparable fluxes . in the previous subsection , we showed that the x - ray emission is so strong that the magnetic field in the hot spot must be a factor of @xmath16 14 below equipartition if the x - rays are produced by inverse compton scattering from the electrons that generate the _ observed _ radio emission . even then , the observed x - ray spectrum is different to the predicted one ( fig . 8) , and so this model ( model 1 , table 4 ) may be ruled out . the other process capable of generating the observed x - rays is synchrotron radiation . however , the x - ray emission of the hot spot is not a smooth continuation of the radio optical synchrotron spectrum . in model 2 , we contrived to fit the x - ray spectrum by a combination of synchrotron emission from an extension of the radio - optical spectrum and synchrotron self - compton from the radio - optical synchrotron - emitting electrons . in view of the unsatisfactory nature of these models , we now consider inverse compton and synchrotron models involving hypothetical new electron populations . _ i ) x - rays as inverse compton radiation _ for a successful inverse compton scattering model , a low energy population of relativistic electrons with an index of the energy spectrum p @xmath2 2@xmath83 + 1 is needed . the required value of p is thus p@xmath84 @xmath2 3.14@xmath85 . 9 shows a model ( model 3 , table 4 ) for the hot spot in which a synchrotron - emitting component with a larger spectral index than that observed in the radio has been added at low radio frequencies . the spectrum of the hot spot has not been measured below 327 mhz , but the extension of this hypothetical spectrum down to lower frequencies does not exceed the measured _ integrated _ flux density of pictor a ( the lowest such measurements are at @xmath16 80 mhz , see prm ) . it is envisaged that the population of electrons which emits the _ observed _ synchrotron radio emission is in a relatively strong ( e.g. equipartition , table 3 ) magnetic field , so that the synchrotron self- compton emission from this _ electron _ population is negligible , as shown above . however , the observed radio - optical synchrotron _ radiation _ , as well as the radio synchrotron radiation of the hypothesised component , is available for scattering by the hypothetical low energy electron population . as can be seen from fig . 9 , the low energy electrons must radiate predominantly through the inverse compton channel , requiring a very weak magnetic field ( model 3 , table 4 ) which is @xmath16 100 times weaker than the equipartition field in the core of the western hot spot . the calculation shows that the second order compton scattered component is @xmath20 2 orders of magnitude weaker than the first order compton scattered component in the chandra band . we find that the spectrum of the low energy electrons has to be steep ( p = 3.3 ) as shallower spectra ( p @xmath12 3.3 ) cause the inverse compton spectrum to increase above @xmath86 hz , contrary to our x - ray observations . taking this model to its extreme , the hypothetical electron population could be in a region free of magnetic field , and thus radiate no radio emission at all . this requirement for a weak or absent magnetic field associated with this population is difficult to understand in view of the finding that the x - ray , optical and radio emitting regions of the western hot spot are co - spatial ( section 3.2.1 and fig . high resolution radio observations at lower frequencies would provide stronger constraints on this model . _ ii ) x - rays as synchrotron radiation _ alternatively , the x - ray emission of the hot spot could be synchrotron radiation from a separate population of electrons . in such models ( models 4a and b ) , the energy index of the x - ray emitting electrons would be p@xmath84 @xmath2 3.14@xmath85 , similar to the above inverse compton model . in view of the short synchrotron loss times ( a few years for a 5 kev - emitting electron in the equipartition field ) , the energy index at injection would be p@xmath87 @xmath2 2.14@xmath85 in a steady state , continuous injection model . the synchrotron break frequency must thus be below the x - ray band but above 10@xmath88 hz ( otherwise the emission would exceed the observed flux at the latter frequency ) . models 4b and 4a ( table 4 ) , in which we have adopted the equipartition field , represent these two extremes for the synchrotron break frequency , respectively . thus , in both these models , the extrapolation of the x - ray spectrum to lower frequencies does not exceed the observed radio , infrared or optical flux . interestingly , the required value of p@xmath87 is the same to within the errors as the canonical index for particle acceleration by a strong shock ( p = 2 , e.g. bell 1978 ; blandford & ostriker 1978 ) . in this picture , electrons would have to be continuously reaccelerated by such shocks on pc scales . unfortunately , there is no information on the radio spectrum of the jet . we have only prm s statement that the average excess brightness of the jet above the background lobe emission is estimated to be about 10 mjy ( 75 beam)@xmath10 at @xmath89 20 cm . over the length of the detected x - ray jet , we then estimate a total 20 cm flux density of 152 mjy , with large uncertainties . the spectral index between 20 cm and 1 kev is @xmath90 = 0.87 , which is similar to @xmath91 = 0.85 , the spectral index for the _ entire _ radio source for @xmath71 @xmath92 400 mhz ( prm ) . this value of @xmath90 may also be compared with the index within the chandra band of @xmath93 = 0.94@xmath3 . given the errors , it is not out of the question that the jet has a constant spectral index from ghz frequencies to 10 kev , but the radio spectrum is needed to check this . simply joining the radio and x - ray flux densities with @xmath94 = 0.87 implies a total optical magnitude for the jet of v @xmath2 23 mag , and an optical surface brightness of v @xmath2 29 mag ( arc sec)@xmath49 . the fact that the jet has not been detected in the optical is consistent with these numbers . some estimated parameters of the jet are given in table 5 . it is apparent that the microwave background radiation dominates the radiant energy within the jet , unless the radius of the radio synchrotron - emitting region is much smaller than the observed radius of the x - ray jet . the x - rays from the jet could be either inverse compton or synchrotron radiation , which we discuss in turn . evaluation of these models is limited by our ignorance of the jet s radio spectrum . for concreteness , we shall adopt a jet radius of 1@xmath18 ( 950 pc ) for both the radio and x - ray emission . _ i ) x - rays as inverse compton radiation _ we consider first inverse compton models in which doppler boosting is unimportant , and show that such models of the jet s x - ray emission are implausible . we then consider models including doppler boosting . _ a ) insignificant doppler boosting by the jet _ adopting @xmath95 = 0.87 , s@xmath96 = 152 mjy for the part of the jet detected in x - rays ( section 4.3.1 ) , and the equipartition field h@xmath97 = 2.3 @xmath8 10@xmath98 gauss , the predicted inverse compton scattered x - ray flux falls a factor of @xmath2 500 below that observed . in order to match the observed x - ray flux , we need to reduce the magnetic field to h@xmath97 @xmath2 7 @xmath8 10@xmath99 gauss , a factor of @xmath2 30 below equipartition . alternatively , as for the hot spot , we could invoke a radiation beam from the nucleus to boost the radiation density in the jet . to obtain the observed x - ray flux from the jet while retaining the equipartition field requires an ( isotropic ) nuclear luminosity of @xmath16 2 @xmath8 10@xmath78 erg s@xmath10 , which is implausibly high . we could also suppose , as we did for the hot spot , that the _ observed _ radio jet contains a stronger ( e.g. equipartition ) magnetic field and there is , in addition , a population of electrons in a weak or absent field . because we do not know the radio spectrum of the jet , and have thus not been able to demonstrate that the radio and x - ray spectral indices are different , we can not prove that the radio- and x - ray - emitting electrons represent different populations , as we could for the hot spot . the following simple fact argues against inverse compton scattering for the jet s x - ray emission _ if doppler boosting is unimportant _ : in the radio band , the western lobe dominates the jet , while in x - rays the converse is true . the ratio of the rates of energy loss to inverse compton scattering and synchrotron radiation is @xmath100 the radiant energy density in both the jet and western lobe is dominated by the microwave background radiation . the equipartition magnetic field in the jet is @xmath2 2.3 @xmath8 10@xmath98 gauss ( for r@xmath101 = 1@xmath18 , see table 5 ) and that in the lobe is @xmath16 5 @xmath8 10@xmath102 gauss . thus one expects that the ratio ( de / dt)@xmath103/(de / dt)@xmath104 should be larger for the lobe than the jet . given that the lobe s radio synchrotron radiation overwhelmingly dominates that of the jet , the lobe s inverse compton emission should dominate the jet by an even larger factor , contrary to observation . however , the jet s inverse compton emission would be larger if either a ) its magnetic field is well below equipartition , or b ) a narrow beam of radiation is emitted by the nucleus along the jet , providing a larger @xmath60 than the microwave background , as discussed above . both of these possibilities are _ ad hoc _ and so we consider the prominence of the jet compared to the lobe in the chandra image as an argument against inverse compton scattering in the absence of doppler boosting . _ b ) significant doppler boosting by the jet _ the above discussion neglects relativistic boosting or diminution by possible bulk relativistic motion of the jet . there is now a strong case for energy transport at bulk relativistic velocities to the hot spots in frii sources ( e.g. bridle 1996 ) . for emission which is isotropic in the rest frame and has a power law spectrum , the observed flux density , @xmath105 , is related to the flux density in the rest frame , @xmath106 , by @xmath107 where @xmath108^{-1}$ ] , @xmath0 is the lorentz factor of the bulk flow , @xmath58 is the bulk velocity in units of the speed of light , and @xmath5 is the angle between the velocity vector and the line of sight ( e.g. urry & padovani 1995 ) . this equation assumes that the emission comes from a discrete , moving source . for a smooth , continuous jet , the exponent 3+@xmath13 becomes 2+@xmath13 ( begelman , blandford & rees 1984 ) . equation ( 6 ) will describe relativistic boosting or diminution of the jet s synchrotron radiation , as long as that radiation is isotropic in the jet s rest frame . as noted above ( table 5 ) , the radiation density in the jet is dominated by the microwave background radiation , which is isotropic in the observer s frame and anisotropic in the rest frame of the jet . in this case , the principal dependence of the inverse compton scattered radiation on @xmath5 is given by ( dermer 1995 ) @xmath109 where an additional term which depends slowly on @xmath110 has been omitted ( begelman & sikora 1987 ; dermer 1995 ) . retaining the discrete source model , the ratio of inverse compton scattered to synchrotron spectral flux density is then @xmath111 thus the ratio of inverse compton to synchrotron flux is increased by doppler boosting ( @xmath34 @xmath92 1 ) . our goal is to develop successful models for the jet s x - ray emission for various assumed angles @xmath5 . selection of a given @xmath5 defines a maximum value of @xmath34 ( @xmath112 @xmath2 1/@xmath5 ) , which is achieved when @xmath0 @xmath2 1/@xmath5 . assuming @xmath112 ( which is the most optimistic choice , since it minimises the required reduction of h below equipartition at a given @xmath5 ) , we have computed the value of h required to reproduce the observed x - ray to radio flux ratio , and compared it with the equipartition field which would be inferred from the synchrotron emission by an observer in the rest frame of the jet ( estimated by `` deboosting '' the observed radio flux ; h@xmath22 @xmath16 h@xmath113 ) . the results are given in table 6 . for our canonical @xmath5 = 23@xmath6 ( section 4.1 ) , the field must be at least a factor of 6 below equipartition . an equipartition model demands @xmath5 = 8@xmath6 , which is surely too small for a lobe - dominated frii radio galaxy like pictor a ( see discussion in section 4.1 ) . such an angle would make pictor a one of the largest known radio galaxies ( length 3 mpc ) . we conclude that inverse compton scattering of the microwave background is a viable model for the x - ray emission of the jet of pictor a , but requires a magnetic field which is substantially below equipartition . _ ii ) x - rays as synchrotron radiation _ as already noted , the available data are consistent with a single power law with @xmath90 = 0.87 from 1.4 ghz to 10 kev . this is , however , an unlikely physical situation for a purely synchrotron spectrum in view of the short energy loss times of the x - ray emitting electrons to synchrotron and inverse compton radiation . for electrons emitting synchrotron radiation at 1.4 ghz and 1 kev in the equipartition field of 2.3 @xmath8 10@xmath98 gauss , the times to lose half their energy are @xmath2 4 @xmath8 10@xmath114 and 300 yrs , respectively . if the energies of the radio - emitting electrons are not significantly reduced by synchrotron and inverse compton losses , we would expect the radio spectrum to be flatter than the x - ray spectrum ; in the simplest case involving continuous injection , @xmath95 = @xmath93 0.5 @xmath2 0.4 . it would thus be valuable to measure @xmath95 . for a synchrotron model in which the radiation is isotropic in the frame of bulk jet motion , the observed ratio of flux in the approaching side of the jet to the receding side is given by @xmath115 from x - ray observations @xmath116 and @xmath117 at the brightest part of the x - ray jet . assuming @xmath5 @xmath92 23@xmath6 ( section 4.1 ) , then @xmath58 @xmath92 0.3 . if the x - rays are the result of inverse compton scattering of the microwave background radiation , the exponent 3+@xmath13 in equation ( 9 ) becomes 4 + 2@xmath13 ( cf . equation 7 ) and a somewhat smaller lower limit to @xmath58 is obtained . our chandra study of pictor a has shown that the x - ray emissions from the jet and the western hot spot are non - thermal . the spectra of both are well described by an absorbed power law with ( flux density ) spectral index @xmath13 @xmath2 1.0 . the x - ray spectrum of the hot spot is not a smooth extension of the radio to optical synchrotron spectrum , which turns down or cuts off near @xmath16 10@xmath17 hz . inverse compton scattering of the synchrotron radio photons by the relativistic electrons responsible for the radio emission ( i.e. a synchrotron self - compton model ) may be ruled out for the hot spot s x - ray emission , as the predicted spectrum differs from that observed . we considered the possible existence of a population of relativistic electrons in the hot spot that radiates synchrotron emission at frequencies below that at which its spectrum has been measured ( i.e. @xmath12 327 mhz ) . by choosing an appropriate index for the energy spectrum of these electrons , we constructed a successful synchrotron self - compton model for the x - rays at the price of reducing the magnetic field to @xmath16 1% of equipartition ( see fig . 9 and model 3 in table 4 ) . more generally , relativistic electrons could exist in regions with weak or absent magnetic fields . since the properties of such electron populations are unconstrained by radio observations , one can always create inverse compton models that match the x - ray spectra . it is , however , then difficult to understand why the x - rays from both the jet and the hot spot should correlate so well with synchrotron radio emission , which must arise in a relatively strong magnetic field . the x - ray spectrum of the hot spot may be reproduced in a composite synchrotron plus synchrotron self - compton model . in this picture , the spectral index of the synchrotron radiation is supposed to increase by 0.5 above the break near 10@xmath17 hz , as would be expected in a continuous injection model . addition of this synchrotron emission to the synchrotron self - compton emission expected for a magnetic field a factor of 9 below equipartition reproduces the observed spectrum ( see fig . 8 and model 2 , table 4 ) . the model is contrived , requiring similar fluxes from the two components in the chandra band , but can not be ruled out . if the jet is non - relativistic , inverse compton scattering is an implausible model for its x - ray emission , for it requires a magnetic field a factor of 30 below equipartition . further , it is hard to understand why the jet is brighter than the lobe ( the opposite of the situation in the radio ) in such a model . if the jet is relativistic , these difficulties are eased , and we consider inverse compton scattering by such a jet off the microwave background a viable mechanism . however , the magnetic field must still be well below equipartition for plausible angles between the jet and the line of sight in this lobe - dominated , frii radio galaxy . synchrotron radiation is a plausible x - ray emission process for both jet and hot spot . strong , non - relativistic shocks are believed to accelerate relativistic particles yielding an energy spectrum n(e ) @xmath14 e@xmath118 with p = 2 at injection . since the half lives of x - ray emitting electrons to synchrotron losses are very short ( @xmath16 years ) , the spectrum steepens and a synchrotron spectral index of @xmath13 = 1.0 is expected , in excellent accord with observations ( models 4a , b , table 4 ) . the separate population of radio - optical synchrotron emitting electrons remains unexplained ; the radio spectral index - @xmath119 = 0.740 @xmath1 0.015 - is close to the average for non - thermal radio sources . various processes , including acceleration in weak shocks , synchrotron losses and effects of tangled fields ( see summary in longair 1994 , ch . 21 ) , have been invoked to account for the difference between the typical index seen in radio sources and the value @xmath13 = 0.5 expected in the canonical model . hot spots are , of course , associated with two shocks - an internal , mildly relativistic shock in the jet and a non - relativistic bow shock in the intergalactic medium . both may reasonably be expected to accelerate cosmic rays and the resulting fluxes and energy spectra may differ . as well as being directly accelerated in shocks , high @xmath61 relativistic electrons may result from a ` proton induced cascade ' initiated by photopion production ( e.g. sikora et al . 1987 ; biermann & strittmatter 1987 ; mannheim & biermann 1989 ; mannheim , krlls & biermann 1991 ) . in this process , relativistic proton - photon collisions create @xmath120 , @xmath121 and @xmath122 . the last two decay into relativistic electrons , positrons , neutrinos and antineutrinos ( e.g. biermann & strittmatter 1987 ) . the process thus provides a supply of synchrotron x- and @xmath61-ray emitting electrons and positrons from high energy protons accelerated by shocks . mannheim , krlls & biermann ( 1991 ) considered this process for production of synchrotron x - ray emission in radio galaxy hot spots . their calculations suggest that the proton induced cascade produces x - ray emission about an order of magnitude weaker and with a harder spectrum than is observed for the western hot spot of pictor a. as they note , the predicted luminosity can be increased by increasing the number of relativistic protons or the number of photons ( the latter might occur , for example , if the hot spot is illuminated by a beam of radiation from the galaxy nucleus ) . this process seems promising and further calculations of the expected x - ray luminosity and spectrum of a proton induced cascade would be worthwhile . our discussion of the western hot spot has assumed it moves outwards non - relativistically . while this is established for the population of radio hot spots in general , it is not necessarily true for pictor a. the fact that the western hot spot is on the near side and is much brighter than the eastern hot spots at both radio and x - ray wavelengths raises the possibility of relativistic outflow . some of the problems with inverse compton models of the hot spot might then be eased ( as they are for the jet ) , but the difference between the radio and x - ray spectral indices remains a difficulty . this research was supported by nasa through grant nag 81027 and by the graduate school of the university of maryland through a research fellowship to asw . we are extremely grateful to rick perley for providing the radio images published by prm in numerical form . we also wish to thank the staff of the chandra science center , especially dan harris and shanil virani , for their help . bridle , a. h. 1996 , in energy transport in radio galaxies and quasars , asp conference series vol . 100 , eds . e. hardee , a. h. bridle and j. a. zensus , p 383 ( astronomical society of the pacific : san francisco ) raymond - smith & @xmath123 & @xmath124 & solar & @xmath125 & & & 166 / 97 + raymond - smith & 4.2 ( frozen ) & @xmath126 & solar & @xmath127 & & & 232 / 99 + raymond - smith & @xmath128 & @xmath129 & @xmath130 solar & @xmath131 & & & 101 / 96 + raymond - smith & 4.2 ( frozen ) & @xmath132 & @xmath133 solar & @xmath134 & & & 122 / 98 + power law & 7.1@xmath41 & & & & 2.07@xmath40 & ( 1.34@xmath135 ) @xmath8 10@xmath46 & 95 / 98 + lcccccccc raymond - smith & 4.2 ( frozen ) & @xmath136 & solar & @xmath137 & & & 27 / 15 + raymond - smith & 4.2 ( frozen ) & @xmath138 & @xmath139 solar & @xmath140 & & & 19 / 14 + power law & 5.8@xmath52 & & & & 1.94@xmath3 & ( 1.45@xmath141 ) @xmath8 10@xmath98 & 15 / 14 + lc l(327 mhz - optical ) & @xmath2 3 @xmath8 10@xmath142 erg s@xmath10 + l(0.1 - 10 kev ) & 4.9 @xmath8 10@xmath43 erg s@xmath10 + h@xmath143 ( bright core ) & 4.7 @xmath8 10@xmath46 gauss + h@xmath143 ( filament ) & 9.1 @xmath8 10@xmath98 gauss + @xmath60 ( core , synchrotron radn ) & @xmath16 8 @xmath8 10@xmath144 erg @xmath45 + @xmath60 ( filament , synchrotron radn ) & @xmath16 1 @xmath8 10@xmath42 erg @xmath45 + @xmath60 ( microwave background ) & 4 @xmath8 10@xmath42 erg @xmath45 + @xmath60 ( galaxy optical light ) & 4 @xmath8 10@xmath145 erg @xmath45 + 1 & 5.3 & 2.48 & @xmath146 & & & @xmath147 & @xmath77 & 0.07 + 2 & 2.2 & 2.48 & @xmath146 & @xmath148 & & @xmath147 & @xmath81 & 0.11 + 3 & 485.0 & 3.30 & @xmath149 & & @xmath150 & @xmath147 & @xmath151 & 0.01 + 4a & @xmath152 & 2.14 & & @xmath153 & & @xmath147 & @xmath154 & 1 + 4b & @xmath155 & 2.14 & & @xmath156 & & @xmath147 & @xmath154 & 1 + lc h@xmath143 & 2.3 @xmath8 10@xmath98 ( r@xmath157/1 arc sec)@xmath158 gauss + @xmath60 ( synchrotron radn)@xmath159 & @xmath16 1 @xmath8 10@xmath160 ( r@xmath157/1 arc sec)@xmath10 erg @xmath45 + @xmath60 ( microwave background ) & 4 @xmath8 10@xmath42 erg @xmath45 + @xmath60 ( galaxy optical light ) & @xmath16 8 @xmath8 10@xmath160 ( r/1 arc min)@xmath49 erg @xmath45 + lcccccccc any value@xmath161 & 1 & 1 & 7@xmath810@xmath99 & 2.3@xmath810@xmath98 & 0.03 + 23@xmath6 & 2.6 & 2.6 & 2@xmath810@xmath102 & 1.2@xmath810@xmath98 & 0.16 + 8@xmath6 & 7.2 & 7.2 & 6@xmath810@xmath102 & 6@xmath810@xmath102 & 1 +

we investigate the properties of conduction electrons in single - walled armchair carbon nanotubes ( swnt ) in the presence of both transverse electric and magnetic fields . we find that these fields provide a controlled means of tuning low - energy band structure properties such as inducing gaps in the spectrum , breaking various symmetries and altering the fermi velocities . we show that the fields can strongly affect electron - electron interaction , yielding tunable luttinger liquid physics , the possibility of spin - charge - band separation , and a competition between spin - density - wave and charge - density - wave order . for short tubes , the fields can alter boundary conditions and associated single - particle level spacings as well as quantum dot behavior .

0812.1851

the astounding range of experimental and theoretical studies performed on carbon nanotubes @xcite has revealed a spectrum of physics characteristic of strongly correlated low - dimensional electronic systems @xcite . the underlying graphene lattice structure of these tubes uniquely affects band structure , effective dimensionality , coulomb interaction effects , and the quantum dot behavior exhibited by short nanotube segments . the band structure shows differing behavior depending on various factors such as chirality , applied gate potentials , boundary conditions at the tube ends and mechanical stress @xcite . in single - walled armchair nanotubes ( swnt ) @xcite , which are the entities of interest here , gapless linearly dispersing modes endow the nanotube with its peculiar quantum wire properties . as described theoretically and ascertained experimentally , interactions within the modes of this effectively one - dimensional system cause it to behave as a luttinger liquid characterized by non - ohmic conductances @xcite . tubes placed between tunnel barriers act as quantum dots @xcite , which , while displaying zero dimensional physics such as coulomb blockade behavior , retain some higher dimensional traits such as hosting plasmons typical of one dimension and band degrees of freedom attributed to the underlying graphene lattice . potentially invaluable to applications , these nanotube quantum dots have been proposed as elements of quantum devices and the quantum states of blockaded electrons have been regarded as candidates for units of quantum information @xcite . in each of these aspects , the presence of applied fields can dramatically alter the nanotube s behavior ; here we present an extensive study of the effects of electric and magnetic fields applied transversally to the axis of the nanotube . at the level of the band structure , it has been shown that a parallel magnetic field can have the striking effect of converting a metallic tube to a semiconducting one by way of inducing a gap @xcite , and vice - versa , an effect discernible in conductance , coulomb blockade and scanning tunneling microscope ( stm ) measurements . here , instead of a parallel field , we discuss transverse field configurations ( both electric and magnetic ) and the conditions under which a band gap opens up or the spectrum remains gapless in armchair swnts . in the latter case , we demonstrate , via band - structure calculations , simultaneous breaking of the valley degeneracy ( of the two distinct dirac points ) , the left - right - mover degeneracy , and the particle - hole symmetry . moreover , the fields yield a non - negligible reduction in the fermi velocity of conduction electrons traveling along the tube . we show that for certain configurations of fields , the ground state of the tube can even be made to carry finite current . transverse fields provide an excellent means of altering the ratio of interaction strength to the fermi energy in swnts . this makes nanotubes potentially the only systems to date in which the associated luttinger liquid physics can be tuned in a controlled fashion . as described in previous work , either an electric field @xcite or a magnetic field @xcite alone suffices to change the value of the luttinger liquid parameter from that measured in field - free environments . the magnitude of the electric fields required to bring about a significant change are well within current experimental reach @xcite . here , we find that our approach reproduces these results . we show that such a tuning of luttinger parameters can mediate a transition from the system showing tendencies towards spin - density - wave ( sdw ) ordering to that of charge - density - wave ( cdw ) ordering . furthermore , in addition to the tuning of the luttinger parameter presented in previous works for the net charge density @xcite , we find that luttinger - type interactions become manifest in modes associated with the density differences between nanotube bands as well . thus we predict that akin to spin - charge separation , transverse fields can induce a spin - charge - band separation wherein the three degrees of freedom move at different velocities . the above results are discussed in the case of an infinite system . for short tubes or finite length segments formed by tunnel barriers , boundary effects need to be taken into account . we find that applied fields influence multiple aspects of short nanotubes . first , fields can alter the single - particle energy level spacing of the tube . here , we carefully account for the effect of the tube ends in the case that the left and right movers travel at different speeds . second , the charging energies become field dependent , and third , the plasmon spectrum varies in accordance with the first two effects . we have conducted a comprehensive analysis of the short nanotube as a finite - sized luttinger liquid and show how all three effects can be captured . within this description , we discuss the structure of field - dependent coulomb blockade peaks and how the presence of both electric and magnetic fields acts as a means of manipulating quantum states of the dot ( the effect of a magnetic field alone has recently been discussed by bellucci and onorato @xcite ) . ) and electric fields . the carbon atoms belonging to the a and b sublattices are indicated by dark ( blue ) and light ( green ) shading , respectively.,width=8 ] the outline of this paper is as follows . in section ii we present the formulation and results of our band structure calculation . in section iii we formulate an effective one - dimensional hamiltonian which takes into account field effects . we bosonize this hamiltonian , describing interaction effects in terms of luttinger liquid physics . in section iv we investigate the various luttinger liquid phases and the feasibility of using fields to access such phases . in section v we discuss field - tuned quantum dot physics . finally , in section vi we present the highlights of our results and discuss their relevance to experiments . we briefly recapitulate the band structure of an infinitely long armchair tube in the absence of any fields @xcite . the electronic properties of graphene are well described by a tight - binding model in which electrons hop between nearest neighbors of the underlying hexagonal bipartite lattice ( sublattices here labeled @xmath0 and @xmath1 ) with an associated energy ( the hopping integral ) of @xmath2 ev . an armchair carbon nanotube can be regarded as a sheet of graphene rolled along the @xmath3-direction ( for notation of the chirality , see e.g. ref . @xcite ) , denoted by @xmath4 . this gives rise to states of quantized momentum @xmath5 where @xmath6 is the circumference of the tube , and @xmath7 where @xmath8 nm is the nearest carbon - carbon distance . the resulting series of one dimensional bands can be described by the wavevectors @xmath9 where @xmath10 is the quasimomentum parallel to the tube s axis and @xmath11 is the state s angular momentum about the tube s circumference . a convenient set of basis states is given by the following linear combination of atomic orbitals @xmath12 where @xmath13 is the @xmath14-electronic state of the atom located at @xmath15 . the sum runs over the @xmath16 atoms in the unit cell that belong to either the a or b sublattice . at half filling , the associated dispersion has low energy excitations near the so - called dirac points of the form @xmath17 where @xmath18 m / s , @xmath19 and @xmath20 . thus , @xmath21 label the two inequivalent fermi points . the setup of interest is shown in fig . [ fig : setup ] . an external magnetic field is applied in the negative @xmath22-direction ; an applied ( transverse ) electric field makes an angle @xmath23 with the magnetic field . these fields give rise to scalar and vector potentials @xmath24 respectively ( @xmath25 ) . the @xmath26-axis runs parallel to the tube s axis and the additional coordinate @xmath27 measures the circumferential distance starting from the negative @xmath22-axis ( a positive value of @xmath27 corresponds to a counterclockwise rotation as one looks along the @xmath26-axis in the positive direction ) . these external potentials are easily accommodated within the tight binding approach . in eq . ( [ eq : mag ] ) we have selected a gauge that is independent of @xmath26 and thus @xmath10 remains a good quantum number . the hopping matrix elements in the presence of the fields are given by @xmath28 where the sum runs over nearest neighbors @xmath15 , @xmath29 and @xmath30 is the aharonov - bohm phase associated with the magnetic field @xcite . the dimensionless parameter @xmath31 is given by @xmath32 . numerically , for an @xmath3 nanotube , the magnetic field in teslas is related to the dimensionless parameter @xmath31 via @xmath33 t. the scalar potential gives rise to an on - site potential described by the matrix element @xmath34 for @xmath35 where @xmath36 . the electric field strength in v / nm is related to @xmath37 by @xmath38 v / nm for a tube with chiral vector @xmath3 . that these matrix elements mix states of different angular momentum has a straightforward classical analog : a charged particle on the surface of a cylinder will circulate around its circumference as a result of the applied fields we have studied the effects of the fields perturbatively in @xmath31 and @xmath37 . in the vicinity of the fermi points , the left and right moving bands are nearly degenerate , so care must be taken in applying perturbation theory . the details of this calculation are presented in appendix [ app : band ] where we carry out perturbation theory to second order . there are three cases of interest that we summarize below . an illustration of these three cases is shown in figs . [ fig : spectrume ] , [ fig : spectrumb ] , [ fig : spectrumvbn ] , and [ fig : spectrumvb ] . while some of the effects discussed are rather small for standard swnts , we note that our band structure analysis can be applied to multi - walled nanotubes as well in which case a larger radius yields more pronounced effects . carbon nanotube near the @xmath39 dirac fermi point ( field - free value @xmath40 indicated by the vertical line ) in the presence of an external perpendicular field with @xmath41 ( black dotted ) , @xmath42 ( red dot - dashed ) , @xmath43 ( blue solid ) as the crossing moves to the right . the horizontal axis indicates the value of @xmath44 , where @xmath45 ; the vertical axis is given in units of @xmath46 , the hopping integral ( @xmath47 ev ) . [ fig : spectrume],width=8 ] carbon nanotube near the @xmath39 dirac fermi point ( field - free value @xmath40 indicated by the vertical line ) in the presence of a magnetic field @xmath48 ( black dotted ) , @xmath42 ( red dot - dashed ) , @xmath43 ( blue solid ) ( as the crossings move to the right ) . the horizontal axis indicates the value of @xmath44 , where @xmath45 ; the vertical axis is given in units of @xmath46 , the hopping integral ( @xmath47 ev ) . [ fig : spectrumb],width=8 ] for a single field , the most salient features of our band structure calculation are the reduction in fermi velocity and the shift in fermi momentum . semiclassically , the reduction in fermi velocity can be ascribed to the deflection of the electrons by the fields leading to a reduction in the velocity component along the tube . furthermore the bands remain gapless . for a magnetic field , we find a reduced fermi velocity given by @xmath51 the first order correction in @xmath31 vanishes because @xmath52 must be an even function of @xmath31 by symmetry . the term @xmath53 is a function of @xmath16 and is given by eq . ( [ eq : v1 ] ) . this term depends on the geometric details of an armchair tube ; for large tubes @xmath54 . an experimental observation of such a reduction will require strong fields . for example , a ( 20,20 ) armchair tube with a 20 t field gives a @xmath55 reduction in the fermi velocity . for an electric field we have @xmath56 a field of strength @xmath57 v / nm corresponds to a reduction in the fermi velocity of a ( 10,10 ) tube of roughly @xmath58 . the term @xmath59 for large tubes ; its exact form is given by eq . ( [ eq : v2 ] ) . as mentioned in the introduction , the reduction of the fermi velocity has been noted by several authors @xcite . in these papers the carbon nanotube was modeled as a smooth cylinder and the low - energy electronic behavior was put in by hand . our results are in agreement with these results in the limit of small fields and large @xmath16 ( in the regime that perturbation is valid ) . additionally , by taking into account the geometry of the armchair nanotube we find that either a magnetic or electric field alone will shift the fermi points . that is , the nanotube still has the same low energy spectrum with renormalized values of @xmath60 ( whose precise form is given by eq . ( [ eq : kf ] ) ) . the band structure of a @xmath61 nanotube in the presence of electric and magnetic fields of various strength is shown in figs . [ fig : spectrume ] and [ fig : spectrumb ] , respectively . carbon nanotube near @xmath62 dirac fermi point ( field - free value @xmath63 indicated by the vertical dashed line ) in the presence of transverse electric and magnetic field ( @xmath64 and @xmath65 ) . the angle between @xmath66 and @xmath67 being @xmath68 ( black dotted ) , @xmath69 ( red dot - dashed ) , @xmath70 ( blue solid ) ( from outer to inner ) . the horizontal axis indicates the value of @xmath44 , where @xmath45 ; the vertical axis is given in units of @xmath46 , the hopping integral ( @xmath47 ev ) . [ fig : spectrumvbn],width=8 ] carbon nanotube near @xmath39 dirac fermi point ( field - free value @xmath71 by the vertical line ) in the presence of transverse electric and magnetic field ( @xmath64 and @xmath65 ) . the angle between @xmath66 and @xmath67 being @xmath68 ( black dotted ) , @xmath69 ( red dot - dashed ) , @xmath70 ( blue solid ) ( from outer to inner ) . the horizontal axis indicates the value of @xmath44 , where @xmath45 ; the vertical axis is given in units of @xmath46 , the hopping integral ( @xmath47 ev ) . [ fig : spectrumvb],width=8 ] mutually perpendicular fields break both the time - reversal and particle - hole symmetry of the band structure . the left and right movers now move with different speeds . for a magnetic field in the negative @xmath22-direction and an electric field in the positive @xmath74-direction we have @xmath75 where @xmath76 is @xmath77 for right movers and @xmath78 for left movers . the expressions for @xmath79 is given by eq . ( [ eq : v3 ] ) ; for large tubes @xmath80 . for the fields we consider , @xmath81 is roughly the same order of magnitude as @xmath82 and thus it is natural to expect the band structure will mimic the behavior of a classical velocity selector . indeed the asymmetry of the velocity in the right- and left - moving branches is expected from a elementary consideration . for a charge particle moving on a smooth cylinder in the presence of mutually perpendicular transverse magnetic and electric fields , the direction of the force caused by the magnetic field depends on whether the particle is moving along the tube in one direction or another , whereas the electric field remains the same . hence , for fixed kinetic energy , the force in the transverse direction causing the particle to spiral is different for different directions of motion , and giving rise to different velocities along the axial direction . another prominent feature of the band structure in this case is a relative energy shift with respect to the two fermi points . near the two dirac points @xmath83 we have @xmath84 where @xmath85 ( @xmath20 for @xmath86 ) . by definition , @xmath87 is the momentum for which the left and right moving bands for a given fermi point are degenerate and again is generally different from @xmath88 in the presence of fields . the precise forms of @xmath60 and @xmath89 are given by eqs . ( [ eq : kf ] ) and ( [ eq : ds ] ) , respectively . the solid blue lines in figs . [ fig : spectrumvbn ] ( for @xmath90 ) and [ fig : spectrumvb ] ( @xmath91 ) indicate low - energy dispersion of a @xmath61 nanotube in the presence of crossed electric ( @xmath92 ) and magnetic ( @xmath65 ) fields . the presence of both electric and magnetic fields will generically open gaps at the fermi points . the size of this gap is given by @xmath94 for example , a ( 15,15 ) tube parallel electric and magnetic fields 1 v / nm and 10 t respectively gives @xmath95 mev . for electric and magnetic fields which are not parallel , this gap is weakly indirect . the degeneracy associated with the graphene fermi points stems from the equivalence of the two sublattices . for example , a two - dimensional honeycomb lattice with a and b sublattices composed of different types of atoms would generically have a gap @xcite . now , in a nanotube , gaps can arise for different reasons . for example , the gap associated with a semiconducting tube occurs because the quantized bands miss the dirac point of the underlying graphene lattice . such a gap can be closed by applying a magnetic field ( of a specific strength ) parallel to the tube s axis . in the present case however , the gap arises from an energy difference associated with the two sublattices and therefore can not be so closed . not surprisingly , the gap described here vanishes precisely when the electric and magnetic forces on a classical charged particle traveling along the tube are either parallel or antiparallel . from figs . [ fig : spectrumvbn ] and [ fig : spectrumvb ] we clearly see that as the angle between the fields varies , the particle - hole symmetry and the valley degeneracy , as well as the symmetry between left- and right - moving spectra , are broken . the degree of breaking of these symmetries is greatest in the case that the fields are perpendicular . on the other hand , these symmetries are preserved when the fields are parallel or anti - parallel and the gap is at its maximum . this angle - dependent gap can be manifest in the transport measurement , e.g. , the conductance can be @xmath96 , or @xmath97 or zero , depending on whether there are @xmath98 , @xmath99 or zero conducting channels ( including spin ) . in order to study the low - energy physics of electronic excitations about the fermi energy for all the cases described above , we can identify the fermionic operator on the cylindrical surface of the tube as approximately @xmath100 where @xmath101 is the one - dimensional field operator at the point @xmath26 along the tube axis associated with the a and b sublattices ( @xmath102 ) , fermi points @xmath40 ( @xmath20 ) , and spin @xmath103 . the bloch wavefunction @xmath104 retains detailed information about the response of the electronic wave functions to the applied fields @xcite . while the sublattice basis ( indexed by @xmath105 ) is convenient for discussing these wavefunctions , it does not diagonalize the hopping interaction . we therefore transform to the right and left moving basis ( @xmath106 ) via the transformation @xmath107 where @xmath108 @xcite . this gives a kinetic energy term @xmath109 where , as shown above a generic field configuration can give rise to the possibility that @xmath110 . the gap which arises as a result of the presence of both electric and magnetic fields [ eq . ( [ eq : gap ] ) ] can be incorporated into our hamiltonian by a mass term @xmath111 note that this mass term is of a different form than that considered in the exhaustive zero - field study by egger et al.@xcite ( which had the form @xmath112 ) . similarly , the effect of the shift associated with the fermi points can be described by the term @xmath113 in this section we specialize to the case of gapless modes for which interactions can be easily incorporated . as the simplest case , when only a magnetic or electric field is present , the kinetic piece of the hamiltonian given in eq . ( [ eq : hkinetic ] ) now has @xmath114 and @xmath115 , i.e. , @xmath116 we use standard approaches such as bosonization to study the effect of fields . where appropriate , we include discussions for the asymmetric case of @xmath117 . we closely follow the approach of ref . @xcite whose lucid pedagogical exposition we do not repeat but instead confine our discussion to the new field - dependant features . the presence of fields does not alter the fact that the electrons on a tube are locked into their lowest energy radial mode . hence , it is possible to study the low energy excitations in the presence of interactions using an effective bosonized 1d hamiltonian . bosonization offers a great simplification since many of the quartic interaction terms in the fermionic language become quadratic once they are bosonized . the interaction term takes the general form @xmath118 where @xmath119 is the field in eq . ( [ eq : wvfn ] ) describing low - energy electrons . following @xcite , we employ the form of the coulomb interaction on the surface of a cylinder given by @xmath120 where @xmath121 is the radius of the tube and @xmath26 and @xmath27 the coordinates defined in the previous section and @xmath122 is roughly the thickness of the graphene sheet @xcite . the form of the interaction in eq . ( [ eq : hint1 ] ) is explicitly given in terms of two - dimensional integrals . an effective interaction term involving purely linear integrals along the tube s axis can be obtained by expressing @xmath119 in terms of the linear and circumferential fields as in eq . ( [ eq : wvfn ] ) and integrating out the circumferential degrees of freedom from eq . ( [ eq : hint1 ] ) . here we remark that integration over the circumferential degrees of freedom generates field - induced terms that give rise to novel physics and can be traced to the dependence of the ground state wave function on non - zero angular momentum states in the circumferential direction . the resulting effective interaction involves two - particle scattering processes between fermions moving along the tube axis denoted by the fields @xmath123 . the associated scattering processes can be classified according to whether the incoming particles preserve their fermi point quantum number @xmath85 when scattered , - forward scattering ( @xmath124)- , or scatter across the fermi surface - backscattering ( @xmath125 ) . in ref . @xcite , a further distinction is made in the forward scattering events depending on whether the interaction potential is homogeneous over the circumference of the tube ( @xmath126 ) or is able to distinguish the microscopic differences which arise between sublattices ( @xmath127 ) . at this point , in order to obtain an effective low - energy description of the interacting electrons , the one - dimensional fermionic operators can be bosonized as in ref . @xcite : @xmath128.\ ] ] where @xmath129 the bosonic fields @xmath130 s satisfy the commutation relations @xmath131=-i\pi r\delta_{rr'}\delta_{\sigma\sigma'}{\rm sgn}(x - x'),\ ] ] where @xmath132 denotes the left- and right - movers , @xmath133 indicates the fermi points , and @xmath134 represents the spin direction ( @xmath135 ) . the @xmath136 s are the so - called klein factors ; they enforce the anticommutation relations between different channels , @xmath137 moreover , the fields @xmath138 s ( with @xmath139 and @xmath140 ) and their dual fields @xmath141 [ both are linear combinations of @xmath142 in turn satisfy @xmath143=-\frac{i}{2 } \delta_{jj'}\delta_{\delta\delta'}{\rm sgn}(x - x')\ ] ] the effective density in a given channel takes the form @xmath144 the kinetic energy associated with the linearly dispersing fermionic modes is quadratic in the bosonized fields . as for the interactions , the dominant contributions also come from quadratic terms reflecting net density - density type interactions . as in the field - free case , the @xmath126 process has one such contribution from the usual coulomb form involving the net charge density , which in the bosonized representation is given by @xmath145 where @xmath146 is the fourier transform of @xmath147 and @xmath148 is a function of @xmath16 ( see eq . ( [ eq : c ] ) ) . the presence of either an electric or magnetic field gives rise to additional quadratic terms in the @xmath126 process . these terms have their origin in the non - zero angular momentum components of the circumferential wavefunction @xmath149 in eq . ( [ eq : wvfn ] ) . a detailed accounting of the radial wave functions ( see appendix b ) shows that an electric field contributes a term @xmath150 whereas a magnetic field provides a contribution @xmath151 in the presence of mutually perpendicular transverse electric and magnetic fields , there is an additional contribution to the interaction @xmath152 the values of @xmath153 , @xmath154 , and @xmath155 are given in appendix b. the luttinger liquid hamiltonian takes into account all these quadratic terms ; other terms emerging from the interaction potential are sub - dominant and can be considered perturbatively . an intuitive physical picture for the origin of the field - dependant interaction terms above can be obtained by noting that in eq . ( [ eq : rhora ] ) the charge density for @xmath156 is of the form @xmath157 . this is broadly consistent with the magnetic field coupling to momentum via @xmath158 and hence to @xmath159 , resulting in a term @xmath160 . moreover , the electric potential differs slightly between adjacent @xmath0 and @xmath1 sites , and in turn couples to left- and right - movers differently , resulting in a term @xmath161 . equation ( [ eq : hint1 ] ) thus gives rise to interactions among the charge densities @xmath162 and current @xmath163 . we therefore obtain the following nonvanishing terms : ( 1 ) @xmath164 , ( 2 ) @xmath165 , and ( 3 ) @xmath166 . hence , one expects the resulting interaction terms given above . the sum of the kinetic energy and interactions described by eqs . ( [ eq:1dke2 ] ) , ( [ eq : fs0 ] ) , ( [ eq : fse ] ) , and ( [ eq : fsb ] ) gives the total hamiltonian that we focus on in this paper @xmath167 for the case of a single field ( either magnetic or electric ) the dispersion remains symmetric and therefore eq . ( [ eq : h_tot ] ) can be written in the form @xmath168 . \label{eq : ll}\ ] ] @xmath169 reflects no interactions in the @xmath170 sector and @xmath171 reflects repulsive interactions . for the asymmetric case , the diagonalization of the luttinger liquid hamiltonian is technically more complicated but conceptually simple given the quadratic form of the relevant terms @xcite . below we discuss the form of the various luttinger parameters and the related physics . of a 500 nm long nanotube as a function of electric field strength @xmath173 ( in v / nm ) for different values of @xmath16 : @xmath174 ( solid blue line ) , @xmath175 ( red dot - dashed line ) , and @xmath176 ( black dotted line ) . [ fig : kcplus],width=8 ] as seen above , for just an electric field present , interactions only affect the net charge sector @xmath177 and renormalize the velocity in this sector @xmath178 where @xmath179 is the lower cutoff provided by the length @xmath180 of the tube . this yields a luttinger parameter value @xmath181 in the low - field limit , where @xmath182 is the field reduced fermi velocity calculated in the previous section . figure [ fig : kcplus ] shows the dependence of @xmath183 on @xmath37 . in the other sectors , the luttinger parameters retain their non - interacting value @xmath184 . the luttinger model predicts power law behavior for the tunneling density of states @xcite with exponents @xmath185 and @xmath186 for tunneling into the end or bulk of a tube respectively . a smaller value of @xmath183 implies stronger repulsive interactions in the net charge sector . here , two tendencies compete ; in eq . ( [ eq : kcpe ] ) , @xmath187 increases the interaction strength relative to the kinetic energy . on the other hand , the direct effect of the field , as reflected in the @xmath188 term in eq . ( [ eq : kcpe ] ) is to decrease the relative interaction strength . the former effect is dominant of the latter , hence as shown in fig . [ fig : kcplus ] , @xmath183 is monotonically decreasing in increasing field strength . of a 500 nm long nanotube as a function of magnetic field strength @xmath1 ( in teslas ) for different values of @xmath16 : @xmath174 ( solid blue line ) , @xmath175 ( red dot - dashed line ) , and @xmath176 ( black dotted line ) . [ fig : kcminus],width=8 ] in the presence of a magnetic field , the luttinger parameter deviates from the non - interacting value of unity not only in the net charge sector , @xmath190 but also in the relative charge sector , @xmath191 the spin sectors remain unaffected ; @xmath192 . for both cases @xmath177 and @xmath193 , the reduction in the fermi velocity @xmath194 increases the relative interaction strength , thereby decreasing the values of @xmath195 . in addition , as reflected in the @xmath196 coefficient in eq . ( [ eq : kcmb ] ) , the field directly decreases @xmath197 via the interaction term of eq . ( [ eq : fsb ] ) discussed above . figure [ fig : kcminus ] shows the dependence of @xmath197 on the magnetic field @xmath1 . in this case , the luttinger model predicts power law behavior for the tunneling density of states with exponents @xmath198 and @xmath199 . unfortunately , this effect is small ; even for fields as large as @xmath200 , @xmath197 is reduced from unity only by about one half of a percent for a ( 20,20 ) tube ( see fig . [ fig : kcminus ] ) . nevertheless , for multi - wall nanotubes , the outermost layer can have @xmath16 as large as 100 , and hence , the field only needs to be as large as @xmath201 to see a 0.5 % reduction . the deviation of @xmath197 from unity leads to the fascinating prospect of spin - band - charge separation . in one - dimensional systems , the possibility of spin - charge separation stemming from different interactions within the two sectors has been extensively discussed and observed in the case of etched quantum wires @xcite . here we predict that nanotubes in transverse magnetic fields can undergo yet another separation due to the interactions in the @xmath193 sector . thus we propose that in this case , the four modes travel at three different velocities , @xmath202 , @xmath203 and @xmath204 . of an @xmath174 ( solid blue line ) , @xmath175 ( red dot - dashed line ) , and @xmath176 ( black dotted line ) carbon nanotube as a function of electric field strength e ( in v / nm ) in the presence of a 5 t magnetic field ( tube has @xmath205 ) . ( [ eq : vbe]),width=8 ] of an @xmath174 ( solid blue line ) , @xmath175 ( red dot - dashed line ) , and @xmath176 ( black dotted line ) carbon nanotube as a function of electric field strength e ( in v / nm ) in the presence of a 5 t magnetic field ( tube has @xmath205 ) . see eq . ( [ eq : vbe]),width=8 ] as shown in fig . [ fig : kcminus ] , the spin - band separation discussed above for purely electric fields is a small effect even for very large fields . the prospect of observing this effect is greatly improved for crossed electric and magnetic fields . in this case , the term which mixes the @xmath177 and @xmath193 sectors is given by eq . ( [ eq : fsbe ] ) and the values of @xmath155 are significantly larger than the corresponding terms @xmath153 and @xmath154 ( see appendix b ) . as mentioned previously , the case of crossed fields gives rise to an asymmetric dispersion , and it leads to cross terms @xmath206 with coefficients proportional to the difference @xmath207 , which is quadratic in the external fields . the full treatment of these and the cross term in eq . ( [ eq : fsbe ] ) is beyond the scope of the current paper . however , it is possible to get a sense of the order of magnitudes involved by noting that in the present case the dominant field effect is given by eq . ( [ eq : fsbe ] ) . we shall in the following study the effect of this cross term by ignoring the asymmetry in the dispersion . the relevant terms in the hamiltonian are @xmath208\nonumber\\ & & + \frac{v_{f}}{2}\int dx \big[(\partial_x\theta_{c-})^2 + ( \partial_x\phi_{c-})^2\big]\nonumber\\ & & + g\int dx(\partial_x\theta_{c-})(\partial_x\phi_{c+}),\end{aligned}\ ] ] with @xmath209 defining the field - free values . in diagonalizing the resulting hamiltonian , care is required to ensure that transformed fields respect commutation relations such as those of eq . ( [ eq : comm ] ) . it is straightforward to show that the resulting plasmonic modes retain linear dispersions having associated velocities and luttinger parameters given by @xmath210 \nonumber \\ k_{c\pm } & \equiv & v_f / v_{\pm } \label{eq : vbe}\end{aligned}\ ] ] where @xmath211 is the coefficient of the @xmath212 term and is given by @xmath213 . the associated tunneling density of states into the bulk of the tube is given by @xmath214 this expression differs from the usual form ( see the expressions below eq . ( [ eq : kcpe ] ) and below eq . ( [ eq : kcpe ] ) ) because of the unusual charge - current coupling term in the interactions . figure [ fig : kcminus2 ] shows the dependence of @xmath197 on electric field for a @xmath215 nanotube ( with @xmath216 ) in the presence of a 5 t magnetic field . interestingly , the fields render @xmath197 larger than unity , reflecting tendencies towards perfect conduction in the @xmath193 sector . furthermore , the values attained by @xmath197 show significant deviation from unity , making it feasible to observe the proposed spin - charge - band separation for these field configurations . figure [ fig : kcplus2 ] shows a plot of @xmath183 for the same situation . the field - tuning of the luttinger parameters discussed in the previous section offers a viable way of tuning the groundstate of the nanotube through different phases and different ordering tendencies . in the absence of fields , egger and gogolin @xcite performed an involved analysis using renormalization group arguments , refermionization and considerations of various susceptbilities to predict the ordering tendencies of the nanotube as a function of luttinger parameters and temperature . in particular , in the luttinger liquid regime which is the easiest to experimentally access , wherein all four sectors @xmath217 and @xmath218 , remain ungapped , the prediction is that for the range of interaction values @xmath219 , the system tends to show an inter - sublattice spin - density wave ( sdw@xmath14 ) ordering while for @xmath220 it tends to show inter - sublattice charge - density wave ( cdw@xmath14 ) ordering , where the corresponding operators are defined , respectively , as @xmath221 , and @xmath222 , . these analyses involved considering operators associated with certain orderings and determining the slowest decaying , equivalently , the most relevant operator ( i.e. , of the smallest scaling dimension ) . here , we discuss the key changes that occur in the luttinger regime in the presence of fields . we focus on the luttinger liquid regime and consider the manner in which the field - tuned change in luttinger parameters affect various susceptibilities . we only consider the cases where either only an electric or magnetic field is present ; the cases when both fields are present are extremely involved and beyond the scope of this paper . we do not take into account the effect of non - quadratic bosonic terms generated by the fields ; even if relevant , we expect that the bare coupling associated with these terms is so small that they only come into play at very low temperatures and not in the luttinger liquid regime . * case of @xmath223 , @xmath224 - * for the case of only an electric field present , as discussed above , the effect of the field goes purely into changing the value of @xmath183 . given that experimentally the value of @xmath183 is around and oftentimes higher than @xmath225 and that the field tends to reduce the value of @xmath183 , the electric field provides a unique means of tuning from a tendency towards ( sdw@xmath14 ) ordering to that of ( cdw@xmath14 ) ordering ( see for example figure [ fig : kcplus ] ) . * case of @xmath226 , @xmath227 - * the case of only a magnetic field present , as discussed above , presents a slightly more complex situation in which both @xmath183 and @xmath197 deviate from unity . as a result , various susceptibilities acquire a @xmath197 dependance in their scaling behavior . for instance , operators associated with intra - sublattice ordering such as @xmath228 and @xmath229 , which in the absence of fields are marginal , both acquire a scaling dimension @xmath230 . tendencies for superconducting order become slightly stronger , though still irrelevant ; the singlet pairing operator @xmath231 acquires the scaling dimension @xmath232 . to determine which ordering dominates , we consider the most relevant candidates : the cdw@xmath14 and sdw@xmath14 operators , both of which have scaling dimension @xmath233 ; parts of the second order cdw@xmath14 operator denoted by @xmath234 that have scaling dimension @xmath235 ; and a fourth order cdw@xmath14 operator denoted by @xmath236 which has scaling dimensions @xmath237 . comparing these exponents shows that @xmath238 and @xmath239 are more relevant than @xmath236 for @xmath240 , a condition easier to satisfy in the presence of fields than in the field - free case since @xmath197 can then be less than 1 . now @xmath241 is more relevant than @xmath236 for @xmath242 . for @xmath241 to be more relevant than @xmath238 and @xmath239 requires @xmath243 , a condition requiring inaccessibly strong interactions . finally , to determine whether @xmath238 or @xmath239 dominates , we appeal to the arguments of ref . @xcite ; at lower temperatures where the physics is dominated by certain strong coupling fixed points , pinning of the @xmath244 mode suppresses @xmath238 , making its magnitude in the luttinger phase smaller than that of @xmath239 . though the methods of refermionization employed to reach this conclusion are no longer valid for arbitrary values of @xmath245 , the strong coupling analysis still holds and we believe that a similar conclusion can be reached for the finite magnetic field situation . to summarize our results , while a detailed analysis , and considerations of operators that are not taken into account in ref . @xcite are in order , the most likely scenario is that the nanotube in the luttinger regime for the @xmath189 , @xmath49 case is dominated by sdw@xmath14 ordering tendencies for @xmath240 ( the more likely scenario gives that the deviation of @xmath197 from unity is not very large ) and cdw@xmath14 ordering tendencies for @xmath246 . for high resistance contacts or sufficiently low temperatures @xcite , the nanotube shown in fig . [ fig : setup ] is only weakly coupled to the leads , thus forming a quantum dot @xcite . the behavior of such dots and related coulomb blockade effects have been extensively studied by theory and experiment @xcite . typical of quantum dot physics , coulomb blockade peaks have provided information on single - particle level spacings and charging energies associated with the dot , as well as phase shifts due to scattering at the edges of the nanotube dot . moreover , under certain conditions , the nanotube quantum dot has revealed a periodicity of four associated with the degeneracy emerging from the band and spin degrees of freedom . recent work has also investigated the effects of a transverse magnetic field on quantum dot behavior and the associated single - particle and charging energies @xcite . here , we study the role of boundary conditions on nanotube quantum dot physics , which requires subtle considerations in the presence of fields . we then discuss the quantum dot behavior described by a finite - sized version of the nanotube luttinger liquid hamiltonian which takes into account relevant boundary conditions . as is well known , finiteness of the tube length leads to a quantized single particle spectrum which depends on the boundary conditions at the ends of the tube . we assume that the wavefunctions at a given end are related by @xmath247 where @xmath248 is a matrix which depends on the the microscopic details of the tube end but is assumed to be energy independent . we specialize to the case that the boundary conditions do not affect spin ; that is we take @xmath249 , where @xmath250 is the kronecker delta function . we thus assume the absence of magnetic impurities and local moments at the tube ends . in order to obtain the appropriate boundary conditions for the case of an asymmetric dispersion , we demand that the first quantized kinetic energy operator @xmath251 together with the boundary conditions is self - adjoint @xcite . this treatment does not account for the effect of interactions on the boundary conditions which would be more naturally discussed in terms of the bosonic fields . such an analysis shows that there is an additional term in the current proportional to @xmath252 which vanishes for the density - density interaction considered here @xcite . by definition , @xmath253 since the boundary effects are assumed to be independent of spin , we have dropped the spin index . for an arbitrary spinor @xmath254 with @xmath255 where @xmath256 and @xmath257 are both two - component spinors in the fermi point basis ( @xmath258 ) , self - adjointness gives @xmath259 integrating the left - hand side of this equation by parts gives @xmath260_{x = 0,l } \\ & & + \langle \hat{h}_0 \psi , \psi \rangle\end{aligned}\ ] ] self - adjointness is satisfied as long as the boundary terms vanish , and this leads to @xmath261 with @xmath262 unitary . the details of the s - matrix can vary for each experimental set - up and depend on physical attributes such as the substrate , the hardness of the confining potential offered by the leads and the orientation of the tube s sublattices with respect to the leads . these parameters can be incorporated as variables in the boundary conditions which can then be utilized to obtain the single - particle spectrum . the most general version of these boundary conditions are outlined in ref . @xcite via the effective - mass model . for a given s - matrix the spectrum of single particle states can be determined by applying the condition of eq . ( [ eq : bc ] ) at both ends and demanding that both the left and right movers have the same energy . the two fermi points give rise to two sets of bands . the energy between two adjacent states in the _ same _ band is equal to @xmath263 where @xmath264 is the harmonic mean , @xmath265 however , the energy offset of the bands from the fermi energy depends on the details of the s - matrix . in general , the two fermi points will give rise to two sets of energy states given by @xmath266 and @xmath267 where @xmath268 . since we are ultimately interested in the spacing between coulomb blockade peaks , we focus on the energy difference between bands which we define as @xmath269 ( and for convenience we define @xmath270 such that @xmath271 ) . we examine two special cases for the s - matrix ; deriving the @xmath270 for the most general scattering matrix would be a straightforward extension . first , consider the case in which the tube ends do not mix the fermi points , though we allow the phase shift the electron suffers at the tube end @xmath272 to be different for the two fermi points ( @xmath20 ) and the two tube ends ( @xmath273 ) . in this case we have @xmath274 , @xmath275 and @xmath276 . the energy offset between the bands takes the form @xmath277 , \label{eq : bandd}\ ] ] where @xmath278 and @xmath279 and @xmath280 is the greatest integer less than or equal to @xmath26 , and the quantity @xmath281 is the field induced offset between the two fermi points as defined in eq . ( [ eq : mdiv ] ) . ( see appendix [ ap:3 ] for derivation . ) now , consider an electron that is completely scattered into the opposite fermi point at both boundaries . for simplicity we take @xmath282 . in this case , the splitting between bands takes the form @xmath283 . \label{eq : bandod}\ ] ] for the limiting case of no fields ( this also means that @xmath284 ) , one expects the existence of sets of four single - particle states , namely two degenerate sets of spin states and two sets of band states whose energy splitting depends on the various phase shifts and the extent to which modes at the two fermi points mix . for no fermi point mixing , the interband splitting is @xmath285 while for complete fermi point mixing , the splitting is @xmath286 $ ] . coulomb blockade experiments have shown an interband band splitting of about 10% @xcite of the @xmath287 . such a persistent approximate degeneracy in band energies for a range of tubes @xcite suggests that the magnitude of fermi point mixing in these samples is minimal . as discussed above , the boundary conditions in a given experiment are not directly observable since @xmath270 depends on several parameters . fields provide a way of controlling @xmath270 as well as studying its physical origin in a particular sample . by scanning through various field strengths , the variation of the band offset can reveal information about the nature of boundary scattering . for example , the extent to which a given tube interpolates between the two expressions given in eqs . ( [ eq : bandd ] ) and ( [ eq : bandod ] ) can be used to determine the importance of ( fermi point ) backscattering at the tube ends . in the case of a _ natural _ band degeneracy in a tube ( that is , no fermi point mixing at the ends of the tube and @xmath288 ) , both electric and magnetic fields need to be applied to break the degeneracy ; the magnitude of the subband splitting as a function of fields can be extracted from eq . ( [ eq : bandd ] ) by setting @xmath288 . an alternative approach for breaking the subband degeneracy was explored by ref . @xcite in which a nonuniform external potential along the tube was applied . however , this approach becomes infeasible for the case of a diagonal scattering matrix since it relies on band curvature away from half - filling . thus combining electric and magnetic fields can provide a means of breaking and tuning the degeneracy of the quantum states of electrons inhabiting the nanotube quantum dot . of the four possible states discussed above , where the direction of spin is defined with respect to the magnetic field , an extra electron would occupy the ground state , which can be chosen to be any of the four depending on the direction of the fields . the quantum state can be characterized by a superspin inhabiting a @xmath289 band and spin space . the enhanced control of the spectrum of nanotubes that fields offer would obviously have important implications for any potential quantum information applications . we now consider quantum dot phenomena by taking into account interaction effects in addition to the single - particle level spacing analysis of the previous subsections . to present a coherent picture , we work within the context of the luttinger liquid description for field configurations that retain gapless modes . this approach neglects the exchange energy within a dot , which while shown to be present , is often much smaller in magnitude than the level spacing and interaction energy @xcite . while our treatment captures salient features of quantum dot behavior , a full analysis of the luttinger liquid formulation for the most general boundary conditions is yet to be performed . following the method of ref . @xcite ( see also refs . @xcite ) , for a finite sized version of eq . ( [ eq : h_tot ] ) , we decompose the bosonic fields @xmath290 and @xmath291 into sums of topological modes @xmath292 , corresponding to a net occupation number of the @xmath170 sector @xmath293 and harmonic modes corresponding to plasmons . for simplicity , we consider the case of no fermi point scattering so that the fermi point basis as defined by eqs . ( [ eq : basisdef ] ) and ( [ eq : bosondensity ] ) and the band basis which diagonalizes the boundary conditions in the previous subsection coincide ( and therefore take @xmath294 ) . assuming the boundary conditions derived in the previous section , we integrate out the @xmath26-dependence for the topological sector in the finite sized version of eq . ( [ eq : h_tot ] ) . the resulting hamiltonian associated with charging energy for each topological sector takes the form @xmath295 where @xmath296 and @xmath297 is equal to the interaction energy of a given mode . the interaction energy for the net charge sector comes from the forward scattering contribution of eq . ( [ eq : fs0 ] ) to yield @xmath298 . the contribution due to the electric field given by eq . ( [ eq : fse ] ) is found to be fourth order in fields and can thus be neglected . the magnetic field , however , does contribute to the charging energy ; from eq . ( [ eq : fsb ] ) , it can be shown that @xmath299 . this expression represents an upper bound that assumes the limit of no fermi - point mixing . the topological modes correspond to the addition of charges onto the dot ; in addition , plasmon modes that correspond to harmonic vibrations of the densities in the various sectors are present . in principle , the procedure employed in ref . @xcite to derive the structure of these plasmonic excitations can be generalized to the case of the asymmetric dispersion by incorporating the asymmetric description in ref . @xcite . here we forego such a derivation ; most quantum dot experiments involving adiabatic tuning of parameters such as gate voltage and thus probe purely ground state properties determined by the topological sectors . in the presence of a gate voltage @xmath300 , the hamiltonian associated with the topological modes of the nanotube quantum dot is given by @xmath301 where @xmath302 is essentially @xmath303 and the term @xmath304 accounts for the zeeman splitting . the hierarchy in energy scales can be summarized for a typical tube length of @xmath305 nm . the intrasubband splitting is @xmath306 mev . therefore we have that @xmath307 mev . the charging energy is then @xmath308 mev . consider a tube for which the ends of the tube do not appreciably change the fermi point . in that case the sign of @xmath270 can be changed by reasonable electric and magnetic field values ( for example , this is true for an @xmath309 tube , a 6 t magnetic field and an electric field of order 1 v / nm ) . finally , @xmath310 mev / t and therefore for most situations the zeeman splitting will be at least an order of magnitude smaller than the other effects considered here . the hamiltonian in eq . ( [ eq : ham_qtmdot ] ) forms a starting point for analyzing quantum dot and coulomb blockade behavior in short nanotubes . typically , conductance is measured across the tube as a function of the applied gate voltage @xmath300 and a bias voltage @xmath311 ; while for the most part , energetic costs impede the flow of electrons , at special degeneracy points that equally favor an occupation of @xmath312 and @xmath313 electrons , zero bias coulomb blockade conductance peaks can be observed . given that the occupation numbers @xmath314 with @xmath315 are good quantum numbers , the net energy of the system @xmath316 for a given configuration of electrons corresponds to a given combination of eigenvalues of @xmath314 . the equilibrium configuration of electrons on the quantum dot can thus be derived by minimization , i.e. , by requiring @xmath317 for all @xmath314 sectors , subject to the physical constraint that that electron occupation numbers @xmath318 take on integer values . the relationship between these two bases is given by @xmath319 and @xmath320 . as an illustrative example , consider the coulomb blockade situation for the simple case of no fields , no band or zeeman splitting ( @xmath321 and only the charging energy in the @xmath177 sector is non - zero ) . applying the condition that @xmath322 gives @xmath323 and @xmath324 ( where @xmath302 is varied by @xmath300 ) . these conditions suggest that all the states at a given energy level will fill before filling the subsequent energy level . now , as the chemical potential is increased , the first extra electron added onto the dot can occupy any of the degenerate states of the @xmath325 space . suppose that this electron goes in the @xmath39 band with its spin up . the state of the tube is then characterized by the quantum numbers @xmath326 . further increasing the chemical potential by an amount @xmath327 will add the next electron to any of the remaining three states . for example if the filling obeys hund s rule ( i.e. , assuming the exchange interaction which we have thus far neglected ) then band @xmath62 would be filled with spin up electron thus giving the state @xmath328 . an additional third electron can occupy @xmath329 or @xmath330 state . the energy cost for adding each of these extra electrons is @xmath327 , reflecting the coulomb charging energy . however , the fifth extra electron requires a chemical potential increase of @xmath331 , reflecting the coulomb energy as well as the excitation energy required to occupy the next energy level . this analysis , executed within the luttinger liquid description , reproduces the periodicity of four observed in experiment @xcite . in the presence of fields and intrinsic subband splitting , an analysis similar to the one above can be performed for altered values of @xmath332 and @xmath333 s and the orders of magnitude discussed after eq . ( [ eq : ham_qtmdot ] ) . we take @xmath334 and @xmath335 . furthermore , suppose that for a given tube that a nonzero magnetic field gave @xmath336 . in that case , the tendency to minimize @xmath337 would give rise to a shell - filling opposite to the usual hund - like filling . for example , the order in which states are filled could take te forn @xmath338 or equivalently @xmath339 . instead of the coulomb blockade peaks being of periodicity 4 described above ( spaced apart by @xmath340 , spacings become @xmath341 . the effect of shell - filling in nanotube quantum dots has been investigated experimentally by liang et al . the four - electron periodicity was clearly observed via transport measurement . parameters of charging energy and exchange energy were determined . the above results , including field - dependent luttinger parameters , the effect of boundary conditions , the band splitting , asymmetric dispersions and additional two - electron periodicity can thus be studied in a similar setup with transverse fields . additionally , the fabry - perot transmission resonances in the presence of a transverse magnetic field predicted by bellucci and onorato @xcite , as well as possible resonances by both transverse electric and magnetic fields ( and their relative angles ) , can be investigated in the same experimental setup . while our arguments here have been confined to adiabatic tuning and zero - bias conductance , our approach can be used to investigate non - equilibrium phenomena , temperature dependences , higher order tunneling events such as cotunneling and non - adiabatic tuning . each of these considerations , which is beyond the scope of this paper , would involve excitations of the plasmonic modes . we have investigated the effects of transverse electric and magnetic fields on armchair carbon nanotubes . we found that fields can break several symmetries inherent to the carbon nanotubes the valley degeneracy , the left - right - mover degeneracy , and the particle - hole symmetry . the magnitude of a gap in the nanotube spectrum can be continuously tuned by varying the strength and the relative angle of the two fields . we also found that the electron - electron interaction is modified by both fields and thus luttinger - liquid parameters can be tuned by fields . in particular , an interesting consequence is the possibility of spin - charge - band separation . we also discussed how the fields can be used to study boundary effects in finite sized tubes and to describe the coulomb - blockade physics in presence of fields . each of these salient features can become manifest in experiment , some in dramatic ways . while we summarize these experimental signatures here , details of the physics and orders of magnitude can be found in the relevant section . at the band structure level , the reduction of the fermi velocity can be observed by measuring the particle level spacing in a finite size tube . the shift in the fermi momentum induced by either field may be detected by virtue of the associated friedel oscillations around a dopant or impurity using a stm ( see for example @xcite ) for both fields present at an arbitrary angle to one another , a continuous conduction gap occurs at the fermi energy , discernible via direct conductance measurements , shifts in coulomb blockade peaks and stm measurements @xcite . in the transport measurement , the conductance peak can vary from @xmath68 to @xmath97 , and to @xmath342 , depending on the fields and the chemical potential . perhaps the most dramatic prediction of band structure analysis is that an electric and a magnetic field perpendicular to one another and the tube axis would give rise to a current carrying ground state . thus , a tube subject to this field configuration placed across two leads should induce a measurable current even in the absence of an applied voltage drop across the leads . in the regime in which swnts exhibit luttinger liquid behavior , strong enough fields can give rise to significant changes . the value of the luttinger parameter associated with the net charge sector can be tuned via either an electric field or a magnetic field or both . furthermore , the presence of an electric field gives rise to density - density interactions associated with the difference in densities in the two bands and results in the deviation of the associated luttinger parameter @xmath197 from its non - interacting value of unity . the tunneling density - of - states , a quantity ubiquitous to a range of experiments including scanning tunneling microscope studies and conductance measurements @xcite , would reflect these changes in its power - law behavior . as discussed in sec.[sec : luttinger ] , changing the luttinger parameter via fields can result in tuning through phases having charge - density - wave order or spin - density - wave order ; such phases are potentially measurable in stm and neutron scattering experiments . another exciting prospect comes about in the tuning of @xmath197 , namely , that of spin - band - charge separation . in the past , momentum resolved tunneling experiments have resolved charge and spin modes moving at different velocities in quantum wires @xcite ; in principle , a more elaborate version of such an experiment could detect charge , spin and band modes moving at three different velocities in nanotubes . in the quantum dot regime , the application of fields acts as a controlled means of changing the coulomb blockade structure of the dot , and could potentially have a plethora of applications . for tubes that preserve the four - fold degeneracy emerging from spin and band degrees of freedom in the absence of fields , the presence of fields can serve to break this degeneracy . for tubes that show a lack of degeneracy , fields provide a way of determining the origin of degeneracy breaking . this effect has potential applications to quantum information . through achieving a desired amount of degeneracy breaking for the four states , fields can be an effective means of initializing the quantum state of an extra electron added on to the dot . having initialized a quantum state , transitions can be induced to other states . additionally , as has been demonstrated for semi - conducting quantum dot spin states @xcite , superpositions can be created by applying oscillating fields . for the energies quoted above , oscillation frequencies would be on the order of @xmath343 hz . to conclude , transverse fields induce a rich range of physical effects in the electronic properties of swnts from band structure effects to one - dimensional behavior to quantum dot physics . several of these features are very much within the reach of current experimental capabilities and are of both fundamental and applied value . we would like to acknowledge m. stone for illuminating discussions . one of us ( wd ) would like to thank sarang gopalakrishnan for discussions on the various symmetries of the band structure and jeremy mcminis for his assistance in preparing the nanotube figure ( fig . this work was supported by the nsf under the grant dmr-0605813 ( sv , wd ) . tcw was supported by iqc , nserc and orf . in the absence of any fields , the eigenstates of an infinitely long armchair nanotube are superpositions of the states @xmath344 defined by eq . ( 1 ) . for the particular case of an armchair nanotube eq . ( 4 ) gives @xmath345 \right ) \bigg\}.\end{aligned}\ ] ] where @xmath346 . for small magnetic fields ( @xmath347 ) we have @xmath348 for @xmath349 and @xmath350 applying perturbation theory near the fermi points requires care because of the near degeneracy of the left and right movers . in this case , the usual procedure of first diagonalizing the nearly degenerate subspace fails because the matrix elements within this subspace vanish to the order we are working . however , there are matrix elements for transitions to other energy levels ( @xmath351 ) and these matrix elements give rise to an effective interaction between states in the nearly degenerate subspace . the subspace can be diagonalized once these additional interactions are taken into account @xcite . for the gapless case , the low energy spectrum near half - filling is @xmath352the renormalized fermi velocity is given by @xmath353 where @xmath354 @xmath355 @xmath356 and @xmath357 \label{eq : kf}\ ] ] the shift between the two fermi points ( see eq . ( 10 ) ) is given by @xmath358 for the case of mutually perpendicular fields discussed in section ii , the electronic densities are given by the vector @xmath359 where the upper and lower components are the electronic densities over the @xmath0 and @xmath1 sublattices respectively . the constants @xmath360 and @xmath361 are given by @xmath362 in order to find the form of the effective interaction @xmath363 we need account for not only the radial dependence of the wave functions but also for the physical separation between the sublattices . although the factorization performed in eq . [ eq : wvfn ] is an approximation , we may still account for the physical separation of the sublattices . we follow the approach introduced in @xcite . @xmath364 where @xmath365 is a shorthand for the coulomb interaction with an offset @xmath366 , that is @xmath367 where the right hand side of this equation is given by eq ( [ eq : coulomb ] ) . the constant @xmath148 which appears in eq . ( [ eq : one ] ) is given by @xmath368.\end{aligned}\ ] ] similarly we find that the values of @xmath153 and @xmath154 defined in section 3 are given by @xmath369 @xmath370 and @xmath371,\end{aligned}\ ] ] where @xmath372 \right . \\ & + & \left . \ln \left [ \cos^2 \frac{z+a_{c}}{2r } + \left ( \frac{a_z}{2r } \right)^2 \right ] \right\ } \cos \frac{z}{r},\end{aligned}\ ] ] @xmath373 \right.\\ & - & \left . \ln \left [ \cos^2 \frac{z+a_{c}}{2r } + \left ( \frac{a_z}{2r } \right)^2 \right ] \right\ } \sin \frac{z}{r},\end{aligned}\ ] ] @xmath374\end{aligned}\ ] ] and @xmath375 the values of @xmath148 , @xmath153 , @xmath154 and @xmath155 have been tabulated for various tube sizes and are displayed below . [ cols=">,>,>,>,>",options="header " , ] here we illustrate the case in which the tube ends do not mix fermi points ; that is @xmath376 , @xmath275 and @xmath276 . assume the wavefunctions are @xmath377 the energy spectrum for each branch is described by eq . ( [ eq : mdiv ] ) , e.g. , @xmath378 . the self - adjointness condition ( [ eq : bc ] ) at @xmath379 gives @xmath380 this gives constraints on the four momenta @xmath381 where @xmath382 are arbitrary integers . for the @xmath39 branch , the energy levels corresponding to @xmath383 and @xmath384 should be equal ( in order for the wavefunction to represent an eigenstate ) . hence @xmath385 which gives @xmath386 and thus @xmath387 where @xmath388 . similarly , for @xmath62 , we have @xmath389 hence , we arrive at the interband energy difference @xmath390 where @xmath391 and the @xmath392 are selected so that @xmath393 is less than the intraband spacing . hence we obtain eq . ( [ eq : bandd ] ) . a similar consideration leads to eq . ( [ eq : bandod ] ) . o. auslaender , a. yacoby , r. de picciotto , k. w. baldwin , l. n. pfeiffer , and k. w. west , science * 295 * , 825 ( 2002 ) . o. m. auslaender , h. steinberg , a. yacoby , y. tserkovnyak , b. i. halperin , k. w. baldwin , l. n. pfeiffer , and k. w. west , science * 308 * , 88 ( 2005 ) .

we study the connes kreimer hopf algebra of renormalization in the case of gauge theories . we show that the ward identities and the slavnov taylor identities ( in the abelian and non - abelian case respectively ) are compatible with the hopf algebra structure , in that they generate a hopf ideal . consequently , the quotient hopf algebra is well - defined and has those identities built in . this provides a purely combinatorial and rigorous proof of compatibility of the slavnov taylor identities with renormalization . graphs - st

hep-th0610137

the combinatorial structure underlying renormalization in perturbative quantum field theory was transparent in the original approach of bogogliubov , hepp , parasiuk and zimmerman ( cf . for instance ( * ? ? ? * ch.5 ) ) . it was realized by kreimer in @xcite that this structure is in fact organized by a hopf algebra of rooted trees . one year later , connes and kreimer @xcite reformulated this combinatorial structure directly in terms of a ( commutative ) hopf algebra of feynman graphs and understood the bphz - procedure as a birkhoff decomposition in the group that is dual to this hopf algebra . in physics , however , one usually works in the setting of functional calculus since although defined only formally functional integrals are particularly well - suited for the perturbative treatment of quantum gauge theories . for example , the slavnov taylor identities that are the reminiscents of the gauge symmetry of the classical field theory , can be gathered elegantly in a single equation known as the _ zinn - justin equation _ , and involving the effective action . although this approach is very powerful in showing for example renormalizability of gauge field theories , the applicability of the graph - per - graph approach of the bphz - renormalization procedure is not so transparent any more , and the same holds for the combinatorial structure underlying it . recent developments @xcite give more insight in the combinatorial aspects of non - abelian gauge theories . in @xcite , we considered the hopf algebra of feynman graphs in quantum electrodynamics and found that certain ward takahashi identities can be imposed as relations on this hopf algebra . in this article , we continue to explore the combinatorial structure of gauge theories in terms of a hopf algebra . more precisely , we will show that the slavnov taylor identities between the coupling constants can be implemented as relations on the hopf algebra , that is , in a way compatible with the counit , coproduct and antipode . this then provides a combinatorial proof of the compatibility of the slavnov taylor identities with renormalization . we start in section [ sect : prel ] with a precise setup of the hopf algebra of feynman graphs in a generic theory , including gauge theories , and derive a formula for the coproduct on 1pi green s functions . such green s functions are sums of all 1pi graphs with a certain fixed external structure , including symmetry factors . it is the latter that make this derivation slightly involved . section [ sect : ward ] we will be a warming - up for the non - abelian case , by considering quantum electrodynamics , which is an abelian gauge theory . using the expression for the coproduct on the green s functions , we show that certain ward identities can be imposed as ( linear ) relations on the hopf algebra . in other words , they define a hopf ideal . the case of non - abelian gauge theories will be consided in section [ sect : st ] , where we will show that the slavnov taylor identities define quadratic relations in the hopf algebra . in fact , as we will see , it is in the very nature of the combinatorial factors that are involved that the slavnov taylor identities appear . we have added two appendices . in the first , we rederive the compatibility of ward identities with the hopf algebra structure in qed obtained in section [ sect : ward ] from our previous result on ward takahashi identities in @xcite . in the second appendix , we list some useful basic combinatorial identities used throughout the text . we start with some definitions on feynman graphs and their symmetries , thereby making precise several properties needed later . the feynman graphs we will consider are built from a certain set of edges and vertices @xmath0 , and we write @xmath1 . for example , in @xmath2-theory , the set @xmath3 contains the bi- and trivalent vertex and @xmath4 the straight line , but more interesting theories such as gauge theories contain different types of edges and vertices ( for example involving curly , dotted and straight lines ) corresponding to different particles . more precisely , we have the following definition @xcite . a feynman graph @xmath5 is given by a set @xmath6 of vertices each of which is an element in @xmath3 and @xmath7 of edges in @xmath4 , and maps @xmath8 that are compatible with the type of vertex and edge as parametrized by @xmath3 and @xmath4 , respectively . the set @xmath9 labels the external lines , so that @xmath10 for all @xmath11 . the set of external lines is @xmath12 and its complement @xmath13 in @xmath7 is the set of internal lines . we remark that the elements in @xmath14 can thus be labeled as @xmath15 where @xmath16 and we understand this labeling as being fixed . with this definition , the notion of a graph automorphism can be defined as follows . an automorphism of a feynman graph @xmath5 is given by an isomorphism @xmath17 from @xmath6 to itself , and an isomorphism @xmath18 from @xmath7 to itself that is the identity on @xmath14 and such that for all @xmath19 , @xmath20 moreover , we require @xmath17 and @xmath17 to respect the type of vertex / edge in the set @xmath0 . + the automorphism group @xmath21 of @xmath5 consists of all such automorphisms ; its order is called the symmetry factor of @xmath5 and is denoted by @xmath22 . similarly , there is a notion of an isomorphism of two graphs @xmath5 and @xmath23 as a pair of maps that intertwines the maps @xmath24 as in eq . . we remark that we correct in this way for the apparent orientation given by the two maps @xmath25 and @xmath26 and we stress that the fermionic lines are unoriented . we take the complex character of the fermionic fields into account by summing over all possible orientations once we apply the feynman rules . the above definition of automorphism differs from the usual notion of graph automorphism ( cf . for instance @xcite ) in that the latter might also permute the elements in @xmath27 when understood as external vertices . in the above notation , such an automorphism of @xmath5 would be given by an isomorphism @xmath17 from @xmath28 to itself , and an isomorphism @xmath18 from @xmath7 to itself such that equation holds . @xmath29 @xmath30 @xmath29 @xmath31 note that for @xmath32 the disjoint union of @xmath33 graphs , the symmetry factor is given by @xmath34 where @xmath35 are the numbers of isomorphic ( with fixed external lines ) connected components of @xmath23 . equivalently , one has for a 1pi graph @xmath23 , @xmath36 with @xmath37 the number of connected components of @xmath38 that are isomorphic to @xmath23 . if @xmath5 is a connected feynman graph with external lines labeled by @xmath39 , we can construct another graph @xmath40 , by permutating the external lines by an element @xmath41 , respecting the type of external lines . the graph @xmath40 is given by the same sets @xmath6 and @xmath7 but with maps @xmath42 this permutation affects the labeling of the external lines by @xmath43 , which explains the terminology permutation of external lines ; we write @xmath44 for the edge in @xmath40 corresponding to an edge @xmath19 under the permutation @xmath41 . [ defn : triv - perm ] a permutation @xmath41 of the external lines of @xmath5 is called trivial if there exists an isomorphism between @xmath40 and @xmath5 , leaving the labeling of the external lines fixed . + the number of non - isomorphic graphs @xmath40 obtained by a permutation @xmath41 of the external lines of @xmath5 , is denoted by @xmath45 and extended to disconnected graphs by @xmath46 . [ lma : perm ] a permutation @xmath41 of the external lines of @xmath5 is trivial if and only if there exists an automorphism @xmath47 of the graph @xmath5 not necessarily leaving the external lines fixed , such that @xmath48 . firstly , if @xmath41 is trivial , there exists an isomorphism @xmath49 and the pair @xmath50 is an automorphism @xmath47 of @xmath5 ( without fixed external vertices ) , since , @xmath51 on the other hand , such an automorphism @xmath47 is given by two maps @xmath17 and @xmath18 , where @xmath17 is the product of two permutations of the disjoint sets @xmath6 and @xmath27 , say @xmath52 and @xmath41 , respectively . correspondingly , @xmath41 acts on @xmath14 by permutation , so that also @xmath53 . this factorization gives rise to an isomorphism @xmath54 from @xmath40 to @xmath5 , which leaves external lines fixed . @xmath55 for the purpose of renormalization , one is mainly interested in one - particle irreducible feynman graphs which have _ residues _ that are elements in the set @xmath0 . a feynman graph is called one - particle irreducible ( 1pi ) if it is not a tree and can not be disconnected by removal of a single edge . the residue @xmath56 of a feynman graph @xmath5 is defined as the vertex / edge the graph corresponds to after collapsing all its internal edges to a point . for example , we have @xmath57 we restrict to the class of feynman graphs @xmath5 for which @xmath58 and will denote a generic graph with residue @xmath59 by @xmath60 . if it also has loop number @xmath61 , we denote it by @xmath62 . the hopf algebra @xmath63 of feynman graphs is the free commutative algebra generated by all 1pi feynman graphs , with counit @xmath64 unless @xmath65 , in which case @xmath66 , coproduct , @xmath67 and antipode given recursively by , @xmath68 [ defn : ins ] an insertion place for a ( connected ) graph @xmath69 in @xmath5 is the subset of @xmath70 } \cup \gamma^{[1]}$ ] consisting of vertices / internal edges of the form @xmath71 . it can be extended to disconnected graphs @xmath72 by giving @xmath33-tuples of insertion places for @xmath73 , thereby allowing several insertions of the connected components with residue @xmath74 in @xmath4 on the same internal edge in @xmath5 of the form @xmath74 . the number of insertion places for @xmath69 in @xmath5 is denoted by @xmath75 . an explicit expression for @xmath75 can be obtained as follows @xcite . let @xmath76 be the number of vertices / edges @xmath74 in @xmath70 } \cup \gamma^{[1]}$ ] , for @xmath59 . moreover , let @xmath77 be the number of connected components of @xmath69 with residue @xmath74 . since insertion of a vertex graph ( i.e. with residue in @xmath3 ) on a @xmath78}$ ] prevents a subsequent insertion at @xmath79 of a vertex graph with the same residue , whereas insertion of an edge graph ( i.e. with residue in @xmath4 ) creates two new egdes and hence two insertion places for a subsequent edge graph , we find the following expression , @xmath80 indeed , the binomial coefficients arises for each vertex @xmath79 since we are choosing @xmath81 out of @xmath82 whereas for an edge @xmath83 we choose @xmath84 out of @xmath85 _ with repetition_. we extend this definition to empty graphs by defining @xmath86 for a 1pi graph @xmath69 , and @xmath87 for a disconnected graph @xmath69 . our expression for @xmath75 differs slightly from the one given in @xcite where additional factors of @xmath88 are present for @xmath59 . it turns out that the above expression appears naturally in the coproduct on 1pi green s functions ( see below ) . a few examples are in place : @xmath89 an insertion of a connected graph @xmath69 at the insertion place @xmath90 in @xmath5 , is given by a bijection between the set @xmath91 of external lines of @xmath69 and the set @xmath92 . if @xmath93 , @xmath92 denotes the set of lines attached to the vertex @xmath90 , and if @xmath94 , @xmath92 denotes the set of adjacent edges to any internal point of @xmath90 . the graph obtained in this way is denoted by @xmath95 . + two insertions @xmath96 and @xmath97 are called equivalent if @xmath98 and @xmath99 for some trivial permutation @xmath41 of the external lines of @xmath69 . the set of all insertions of @xmath69 in @xmath5 up to equivalence is denoted by @xmath100 ; it consists of equivalence classes @xmath101 $ ] . this equivalence relation on insertions makes sense in that @xmath102 whenever @xmath103 . we extend @xmath104 to disconnected graphs @xmath69 as follows . if @xmath105 is the disjoint union of @xmath33 graphs , the set @xmath104 of insertions of @xmath69 in @xmath5 is defined as the set of @xmath106tuples of pairs @xmath107 , \ldots , [ x_n,\phi_n]\right)$ ] , where @xmath108 \in x(\gamma , \gamma_1)$ ] and @xmath109 $ ] is an element in @xmath110 which is not part of any of the inserted graphs @xmath111 for @xmath112 . the cardinality of @xmath104 is the number @xmath75 of insertion places for @xmath69 in @xmath5 times the number @xmath113 of non - trivial permutations of the external lines of @xmath69 . we also need the following generalization for the number of insertion places . [ defn : ins3 ] let @xmath114 be three ( disjoint unions of ) 1pi graphs . we define @xmath115 to be the number of places to insert @xmath69 into @xmath5 ( say , at @xmath90 using @xmath116 ) and then subsequently insert @xmath117 in @xmath95 . in other words , @xmath118 \in x(\gamma,\gamma)}{\gamma \circ_{(x,\phi ) } \gamma~|~\gamma'}.\ ] ] moreover , we set @xmath119 and @xmath120 for two 1pi graphs @xmath121 and @xmath122 if one of the two graphs @xmath121 is disconnected . the factor @xmath123 corrects for the overcounting due to the several ( non - equivalent ) ways to insert @xmath69 into @xmath5 at a particular place . note that automatically @xmath124 and if @xmath125 , we have @xmath126 suppose @xmath69 is a 1pi graph . there is a natural action of @xmath21 on @xmath104 given by @xmath127 one readily checks that this action respects the equivalence relation on insertions , and therefore acts on the equivalence classes @xmath101 $ ] . moreover , an element @xmath128 naturally induces an isomorphism @xmath95 to @xmath129 for an element @xmath101 $ ] in @xmath104 , we denote by @xmath130 the number of graphs @xmath117 in @xmath131 that are images of @xmath69 under some element in @xmath132 . moreover , @xmath133 denotes the number of orbits @xmath134 $ ] such that @xmath135 . both definitions are independent of the choice of a representative @xmath96 as well as the choice of the element @xmath136 $ ] in the orbit . indeed , an element @xmath47 in @xmath21 will induce a natural isomorphism @xmath137 . [ lma : orbit ] suppose @xmath69 is a 1pi graph and let @xmath138 . the length of the orbit @xmath139 $ ] is given by @xmath140 \right| = \frac{\sym(\gamma ) \sym(\gamma ) m(x,\phi)}{\sym(\gamma \circ_{(x,\phi ) } \gamma)}.\ ] ] we use the orbit - stabilizer theorem , stating in this case that the orbit @xmath139 $ ] is isomorphic to the left cosets of the stabilizer @xmath141}$ ] in @xmath21 . in particular , we have for its length , @xmath142|= \left[\aut ( \gamma ) : \aut ( \gamma)_{[x,\phi ] } \right]=\frac{|\aut(\gamma)|}{|\aut(\gamma)_{[x,\phi]}|}.\ ] ] the order of @xmath141}$ ] can be computed as follows . let @xmath143 be the subgroup of @xmath132 consisting of automorphisms that map @xmath69 to itself ( but possibly permuting the external lines of @xmath69 ) . there is a short exact sequence of groups @xmath144 } \to 1.\ ] ] indeed , the image @xmath145 inside @xmath21 of an element @xmath47 in @xmath146 is defined by restricting @xmath47 to @xmath147 and by the identity map on the vertex @xmath90 . then , by lemma [ lma : perm ] , @xmath145 might permute the edges connected to the vertex @xmath90 but always in a trivial way , since @xmath47 induces an automorphism of @xmath69 not necessarily leaving its external lines fixed . therefore , @xmath148 for some trivial permutation @xmath41 of @xmath149 , so that it is an element in the fixed point subgroup @xmath150}$ ] . moreover , the kernel of the map that sends such a @xmath47 to @xmath145 consists precisely of those elements in @xmath146 that correspond to the identity on @xmath5 ; in other words , these are automorphisms of @xmath69 that leave external lines fixed . we conclude that the quotient group @xmath151 is isomorphic to @xmath150}$ ] . since @xmath152 is generated by the elements in @xmath143 and automorphisms that map @xmath69 isomorphically to a subgraph @xmath117 of @xmath5 , we see that @xmath153 combining these results , we conclude that @xmath142|= \frac{|\aut(\gamma)|~|\aut ( \gamma)|}{|\aut(\gamma\circ_{(x,\phi)}\gamma)_\gamma| } = \frac{\sym(\gamma)\sym(\gamma)m(x,\phi)}{\sym(\gamma\circ_{(x,\phi ) } \gamma)}.\ ] ] as a final preparation to the next section , we will write the coproduct as a sum of maps @xmath154 , with @xmath69 a disjoint union of 1pi graphs ( with fixed external lines ) . it is given by @xmath155 and defined to be zero if @xmath5 contains no subgraphs isomorphic to @xmath69 . in particular , @xmath156 is the identity map , @xmath157 and @xmath158 if @xmath159 . however , since only subgraphs isomorphic to @xmath69 enter in this formula hence no reference is made to a particular labeling of the external lines of @xmath69 we have to correct by a factor of @xmath113 if we are to sum over all disjoint unions of 1pi graphs _ with fixed external lines _ , @xmath160 we recall the following combinatorial factor from @xcite ; for a given @xmath161 , we denote by @xmath162 the number of subgraphs @xmath163 in @xmath5 such that @xmath164 . with this definition , we can write @xmath165 which also yields the following formula for the coproduct , @xmath166 from this last formula , one easily derives the lie bracket on feynman graphs as derived in @xcite . indeed , one can define a pre - lie product between 1pi graphs @xmath167 by duality @xmath168 with the pairing given by @xmath169 if @xmath170 and zero otherwise . this pre - lie product defines a lie bracket by @xmath171= \gamma_1 \ast \gamma_2 - \gamma_2 \ast \gamma_1 $ ] with @xmath172 given explicitly by @xmath173 [ lma : cop - gamma ] if @xmath5 and @xmath69 are nonempty ( connected ) 1pi graphs , then @xmath174 we have to count the number of subgraphs @xmath163 of @xmath95 such that there is an isomorphism @xmath175 . this isomorphism can be trivial in the sense that there exists an element in @xmath132 mapping @xmath117 to @xmath69 . otherwise , the existence of such a isomorphism implies that there is an isomorphism @xmath102 , with @xmath176 the image in @xmath5 of @xmath177 in the quotient @xmath178 ; such an isomorphism maps @xmath69 in @xmath95 to a certain subgraph @xmath117 of @xmath5 . moreover , @xmath101 $ ] and @xmath136 $ ] are in disjoint @xmath21-orbits , since if @xmath179 , the isomorphism would be the composition of an element in @xmath132 and an element in @xmath21 . we claim that all subgraphs @xmath117 obtained in this way ( for disjoint orbits ) are all different subgraphs of @xmath5 , and can not be the image of @xmath69 under the action of an element in @xmath132 . this would then lead to @xmath180 many subgraphs @xmath117 of @xmath181 satisfying @xmath182 . let @xmath101 , [ x',\phi ' ] , [ x'',\phi '' ] \in x(\gamma,\gamma)$ ] be in disjoint orbits and suppose that there are isomorphisms @xmath183 mapping @xmath69 to subgraphs @xmath117 and @xmath184 in @xmath5 , respectively . if @xmath117 and @xmath184 coincide ( up to an isomorphism @xmath185 ) , then the composition @xmath186 gives an isomorphism from @xmath187 to @xmath188 mapping @xmath69 to itself . it therefore induces an element in @xmath21 that sends @xmath136 $ ] to @xmath189 $ ] , which can not be true . we conclude that @xmath117 and @xmath184 are different subgraphs of @xmath5 . on the other hand , if there is an element @xmath116 in @xmath132 that maps @xmath69 to such a subgraph @xmath190 , the composition @xmath191 would map @xmath192 to @xmath95 isomorphically , sending @xmath69 to itself . again , such a map must be induced by an element in @xmath21 mapping @xmath101 $ ] to @xmath136 $ ] , contradicting our assumptions . [ lma : defect ] let @xmath193 be as above . then , @xmath194 where @xmath195 is defined by @xmath196 \in x(\gamma , \gamma ' ) } \frac{\sym(\gamma \circ_{(y,\psi ) } \gamma')}{\sym(\gamma\gamma ' ) } \delta_{\gamma \circ_{(y,\psi ) } \gamma'}.\ ] ] consider @xmath197 on a 1pi graph @xmath5 ; if @xmath69 and @xmath117 appear as disjoint subgraphs of @xmath5 , this expression is given by @xmath198 , up to a factor of @xmath199 which corrects for the overcounting . indeed , let @xmath200 denote all subgraphs of @xmath5 that are isomorphic to @xmath69 . if @xmath201 , then @xmath202 leading precisely to the factor @xmath203 . on the other hand , if @xmath204 , then both terms vanish . in the case that @xmath5 contains a subgraph @xmath205 such that @xmath206 , we find a discrepancy between the two terms which is given by the following sum , @xmath207 here @xmath208 is by definition the number of disjoint subgraphs of @xmath209 that are isomorphic to @xmath117 and such that @xmath210 , which do indeed all contribute to @xmath198 . we replace the above sum by a sum over insertion places of @xmath117 in @xmath69 , while correcting for the equivalent insertions . the latter correcting factor is given as the number of elements @xmath211 \in x(\gamma,\gamma')$ ] such that @xmath212 . such an isomorphism can be induced by an element @xmath213 , with @xmath211 = g[y,\phi]$ ] but leaving @xmath117 untouched , leading to a factor of @xmath214|$ ] . the number of isomorphisms @xmath212 that are not induced by such an element , is given precisely by the factor @xmath215 . thus , on inserting the expression for @xmath216 derived in lemma [ lma : cop - gamma ] , we infer that , @xmath217 \in x(\gamma,\gamma ' ) } \frac{m(y,\psi)n(y,\psi)}{n(y,\psi ) |\aut(\gamma)[y,\psi]| } ~ \delta_{\gamma\circ_{(y,\psi ) } \gamma ' } ( \gamma)\\ & = \sum_{[y,\psi ] \in x(\gamma,\gamma ' ) } \frac{\sym(\gamma \circ_{(y,\psi ) } \gamma')}{\sym(\gamma\gamma ' ) } ~ \delta_{\gamma\circ_{(y,\psi ) } \gamma'}(\gamma),\end{aligned}\ ] ] where we have applied lemma [ lma : orbit ] in going to the second line . we have also used equation to replace @xmath218 by @xmath219 . our main result of this section is the following . [ prop : linear ] the coproduct takes the following form on the 1pi green s functions : @xmath220 where the sums are over all 1pi graphs @xmath5 with the indicated residue and loop number , and graphs @xmath69 at the indicated loop order that are disjoint unions of 1pi graphs . since @xmath221 , this would follow from the following equality , for @xmath69 any disjoint union of 1pi graphs at loop order @xmath222 and @xmath223 an auxiliary graph , @xmath224 indeed , putting @xmath225 and summing over @xmath69 then gives the desired result . we show that equation holds by induction on the number of connected components of @xmath69 . [ lma : ind1 ] if @xmath69 is a 1pi graph , then equation holds . if @xmath226 , there is nothing to prove , since @xmath227 , @xmath228 and @xmath229 . we claim that the following equality holds for @xmath230 , @xmath231\in x(\tilde\gamma,\gamma ) } \hspace{-1 mm } \frac{{\tilde\gamma \circ_{(x,\phi ) } \gamma~|~\gamma_0 } } { |\gamma|_\vee n(x,\phi ) | \aut(\tilde\gamma)[x,\phi ] |\sym(\tilde\gamma \circ_{(x,\phi ) } \gamma ) } ~ n(\tilde \gamma \circ_{(x,\phi ) } \gamma , \gamma , \tilde\gamma).\ ] ] indeed , one can replace the sum on the left - hand - side over @xmath5 by a sum over insertion places of @xmath69 in @xmath232 ( so that @xmath233 for some @xmath101 \in x(\gamma,\gamma)$ ] , and also @xmath234 ) , provided one divides by a combinatorial factor counting the number of equivalent insertions . this factor is given as the number of elements @xmath136 \in x(\gamma,\gamma)$ ] such that @xmath235 , in which case @xmath236 and also @xmath237 . such an isomorphism @xmath235 can be induced by an element in @xmath238 with @xmath136 = g[x,\phi]$ ] but leaving @xmath69 untouched . this leads to division by the length of the orbit @xmath239 $ ] . otherwise , an isomorphism from @xmath240 to @xmath241 has to map @xmath69 to an isomorphic subgraph @xmath242 . in that case , it can not be induced by an element in @xmath243 , leading precisely to the additional factor of @xmath133 . equation now follows directly by inserting the expressions obtained in lemma [ lma : orbit ] and [ lma : cop - gamma ] in the above equation and summing over all 1pi graphs @xmath232 , as in equation . we also noted on the way that by definition @xmath244\in x(\tilde\gamma,\gamma ) } { \tilde\gamma \circ_{(x,\phi ) } \gamma~|~\gamma_0 } = { \tilde\gamma~|~{\gamma~|~\gamma_0}}.\ ] ] the case @xmath245 arises whenever @xmath246 and @xmath247 , in which case the combinatorial factors @xmath248 and @xmath249 coincide . assume now that equation holds for @xmath69 a ( non - empty ) disjoint union of 1pi graphs of loop order @xmath250 . we will prove that it also holds for the disjoint union @xmath251 of it with a non - empty 1pi graph @xmath117 of loop order @xmath252 . an application of lemma [ lma : defect ] yields , @xmath253 since @xmath117 is a 1pi graph , we can apply lemma [ lma : ind1 ] to the first term , which gives for the sum over all graphs @xmath254 , @xmath255 using also equation . the induction hypothesis that is , validity of eq . in the case of @xmath69 now yields , @xmath256 combining once more the symmetry factors @xmath257 and @xmath258 with the help of @xmath259 . for the second term in equation , we can use the induction hypothesis on @xmath260 to show that @xmath261 \in x(\gamma,\gamma ' ) } \frac{{\gamma~|~{\gamma\circ_{(y,\psi ) } \gamma'~|~}}{\gamma_0}}{|\gamma'|_\vee\sym(\gamma\gamma ' ) \sym(\gamma ) } ~ \gamma,\end{aligned}\ ] ] since @xmath262 . we conclude with the following equality , @xmath263 \in x(\gamma , \gamma')}{\gamma~|~{\gamma\circ_{(y,\psi ) } \gamma'~|~\gamma_0}},\ ] ] which follows easily from definition [ defn : ins3 ] . indeed , by definition @xmath264 \in x(\gamma , \gamma ) } \sum_{[x',\phi ' ] \in x(\gamma \circ_{(x,\phi ) } \gamma , \gamma ' ) } \hspace{-2 mm } { ( \gamma \circ_{(x,\phi ) } \gamma ) \circ_{(x',\phi ' ) } \gamma'~|~\gamma_0},\end{aligned}\ ] ] which counts the number of places to insert @xmath265 and then @xmath223 in @xmath5 . subtraction of the number of such places with @xmath117 sitting inside @xmath69 , leads precisely to the number of places to subsequently insert @xmath266 and @xmath223 in @xmath5 . this section will be a warming - up of what is to come in the next section concerning non - abelian gauge theories . quantum electrodynamics is an abelian gauge theory , and as a consequence , the slavnov taylor identities ( cf [ defn : st ] below ) become much more simple . more precisely , they become linear in the graphs and also known as ward identities @xcite ( see defn . [ defn : ward ] below ) . we first make some observations about feynman graphs in qed . we then proceed to prove compatibility of these ward identities with the hopf algebra of renormalization . in ( massless ) quantum electrodynamics , there is only the vertex of valence three , describing the interaction of the photon with a pair of electrons . there are two types of edges corresponding to the photon ( wiggly edge ) and the electron ( straight edge ) . summarizing , we have in the notation of the previous section : @xmath1 with @xmath267 in particular , this means that in the process of renormalization , only three types of graphs are of importance : the vertex graph , the electron self - energy graph and the vacuum polarization . correspondingly , we define the following 1pi green s functions , @xmath268 with @xmath269 , and where the sum is over all 1pi feynman graphs with the indicated residue . when this sum is restricted to 1pi graphs with loop number @xmath61 , we denote this _ green s function at loop order @xmath61 _ by @xmath270 . in particular , @xmath271 is understood as the 1pi graph with loop number zero , which is the empty graph ; hence @xmath272 . the ( regularized ) feynman rules can be understood as an algebra map @xmath116 from @xmath63 to the field @xmath250 of laurent series in the regularization parameter @xcite . under this map , the above green s functions are mapped to the corresponding feynman amplitudes . in particular , we see that the ( unrenormalized ) effective action can be written as : @xmath273 a simplification in qed , is that the number @xmath75 of insertion places only depends on the number of loops , the residue of @xmath5 and @xmath69 . let @xmath274 denote the number of electron lines of @xmath74 and @xmath275 the number of photon lines . thus , if @xmath5 has residue @xmath74 , then @xmath274 is the number of external electron lines of @xmath5 and @xmath275 the number of its external photon lines . also recall the notation @xmath276 for the number of vertices / edges in @xmath5 of type @xmath277 , introduced just below definition [ defn : ins ] . the number of loops @xmath61 of the graph @xmath5 is then given by ( see for instance ( * ? ? ? * section 10.1 ) ) , @xmath278 where the number of vertices @xmath279 in @xmath5 with residue @xmath74 can be expressed as , @xmath280 combining these equalities , we arrive at the following relations , @xmath281 from which we infer that the number @xmath75 can be expressed in terms of @xmath61 , @xmath74 and @xmath69 . we denote this number by @xmath282 . explicitly , we have for example , @xmath283 in qed , there are relations between the ( amplitudes of the ) green s functions @xmath284 and @xmath285 ; these relations play an important role in the process of renormalization . we will show here that they can be implemented on the hopf algebra , in a way that is compatible with the coproduct . [ defn : ward ] the ward elements @xmath286 at loop order @xmath287 are defined by @xmath288 where the sum is over 1pi feynman graphs with loop number @xmath61 and with the indicated residue . moreover , we set @xmath289 . before stating our main result on the compatibility of the ward elements with the coproduct , we introduce the following combinatorial factor , @xmath290 note the mixture between the two factors @xmath291 and @xmath292 of equation above . if the vertex graphs in @xmath69 are labeled as @xmath293 , one could also define @xmath294 recursively by , @xmath295 for a vertex graph @xmath296 while setting @xmath297 if @xmath298 . [ thm : ward ] for any @xmath299 , we have @xmath300 consequently , the ideal @xmath301 generated by the ward elements @xmath286 for every @xmath61 is a hopf ideal in @xmath63 , @xmath302 essential for the proof will be the relation between the two factors @xmath303 and @xmath304 displayed in equation . using pascal s rule , eq . , we derive the following relation between the two numbers : @xmath305 before inserting this in the expression for @xmath306 derived in proposition [ prop : linear ] , we observe that the second term is just @xmath307 with @xmath209 the graph @xmath69 with one vertex graph @xmath308 subtracted , _ times a factor of _ @xmath309 . similarly , the third term is @xmath307 where now @xmath205 is @xmath69 with one electron self - energy graph @xmath310 subtracted , _ times a factor of _ @xmath311 . note that in the respective cases @xmath298 and @xmath312 , the above two terms vanish . we then find that , @xmath313 the above factors @xmath314 and @xmath315 are precisely canceled when replacing the sum over @xmath69 by a sum over @xmath310 and @xmath205 . indeed , in doing so , a factor of @xmath316 is needed to correct for overcounting , @xmath317 by the very definition of @xmath318 , and a similar result holds for @xmath308 . we thus have respective factors of @xmath314 and @xmath315 in the denominator , and combining @xmath319 with @xmath320 ( and similarly for the analogous expression for @xmath308 ) using equation yields , @xmath321 in this section , we come to the main purpose of this article , and show compatibility of the slavnov taylor identities with the hopf algebraic structure of renormalization of non - abelian gauge fields . before that , we carefully describe the setting of non - abelian gauge theories . in particular , we start by describing the graphs that are allowed in such theories and list some combinatorial properties of the number @xmath75 of insertion places defined in section [ sect : ins ] . these properties will be essential in the proof of compatibility of the slavnov taylor identities with the coproduct . in order to make the following as concrete as possible , we work in the setting of the non - abelian gauge theory _ quantum chromodynamics _ ( qcd ) . it describes the interaction between quarks ( the fermions ) via gluons ( the gauge bosons ) . in contrast with quantum electrodynamics described previously , there are now three vertices of valence three , describing the interaction of the fermion and ghost with the gluon , as well as the cubic gluon self - interaction . in addition , there is the quartic gluon self - interaction . this means that the feynman graphs are built from the following two sets of vertices and edges : @xmath322 where the plain , dotted and curly lines represent the quark , ghost and gluon , respectively . corresponding to the elements in @xmath0 , we define 7 ( 1pi ) green s functions , @xmath323 with the sum over all 1pi feynman graphs with the indicated residue , i.e. @xmath324 . [ rem : eff - action ] as for quantum electrodynamics , the ( regularized ) feynman rules for qcd ( as listed for instance on page 34 of @xcite ) can be understood as an algebra map @xmath116 from @xmath63 to the field @xmath250 of laurent series in the regularization parameter @xcite . the ( unrenormalized ) effective action can be written as , @xmath325 contrary to the case of qed ( cf . [ subsect : graphs - qed ] ) , there are no relations in qcd expressing the numbers @xmath326 of vertices / edges of @xmath5 of the type @xmath74 in terms of the loop number and the residue of a graph @xmath5 . however , we do have relations between them . denote by @xmath327 the number of lines attached to @xmath74 that are of the type @xmath328 ; by convention , @xmath329 for an edge @xmath330 . set @xmath331 , it is the total number of lines attached to @xmath74 . [ lma : m ] let @xmath5 be a ( qcd ) 1pi feynman graph at loop order @xmath61 and with residue @xmath74 . then , the following relations hold @xmath332 equation ( a ) is the usual expression of the betti number of the graph @xmath5 ; it can be proved by induction on the number of edges of the graph @xmath5 . equations ( b ) and ( d ) follow from the statement that each of these two vertices involve exactly two quark or two ghost lines , respectively . we prove that equation ( c ) holds by induction on the number of edges . first , denote by @xmath333 and @xmath334 the number of all trivalent and quadrivalent vertices of @xmath5 , respectively . since the type of vertices and edges is irrelevant for eq . ( c ) , we have to show validity of , @xmath335 for a graph @xmath5 consisting of @xmath333 trivalent and @xmath334 quadrivalent vertices of a single fixed type . induction on the number of edges starts with the graph consisting of only one edge , for which @xmath336 , by definition . then , if we add an edge to a graph @xmath5 for which eq . holds , we have the following possibilities to connect the new edge : [ cols="<,>,>,>,>",options="header " , ] acting by @xmath337 , while retaining admissibility at each step . clearly , two admissible vectors differ by integers in each of the entries @xmath338 for @xmath339 . since equations ( a)-(d ) in lemma [ lma : m ] impose linear constraints on the entries of @xmath340 , it is enough to consider the cases @xmath341 for each of the three edges individually ( thus leaving the other @xmath338 invariant ) . then , equations ( a)-(d ) become four independent constraints on the four remaining entries @xmath342 , @xmath343 , the solution of which is displayed in the above table . contrary to the linear ward identities that we have encountered in qed , there are now quadratic relations between green s functions that reflect the non - abelian nature of the gauge symmetry of the corresponding classical field theory . for a derivation of the slavnov taylor identities , we refer the reader to the standard text books on quantum field theory , such as ( * ? ? ? * sect . 17.1 ) or ( * ? ? ? 21.4 - 5 ) . pictorially , we have the following three identities : @xmath344 where the blob stands for the 1pi green s function corresponding to the indicated external structure . in the hopf algebraic setting of renormalization , this motivates the following definition . [ defn : st ] the slavnov taylor elements are defined by @xmath345 note that these elements involve both a linear and quadratic part , for instance,@xmath346 the choice for using the same labelling @xmath347 for the slavnov taylor identities as for the admissible steps in lemma [ lma : steps ] is not coincidental , but motivated by the next lemma . first , define for vectors @xmath348 and @xmath349 the following constant , @xmath350 a glance back at section [ sect : ins ] , makes one realize that whenever @xmath351 and @xmath352 arise from two graphs @xmath5 and @xmath69 ( i.e. @xmath353 ) , this constant becomes the number @xmath75 of insertion places of @xmath69 in @xmath5 . we also introduce the following standard basis @xmath354 of vectors corresponding to the elements @xmath59 by setting @xmath355 . [ lma : steps - st ] the following equation holds for @xmath356 , @xmath357 & = \st^a ~ \sum_\gamma c \begin{pmatrix } \vec{m } - \vec{f}_\gluq \\ \vec{n } \end{pmatrix } \frac{\gamma}{\sym(\gamma)}~.\end{aligned}\ ] ] first observe that the cases @xmath358 and @xmath359 can be treated simultaneously by exchanging external electron lines with external ghost lines . after applying pascal s rule eight times , we obtain @xmath360 now , the analogue of equation in the setting of qcd yields , when applied to these six terms , precisely the above formula in the case @xmath361 ( and similarly for @xmath362 ) . in like manner , one shows that @xmath363 from which , with the proper analogue of eq . , one concludes the above formula for @xmath364 . the above lemma will be essential in the proof of our main result , stating compatibility between the slavnov taylor elements and the hopf algebra structure of renormalization . [ thm : st ] the ideal @xmath301 generated by the slavnov taylor elements is a hopf ideal , i.e. @xmath365 by the very definition of the ideal @xmath301 the second condition is trivially satisfied whereas the third follows from the first , by the recursive formula for the antipode in equation . it is thus enough to show that @xmath366 for @xmath356 . clearly , the primitive part of the coproduct @xmath367 is of the desired form , by the very definition of being the primitive part . we will apply proposition [ prop : linear ] to determine the non - primitive part . starting with the second term in @xmath368 , we find that , @xmath369 the first two terms combine , since [ cop : gluq - qua - terms ] @xmath370 \frac{\gamma}{\sym(\gamma ) } \nn \\ & = \sum_\gamma c \begin{pmatrix } \vec{m}_{\gamma_1 } - \vec{f}_\qua \\ \end{pmatrix } \frac{\gamma}{\sym(\gamma)}~,\end{aligned}\ ] ] by an application of pascal s rule . on the other hand , the second two terms contribute with @xmath371 finally , the quadratic terms becomes @xmath372 using vandermonde s identity . unfortunately , the three coefficients in eq . do not coincide , and it is thus not immediately clear how to combine the terms in eq . in order to obtain @xmath368 on the second leg of the tensor product . however , we claim that they differ by a finite number of combinations of the steps , and , so that they can be combined _ at the cost of adding terms in _ @xmath373 , exactly those that appear in lemma [ lma : steps - st ] . indeed , in the case of one rewrites trivially , @xmath374 and observes that the vector @xmath375 is @xmath376-admissible if @xmath69 is at loop order @xmath250 . also , regarding eq . , one notes that @xmath377 is @xmath376-admissible . since @xmath378 is @xmath376-admissible by definition , an application of lemma [ lma : steps ] in combination with lemma [ lma : steps - st ] shows that all terms in eq . can be combined to give @xmath379 _ modulo _ elements of the form @xmath380 for some @xmath381 . a completely analogous argument shows that the same conclusion holds for @xmath382 and @xmath383 . after having derived the necessary combinatorial identities , and obtained a formula for the coproduct on 1pi green s functions , we have showed that the slavnov taylor identities in quantum chromodynamics can be implemented on the hopf algebra of feynman graphs as relations defining a hopf ideal . since the map @xmath384 defined by the ( regularized ) feynman rules ( cf . remark [ rem : eff - action ] ) vanishes on this ideal , it factors through an algebra map from the quotient hopf algebra @xmath385 to the field @xmath250 of laurent series in the regularization parameter . since @xmath386 is still a commutative ( but non - free ) connected hopf algebra , there is a birkhoff factorization @xmath387 ( * ? ? ? * theorem 4 ) for the convolution product @xmath172 in @xmath388 . the two algebra maps @xmath389 are given on 1pi graphs by the following recursive formula , @xmath390 where @xmath391 is the projection on the pole part in @xmath250 . it was realized in @xcite that @xmath392 and @xmath393 precisely give the renormalized feynman amplitude and the counterterms , respectively , corresponding to the graph @xmath5 . since they are algebra maps from @xmath386 to @xmath250 , we conclude that they automatically satisfy the slavnov taylor identities . moreover , the compatibility of the slavnov taylor identities with the hopf algebra structure implies validity of the ` gauge theory theorem ' in ( * ? ? ? 5 ) ( see also @xcite ) . the latter states that there exists a certain sub hopf algebra and that the map @xmath394 , where @xmath395 is a normalized version of @xmath33 defined just above eq . , is a closed hochschild cocycle . the author would like to thank matilde marcolli for several discussions . we give an alternative proof of the above theorem stating compatibility of the ward identities with the coproduct , which is based on our previous result @xcite on the compatibility of the so - called _ ward takahashi identities _ with the coproduct . such ward takahashi identities are identities between individual graphs , and an expression of the above ward elements @xmath286 in terms of the ward takahashi elements ( to be introduced below ) aloows us to deduce theorem [ thm : ward ] from this result . let us start by introducing a map similar to @xmath154 introduced in the section [ sect : ins ] . for a disjoint union of 1pi graphs @xmath69 and a 1pi electron self - energy graph @xmath310 , we define on a 1pi graph @xmath5 , @xmath396 where @xmath397 is the graph @xmath398 with an external photon line attached to the edge @xmath330 corresponding to @xmath399 in the quotient . in the case that @xmath69 is the empty graph , this map will be denoted by @xmath400 and when in addition @xmath401 , we set @xmath402 . two examples that might help the reader to see what is going on are , @xmath403 the analogue of @xmath162 is given by @xmath404 ; it is defined to be the number of subgraphs @xmath117 of the electron self - energy graph @xmath5 that are isomorphic to the electron self - energy graph @xmath69 , and such that @xmath405 is isomorphic to the vertex graph @xmath23 . we also introduce the following combinatorial factor , to be used below . the compatibility of the so - called ward takahashi identities with the coproduct was derived in @xcite and will be recalled in proposition [ prop : wt ] below ; for completeness , we also restate its proof . it gives relations between individual graphs , and will be used in the alternative proof of theorem [ thm : ward ] , involving relations between green s functions . we first recall some of the notation in that paper . given any electron self - energy graph @xmath5 , we can label the internal electron edges from @xmath406 to @xmath407 . if we fix such a labelling , we denote by @xmath408 the graph @xmath5 with an external photon line attached to the electron line @xmath409 . [ prop : wt ] let @xmath410 be a 1pi electron self - energy graphs and define the corresponding ward takahashi element by @xmath411 , with the sum over internal electron lines in @xmath5 , moreover , set @xmath412 . then , @xmath413 where the sum is over disjoint unions of 1pi graphs @xmath69 ( including the empty graph ) and 1pi electron self - energy graphs @xmath310 . we start by computing @xmath414 ; we split the sum over subgraphs into two terms , those for which @xmath69 contains the electron line @xmath409 , and those for which it does not , @xmath415 where @xmath416 denotes the disjoint union of the electron self - energy graphs in @xmath69 . the absence of terms in which @xmath417 is due to the fact that there are no vertices with more than one photon line , so that the corresponding graph @xmath418 does not appear as a subgraph of @xmath408 in the coproduct . the first term can be written as , @xmath419 by decomposing the sum over all @xmath409 that are part of the subgraph @xmath416 into the connected components @xmath310 of @xmath416 , thereby also noting that if @xmath420 , the quotient @xmath421 is nothing else then @xmath422 . on the other hand , the second term can be split into two parts , one for which @xmath409 is an external electron edge for one of the electron self - energy graphs in @xmath69 , and one for which it is not , @xmath423 here we used the fact that although an electron self - energy graph has two external electron edges , they might be in common with another electron self - energy graph that is also part of @xmath69 . this means that for a block of electron self - energy graphs that sit inside @xmath5 by concatenation , there is a term @xmath424 for each of such graphs , plus one for the closing `` fence pole '' . indeed , the number of external electron edges of such a block is precisely the number of its constituents plus one . the extra term combines with the second term , since it precisely gives the term @xmath425 with @xmath426 the electron edge corresponding to @xmath310 so as to complete the sum over all internal electron edges in the last term . [ eq : w - wt ] there are the following relations between @xmath427 and the ward elements @xmath286 , @xmath428 since every vertex graph can be written as @xmath408 for an electron self - energy graph @xmath5 , we only have to check that the symmetry factors turn out right . we first observe that @xmath429 coincides with the order of the isotropy group @xmath430 of the electron edge @xmath409 in @xmath21 . moreover , two graphs @xmath408 and @xmath431 are isomorphic if and only if @xmath409 and @xmath432 are in the same orbit under the action of @xmath21 . an application of the orbit - stabilizer theorem shows that such an orbit has length @xmath433 , so that , @xmath434 from which the result follows at once . before continuing our derivation of theorem [ thm : ward ] from proposition [ prop : wt ] , we derive a result anologous to lemma [ lma : defect ] for the map @xmath435 . instead of the insertion of graphs of definition [ defn : ins ] that appeared in lemma [ lma : defect ] , it involves insertion of an electron self - energy graph into a vertex graph , defined as follows . [ defn : insertion ] an insertion of a 1pi electron self - energy graph @xmath69 into a 1pi vertex graph @xmath5 is given by an isomorphism @xmath116 from @xmath91 to the two electron lines connected to the vertex to which the external photon line is attached , after removal of this photon line . the resulting electron self - energy graph is denoted by @xmath436 . for example , @xmath437 [ lma : defect - e ] let @xmath69 and @xmath310 as above and let @xmath314 denote the number of 1pi vertex graphs in @xmath69 as before . then , @xmath438 where @xmath439 is defined in terms of the vertex graph @xmath293 in @xmath69 by , @xmath440}\frac{n^e(\gamma_l { ~\#_\phi~ } \gamma_e,\gamma_e,\gamma_l)}{n(\gamma-\gamma_l,\gamma_l ) } \frac{|\gamma_l { ~\#_\phi~ } \gamma_e|_\vee}{|\gamma_l|_\vee } ~ \delta^e_{\gamma-\gamma_l , \gamma_l { ~\#_\phi~ } \gamma_e}.\ ] ] similar to the proof of lemma [ lma : defect ] , we introduce a map @xmath441 ; it corrects for the overcounting in the cases that there is a subgraph @xmath205 of @xmath5 containing @xmath442 , such that @xmath443 . indeed , quotienting @xmath5 by such a @xmath205 does not appear in @xmath444 , although it does in @xmath445 . we then write , @xmath446 } \frac{n^e(\gamma_l { ~\#_\phi~ } \gamma_e,\gamma_e,\gamma_l)}{n(\gamma-\gamma_l,\gamma_l ) } \frac{|\gamma_l { ~\#_\phi~ } \gamma_e|_\vee}{|\gamma_l|_\vee } ~\delta^e_{\gamma-\gamma_l,\gamma_l { ~\#_\phi~ } \gamma_e}(\gamma),\end{aligned}\ ] ] since each such graph @xmath205 has to be isomorphic to the graph obtained by inserting @xmath310 ( using @xmath447 ) into one of the vertex graph components @xmath448 of @xmath69 , with an additional factor of @xmath449 due to the multiplicity arising on the left - hand - side . on the other hand , division by @xmath450 corrects for insertions of @xmath310 into isomorphic @xmath451 ; indeed , if @xmath452 then clearly @xmath453 . finally , the factor @xmath454 arises from the difference of summing over subgraphs isomorphic to @xmath451 and subgraphs isomorphic to @xmath455 . for any @xmath299 , we have @xmath300 consequently , the ideal @xmath301 generated by the ward elements @xmath286 for every @xmath61 is a hopf ideal in @xmath63 , @xmath302 firstly , with @xmath286 being related to @xmath427 by the above eq . ( and understood to be zero for @xmath456 ) , proposition [ prop : wt ] implies that @xmath457 the second term can be easily written in the desired form since with proposition [ prop : linear ] , @xmath458 which is indeed an element in @xmath459 . for the first term , we derive the result from the following equation , @xmath460 for two graphs @xmath69 and @xmath310 at respective loop order @xmath252 and @xmath250 . in fact , validity of this equation implies that @xmath461 which , on its turn , is an element in @xmath373 . we prove that equation holds by induction on the number @xmath314 of 1pi vertex graphs in @xmath69 . if this number is zero , lemma [ lma : defect - e ] takes a simple form and we conclude that in this case , @xmath462 since for each 1pi vertex graph @xmath232 , there are precisely @xmath463 many electron self - energy graphs @xmath5 that result in @xmath232 after quotienting by a subgraph isomorphic to @xmath310 and connecting an external photon line to the corresponding internal electron line . thus , an application of equation shows validity of eq . , since by definition @xmath464 . let us then suppose that eq . holds for @xmath465 and number the vertex graphs in @xmath69 by @xmath73 . if @xmath296 is another 1pi vertex graph , then lemma [ lma : defect - e ] yields @xmath466 } \frac{n^e(\gamma_l { ~\#_\phi~ } \gamma_e,\gamma_e,\gamma_l ) } { |\gamma\gamma_{n+1}-\gamma_l|_\vee |\gamma_l { ~\#_\phi~ } \gamma_e|_\vee \gamma_e|_\vee \sym(\gamma ) n(\gamma\gamma_{n+1}-\gamma_l,\gamma_l ) } \sum_{[\phi ] } \delta^e_{\gamma\gamma_{n+1}-\gamma_l , \gamma_l { ~\#_\phi~ } \gamma_e } ( \gamma),\end{gathered}\ ] ] after insertion of the definition of @xmath467 and an application of definition [ defn : triv - perm ] . the first term can be reduced by applying eq . to @xmath468 and eq . to @xmath469 . for the second term , we can apply the induction hypothesis , since @xmath470 has @xmath33 1pi vertex graph components . with the following two equalities for the combinatorial factors : @xmath471 which can be obtained from eq . and the observation that @xmath472 counts precisely the automorphisms of @xmath473 that do not come from an automorphism of @xmath451 or @xmath310 , we derive @xmath474 note that the sum over @xmath475 $ ] has been cancelled against the factor @xmath476 in the denominator . a glance back at the recursive definition of @xmath477 in equation then completes the proof . for completeness , we list some combinatorial identities used throughout the text . firstly , there is the well - known pascal s rule :

infrared images of the q 0957 + 561 gravitational lens obtained with the hubble space telescope show two large ( ) lensed images of the quasar host galaxy . parts of the host galaxy are doubly - imaged like the quasar , while other parts are quadruply - imaged . the distortions of the host galaxy offer the best probe yet of the global structure of the lensing potential , which is essential for determining the hubble constant from the measured time delay . the distortions are inconsistent with the predictions of previously published lens models , which invalidates those models and their implications for . new models show that the distortions finally break the long - standing degeneracy between the shape of the lens galaxy and the tidal shear contributed by the cluster containing the lens galaxy . the shape of the lens galaxy s mass distribution must be remarkably similar to the shape of its luminosity distribution , and most models that produce reasonable values for the hubble constant roughly match the observed ellipticity gradient and isophote twist of the lens galaxy . also , the cluster must be non - spherical and produce a relatively small tidal shear . although there are still degeneracies in the lens models that lead to a 25% uncertainty in the derived value of the hubble constant , there are also strong prospects for new observations to further improve the constraints and reduce the uncertainties . psfig

astro-ph0001500

gravitational lenses offer an attractive independent method of determining the hubble constant ( @xmath2 ) on cosmological scales without the systematic difficulties associated with the local distance ladder ( refsdal 1964 , 1966 ) . six of the more than 50 known gravitational lenses now have time delays that can be used to estimate @xmath2 : b 0218 + 357 ( biggs et al . 1999 ) ; q 0957 + 561 ( schild & thomson 1995 ; kundi et al . 1997 ; haarsma et al.1999 ) ; pg 1115 + 080 ( schechter et al . 1997 ; barkana 1997 ) ; b 1600 + 434 ( hjorth et al . 1999 ) ; b 1608 + 656 ( fassnacht et al.1999 ) ; and pks 1830@xmath3211 ( lovell et al . 1998 ) ; preliminary @xmath2 estimates for five of these systems have been compiled by koopmans & fassnacht ( 1999 ) . once a time delay is accurately determined , the uncertainties in the derived value of the hubble constant are due almost entirely to the systematic uncertainties in the model for the lensing potential , plus `` cosmic variance '' due to the effects of weak density inhomogeneities along the line of sight ( seljak 1994 ; barkana 1996 ) . only two of the time delay lenses have been modeled in sufficient detail to fully understand the systematic uncertainties created by their geometries : q 0957 + 561 ( falco , gorenstein & shapiro 1985 , 1991 ; kochanek 1991 ; bernstein , tyson & kochanek 1993 ; grogin & narayan 1996 ; chartas et al . 1998 ; barkana et al . 1999 ; bernstein & fischer 1999 ; chae 1999 ; romanowsky & kochanek 1999 ) , and pg 1115 + 080 ( schechter et al . 1997 ; courbin et al . 1997 ; keeton & kochanek 1997 ; saha & williams 1997 ; impey et al . 1998 ) . in q 0957 + 561 , the lens consists of a brightest cluster galaxy and its parent cluster , and the value of the hubble constant depends on the mass balance between the two components . judgments about the ability of models to determine the correct mass balance vary significantly : optimistic estimates yield @xmath4 at 95% confidence ( grogin & narayan 1996 , using the stellar dynamical models of romanowsky & kochanek 1999 ) ; while more pessimistic estimates yield @xmath5 ( kochanek 1991 ; bernstein et al . 1993 ; bernstein & fischer 1999 ) . in pg 1115 + 080 , the lens is one of the brighter galaxies in a small group , but the four - image geometry easily determines the relative roles of the primary lens galaxy and the group . instead , the value of the hubble constant depends on the assumed radial mass distribution of the primary lens galaxy . galaxy models with dark matter and mass distributions consistent with the best estimates for early - type galaxies lead to low values for the hubble constant ( @xmath6 ; impey et al . 1998 ) . impey et al . ( 1998 ) discovered an einstein ring image of the quasar host galaxy in the pg 1115 + 080 lens , and noted that the geometry of the ring could be used to break the degeneracy in the lens models . here we report the discovery of the host galaxy in the q 0957 + 561 lens and discuss the implications for lens models and the hubble constant . in 2 we describe our observations . in 3 we discuss the data and methods used to model the system , summarizing previous work and introducing our new data and techniques . in 4 we examine the arcs predicted by previously published lens models and demonstrate that they fail to match the observed arcs . in 5 we present new models that match all data , including the arcs , and discuss their implications for the properties of the lens galaxy and cluster and for @xmath2 . in 6 we summarize our results and discuss prospects for further improving the constraints on this lens system and on @xmath2 . we observed q 0957 + 561 with the hubble space telescope as part of the cfa arizona space telescope lens survey ( castles ; lehr et al . 1999 ; falco et al . 1999 ) . using the nic2 camera , we obtained a 2800-second f160w ( h band ) image of q 0957 + 561 divided into four dithered exposures . a log of the observations is presented in table 1 . we reduced the images using nicred , a custom reduction package developed for castles ( lehr et al . we also reanalyzed archival wfpc2 images , including a 32200-second f555w ( v band ) exposure and a 2620-second f814w ( i band ) exposure ( bernstein et al . 1997 ) . in the optical images , q 0957 + 561 fell in chip wf3 of the wfpc2 camera ( with pixel size @xmath001 , compared to @xmath00076 for nic2 ) . because the images a and b of q 0957 + 561 were saturated in the wfpc2 images , we derived our astrometry of a , b and the lens galaxy g1 exclusively from our unsaturated nic2 data . figure 1a shows our combined h band image , which prominently shows the lens galaxy and the two quasar images . the quasar host galaxy is visible as a faint arc next to the quasar a image . table 2 summarizes a photometric model consisting of point sources for the two quasar images and elliptical de vaucouleurs models for the main lens galaxy ( g1 ) and a neighboring galaxy ( g2 , not seen in figure 1 ) . the lens galaxy is known to have a small ellipticity gradient and isophote twist ( bernstein et al . 1997 ) , but we used only a simple elliptical model . figure 1b shows the residuals after subtracting the psf - convolved photometric model from the original image , and figure 1c shows these residuals convolved with a @xmath00076 ( 1 pixel ) fwhm gaussian to enhance the visibility of low surface brightness features . the residual image shows the host galaxy image near quasar a to be an arc distorted tangentially relative to the lens galaxy . it also reveals an extended asymmetric arc near the b quasar image . if the residuals near quasar b were created by the error of fitting a simple elliptical surface brightness model to a galaxy with a radially varying ellipticity and orientation , we would expect them to have reflection symmetry through the center of the galaxy . because the arc near quasar b lacks such symmetry , we conclude that it is the lensed counterpart of the arc near quasar a , and that both are images of the quasar host galaxy . the residual h band image also shows two additional sources , one east of quasar a and another just west of quasar b ( labeled g3 and g4 in figure 1c ) . their positions and photometric properties are given in table 2 . these objects correspond in position and v magnitude to two of the faint sources seen in the deep v band image and labeled blobs 7 and 1 ( respectively ) by bernstein et al . their colors ( especially v@xmath3i ) are similar to those of g1 and g2 and suggest that they are probably faint cluster members . whether or not they are associated with the cluster , it is clear that they are not multiply imaged and are not associated with the quasar source . the fact that g4 is not part of the host galaxy arc will be important to remember when examining the arc structure predicted by lens models ( see 3.3 ) . the lensed host galaxy is not detected at significant levels in the shallow i band image , although a visual examination suggests that low - level residuals are present and should be detected at a significant level with a longer integration . the deep v band image does not have large extended arcs , but it does show seven faint `` blobs '' and a thin arclet ( bernstein et al . 1997 ) . as noted above , at least two of the blobs are probably faint cluster galaxies . however , blobs 2 and 3 and the arclet have been identified as possible lensed features , and lens models support this interpretation ( barkana et al . 1999 ; bernstein & fischer 1999 ; chae 1999 ) . since the v band corresponds to rest - frame uv at the quasar redshift @xmath1 , these features probably correspond to discrete star forming regions in the host galaxy . we follow previous models and include blobs 2 and 3 plus two `` knots '' in the arclet as model constraints ( see 3.2 ) . we estimated the fluxes of the host galaxy images using the iraf task polyphot , which measures fluxes within polygonal apertures . we traced polygonal apertures along the edges of the distorted images of the host galaxy at the @xmath7 level above the sky . we found that both lensed images of the host galaxy have h band brightnesses of @xmath018.4 mag , and a surface brightness of @xmath020.6 mag / arcsec@xmath8 ( both images have approximately the same area ) . the main source of uncertainty for the brightness estimates is the relatively large size of the residuals from the subtraction of the quasar images , compared to the ( low ) brightness levels of the arcs ; we estimate the uncertainties to be @xmath00.3 mag . given an acceptable lens model we can map the host galaxy images to the source plane to obtain a map of the unlensed host galaxy ( see 3.3 ) . in principle we could use the source maps to measure the photometric properties of the host galaxy . however , in practice the imperfect quasar subtraction corrupts the flux in the bright central regions of the source and hinders the measurement . one robust statement we can make is that part of the host galaxy is doubly - imaged like the quasar , but part of it is quadruply - imaged ( see 4 ) . this accounts for the shape differences between the a and b arcs : the a arc is a single distorted image of the host galaxy , while the b arc is a composite of three images that straddle the lensing critical line . q 0957 + 561 is the most thoroughly studied gravitational lens . the system comprises a radio - loud quasar at redshift @xmath1 lensed into two images by a brightest cluster galaxy and its parent cluster at redshift @xmath9 ( walsh , carswell & weymann 1979 ; young et al . 1980 ) . there is a time delay of @xmath10 days between the images ( schild & thomson 1995 ; kundi et al . 1997 ; haarsma et al . vlbi observations have resolved each image into a core and @xmath080 milli - arcsecond jet ( garrett et al.1994 ) . deep optical images have uncovered faint `` blobs '' and `` knots '' that are probably lensed image pairs of star forming regions in the quasar host galaxy ( bernstein et al . our observations have revealed infrared arcs representing distorted images of the quasar host galaxy . in this section we discuss how to use these and other data to constrain models of the system and values for the hubble constant @xmath2 . in 3.1 we review the theory of lens modeling and discuss important model degeneracies . in 3.2 and 3.3 we discuss previous and new observational constraints on models . in 3.4 and 3.5 we describe classes of models applied to the lens galaxy and cluster in q 0957 + 561 , summarizing previous classes and introducing a new one . finally , in 3.6 we discuss how we apply the observational constraints to our new class of models . in a multiply - imaged gravitational lens system , the light from a distant source is deflected by the gravitational potential of foreground objects so that we observe multiple images of the source . the lensing potential is usually dominated by a single galaxy , although there may be non - negligible perturbations from other objects nearby . basic lens theory is presented in the book by schneider , ehlers & falco ( 1992 ) , and we quote the relevant results here . the lensing potential @xmath11 is determined by the two - dimensional poisson equation @xmath12 , where @xmath13 is the surface mass density in units of the critical surface density for lensing ( in angular units ) , @xmath14 where @xmath15 and @xmath16 are angular diameter distances from the observer to the lens and source , respectively , and @xmath17 is the angular diameter distance from the lens to the source . the lensing potential deflects a light ray so that the angular position @xmath18 of the source on the sky and the angular position @xmath19 of an image are related by the lens equation , @xmath20 there is an image corresponding to each solution @xmath21 of this equation . lensing introduces a time delay between the ray paths of two images of the same source . the time delay between images at positions @xmath21 and @xmath22 is @xmath23 , \label{eq : tdel}\ ] ] where @xmath24 is the redshift of the lens . this equation is the basis of attempts to use lensing to determine the hubble constant @xmath2 . by measuring light curves of images one can determine the time delay @xmath25 . a lens model gives the term in square brackets in eq . ( [ eq : tdel ] ) . the combination of distances is @xmath26 and only weakly dependent on other cosmological parameters . a lens model consists of a description of the lensing potential @xmath11 . the observed images provide two principle constraints on such a model . first , roughly speaking an image at projected distance @xmath27 from the main lens galaxy measures the enclosed mass @xmath28 . in q 0957 + 561 and many 2-image gravitational lenses , the two images lie at different distances @xmath29 , so they measure two masses @xmath30 and @xmath31 and hence constrain the mass profile ( e.g. grogin & narayan 1996 ) . second , with at least three well - determined positions ( two images and the lens galaxy , or four images ) , the images determine the quadrupole moment of the net potential . however , there are four properties of the mass distribution that must be determined for a complete description of the model : the mass profile of the main lens galaxy ; the shape ( ellipticity and orientation ) of the main lens galaxy ; the shear from the the gravitational tidal field induced by objects near the lens galaxy ; and the amount of gravitational focusing ( or `` convergence '' ) contributed by the environment of the lens galaxy . with more unknown quantities than constraints , there are two common degeneracies in the lens models . first , to lowest order the convergence @xmath32 contributed by the environment can not be determined by lens models , which leads to the so - called `` mass sheet degeneracy '' ( falco et al . if a lensing potential @xmath33 fits the observed data , then any potential @xmath34 will fit the data equally well . the @xmath32 term is equivalent to the potential from a uniform mass sheet with surface density @xmath35 . the only potentially observable difference between the lens models represented by @xmath11 and @xmath36 is in the predicted time delays , @xmath37 however , this effect is not observable if we want to use gravitational lensing to determine the hubble constant because it simply translates into a scaling of the inferred value for @xmath2 . if @xmath2 and @xmath38 are the values inferred from the two lens models , then @xmath39 this analysis has used only the lowest order term of the contribution from the environment , but the higher order terms can not necessarily eliminate the degeneracy ( e.g. chae 1999 ) . the mass sheet degeneracy is important for q 0957 + 561 because the parent cluster of the lens galaxy contributes a significant convergence . the only way to break this degeneracy is to obtain an independent mass constraint to determine the relative contributions of the main lens galaxy and the environment to the mass enclosed by the images ( see 3.2 ) . the second important degeneracy is between the shape of the main lens galaxy and the shear from the gravitational tidal field of objects in the environment of the lens galaxy . because the images determine only the quadrupole moment of the total potential , there can be a wide range of parameter space in which the lens galaxy ellipticity and the external shear combine to produce the required joint quadrupole ( e.g. keeton et al . consider a 2-image lens with image positions @xmath40 and @xmath41 , a galaxy potential @xmath42 , and an external shear with amplitude @xmath43 and direction @xmath44 . fitting the image positions exactly is equivalent to solving the equation @xmath45 where the shear is described by the tensor @xmath46 .\ ] ] no matter what galaxy potential @xmath42 is used , it is straightforward to solve eq . ( [ eq : shr ] ) for the shear parameters @xmath43 and @xmath44 . in other words , one can fit the image positions in a 2-image lens to arbitrary precision for _ any model lens galaxy _ , although in some cases the inferred shear @xmath43 may be unphysically strong . constraints from the images fluxes or additional images often break this degeneracy , although the @xmath47 range of models may still be large . in previous models of q 0957 + 561 , the additional constraints came primarily from the vlbi observations of the quasar images , which show several discrete components with @xmath00.1 milli - arcsecond errorbars ( garrett et al . 1994 ; barkana et al . it is easy to be misled about the true structure of the potential , however , when using such strong constraints ( see kochanek 1991 ; bernstein et al . 1993 ; bernstein & fischer 1999 ) . internal structure in the lens galaxy , such as isophote twists or lumpiness in the mass distribution , may affect the images at sub - milli - arcsecond scales ( e.g. mao & schneider 1998 ) . neglecting that structure forces the models to adjust large - scale model components in order to fit small - scale constraints , leading them to converge to a best - fit solution that is well defined ( i.e. with no apparent degeneracy ) but incorrect . in q 0957 + 561 , the observed ellipticity gradient and isophote twist suggest that the galaxy does have important internal structure ( bernstein & fischer 1997 ) , and in 4 we show that models neglecting this structure indeed converged to incorrect solutions . in q 0957 + 561 the high - resolution vlbi maps of the quasar images provide strong position constraints , plus somewhat weaker constraints on the relative magnification matrix between the images ( garrett et al . 1994 ; barkana et al . the optical blobs and knots offer additional but weaker position constraints . the knots are particularly useful because they appear to represent a pair of `` fold '' images and thus require the lensing critical line to pass between them ( see bernstein & fischer 1999 ) . the constraints from these images leave two of the three model degeneracies discussed in 3.1 : the mass sheet degeneracy due to the convergence provided by the cluster ; and the degeneracy between the lens galaxy shape and the cluster shear . several steps have been taken to break the mass sheet degeneracy by determining the relative masses of the lens galaxy and cluster . first , rhee ( 1991 ) , falco et al . ( 1997 ) , and tonry & franx ( 1999 ) measured the lens galaxy s central velocity dispersion to be @xmath48 . romanowsky & kochanek ( 1999 ) used stellar dynamical models to translate this into constraints on the lens galaxy mass and concluded that @xmath49 of the image separation is contributed by the galaxy ( and the rest by the cluster convergence ) . second , fischer et al . ( 1997 ) detected a weak lensing signal from the cluster , which bernstein & fischer ( 1999 ) translated into an estimate of the mean surface density of all mass inside an aperture of radius 30 centered on the lens galaxy , @xmath50 furthermore , bernstein & fischer ( 1999 ) showed that the surface mass density @xmath51 of the cluster at the position of the lens galaxy is given by @xmath52 where @xmath53 is the mean mass density of a model lens galaxy . in this way one can combine the weak lensing measurement with a lens model to estimate @xmath54 ( see 5.3 ) . there have been other attempts to measure the cluster mass , using the velocity dispersion of the galaxies in the cluster ( @xmath55 ; garrett , walsh & carswell 1992 ; angonin - willaime , soucail & vanderriest 1994 ) or the x - rays from the cluster gas ( chartas et al . 1998 ) , but they are still limited by significant systematic uncertainties . although measurements of the cluster surface density are important ( especially for determining @xmath2 , see 5.4 ) , they do not break the degeneracy between the lens galaxy shape and the cluster shear . another approach is to constrain the shape of the lens galaxy s mass distribution . the orientation of the lens galaxy s light distribution should offer a guide to the orientation of its mass distribution , because other lenses suggest that mass and light are at least roughly aligned ( typically within @xmath010 ; keeton , kochanek & falco 1998 ) . an important detail , though , is that the lens galaxy has a radially varying ellipticity and orientation ( bernstein et al . 1997 ) : the ellipticity increases from @xmath00.1 inside a radius of 1 to @xmath00.4 outside 10 , and the pa varies from @xmath040 to @xmath060 ( albeit with large uncertainties ) . these variations suggest that the galaxy s projected mass distribution may not have simple elliptical symmetry , and they may well need to be incorporated into lens models . the host galaxy arcs act as an extensive set of position constraints that may break the degeneracy between the lens galaxy shape and the cluster shear . although the intrinsic source structure is unknown , the simplicity of the lensing geometry means that we can model the arcs self - consistently using a modified version of the ring cycle algorithm of kochanek et al . ( 1989 ) . specifically , we know that there is a one - to - one mapping between the a image of the host galaxy and the source , so for a given lens model we can project the a arc back to the source plane to obtain the correct source structure for that lens model . we can then reproject that source onto the image plane to predict the structure of the b arc . since the arcs are very large compared to the size of the psf , there is no need to include the effects of the psf . for the arc map we use the smoothed residual image ( figure 1c ) . this technique produces maps of the model arcs that can be compared visually to the observed arcs ( see figures 24 below ) . however , for modeling we want to quantify the differences , so we define a @xmath56 term for the arcs using a pixel - by - pixel comparison of the observed and model surface brightnesses , @xmath57 the factor of 2 in the denominator enters because the source is constructed from arc a , so it and the observed b arc each have errorbars of @xmath58 . to compute @xmath59 we mask the arc map and use only the region near the b arc , omitting regions near the b quasar and the core of the lens galaxy where imperfect subtraction makes the residual flux unreliable . we also omit the region around the object g4 where the flux is not due to the quasar host galaxy ( see 2 ) . the masked region is shown in figure 1c . to count the number of arc constraints , we count the number of pixels inside the mask for which either the observed arc or the model arc is more than @xmath60 above the sky and take this to be the number of arc constraints @xmath61 . typically @xmath62 . because the source is fully determined from the a arc , we quote @xmath63 as an effective @xmath64 for the arc . the fact that our best models produce @xmath65 ( see 5 ) suggests that this is a reasonable counting of the constraints . q 0957 + 561 has been modeled extensively ; we summarize results from the recent models by grogin & narayan ( 1996 ) , barkana et al.(1999 ) , bernstein & fischer ( 1999 ) , and chae ( 1999 ) in table 3 . ( technical details of comparing the models are discussed in appendix a. ) the models fall into two main families based on how they treat the main lens galaxy . first is the `` power law '' family , in which the lens galaxy is modeled with a circular or elliptical surface density with a softened power law profile : @xmath66 , where @xmath67 is an ellipsoidal coordinate , @xmath68 is a core radius , and @xmath69 is the power law exponent such that @xmath70 asymptotically . the best - fit models typically have a small core radius and a power law @xmath71 corresponding to a profile slightly shallower than isothermal . in the second family , called `` fgs '' models after being introduced by falco , gorenstein & shapiro ( 1991 ) , the lens galaxy is treated as a circular or elliptical king model , plus a central point mass to account for a mass deficit in the core of the king model . in almost all of the previous models the lens galaxy was assumed to have a projected mass distribution with circular or elliptical symmetry . however , as discussed in 3.1 and 3.2 the observed ellipticity gradient and isophote twist suggest that this assumption may not be correct . the key improvement in the models is the addition of internal structure in the lens galaxy s mass comparable to that seen in its light . while we do not expect the mass to trace the light exactly , it should be given the same freedoms in order to avoid oversimplifying the models . bernstein & fischer ( 1999 ) did add freedom to the galaxy s radial and angular structure by using independent power laws in different radial zones , but this led to unphysical models with density discontinuities that can not match normal rotation curves . we introduce models that allow similar radial and angular freedom while keeping the density smooth . we start with the pseudo - jaffe ellipsoid , whose projected density distribution has elliptical symmetry and a profile that is roughly flat inside a core radius @xmath68 , falls as @xmath72 out to a cut - off radius @xmath73 ( yielding a rotation curve that is approximately flat ) , and then falls as @xmath74 to maintain a finite mass ; a detailed definition is given in appendix b. we then construct `` double pseudo - jaffe '' models comprising two concentric pseudo - jaffe components with different scale lengths , ellipticities , and orientations . we fix the model galaxy to its observed position , which leaves 10 galaxy parameters : a mass parameter ( @xmath75 ) , ellipticity ( @xmath76 ) , orientation angle ( @xmath77 ) , core radius ( @xmath78 ) , and cut - off radius ( @xmath79 ) for each component ( @xmath80 ) . the double pseudo - jaffe model provides a great deal of freedom in the lens galaxy . first , it is essentially a smooth generalization of both the power law and the fgs models . when the inner pseudo - jaffe component is compact , it mimics the point mass in fgs models while the outer component is similar to the king model ( see appendix b ) ; and when the cut - off radius of the inner component is comparable to the core radius of the outer component , the two components combine to produce nearly a smooth @xmath81 power law . second , with this model we can mimic the observed internal structure of the lens galaxy : by adjusting the ellipticities and orientations we can produce a smooth ellipticity gradient and isophote twist . an important difference from the broken power law models of bernstein & fischer ( 1999 ) is that the double pseudo - jaffe models have smooth density profiles . we discuss the physical properties of sample double pseudo - jaffe models in 5.3 . previous models have used several different methods for including the cluster s contribution to the lensing potential . the simplest approach is to expand the cluster potential in a taylor series and keep only the lowest significant ( 2nd order ) term , which describes the tidal shear produced by the cluster . however , this approximation is thought to be poor for q 0957 + 561 because the cluster is close to the lens and the vlbi errorbars are small , so the 3rd order terms are larger than the errorbars . barkana et al . ( 1999 ) and chae ( 1999 ) included all the higher order terms implicitly by introducing a mass distribution to represent the cluster , but degeneracies related to the cluster shape and density profile forced them to make assumptions about the cluster properties . kochanek ( 1991 ) and bernstein & fischer ( 1999 ) instead used a taylor series with general 3rd order terms , @xmath82 the 2nd order @xmath54 term produces the mass sheet degeneracy , so it is usually omitted from the lens models and constrained independently ( see 3.1 and 3.2 ) . the 2nd order @xmath43 term represents the tidal shear from the cluster . the 3rd order @xmath83 and @xmath84 terms arise from the gradient of the cluster density and the @xmath85 component of the cluster mass ( in a frame centered on the lens galaxy ; see bernstein & fischer 1999 ; keeton 2000 ) . the three direction angles @xmath86 are written here as position angles ( i.e. measured east of north ) . we follow bernstein & fischer ( 1999 ) and use the 3rd order taylor series ( omitting the @xmath54 term ) . although we simply fit for the taylor series parameters , it is important to understand how they relate to physical properties of the cluster and what their reasonable ranges are . in popular cluster mass models the cluster amplitudes @xmath87 and direction angles @xmath86 have the following properties ( keeton 2000 ) : * circularly symmetric mass distribution : the three angles all point to the center of the cluster . * singular isothermal ellipsoid : the shear angle @xmath44 points to the cluster center and the shear amplitude equals the convergence from the cluster , @xmath88 . both of these results hold for all positions and all values of the axis ratio @xmath89 . the gradient amplitude @xmath83 is bounded by @xmath90 , where @xmath91 is the distance from the lens galaxy to the cluster . * softened isothermal ellipsoid : outside of the core the results for the singular isothermal ellipsoid are approximately true . for example , for clusters with an axis ratio @xmath92 the shear angle @xmath44 points to within @xmath93 of the cluster center for @xmath94 , where @xmath68 is the core radius . the cluster has @xmath95 , but @xmath96 only for @xmath97 . * an ellipsoid with the `` universal '' dark matter profile of navarro , frenk & white ( 1996 ) : for clusters with an axis ratio @xmath92 , the shear angle @xmath44 points to within @xmath98 of the cluster center for @xmath99 , where @xmath100 is the scale radius in the `` universal '' profile . the cluster generally has @xmath101 for @xmath102 and @xmath103 for @xmath104 . we let the shear and gradient parameters @xmath105 vary freely , but we keep these general relations in mind when interpreting the results . to avoid an explosion of parameters , we follow bernstein & fischer ( 1999 ) and use a `` restricted '' 3rd order cluster with @xmath106 and @xmath107 ( as for a singular isothermal sphere ) . this model is useful because it requires only four parameters for the cluster , and it can provide model - independent evidence for a non - circular cluster if @xmath108 . we also experimented with models using a singular isothermal mass distribution for the cluster , and we discuss these models briefly but do not quote detailed results ( 5 ) . our models have 10 galaxy and 4 cluster parameters , but fortunately we need not examine all of them explicitly . first , the general features of the lens allow us to fix two scale radii . the lack of a central image requires that the galaxy be nearly singular , so we set the core radius of the inner pseudo - jaffe component ( @xmath109 ) to zero . also , previous lens models suggest that the galaxy needs to have mass extending at least to the distant a image ( @xmath110 from the galaxy ) , so the cut - off radius of the outer pseudo - jaffe component ( @xmath111 ) should be larger than @xmath110 . the model should not be very sensitive to any particular value larger than this , so we fix @xmath112 ( similar to bernstein & fischer 1999 ) . second , we can use linear techniques for the two galaxy mass parameters ( @xmath113 and @xmath114 ) and the two cluster amplitudes ( @xmath43 and @xmath83 ) because they enter the potential as simple multiplicative factors . ( this is a variant of solving for the shear parameters in eq . [ eq : shr ] . ) if we take the strong vlbi constraints on the quasar core and jet positions to be exact ( see kochanek 1991 ; bernstein & fischer 1999 ) , they lead to four linear constraint equations : @xmath115 where a@xmath116 and b@xmath116 denote the two quasar cores while a@xmath117 and b@xmath117 denote the brightest jet components ; the positions are given by barkana et al . the four constraint equations can be solved explicitly to write the four linear parameters as functions of the remaining ( non - linear ) parameters . this technique ensures that the vlbi constraints are fitted exactly , while reducing by four the number of parameters that must be examined directly . we are left with 8 explicit parameters : the ellipticities @xmath118 and @xmath119 , orientation angles @xmath120 and @xmath121 , and scale lengths @xmath122 and @xmath123 of the galaxy ; and the direction angles @xmath44 and @xmath124 of the cluster . we evaluate models in this parameter space using the remaining constraints : the flux ratios of the quasar cores and jets and the positions of the optical blobs and knots ( taken from bernstein & fischer 1999 ) , and the structure of the arcs ( see 3.3 ) . techniques and results for our new models are discussed in 5 . previous models of q 0957 + 561 ( summarized in table 3 ) posited that the vlbi observations and the optical blobs and knots provided strong enough constraints to break the degeneracy between the lens galaxy shape and the cluster shear . as we discussed in 3.1 , however , it is dangerous to use smooth circular or elliptical lens models with high - precision constraints , because the oversimplified models may be forced to converge on a best - fit solution with the wrong global structure for the potential . until now we have lacked the constraints to test this concern , although the fact that previous models gave wildly difference shapes for the lens galaxy certainly suggests that the models were not robust . the host galaxy arcs finally provide an extensive set of constraints for testing the previous models . the first clue comes from comparing the qualitative features of the arcs and the existing models . despite differences in details , most of the previous models have a lensing potential with a strong tidal shear from the cluster ( @xmath125@xmath126 ) . the galaxy dominates the potential for the close image ( b ) , but the strong shear means that the cluster dominates the potential for the distant image ( a ) . the distinction is important because in general a round host galaxy produces an image distorted tangentially relative to the mass that dominates the potential . as a result , for a circular host galaxy most previous models of q 0957 + 561 would predict a b arc tangential to the galaxy ( as observed ) , and an a arc tangential to the cluster i.e.stretched _ radially _ relative to the lens galaxy , opposite what is observed . these lens models can produce an a image tangential to the galaxy only if the source galaxy is highly flattened and oriented at just the right angle . in other words , we must either invoke a very special source configuration or conclude that the standard q 0957 + 561 models have generic problems . we can quantify these problems by using the arc modeling technique described in 3.3 to predict the structure of the b arc for each model and compare it to the observed b arc . ( recall that this technique uses the observed a arc to construct the source , so models always reproduce the a arc correctly and hence are evaluated by examining the b arc . ) figures 2 and 3 show the observed and predicted models arcs for previous power law models , while figure 4 shows the arcs for fgs models ; figure 5 shows the intrinsic source inferred for two of the models . table 3 includes the quantitative @xmath127 estimates for the models . all of these models are inconsistent with the structure of the b arc , both visually and in terms of @xmath127 . although the predicted arcs differ among the various models , there are two common features . first , in some of the models ( notably those with a circular lens galaxy ) the predicted b arc has a pair of bright ridges not seen in the observed b arc . models that produce such ridges are strongly inconsistent with the data ; they have @xmath127 statistics no better than 5.0 and as poor as 10.8 . second , the models generally fail to predict the correct shape for the b arc at low surface brightness levels . the predicted arcs have too much curvature and do not match the northeast extension of the observed arc . the failure to match even the rough shape ( extent and curvature ) of the b arc illustrates the point explained above : if the potential near the a image is dominated by the cluster shear , reproducing the a arc requires a flattened source ( e.g. the barkana / spemd model in figure 5 ) that is distorted into a highly curved b arc . the existing models can not match the arcs better than @xmath128 , which given the large number of arc constraints means that these models are formally excluded at an extremely high significance level . the models arcs also illustrate a crucial limitation of the broken power law models of bernstein & fischer ( 1999 ) , represented here by the dm2+c2 model in figure 3 . these models use independent elliptical power law models in different radial zones to mimic the ellipticity and orientation variations in the observed lens galaxy . the problem is that the density is discontinuous across the zone boundary , which leads to a discontinuity in the predicted b arc . the discontinuity does not affect the quantitative @xmath127 statistic because it occurs outside our mask , but it is qualitatively unacceptable . combining this with the inability of the broken power law models to match normal rotation curves emphasizes that the models are physically unacceptable . the need for a more physically reasonable way to include ellipticity and orientation variations motivates our introduction of the double pseudo - jaffe models . although the previous models do not correctly reproduce the distortions of the host galaxy , they do reveal a model - independent qualitative feature of the lensing . part of the quasar host galaxy is doubly - imaged like the quasar , but a small region crosses the lensing caustic and is quadruply - imaged ( see figure 5 ) . arc a is a single distorted image of the host galaxy , but arc b is a combination of the three remaining images of the quadruply - imaged and the one remaining image of the doubly - imaged region . this fact explains why the a and b arcs have such different geometries . it also implies that with the current observations of the arcs , most of the constraints come from the small region of the host galaxy that is quadruply - imaged . a deeper image of the arcs should reveal a complete einstein ring that would strengthen the constraints by using more of the host galaxy ( see 6 ) . in summary , the host galaxy arcs in q 0957 + 561 offer a strong new probe of the global shape of the lensing potential , and they show that existing lens models converged to the wrong potential . generically , the previous models have too much shear from the cluster . the failure to fit the arcs is a common problem of models that combine strong constraints from the sub - milli - arcsecond structure of the quasar jets with oversimplified circular or elliptical lens models . it rules out existing models of the system , together with the bounds on @xmath2 drawn from them . two goals motivate our search for new models of q 0957 + 561 . first , if we want to use the vlbi constraints with sub - milli - arcsecond precision , we must be sensitive to details of the lens galaxy structure such as its radially varying ellipticity and orientation . we used double pseudo - jaffe models ( see 3.4 ) to incorporate such structure in a smooth and physically reasonable way . second , we want to incorporate the powerful constraints from the host galaxy arcs into the modeling process , instead of using them for _ a posteriori _ tests as in 4 . in this way we hope to break the environment degeneracy in a robust way . when we began to examine new models we found a range of solutions consistent with the data , so we adopted monte carlo techniques to sample this range . in this section we first describe the monte carlo techniques ( 5.1 ) , and then discuss the models and their implications for breaking the degeneracy between the galaxy and cluster ( 5.2 ) , the physical properties of the lens galaxy and cluster ( 5.3 ) , and the hubble constant ( 5.4 ) . to explore our model space , we picked random values for the non - linear model parameters , using restricted but physically - motivated ranges : * galaxy ellipticity : there are no successful models with an outer ellipticity below 0.3 . we believed models with an ellipticity larger than @xmath129 to be implausible . hence we considered @xmath130 for the outer ellipticity . in models with an ellipticity gradient the inner ellipticity can be small , so we considered @xmath131 . * galaxy orientation : in the observed lens galaxy the pa varies from @xmath040 to @xmath060 ( bernstein et al . 1997 ) . we expect the mass to be roughly aligned with the light ( to within @xmath010 ; keeton et al . 1998 ) , so we considered @xmath132 for the inner and outer position angles . * galaxy scale lengths : we fixed the inner pseudo - jaffe component to be singular ( @xmath133 ) and the outer component to have a cut - off radius @xmath112 ( see 3.6 ) . for the cut - off radius of the inner component ( @xmath122 ) and the core radius of the outer component ( @xmath123 ) we considered @xmath134 . * cluster direction angles : galaxy counts and the weak lensing measurement suggest that the cluster has its mass concentration toward @xmath135 ( fischer et al . 1997 ) . for reasonable cluster models the shear angle @xmath44 points roughly to the cluster center ( see 3.5 ) , so we considered shear angles in the range @xmath136 . we were less restrictive with the gradient angle @xmath124 and required only that it be in the same quadrant as the cluster , @xmath137 . given values for these parameters , we fixed the four remaining parameters ( @xmath113 , @xmath114 , @xmath43 , @xmath83 ) using the constraints from the quasar core and jet positions ( see 3.6 ) . our limited parameter ranges may omit some models that are formally consistent with the data , but they span what we expect for physically reasonable models . we tabulated models that fit the data at the 95% confidence level . with @xmath138 constraints this is equivalent to @xmath139 ( e.g. press et al . 1992 , 6.2 and 15.6 ) . we applied these thresholds to the constraints from the quasars and the optical blobs and knots both separately and jointly : we required @xmath140 for the flux ratios of the quasar cores and jets ( 2 constraints ) , @xmath141 for the positions of the optical blobs and knots ( 4 constraints ) , and @xmath142 for the 6 joint constraints . we did not impose constraints from the less reliable blob and knot flux ratios . we considered different thresholds for the arc constraints , as discussed below . note that it was not useful to combine all the constraints into a total @xmath64 because the number of constraints from the arcs is so large . the monte carlo technique is not especially efficient ; we examined @xmath143 models and found of order 1000 models consistent with the data . however , examining that many models is not prohibitively time consuming . most of the models can be ruled out quickly because they fail the flux and blob / knot @xmath56 cuts . for the remaining models , we first computed a fast @xmath127 using a 4 times undersampled arc map ; only for promising models did we compute @xmath127 using the full arc map . the benefit of this brute force approach is the ability to identify and sample the wide range of models consistent with the data , and thereby estimate the full range of possible @xmath2 values . we first considered models in which the inner and outer components of the model galaxy were fixed to have the same ellipticity and orientation ; we found @xmath144 models with @xmath145 , whose properties are summarized in figure 6 . next we allowed the inner and outer galaxy components to have different shapes , mimicking ellipticity and orientation gradients . for simplicity , we refer to such models as having a `` twist , '' even though that term formally describes only an orientation gradient . we found @xmath146 twist models with @xmath147 , whose properties are summarized in figure 7 . table 4 gives parameters and figure 8 shows the predicted arcs for sample models of both types . we note that the number of successful models with a twist is smaller than the number of successful models without a twist . introducing the twist enlarges the parameter space without necessarily increasing the space of successful models by the same amount ; hence sampling the same total number of models yields fewer acceptable models . in the following sections we discuss models with @xmath148 , but our conclusions would not change substantially if we lowered this threshold . the host galaxy arcs finally break the degeneracy between the lens galaxy shape and the cluster shear in q 0957 + 561 . first , they constrain the lens galaxy shape . among the previous models , the three with the best fits to the arcs have @xmath149 ( table 3 ) . in the new models , the bounds on the ( outer ) orientation of the lens galaxy are @xmath150 in models without a twist ( figure 6d ) and @xmath151 in models with a twist ( figure 7f ) . these ranges are consistent with the observed galaxy s orientation of about @xmath152 ( bernstein et al . in other words , the arcs not only constrain the model galaxy s orientation , they require that its mass distribution be at least roughly aligned with the light distribution ; such alignment is also seen in other lens systems ( keeton et al . also , in the new models the bounds on the ( outer ) ellipticity of the galaxy are @xmath153 in models without a twist ( figure 6c ) and @xmath154 in models with a twist ( figure 7e ) . these ranges are slightly higher than the ellipticity seen in the outer parts of the galaxy ( bernstein et al . 1997 ) , but there is no _ a priori _ reason to expect the ellipticities of the dark halo and the light to be similar . these constraints on the lens galaxy shape span models with a wide range of lens galaxy mass profiles ( controlled by the scale radii @xmath122 and @xmath123 ; see figures 6b and 7b ) , ellipticity profiles ( with or without a twist ) , and cluster contributions . thus we believe that they are robust , and that the ability of the host galaxy arcs to finally constrain the galaxy shape is the most important result of our new models . second , the arcs place limits on the lensing contribution of the cluster . the new models have moderate or even small cluster shears ( typically @xmath155 ; figures 6e and 7 g ) , which contrasts with the strong shears in previous models . a small shear is required by the fact that both arcs are tangential to the lens galaxy ( see 4 ) . also , including the 3rd order cluster term is important , as expected from the proximity of the cluster to the lens ( kochanek 1991 ) . most models have a cluster gradient amplitude in the range @xmath156 ( figures 6 g and 7i ) , so at the position of image a the 3rd order term is comparable to or even larger than the 2nd order term ( see eq . [ eq : phiclus ] ) . because the cluster amplitudes are small , the corresponding direction angles @xmath44 and @xmath124 are poorly constrained ( figures 6f , 6h , 7h , and 7j ) . nevertheless , the fact that they usually differ implies that the cluster can not be spherical ( see 3.5 ) . unfortunately , even the arcs can not uniquely determine the model . first , the cluster angles can adjust to accommodate the narrow but finite range of lens galaxies given above . second , in models with a twist the inner component can take on all values of ellipticity and orientation in the ranges we allowed ( figures 7c and 7d ) . this is because the inner component is usually compact , so like the point mass in fgs lens models it serves mainly to correct a central mass deficit and the model is largely insensitive to the distribution of that mass . both of these degeneracies affect the potential enough to produce a disappointingly wide range of hubble constant values ( see 5.4 ) . these conclusions are drawn from models using a taylor series for the cluster potential . we also examined models treating the cluster using an elongated isothermal model with an arbitrary position , axis ratio , and orientation . the above conclusions about the galaxy properties still pertain , with one modification : in models with a twist the distribution of outer orientation angles gains a tail down to @xmath157 . the above conclusions about the cluster contributions also hold , except that the new bounds on the shear amplitude are @xmath158 . since an isothermal cluster has @xmath159 , it is hard to obtain a small shear without essentially eliminating the cluster . we introduced the double pseudo - jaffe models in order to smoothly include ellipticity and orientation variations in the lens galaxy , but we must ask whether the models are physically plausible . figure 9 shows the estimated density and circular velocity profiles for the models in table 4 . these profiles are only estimates because they require the full 3-d mass distribution , while lens models give only the projected distribution . we assumed that the intrinsic distributions are oblate spheroids viewed edge - on , and we computed the density and circular velocity profiles in the equatorial plane , neglecting any twist . the high inferred circular velocities ( @xmath160 ) are misleading , because our models omit any convergence from the cluster and @xmath161 , and because converting to an observed velocity dispersion involves systematic uncertainties in the stellar dynamics ( see romanowsky & kochanek 1999 ) . the qualitative features , however , are plausible . the inner and outer galaxy components generally combine to produce a rotation curve that is approximately flat , although in several models it turns up at small radii because the inner component of the pseudo - jaffe models is compact . this effect , which also appears in fgs models because of the point mass , occurs because lensing constrains only the total mass enclosed by the b image ( @xmath162 ) , and if that mass is compact then it produces a rising rotation curve . we conclude that the fitted double pseudo - jaffe models are fairly reasonable . as for the cluster , recall that the host galaxy arcs require the shear to be small . at the same time , two independent estimates of the cluster mass suggest that the convergence from the cluster is @xmath163 . first , stellar dynamical modeling of the lens galaxy constrains its mass such that explaining the image separation requires a cluster convergence of @xmath164 ( romanowsky & kochanek 1999 ) . second , fischer et al . ( 1997 ) and bernstein & fischer ( 1999 ) used weak lensing to measure the total mass inside an aperture centered on the lens galaxy . subtracting the mass of a model lens galaxy then gives a model - dependent estimate of the remaining cluster mass ( see eq . [ eq : kc ] ) . with this technique all of our models yield @xmath165 . these results are surprising because with popular cluster models it is difficult to produce a shear that is much smaller than the convergence ( keeton 2000 ) . specifically , with a singular isothermal ellipsoid , @xmath166 for all cluster positions , orientations , and axis ratios . introducing a core radius allows @xmath167 , but the shear is substantially smaller than the convergence only if the lens galaxy is near or within the cluster core . if the cluster halo has the `` universal '' or nfw density profile ( navarro et al . 1996 ) , @xmath168 only inside @xmath169 , where @xmath100 is the nfw scale length . for the observed cluster at redshift @xmath9 with a velocity dispersion of @xmath0700 ( angonin - willaime et al . 1994 ) , the nfw scale length would be @xmath170 @xmath171 kpc or @xmath172 ( e.g. navarro et al . 1996 ) . combining these results with the estimates of the cluster contribution ( @xmath173 or even @xmath174 , and @xmath163 ) suggests two possible conclusions about the cluster . on the one hand , if the cluster is ellipsoidal it must have a core or scale radius large enough to encompass the galaxy . having the lens galaxy be in the cluster core would not be too surprising since it is the brightest galaxy in the cluster . on the other hand , the assumption of an ellipsoidal model for the cluster may not be correct , perhaps because of substructure . finally , we want to combine the lens models and the observed time delay and use eq . ( [ eq : tdel ] ) to determine the hubble constant @xmath2 . as discussed in 3.1 , the inferred value of @xmath2 depends weakly on the other cosmological parameters ; we quote results assuming @xmath175 and @xmath176 and note that they would increase by 5.8% ( 4.5% ) for an @xmath177 open ( flat ) cosmology . also , the mass sheet degeneracy implies @xmath178 ( see eq . [ eq : hk ] ) , and @xmath54 can not be constrained by lens models . thus if we write @xmath179 , lensing directly measures only the combination @xmath180 , which is what we quote in tables 3 and 4 and figures 6 and 7 . we need an independent estimate of @xmath54 in order to constrain @xmath2 itself . we saw in 5.2 that the host galaxy arcs help break the degeneracy between the lens galaxy shape and the cluster shear , but they do not eliminate it entirely . the remaining freedom in the models is small in terms of the properties of the lens galaxy but large in terms of the hubble constant . models without a twist yield @xmath181 , with a tail down to @xmath182 ( figure 6 ) . allowing a twist in the lens galaxy broadens the distribution down to @xmath183 ( figure 7 ) . in other words , within the lens models there is a @xmath184 variation in inferred values for the hubble constant ; this is independent of uncertainties in @xmath54 . most of this variation is related to a strong correlation between the lens galaxy ellipticity and the hubble constant ( figures 6c , 7c , and 7e ) : flatter galaxies have deeper central potential wells when normalized to produce the same images . although the model uncertainties limit our ability to use q 0957 + 561 to constrain the hubble constant , we might invert our thinking and ask what properties of the lens model are required to be consistent with local distance ladder determinations of @xmath2 ( e.g. mould et al . 1999 ) . assuming a cluster convergence @xmath163 ( see 5.3 ) , obtaining @xmath185 requires a lens galaxy that is relatively round in the center ( @xmath186 ) and moderately flattened in the outer parts ( @xmath187 ) . these bounds bolster the suggestion that the lens galaxy must have a mass distribution with an ellipticity gradient fairly similar to that seen in the light distribution . we have detected large , distorted images of the quasar host galaxy in the gravitational lens q 0957 + 561 . in the h band ( rest frame r ) , the host galaxy appears as two long ( @xmath05 ) arcs stretched tangentially relative to the lens galaxy . previously published models of q 0957 + 561 fail to predict the correct shape for the host galaxy arcs . this failure rules out those models and any conclusions about the value of the hubble constant drawn from them . the problem with the previous models is that they oversimplified the mass distribution for the lens galaxy , so they had to adjust the large - scale features of the model ( the galaxy shape and the cluster shear ) in order to fit small - scale constraints ( the vlbi jets , with @xmath00.1 milli - arcsecond errorbars ) . without strong constraints on the global shape of the lensing potential , each class of models happily converged to a best - fit solution with the wrong global structure . the failure of these models presents two important lessons ( also see kochanek 1991 ; bernstein et al . 1993 ; mao & schneider 1998 ; bernstein & fischer 1999 ) . first , when the constraints are extremely precise it is necessary to include the full complexity of the structure of the lens galaxy . second , it is important to carefully explore the full range of possible models before leaping to conclusions about the value of the hubble constant and its uncertainties . two improvements are crucial to finding better models . first , the host galaxy arcs finally provide enough constraints to separately determine both the lens galaxy shape and the cluster shear , and they must be incorporated into the modeling process . second , models for the lens galaxy must allow internal structure similar to that seen in the observed galaxy , which for q 0957 + 561 means a radially varying ellipticity and orientation . although the mass need not exactly trace the light , it must be given the same freedoms lest we oversimplify the models . we introduced a new class of lens models , the double pseudo - jaffe models , that smoothly incorporate an ellipticity gradient and isophote twist . our models lead to several new conclusions about the system . first , the shape of the lens galaxy s mass distribution must be surprisingly similar to the shape of its luminosity distribution . the mass distribution must be moderately flattened and roughly aligned with the light distribution . such alignment between the mass and the light is also seen in other gravitational lenses ( keeton et al . 1998 ) , but was not seen in previous models of q 0957 + 561 . second , the shear from the cluster must be small ( @xmath173 ) , in marked contrast with the moderate or strong shears seen in previous models ( @xmath125@xmath126 ) . the small shear combined with an estimated convergence @xmath163 implies that the cluster potential must be approximately centered on the lens galaxy . unfortunately , the current observations of the host galaxy can not fully determine the lens model . there are still freedoms related to several direction angles that describe the cluster and to the amplitudes of the ellipticity and orientation gradients , and they leave a 25% uncertainty in the inferred value of the hubble constant . however , there are two promising prospects for further improving the constraints on this systems . first , the substantial progress we have found is based on a 2800-second h band image . a deeper image with a refurbished nicmos camera should show a complete einstein ring image . since an einstein ring probes the potential all the way around the lens galaxy , it is extremely useful for determining both the lens model and the intrinsic shape of the host galaxy ( see keeton , kochanek & mcleod 2000 ) . filling in the gaps in the q 0957 + 561 ring should eliminate the remaining uncertainties in the models . second , x - rays from the cluster gas have been detected ( chartas et al . 1998 ) , but the image resolution was poor and the signal was dominated by x - rays from the quasar images . new high - resolution x - ray observations to map the cluster gas would constrain the cluster potential and help determine the cluster angles that are still unknown ( namely the angle to the cluster center and the angle of the cluster s density gradient at the lens ) . combining deeper infrared and x - ray imaging would thus dramatically improve the constraints on the models and allow consistency checks of the cluster potential . improved x - ray imaging would also further improve the mass estimates of the cluster for breaking the mass sheet degeneracy . distorted images of the host galaxy have now been observed in four of the time delay lenses ( q 0957 + 561 , pg 1115 + 080 , b 1600 + 434 , and b 1608 + 656 ; see impey et al . 1998 ; kochanek et al . as we have illustrated , images of the host galaxy are a powerful constraint on models of the system and hence on the uncertainties in the value of @xmath2 derived from the time delay measurements . to date , host galaxies have been thought of mainly as pleasant bonuses in relatively shallow images targeting the lens galaxies . taking full advantage of the host galaxies will require deeper images focused on the host galaxies themselves . such images hold great promise for breaking common degeneracies in lens models and allowing gravitational lensing to map in detail the mass distributions of distant galaxies , to probe the potentials of lens galaxy environments , and to determine a robust and independent measurement of the hubble constant at cosmological distances . acknowledgements : we thank gary bernstein for discussions and comparisons with new optical data . support for the castles project was provided by nasa through grant numbers go-7495 and go-7887 from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc . , under nasa contract nas 5 - 26555 . table 3 summarizes results from models of q 0957 + 561 by grogin & narayan ( 1996 ) , barkana et al . ( 1999 ) , bernstein & fischer ( 1999 ) , and chae ( 1999 ) . the various authors reported results in different ways . sorting out the differences can be confusing , so we converted all the results to a standard form and explain the conversions here . first , bernstein & fischer ( 1999 ) and chae ( 1999 ) quoted an ellipticity parameter @xmath188 that is related to the axis ratio @xmath89 by @xmath189 . we report the true ellipticity @xmath190 . ( bernstein & fischer called their ellipticity parameter @xmath191 , but it played the same role as @xmath188 here . ) second , different authors quoted results for the hubble constant in different ways . in general , lens models can determine only the combination @xmath180 where @xmath192 and @xmath193 is the surface mass density of the cluster ( in critical units ) at the position of the lens . models that use a multipole expansion for the cluster ( an external shear or 3rd order cluster , see eq . [ eq : phiclus ] ) offer no constraints on @xmath54 . models that use an actual mass distribution for the cluster do predict @xmath54 : an isothermal sphere that has core radius @xmath68 and is located a distance @xmath194 from the lens produces @xmath195 and shear @xmath196 $ ] , where @xmath197 and @xmath198 is a mass parameter such that @xmath199 . however , the @xmath54 prediction is not unique because varying the cluster s shape and profile can change @xmath54 ( and @xmath200 ) without changing the goodness of fit ( see chae 1999 for examples ) . hence we describe these cluster models in terms of an equivalent external shear with magnitude @xmath201 ( see grogin & narayan 1996 ; barkana et al . 1999 ) . in table 3 we quote the magnitude of the external shear ( @xmath43 , or @xmath202 in the isothermal sphere cluster models ) and the hubble constant combination @xmath180 . we note that if the cluster is a singular isothermal sphere , @xmath203 ( see barkana et al.1999 ) . the estimates of @xmath180 can be translated into actual estimates for @xmath2 using the weak lensing measurement of the cluster mass ( see 3.2 ) . third , different authors used slightly different definitions of @xmath56 . we did _ not _ convert @xmath56 values ; we used what the authors reported . finally , we corrected several apparent typographical errors . barkana et al . ( 1999 ) claimed that their galaxy orientation angle @xmath204 was a position angle , but in fact the position angle was @xmath205 and this is what we quote . bernstein & fischer ( 1999 ) quoted angles measured from the positive @xmath206-axis . their formalism is self - consistent , except that their eq . ( 16 ) needs a minus sign in front of the @xmath43 term . we report the bernstein & fischer ( 1999 ) angles as position angles . finally , for the bernstein & fischer ( 1999 ) 3rd order cluster models there appears to be a difference of a factor of 2 between the amplitude of 3rd order cluster term as written in their eq . ( 16 ) and as reported in their table 3 . we quote values consistent with their eq . a standard jaffe ( 1983 ) model has a 3-dimensional density distribution @xmath207 where @xmath67 is an ellipsoidal coordinate and @xmath73 is the break radius . for lensing it is more convenient to use a modified density distribution @xmath208 where @xmath73 is again the break radius , and we have added a core radius @xmath209 . for this model the projected surface mass density , in units of the critical surface density for lensing , is @xmath210 , \label{eq : density}\ ] ] where the mass normalization parameter @xmath198 is chosen to match the lensing critical radius in the limit of a singular isothermal sphere ( @xmath211 , @xmath212 , @xmath213 ) . ( [ eq : density ] ) defines what we call the pseudo - jaffe ellipsoid . its projected surface density is roughly constant for @xmath214 , falls as @xmath215 for @xmath216 , and falls as @xmath217 for @xmath218 ; its total mass is @xmath219 . the ellipsoid coordinate @xmath67 can be written in terms of the projected axis ratio @xmath89 and the position angle @xmath220 ( measured east of north ) as @xmath221 .\ ] ] the pseudo - jaffe lens model is easy to compute because it is written as the difference of two softened isothermal ellipsoids , whose analytic lensing properties are known ( kassiola & kovner 1993 ; kormann , schneider & bartelmann 1994 ; keeton & kochanek 1998 ) . this model has been used previously by de zeeuw & pfenniger ( 1988 ) , brainerd , blandford & smail ( 1996 ) , and keeton & kochanek ( 1998 ) . in the limit of a singular model ( @xmath222 ) , the pseudo - jaffe model is an example of a general class of `` cuspy '' lens models with @xmath223 . these models are a more realistic family for real galaxies than the softened power law models , and are discussed by muoz , kochanek & keeton ( 2000 ) . the pseudo - jaffe model is related to the king model used in fgs lens models . the standard approach is to approximate the king model with a combination of isothermal models ( young et al . 1980 ) , @xmath224 the king model has a single scale radius @xmath100 , while the pseudo - jaffe model has independent core and break radii . also , in the king model the coefficients of the two terms differ , while in the pseudo - jaffe model they are the same . falco , e. e. , kochanek , c. s. , lehr , j. , mcleod , b. a. , impey , c. d. , keeton , c. r. , peng , c. y. , & rix , h .- w . 1999 , to appear in gravitational lensing : recent progress and future goals , asp conference series , eds . t. brainerd & c. s. kochanek ( also preprint astro - ph/9910025 ) kochanek , c. s. , falco , e. e. , impey , c. d. , lehr , j. , mcleod , b. a. , & rix , h .- w . 1999 , in after the dark ages : when galaxies were young ( 9th annual october astrophysics conference in maryland ) , eds . s. holt & e. smith ( also preprint astro - ph/9811111 )

we present photometry of 320,917 stars with throughout the ori star - forming region . using the more spatially limited spectroscopic surveys of dolan & mathieu to define the color - magnitude domain of young low - mass members of the association , and removing statistically the field stars in this domain , we use our photometry to identify a representative pms population throughout the interior of the molecular ring . the spatial distribution of this population shows a concentration of pms stars around ori and in front of the b35 dark cloud . however , few pms stars are found outside these pockets of high stellar density , suggesting that star formation was concentrated in an elongated cloud extending from b35 through ori to the b30 cloud . we find a lower limit for the global stellar mass of about 500 we find that the global ratio of low- to high - mass stars is similar to that predicted by the field initial mass function , but this ratio varies strongly as a function of position in the star - forming region . locally , the star - formation process does not produce a universal initial mass function . using our derived stellar ages across the region , we construct a history of the star - forming complex . this history incorporates a recent supernova to explain the distribution of stars and gas today . we infer that most of the present molecular ring was formed by ejecta from the center driven by the supernova blast about 1 myr ago . however , we suggest that the b30 and b35 clouds were primordial , and massive enough to be mostly little disturbed by the shock . the stars which we see today trace the former extent of the cloud complex . given the kinematics of the stellar population , we predict that the association will disperse into the field within a few tens of myr . the gas will be dispersed on a similar time scale , or faster if ori becomes a supernova before it escapes the region .

astro-ph0110160

much of our understanding of the process of low - mass star formation arises from studies of individual young stars . historically , the most popular regions for the study of these stars are nearby dark clouds , such as taurus - auriga , chameleon , lupus and @xmath3 oph . in these regions , the t tauri stars interact with the local interstellar medium with parsec - long jets and substantial molecular outflows , but their spheres of influence are usually smaller than their separations . thus models of star formation which arise from observations of stars in such t associations often consider only stars in isolation . recent advances in technology have permitted detailed studies of low - mass stellar populations in ob associations . the most massive members of such ob associations have large spheres of influence and may assail the entire star - forming region with dissociating or ionizing radiation or disrupting winds . thus the massive stars must be integrated into the formation process of associated low - mass stars . can the massive stars terminate nearby star formation by destroying the gas clouds ? alternatively , can they enhance star formation by compressing the clouds ? could both of these effects occur in different regions of the same star - forming complex ? can these effects change the resulting spatial and mass distributions of young stars ? recent studies are beginning to answer affirmatively to all of these questions . in cases where the massive stars ionize clouds and reveal embedded stars ( e.g. eagle nebula , @xcite ) or evaporate circumstellar disks of existing stars ( e.g. trapezium , @xcite ) , it is clear that massive stars have _ terminated _ star formation . on the other hand , @xcite and @xcite show that star formation in the upper scorpius ob association was likely _ triggered _ by a nearby supernova , since all the stars were formed within a very short timespan . for many years , star formation was hypothesized to be bimodal : low - mass and high - mass stars were thought to form under very different conditions ( c.f . @xcite ) . with the notable exception of the trapezium , evidence for low - mass stars near ob stars was lacking until the 1980s . since then , however , young low - mass stars have been shown to be ubiquitous . in particular the rosat all - sky survey found young , low - mass stars associated with ob stars throughout the gould belt ( @xcite ) . since this realization , surveys have demonstrated that the initial mass functions ( imfs ) of several ob associations are consistent with the field imf ( e.g. , @xcite in upper sco , @xcite in the trapezium ) . but these statements are for the global imf of the star formation regions . in many associations , the ob stars are more spatially concentrated than the rest of the stars . except perhaps in very high density environments like the trapezium , these associations are too young for dynamical mass segregation . perhaps some aspect of the star - forming process skews the local imf towards high - mass stars in some locations and low - mass stars in others . to make a comprehensive exploration of the effect of massive stars on low - mass star formation , this paper employs a photometric survey for young low - mass stars throughout the @xmath1 orionis ob association . the @xmath1 ori region is a superb laboratory for studying the evolution of a molecular cloud into an association of low- and high - mass stars . the star - forming complex contains a tight knot of ob stars encircled by a 40 pc diameter ring of dense molecular gas and dust ( see figure 15 of * ? ? ? notable features of this clumpy ring include the large b30 cloud on the northwest edge and the elongated b35 cloud protruding inward from the eastern side . by contrast , the interior of this ring is nearly devoid of dense gas . it is this relative transparency that has made the @xmath1 ori region both understudied in the past and appealing for the present study of the young stellar population : the lack of opacity reveals the entire member population as well as vast numbers of background field stars . until recently , comprehensive studies have been heavily biased towards stars with strong emission features such as h@xmath4 @xcite or x - rays ( * ? ? ? * and several subsequent works ) . in ( * ? ? * hereafter paper i ) and ( * ? ? ? * hereafter paper ii ) , we studied a spatial subset of the @xmath1 ori region , focusing on a 6-long region extending from b30 to b35 through the central concentration of ob stars . for nearly every star in that region with @xmath5 and redder than a 30 myr @xcite isochrone ( 3618 stars ) , we obtained a high - resolution spectrum with the wiyn multi - object spectrograph ( mos ) . from these , we identified 266 pre - main - sequence ( pms ) stars via the presence of strong lithium @xmath16708 absorption , a secure diagnostic of youth . using model evolution tracks to determine ages , we found that low - mass star formation started simultaneously with the ob stars 6 - 8 myr ago and has accelerated since then . however , we found a marked difference in recent star formation rates at different locations : within 2 of the ob stars , there are almost no stars younger than 1 myr while further away stars of this young age are plentiful . these differing age distributions led to our conclusion that there was a supernova 1 myr ago which disrupted further star birth in the vicinity of the ob stars . the model evolution tracks also yielded masses for the pms stars . compared with the massive ob stars , we concluded that globally the imf of the @xmath1 ori star - forming region was consistent with the field imf , but there were significant local variations : the high - mass stars dominated the central region while low - mass stars dominated at larger radii . those low - mass stars closest to the ob concentration also show a remarkable lack of h@xmath4 emission at all ages . we attributed this to fuv radiation from the massive stars destroying the circumstellar disks of those stars . the age and mass data for the pms stars showed us that nearly all of the detected pms stars lie in a region of the versus @xmath6 color - magnitude diagram brighter and redder than the bulk of the field stars . this result suggested that photometric analysis would be an effective tool for identifying and removing well over 99% of the field stars . to exploit this feature , we have imaged most of the @xmath1 ori region in and produced a comprehensive stellar photometric database . with these data , we identify all stars in the region of the cmd where pms members of the association are likely to lie . using the subset of fields where we have spectroscopic data ( paper 2 ) , we construct a model of the field star distribution . statistically removing the field stars , we are left with a stellar sample representative of the true pms population . this representative sample allows us to examine the spatial distribution of the young population . this sample also permits us to count the total number of low - mass members , and thereby compare the association imf to the field imf . in combination with the high- and low - mass stellar age distribution derived in paper ii , we develop in this paper a comprehensive history of star formation in the @xmath1 ori association . in 2 of paper ii we presented a summary of our broad - band photometric survey of the @xmath1 ori region . this paper presents a more detailed description and analysis of those photometric data . unlike paper ii , we discuss only the `` 1999 photometry '' in the following . in fact , some stars with older photometry from 1997 and 1998 are included in the dataset presented here where we have rare , accidental holes in our 1999 sky coverage , but this accounts for just 0.1% ( 361 of 320917 ) of the stars , so we ignore those older datasets in the following discussion . during december 1520 , 1999 we used the mosaic imager on the kpno 0.9 m telescope to survey about 60 degrees@xmath7 of the @xmath1 ori star - forming region in . this region includes the central concentration of ob stars , portions of the encircling ring of molecular clouds , and the nearly gas - free region in between . the mosaic imager is highly effective for such a survey with its nearly 1 field of view and pixel size of 0.42 . in total , we observed 61 fields with filters . table [ tab : obs ] summarizes the conditions of the observing run . a more detailed analysis of individual fields from each night is included in [ sec : data - quality ] . in general , the run was of high quality , but thin clouds affected a few fields , as described below . the full moon was very close to the star - forming region on the last nights , causing the signal - to - noise for stars of a given magnitude to be lower . prior to the run , we had selected 83 fields to observe , but anticipating that weather would prevent us from finishing all of these , we observed the fields in a priority order based on the following criteria : fields which had been observed in 1997 and 1998 ( to provide improved photometry for those crucial regions ) , fields adjacent to and overlapping the 1997 and 1998 fields , fields which overlapped archival rosat pointings ( for future comparisons with the x - ray data ) , and regions which appeared interesting in the @xcite co map ( particularly , the b223 cloud and the small , elongated cloud south of b35 ) . figure [ fig : nightmap ] shows the location of the observed fields . the field numbers marked on figure [ fig : nightmap ] indicate our priority order in the sense that a lower number is higher priority . table [ tab : chxref ] presents a cross - reference to the field names used in paper ii . we present the center coordinates for these fields as observed in table [ tab : fields ] , as well as the night on which the field was observed ( cross - referenced to table 1 ) . ( for those fields not observed , a 0 is given for night observed and the intended field center is provided for possible future extension of the survey . ) due to problems with telescope pointing , the actual field centers are often 1520 ( and up to 1.3 for some fields on night 1 ) away from the intended coordinates . since we usually overlapped adjacent fields by 5 , this is typically not a problem in terms of areal coverage . for each observed field , we took three exposures per filter , each shifted 100 pixels ( about 43 ) in right ascension and declination . this is the minimum number needed to cover the inter - ccd gaps in the mosaic imager . for almost 97% of the image , where there are no gaps or bad pixels , the multiple exposures increase the signal by three when combined . for just over 3% of the field , a star will lie on a gap in one of the three exposures . our photometry of such a star would only use signal from the other two , which corresponds to a 19% signal - to - noise degradation relative to the regions with full coverage . finally , 0.1% of the field lies on the intersection of gaps from two images , which translates to a 43% lower signal - to - noise than when all three images can be used . for each exposure , we used the following exposure times : 18 sec in @xmath8 , 15 sec in @xmath6 , 27 sec in @xmath9 . these times were chosen to yield signal - to - noise better than 100 in in the coadded images for a star with @xmath10 and = 1 , while preventing @xmath11 stars from saturating . for photometric calibration , we took single exposures of 710 fields each night containing @xcite standard stars . we achieved nearly uniform sampling in airmass from at least 1.2 to 2.0 each night . we always observed one or more of these calibration fields before the first and after the last science field of each night . we also obtained ten bias exposures , five dome flats and a few twilight flats per filter each night . over the course of the six nights , we gathered enough twilight flats to make one high signal - to - noise flat in each filter . we discuss the relative merits of dome and twilight flats in the next section . we reduced all of our images with iraf - mscred v3.2.3 , closely following the data reduction guide written for v2.0 by f. valdes . first , for each night we combined the bias images into a single image . it is clear from these images that the bias voltages for the mosaic ccds are unstable at the 23 adu level over a timescale of minutes , so for all subsequent images , we subtracted both the overscan strip and the combined bias image for the appropriate night . also , for each night we combined the dome flats for each filter into a single image . dividing dome flats from different nights as a test , we immediately noticed substantial deviations from unity , often in the form of strong gradients . communication with g. jacoby led us to believe that this may have arisen because we were insufficiently careful when pointing the telescope at the dome white spot when taking these flats . the small spot compared to the wide field of the camera requires pointing to within 1 to prevent large vignetting . this problem led us to reject the dome flats . instead , we used the twilight flats for flat fielding . first we selected the twilight flats from each night with mean count levels of 7,00015,000 adu ( 9 in @xmath8 , 10 in @xmath6 and 15 in @xmath9 ) . we combined these to create one combined flat per filter for the entire run . contrary to typical techniques , we did not pre - flatten these with the dome flats . we performed initial image reductions on the landolt fields with mscred - ccdproc using the twilight flats . for each field we then found a single star from the tycho catalog @xcite with @xmath12 to roughly register the images using msczero . we computed more detailed astrometry for each image with msccmatch using 5001000 stars from the usno - a2.0 catalog @xcite with @xmath13 . this typically yielded astrometric solutions with rms errors of 0.7 to 0.9 in right ascension and 0.5 to 0.8 in declination . using these solutions , we resampled the images to the sky tangent plane with mscimage . this latter step was essential to correct the non - uniform plate scale caused by the field corrector on the telescope . additionally , we found that redoing the astrometric solution with ccmap after resampling allows us to improve the rms coordinate residuals to about 0.6 in right ascension and 0.45 in declination . the greatest improvements occur at the edges and corners of the image , where the position implied by the coordinates may move by up to a pixel or two . we see this improvement because ccmap is more flexible than msccmatch , and it appears that there are minor imperfections in the noao - provided plate solution for the mosaic field . finally , using the reduced landolt field images we manually identified and centroided each landolt star with imexamine and ran phot on those stars . we used the relative magnitudes reported by phot to test the photometricity of each night . with one small exception , we found that nights 13 , 5 and 6 are all of high quality with rms fit residuals of 0.030 ( @xmath14 ) , 0.027 ( @xmath9 ) , 0.027 ( @xmath6 ) and 0.029 ( @xmath8 ) mag . the one exception is the last hour of night 1 , when thin clouds appeared unexpectedly . these clouds ruined the last landolt field and damaged the three previous science fields . the repair of these three fields is described below in [ sec : data - quality ] . as significant difference in the photometric parameters were not found between the nights , we combined the observations of standards on all photometric nights to compute the ccd and atmospheric parameters for the run . to accomplish the latter , we used fitparams for each filter to solve for four coefficients : the zero - point , color , airmass , and color - airmass terms . the initial stages of the reduction of the science fields were identical to those of the landolt fields . however , we employed the additional step of combining the three exposures per filter for the survey fields . after resampling the images to the sky tangent plane , we median - combined the multiple exposures with mscstack to make a single image . because the mscimage task allows one to register images while resampling , there was no need for a separate registration step . next , we identified the targets for which we wish to compute photometry . rather than attempting to identify sources in our images with a tool like daofind , we have adopted the point - source detections from the usno catalog . this choice has both positive and negative consequences . the main advantage is that we can trust that our targets are genuine stars and not extended sources , cosmic rays or ccd artifacts . in addition , we do not have to attempt the difficult task of cross - referencing our target detections from the different filters : we use the same target list for all filters . the main disadvantage is that the palomar survey , on which the usno catalog is based , has slightly coarser resolution than the mosaic , so a close pair which is resolvable in our field could appear blended as a single star in the usno catalog . another potential problem is that the usno coordinates are epoch 19501956 , and so may be inaccurate for epoch 1999 . this concern is minimized because ( 1 ) @xmath1 ori is very near the anti - center of the solar motion so solar reflex motion is small , ( 2 ) we suspect relative proper motions of member stars of the association are quite low , as they are in the orion nebula cluster ( most recently measured as @xmath15 mas / yr ; @xcite ) , and ( 3 ) we are using large apertures . using the astrometric solution from ccmap above , we transform the ra / dec coordinates into pixel coordinates for all usno - a2.0 stars with @xmath16 . we then performed aperture photometry with phot on the usno positions . because the @xmath1 ori region is not crowded , we used a generous aperture radius of 12 pixels ( about 5 ) , as we did for the landolt stars . for background subtraction , we used a sky annulus with inner and outer radius of 18 and 27 pixels respectively . following the phot task , we used the portion of the landolt solution with the invert task to compute calibrated magnitudes for each star . after removing the invalid stars ( ones flagged by iraf as saturated or on the edge of the image ) , this resulted in photometry of 30009000 stars per field ( varying primarily as a function of galactic latitude and cloud density in the star - forming complex ) . in table [ tab : data ] we present @xmath9 , @xmath6 and @xmath8 photometry for all stars , sorted by right ascension . where we have multiple measurements , we average the @xmath9 , @xmath6 and @xmath8 quantities . the quoted errors , derived in the following section , are separated into the three domains where we are limited by systematics ( @xmath5 ) , photon noise ( @xmath17 ) and sky background ( @xmath18 ) . the final three columns contain j2000.0 coordinates from the usno - a2.0 catalog , and the field(s ) in which the star was observed . figure [ fig : manystars ] shows the position of all survey stars with @xmath19 . with these 138,576 stars marked as black dots , regions of high extinction show up clearly in white , signifying an absence of background stars . these regions closely correlate with the iras infrared image of the dense molecular clouds shown in figure 16 from paper ii . thus figure [ fig : manystars ] vividly demonstrates that most of the stars in the dataset are background stars . with three exposures per field and overlapping regions between fields , we have many opportunities for internal quality checks . paper ii discussed in detail the quality of a subset of the photometry , primarily from nights 1 and 2 . here we discuss the entire dataset , first in a global sense and then followed by an investigation into differences between fields . in paper ii we evaluated our internal errors based on comparisons of photometry derived from each of the three exposures for each field and filter . there , we found an internal limiting precision of 0.016 mag in all bands for bright stars ( @xmath20 ) . for single exposures , stars with @xmath21 are photon - limited while stars fainter than @xmath22 are limited by sky brightness . these limits can be seen in the top panel of figure [ fig : photerr ] which shows the measured magnitude differences derived from any two of the three @xmath6 exposures of field 110 . however , as we will show below , external systematic effects dominate stars brighter than @xmath10 , so the 0.016 mag precision is not achieved . our external errors can be assessed by comparison of photometric measurements of any star observed more than once . such multiple observation come from fields which overlap each other on edges or corners , totaling to 44414 , 3052 , and 399 stars with 2 , 3 , or 4 observations respectively . for these stars , we computed standard deviations of @xmath6 magnitude measurements . to study the distribution of standard deviations , we divided the stars into 6 bins according to their mean @xmath6 magnitude . inverse gaussian analysis of each of the magnitude intervals found the standard deviations of these distributions to be 0.12 ( @xmath23 ) , 0.069 ( @xmath24 ) , 0.047 ( @xmath25 ) , 0.041 ( @xmath26 ) , 0.039 ( @xmath27 ) and 0.037 ( @xmath28 ) mag . these numbers demonstrate that photometric quality is only weakly a function of magnitude for stars with @xmath29 . the asymptotic approach to a standard deviation of about 0.030 mag indicates a systematic limit on our photometric precision . for the faintest stars , the photometry deteriorates rapidly as the moonlit sky brightness begins to dominate . thus we use the dispersions quoted above as our photometric uncertainty in table [ tab : data ] for stars fainter than @xmath10 . however , for stars brighter than @xmath10 this analysis underestimates the quality of the photometry . we have found that the field - to - field overlap regions at the edges of the images ( from where all of our multiple measurements derive ) suffer significantly worse photometry than the rest of the images . there are two main systematic effects which degrade the photometric quality . first , when we compare photometry between fields which overlap north - south , we find that there is a systematic offset . this offset has a mean value of 0.029 ( @xmath9 ) , 0.047 ( @xmath6 ) or 0.051 ( @xmath8 ) mag measured from all multiply measured stars with @xmath30 in the 64 pairs of north - south adjacent fields . these offsets are in the sense that stars at the north edge of the south - most field appear fainter than the same stars observed in the south edge of the north - most field . measuring this magnitude difference as a function of position within the overlapping strip , we find that the most deviant photometry comes from within @xmath311 of the very edge of the chip , and improves rapidly toward the interior of the chip . we find the dispersion of magnitude _ differences _ to be 0.048 mag in all three filters , corresponding to a true measurement precision of a single observation of 0.034 mag . for the 47 fields which overlap east - west , on the other hand , we find mean offsets near zero in all filters . the single measurement dispersion is 0.030 mag , similar to the difference dispersion found at the north - south edges . given the restriction of the north - south photometric offsets to the outer couple arcminutes of the field of view and the uniform photometric quality with position in the east - west overlap regions , we are confident that photometry over almost all of a field is of uniform quality . in the absence of independent photometry , we choose to minimize the impact of the north - south offset by averaging together all multiple measurements . we nonetheless caution that the photometry of stars within a few arcminutes of the north or south boundary of a field is likely less accurate than the rest of the stellar sample . the second effect that systematically degrades the photometric quality is the proximity of the moon on the later nights of the run . by night 6 , the moon was 97% full and only 18 northwest of @xmath1 ori . these conditions forced us to select targets in the southeast quadrant of the star - forming region until the moon was occulted by the dome in the last third of the night . these conditions inflated the resulting photometric dispersion between overlapping fields by 15% or more for stars with @xmath32 . to more accurately quantify the effect of the moon , we recomputed the photometric precision excluding all fields from nights 5 and 6 , which translates to excluding almost 50% of the multiply - measured stars . the offsets ( either north - south or east - west ) did not change appreciably , but the dispersion of the magnitude differences decreased to 0.030 ( north - south ) or 0.029 ( east - west ) mag for a single field . comparable tests including _ just _ nights 5 and 6 yield 0.036 and 0.033 mag . in conclusion , we adopt an intermediate estimate of 0.032 mag for our photometric precision in all filters . the photometry may be worse near the north and south edges of the fields , but those are regions where we have multiple measurements whose combination will reduce the errors . in addition , we warn that this precision varies from field to field . in particular we note that the slightly lower - quality photometry from later nights of the run are at systematically larger radii from @xmath1 ori . finally , we found a few fields to have anomalous photometry . we observed six instances ( fields 109 , 110 , 112 , 128 , 138 and 173 ) where one exposure in one filter had photometry which disagreed with the other two exposures by between 0.02 and 0.10 mag . in one additional case ( field 111 ) , one exposure in each of the three filters suffered similar defects . figure [ fig : photerr ] shows data from field 110 as an example comparing the three exposures in a normal filter alongside three exposures from a problematic filter . from this comparison , one can see the obvious 0.030 magnitude offset in exposure 2 of the @xmath8 filter . in all cases , the observed discrepancies were in the sense that the anomalous exposure generated fainter magnitudes than the others . we are not certain about the cause of these irregularities . it could be very thin cirrus which we did not notice at the telescope . or , perhaps more likely , it could be due to partial occultation by the dome , since we had problems with the dome drive motor overshooting its target azimuth throughout the run . we have responded to these problems in the data with two prodedures . for fields 138 and 173 , which each had small offsets of about 0.03 mag for one exposure , we simply applied a correction to that one exposure to match the others before the coadding step . a followup comparison with the overlapping neighbors show that this correction was successful . for the rest of the problematic fields , we simply discarded the errant exposure and combined the other two exposures alone . subsequent checking against the adjacent fields showed that photometry derived from those remaining exposures is of high quality . discarding one exposure caused us to miss a few stars in the inter - chip gaps in the affected filter , but the effect is minor . in addition , we have three fields ( fields 4 , 6 and 106 , observed in this order of time ) which were apparently contaminated by unnoticed cirrus at the end of night 1 . the gradual arrival of these clouds is obvious when the these fields are compared to adjacent fields observed on different nights . the fields surrounding the affected ones all had very high quality photometry which allowed us to compute precise offsets . the first two filters of field 4 are unaffected while the third ( @xmath8 ) suffers 0.015 mag of extinction . next , field 6 suffers extinction in all three filters of 0.032 ( @xmath9 ) , 0.053 ( @xmath6 ) and 0.044 ( @xmath8 ) mag . finally , field 106 suffers 0.092 ( @xmath9 ) , 0.144 ( @xmath6 ) and 0.299 ( @xmath8 ) mag of extinction . we used these measured offsets as corrections which we applied to the final photometry for each of these fields . since the corrections are small ( at least for fields 4 and 6 ) , the errors in the color correction should also be small . thus , our only outstanding concern for these fields is that the extinction might have been non - uniform . the photometric offsets for the opposing edges were similar , so we suspect that the data are of acceptable quality . in summary , we have obtained photometric data for 320,917 stars over 60 degrees@xmath7 with a typical photometric precision of 0.032 mag in all bands for stars brighter than @xmath10 . our goal is to use these photometric data to identify a statistical sample of pms stars in the @xmath1 ori star - forming region . we have demonstrated in papers i and ii that using spectroscopic data combined with multicolor photometry we can securely identify young association members despite the fact that field stars outnumber the members by factors of several hundred . in this section , we demonstrate that we can use the knowledge gained from our spectroscopic survey to identify a statistical sample of stars which closely resemble the pms population ( in number , spatial distribution , and photometry ) using the photometric data alone . as a proxy for the true pms population , we will use this sample to study the spatial distribution of young , low - mass stars over the entire @xmath1 ori region and to compute the imf of the star - forming complex . our technique relies on the fact that pms stars have @xmath6 vs. cmd positions rather different from most field stars . additionally , we use the information from papers i and ii that the pms stars lie in a small range of ages at about the same distance . exploiting these characteristics allows us to maximize pms stars while minimizing field stars in a photometrically - selected sample . we use the @xmath6 , and measurements of confirmed pms stars to define photometric boundaries in which the field star contamination will be as low as possible . the first step is to define a region of the cmd which has a high ratio of pms stars to field stars . we use our spectroscopic pms population from paper ii as a calibration sample , or control group , to define such a region before applying it to the fields where we have no spectroscopic information . figure [ fig : control ] shows the location of these spectroscopically - identified pms stars along with a large population of field stars in the same lines of sight . using a line at @xmath10 along with isochrones and mass tracks from the @xcite stellar evolution model as boundaries , we can isolate a region in the cmd which has many pms stars and few field stars . specifically , we have selected an isochrone at 4 myr and a mass track at 0.6 @xmath2 ( both at 450 pc ) as optimal boundaries to minimize field contamination . we further cull field stars from the spectroscopic control sample using the colors . figure [ fig : cullvr ] shows all of the field and pms stars which fulfill the criteria in the @xmath6 vs. domain . for this sub - sample , the pms stars form a well - defined locus in the @xmath6 vs. cmd , while the field stars are much more broadly distributed , particularly brighter and redder than the pms stars . we have chosen a pair of boundaries , shown in the figure , which remove 28% of the field stars at the expense of only 1% of the pms stars . the resulting , and @xmath6 boundaries yields roughly equal numbers of pms ( 179 ) and field ( 232 ) stars . next we use these photometric boundaries to select candidate pms stars in survey fields for which we do not have spectroscopic information . figure [ fig : agemap ] shows the spatial distribution of all stars which fall within the photometric criteria defined above . inset in that figure is an @xmath6 vs. cmd showing those same stars . comparison with figure [ fig : manystars ] gives a sense of the large extent to which field stars have been removed . the most obvious feature of this map is that there are stars everywhere in the region of interest . based on the analysis of the previous section , more than half of these are field stars which should have a nearly uniform distribution . close examination of figure [ fig : agemap ] shows that there are excesses of stars in particular areas . notably , there is a large excess centered on @xmath1 ori . there is also a distinct clustering of stars near b35 . thus , the photometric selection technique is successfully identifying pms populations in regions where we had previously found enhancements of pms stars by spectroscopic techniques . additionally , the figure appears to show a ring of higher stellar density at the edges of the surveyed region , coinciding with areas where the field star density is shown to be low in figure [ fig : manystars ] . however , as we discuss in the next section , these density enhancements may be artifacts caused by reddening in the dark clouds rather than enhancements of pms stars . when searching for density enhancements in figure [ fig : agemap ] , one implicitly assumes that the field star distribution is uniform . however , if a large fraction of the field stars are more distant than the @xmath1 ori association as indicated by figure [ fig : manystars ] , the reddening due to dense gas in the star - forming region can move field stars into our photometric selection region . this change in the morphology of the field cmd can cause an apparent excess in the vicinity of a dusty cloud . that is , the number of non - pms stars varies as a function of extinction in the star - forming region . to estimate the magnitude of the extinction , we use the following simple procedure . first we refer back to figure 2 of paper ii , which shows the distribution of co j=1@xmath330 emission in the star - forming region @xcite . the highest contour is 19.2 k km s@xmath34 . @xcite quote a conversion from co brightness temperature to h@xmath35 column density of @xmath36 ( k km s@xmath34 ) from @xcite for orion . thus the strongest co emission traces h@xmath35 of column density @xmath37 @xmath38 . doubling this number to account for the two hydrogen atoms per molecule and using the @xcite conversion from hydrogen column density to @xmath39 color excess of @xmath40 @xmath38 mag@xmath34 , we find the peak reddening is roughly @xmath41 mag . using a general - to - selective extinction ratio of @xmath42 and the @xcite extinction law , we find the peak absorption to be on the order of @xmath43 mag . if instead we use a higher value of @xmath44 , we find @xmath45 mag . certainly , this calculation is only an approximation ( for example it is not clear if our application of the bohlin , savage & drake conversion is appropriate for this region . however it suffices to demonstrate that the dense clouds have non - negligible extinction , but are not opaque . thus , there could be a couple of magnitudes of extinction for background stars where the molecular clouds are densest . the consequence of this is that many of the blue field stars could be reddened into the photometric boundaries we defined above . to test this hypothesis , we take a field where there is very little molecular gas ( as inferred from the co map ) and artificially redden it by a constant extinction , adding some random scatter to simulate varying conditions within the field . figure [ fig : redtest ] shows an example of this test performed on field 151 . the upper left panel shows the unaltered cmd of that field . to create the cmd in the lower left panel , we added an extinction ( @xmath46 ) randomly selected from a gaussian distribution with a mean of 1 magnitude and a standard deviation of 0.5 mag ( discarding any selected extinction below zero ) . in this example we compute the @xmath47 reddening from @xmath46 using @xmath48 . as a comparison , we include in the lower right panel the cmd of field 65 , which should be genuinely reddened as it is projected on the b30 dark cloud where the co has a high column density . note the strong similarity between the lower two panels of figure [ fig : redtest ] showing one artificially and one physically extincted field . we take this similarity to mean that ( 1 ) the on - cloud fields have similar background cmds to the off - cloud fields , but they have been extincted and reddened ; ( 2 ) most of the field stars are in the background ; and ( 3 ) our extinction estimates calculated above are reasonable . most importantly , the number of stars in the photometric selection region rises rapidly with increasing extinction . therefore , we conclude that the apparent excess stellar density projected on the regions of the dark clouds in figure [ fig : agemap ] is likely an artifact of the extinction of background field stars caused by those clouds . this also means that the probability of a bright , red star projected on a dark cloud being a young member of the association is much smaller than one with similar photometry projected on a transparent region . the sample of stars we constructed above contains stars with photometry consistent with pms members of the association . however , that sample still contains a substantial number of field stars which are photometrically indistinguishable from pms stars in . to study the imf and spatial distribution of the low - mass @xmath1 ori members , we would like our proxy population to contain stars with not only the correct photometric properties ( thus ages and masses ) , but also to contain the correct number of stars . ideally , we would simply subtract a constant number of stars from each field to arrive at a statistical estimate of the number of pms stars in those fields , but unfortunately the field star distribution is not uniform , because the background star density decreases with increasing galactic latitude . therefore , in this section we describe a technique we have used to determine the number of field stars using the assumption that the unreddened field - star cmd has the same morphology across the entire complex . that is , we assume that the ratio of field stars in one region of the cmd to another region is a constant . we establish this ratio using the spectroscopically studied control fields where we can securely create a sample with only field stars and no pms stars . then we use this ratio to subtract field stars in each of the rest of the fields . this yields a measure of the number of pms stars in each field . note that , as described in the last section , the regions with substantial reddening do not have the same cmd morphology as unreddened regions , so this technique does not work well for fields projected on the dark clouds . in detail , we first identified a region of the cmd which is free of pms stars ( which we call the `` pms - free region '' in contrast to the `` pms candidate region '' discussed in earlier sections ) . we defined this by the boundaries @xmath49 and @xmath50 . several criteria entered into our selection of this particular region . first , this region is rich with field stars , as seen in figure [ fig : control ] , for example . the more field stars there are , the less error is introduced into our calculations by counting noise . second , this region is faint enough and blue enough that it can not contain any pms stars . in fact , this box is almost entirely below the zams at 450 pc , let alone near the pms stars . third , it is bright enough that photometric is negligible . totaling over all the control fields except those with poor spectroscopic completeness ( as identified in table 3 of paper ii ) and those projected on the densest portions of b30 , we find a ratio of 0.033 for the number of field stars in the pms - candidate zone to the number of stars in the pms - free region . the counting error associated with this ratio is about 15% . next , for each of the fields in our survey we count the number of stars in the pms - free region and use the above ratio to compute the expected number of field stars in the pms - candidate region . the difference between this result and the observed number of pms candidates is an unbiased estimate of the true number of pms stars . finally , for the purpose of creating a map of the pms stars , we subtract the derived number of field stars from the survey fields . this removal is statistical , since we do not really know which are field stars and which are pms stars . thus , for each field we randomly remove a number of stars equal to the expected number of field stars leaving a representative sample of pms candidates with approximately the same numbers , photometric properties , and spatial distribution of the true pms population . the random nature of the removal technique means that the pms sample is only representative on a field - by - field basis . that is , we find an accurate estimate of the number of stars in each 1 field , but we do not know their true positions within that field . because the fields which are projected on the molecular clouds suffer reddening of the background stars , our technique is invalid there ( although it has been applied to all fields for simplicity ) . figure [ fig : pmsmap ] shows the statistical pms population after subtraction of field stars . figure [ fig : pmsmap2 ] shows the same map as figure [ fig : pmsmap ] rotated into galactic coordinates and overlayed on the co map of @xcite . stellar density enhancements are clearly seen near both @xmath1 ori and b35 , confirming similar observations made from figure [ fig : agemap ] . _ perhaps more remarkably , much of the region inside the molecular ring is devoid of pms stars . _ these two figures show a large number of low - mass stars near @xmath1 ori and the neighboring b stars . the stars around @xmath1 ori are present primarily within the 2 circle marked on figure [ fig : pmsmap ] . out to that radius , this distribution is consistent with an @xmath51 density distribution centered on @xmath1 ori . this is reminiscent of the simple kinematic model of the stellar expansion we constructed in paper i where we explored ( and could not rule out ) the possibility that all of the stars around @xmath1 ori had formed in a tight trapezium - like cluster and then kinematically dispersed when the parent gas was dispersed . the essential new information from this study is the spatial extent of the population of pms stars around @xmath1 ori . even excluding stars in the vicinity of b35 , the distribution of pms stars extends as far as 16 pc from @xmath1 ori . for a one - dimensional velocity dispersion of 2.5 km / sec ( as found from radial velocities in paper i ) , a typical star would move 2.55 pc in 12 myr , while extreme velocity stars might travel as much as 7.515 pc . as such it is possible that the spatial extent of the present pms population around @xmath1 ori derives in large part from ballistic expansion of an initially more concentrated population after a recent gas dispersal event . even so , given the substantial fraction of stars found 1015 pc from @xmath1 ori , it seems unlikely that the distributed population derives entirely from a trapezium - like cluster around @xmath1 ori . indeed , as noted in paper i , it would be odd if the low - mass stars of such a cluster dispersed while the ob stars did not . thus we conclude that star formation occurred over a distributed region around @xmath1 ori , but quite possibly over a spatial extent more limited than the presently observed spatial distribution of pms stars . the edge of this central population of pms stars abuts on a clump of stars near the b35 cloud . in figure [ fig : pmsmap2 ] , one can see that this group extends toward @xmath1 ori from the densest portion of the molecular cloud . however , it is somewhat surprising that there appear to be no pms candidates on the `` neck '' of the b35 cloud , the elongated portion which connects the dense `` head '' to the ring . clearly the star - formation rate has differed significantly between the front and the back of this cloud . finally , the _ lack _ of pms candidates outside of the central population is equally important . figure [ fig : pmsmap ] shows this dearth most clearly in the region between the two circles . ( again , the surface density enhancements outside the larger circle are not valid representations of the pms population . ) thus , within the present ring of molecular clouds we find that star formation occurred only within a 2-radius circle around @xmath1 ori and near the b35 cloud . we also know from paper ii that star formation has been active in the b30 cloud . it remains to be seen whether any star formation has occurred elsewhere within the molecular ring . using our statistical count of pms stars to represent the true pms population , we assess the association imf in this section . we restrict our sample to @xmath52 , which minimizes both incompleteness at the faint end ( see 5.3 of paper ii ) and contamination by the field , respectively . in the census completed in section 3.5 , we used an age limit of 4 myr . however , in paper ii we noted that the onset of star formation occurred about 8 myr ago . thus a fair assessment of the imf should include stars of at least that age . a drawback is that in extending our census to an 8 myr age limit , the number of field stars increases faster than the number of pms stars . for stars with with @xmath52 there are 60% more total stars in the expanded selection region of the cmd , but in the spectroscopic survey we found only a 10% increase in pms stars in this region ( 62 pms stars younger than 8 myr vs. 56 younger than 4 myr ) . thus we consider two different methods of counting pms stars to 8 myr . first , we simply redo the census with an age cutoff of 8 myr ( which involves re - doing the field star calibration and rejection of [ sec : search - remove ] ) . in this case , we count 142 stars with @xmath52 within the dashed , 3 circle on figure [ fig : pmsmap ] . second , we consider the number of pms candidates younger than 4 myr ( 102 ) and extrapolate to 8 myr using the ratio of pms stars younger than 8 myr ( 62 ) to those younger than 4 myr ( 56 ) found in our spectroscopic survey . this yields an estimate of ( 102@xmath5362/56 ) = 113 pms stars younger than 8 myr over the entire star - forming region . this latter computation has the drawback that it implicitly assumes the same star - formation history throughout the entire region . since we do not know which count is more accurate , we will use both the direct count ( 142 pms stars ) and the extrapolated count ( 113 pms stars ) in the calculations below . we note that our uncertainty in the calibration of the field star density of 15% yields a 20% uncertainty in the number of pms candidates . using the ( * ? ? ? * hereafter ms ) field imf , 142 stars with @xmath52 predicts 44 ob stars ; a count of 113 pms stars predicts 35 ob stars . in fact , there are only 24 ob stars ( as identified in paper ii via the ppm catalog ; @xcite ) in projection within the region of the low - mass star census . application of the monte carlo test described in paper i shows that the difference between the observed number and the prediction suggests a difference in the imfs at the 99.8% confidence level for the direct count and at the 95.8% confidence level for the corrected count . thus , according to the ms imf , the low - mass stars are mildly over - represented in the star - forming region . this over - representation is further increased by the fact that there are very likely non - members included in the ob census , as discussed in 3 of paper ii . we have also tested the field imf formulation of ( * ? ? ? * hereafter ktg ) , which employs three power - law fits to the mass function instead of the log - gaussian used by ms . the ktg field imf predicts that , given the direct count of 142 stars between 0.4 and 0.6 @xmath2 , we should see 20 ob stars . this is more than a factor of two lower than the ms prediction above , and is consistent with the observed number of ob stars . the corrected count of 113 pms stars predicts 16 ob stars which is also consistent with the observed number . clearly the systematic uncertainties in the field imf dominate these comparisons . given that only in the most extreme case can the difference between the field imf and the association imf be considered significant , we conclude that the global initial mass function in @xmath1 ori is not distinguishable from that in the field . on the other hand , local variations of the imf are significant , as reported in papers i and ii : ( 1 ) in the central 1 where the ob stars are concentrated , the low - mass stars are _ deficient _ by a factor of two ; ( 2 ) outside of this central region they are _ overabundant _ by a factor of three . the spatial distribution of star formation is significantly _ mass - biased_. our count of stars with @xmath54 plus the 24 ob stars implies a total mass of 450 to 650 @xmath2 for stars above 0.1 @xmath2 within the molecular ring , depending on the pms count ( 113 or 142 ) and choice of field model ( ms or ktg ) . these mass estimates are a factor of two higher than the the lower limit on the total mass stated at the end of paper ii , but here we also include more than a factor of three more spatial coverage . compared to the total cloud mass of @xmath55 @xmath2 calculated by @xcite ( including molecular , neutral and ionized gas ) , this implies a star formation efficiency of 1%2% . the molecular cloud mass computed with higher quality co data by @xcite matches that of maddalena & morris for the molecular gas ( @xmath56 @xmath2 ) , but they do not include or mass , so we can not compute an independent efficiency from their work . one must note that these are mass estimates of the present - day clouds . it is likely that a significant fraction of the natal cloud could be sufficiently heated or dispersed by now that it is not included in the above estimates . but whether the efficiency is 1%2% or a factor of two lower , it is similar to other estimates of global star - formation efficiency in molecular clouds ( e.g.@xcite ) . we have created a statistical representation of the entire young stellar population inside of the molecular ring . we have found a large concentration of pms candidates centered on @xmath1 ori as well a concentration near the b35 dark cloud . more thorough studies of a subset of these stars via wiyn spectroscopy in papers i and ii definitively showed that these populations are composed of genuine young members . in addition , the spectroscopic investigation showed that there are many pms stars projected on the b30 dark cloud ( on the northwest side of the molecular ring ) , where our purely photometric analysis does not work well . there are no other concentrations of young stars within the molecular ring . this supports the model proposed by @xcite that the initial cloud distribution was elongated , extending ( at least ) from b30 through the center to b35 . thus , the current distribution of young stars is seen as a fossil of the parent molecular cloud , tracing the densest parts of that structure . our identification of this fossil is only possible because ( 1 ) the cloud is mostly dispersed today and ( 2 ) the stars have not moved much from their places of birth . this latter point is supported by the kinematic evidence presented in 3.4 and 4.4 of paper i. to recap those arguments , the high ob - star proper motions and pms - star radial velocities in combination with the present tight spatial concentration of those populations constrain the dynamical timescale to 12 myr , at most . before that , the stars must have been gravitationally bound , presumably by the parent cloud . but today there are virtually no signs of that parent cloud remaining within the central 2 radius , or 16 pc radius at the 450 pc distance of @xmath1 ori . we have suggested that the cloud was rapidly dispersed by a supernova about 1 myr ago . we require a supernova instead of more conventional action by ob star winds and radiation primarily to achieve the short timescale : @xcite calculate that more than 3 myr is needed for the ob stars alone to carve out an region of 16 pc radius . in 3 myr , the proper motion of those ob stars would have carried them outside of the molecular ring , which is clearly inconsistent with observations . in addition , the dearth of pms stars younger than 12 myr within 16 pc of @xmath1 ori ( discussed in detail in paper ii ) , despite the abundance of such young stars in the more distant dark clouds b30 and b35 , demonstrates that an event centered near @xmath1 ori terminated all low - mass star formation in its vicinity quite recently . the probable mechanism for this termination is the dispersal ( again by supernova ) of the central cloud from which the stars were forming . with this evidence as foundation , we construct the following history of the star - forming region . in the course of this discussion , the reader may wish to refer to panels a , b , c and d of figures [ fig : schemzoom ] and [ fig : schematic ] , which show schematic representations of this history . about 810 myr ago , the @xmath1 ori region was composed of a starless , roughly linear string of dense molecular clouds ( panel a ) . the most massive lobes of this cloud chain were the large central core , the progenitors of the present - day b30 and b35 dark clouds , and ( south of these ) the b223 cloud . the elongated structure of this cloud complex connected to the other linear structures to the northwest and southeast which we see today in the co map shown in panel d of figure [ fig : schematic ] . over the next few myr , stars began to form in the densest portions of this cloud chain . at 6 myr ago , a dozen ob stars formed near @xmath1 ori s present - day position ( panel b ) . the local gas density remained high enough that they could not ionize or disperse more than a small fraction of the gas . thus , they remained bound to their natal cloud . at the same time , the birth rate of low - mass stars increased in all productive areas of the star - forming complex . many of these were unaffected by the massive stars many parsecs away , but the closest pms stars passed in close proximity to the ob stars and lost their circumstellar disks to the fuv radiation ( paper i ) . then , shortly before the present day , one of the o stars ( perhaps a binary companion of @xmath1 ori itself ) became a supernova ( panel c ) . the supernova blast encountered a non - uniform medium filled with a small region forming around the ob stars , dense gas in the immediate vicinity of the ob stars , a few massive clouds about 1015 parsecs away , and the rest of the volume filled by somewhat lower density molecular gas . the blast quickly dispersed all of the parent core , creating the molecular ring , the large region , and the nearby structures @xcite . however , when the shock reached the more distant , massive b30 and b35 clouds it swept around them . thus today we see the fossil distribution of young stars within the molecular ring , as well as the remnants of the b30 and b35 clouds within the ionized region ( panel d ) . in this scenario we suggest that much of the gas from the parent core has moved as a consequence of the supernova , while the other massive clouds have moved little . this picture differs from the all - expanding model of @xcite and the all - stationary model of paper ii . the maddalena & morris model is contradicted because the accelerating gas should have left behind all stars formed therein , whereas we find stars of all ages still projected on b30 and b35 ( implying that these clouds have not moved since the oldest stars were born ) . also incorrect is our paper ii model of the star - forming region as a single giant cloud with a hole torn from its center , because we have found an annulus inside the co ring absent of pms stars . this absence implies that the annulus never had clouds of high enough density to form stars . examining figure [ fig : pmsmap ] , one can see the co ring is not actually centered on @xmath1 ori . we interpret this as evidence of non - uniform gas density : the supernova shock expanded more easily towards ( and around ) the smaller b35 cloud to the east than towards b30 to the northwest . over several hundred thousand years , the supernova shock snowplowed the gas outward to where we see the cold ring today . because the stars have a radial velocity similar to the b30/35 clouds ( comparing our data to the @xcite and @xcite co data ) while the gas has a mean velocity several km s@xmath34 blueward ( from the wisconsin h@xmath4 mapper survey ; l. m. haffner 1998 , private communication ) , it seems the shock must have expanded more readily into lower density gas in the foreground . certainly , the b30 and b35 clouds have not been totally immune to the shock wave . the inward - facing sides of the clouds have been heavily eroded by the blast . we see a signature of this in the strong density gradients in the molecular clouds facing towards @xmath1 ori , seen in figure [ fig : pmsmap2 ] . this is a consequence of either compression by the supernova ( and/or subsequent ob winds ) or by the dispersal of the outer , low - density envelope of the cloud . these same faces show bright rims today @xcite as they are continuing to be ionized by the remaining ob stars . we also see evidence of the partial destruction of b30 and b35 in the pms stars which lie slightly in front of the clouds , tracing the former extent of the gas . we reiterate that because the stars in front of the clouds have a wide range of ages matching the age span of the entire association , this enhanced pms population in front of the clouds is likely a sign of cloud destruction , not cloud acceleration or triggered star formation . it appears that the supernova had a more devastating effect on b35 than b30 , since the former appears to have ceased embedded star formation while the latter still contains many far - ir sources @xcite . all of the pms candidates near b35 in figure [ fig : pmsmap2 ] lie on the side of the cloud facing @xmath1 ori ; none lie on the low - surface - density `` neck '' connecting b35 to the co ring . we suspect that the b35 cloud will vanish in a few myr , especially if @xmath1 ori becomes a supernova soon . there are still enigmas left in the structure of the gas which are not explained fully by our scenario . first is the history of the b223 dark cloud in the southwest of the star - forming complex . @xcite noted that b223 is blueshifted relative to the rest of the clouds and interpreted this as evidence of expansion of the gas . however , we have found that all of the pms stars are at a similar radial velocity to b30 and b35 , not at a system velocity intermediate between b30 , b35 and b223 as would be expected if all clouds were expanding from a common center . the co surface density for the b223 cloud is quite high , such that its mass is likely comparable to that of b30 . thus , it seems likely that if b30 was unmoved by the supernova , b223 should be similarly stationary . the radial - velocity difference of b223 suggests that either ( 1 ) b223 was not originally connected to the @xmath1 ori progenitor cloud , or ( 2 ) b223 is the result of the projection or collision of two clouds : one ejected from the @xmath1 ori vicinity and a denser , unrelated one . this latter scenario is appealing , since the velocity profile presented by @xcite shows that b223 spans almost 10 km s@xmath34 in radial velocity , which could be a sign of two superimposed clouds . another puzzle is the presence of a pair of low - surface - density co clouds projected very near to @xmath1 ori ( figure [ fig : pmsmap2 ] ) . if these clouds are at the same distance as the ob stars , then it is unclear how they could have survived the massive star radiation , let alone a supernova . instead , we suspect that these thin clouds must be in front of or behind the stellar association . a detailed velocity map with the sensitivity of the @xcite survey might help to solve this dilemma . the scenario we have described includes stars forming near the massive , stationary clouds , but not elsewhere . in the vicinity of the co ring , where we can not photometrically sift the pms candidates from the field stars , we do not know how many members may exist . our model of expanding , low - mass cloudlets implies that there should be no old stars in the majority of the ring . a detailed spectroscopic survey in the vicinity of these clouds would answer the question of whether star formation is occurring there or not . we suspect that there could be more stars forming just outside the ring , south of betelgeuse , where the @xmath1 ori region reaches the `` northern filament '' identified by @xcite . the total mass of stars identified as candidate members is roughly 500 @xmath2 , which is comparable to an open cluster , but spread over a much larger area ( about 3040 pc diameter ) . the kinematics indicate that this association will be completely dispersed into the field within a few @xmath53 10 myr . thus , the @xmath1 ori region may be representative of the process of star - formation in moderate - mass molecular clouds . we have presented photometry of over 300,000 stars in the @xmath1 ori region . using the spectroscopic survey from papers i and ii to define the photometric characteristics of pms stars , we culled more than 99.9% of the field stars from our sample . further statistical field star subtraction then leaves a representative sample of pms candidates which has the same spatial distribution and population size as the true pms stars in the association . this sample presented us with a representative snapshot of the present - day low - mass young stellar population of the star - forming complex . in combination with the more detailed , but spatially - limited , results from papers i and ii , this snapshot allowed us to deduce the chronology of the region : * 10 myr ago a long chain of molecular gas extended from east to west across the present - day star - forming region , including three particularly massive clouds . * 6 myr ago stars formed in the most massive clouds , but the onset of formation was not sudden . instead , the birth rate increased gradually over many million years . numerous ob stars were born in the central cloud , but were very rare elsewhere . * 1 myr ago a supernova exploded , shredding the central cloud and thus unbinding the central stellar population . a ring of gas was pushed from the center region . * today star formation continues in the b30 and b35 clouds , but has ceased in the vicinity of the supernova epicenter . * future the termination of star birth in b35 is imminent , as is the escape of the ob stars from their central position . all of the stars will disperse into the field over the next 10 myr or so . the gas may also be further dispersed by subsequent supernovae . we have found that globally this star - forming region has generated a mass distribution similar to the field population , but the imf is spatially non - homogeneous : the center region strongly favors massive stars , while the periphery is heavily biased toward low - mass stars . if the @xmath1 ori region is typical of star formation in medium - mass molecular clouds , then this history tells us that these star / cloud associations are short - lived : they terminate themselves from within and are not detectable for more than a few tens of myr . it is not clear what signals the commencement or acceleration of star birth : in the @xmath1 ori region , stars began to form at a low rate everywhere at about the same time , but most of the stars were formed recently . it would be very enlightening to identify analogues of the @xmath1 ori region at earlier and later stages of evolution to study the turning points of star formation history . monet , d. , bird , a. , canzian , b. , harris , h. , reid , n. , rhodes , a. , sell , s. , ables , h. , dahn , c. , guetter , h. , henden , a. , leggett , s. , levison , h. , luginbuhl , c. , martini , j. , monet , a. , pier , j. , riepe , b. , stone , r. , vrba , f. & walker , r. 1996 , usno - a1.0 , ( u.s . naval observatory , washington dc ) ccccl 1 & dec 15 & 11 & very good , then cirrus just before dawn + 2 & dec 16 & 14 & cirrus at dusk , then very good + 3 & dec 17 & 13 & perfect + 4 & dec 18 & 0 & thick clouds , no data + 5 & dec 19 & 6 & cirrus early , then clear with bright moon + 6 & dec 20 & 17 & full moon , no clouds + ccccccc lori j051750.9 + 103824 & 16.738 & 15.993 & 16.838 & 05 17 50.916 & 10 38 23.84 & 146 + lori j051751.2 + 102548 & 18.714 & 17.319 & 16.622 & 05 17 51.197 & 10 25 47.72 & 146 + lori j051751.5 + 105655 & 15.396 & 14.862 & 14.282 & 05 17 51.502 & 10 56 55.41 & 146 + lori j051751.5 + 103716 & 16.674 & 15.930 & 15.304 & 05 17 51.515 & 10 37 16.45 & 146 + lori j051751.6 + 105902 & 17.676 & 17.070 & 16.314 & 05 17 51.593 & 10 59 01.69 & 146 +

bernard et al . ( 2015 ) study an optimal insurance design problem where an individual s preference is of the rank - dependent utility ( rdu ) type , and show that in general an optimal contract covers both large and small losses . however , their contracts suffer from a problem of moral hazard for paying more compensation for a smaller loss . this paper addresses this setback by exogenously imposing the constraint that both the indemnity function and the insured s retention function be increasing with respect to the loss . we characterize the optimal solutions via calculus of variations , and then apply the result to obtain explicitly expressed contracts for problems with yaari s dual criterion and general rdu . finally , we use a numerical example to compare the results between ours and that of bernard et al . ( 2015 ) . * keywords : * optimal insurance design , rank - dependent utility theory , yaari s dual criterion , probability weighting function , moral hazard , indemnity function , retention function , quantile formulation .

1509.04839

optimal insurance contract design is an important problem , manifested not only in theory but also in insurance and financial practices . the problem is to determine the optimal amount of compensation as a function of the loss called indemnity so as to maximize the insured s satisfaction , subject to the participation constraint of the insurer . in the insurance literature , most of the work assume that the insurer is risk neutral while the insured is a risk - averse expected utility ( eu ) maximizer ; see e.g. arrow ( 1963 ) , raviv ( 1979 ) , and gollier and schlesinger ( 1996 ) . in this case , the optimal contract is in general a deductible one that covers part of the loss in excess of a deductible level . however , the eu theory has received many criticisms , for it fails to explain numerous experimental observations and theoretical puzzles . in the context of insurance contracting , the classical eu - based models can not explain some behaviors in insurance demand such as that for small losses ( e.g. demand for warranties ) ; see a detailed discussion in bernard et al . ( 2015 ) . in order to overcome this drawback of the eu theory , different measures of evaluating uncertain outcomes have been put forward to depict human behaviors . a notable one is the rank - dependent utility ( rdu ) proposed by quiggin ( 1982 ) , which consists of a concave utility function and an inverse-@xmath0 shaped probability weighting ( or distortion ) function . through the probability weighting , the rdu theory captures the common observation that people tend to exaggerate small probabilities of extremely good and bad outcomes . with the development of advanced mathematical tools , the rdu preference has been applied to many areas of finance , including portfolio choice and option pricing . on the other hand , barseghyan et al . ( 2013 ) use data on households insurance deductible decisions in auto and home insurance to demonstrate the relevance and importance of the probability weighting and suggest the possibility of generalizing their conclusions to other insurance choices . there have been also studies in the area of insurance contract design within the rdu framework ; see for example chateauneuf , dana and tallon ( 2000 ) , dana and scarsini ( 2007 ) , and carlier and dana ( 2008 ) . however , all these papers assume that the probability weighting function is convex . bernard et al . ( 2015 ) are probably the first to study rdu - based insurance contracting with _ inverse-@xmath0 shaped _ weighting functions , using the quantile formulation originally developed for portfolio choice ( jin and zhou 2008 , he and zhou 2011 ) . they derive optimal contracts that not only insure large losses above a deductible level but also cover small ones . however , their contracts suffer from a severe problem of moral hazard , since they are not increasing with respect to the losses . is increasing if @xmath1 whenever @xmath2 . we say @xmath3 is `` strictly increasing '' if @xmath4 whenever @xmath2 . similar conventions are used for `` decreasing '' and `` strictly decreasing '' functions . ] as a consequence , insureds may be motivated to hide their true losses in order to obtain additional compensations ; see a discussion on pp . 175176 of bernard et al . ( 2015 ) . this paper aims to address this setback . we consider the same insurance model as in bernard et al . ( 2015 ) , but adding an explicit constraint that both the indemnity function and the insured s retention function ( i.e. the part of the losses to be born by the insured ) must be globally increasing with respect to the losses . this constraint will rule out completely the aforementioned behaviour of moral hazard ; yet mathematically it gives rise to substantial difficulty . the approach used in bernard et al . ( 2015 ) no longer works . we develop a general approach to overcome this difficulty . specifically , we first derive the necessary and sufficient conditions for optimal solutions via calculus of variations . then we deduce explicitly expressed optimal contracts by a fine analysis on these conditions . an interesting finding is that , for a good and reasonable range of parameters specifications , there are only two types of optima contracts , one being the classical deductible one and the other a three - fold " one covering both small and large losses . the remainder of the paper is organized as follows . section 2 presents the optimal insurance model under the rdu framework including its quantile formulation . section 3 applies the calculus of variations to derive a general necessary and sufficient condition for optimal solutions . we then derive optimal contracts for yaari s criterion and the general rdu in sections 4 and 5 , respectively . section 6 provides a numerical example to illustrate our results . finally , we conclude with section 7 . proofs of some lemmas are placed in an appendix . in this section , we present the optimal insurance contracting model in which the insured has the rdu type of preferences , followed by its quantile formulation that will facilitate deriving the solutions . we follow bernard et al . ( 2015 ) for the problem formulation except for one critical difference , which we will highlight . let @xmath5 be a probability space . an insured , endowed with an initial wealth @xmath6 , faces a non - negative random loss @xmath7 supported in @xmath8 $ ] , where @xmath9 is a given positive scalar . he chooses an insurance contract to protect himself from the loss , by paying a premium @xmath10 to the insurer in return for a compensation ( or _ indemnity _ ) in the case of a loss . this compensation is to be determined as a function of the loss @xmath7 , denoted by @xmath11 throughout this paper . the _ retention _ function @xmath12 is thereby the part of the loss to be borne by the insured . for a given @xmath7 , the insured aims to choose an insurance contract that provides the best tradeoff between the premium and compensation based on his risk preference . in this paper , we consider the case when insured s preference is of the rdu type . this rdu preference consists of two components : a @xmath13 function @xmath14 and a probability @xmath15 function @xmath16\mapsto [ 0,1]$ ] . let us denote by @xmath17 the rdu of the final ( random ) wealth @xmath18 of an insured , which is a choquet integral of @xmath19 with respect to the capacity @xmath20 , i.e. , @xmath21,\end{aligned}\ ] ] where @xmath22 is the cumulative distribution function ( cdf ) of @xmath18 . assuming that @xmath23 is differentiable , we can rewrite @xmath24 if @xmath23 is inverse-@xmath0 shaped , that is , it is first concave and then convex ; see figure 1 , then the above expression shows that the role @xmath23 plays is to overweigh both tails of @xmath18 when evaluating the mean of @xmath19 . on the other hand , if the insurer is risk - neutral and the cost of offering the compensation is proportional to the expectation of the indemnity , then the premium to be charged for an insurance contract should satisfy the participation constraint @xmath25,\ ] ] where the constant @xmath26 is the _ safety loading _ of the insurer . it is natural to require an indemnity function to satisfy @xmath27 a constraint that has been imposed in most insurance contracting literature . if the insured s preference is dictated by the classical eu theory , then the optimal contract is typically a deductible contract which automatically renders the indemnity function increasing ; see e.g. arrow ( 1971 ) and raviv ( 1979 ) . however , for the rdu preference the resulting optimal indemnity may not be an increasing function , as shown in bernard et al . this may potentially cause moral hazard as pointed out earlier . similarly , a non - monotone retention function may also lead to moral hazard . consequently , to include the increasing constraint on the contract has been an outstanding open question . in this paper , we require the indemnity function to satisfy @xmath28 and @xmath29 . in other words , we constrain _ both _ indemnity and retention functions to be globally increasing . we can now formulate our insurance contracting problem as @xmath30\leqslant \pi , \\ & \quad i(\cdot)\in \mathbb{i } , \end{array}\end{aligned}\ ] ] where @xmath31 and @xmath6 and @xmath10 are fixed scalars . for any random variable @xmath32 a.s . , define the quantile function of @xmath33 as @xmath34.\ ] ] note that any quantile function is nonnegative , increasing and left - continuous ( ilc ) . we now introduce the following assumptions that will be used hereafter . [ assump:21 ] the random loss @xmath7 has a strictly increasing distribution function @xmath35 . moreover , @xmath36 is absolutely continuous on @xmath37 $ ] . [ assump:22 ] ( concave utility ) the utility function @xmath14 is strictly increasing and continuously differentiable . furthermore , @xmath38 is decreasing . [ assump:23 ] ( inverse-@xmath0 shaped weighting ) the probability weighting function @xmath39 is a continuous and strictly increasing mapping from [ 0,1 ] onto [ 0,1 ] and twice differentiable on @xmath40 . moreover , there exists @xmath41 such that @xmath42 is strictly decreasing on @xmath43 and strictly increasing on @xmath44 . furthermore , @xmath45 and @xmath46 . the first part of assumption [ assump:21 ] , crucial for the quantile formulation , is standard ; see e.g. raviv ( 1979 ) . note a significant difference from bernard et al . ( 2015 ) is that here we allow @xmath7 to have atoms ( which is usually the case in the insurance context ) . for example , let @xmath7 be distributed with @xmath47 for @xmath48 $ ] , where @xmath49 and @xmath50 . then , @xmath7 satisfies assumption [ assump:21 ] , and has an atom at @xmath51 with the probability @xmath52 . this assumption also ensures that @xmath53 $ ] , a fact that will be used often in the subsequent analysis . next , assumption [ assump:22 ] is standard for a utility function . finally , assumption [ assump:23 ] is satisfied for many weighting functions proposed or used in the literature , e.g. that proposed by tversky and kahneman ( 1992 ) ( parameterized by @xmath54 ) : @xmath55 figure 1 displays this ( inverse-@xmath0 shaped ) weighting function ( in blue ) when @xmath56 . [ figure : distortion ] shaped weighting function satisfying assumption [ assump:23 ] . the marked points @xmath57 and @xmath58 will be explained later . , title="fig:",width=4 ] in practice , most of the insurance contracts are not tailor - made for individual customers . instead , an insurance company usually has contracts with different premiums to accommodate customers with different needs . each contract is designed with the best interest of a representative customer in mind so as to stay marketable and competitive , while maintaining the desired profitability ( the participation constraint ) . an insured can then choose one from the menu of contracts to cater for individual needs . problem is therefore motivated by the insurer s making of this menu . if the premium @xmath59 $ ] , then @xmath60 ( corresponding to a full coverage ) is feasible and maximizes the objective function in problem pointwisely ; hence optimal . to rule out this trivial case , henceforth we restrict @xmath61 $ ] . moreover , we assume @xmath62-m\geqslant 0,\end{aligned}\ ] ] to ensure that the policyholder will not go bankrupt . this is because @xmath63 , @xmath64)$ ] . it is more convenient to consider the retention function @xmath65 instead of @xmath66 in our study below . letting @xmath67-\frac{\pi}{1+\rho } \in ( 0,e[x]),\ w:=w_0-(1+\rho)e[x]>0,\ w_\delta:=w+(1+\rho)\delta\equiv w_0-\pi$ ] , we can reformulate in terms of @xmath68 : @xmath69\geqslant \delta , \\ & \quad r(\cdot)\in \mathcal{r } , \end{array}\end{aligned}\ ] ] where @xmath70 the objective function in is not concave in @xmath71 ( due to the nonlinear weighting function @xmath23 ) , leading to a major difficulty in solving . however , under assumption [ assump:23 ] , we have @xmath72 \\ = & & \int_0 ^ 1 u(f_{w_\delta - r(x)}^{-1}(z))t'(1-z)dz= \int_0 ^ 1 u(w_\delta - f_{r(x)}^{-1}(1-z))t'(1-z)dz \\ = & & \int_0 ^ 1 u(w_\delta - f_{r(x)}^{-1}(z))t'(z)dz,\end{aligned}\ ] ] where the third equality is because @xmath73 except for a countable set of @xmath74 . on the other hand , @xmath75\geqslant \delta$ ] is equivalent to @xmath76 . the above suggests that we may change the decision variable from the random variable @xmath71 to its quantile function @xmath77 , with which the objective function of becomes concave and the first constraint is linear . it remains to rewrite the monotonicity constraint ( represented by the constraint set @xmath78 ) also in terms of @xmath77 . to this end , the next lemma plays an important role . [ lemma : quantile ] under assumption [ assump:21 ] , for any given @xmath79 , we have @xmath80.\ ] ] first , by the monotonicity of @xmath68 , we have @xmath81 , so by the definition of @xmath82 , we conclude that @xmath83 . it suffices to prove the reverse inequality . consider two cases . * @xmath84 : in this case , we have @xmath85 as quantile functions are nonnegative . * @xmath86 : it suffices to prove that @xmath87 for any @xmath88 . take @xmath89 such that @xmath90 . by the continuity and monotonicity of @xmath68 , there exists @xmath91 such that @xmath92 and @xmath93 . then , @xmath94 , where we have used the fact that @xmath35 is strictly increasing under assumption [ assump:21 ] . the claim is thus proved . in view of the above results , we can rewrite as the following problem , in which the decision variable is @xmath95 ( denoted by @xmath96 for simplicity ) : @xmath97 where @xmath98 . in the absence of an explicit expression the constraint set @xmath99 is hard to deal with . the following result addresses this issue . note the major technical difficulty arises from the possible existence of the atoms of @xmath7 . under assumption [ assump:21 ] , we have @xmath100 \}.\ ] ] we denote the right hand side of by @xmath101 . for any @xmath102 , there exists @xmath79 such that @xmath103 . for any @xmath104 , define @xmath105 : r(x)=g(a)\}$ ] , @xmath106 : r(x)=g(a)\}$ ] , define @xmath107 and @xmath108 similarly . let us show that @xmath109 . in fact , by definition , @xmath110 suppose @xmath111 for some @xmath112 . then by monotonicity , @xmath113 where we have used the fact that @xmath114 to get the last inequality . this leads to a contradiction ; hence it must hold that @xmath115 . similarly , we can prove @xmath116 . then we have @xmath117 this inequality shows that @xmath118 is absolutely continuous since @xmath119 is an absolutely continuous function under assumption [ assump:21 ] . furthermore , it also implies @xmath120 a.e . @xmath121.$ ] so we have established that @xmath122 . to prove the reverse inclusion , take any @xmath123 and define @xmath124 . it follows from assumption [ assump:21 ] that @xmath125 and @xmath126 @xmath127 . hence @xmath128 . it now suffices to show @xmath129 for any @xmath130 . if @xmath131 , then @xmath132 holds otherwise , for any @xmath133 , there exists @xmath91 such that @xmath134 by the continuity of @xmath68 . then by the monotonicity of @xmath68 and @xmath96 , we have @xmath135 which means @xmath132 . using the same notation , @xmath136 , as above , and noting that @xmath137 , we have @xmath138 by the definition of @xmath136 and the continuity of @xmath68 . moreover , it follows from @xmath139 that @xmath140 . therefore , @xmath141 holds by monotonicity . the desired result follows . to solve , we apply the lagrange dual method to remove the constraint @xmath142 and consider the following auxiliary problem : @xmath143dz -\lambda\delta , \\ \mathrm{s.t . } & \quad g(\cdot)\in { \mathbb{g}}. \end{array}\ ] ] the existence of the optimal solutions to and ( for each given @xmath144 ) is established in appendix b , while the uniqueness is straightforward when the utility function @xmath145 is strictly concave . to derive the optimal solution to , we first solve to obtain an optimal solution , denoted by @xmath146 . then we determine @xmath147 by binding the constraint @xmath148 . a standard duality argument then deduces that @xmath149 is an optimal solution to . finally , an optimal solution to is given by @xmath150 @xmath151 $ ] and that to ( 1 ) by @xmath152 @xmath151 $ ] . so our problem boils down to solving . however , in doing so the constraint that @xmath153 in @xmath154 poses the major difficulty compared with bernard et al . in this section , we derive a necessary and sufficient condition for a solution to be optimal to . assume @xmath146 solves with a fixed @xmath155 . let @xmath156 be arbitrary and fixed . for any @xmath157 , set @xmath158 . then @xmath159 . by the optimality of @xmath146 and the concavity of @xmath145 , we have @xmath160 dz-\int_0 ^ 1 \left[u(w_\delta-\widetilde{g}_\lambda(z))t'(z)+\lambda \widetilde g_\lambda(z)\right]dz\right\ } \\ \nonumber = & & \frac{1}{\varepsilon}\left\{\int_0 ^ 1 \left[(u(w_\delta - g^\epsilon(z))-u(w_\delta-\widetilde{g}_\lambda(z)))t'(z)+ \lambda ( g^\epsilon(z)- \widetilde{g}_\lambda(z))\right]dz\right\}\\ \nonumber \geqslant & & \frac{1}{\varepsilon}\left\{\int_0 ^ 1 \left[(u'(w_\delta - g^\epsilon(z)))(w_\delta - g^\epsilon(z)-w_\delta+\widetilde{g}_\lambda(z))t'(z)+ \lambda ( g^\epsilon(z)- \widetilde{g}_\lambda(z))\right]dz\right\}\\ \nonumber \underrightarrow{\epsilon \downarrow 0 } & & \int_0 ^ 1 \left[(u'(w_\delta-\widetilde{g}_\lambda(z)))(\widetilde{g}_\lambda(z)-g(z))t'(z)+ \lambda ( g(z)- \widetilde{g}_\lambda(z))\right]dz \\ = & & \int_0 ^ 1 \left[u'(w_\delta-\widetilde{g}_\lambda(z))t'(z)-\lambda \right](\widetilde{g}_\lambda(z)-g(z))dz.\end{aligned}\ ] ] define @xmath161 dt,\quad z\in[0,1].\end{aligned}\ ] ] then yields @xmath162(\widetilde{g}_\lambda(z)-g(z))dz = \int_0 ^ 1 \int_0^z ( \widetilde{g}_\lambda'(t)-g'(t ) ) dt dn_\lambda(z ) \\ = & \int_0 ^ 1 \int_t^1 ( \widetilde{g}_\lambda'(t)-g'(t ) ) dn_\lambda(z)dt = \int_0 ^ 1 n_\lambda(t ) ( g'(t)-\widetilde{g}_\lambda'(t))dt,\end{aligned}\ ] ] leading to @xmath163 in other words , @xmath164 maximizes @xmath165 over @xmath166 . therefore , a necessary condition for @xmath146 to be optimal for is @xmath167 dt < 0 , \\ \in [ 0,(f^{-1}_x)'(z ) ] , & \quad \text { if } n_\lambda(z)=\int_z^1 [ \lambda- u'(w_\delta-\widetilde{g}_\lambda(t))t'(t ) ] dt = 0,\\ ( f^{-1}_x)'(z ) , & \quad \text { if } n_\lambda(z)=\int_z^1 [ \lambda- u'(w_\delta-\widetilde{g}_\lambda(t))t'(t ) ] dt > 0 . \end{cases}\end{aligned}\ ] ] it turns out that completely characterizes the optimal solutions to . [ theorem : main ] a function @xmath146 is an optimal solution to if and only if @xmath168 and @xmath146 satisfies . we only need to prove the `` if '' part . for any feasible @xmath96 in @xmath154 , we have @xmath169t'(z ) dz + \int_0 ^ 1 \lambda ( \widetilde{g}_\lambda(z)- { g}(z))dz \\ \geqslant & \int_0 ^ 1 u'(w_\delta-\widetilde{g}_\lambda(z))({g}(z ) - \widetilde{g}_\lambda(z))t'(z ) dz - \int_0 ^ 1 \lambda ( { g}(z ) - \widetilde{g}_\lambda(z ) ) dz \\ = & \int_0 ^ 1 n_\lambda'(z)({g}(z ) - \widetilde{g}_\lambda(z ) ) dz = \int_0 ^ 1 n_\lambda(t ) ( \widetilde{g}_\lambda'(t ) - { g}'(t))dt \geqslant 0.\end{aligned}\ ] ] hence , @xmath146 is optimal for . the above theorem establishes a general characterization result for the optimal solutions of . this result , however , is only implicit as an optimal @xmath146 appears on both sides of . moreover , the derivative of @xmath170 is undetermined when @xmath171 . in the next two sections , we will apply this general result to derive the solutions . when @xmath172 , the corresponding @xmath173 reduces to the so - called yaari s dual criterion ( yaari 1987 ) . in this section we solve our insurance problem with yaari s criterion by applying theorem [ theorem : main ] . in this case , the condition is greatly simplified . indeed , when @xmath172 , reduces to + @xmath174 , & \quad \text { if } \int_z^1 ( \lambda - t'(t ) ) dt=\lambda ( 1-z)-(1-t(z ) ) = 0 , \\ ( f^{-1}_x)'(z ) , & \quad \text { if } \int_z^1 ( \lambda - t'(t ) ) dt=\lambda ( 1-z)-(1-t(z ) ) > 0 . \end{cases}\end{aligned}\ ] ] it should be noted that although @xmath172 is not strictly concave here , the uniqueness of optimal solution to is implied by the characterizing condition . to apply , we need to compare @xmath155 and @xmath175 . define @xmath176 , @xmath177 . [ desp ] the function @xmath178 is a continuous function on @xmath179 . moreover , under assumption [ assump:23 ] , there exists a unique @xmath180 such that @xmath178 is strictly decreasing on @xmath181 $ ] and strictly increasing on @xmath182 . we have @xmath183 , @xmath177 . let @xmath184 . then @xmath185 . it follows from assumption [ assump:23 ] that @xmath186 for @xmath187 and @xmath188 for @xmath189 . moreover , @xmath190 , @xmath191 , and @xmath192 ( noting @xmath193 is strictly convex on @xmath194 $ ] ) . so , there exists @xmath180 such that @xmath195 for @xmath196 and @xmath197 for @xmath198 . the desired result follows . clearly , @xmath199 . set @xmath200 . from the proof of lemma [ desp ] , @xmath57 is determined by @xmath201 . let @xmath202 $ ] be the unique scalar such that @xmath203 or @xmath204 . see figure 1 for the locations of the points @xmath57 and @xmath58 . now , we proceed by considering three cases based on the value of @xmath155 . @xmath205 . in this case , @xmath206 $ ] . it then follows from @xmath207 that @xmath208 ; hence @xmath209 @xmath210 $ ] . thus the corresponding retention @xmath211 $ ] and indemnity @xmath212 @xmath151 $ ] , namely , the optimal contract is a full insurance contract . @xmath213 . by lemma [ desp ] , there exist unique @xmath214 and @xmath215 such that @xmath216 . accordingly , we have @xmath217 hence , leads to the following function : @xmath218 the corresponding retention and indemnity functions are , respectively , @xmath219 and @xmath220 the corresponding indemnity function is illustrated by figure 2 . qualitatively , the insurance covers not only large losses ( when @xmath221 ) but also _ small _ losses ( when @xmath222 ) , and the compensation is a constant for the median range of losses . we term such a contract a _ threefold _ one . the need for small loss coverage along with its connection to the probability weighting are amply discussed in bernard et al . however , in bernard et al . ( 2015 ) the optimal indemnity is strictly decreasing in some ranges of the losses . such a contract may incentivize the insured to hide partial losses in order to get more compensations . in contrast , both our indemnity and retention are increasing functions of the loss , which will rule out this sort of moral hazard . @xmath223 by lemma [ desp ] , there exists a unique @xmath224 $ ] such that @xmath225 . thus @xmath226 by , we have @xmath227 so @xmath228 this contract is a standard deductible contract in which only losses above a deductible point will be covered . we now summarize our results . define @xmath229 and let @xmath230 . clearly @xmath231 $ ] . [ prop : yaari ] under yaari s criterion , @xmath172 , and assumptions [ assump:21 ] and [ assump:23 ] , we have the following conclusions : a. if @xmath232 , then the optimal solution to is @xmath233 . b. if @xmath234 , then the optimal solution to is @xmath235 where @xmath236 is the unique pair satisfying @xmath237 and @xmath238 . c. if @xmath239 $ ] , then the optimal solution to is @xmath240 where @xmath241 the unique number satisfying @xmath242 and @xmath238 . a. when @xmath243 , the optimal solution to is trivially @xmath233 . b. when @xmath234 , there exists a unique pair @xmath236 such that @xmath237 and @xmath238 where @xmath244 is defined by . the existence of this pair follows from the condition that @xmath245 and the definition of @xmath246 , whereas the uniqueness comes from the requirement that @xmath247 and @xmath248 . letting @xmath249 , it is easy to show that @xmath250 satisfies under @xmath155 , corresponding to the aforementioned case 4.2 . this implies that @xmath250 is optimal for under @xmath251 . c. when @xmath239 $ ] , a case corresponding to case 4.3 , the desired result can be derived similarly as in ( ii ) . the proof is completed . we are now in the position to state our main result in terms of the premium @xmath10 and the indemnity function @xmath11 . denote by @xmath252-k_c)$ ] . under yaari s criterion , @xmath172 , and assumptions [ assump:21 ] and [ assump:23 ] , the optimal indemnity function @xmath253 to problem is given as a. if @xmath59 $ ] , then @xmath254 @xmath151 $ ] . b. if @xmath255 $ ] , then @xmath256 where @xmath236 is the unique pair satisfying @xmath237 and @xmath257=\frac{\pi}{1+\rho}$ ] . c. if @xmath258 , then @xmath259 where @xmath241 is the unique scalar satisfying @xmath242 and @xmath257=\frac{\pi}{1+\rho}$ ] . since @xmath260-\frac{\pi}{1+\rho}$ ] , the constraint @xmath75\equiv \int_0 ^ 1 { g}_{r(x)}(z ) dz = \delta$ ] is equivalent to that @xmath261=e[x]-e[r(x)]= \frac{\pi}{1+\rho}$ ] . the desired result is then a direct consequence of proposition [ prop : yaari ] . the economic interpretation of this result is clear . when the premium is small ( @xmath258 ) , the insurance only compensates large losses in excess of certain amount . when the premium is in middle range ( @xmath255 $ ] ) , the contract is a threefold one , covering both small and large losses . when the premium is sufficiently large ( @xmath59 $ ] ) , it is a full coverage . it is interesting to investigate the comparative statics of the point @xmath262 ( in terms of @xmath58 ) that triggers the coverage for small losses . in fact , as @xmath263 , we have @xmath264 however , @xmath265-k_c)$ ] ; hence @xmath266 this implies that the insurer is more willing to be protected against small losses if his weighting function has a bigger @xmath58 . this is consistent with the fact that a bigger @xmath58 renders a larger concave domain of the probability weighting that overweighs small losses ( refer to figure 1 ) . in this section we study the general rdu model in which the utility function is strictly concave . compared with the yaari model , solving the corresponding insurance problem calls for a more delicate analysis . for any twice differentiable function @xmath3 with @xmath267 , define its _ arrow - pratt measure of absolute risk aversion _ @xmath268 . we now introduce the following assumptions . [ scu ] ( strictly concave utility ) the utility function @xmath14 is strictly increasing and twice differentiable . furthermore , @xmath38 is strictly decreasing . [ ap ] a. the function @xmath269 is decreasing on @xmath270 . b. @xmath271 @xmath272 $ ] . assumption [ scu ] is to replace assumption [ assump:22 ] , ensuring a genuine rdu criterion . assumption [ ap]-(i ) requires that the absolute risk aversion measure of the utility function @xmath145 be decreasing , which holds true for many frequently used utility functions including logarithmic , power and exponential utilities . in general , experimental and empirical evidences are consistent with the decreasing absolute risk aversion ; see e.g. friend , irwin and blume , marshall ( 1975 ) . on the other hand , @xmath273 , @xmath274 $ ] , measures the level of probability weighting for small losses . the economical interpretation of assumption [ ap]-(ii ) is , therefore , that the degree of the insured s concern for small losses is sufficiently large relative to the absolute risk aversion of the utility function . note that assumption [ ap]-(ii ) is automatically satisfied when @xmath275 , @xmath276 $ ] , which is equivalent to @xmath277 . in practice , @xmath278 is a plausible assumption for many insurance products such as automobile and house insurance . on the other hand , @xmath57 is very small for many commonly used inverse-@xmath0 shaped weighting functions . take tversky and kahneman s weighting function as an example , @xmath279 when @xmath280 , @xmath281 when @xmath56 , and @xmath282 when @xmath283 . in these cases , assumption [ ap]-(ii ) holds automatically . problem has trivial solutions in the following two cases . when @xmath243 , the optimal solution is @xmath284 @xmath285 $ ] , corresponding to a full coverage . when @xmath260 $ ] , the optimal solution is @xmath286 @xmath285 $ ] as it is the only feasible solution , corresponding to no coverage . so we are interested in only the case @xmath287 $ ] . it follows from proposition [ prop - dual ] in appendix c that there exists @xmath288 such that @xmath289 is optimal solution to under @xmath288 and @xmath290 . furthermore , recall that we have proved that has a unique solution when @xmath145 is strictly concave and provides the necessary and sufficient condition for the optimal solution . [ lemma:51 ] for any @xmath166 , if there exists @xmath291 such that @xmath292 dt = 0 $ ] , then @xmath293 . from @xmath294 , it follows @xmath295 . hence , if @xmath296 , then @xmath297 dt \\ \leqslant & & \int_z^1 \left [ \lambda- u'(w_\delta - g(z))t'(t ) \right ] dt = \frac{\lambda}{t'(z)}(1-z ) \left[t'(z)-\frac{1-t(z)}{1-z}\right]<0,\end{aligned}\ ] ] where the last inequality is due to lemma [ lemma : a1]-(i ) in appendix a , noting @xmath296 . this is a contradiction . [ lemma:52 ] under assumption [ ap ] , for any @xmath298 , @xmath299 is a strictly decreasing function of @xmath74 on @xmath181 $ ] . noting @xmath300 , it follows from assumption [ ap ] that @xmath301 @xmath272 $ ] . now , we compute the partial derivative of @xmath299 with respect to @xmath302 $ ] : @xmath303 \\ < & & u'(w_\delta - g(z))t'(z)\left[a_u(w_\delta - g(z))g'(z)-a_u(w_\delta - f^{-1}_x(z))(f^{-1}_x)'(z)\right ] \\ \leqslant & & u'(w_\delta - g(z))t'(z)\left[a_u(w_\delta - g(z))(f^{-1}_x)'(z)-a_u(w_\delta - f^{-1}_x(z))(f^{-1}_x)'(z)\right ] \\ \leqslant & & u'(w_\delta - g(z))t'(z)\left[a_u(w_\delta - f^{-1}_x(z))(f^{-1}_x)'(z)-a_u(w_\delta - f^{-1}_x(z))(f^{-1}_x)'(z)\right ] \\ = & & 0.\end{aligned}\ ] ] the proof is complete . now , for any @xmath304 , we have @xmath305 dt \leqslant & & \int_z^1 \left [ \widehat{\lambda}u'(w_\delta)- u'(w_\delta)t'(t ) \right ] dt \\ = & & u'(w_\delta ) \int_z^1 [ \widehat{\lambda}-t'(t ) ] dt = u'(w_\delta)(1-z)\left[\widehat{\lambda}-\frac{1-t(z)}{1-z}\right ] < 0,\end{aligned}\ ] ] where the last inequality is due to lemma [ desp ] . hence @xmath306 @xmath307 $ ] is the only solution satisfying . however , @xmath308 , a contradiction . therefore , only when @xmath309 is it possible for to hold . fixing @xmath309 , we now analyze the shape of the function @xmath146 that satisfies . assume that @xmath310 . we have @xmath311 and @xmath312 since @xmath313 . so , @xmath314 when @xmath74 is close to 1 since @xmath315 for such @xmath74 . hence , @xmath316 $ ] for some @xmath317 , at which @xmath318 and @xmath315 for @xmath319 . next , we consider three cases respectively depending on the value of @xmath320 . \(a ) if @xmath321 ( i.e. @xmath322 ) , then we have , @xmath323 @xmath324 dt < & & \int_z^1 \left [ \widehat{\lambda}u'(w_\delta - k)- u'(w_\delta - k)t'(t ) \right ] dt \\ = & & u'(w_\delta - k)(1-z)\left[\widehat{\lambda}-\frac{1-t(z)}{1-z}\right ] \leqslant 0.\end{aligned}\ ] ] it then follows from that @xmath325 . however , @xmath326 , or @xmath327 , leading to a contradiction . so , this case in fact will not take place . \(b ) if @xmath328 , then @xmath89 should be @xmath57 . this is because @xmath329 dt=0 $ ] and @xmath330 dt=\frac{\lambda}{\widehat{\lambda}}(1-z)(\widehat{\lambda}-\frac{1-t(z)}{1-z})<0 $ ] for @xmath331 by lemma [ desp ] . moreover , @xmath332=0 . by lemma [ lemma:52 ] , @xmath333 strictly increases with respect to @xmath302 $ ] . it follows that @xmath334 for @xmath335 . then implies @xmath314 for @xmath336 . as a result , @xmath337 , or @xmath338 , which is a contradiction . so , again , this case will not occur . \(c ) if @xmath339 , then @xmath340 exists . by lemma [ lemma:51 ] , we have @xmath341 . hence , there may or may not exist @xmath342 such that @xmath343 and @xmath344 for @xmath345 . we now discuss four subcases depending on the existence and location of @xmath346 . ( c.1 ) if @xmath346 does not exist or @xmath347 ( i.e. @xmath344 for @xmath348 ) , then by , @xmath349 for @xmath348 . combined with the fact that @xmath350 , we have : @xmath351 this corresponds to a deductible contract . ( c.2 ) if @xmath346 exists and @xmath352 $ ] , then @xmath349 for @xmath345 in view of . combining the property of @xmath89 and @xmath346 , we deduce @xmath353 . then , using lemma [ lemma:52 ] , we have @xmath354 for @xmath355 . it follows from that @xmath314 for @xmath356 . in this case , we can express @xmath146 as follows @xmath357 this is the threefold contract , depicted in figure 2 . ( c.3 ) if @xmath346 exists and @xmath358 ( recall that @xmath359 is the turning point where the weighting function @xmath193 changes from being concave to convex ) , then a similar analysis as in case c.2 shows that @xmath360 and @xmath361 . this means @xmath362 . however , @xmath363 and @xmath364 , which is a contradiction . so , this case is not feasible . ( c.4 ) if @xmath346 exists and @xmath365 $ ] , then @xmath361 . we prove @xmath366 $ ] . in fact , if it is false , then there exists @xmath367 such that @xmath368 dt=0 $ ] and @xmath369 dt<0 $ ] for @xmath370 . however , @xmath371 for @xmath370 since @xmath365 $ ] . so , @xmath368 dt<(\lambda - u'(w_\delta-\widetilde{g}_\lambda(z_2))t'(z_2))(z_2-z_3)<0 $ ] , arriving at a contradiction . therefore , @xmath372 . from @xmath373 dt=0 $ ] , it follows @xmath374 . however , @xmath375 dt \\ > & & \int_{z_2}^{z_1 } \left [ u'(w_\delta+f^{-1}_x(z_2)-f^{-1}_x(z_1))\frac{1-t(z_1)}{1-z_1}- u'(w_\delta+f^{-1}_x(z_2)-f^{-1}_x(z_1))t'(t ) \right ] dt \\ = & & u'(w_\delta+f^{-1}_x(z_2)-f^{-1}_x(z_1))(z_1-z_2 ) \left [ \frac{1-t(z_1)}{1-z_1}- \frac{t(z_1)-t(z_2)}{z_1-z_2 } \right]>0 , \ ] ] where the last inequality follows from lemma [ lemma : a1]-(ii ) in appendix a. this is a contradiction . so , the current case will not occur either . to summarize , for any @xmath376 , only deductible and threefold contracts are possibly optimal , stipulated in case c.1 and case c.2 above . next , we investigate these two cases more closely . define a function @xmath377 on @xmath378 $ ] as follows : @xmath379 dt.\end{aligned}\ ] ] then , by using lemma [ lemma:52 ] and the fact that @xmath380 , we have @xmath381 dt < 0.\end{aligned}\ ] ] recalling that @xmath204 , we have @xmath382 dt \\ = & & \int_0^c [ u'(w_\delta - f^{-1}_x(c))-u'(w_\delta - f^{-1}_x(t))t'(t ) ] dt \\ > & & u'(w_\delta - f^{-1}_x(c))c - \int_0^c [ u'(w_\delta - f^{-1}_x(c))t'(t ) ] dt = 0 . \ ] ] moreover , we take the derivative of @xmath383 with respect to @xmath384 $ ] to obtain @xmath385 hence , there exists a unique point @xmath386 such that @xmath387 , @xmath388 for @xmath389 , and @xmath390 for @xmath391 . define @xmath392 and @xmath393 . [ prop:51 ] if @xmath394 $ ] , then the optimal solution to is @xmath395 where @xmath3 is the unique scalar such that @xmath396 and @xmath248 . the existence of @xmath3 follows from the monotonicity of @xmath397 with respesct to @xmath3 immediately . denoting @xmath398 , we need to show that @xmath250 satisfies with @xmath399 . first , it is straightforward that @xmath400 dt = 0 $ ] . next , we are to prove that @xmath401 dt>0 $ ] @xmath402 . we divide the proof into two cases . * if @xmath403 , then @xmath404 dt \\ \geqslant & \int_z^f \left [ \frac{u'(w_\delta - f^{-1}_x(f))(1-t(f))}{1-f}-u'(w_\delta - f^{-1}_x(f))t'(t ) \right ] dt \\ = & u'(w_\delta - f^{-1}_x(f))(f - z)\left[\frac{1-t(f)}{1-f}-\frac{1-t(z)}{1-z}\right ] > 0,\end{aligned}\ ] ] where the last inequality is due to lemma [ desp ] . * if @xmath405 and @xmath406 , then by lemma [ lemma:52 ] and the result above , we have @xmath404 dt \\ = & \int_z^a \left [ \frac{u'(w_\delta - f^{-1}_x(f))(1-t(f))}{1-f}-u'(w_\delta - f^{-1}_x(t))t'(t ) \right ] dt \\ & + \int_a^f \left [ \frac{u'(w_\delta - f^{-1}_x(f))(1-t(f))}{1-f}-u'(w_\delta - f^{-1}_x(t))t'(t ) \right ] dt \\ > & \int_z^a \left [ \frac{u'(w_\delta - f^{-1}_x(f))(1-t(f))}{1-f}-u'(w_\delta - f^{-1}_x(t))t'(t ) \right ] dt \\ \geqslant & \int_z^a \left[u'(w_\delta - f_x^{-1}(z))t'(z)-u'(w_\delta - f^{-1}_x(t))t'(t ) \right ] dt > 0.\end{aligned}\ ] ] if @xmath405 and @xmath407 holds , then @xmath408 dt \\ = & \int_{l_\delta}^f \left [ \frac{u'(w_\delta - f^{-1}_x(f))(1-t(f))}{1-f}-u'(w_\delta - f^{-1}_x(t))t'(t ) \right ] dt \\ & + \int_z^{l_\delta } \left [ \frac{u'(w_\delta - f^{-1}_x(f))(1-t(f))}{1-f}-u'(w_\delta - f^{-1}_x(t))t'(t ) \right ] dt \\ \geqslant & \int_z^{l_\delta } \left [ \frac{u'(w_\delta - f^{-1}_x(f))(1-t(f))}{1-f}-u'(w_\delta - f^{-1}_x(t))t'(t ) \right ] dt \\ \geqslant & \int_z^{l_\delta } \left [ \frac{u'(w_\delta - f^{-1}_x(l_\delta))(1-t(l_\delta))}{1-l_\delta}-u'(w_\delta - f^{-1}_x(t))t'(t ) \right ] dt \\ = & \int_0^{l_\delta } \left [ \frac{u'(w_\delta - f^{-1}_x(l_\delta))(1-t(l_\delta))}{1-l_\delta}-u'(w_\delta - f^{-1}_x(t))t'(t ) \right ] dt \\ & - \int_0^z \left [ \frac{u'(w_\delta - f^{-1}_x(l_\delta))(1-t(l_\delta))}{1-l_\delta}-u'(w_\delta - f^{-1}_x(t))t'(t ) \right ] dt \\ = & - \int_0^z \left [ \frac{u'(w_\delta - f^{-1}_x(l_\delta))(1-t(l_\delta))}{1-l_\delta}-u'(w_\delta - f^{-1}_x(t))t'(t ) \right ] dt > 0,\end{aligned}\ ] ] where the last inequality is due to @xmath409 as @xmath410 and the fact that @xmath411 is strictly decreasing on @xmath181 $ ] . the claim follows now . [ lemma:53 ] if @xmath412 , then the corresponding optimal contract is not a deductible one . there exists @xmath288 such that @xmath289 satisfies under @xmath288 and @xmath148 ( see appendix c ) . if @xmath289 corresponds to a deductible contract , then there exists @xmath413 ( since @xmath414 , we have @xmath415 ) such that @xmath416 since @xmath289 satisfies , we have @xmath417 dt = 0 $ ] , or @xmath418 . on the other hand , @xmath419 dt \geqslant 0 $ ] for @xmath420 $ ] . however , by the definition of @xmath421 , @xmath422 dt < 0 $ ] as @xmath415 . since @xmath423 is a continuous function , a contradiction arises . it follows from lemma [ lemma:53 ] that , if @xmath424 , the optimal contract ( which always exists ) can only be a threefold one , corresponding to case c.2 . we are now led to the following proposition . [ prop:52 ] if @xmath412 , then the optimal solution to is given as @xmath425 where @xmath426 are such that @xmath427 , @xmath428 dt=0 $ ] and @xmath429 . the conclusion is a direct consequence of lemma [ lemma:53 ] . note that any pair @xmath430 satisfying the requirements in proposition [ prop:52 ] leads to an optimal solution to . therefore such a pair @xmath430 is unique as the optimal solution to is unique . proposition [ prop:51 ] and proposition [ prop:52 ] give two qualitatively distinct optimal contracts for any given @xmath287 $ ] , and the two cases are divided depending on whether or not @xmath431 . however , @xmath432 in general depends on @xmath251 in an implicit and complicated way ; so it is hard to compare @xmath251 and @xmath432 . nevertheless we are able to treat at least two cases where @xmath269 is either a constant or strictly decreasing in @xmath74 . first , assume that the utility function exhibits constant absolute risk aversion , e.g. @xmath433 @xmath434 . then it is easy to see from that @xmath421 is independent of @xmath251 , and hence so is @xmath432 . in this case , denote @xmath435 and @xmath436-k)$ ] . then we have the following result . [ thm:54 ] assume that assumptions [ assump:21 ] , [ assump:23 ] , and [ ap ] hold , and that @xmath437 exhibits constant absolute risk aversion . then the optimal indemnity function @xmath253 to problem is given as a. if @xmath438 $ ] , then @xmath439 for @xmath440 $ ] . b. if @xmath441 $ ] , then @xmath442 where @xmath430 is the unique pair satisfying @xmath427 , @xmath428 dt=0 $ ] , and @xmath257=\frac{\pi}{1+\rho}$ ] . c. if @xmath443 , then @xmath444 where @xmath3 is the unique scalar satisfying @xmath257=\frac{\pi}{1+\rho}$ ] . the result follows from propositions [ prop:51 ] , [ prop:52 ] and the fact that @xmath432 is a constant for any @xmath287 $ ] . now , we study the case in which @xmath269 is strictly decreasing . we need the following lemma . [ lemma:55 ] if @xmath445 $ ] , then @xmath446 . according to definition of @xmath447 , we have @xmath448 dt=0.\end{aligned}\ ] ] since @xmath449 , we have @xmath450 for @xmath451 by lemma [ lemma : a2 ] in appendix a. hence @xmath452 dt<0.\end{aligned}\ ] ] as a result @xmath453 , @xmath454 . since @xmath455 for @xmath456 , we get @xmath457 . define @xmath458 on @xmath459 $ ] . then @xmath460 . hence , @xmath461 is a continuous and strictly increasing function . determine @xmath462 and @xmath463 by @xmath464 and @xmath465 , and set @xmath466 and @xmath467 . finally , define a function @xmath468 on @xmath378 $ ] as follows : @xmath469 dt.\end{aligned}\ ] ] [ prop:53 ] assume that assumptions [ assump:21 ] , [ assump:23 ] , and [ ap ] hold and that @xmath470 is strictly decreasing . then the optimal solution to is given as a. if @xmath232 , then @xmath471 @xmath472 . b. if @xmath473 , then @xmath474 where @xmath426 are such that @xmath427 , @xmath428 dt=0 $ ] , and @xmath429 . c. if @xmath475 , then let @xmath476 such that @xmath477 . if @xmath478 , then @xmath479 where @xmath426 are such that @xmath427 , @xmath428 dt=0 $ ] , and @xmath429 . if @xmath480 , then @xmath395 where @xmath3 is such that @xmath481 and @xmath248 . d. if @xmath482 $ ] , then @xmath483 where @xmath3 is such that @xmath484 and @xmath248 . ( i),(ii ) and ( iv ) are direct consequences of proposition [ prop:51 ] , [ prop:52 ] and lemma [ lemma:55 ] . for ( iii ) , there is a unique @xmath476 such that @xmath477 , which follows from the definition of @xmath485 , @xmath486 and the fact that @xmath461 is a continuous and strictly increasing function . if @xmath478 , then @xmath487 ; hence @xmath488 . therefore , @xmath431 . the desired result follows from proposition [ prop:52 ] . the proof for @xmath480 is similar . let us give the result in terms of premium and indemnity function . assume that assumptions [ assump:21 ] , [ assump:23 ] , and [ ap ] hold , and that @xmath470 is strictly decreasing . then the optimal indemnity function @xmath253 to problem ( 1 ) is given as a. if @xmath489 $ ] , then @xmath254 @xmath151 $ ] . b. if @xmath490-\widetilde{\delta})\leqslant \pi < ( 1+\rho)e[x]$ ] , then @xmath491 where @xmath430 is the unique pair satisfying @xmath427 , @xmath428 dt=0 $ ] , and @xmath257=\frac{\pi}{1+\rho}$ ] . c. if @xmath492-\overline{\delta } ) < \pi < ( 1+\rho)(e[x]-\widetilde{\delta})$ ] , then let @xmath476 such that @xmath493-\frac{\pi}{1+\rho}$ ] . if @xmath478 , then @xmath442 where @xmath430 is the unique pair satisfying @xmath427 , @xmath428 dt=0 $ ] , and @xmath257=\frac{\pi}{1+\rho}$ ] . if @xmath480 , then @xmath444 where @xmath241 is the unique number satisfying @xmath494 and @xmath257=\frac{\pi}{1+\rho}$ ] . d. if @xmath495-\overline{\delta})$ ] , then @xmath444 where @xmath241 is the unique number satisfying @xmath496 and @xmath257=\frac{\pi}{1+\rho}$ ] . it follows easily from proposition [ prop:53 ] . in this section , we use a numerical example to illustrate our result with a given premium @xmath10 . we take the same numerical setting as in bernard et al . ( 2015 ) for a comparison purpose : the loss @xmath7 follows a truncated exponential distribution with the density function @xmath497 , where the intensity parameter @xmath498 , and @xmath499 . the initial wealth @xmath500 , and @xmath501 with @xmath502 . moreover , @xmath503 and @xmath504 . finally , the weighting function @xmath505 with @xmath56 . we can verify that the assumptions of theorem [ thm:54 ] is satisfied under this setting . the optimal indemnity obtained by bernard et al . ( 2015 ) is plotted in blue in figure 3 . we note that in the result of bernard et al . ( 2015 ) , if the loss is between 1 and 2 , then the insured has the incentive to hide part of the loss in order to be paid with a larger compensation . by contrast , our indemnity function , depicted in red , is increasing and any increment in compensations is always less than or equal to the increment in losses . it effectively rules out the aforementioned behavior of moral hazard . in this paper , we have studied an optimal insurance design problem where the insured has the rdu preference . there are documented evidences proving that this preference captures human behaviors better than the eu preference . the main contribution of our work is that our optimal contracts are monotone with respect to losses , thereby eliminating the potential problem of moral hazard associated with the existing results . an interesting conclusion from our results is that , under our assumptions ( in particular assumption [ ap]-(ii ) ) , there are only two types of non - trivial optimal contracts possible , one being the classical deductible and the other the threefold contract covering both small and large losses . while we have demonstrated that assumption [ ap]-(ii ) holds for many economically interesting cases , removing this assumption remains a mathematically outstanding open problem . in this part , we prove some lemmas which have been used in section 5 . [ lemma : a1 ] assume @xmath506\mapsto [ 0,1]$ ] satisfies assumption [ assump:23 ] . we have the following results : + ( i ) if @xmath507 , then @xmath508 . + ( ii ) if @xmath509 and @xmath510 , then @xmath511 . \(i ) if @xmath512 , then @xmath513 . if @xmath514 , then @xmath508 since @xmath193 is convex and strictly increasing on @xmath194 $ ] . + ( ii ) since @xmath515 , @xmath516 , @xmath517 , and @xmath518 are all strictly positive , we have @xmath519 however , @xmath520 follows from lemma [ desp ] . for fixed @xmath521 , define @xmath522 on @xmath523 . [ lemma : a2 ] if @xmath524 is strictly decreasing , then @xmath525 is a strictly increasing function on @xmath526 . we take derivative : @xmath527 > 0.\end{aligned}\ ] ] hence , we get the result . we first prove that the constraint set @xmath99 is compact under some norm . we consider all the continuous functions on @xmath37 $ ] , denoted as @xmath528 $ ] . define a metric between @xmath529 as @xmath530 , @xmath531 $ ] . clearly , @xmath528 $ ] is a metric space under @xmath26 . by arzela ascoli s theorem , for any sequence @xmath532 in @xmath154 , there exists a subsequence @xmath533 that converges in @xmath528 $ ] under @xmath26 . [ lemma : b1 ] the feasible set @xmath534 is compact under @xmath26 . for any sequence @xmath532 in @xmath154 , there exists a subsequence @xmath533 that uniformly converges in @xmath535 $ ] . we now prove that @xmath536 . if there exist @xmath537 such that @xmath538 , then take @xmath539 . if follows from the uniform convergence that there exists @xmath540 such that @xmath541 @xmath542 . hence , @xmath543 @xmath542 , which is a contradiction . this proves that @xmath544 @xmath545 . similarly , we can prove that @xmath546 . the existence of optimal solutions to and can be established now . for example , for , let @xmath547 be the optimal value of under given @xmath155 and @xmath251 . we can take a sequence @xmath532 in @xmath154 such that @xmath548 . then , according to lemma [ lemma : b1 ] , there exists a subsequence @xmath533 converging to @xmath250 in @xmath154 and @xmath250 is optimal solution to . for , the proof is similar . for the following lemma , refer to komiya ( 1988 ) for an elementary proof . ( sion s minimax theorem ) let x be a compact convex subset of a linear topological space and y a convex subset of a linear topological space . if @xmath3 is a real - valued function on @xmath549 such that @xmath550 is continuous and concave on @xmath33 @xmath551 , and @xmath552 is continuous and convex on @xmath7 @xmath553 , then , @xmath554 let @xmath251 be given with @xmath287 $ ] . denote by @xmath250 the optimal solution to under @xmath251 ( it is easy to show @xmath429 ) and by @xmath555 the optimal solution to under @xmath155 and @xmath251 . denote by @xmath556 and @xmath557 be respectively the optimal values of and . therefore , by using the convexity of @xmath557 , we can find the optimal @xmath568 $ ] minimizes the right part , and satisfies that @xmath569 . moreover , @xmath570 dz = u_\delta(g^*(\cdot))=v(\delta).\end{aligned}\ ] ] the second equality comes from the fact that @xmath250 is the optimal solution to under @xmath251 ; hence @xmath429 . by @xmath569 and @xmath571 , we have @xmath250 is optimal solution to under given @xmath288 . and , by uniqueness of optimal solutions to , we know that @xmath250 is the unique optimal solution to under given @xmath288 and satisfying @xmath429 .

using a new _ xmm - newton _ observation , we have characterized the x - ray properties of the middle - aged radio - quiet-ray pulsar j0357 + 3205 ( named morla ) and its tail . the x - ray emission from the pulsar is consistent with a magnetospheric non - thermal origin plus a thermal emission from a hot spot ( or hot spots ) . the lack of a thermal component from the whole surface makes morla the coldest neutron star in its age range . we found marginal evidence for a double - peaked modulation of the x - ray emission . the study of the 9-long tail confirmed the lack of extended emission near the pulsar itself . the tail shows a very asymmetric brightness profile and its spectrum lacks any spatial variation . we found the nebular emission to be inconsistent with a classical bow - shock , ram - pressure dominated pulsar wind nebula . we propose thermal bremsstrahlung as an alternative mechanism for morla s tail emission . in this scenario , the tail emission comes from the shocked interstellar medium ( ism ) material heated up to x - ray temperatures . this can fully explain the peculiar features of the tail , assuming a hot , moderately dense interstellar medium around the pulsar . for a bremsstrahlung - emitting tail , we can estimate the pulsar distance to be between 300 and 900 pc . a pulsar velocity of km s is required - which would make morla the pulsar with the largest velocity - and high inclination angles ( ) are preferred . we propose morla s nebula as the first example of a new `` turtle s tail '' class of thermally - emitting nebulae associated to high velocity pulsars .

1212.6664

the dramatic increase in the number of known gamma - ray pulsars since the launch of the _ fermi gamma - ray space telescope _ offers the first opportunity to study a sizeable population of these high - energy sources . the _ fermi _ large area telescope ( lat , @xcite ) has discovered pulsed @xmath0-ray signals from more than 100 objects @xcite , revolutionizing our view of them and marking the birth of new high - energy pulsar sub - families , such as millisecond @xcite and radio - quiet @xmath0-ray pulsars @xcite , that are as numerous as the classical one . the wealth of detections confirms the importance of the @xmath0-ray channel in the overall energy budget of rotation - powered pulsars and points to emission models in which the @xmath0-ray production occurs in the outer magnetosphere along open - field lines ( outer gap / slot gap , @xcite ) and paves the way for our understanding of the three - dimensional structure and electrodynamics of neutron star magnetospheres . in a sense , the most important objects to constrain pulsar models are the `` extreme '' ones , accounting for the tails of the population distribution in energetics , age and magnetic field . in this respect , psr j0357 + 3205 is one of the most interesting pulsars discovered by the lat . j0357 + 3205 , which we name morla " ( cf . the neverending story " , m. ende ) for its slow rotation among _ fermi_-lat pulsars ( p=444 ms ) , was discovered in a blind search during the first three months of _ fermi_-lat scanning @xcite . it is located off the galactic plane , at a latitude of b@xmath5 - 16@xmath4 . while morla is not the oldest rotation - powered pulsar ( @xmath6=5.4 @xmath7 10@xmath8 yr ) - with an age similar to the well - known `` three musketeers '' @xcite - it is the slowest rotator among _ fermi _ pulsars . its spin - down luminosity , as low as @xmath9 = 5.9 @xmath7 10@xmath10 erg s@xmath2 , makes it one of the non - recycled @xmath0-ray pulsars with the smallest rotational energy loss detected so far . this suggests that psr j0357 + 3205 is quite nearby : by scaling its @xmath0-ray flux using the `` pseudo - distance '' relation from @xcite , we obtain a distance of @xmath1500 pc , making morla a natural target for x - ray observations . exploiting a joint _ chandra _ and _ noao _ program , @xcite found the x - ray counterpart of the @xmath0-ray pulsar to be located at r.a.(j2000 ) = 03@xmath1157@xmath1252@xmath13.32 , dec.(j2000 ) = + 32@xmath40520.6 `` ( 0.25 '' error radius , 68% confidence level ) . the pulsar x - ray emission was found to be consistent with a purely magnetospheric , non - thermal origin leading to a value of the @xmath0-to - x flux ratio of @xmath11000 , well below the mean for radio - quiet pulsars ( @xmath19000 , @xcite ) . unfortunately , the poor statistics precluded a clear detection of a thermal component in the neutron star spectrum as well as a full characterization of its column density and distance . nevertheless , the _ chandra _ data unveiled the presence of an extended structure protruding from the pulsar and extending @xmath39 in length , which would correspond to @xmath11.3 pc for a pulsar distance of 500 pc , and @xmath11.5 in width . tail `` is apparently detached from the pulsar , without detection of nebular counts until 50 '' from the pulsar ; it shows an unusual , asymmetric morphology , with the brightnest part far from the pulsar . _ noao _ optical images , taken in v and k bands , show no hints of correlated , diffuse emission . no extended source was detected either in public radio or infrared surveys . the lack of any h@xmath14 emission around the pulsar @xcite points to a full ionized , likely hot ( @xmath15 k ) , ism . + recently , using a new _ chandra _ observation , @xcite detected the pulsar s proper motionof 165@xmath1630 mas yr@xmath2 , in the direction away from the tail and parallel to the galactic plane . assuming a pulsar distance of 500 pc , this results in a lower limit velocity of 390 km s@xmath2 , making it a relatively high - velocity pulsar @xcite . many extended tails of x - ray emission have been discovered , associated with a number of rotation powered pulsars ( e.g. @xcite ) within the framework of a ram - pressure dominated pulsar wind nebula ( see @xcite for a review ) . however , the unusual phenomenology of the feature we observed is barely consistent with such a picture . there is no hint of diffuse emission in the pulsar surroundings , where bright emission from the wind termination shock should be seen ( as observed in other known cases ) . moreover , the asymmetric brightness profile of the tail , with its sharp north - eastern edge and its broad maximum at large distance from the pulsar is remarkably different from any other observed structure ( with the possible exception of the peculiar feature associated with psr b2224 + 65 , powering the guitar nebula - see @xcite ) . the mere existence of the tail is also problematic for a pulsar with such a low @xmath9 as j0357 + 3205 ( see e.g. @xcite ) . in order to assess the nature of the tail and its relationship with the pulsar , as well as to better constrain the pulsar emission properties and physics , we obtained a deep observation of the pulsar and its peculiar tail with _ xmm - newton_. exploiting its high spectral resolution , we were able to disentangle the components of the pulsar spectrum and constrain its distance based on the absorption column density . a complete study of the timing behavior of morla in x - rays was carried out as well . finally , we studied the properties of the tail , in order to characterize the mechanisms responsible for its emission . our deep _ xmm - newton _ observation of morla started on 2011 september 15 at 02:37:18 ut and lasted 111.3 ks . the pn camera @xcite of the epic instrument was operating in large window mode ( time resolution of 47.7 ms over a 14 @xmath7 27 field of view ( fov ) ) , while the mos detectors @xcite were set in full frame mode ( 2.6 s time resolution on a 15 radius fov ) . the thin optical filter was used for the pn camera while we chose to use the medium filter for the mos detectors due to the presence of moderately bright ( m@xmath179 ) stars . we used the _ xmm - newton _ science analysis software ( sas ) v11.0 . after standard data processing , using the epproc and emproc tools , and screening for high particle background time intervals ( following @xcite ) , the good , dead - time corrected exposure time was 98.5 ks for the pn and 108.3 ks for the two mos detectors . in order to fully characterize both the pulsar and the nebula , we also used the three available _ chandra_/acis @xcite observations of the field , performed on 2009 october 25 , 26 , and 2011 december 25 , lasting 29.5 , 47.1 , and 29.0 ks respectively ( these data sets were included in the investigations by @xcite and @xcite ) . we retrieved level 2 " data from the _ chandra _ science archive and used the _ chandra _ interactive analysis of observation ( ciao ) software v.4.2 . we also combined the _ spectra , response matrices and effective area files using the mathpha , addarf and addrmf tools . figure [ fig1 ] shows the 0.3 - 6 kev exposure - corrected _ xmm - newton _ fov image , where the pn and two mos images were added with the ximage tool . we applied a gaussian filter with a kernel radius of 5 " . the image clearly shows the straylight effect , probably due to the bright source x persei that dominates the x - ray sky @xmath170 to the southwest of morla , with a flux of @xmath110@xmath18 erg @xmath19 s@xmath2 . x - ray straylight in epic is produced by photons which experience single reflection by the mirror hyperbolae and which reach the sensitive area of the camera and produce near - annular structures in the image . in order to study the background sources in the _ xmm - newton _ observation , we evaluated the spectrum of our straylight annuli . this is consistent , at low energies , with the x persei spectrum found by @xcite , thus we added and froze these components in the spectral model of sources affected by this phenomenon . source detection using maximum likelihood fitting was done simultaneously on each of the epic - pn , mos1 , and mos2 using the sas tool edetect_chain . this tool runs on the event lists and invokes several other sas tools to produce background , sensitivity , and vignetting - corrected exposure maps . a likelihood threshold of 10 was used for source detection , corresponding to a significance level of 3.6@xmath20 . figure [ fig1 ] shows the brightest sources in the pn field of view ; they were analyzed in order to produce x - ray spectra and find optical counterparts . the first 7 objects were also observed by chandra so that we used both the telescope datasets to obtain radial profiles and spectra . + first , we analyzed their radial brightness profile and compared them with the _ xmm - newton _ and _ chandra _ theoretical point spread functions ( psfs ) . we found sources 8 and 14 to be the superimposition of two different point - like sources . using the 2011 chandra data , source 7 is incompatible with a simple point - like source . this is in agreement with results from xmm - newton data - though heavily affected by straylight photons - of a point - like source surrounded by extended emission . the nasa / ipac extragalactic database ( ned ) reports one radio source compatible with the point - like emission , identifying it as an active galactic nucleus ( agn ) in a cluster . + each spectrum was then fitted either with an absorbed powerlaw , well - suited for agns as well as pulsars and an absorbed apec , well - suited for stellar coronae . only source 9 requires a double apec , quite common for stellar coronae ; for this source we thus can not exclude the superimposition of two different stars closer than @xmath12 " . three sources ( 11 , 13 , and 14 ) can not be described by the simple models considered here ; deeper studies on the serendipitous sources are beyond the aim of this paper so they were excluded from our analysis . + optical counterparts were searched in the deep ( 4hr 26min exposure ) optical and near - infrared images collected with the 4 m mayall telescope on 2009 november 10 , using the large - field ( 36@xmath736 ) mosaic ccd imager @xcite . all the results of our x - ray and optical analyses on background sources are reported in table [ tab-1 ] . + we found 2 sources ( 8 and 9 ) that are compatible with a star interpretation both in the x - ray and optical bands . these two objects are likely located closer to us than morla , given their lower value of the absorption column . deeper optical observations with different filters would be required to find their distances from the optical absorption . all the remaining objects are extragalactic , while source 17 presents a very high value of the x - ray to optical flux ratio . this can be described as an obscured agn @xcite . it is possible that the emission actually comes from two different sources ( thus lowering the x - ray to optical flux ratio ) , though the low statistics prevent us from studying the brightness profile of the source . + all the agns were fitted together , linking their n@xmath21 to assess the average galactic absorption column . the resulting galactic value n@xmath22 = 2.10@xmath160.09 @xmath7 10@xmath23 @xmath19 is in good agreement with the theoretical one predicted by @xcite and @xcite for objects at the limit of the galaxy ( @xmath1500 pc ) . crrrrr object & coordinates@xmath24 & spec@xmath25 & n@xmath21 ( 10@xmath23 @xmath19 ) & f@xmath25/f@xmath26 & type + 1 & 03:57:47.50,+32:03:45.10 & pow & 1.83@xmath160.20 & 5.0 & agn + 2 & 03:57:50.95,+32:01:47.70 & pow / apec & 2.11@xmath27/1.36@xmath28 & 6.5 & agn + 3 & 03:57:53.90,+32:00:58.50 & pow / apec & 1.02@xmath29/0.66@xmath30 & 9.1 & agn + 4 & 03:58:21.95,+31:59:11.65 & pow & 1.87@xmath160.22 & 1.8 & agn + 5 & 03:58:15.25,+32:00:17.35 & pow & 1.91@xmath31 & 1.0 & agn + 6 & 03:58:00.29,+31:58:01.90 & pow / apec & 2.97@xmath32/2.00@xmath33 & 1.5 & agn + 7 & 03:57:39.35,+32:01:03.00 & pow / apec & 2.68@xmath160.60/1.20@xmath34 & 9.2 & agn + 8a & 03:57:12.7,5 + 32:00:50.10 & apec & 1.25@xmath35 & 0.0056 & star + 8b & 03:57:12.95,+32:00:42.65 & apec & 0.49@xmath36 & 0.0028 & star + 9 & 03:57:07.35,+32:02:59.65 & 2apec & 0.21@xmath37 & 0.0060 & star + 10 & 03:57:46.30,+31:57:24.50 & pow & 1.98@xmath38 & 3.6 & agn + 12 & 03:57:20.20,+32:03:14.95 & pow / apec & 1.69@xmath39/0.98@xmath40 & 0.12 & agn + 15 & 03:58:40.80,+32:02:27.35 & pow & 1.87@xmath41 & 2.8 & agn + 16 & 03:58:20.35,+32:08:24.85 & pow & 1.83@xmath42 & 1.0 & agn + 17 & 03:58:17.45,+32:08:50.20 & pow / apec & 1.73@xmath160.72/0.94@xmath43 & 400 & agn + ) . in the lower - right part of the figure the straylight effect is apparent.[fig1 ] ] in order to optimize the pulsar extraction radius we evaluated the brightness profile of the source to maximize both the signal - to - noise ratio ( s / n ) and the number of counts . the psf of the epic - pn camera onboard _ xmm - newton _ is best described by an off - axis , energy - dependent king function @xcite . the full width half maximum of the psf for an on - axis source at 1.5 kev is typically less than 12.5 `` for the pn camera and 4.4 '' for the two mos detectorss . our results are in agreement with those published by @xcite , based on _ chandra _ data , with no excesses within @xmath44 of the pulsar , with the exception of its bright tail . we found the best extraction radius to be 18 `` , containing 2022 , 515 and 540 counts in the 0.3 - 10 kev energy range in the pn and the two mos detectors respectively , with a background contribution of @xmath110% . we generated an ad hoc response matrix and effective area files using the sas tools rmfgen and arfgen . we also analyzed the _ chandra _ spectra of the source using a 1.5 '' extraction radius ( 698 pulsar counts with a background contribution of 0.4% ) ; we added the three different spectra with the tool mathpha . response matrix and effective area files were generated using the ciao script psextract . due to the presence of the pulsar tail , backgrounds were evaluated from a circle on the same ccd as the pulsar and without serendipitous x - ray sources , with a radius of 10 `` for _ chandra _ and 50 '' for the xmm telescopes . + for the pulsar emission , we tried a simple powerlaw ( pow ) , the combination of a powerlaw and a blackbody ( bb ) as well as the combination of a powerlaw and a magnetized neutron star atmosphere model ( nsa in xspec - assuming a neutron star with a radius of 13 km , mass of 1.4 m@xmath45 and a surface magnetic field of 10@xmath46 g ) . while the simple powerlaw gives a fit only marginally acceptable ( reduced chi square @xmath47=1.22 , 134 degrees of freedom ( dof ) ) , both the composite models yield better fits ( nsa+pow : @xmath47=0.96 , 131 dof ; bb+pow @xmath47=0.97 , 132 dof ) . an f - test performed comparing a simple powerlaw with the composite spectra results in a chance probability of less than 10@xmath48 , pointing to a significant improvement from adding a thermal component . the pulsar emission is well described by a powerlaw component with a photon index of @xmath49 = 2.28@xmath50 ( 90% confidence errors ) and a blackbody component with a temperature of kt = 94.0@xmath51 ev and an emitting radius at r = 0.45@xmath52 km ( where @xmath53 is the distance of the pulsar in units of @xmath54 pc ) , absorbed by a column density of n@xmath21 = 1.20@xmath55 @xmath7 10@xmath23 @xmath19 . the 0.3 - 10 kev non - thermal pulsar luminosity is l@xmath56 = 1.63@xmath57 @xmath7 10@xmath58 erg s@xmath2 , while the thermal one l@xmath56 = 1.14@xmath59 @xmath7 10@xmath58 erg s@xmath2 . as expected , the nsa model yields a lower temperature of 41.1@xmath60 ev and a larger emitting radius of r = 3.77@xmath61 km , and a powerlaw component with @xmath49 = 2.21@xmath160.19 , absorbed by a column density of n@xmath21 = 1.49@xmath62 @xmath7 10@xmath23 @xmath19 . both results are in agreement with the upper limits on the thermal components found by @xcite . figure [ fig2 ] shows the results of the fit for the two different spectral models . ] in order to carry out a proper timing analysis with epic data , a correct photon arrival time must be assigned to each event detected on board . the transformation from readout sequences by the epic camera to photon arrival times of each photon is performed by the epea ( european photon event analyser , @xcite ) . a problem in the epea can produce time jumps in some observations , which have to be corrected . in order to identify any possible remaining time jumps , one must neglect both the barycentering and the randomization of photon arrival times . the time @xmath63 t between successive events is then calculated and divided by the frame time ( ft ) of the relevant mode ( 47.66384 ms for pn large window mode ) . a time jump is shown to exist when @xmath63t / ft is different from an integer by a quantity larger than a tolerance parameter . only those time jumps which happen to be an integer multiple of the relevant ft would be missed by this method ( see e.g. @xcite ) . + our _ xmm - newton _ data show that a time jump happened in all the three instruments between 2011 - 09 - 15 16:04:45.822 and 16:10:14.310 utc , at about the half of the observation . such event makes it more difficult to find the pulsation of morla , creating a possible misalignment in the pulsar phase assigned to the photons before and after the jump . for this reason , we performed careful timing studies in the two subsets , as well as in the total data set . we evaluated the optimal selection of pulsar events by maximizing the signal - to - noise ratio in the 0.2 - 6 kev energy range as a function of the source extraction region . the optimal choice turned out to be a source extraction radius of 18 " for the pn camera . we selected only pattern 0 photons in the range 0.2 - 0.35 kev in order to reduce the background contribution due to low - energy electronic noise , while in the 0.35 - 6 kev energy range , we used pattern 0 - 4 photons . photon arrival times were converted to the solar system barycenter using the sas tool barycen . according to the accurate _ fermi_-lat ephemeris @xcite , a period of p = 0.44410548661(9 ) s is expected at the epoch of the xmm observation ( the uncertainty on the last digit is computed by propagating the @xmath64 error of the _ fermi_-lat timing solution ) . the uncertainty of @xmath65 s is much smaller than the intrinsic resolution of the _ xmm - newton _ dataset ( @xmath66 s ) . thus , if the spin - down of the pulsar is stable , a single trial folding is appropriate . folding using the full lat ephemeris yields marginal evidence for a modulation of the x - ray emission ( see figure [ figp ] ) . a pearson s @xmath67 test on the folded light curve ( 10 phase bins ) in the 0.2 - 6 kev energy range results in a probability of @xmath68 of sampling a uniform distribution . due to the low signal modulation , a study of the phase - misalignment before and after the time jump was not possible . to evaluate the significance of the detection independently of the phase binning chosen , we computed the corresponding @xmath69 value using the rayleigh test @xcite for 1 and 2 harmonics . this exercise yields probabilities as small as @xmath68 and @xmath70 , respectively , for a chance fluctuation . such a test confirms the hint of a modulation at the pulsar frequency ( and its first harmonic ) in the _ xmm - newton _ data , with a misalignment between the x - ray and gamma - ray peaks ( see figure [ figp ] ) . however , the low number of counts prevented us from disentangling the thermal from the non - thermal pulsation or from performing a phase - resolved spectroscopy analysis . contribution is taken into account . we note that the _ xmm - newton _ time slip ( discussed in section [ plc ] ) could affect our results . pattern 0 events have been selected in the 0.2 - 0.35 kev energy range while pattern @xmath71 4 have been used in the 0.35 - 6 kev range . for each energy band , the relative contributions of all the spectral components are showen . on average the background accounts for @xmath115% of the counts . 80% of the 0.8 - 6 kev counts come from the non - thermal pulsar component while 50% of the 0.2 - 0.8 kev counts come from the thermal one . the lower panel shows the lat light curve of the morla pulsar from @xcite to which the _ xmm - newton _ light curves have been aligned in phase.[figp ] ] the 2009 _ chandra _ observations revealed an extended , bright nebula protruding from the pulsar , extending @xmath19 in length . unfortunately , the 2011 _ chandra _ observation does not cover the entire nebular emission so it can not be used . the extended feature can be easily studied in our _ xmm - newton _ observation due to its lower particle background and higher spectral resolution , compared to the _ chandra _ one . on the other hand , the worse spatial resolution of _ xmm - newton _ prevented us from significantly constraining the properties of the nebula near the pulsar . we therefore concentrate on a higher - scale analysis of the extended feature . first , we produced brightness profiles of the nebula along the pulsar motion ( apm ) and orthogonal to it ( opm ) . in order to maximize the statistical significance of each bin , we carefully evaluated the bin size and the width of extraction boxes , choosing a scale of 90``@xmath750 '' apm and 450``@xmath712.5 '' opm ( see figure [ fig3 ] ) . for each instrument , we excluded the emission region surrounding the pulsar , as well as all the serendipitous sources and ccd gaps , according to our source psf and calibration studies . area calculations for each bin take into account the excluded regions . the presence of straylight annuli , pn ccd gaps , and numerous serendipitous sources in the faintest part of the nebula , toward the south - west , forced us to choose ad hoc boxes in that region , in order to exclude all the contaminants . + the brightness profile for both _ xmm - newton _ and _ chandra _ are shown in figure [ fig4 ] . in order to take into account the different angular resolution of the _ xmm - newton _ instruments , the profiles were convolved with a lorentzian kernel of @xmath49 = 7.125 `` , corresponding to a full - width half maximum ( fwhm ) of the distribution of 6.6 '' , the nominal fwhm of the _ xmm - newton _ mirrors . figure [ fig4 ] shows the smoothed _ chandra _ and _ xmm - newton _ brightness profiles . while minor differences are found in the count rate profiles shown in the figure , taking into account both the different ( energy - dependent ) effective areas , the flux profiles are compatible . the _ xmm - newton _ data show a hint of extended emission , at brightness below the _ chandra _ detection limit , near the pulsar itself . such a difference is possibly due to the higher _ xmm - newton _ sensitivity , that highlights the fainter part of the nebula , right behind the pulsar . no variations in the tail profile are detected in the available observations . + along the pulsar motion , the tail emerges from the background within 50 `` of the pulsar and its flux increases gradually , reaching a flat maximum between 4 and 7. its decrease is faster , though the profile is affected by the two serendipitous agn ( see section [ i m a ] ) and it fades below the background level at about 9 from morla . in the direction orthogonal to the main axis ( opm ) , the tail profile is clearly asymmetric : it shows a fast increase in the northeast direction ( rising to the maximum within 30 '' ) , reaches a sharp maximum , in broad agreement with the neutron star proper motion direction , then the decrease is slower , fading at about 1 - 1.5 in the nearest part of the nebula and 2 in the furthest . such a behavior recalls the projection of a cone , with the apex corresponding to the pulsar position ( or somewhere in the vicinity of the pulsar ) . + brightness profiles of the tail in the soft ( 0.3 - 1.5 kev ) and hard ( 1.5 - 6 kev ) energy bands show no significant discrepancies , with the possible exception of the region immediately next to the pulsar ( @xmath7250 " ) , where a hint of a softening in the spectrum is seen . or more . following the red polygon , we traced in magenta a plausible shape of the tail while the cyan ellipse is the extraction region used for our spectral analyses on the tail . the long green arrow indicates the direction of the pulsar proper motion , while the two shorter arrows represent the 1 sigma confidence errors on this direction . [ fig3 ] ] = 7.125 `` , corresponding to a full width at half maximum ( fwhm ) of the distribution of 6.6 '' , the nominal fwhm of the _ xmm - newton _ mirrors . the integral of the _ chandra _ profile has also been normalized to the integral of the 0.3 - 6 kev _ xmm - newton _ one in order to show any possible discrepancy . the blue vertical lines mark the position of the pulsar . sources 4 and 5 , indicated with an arrow , have been removed , possibly enhancing the decrease in brightness . the flat maximum along the main axis is clearly visible between @xmath1250 `` and 500 '' as well as the asymmetrical shape of the tail in the direction orthogonal to it . the profiles in the soft and hard x - ray bands are almost the same , with a possible exception right behind the pulsar , where the soft band dominates . [ fig4 ] ] while in the previous _ chandra _ observations the spectral analysis of the tail emission was hampered by the low signal - to - noise ratio ( s / n ) - @xmath157% of the total counts coming from the background , - the exceptional sensitivity of _ xmm - newton _ allows for a deep study of the nebular emission , with a background contribution of @xmath135% . + in order to maximize the s / n , we extracted the total spectrum from the brightest part of the nebula , an elliptical region centered at r.a . ( j2000 ) 03:58:06.59 , + 32:02:05.42 , a semi - major axis of 4 and a semi - minor axis of 50 `` . we excluded from the analysis region a circle with 20 '' radius around source 5 and the gaps in the ccds . we extracted the background from source - free circular regions uncontaminated by the extended emission and located on the same ccd as the tail . we generated the response and effective area files using the sas tools rmfgen and arfgen . _ chandra _ data were treated in the same way as in @xcite and the resulting spectrum was fitted together with the _ xmm - newton _ ones . the spectrum of the nebula is well described ( @xmath47 = 1.06 , 162 dof ) by a non - thermal emission model ( powerlaw with a photon index @xmath49 = 2.07@xmath160.08 ) , absorbed by a column n@xmath21 = ( 2.61@xmath160.23 ) @xmath7 10@xmath23 @xmath19 . figure [ fig5 ] shows confidence contours for n@xmath21 and @xmath49 for the diffuse feature and for the pulsar . the value of n@xmath21 obtained for the tail seems to be in conflict ( at the @xmath73 level ) with both the one obtained for the pulsar , as well as with the galactic value , obtained by fitting the spectra of the serendipitous agns in the field ( see section [ i m a ] ) . such a discrepancy between the values of n@xmath21 led us to consider alternative models . the spectrum of the tail is well described ( @xmath47 = 1.06 , 162 dof ) by a thermal bremsstrahlung model with a temperature kt = 3.75@xmath74 kev , absorbed by the more realistic column n@xmath21 = ( 1.72@xmath160.16 ) @xmath7 10@xmath23 @xmath19 . other similar thermal models such as mekal ( hot diffuse gas @xcite ) , nei ( collisional plasma , non - equilibrium , constant temperature @xcite ) , and pshock ( plane - parallel shocked plasma , constant temperature @xcite ) with redshift 0 fit equally well the spectrum of the tail and require a very low metal abundance . table [ tab-2 ] reports the results of the fit for all the models . in order to study the spatial - spectral variability of the nebula , we divided the elliptical extraction region using the two axes , then we fitted both the absorbed powerlaw and bremsstrahlung models ( see table [ tab-2 ] ) for each semi - elliptical region . in this way we obtained spectra from the semi - elliptical regions nearest to and furthest from the pulsar , and those at the south - west and north - east . no significant ( @xmath64 ) variation was detected down to @xmath10.2 in @xmath49 or @xmath11 kev in kt . from the study of serendipitous sources ( see section 2.1 ) is also shown with errors at 1 , 2 , and 3 @xmath20 indicated by different shades of blue . it is apparent that the galactic value of n@xmath21 and the value obtained by fitting the powerlaw spectrum on the tail are only marginally compatible.[fig5 ] ] crrrrrr model & @xmath47 & n@xmath21 & @xmath49 & kt & abund@xmath24 & flux@xmath75 + & & 10@xmath23 @xmath19 & & kev & & 10@xmath76 erg @xmath19 s@xmath2 + tot - pow & 1.06 & 2.61@xmath160.23 & 2.07@xmath160.08 & - & - & 6.65@xmath160.40 + tot - brem & 1.06 & 1.72@xmath160.16 & - & 3.75@xmath74 & - & 4.95@xmath160.30 + tot - mek & 1.13 & 1.73@xmath160.16 & - & 3.76@xmath77 & @xmath720.099 & 4.57@xmath160.28 + near - brem & 1.15 & 1.76@xmath160.14 & - & 4.72@xmath78 & - & 2.23@xmath79 + far - brem & & & - & 3.39@xmath80 & - & 2.05@xmath160.16 + sw - brem & 1.33 & 1.72@xmath160.14 & - & 4.01@xmath81 & - & 1.83@xmath160.16 + ne - brem & & & - & 4.23@xmath82 & - & 2.48@xmath160.18 + the x - ray spectrum of the morla pulsar is a combination of thermal radiation from hot spot(s ) superimposed on a non - thermal power - law component . a straight estimate of the ns polar cap size , based on a simple centered dipole magnetic field geometry ( see e.g. @xcite ) and a standard ns radius of 10 km , predicts in this case a polar cap radius of about 85 m. the observed emitting radius is about 450 m , significantly smaller than the entire surface of any reasonable ns but definitely bigger than the expected polar cap . this is true both using the blackbody model and the ns atmosphere model , although the latter yields a larger emitting area and a lower temperature . such a large size of the spots has also been found in other well - known pulsars , like psr b0656 + 14 and psr b1055 - 52 @xcite , and it is still poorly understood . the luminosity associated with the morla polar cap is 1.1 @xmath7 10@xmath58 erg s@xmath2 , a factor of a few lower than the luminosity expected from polar caps heated by e@xmath83 currents . however , it is consistent with the expectations from polar cap heating due to the bombardment by particles created only by inverse compton scattered photons @xcite . psr j0357 + 3205 is close to the death line for production of e@xmath83 pairs by curvature radiation photons and this could explain the reduced polar cap heating . + while the majority of x - ray emitting isolated neutron stars ( e.g. @xcite ) , exhibit thermal emission coming from the whole ns surface , no such a component is required to fit the _ xmm - newton _ spectrum . using the blackbody model , for a ns radius of 10 km and a distance of 500 pc , the 3@xmath20 upper limit on the temperature is 4.4 @xmath7 10@xmath8 k. this constraint makes morla the coldest ns in its age range , slightly cooler than the coeval radio - quiet pulsar rx j1856 - 3754 ( e.g. @xcite ) . such a difference can be attributed to different magnetic fields , and thus different magneto - thermal evolution ( e.g. @xcite ) . the non - thermal x - ray luminosity of morla is in broad agreement with the relation between the x - ray luminosity of rotation - powered pulsars and their spin - down luminosity @xcite . the morphology of the tail , its spatial extension - @xmath11.3 pc at 500 pc assuming no inclination with respect to the plane of the sky ( @xmath84 ) , - and the lack of any other related galactic source led de luca et al . 2011 to associate unambiguously the nebular emission with the pulsar . elongated x - ray tails coupled with pulsars are quite common ( e.g. @xcite ) and are usually interpreted within the framework of bow - shock , ram - pressure - dominated pulsar wind nebulae ( e.g. @xcite ) . when a pulsar moves supersonically , the shocked pulsar wind is expected to flow in a region downstream of the termination shock ( the cavity in the interstellar medium created by the moving neutron star and its wind ) , confined by ram pressure . x - rays are produced by synchrotron emission from the wind particles accelerated at the termination shock , which is typically seen ( if it can be resolved ) as the brightest part of the extended structure ( see , e.g. , @xcite ) , according to the expectations from magnetohydrodynamic simulations @xcite . @xcite have detected and measured a very large proper motion for the pulsar : 165@xmath1630 mas yr@xmath2 - 390 km s@xmath2 across the plane of the sky , at 500 pc - in the direction opposite to the tail , thus compatible with such a scenario . the tail shows no other similarities with other , synchrotron - emitting nebulae ( for a complete discussion about morla s synchrotron - emitting model see @xcite ) . the first problem arises from energetic requirements for the emitting particles . according to @xcite , the maximum potential drop between the pole and the light cylinder ( in an aligned pulsar ) is @xmath85 = ( 3 @xmath86/2c)@xmath87 . in the morla case , the electrons are accelerated up to a maximum lorentz factor @xmath8810@xmath89 , which can be considered as an upper limit for the electrons injected in the pulsar wind nebula . the low @xmath90 of morla requires an ambient magnetic field as high as @xmath150 @xmath91 g - while the mean galactic value is @xmath11 @xmath91 g ( @xcite and references therein ) . however , one must also take into account the magnetic field carried by the particles themselves . an independent estimate of the magnetic field in the tail can be obtained from the measured synchrotron surface brightness @xcite . taking into account the uncertainty on the parameters we can conclude that a reasonable value of the magnetic field in the tail is in the range @xmath92 @xmath91 g , consistent with the value of @xmath150 @xmath91 g obtained before . the second problem is the lack of diffuse emission surrounding the pulsar , where the emission from the wind termination shock should be brightest , as observed in all the other known cases ( e.g. @xcite ) . such a shock could be unresolved by _ chandra _ only in the case of an extremely large ambient density ( several hundred atoms per @xmath93 ) and/or a pulsar speed higher than some thousands km s@xmath2 . taking into account the measure of the proper motion , the distance of the pulsar should be larger than 1.3 kpc or the inclination of the motion should be higher than 70@xmath4 . such conditions imply a tail at least 3.5 pc long , so that it becomes even more difficult to account for the energy of the particles . this picture implies also that a significant fraction of the point - like flux comes from a non - thermal termination shock , reducing dramatically the power - law component of the pulsar . while it is acceptable for the non - thermal component of the pulsar to be a factor of some smaller than the value we obtained @xcite , the lack of nebular emission around the pulsar , where the synchrotron emission should be stronger , remains unexplained . the brightness profile along the main axis of the tail is consistent with no emission within 30 - 50 " of the pulsar , then it slowly increases to reach a broad peak at @xmath14. this behavior is remarkably different from that observed in all the known synchrotron nebulae that show peaks close to the position of their parent pulsar ( e.g. @xcite ) , where the acceleration of the wind particles is the most important . it is very hard to explain in this scenario why the emission peaks at @xmath11 pc from the pulsar . + the asymmetric brightness profile of the tail in the direction perpendicular to its main axis is characterized by a sharp northeastern edge and a slow decay toward the south - west direction . the synchrotron emission from the nebula is only marginally dependent on the ism density , so that ad - hoc and large variations in the ism are demanded , in order to explain this strange profile . even if possible , such a case is hardly believable . + furthermore , the inferred synchrotron cooling time implies a significant spatial - spectral evolution with respect to the distance from the pulsar , as observed in all synchrotron nebulae @xcite . thanks to the unique throughput of _ xmm - newton _ , we could perform a deep spatial - spectral study of the tail , analyzing separately the furthest and the nearest parts of the nebula . no significant differences in the spectral index are detected ( see section [ nspe ] ) . the lack of variation can not be easily explained in terms of a synchrotron nebula . the lack of such a softening would require some particle re - acceleration mechanism within the tail itself . + the excellent throughput of the _ xmm - newton _ instruments has revealed another problem of the synchrotron nebula scenario , connected to the value of the column density ( n@xmath21 ) obtained by fitting the non - thermal model of the nebula . while it is marginally compatible with the value obtained for the pulsar , the inferred n@xmath21 appears to be larger than the galactic one in that direction , measured by fitting serendipitous agns in the field ( see figure [ fig5 ] ) . as reported in section [ nspe ] , a thermal bremsstrahlung model fits equally well the spectrum of the tail and produces an estimate of n@xmath21 in agreement with that of the pulsar and the galactic one . in the bremsstrahlung scenario , the tail emission arises from the shocked ism material heated up to x - ray temperatures . under the assumption of a strong shock - wave and from the junction conditions at the shock front , we can estimate the speed of the pulsar directly from the temperature of the shocked ism material at the head of the bow shock : @xmath94 , where @xmath91 is the molecular weight of the ism gas and we assumed an adiabatic index @xmath95 . the temperature obtained by fitting the tail spectrum with a thermal bremsstrahlung model , @xmath96 kev , implies a pulsar speed @xmath97 km s@xmath2 . from the values of the abundances resulting from the spectral fit , we can assume that @xmath98 , so the speed of the pulsar is about @xmath99 km s@xmath2 , which would make it the largest ever observed in a pulsar . if we combine this result with the pulsar proper motion we can estimate the inclination angle @xmath84 of the pulsar motion ( and thus of its tail ) , @xmath100 acos(@xmath101 ) . moreover , through the value of the photon flux measured using a thermal bremsstrahlung model , it is possible to estimate the number density of the shocked ism gas as @xmath102 atoms @xmath103 . thermal bremsstrahlung emission acts as a cooling mechanism . the time scale for cooling can be estimated as @xmath104 , where @xmath105 is the thermal energy per unit volume and @xmath106 is the volume emissivity . this approximation gives @xmath107 yr , where @xmath108 is the frequency - averaged gaunt factor that typically ranges between @xmath109 and @xmath110 , depending on the plasma temperature @xcite . from our estimates of the temperature and number density of the shocked ism gas we expect @xmath111 yr . considering a pulsar proper motion of 165@xmath1630 mas yr@xmath2 and an extension of the nebula of about @xmath112 , then the age of the tail ( @xmath13300 yr ) is much lower than its cooling time . this result can explain why we have not measured a significant evolution of the temperature of the tail as a function of the distance from the pulsar . on the other hand , the shorter than expected length of the tail can be explained assuming an ad - hoc decrease of a @xmath13 factor in the number density @xmath113 beyond the end of the tail - the volume emissivity of the thermal bremsstrahlung emission is @xmath114 . such a strong dependence on the ism density can explain the asymmetrical shape of the nebula in the direction orthogonal to its main axis : this could be a consequence of an inhomogeneity of the number density in the nebula , quite typical in the ism . from the conservation of mass , energy , and momentum on the two sides of the shock front and from the values measured in the tail ( e.g. the pre - shock temperature @xmath115 and number density @xmath116 ) , we can calculate the same physical quantities outside the nebula : @xmath117 @xmath118\left[(\gamma+1)+(\gamma-1)(m^2 - 1)\right]}{(\gamma+1)^2m^2}\ ] ] where @xmath119 is the mach number and depends on the cone angle of the tail ( @xmath14 ) and an adiabatic index @xmath95 is assumed for the ism gas . in the strong shock - wave limit ( @xmath120 ) @xmath121 atoms @xmath103 while @xmath115 depends on the mach number and thus on the inclination angle @xmath84 . from the geometry of the cone angle of the nebula , a plausible value of @xmath115 is in the range of @xmath122-@xmath123 k , for @xmath124 . the pre - shock temperature is consistent with that of the hot phase of the ism , which fills a large fraction of the galaxy ( e.g. @xcite ) . such a scenario is also in full agreement with the requirement of a full ionized ism by @xcite . we note that the density required by our model is slightly higher than expected for the hot ism : this may result from the concerted action of massive stellar winds and supernova explosions . @xcite report the presence of dense ( 10@xmath125 - 1 atoms @xmath103 ) , hot ( @xmath110@xmath126 k ) envelopes of gas in our galaxy . on the other hand , thermal models such as mekal , nei , and pshock point to a very low metallicity of the ism around the pulsar , making more difficult to explain this dense envelope as the result of a supernova explosion or stellar winds . future multi - wavelength observations of the field and more detailed galactic gas models could better constrain and explain the ism composition around morla . a shock - wave scenario can easily explain the lack of diffuse emission surrounding the pulsar . most of the energy in the pre - shock flow is carried by the ions - the kinetic energy in the streaming of the electrons is less than @xmath127 of the total - but the electrons are generally responsible for the cooling of the plasma : this means that the electron temperature ( @xmath128 ) is responsible for the dynamics of the shock and its emitted spectrum . the electrons must be heated by the ions before the emission becomes detectable and this process takes some time . coulomb heating of the electrons behind the shock proceeds at a rate @xmath129 where @xmath130 is the ion temperature and @xmath131 s @xmath170 kyr is the equipartition time for a fully ionized plasma with a coulomb logarithm of 30 ( where @xmath128 is in units of k and @xmath113 in @xmath103 ) . if all the post - shock energy is in the ions ( @xmath132 ) , then the increase in t@xmath133 due to collisional heating follows : @xmath134 k , where @xmath135 is the age of the tail in units of @xmath136 yr , @xmath116 is the pre - shock number density and @xmath137 is the pulsar speed in units of @xmath138 cm s@xmath2 . thus @xmath139 k , not significantly different from the value observed in the tail . moreover , @xmath140 k within @xmath141 " of the pulsar ( corresponding to @xmath1200 - 300 years , from the measure of the proper motion ) . this fully explains the lack of diffuse emission surrounding the pulsar . so far , no evidence of other pulsars characterized by a thermal bremsstrahlung emitting tail has been reported . this could be explained by the peculiar conditions it requires , e.g. very fast pulsars and an hot and dense ism . with its estimated velocity of 1900 km s@xmath2 , j0357 + 3205 is the fastest moving pulsar known . among all the pulsars listed in the atnf pulsar catalogue @xcite , only 2 have a very high velocity ( @xmath142 km / s ) , reliably measured through an estimate of the distance obtained by other methods than the dispersion measure . psr b2224 + 65 , powering the guitar nebula @xcite and showing an x - ray tail similar to morla @xcite , has a speed of at least @xmath143 km s@xmath2 @xcite . j0821 - 4300 , the central compact object in the puppis a supernova remnant has a speed possibly higher than 1000 km s@xmath2 @xcite . the spin - down power of psr b2224 + 65 is similar to that of morla ( @xmath9=@xmath144 erg s@xmath2 and @xmath145 erg s@xmath2 ) , but the tails are different . while both tails have a similar brightness profile and show no spectral variations , the tail of psr b2224 + 65 is misaligned ( 118@xmath4 ) with respect to the proper motion of the pulsar . alternative explanations have been proposed : according to @xcite the feature could be produced by highly energetic particles which escaped the pulsar wind termination shock and are confined by an interstellar magnetic field . this scenario accounts for a faint counter - feature , detected on the opposite side of the pulsar with respect to the tail , but does not explain the absence of spectral variations along the tail @xcite . another possible scenario is that the observed emission comes from a ballistic jet , similar to those of agns @xcite , but also this picture has some problems . + the key to understand why these two pulsars do not emit through thermal bremsstrahlung lies in the ism characteristics . this emission is detectable in x - rays only if both the ism density and temperature are high enough ( @xmath10.2 atoms @xmath103 and @xmath110@xmath8 - 10@xmath126 k ) . it is hardly believable that the ism around a cco like j0821 - 4300 , lying inside its supernova remnant , could be denser than @xmath110@xmath125 atoms @xmath103 : this can explain the lack of a bremsstrahlung tail . in the case of psr b2224 + 3205 , the ism density , estimated from the h@xmath14 bow - shock emission to be @xmath146 @xmath103 @xcite , is lower than the ism density predicted around morla . no constraints on the ism temperature around psr b2224 + 3205 appear in the literature : it is not excluded that a low temperature of the ism may contribute to the lack of a bremsstrahlung tail . recently , @xcite studied the case of the x - ray source igr j11014 - 6103 , suggesting an association with the supernova remnant msh 11 - 61a @xcite , thus inferring a velocity of 2400 - 2900 km s@xmath2 . igr j11014 - 6103 shows a complex morphology , consisting of a point source , an elongated tail @xmath147 " long , aligned to the direction of the proper motion , and a feature of @xmath14 , nearly perpendicular to the proper motion . spectral and spatial analyses suggest that igr j11014 - 6103 could be a pulsar , therefore providing a natural explanation for the co - aligned feature as a bow - shock ram - pressure - confined pulsar wind nebula . moreover , the feature resembles the tail of morla : its luminosity peaks far from the pulsar and only a few photons are detected near the pulsar . the low statistics prevent us from studying further the similarities between the two tails . however , if the association with msh 11 - 61a is confirmed , its distance would be 10 kpc , implying a tail of @xmath148 pc , comparable to that of morla . xmm - newton _ observation of the field of j0357 + 3205 allowed us to better constrain the distance of the pulsar using the interstellar absorption value . the galactic absorption column in the direction of the pulsar , predicted by @xcite at @xmath1500 pc based on the hi distribution , is fully compatible with our best fit value for the pulsar and the tail , pointing to a lower limit for the pulsar distance of @xmath1300 pc . this result is in broad agreement with the value estimated by scaling the @xmath0-ray flux of the pulsar @xcite : the method hinges upon the observed correlation between the intrinsic @xmath0-ray luminosity and spin - down power of the pulsar @xcite , assuming a beam correction factor of 1 for the @xmath0-ray emission cone of all the pulsars @xcite . by applying this relation , we have a @xmath0-ray efficiency of 0.33 and a distance of @xmath1500 pc . we obtain a new upper limit of @xmath1900 pc by requiring the @xmath0-ray efficiency to be less than 1 . + another important piece of information could come from the serendipitous stars we found in the field : their distance estimates from optical analyses could provide lower limits on the distance of the pulsar . future multi - filter optical observations will provide this limit . + under the hypothesis of a bremsstrahlung - emitting tail , we infer a pulsar velocity of @xmath11900 km s@xmath2 . the estimated value of the pulsar proper motion , 165@xmath1660 mas yr@xmath2 , allows us to set another upper limit for the pulsar distance , being 2.4@xmath149kpc for a purely transversal motion . however , as reported in section [ tail ] , a null inclination leads to unrealistic values ( @xmath110@xmath150 k ) of the ism temperature , so that an inclination angle larger than 60@xmath4 is required . this lowers the upper limit to 1.2@xmath151 kpc , still less constraining than that obtained by computing the pulsar efficiency . for a pulsar with a kick velocity of @xmath11900 km s@xmath2 , a proper motion of 165 mas yr@xmath2 and a distance of 300 pc @xmath152 900 pc , the inclination of the system with respect to the plane of the sky ranges between 68@xmath4 and 83@xmath4 , the density of the ism between 0.15 and 0.3 atoms @xmath103 , and its temperature between 1 and 9 @xmath7 10@xmath8 k. the length of the tail is @xmath16.5 pc , independently of the distance . although the study of the kinematics of the pulsar is beyond the scope of this paper , the small component of the proper motion perpendicular to the galactic plane points to a birth site outside the galactic plane , suggesting that its progenitor might have been a runaway star , as already introduced in @xcite . we have characterized the properties of the middle - aged @xmath0-ray and x - ray pulsar j0357 + 3205 ( morla ) and its tail . the x - ray spectrum of the pulsar is consistent with a magnetospheric non - thermal component plus thermal emission from hot spots ; the lack of any detected thermal component from the whole surface makes morla the coldest ns in its age range . + folding the x - ray photons at the expected pulsar period using the lat ephemeris yields marginal evidence for modulation , with 2 peaks , neither of them aligned with the single gamma - ray peak ( see figure [ figp ] ) . we confirmed the presence of a diffuse emission @xmath19 long , the largest tail of x - ray emission associated with any rotation - powered pulsar . such an extended emission can not be explained in terms of the usual bow - shock ram - pressure - dominated pulsar wind nebula . in fact , the following problems arise : + - the existence of the tail is problematic for a pulsar with such a low @xmath153 as morla ; + - no bow shock has been resolved around the pulsar ; + - the lack of any nebular emission around the pulsar , where the particle acceleration is maximum , can not be explained ; + - the nebular spectrum lacks any spatial evolution , a necessary signature of the radiative cooling of the electrons accelerated at the wind termination shock ; + - the very asymmetric brightness profile in the direction perpendicular to the main axis of the tail requires a large ad - hoc inhomogeneity of the ism around the pulsar ; + - fitting the _ xmm - newton _ nebular spectra with a powerlaw , we found the nebular n@xmath21 to be higher than the galactic value and only marginally in agreement with the n@xmath21 of the pulsar . we propose a thermal bremsstrahlung model as an alternative explanation of the tail emission . in this scenario , the emission comes from the shocked ism material heated up to x - ray temperatures . this model gives full account of the peculiar features of the tail : + - the lack of any detectable spatial evolution in the tail spectrum is due to the long bremsstrahlung cool - down time ; + - the peculiar asymmetries of the brightness profile can be interpreted in terms of small changes in the ism density that strongly affect the brehmsstralung emissivity ; + - most of the energy in the pre - shock flow is carried by the ions , while the electron temperature is responsible for the x - ray emission ; the coulomb heating time of the electrons behind the shock is fully in agreement with the lack of any detected nebular emission near the pulsar ; + - the value of n@xmath21 measured in the tail agrees both with the value obtained for the pulsar and the galactic value . + this scenario allows us to estimate some parameters of the pulsar and of the ism around it . for a bremsstrahlung - emitting tail we estimate the pulsar distance to be between 300 and 900 pc . a pulsar velocity of @xmath11900 km s@xmath2 is required , in agreement with the pulsar proper motion for distances of some hundreds parsecs and a high inclination . the mean density of the ism is required to be @xmath10.2 atoms @xmath103 and the temperature of some 10@xmath8 k. this type of hot gas usually presents lower densities , but a denser phase ( possibly detected by @xcite ) is predicted to be the result of the action of massive stellar winds and supernova explosions . however , the low metallicity we obtained from the spectral fit makes the explanation of this gas envelope more diffcult . + a high inclination of the system ( @xmath370@xmath4 ) is predicted in both models . this value provides further support to the bremsstrahlung model , where the cool - down time demands for a longer tail , and adds to the problems of a synchrotron model . for all these reasons , we believe morla s nebula to be the first example of a new `` turtle s tail '' class of thermally - emitting nebulae , ironically associated with fast moving pulsars . + until now we have no clear evidence of other pulsars characterized by a tail emitting via thermal bremsstrahlung , possibly for the requirements of a very fast pulsar and a hot , dense ism . moreover , energetic pulsars can also generate classic synchrotron nebulae , that may outshine a bremsstrahlung component . the recently discovered nebula associated with the source igr j11014 - 6103 , suspected of being a high - velocity pulsar , resembles the tail of morla : further studies of this feature could lead to a second example of a turtle s tail nebula . this work was supported by the contract swift asi - inaf 1 - 004 - 11 - 0 . we warmly thank the apj referee for the useful comments and suggestions as well as the iasf - inaf staff who constructively participated in the discussion during my `` astrosiesta '' presentation . abdo , a.a . 2009 , science , 325 , 848 abdo , a.a . 2009 , science , 325 , 840 abdo , a.a . 2009 , apj , 700 , 597 abdo , a.a . 2010 , apj , 187 , 460 anders , e. & grevesse , n. 1989 , gecoa , 53 , 197 atwood , w.b . et al . 2009 , apj , 697 , 1071 bandiera , r. 2008 , a&a , 490 , 3 becker et al . 2012 , apj , 755 , 141 bland - hawthorn , j. & reynolds , r. 2000 , arxiv : astro - ph/0006058 borkowski , k.j . et al . 2001 , apj , 548 , 820 buccheri , r. et al . 1983 , a&a , 128 , 245 bucciantini , n. , 2002 , a&a 387 , 1066 bucciantini , n. , et al . , 2005 , a&a 443 , 519 chatterjee , s. & cordes , j.m . 2002 , apj , 575 , 407 cheng , k.s . 1986 , apj , 300 , 500 cordes , j.m . 1993 , nature , 362 , 133 de luca , a. & molendi , s. 2004 , a&a , 419 , 837 de luca , a. et al . 2005 , apj , 623 , 1051 de luca , a. et al . 2011 , apj , 733 , 104 de luca , a. et al . , submitted to apj dickey , j.m . et al . 1990 , ara&a , 28 , 215 gaensler , b.m . , et al . , 2004 , apj 616 , 383 gaensler , b.m . & slane , p.o . 2006 , ara&a , 44 , 17 garmire , g.g . 2003 , spie 4851 , 28 goldreich , p. & julian , w.h . , 1969 , apj 157 , 869 harding , a.k . & muslimov , a.g . , 2001 , apj 556 , 987 harding , a.k . & muslimov , a.g . , 2002 , apj 568 , 862 harding , a.k . & muslimov , a.g . , 2004 , cosp , 35 , 562 hui , c. y. , & becker , w. 2006 , a&a , 457 , l33 hui , c.y . & becker , w. 2007 , a&a 467 , 1209 hui , c.y . 2012 , apj , 747 , 74 jacoby , g. h. et al . 1998 , proc . spie , 3355 , 721 jager , o.c . & djannati - atai , a. 2008 , arxiv:0803.0116 jansson , r. & farrar , g.r . , 2012 , apj , 757 , 14 johnson , s.p . & wang , q.d . 2010 , mnras , 408 , 1216 kalberla , p.m. et al . 2005 , a&a , 440 , 775 kargaltsev , o. et al . 2008 , apj , 684 , 542 kargaltsev o. & pavlov , g.g . , 2008 , in aip conf . 983 , 171 kargaltsev , o. & pavlov , g.g . , 2008 , apj 684 , 542 kaspi , v. , et al . , 2006 , in @xmath154compact stellar x - ray sources@xmath154 , eds . lewin , w. and van der klis , m. , cambridge university press , p. 279 kesteven , m.j.l . 1968 , aujph , 21 , 369 kuster , m. et al . 1999 , spie , 3765 , 673 la palombara , n. & mereghetti , s. 2007 , a&a , 474 , 137 the lat collaboration , in preparation `` the second fermi lat catalog of @xmath0-ray pulsars '' mainieri , v. et al . 2002 , a&a , 393 , 425 manchester , r.n . et al . , 2005 , aj , 129 , 1993 marelli , m. et al . 2011 , apj , 733 , 82 marelli , m. 2012 , arxiv:1205.1748 martin - carrillo , a. et al . 2012 , arxiv:1204.0978 mcgowan , k.e . , et al . , 2006 , apj , 647 , 1300 mewe , r. et al . 1985 , a&as , 62 , 197 padmanabhan , t. 2000 , teoretical astrophysics - vol 1 , cambridge university press pavlov , g.g . et al . , 2003 , apj , 591 , 1157 pons , j.a . 2009 , a&a , 496 , 207 possenti , a. et al . , 2002 , a&a , 387 , 993 read , a.m. 2004 , xmm - ccf - rel-167 sartore , n. et al . 2012 , a&a , 541 , 66 saz parkinson , p.m. et al . 2010 , apj , 725 , 571 struder , l. et al . , 2001 , a&a , 365 , l18 tomsick , j.a . 2012 , apj , 754 , 145 turner , m.j.l . 2001 , a&a , 365 , l27 van der swaluw , e. , 2003 , a&a 404 , 939 watters , k.p . et al . , 2009 , , 695 , 1289 winkler , p. f. , & petre , r. 2007 , apj , 670 , 635 yao , y. & wang , q.d . 2005 , apj , 624 , 751

the version 1.51 of the monte carlo ( mc ) program koralw for all processes is presented . the most important change from the previous version 1.42 is the facility for writing mc events on the mass storage device and reprocessing them later on . in the reprocessing parameters of the standard model may be modified in order to fit them to experimental data . another important new feature is the possibility of including complete corrections to double - resonant-pair component - processes in addition to all background ( non- ) graphs . the inclusion is done with the help of the yfsww3 mc event generator for fully exclusive differential distributions ( event per event ) . technically , it is done in such a way that yfsww3 runs concurrently with koralw as a separate slave process , reading momenta of the mc event generated by koralw and returning the correction weight to koralw . the latter introduces the correction using this weight , and finishes processing the event ( rejection due to total mc weight , hadronization , etc . ) . the communication between koralw and yfsww3 is done with the help of the fifo facility of the unix / linux operating system . this does not require any modifications of the fortran source codes . from the user s point of view , the resulting concurrent mc event generator koralwyfsww3 looks as a regular single mc event generator with all the standard features . * cern - th/2001 - 040 + uthep-01 - 0102 * the monte carlo program koralw version 1.51 and + the concurrent monte carlo koralwyfsww3 + with all background graphs and + first - order corrections to-pair production * s. jadach , * * w. paczek , * * m. skrzypek , * * b.f.l . ward * _ and _ * z. was * _cern , theory division , ch-1211 geneva 23 , switzerland , _ + _institute of nuclear physics , ul . kawiory 26a , 30 - 055 cracow , poland , _ + _department of physics and astronomy , + the university of tennessee , knoxville , tn 37996 - 1200 , usa _ + _ slac , stanford university , stanford , ca 94309 , usa _ + _ institute of computer science , jagellonian university , + ul . nawojki 11 , cracow , poland _ _ to be submitted to computer physics communications _ * work supported in part by the polish government grant kbn 5p03b09320 , the us doe contracts de - fg05 - 91er40627 and de - ac03 - 76sf00515 , the european commission 5-th framework contract hprn - ct-2000 - 00149 , nato grant st.clg.977761 and the polish - french collaboration within in2p3 through lapp annecy . * cern - th/2001 - 040 + april 2001 * * new version summary * _ title of the program : _ koralw , version 1.51 . 0.2 cm _ reference to original program : _ comput . . commun . * 94 * ( 1996 ) 215 ; * 119 * ( 1999 ) 272 0.2 cm _ computer : _ any computer with the fortran 77 compiler under unix or linux operating system 0.2 cm _ operating system : _ unix , linux version 6.x and 7.x 0.2 cm _ programming language used : _ fortran 77 0.2 cm _ high - speed storage required : _ 25 mb 0.2 cm _ keywords : _ + radiative corrections , initial - state radiation ( isr ) , electroweak ( ew ) corrections , leading - logarithmic ( ll ) approximation , heavy boson , four - fermion processes , monte carlo ( mc ) simulation / generation , quantum electrodynamics ( qed ) , quantum chromodynamics ( qcd ) , yennie frautschi suura ( yfs ) exponentiation , standard model ( sm ) , lep2 , next - generation linear colliders ( lc ) . 0.2 cm _ nature of the physical problem : _ + the precise study of-pair production and decay at lep2 requires both non - double - resonant and corrections . so far each of these corrections is available as a separate monte carlo program and there is no monte carlo that could simulate in a complete way both effects at the same time . such a mc event generator would be of importance for example for apparatus simulations or monte carlo based fits . the previous version of koralw included _ all _ non - double - resonant corrections to _ all _ double - resonant four - fermion processes in annihilation . the present version 1.51 allows , for the first time ever , the inclusion on an _ event - per - event basis _ of the respective corrections generated at the same time by the independently running monte carlo program yfsww3 , as well as a reweighting of any earlier generated events with modified parameter sets . 0.2 cm _ method of solution : _ + the monte carlo method used to simulate the all four - fermion final - state processes in the collisions in the presence of multiphoton initial - state radiation has not changed since version 1.42 . adding the corrections generated by yfsww3 is done at the level of the unix / linux operating system with the help of the fifo mechanism ( `` named pipes '' ) . 0.2 cm _ restrictions on the complexity of the problem : _ + for koralw as in version 1.42 ; for koralwyfsww3 as in and . 0.2 cm _ typical running time : _ + approximate times on a pc pentium iii @ 800 mhz for cuts as described in this article : + 5 minutes per 1000 constant - weight ccall events ( koralw stand - alone ) + 50 minutes per 1000 constant - weight ccall events with the correction ( the cmc koralwyfsww3 , max . weight for rejection increased by a factor 2 ) .

hep-ph0104049

after many years of fruitful operation , the lep experiments have been closed down and the lep2 data analysis is approaching its final stage . in the area of @xmath2-pair physics the experimental precision is very high : in order to match it , theoretical calculations must include not only tree - level four - fermion born contributions , with the numerically leading higher - order effects ( mostly qed ) , but also the complete @xmath1 electroweak ( ew ) corrections to @xmath2-pair production @xcite . this applies not only to inclusive quantities such as the total cross section but also to various differential distributions , such as the angular or invariant mass ones . to date there is , however , no single monte carlo ( mc ) event generator that would include simultaneously the complete four - fermion background for massive fermions and the @xmath1 ew corrections to @xmath2-pair mediated processes in _ all _ possible @xmath2-pair decay channels . for example koralw @xcite can generate _ all _ four - fermion final states with the fully massive phase space and the complete born - level four - fermion massive matrix element generated by the grace2 package @xcite . apart from koralw there exists a number of other mc programs for all four - fermion processes at the born level @xcite . the complete @xmath1 corrections to the signal @xmath18 process are implemented only in two mc programs : yfsww3 @xcite and racoonww @xcite . yfsww3 includes the library of electroweak corrections from refs . @xcite . the racoonww program , in addition to @xmath1 corrections to the @xmath3 process , can also calculate the four - fermion corrections in the massless fermion approximation and single , hard , non - collinear , real photon radiation in all four - fermion processes . the massless fermion approximation prevents racoonww from being fully exclusive . also , racoonww seems to still have some problem with providing constant - weight events in the full operational mode . concerning the efficient use of the koralw and yfsww3 mc event generators , the critical open question is : how is it possible to combine their results , so that for every interesting physical observable we get a prediction that includes the complete @xmath1 standard model ( sm ) corrections for the @xmath2-pair production and decay process , keeping sufficient control on the smaller contributions from the `` background diagrams '' ? before we answer this question , let us mention important practical limitations and requirements . for the purpose of the lep2 data analysis it is of paramount importance that the results of yfsww3 and koralw are combined for the _ fully exclusive _ distributions , in other words on an _ event - per - event _ basis . it is not sufficient to combine the predictions of two separate mc runs of yfsww3 and koralw programs for _ inclusive _ observables such as an integrated cross section , asymmetries , single - dimensional angular or @xmath2-mass distributions . such a procedure is not sufficient for full detector simulations and for data analysis , in which the important sm parameter , the mass of the @xmath2 , is fitted to experimental data using a series of the ( fully exclusive ) mc events ! apparently , we are asking whether the koralw and yfsww3 mc programs could be merged into a single new mc event generator , that is a single mc program with a single source code and a single executable object in the machine processor . in principle , this could be done , but not within the time left for the lep2 data analysis . nevertheless , the situation is not completely hopeless and there seem to be some sensible ways out . one possible solution is to combine koralw and yfsww3 into a single tool using events stored on the mass data storage , which we shall call a `` disk file '' or simply a `` disk '' . storing events is done routinely for the purpose of the data analysis anyway . in this scenario , constant - weight events generated with koralw are stored on the disk and later on read by yfsww3 and finally corrected for the missing @xmath1 terms with the help of a special correction weight . the resulting events would be variable - weight ( weighted ) events . this kind of organization is not completely trivial and requires certain `` tuning '' of both programs ; see below for the details . note also that ; for the purpose of fitting the @xmath2 mass , the events generated by koralw and stored on the disk can be corrected in a similar way by koralw , with the weight corresponding to a change of the mass of @xmath2 , or due to any change of the other sm parameters ( any change of the input data of koralw ) . coming back to the above procedure of combining koralw and yfsww3 , we see two important disadvantages : ( i ) running two separate mc programs that communicate through a disk file is inconvenient and ( ii ) the correction weight of yfsww3 may have a long tail , so that it would be difficult or impossible to produce constant - weight ( unweighted ) events through a rejection technique . in this article we present another solution to the above problems and the corresponding mc tool , based on the koralw and yfsww3 mc programs , which is able to provide constant - weight events , implements the @xmath1 corrections for @xmath2-pair production process and includes all of the background diagrams . the present new version 1.51 of the koralw program provides a programming framework for this new solution . contrary to the previous solution where koralw and yfsww3 were communicating through a disk file , here , variable - weight events from koralw are sent immediately , in real time , as an input to yfsww3 , using the `` named pipe '' of the fifo mechanism in the unix / linux operating system . yfsww3 calculates the @xmath1 correction weight and sends it back to koralw with the help of another `` named pipe '' of the fifo . afterward , koralw performs the final rejection according to the total mc weight and invokes hadronization , etc . the important advantage of the above method is that it provides the constant - weight events with the @xmath1 corrected @xmath2-pair process quite efficiently , including all the background diagrams , the higher - order isr corrections , the hadronization , etc . for the fifo - based solution , as compared with the disk - mediated solution mentioned earlier , _ no _ additional modifications of the fortran source codes of both programs are necessary . since koralw and yfsww3 run as two separate , concurrent processes , which communicate with one another , we call this solution a `` concurrent monte carlo ( cmc ) koralw@xmath4yfsww3 '' . from the user s point of view , it acts like a single mc program . to our knowledge , this could be the first important practical application , albeit rather simple , of the concept of `` concurrency '' in the area of the high energy physics monte carlo event generators . we shall also discuss very briefly possible future extensions / improvements of the above technique . the modifications of the yfsww3 program necessary for this technique are discussed in detail in @xcite and in this paper we shall describe them only to a minimum necessary extent . the second group of modifications included in koralw version 1.51 is motivated by the use of koralw to study the background to two - fermion processes due to the emission of a second fermion pair . in short , the modifications provide a number of approximate matrix elements , denoted according to ref . @xcite as isns , fsns , etc . , and a new `` extrapolation procedure '' better suited for the @xmath19-channel - dominated photonic radiation . the layout of the paper is as follows . in section 2 , we discuss various ways of merging koralw with yfsww3 . in particular we show how to do it by means of the fifo ( `` named pipes '' ) mechanism and discuss how to reweight previously generated events from tapes . in the section 3 , we provide some numerical tests of the cmc koralw@xmath4yfsww3 . in section 4 , we describe in detail all modifications of koralw version 1.51 related to the @xmath3 physics and the reweighting procedures , the two - fermion physics and miscellaneous topics , respectively . in section 5 , we explain how to install the version 1.51 of koralw and in section 6 we describe briefly the organization of the source code . in section 7 , we describe in detail various demo programs included in the package , with special emphasis on the practical use of the fifo ( `` named pipes '' ) mechanism and the construction of the concurrent monte carlo . we summarize the paper in section 8 . in an appendix , we describe new and modified program parameters . in this section we shall describe in detail the method of combining results of the mc event generators koralw and yfsww3 at the level of the _ fully exclusive _ differential distribution , such that the resulting distribution features the @xmath1 corrections for the @xmath3 signal process and the background graphs of the four - fermion process ( with the isr corrections ) . as already indicated , this can be done either by using a series of mc events stored on the disk or through the concurrent use of koralw and yfsww3 , which effectively act together as a single mc event generator . in both scenarios the underlying methodology of constructing mc correction weights is the same . it will be described in detail in the following . let us remind the reader that koralw is a dedicated mc event generator with a powerful four - fermion phase - space generator capable of generating every possible four - fermion final state in the complete phase space for massive fermions ( including electrons ) @xcite with the importance sampling due to all possible singularities in the feynman diagrams . koralw uses the exact massive born - level matrix element generated by the grace2 package @xcite . on the other hand , the yfsww3 mc event generator @xcite is the mc program dedicated to the @xmath2-pair production and decay process . it includes the complete @xmath1 ew library of real and virtual corrections to the @xmath2-pair production process of refs . @xcite , along with the multiple photon radiation from the @xmath2-pair ( wsr ) . in the following we will often use the notation non - leading ( nl ) @xmath1 corrections to denote the remaining part of the @xmath1 ew corrections after subtraction of the `` trivial '' leading universal corrections : initial state radiation ( isr ) and coulomb correction ( cc ) . certain features of koralw and yfsww3 are critical on the possibility of combining their results for fully exclusive differential distributions . the most important is that both programs do implement well - defined , fully exclusive , distributions for four final - state fermions and @xmath20 photons , normalized with respect to the standard lorentz - invariant phase space ( lips ) @xmath21 as defined in the pdg @xcite . the four - momenta of the final - state fermions are @xmath22 and of the photons are @xmath23 , @xmath24 . in spite of the fact that both programs use the leading - logarithmic ( ll ) models for the higher - order isr and the final - state radiation ( fsr ) , they do not employ an inclusive approach in which the collinear photon is irreversibly associated with the parent fermion ( structure - function approach ) . the massive kinematics of all fermions allows for full coverage of the phase space . it is also helpful that both programs implement the same cc03 matrix element in the t hooft feynman gauge , which coincides with the gauge - invariant leading pole approximation ( lpa ) of the complete four - fermion born - level matrix element and has , therefore , a well defined physical meaning , see @xcite for more discussion . the @xmath1 nl effect comes as a correction to the above cc03 distribution ; see below for more details . even before the advent of the cmc koralw@xmath4yfsww3 , both programs ( in their unpublished versions ) gradually acquired the capability of writing the four - momenta of generated events , together with some auxiliary information , to an external device and reading them back in order to calculate the correction weight . the correction weight was that corresponding to the modification of the fully exclusive differential distribution : @xmath25 due to a change of the input parameters of the mc event generators . the most important change was due to a variation of the input @xmath2-mass and was instrumental in the fitting of the @xmath2-mass to the lep experimental data using a series of the mc events from koralw or yfsww3 . it is quite likely that this kind of facility will be a standard feature for any future mc event generator aimed at precision measurements in the future experiments . the analysis of the lep data points out in that direction for future developments . as indicated in the introduction , the next non - trivial step was to provide one of the programs , koralw or yfsww3 , with the capability of correcting the fully exclusive distributions using the correction weight calculated by another program . in practice , it has turned out that correcting the events produced with koralw using the correction weight of yfsww3 leads to a lower rejection rate for the final mc weight than the other way around , i.e. correcting the events of yfsww3 with the weight of koralw . the reason is that koralw has a better importance sampling for the background processes ( some of them dominated by the @xmath19-channel exchange ) . koralw does not have a sufficiently good importance sampling for wsr . however , this seems to matter less than the lack of importance sampling for the background processes , particularly the ones with electron(s ) in the final state . the main difficulty in developing the above cross - correcting capabilities of one mc program by another was to match correctly the relative normalizations of the fully exclusive differential distributions in both programs in order to define the correction weight properly ; see below . finally , after both programs have evolved to acquire subprograms for reading and writing events on the external device and calculating the properly normalized correction weights for use by the same or other programs , it was a purely technical exercise to organize both programs in such a way that cross - correcting one program by another could be done `` in flight '' by running _ simultaneously _ koralw and yfsww3 , as two independent concurrent processes . the communication between concurrent processes was organized under unix / linux quite easily with the help of the standard fifo facility of the `` named pipes '' , using the already existing subprograms for reading / writing on / from the external device . in this way , a new solution has emerged , which we call the concurrent monte carlo ( cmc ) koralw@xmath4yfsww3 . from the users point of view it acts as a single monte carlo program with all its regular features . the main advantage of the cmc koralw@xmath4yfsww3 with respect to the previous solution ( with disk files ) is that it can produce efficiently the constant - weight mc events . these events can be fed into a detector mc simulation program and stored on the disk . there is no problem with applying to these events a correction weight due to a change of the @xmath2-mass or other sm parameters . in the first part of this section , we shall discuss the theoretical foundations and requirements for this kind of a merge of two different programs . then , we shall go into the details of the technical realization of the cmc koralw@xmath4yfsww3 solution . the actual modifications of the koralw will be presented in detail in one of the later sections . in the following we shall discuss the physics meaning of the correction weights . after introducing our basic notation and terminology we shall briefly characterize fully exclusive differential distributions in koralw and yfsww3 and define a common reference differential distribution with which we shall define correction weights in both programs . in the following considerations we shall use the differentials @xmath26 , which we always understand as the fully exclusive differential distributions normalized to the @xmath27-particle lorentz - invariant phase space @xmath28 in the following we shall need a clearly defined terminology for the various contributions and ( perturbative ) corrections in @xmath26 . we shall often use objects like @xmath29 . what is the meaning of the symbols used as superscripts and subscripts ? here is the complete list , with explanations : * @xmath30 and @xmath31 : denotes the origin from the mc program : koralw and yfsww3 , respectively . * cc03 : in yfsww3 , we implement the leading pole approximation ( lpa ) to define the double - resonant , @xmath3 , component of the @xmath32 process ; in particular cc03 denotes here the tree - level ( born ) part of lpa , which coincides with the cc03 matrix element in the t hooft feynman gauge . we shall sometimes use @xmath33 or cc03@xmath34 to underline that we use yfsww3 with the version lpa@xmath34 , see refs . * @xmath1 : denotes the lpa with the complete on - shell @xmath1 corrections for the @xmath35 process ; we understand that this includes the cc03 tree - level , one - loop @xmath1 virtual corrections and the exact qed matrix element for photon emission from @xmath3 ( wsr ) . the one - loop ew corrections to decays are included in the present version of yfsww3 as an overall factor and real photon emission is implemented using the approximate treatment of photos @xcite . * @xmath36 : this means that we include the and qed isr correction ( which has not been already included through exponentiation ) ; it is always understood that it is done in the ll approximation ; a similar subscript like @xmath37 is self - explanatory ; in particular subscript @xmath38 means @xmath39 exponentiation ( the born - level matrix element convoluted with the photon emission in the soft - photon approximation ) . * @xmath40 : this means that , close to the @xmath3-threshold , we include properly the non - relativistic coulomb effect , which far from the threshold is matched correctly with the @xmath1 wsr . if not stated otherwise , we use a variant of @xmath40 with the `` screening '' of ref . @xcite , which incorporates the numerically leading part of the non - factorizable qed interference between two @xmath2 decays . * @xmath41 : this denotes the tree - level matrix element for @xmath32 ( the constant @xmath2 width , massive fermions ) , which can be split into the double - resonant cc03 part and the four - fermion - correction due to background diagrams . as already indicated , we can either correct @xmath42 of yfsww3 using the correction of koralw according to @xmath43 or correct @xmath44 of koralw using the correction of yfsww3 according to @xmath45 before we enter into more of the details of how the above is implemented in terms of the mc correction weight , let us briefly characterize the fully differential exclusive distributions generated by yfsww3 and koralw . let us now explicitly define the component distributions in eqs . ( [ y / k ] ) and ( [ k / y ] ) . the fully exclusive differential distribution of koralw is given in should be replaced by @xmath46 . ] eq . ( 4 ) of ref . @xcite : @xmath47 , \end{split } \label{master}\ ] ] where @xmath48 for the definition of the yfs form factor @xmath49 , ir - finite @xmath50 s and other elements in the above distribution we refer the reader to ref . let us only mention that the tree - level four - fermion matrix element is hidden in @xmath51 . the analogous distribution for yfsww3 , see ref . @xcite , is more complicated not only because it features photon emission from @xmath2 s , but also because it includes summation over photon partitions , that is over photon associations to either isr or wsr . this trick is useful for efficient introduction of the ll corrections beyond . as we shall see later , it is relevant for the discussion of the reweighting procedures . for @xmath20 photons , a single partition is represented by the vector @xmath52 , @xmath53 ( @xmath54 for isr and @xmath2 for wsr ) . the sum over partitions is weighted by the partition weight @xmath55 where @xmath56 is adjusted such that @xmath57 and @xmath58 for wsr is defined analogously to @xmath59 of eq . ( [ eq;s - tilde - i ] ) . the exclusive differential distribution of yfsww3 reads as follows : @xmath60 where radiation from the @xmath2-pair is included in @xmath61 furthermore , @xmath62 and @xmath63 denote the four - momenta of the @xmath64 and @xmath65 , respectively . the yfs form factor @xmath66 also includes the wsr . the complete @xmath1 corrections for the @xmath2 pair production process reside in the @xmath67-functions . for the @xmath68 corrections , the sum over real photons extends only over isr photons . see ref . @xcite for an explanation of the rest of the notation . the mc programs koralw and yfsww3 generate the events according to the distributions of eqs . ( [ master])([master ] ) . in fact , both of these distributions come in several variants , with certain physical effects and higher - order radiative corrections switched on / off . it is therefore possible , in principle , to calculate separately any component in the distributions of eqs . ( [ y / k ] ) and ( [ k / y ] ) . in the real mc programs it is , however , difficult or impossible to find a single subprogram , that provides numerically the distributions @xmath69 exactly as defined by eqs . ( [ master])([master ] ) , without any additional factor . in the typical mc program one always deals with the mc weights , which include not only expressions like eqs . ( [ master])([master ] ) , but also some factors representing technicalities of the mc algorithm . the mc weight is @xmath70 , where @xmath71 is the `` target '' distribution defined by the physics model and @xmath72 is the multidifferential distribution actually generated in a given mc program using elementary ( primitive ) mc techniques . @xmath72 is different in koralw and yfsww3 and the mc weight @xmath73 can be different in two mc programs , even if @xmath71 is the same . because of the above specific normalization properties of the mc weights , it is not trivial to construct the correcting weight , which is calculated in one mc program and used in another one . the basic practical methodology relies on introducing a certain `` reference '' differential density @xmath74 , which is identically the same for koralw and yfsww3 , and there is a mc weight in both programs representing @xmath74 . in principle the @xmath74 is a dummy quantity that serves only the purpose of fixing the absolute normalization between the two programs . for @xmath74 understood like this , we could , of course , pick `` any '' distribution even one that does not coincide with any meaningful physical model . however , we shall also need another auxiliary distribution , @xmath75 , which is the maximal common part of the best distributions of the two programs to be merged . in other words , the distribution @xmath75 should include all components / corrections that are present in both koralw and yfsww3 and should not include any correction that is present in only one of them . the simplest possible approach is to choose the reference distribution @xmath74 to be equal to @xmath75 . accordingly , our choice of @xmath74 is the following : @xmath76 as we remember , cc03 we understand in the gauge - invariant way in terms of the lpa . how is the common reference @xmath74 realized in koralw and yfsww3 ? in yfsww3 the starting point is the differential distribution of eq . ( [ master ] ) in order to realize the universal @xmath74 , the radiation from the @xmath2-pair must be switched off . only one partition @xmath77 remains . also the perturbative series of the @xmath50-functions must be truncated to the cc03 matrix element with the isr up to third order in the ll approximation and the ( screened ) coulomb correction . for koralw @xmath74 is equal to the @xmath44 of eq . ( [ master ] ) with the four - fermion born matrix element simplified to the cc03 level , i.e. @xmath78 . we also have to remove another possible source of the difference in @xmath74 as implemented in both programs . the function @xmath79 is really identical in both programs ( for any fermionic four - momenta ) only in the case without additional photon radiation . otherwise , in the presence of additional photons , attention has to be paid to the so - called `` extrapolation / reduction procedure '' . this procedure extrapolates @xmath79 from the 4-body ( four - fermion ) phase - space into the @xmath80-body phase - space ( with the additional @xmath20-photons ) . it is generally not unique and has been defined in a slightly different way in koralw-1.42 and in yfsww3 - 1.16 . this `` extrapolation / reduction procedure '' is marked in eqs . ( [ master])([master ] ) by superscript @xmath81 . we have now modified it in koralw-1.51 to coincide ( optionally ) with that of yfsww3 - 1.16 . in this way , we removed the last source of discrepancy between the @xmath74 as implemented in the two programs . after defining precisely the fully exclusive differential common reference distribution @xmath74 and implementing the corresponding mc weight in both programs , koralw and yfsww3 , the next important step is to check _ numerically _ that the integrated cross section and one - dimensional distributions , such as the distributions of the @xmath2 invariant mass , of the @xmath2 production angle and of the photon energy are ( within statistical errors ) _ identically the same _ , for the mc weights of @xmath74 . such tests were performed very extensively and they have shown the full agreement of the distributions and the cross sections from the two programs , see section [ sec : numeric ] . having completed all the above preparatory steps , we can now precisely define the actual correction weights used in the cmc koralw@xmath4yfsww3 and other similar possible scenarios . equations ( [ y / k ] ) and ( [ k / y ] ) can be rewritten as @xmath82 where @xmath83 let us remind the reader that we have chosen @xmath84 identically the same for koralw and yfsww3 . in the mc realization , the bracket factor on the r.h.s . ( [ + cc03])([k+y ] ) represents the correction weight in a given reweighting procedure ; for a more detailed discussion see the following section [ sec : technics ] . the great practical advantage of introducing @xmath74 is now seen in the above definitions of the corrections @xmath85 and @xmath86 . they are expressed in eqs . ( [ eq - cor4f ] ) and ( [ eq - coro1 ] ) as the mc - generator - independent ratios of @xmath26 s ; however , inside a given mc generator they will be calculated as a ratio of the generator - dependent mc weights ( they are the only available objects there ) . the generator dependence of the mc weights ( from @xmath72 ) cancels out in the ratio of the mc weights . there is also another , multiplicative , way of combining the @xmath1 and four - fermion corrections @xmath87 one can see immediately that it differs from the additive scheme of eqs . ( [ y+k ] ) and ( [ k+y ] ) by the term @xmath88 , which is definitively a part of the higher - order corrections to the background ( non - cc03 ) graphs . as this latter correction has not been calculated so far , one does not know how close the above @xmath88 term is to the actual correction . it can , at best , be treated as a rough indication of the order of magnitude of the true correction . the same general scheme can be applied to reweighting due to any kind of available corrections . for example to correct the events generated with some old versions of koralw one has to define @xmath89 equal to the old setup @xmath90 , calculate the appropriate @xmath91 correction and construct the new distribution @xmath44 @xmath92 in the actual implementation , yfsww3 provides only an approximate version of the correction weight : @xmath93 where the differential distribution of yfsww3 is a simplified version of @xmath94 of eq . ( [ master ] ) , in which the sum over partitions is restricted to one term , in which all photons belong to the isr @xmath95 with the partition weight set to 1 : @xmath96 . as we shall see in the numerical tests presented in the following section , the above approximation is good enough for all practical lep2 applications ; its precision is better than @xmath97 . it is interesting and instructive to look also at the actual form of the @xmath98-corrections in terms of the @xmath50-series and ultimately the feynman graphs for our definition of @xmath99 . for the background - graphs correction we have @xmath100_{\rm cc03 } } \\ & = \frac{\vert m_{}^{4f}(\{p , q\}^{{{\cal r}}})\vert^2 } { \vert m_{}^{\rm cc03}(\{p , q\}^{{{\cal r}}})\vert^2}. \end{aligned } \label{nohot}\ ] ] the last equation follows from the fact that the second- and third - order ll expressions for the isr @xmath101-functions are exactly the same in both @xmath102 and @xmath103 , and also because we have used the same extrapolation / reduction procedures in both distributions ( note that the extrapolation procedure is fixed by the requirement that @xmath74 is identical in koralw and yfsww3 ) . the factorization property of the ll ansatz used in the construction of the isr @xmath104 series , cf . ref . @xcite , is , of course , essential . both distributions in the numerator and denominator of eq . ( [ nohot ] ) are defined in koralw , that is the appropriate mc weights exist for them . the case of @xmath105 , provided by yfsww3 , can be analysed as follows : @xmath106_{\rm cc03 } } \right ) \\ = & \frac{\vert m_{}^{{\cal o}(\alpha)}(\{p , q\}^{{{\cal r}}})\vert^2 + \hbox{h.o.t . } } { \vert m_{}^{\rm cc03}(\{p , q\}^{{{\cal r}}})\vert^2 + \hbox{h.o.t . } } \label{hot } \end{aligned}\ ] ] the additional higher - order terms ( h.o.t . ) in the denominator include the effects due to the isr . the numerator is the matrix element squared for the one - photon isr@xmath107wsr . in the presence of exponentiation , this is true modulo terms , owing to the specific definition of the @xmath108 in the yfs exponentiation , which admits the presence of the virtual corrections , even for a hard photon . for two and more hard photons the biggest correction in the h.o.t . will be of . note that there is _ no _ sum over partitions in the above equation . if , however , we used the exact @xmath86 instead of the approximate @xmath105 the sum over partitions would reappear and the above formula ( and a relation to the feynman graphs ) would become even more complicated . in the following section we shall discuss more practical aspects of combining the two programs into one . in particular we shall present , in detail , the idea of the `` concurrent monte carlo '' realized via the fifo mechanism . the first , most natural way is to add the four - fermion and @xmath1 corrections at the level of the _ inclusive _ observables , that is for integrated cross sections and inclusive one- or two - dimensional distributions ( histograms ) . in this case both sides of eqs . ( [ y / k ] ) and ( [ k / y ] ) are integrated either completely or almost completely , for instance leaving one or two variables unintegrated . the inclusive distributions or the integrated cross sections entering eqs . ( [ y / k ] ) and ( [ k / y ] ) are calculated separately from the independent runs of the two programs and combined according to eqs . ( [ y / k ] ) and ( [ k / y ] ) . the whole procedure is most convenient for the variable - weight events , mainly because one may use the differences of the weights . the application of this inclusive correcting scheme is quite straightforward . suppose that we want to calculate the @xmath2-mass distribution . using the first scheme @xmath109 of eq . ( [ y / k ] ) , we proceed as follows : 1 . make the properly normalized histogram of @xmath110 with a sufficiently long run of the yfsww3 mc with the variable- or constant - weight events . make the properly normalized histogram of @xmath111 with koralw . this can be done by running koralw once ( twice ) with the variable - weight ( constant - weight ) events . add the two histograms . using the second scheme @xmath112 of eq . ( [ k / y ] ) , we proceed in the analogous way : 1 . make the properly normalized histogram of @xmath113 with the run of the koralw with the variable- or constant - weight events . make the properly normalized histogram of @xmath114 with the help of yfsww3 . this can be done by running yfsww3 once ( twice ) with the variable - weight ( constant - weight ) events . add the two histograms . both schemes must give the same results , but in practice it is reasonable to use the one that will require , in the given circumstances , the smaller @xmath115 correction and less fluctuations in the correcting weight . for example , in channels such as @xmath116 , the four - fermion correction turns out to be negligible for energies away from the @xmath2-threshold ; @xmath117 can therefore be completely neglected @xcite . the above inclusive approach has serious limitations . the most important is that it is not fully exclusive ( event per event ) . consequently it can not be used in a data analysis that takes the detector simulation properly into account . on the other hand , this approach can be useful for all kinds of theoretical studies . note that this method was used in ref . @xcite for combining the results of koralw and grc4f . the inclusive method of the previous subsection was not _ event per event _ because the mc events of koralw and yfsww3 were at different random points of the full phase space . consequently , the mc weights of the events from these two generators were also not calculated at the same points of the phase space ; therefore , we could not take their ratios ( apply correcting weights ) . in this section , we shall describe another , more sophisticated , method of combining fully exclusive differential distributions of koralw and yfsww3 , in which both programs work with _ the same _ events , that is the same random points in the @xmath80-dimensional phase space . one of the mc programs , the master mc , generates the event and the other one , the slave mc , instead of generating its own event , reads the event of the master mc and calculates the correction weight due to some missing effect , for instance the @xmath1 correction or the correction due to the missing background diagrams . the transfer of the events from one program to the other can be organized with the help of the external device ( disk or tape ) or `` in flight '' . in the first case , the master mc writes the events on the disk and the slave mc reprocesses the events read from the disk . writing the events on the disk is done routinely in the data analysis anyway . for the description of the second , `` in - flight '' method , see the next subsection . the above procedure of cross - correcting the distributions / events from one mc program with the help of the second mc is not the only one , and is not the simplest one either . a similar procedure is possible , even with a single mc , when the events are stored on the external device and later on , in a separate run , they are corrected by _ the same _ mc program with the correcting weight due to the change of the input parameters ( typically the @xmath2-mass ) . this is very useful for the data analysis , where the constant - weight mc events from a physics mc event generator , like koralw or yfsww3 , are processed through the detector simulation . in this way the cpu - time - consuming reprocessing of the detector simulation is avoided . it also allows fitting the sm parameters to experimental data ( typically the @xmath2-mass ) with full control of the effects due to the detector acceptance . one has to remember , however , that the above procedure provides us essentially with the variable - weight events . if the correction weight is fluctuating very mildly , then this is not a problem ; otherwise , the strongly fluctuating correction weight would lead to higher statistical errors in the calculated observables , and would inhibit the optional transformation of variable - weight events into constant - weight events . as seen from the above discussion , it is essential that the mc event generators can write / read the events into / from disk files . the appropriate tools for reading the events from the disk and calculating the correction weight due to the change of the input parameters and the change of the scattering matrix element have been introduced in the version 1.51 of koralw program presented in this paper , as described in the next section , and in the yfsww3 program as well - physics data analysis . ] , see ref . let us now describe , in more detail , the procedures of cross - correcting the distributions / events from one mc program with the help of the second . there are several scenarios for such a procedure following eqs . ( [ + cc03])([k*y ] ) they differ in the choice of the master / slave mc ( koralw or yfsww3 ) and in the way the correction weight is constructed ( additive or multiplicative ) . let us look closer at the `` symmetric '' scheme of eq . ( [ + cc03 ] ) . it consists of the following steps : 1 . generate the variable- or constant - weight events according to the fully exclusive distribution @xmath99 , using either koralw or yfsww3 , and store all events on the disk . 2 . calculate the correction @xmath85 of eq . ( [ eq - cor4f ] ) using koralw and write it in the disk record for each event . 3 . calculate the correction @xmath86 of eq . ( [ eq - coro1 ] ) using yfsww3 , construct the correction weight @xmath118 , and write it in the disk - record of each event . ( in fact , we have to use the substitute @xmath105 of eq . ( [ eq : deltanlapprox ] ) , since the @xmath86 is not provided by yfsww3 version 1.16 ) . optionally reject events according to @xmath119 . of course , it is not really necessary to reprocess the mc events twice ( total of three runs ) and one may follow the simpler `` asymmetric '' procedure @xmath109 in which yfsww3 is the master mc : 1 . generate the variable- or constant - weight events according to the fully exclusive distribution @xmath94 using yfsww3 . store each event on the disk , together with the value of @xmath86 . 2 . in the second run , for each event use koralw in order to calculate the correction weight , the bracketed expression in eq . ( [ y+k ] ) . optionally , reject the events according to the correction weight . the analogous scenario @xmath112 in which koralw is the master mc looks as follows : 1 . generate the variable- or constant - weight events according to the fully exclusive distribution @xmath103 using koralw . store each event on the disk , together with the value of @xmath85 . 2 . in the second run , for each event use yfsww3 in order to calculate the correction weight , the bracketed expression in eq . ( [ k+y ] ) . optionally , reject the events according to the correction weight . alternatively , one may follow the multiplicative schemes of eqs . ( [ * cc03])([k*y ] ) . as indicated , the above methods provide , in principle , a sample of constant - weight events . however in practice , this is not convenient , as one would have to deal with millions of variable - weight events stored together with their weights , and then to be reprocessed again to include the second correction , and finally to undergo the final rejection . moreover , for the moment only koralw is capable of continuing the event construction ( rejection , decay libraries , etc . ) based on the input events stored on the external device . yfsww3 version 1.16 can only provide the correction weight . consequently , the only practical option to provide constant - weight events is the one in which koralw is the master mc . the use of the variable - weight events from the master mc is , however , possible in the scenario in which the master and the slave programs exchange the events in - flight , without recording them in the ( large ) disk files , see the following subsection . one of the most important results of the modifications of koralw version 1.51 is the possibility of a direct inclusion of the @xmath1 nl corrections calculated by the yfsww3 - 1.16 program into the process of the event generation by koralw . this is not done , however , by simply compiling and linking them together into a single executable , because the source codes of yfsww3 and koralw share a number of subroutines , common blocks and libraries with the same names , but with different contents . it is therefore practically impossible ( without major rewriting of the two source codes ) to merge the two programs into one at the level of the fortran source code . such a merge might also obstruct a further independent development of the programs . once both programs have acquired capabilities of exchanging the events and reprocessing them using the disk file as an intermediate medium , we have noticed that one can follow another approach , avoiding major rewriting of both programs , based on the unix / linux standard facility called the `` names pipes '' or the fifo mechanism . it allows two _ independent _ processes to communicate `` in real time '' through the `` named pipe '' , into which one process writes and from which the other reads , in turn . in our case , the scheme is such that koralw generates an event and writes its four - momenta into the fifo special file . these four - vectors are then read by yfsww3 , which is running in parallel ( concurrently ) and calculates the correcting @xmath1 nl weight , or more precisely the @xmath86 quantity , and then writes it to another fifo special file . in the final step , koralw reads this weight from the fifo special file , includes it in the total weight and finishes the event construction . this concurrent monte carlo koralw@xmath4yfsww3 scheme is schematically depicted in fig . [ fig : flow1 ] . as a result , koralw can now , for example , perform the rejection and provide the _ constant - weight _ events with both the four - fermion background correction @xmath85 and the @xmath1 nl correction @xmath86 to the @xmath2-pair production and decay process taken into account . another application of such a scheme is the reweighting of events with both the four - fermion and @xmath1 corrections included simultaneously in order to perform `` monte carlo fits '' . to summarize it shortly : from the point of view of the user , the cmc program koralw@xmath4yfsww3 works as a single monte carlo generator with all its normal features . this might be , to our knowledge , the first practical working implementation of such a scheme to the problem of the mc event generation in particle physics . remarkably , the efficiency of the cmc koralw@xmath4yfsww3 in generating the _ constant - weight _ events according to the four - fermion background and the @xmath1 nl corrected distributions is not bad . in principle one could worry that an additional yfsww3 weight of the order of a few tens would translate into a similar increase of the maximal weight for rejection and , consequently , into a similar increase of the cpu time needed for the event generation . surprisingly , however , this is not true ! in the additive mode of combining koralw and yfsww3 the increase of the maximal weight that we recorded on the @xmath120 sample of variable - weight events was @xmath121 , for example . overweighted events were of two topologies . the first topology was strongly @xmath3-like with multiple , i.e. at least triple , soft bremsstrahlung and led to overweights below 2 ( in an extreme case with seven photons , we recorded an overweight of 4 ) . the other topology was very far from @xmath3-like ( small values of the @xmath64 , @xmath65 invariant masses ) and gave the highest over - weights ( @xmath122 ) . in the latter cases the use of the on - shell @xmath1 nl library is less justified anyway and these few events can simply be discarded , instead of increasing the maximal weight for the rejection . why was the overweighting so small ? because in koralw to cover the four - fermion - dominated configurations the maximal weights are adjusted higher than what is needed by the @xmath3-like configurations . this extra safety margin turns out to be sufficient to absorb fluctuations of the additional @xmath3-like @xmath1 nl weights . the situation is different for the multiplicative prescription . here , the big four - fermion weight is multiplied by the @xmath1 nl correction , on occasion leading to overweights of the expected order of a few tens . this happens , typically , for a very hard bremsstrahlung photon , with energy of the order of 40 gev . however , as we have explained earlier , the @xmath1 correction for the configurations far away from being @xmath3-like is less justified and the overweights provide merely another reason for using the additive scheme rather than the multiplicative one . let us add a few additional comments on the koralw@xmath4yfsww3 cmc : 1 . the corrections can be combined in an additive or multiplicative way , according to the discussion from the previous subsections . 2 . one may also wonder about other possible source of efficiency loss caused by running two programs at the same time on a single processor . our experience shows , however , that there is very little loss due to an alternate suspension of the programs , performed automatically by the operating system ( in the unix kernel ) . the big advantage of the fifo special files is that they are in practice done by the operating system `` in flight '' , in virtual memory . consequently , there is no need for a huge amount of disk space to store all the millions of generated events needed for generating the sample of the constant - weight events . 4 . in order to introduce the `` named pipes '' , there is _ no need to modify _ in the fortran source code of either of the programs . the only change is the replacement of regular input / output files by the special fifo files , done at the level of the operating system and not in the fortran source code . we give an example of such a procedure in the demo run of the cmc koralw@xmath4yfsww3 , which can be invoked through : make kandy ( kandy stands for `` k and y '' ) . it creates two special fifo files 4vect.data.special and wtext.data.special that will serve for transmitting the four - momenta and the weights between the programs . these special fifo files are then linked to the appropriate input / output file names for koralw and yfsww3 in their local working directories . finally , the two programs are executed with the _ common default data cards _ ! they are modified later on by the user cards located in the local working directory . the proposed scheme of cmc koralw @xmath123 yfsww3 @xmath123 koralw is not the only possibility of realizing the four - fermion background and @xmath1 nl corrected events . alternatively , one can construct the cmc the other way around : yfsww3 @xmath123 koralw @xmath123 yfsww3 ( symbolically yfsww3@xmath4koralw ) . in general , this direction has a serious drawback the four - fermion correction weight from koralw can be very high , especially for the final states with electrons , spoiling the convergence of the series . on the contrary , the yfsww3 weight for the @xmath1 nl correction is well behaved and for lep2 energies it seems not to exceed a few tens . however , if the phase space for generation were restricted to @xmath3-like configurations the situation would be reversed the four - fermion correction weight would be smaller than the @xmath1 nl one and the scheme yfsww3@xmath4koralw would be a good choice . we conclude this section with two technical remarks on the fifo mechanism . they may prove useful in practical implementations of fifos . 1 . the fifo special files must be created ( by the command `` mkfifo _ file_name _ '' ) on a _ local _ disk of a computer on which the program will be executed . this means in particular that it can not be created directly on the afs file system or even on a mounted remote file system . the good location is , for example , the /tmp directory . after creation , however , the fifo special file can be symbolically linked to any location in the mounted or afs file systems . the fifo special file itself requires only 4 kb of disk space . 2 . in certain configurations of the data transfer through named pipes , it may be at some point necessary to flush the output buffers of the programs . the syntax of the fortran command for that operation is call flush(_unit_number _ ) . our present modest application of the parallel processing to mc event generation is schematically depicted in fig . [ fig : flow1 ] . note that on a dual - processor machine , at a certain moment koralw is calculating the four - fermion matrix element while yfsww3 is _ simultaneously _ calculating the electroweak corrections . in the present form , however , koralw@xmath4yfsww3 does not really allow us to profit fully from running on the multiprocessor machine . it is mainly because the calculation of the nl correction takes on average longer than the calculation of the four - fermion correction ; as a result the koralw process is sometimes waiting for yfsww3 before the final reprocessing of the event . also hadronization by jetset , which takes a substantial amount of cpu time and is independent from koralw and yfsww3 , is not delegated to a separate process , but included as a part of koralw . in fig . [ fig : flow2 ] we show another possible future concurrent arrangement , which would work more efficiently on the multiprocessor installation . here , four independent processes work `` in the cascade '' , in such a way that when the last process ( pythia ) finishes to hadronize an event , the first process may have already started to construct the next event . a similar solution was proposed in ref . @xcite , however limited to a solution in which mc generators communicate through disk files ( database ) . of course , there are many other possible variants of such a scheme the best one should be adapted to a particular mc generation problem and to available hardware . the important advantage of such a concurrent arrangement is that it provides `` encapsulation '' for the `` dusted deck programs '' , without the need of laboriously translating them to object - oriented c++ or another oo programming language . it also allows an easy combination of programs written in different programming languages . this may prove to be in practice a rather effective solution , before eventual emergence of the next generation of event generators , written from scratch in the oo environment . although the basic idea of reweighting mc events , produced either by the same mc event generator or the other one , is relatively simple , its actual implementation has to be tested very carefully . such tests are the principal aim of the present section . we shall concentrate on the reweighting of mc events produced by koralw using the correction weight produced by yfsww3 , that is on the `` asymmetric '' procedure @xmath112 described in subsection [ sec : merg - exclu ] . before the programming tools for such a scenario can be fully trusted , we have to perform certain important introductory numerical tests : ( a ) : : we are going to check numerically that the common reference differential distributions @xmath74 generated by koralw and yfsww3 are the same . ( b ) : : for the distribution @xmath124 we are going to test the reweighting tools of koralw and yfsww3 : + ( b.1 ) ; ; self - test of yfsww3 : we shall compare results from the standard run of yfsww3 and from the run @xmath125 in which @xmath126 is obtained by reweighting events generated according to @xmath74 . ( b.2 ) ; ; calibration of @xmath109 using yfsww3 : we shall compare results of @xmath109 ( in which events generated according to @xmath74 by koralw are reweighted with the help of yfsww3 ) with the direct results of yfsww3 and with the results of the @xmath125 scheme . in both kinds of tests , numerical results for the cross sections and the key distributions , for instance the distributions of the @xmath2 mass , @xmath2 scattering angle and photon energy , should agree within statistical errors . we shall use the approximate version of the correcting weight ( @xmath127 from yfsww3 ) . test ( b ) will provide the measure of the quality of the approximation . in all numerical tests in this section , the input parameter set - up and the definitions of event acceptances are that of ref . @xcite , unless stated otherwise . for the convenience of the reader , let us recall briefly these acceptance conditions . 1 . we required that the polar angle of any charged final - state fermion with respect to the beams be @xmath128 . all photons within a cone of @xmath129 around the beams were treated as invisible , i.e. they were disregarded in the calculation of any observable . 3 . the invariant mass of a visible photon with each charged final - state fermion , @xmath130 , was calculated , and the minimum value @xmath131 was found . if @xmath132 or if the photon energy @xmath133gev , the photon was combined with the corresponding fermion , i.e. the photon four - momentum was added to the fermion four - momentum and the photon was discarded . this was repeated for all visible photons . + in our numerical tests we used two values of the recombination cut : @xmath134 let us remark that we have changed here the labeling of these recombination cuts from the slightly misleading bare and calo names used in ref . @xcite . this change allows us to reserve the bare name for a truly bare setup ( without any recombination ) . the integrated cross sections presented here were obtained without any cuts ( labeled in the tables `` no cuts '' ) or with the cut no . 1 from the above acceptance conditions ( labeled in the tables `` with cuts '' ) . description & program & @xmath135[fb ] & @xmath136[fb ] to 0.4 cm & @xmath137 $ ] & @xmath138 $ ] & @xmath139 $ ] + & yfsww3 & @xmath140 & @xmath141 & & & @xmath142 + @xmath143 & koralw & @xmath144 & @xmath145 & @xmath146&@xmath147 & + 200 gev & ( y@xmath148k)/y&@xmath149 & @xmath149 & & & + & yfsww3 & @xmath150 & @xmath151 & & & @xmath152 + @xmath153 & koralw & @xmath154 & @xmath155 & @xmath156 & @xmath156 & + 200 gev & ( y@xmath148k)/y&@xmath157 & @xmath158 & & & + & yfsww3 & @xmath159 & @xmath160 & & & @xmath161 + @xmath162 & koralw & @xmath163 & @xmath164 & @xmath165 & @xmath166 & + 200 gev & ( y@xmath148k)/y&@xmath157 & @xmath167 & & & + + & yfsww3 & @xmath168 & @xmath169 & & & @xmath170 + @xmath143 & koralw & @xmath171 & @xmath172 & @xmath146 & @xmath147 & + 200 gev & ( y@xmath148k)/y&@xmath173 & @xmath174 & & & + & yfsww3 & @xmath175 & @xmath176 & & & @xmath177 + @xmath153 & koralw & @xmath178 & @xmath179 & @xmath180 & @xmath180 & + 200 gev & ( y@xmath148k)/y&@xmath173 & @xmath167 & & & + & yfsww3 & @xmath181 & @xmath182 & & & @xmath183 + @xmath162 & koralw & @xmath184 & @xmath185 & @xmath186 & @xmath187 & + 200 gev & ( y@xmath148k)/y & @xmath188&@xmath158 & & & + the first test of type ( a ) is presented in table [ tab : yr - xtot - withcuts-200 ] , where we check whether the integrated reference cross section @xmath189 defined in the previous section is numerically the same when calculated by koralw and yfsww3 . it is done for three examples of the cc11 class process ( see ref . @xcite for its definition ) at the center - of - mass system ( cms ) energy @xmath190 gev . in addition to the standard @xmath189 of the previous section , we also show results for @xmath135 , which is a variant of @xmath189 in which the isr is switched off . as already indicated , it is the so - called cc03 process ( in the t hooft feynman gauge ) . as we see in table [ tab : yr - xtot - withcuts-200 ] , the relative differences of the reference cross sections @xmath189 and @xmath135 of yfsww3 and koralw are below @xmath191 . this test indicates that the reference differential cross section @xmath74 is implemented correctly through the corresponding mc weight in both koralw and yfsww3 . in table [ tab : yr - xtot - withcuts-200 ] , we also indicate the size of the correction @xmath192 due to the background diagrams and @xmath193 due to the missing , which are defined as follows : @xmath194 note that in table [ tab : yr - xtot - withcuts-200 ] the final - state qcd correction is excluded and @xmath195 . see also ref . @xcite for a description of the qcd correction . ] from @xmath192 ( i.e. @xmath192 is divided by @xmath196 ) . note also that the coulomb correction is taken here without screening . description & program & @xmath135[fb ] & @xmath197[fb ] to 0.4 cm & @xmath198 $ ] & @xmath138 $ ] & @xmath139 $ ] + & yfsww3 & @xmath199 & @xmath200 & & & @xmath201 + @xmath153 & koralw & @xmath202 & @xmath203 & @xmath204 & @xmath205 & + 161 gev & ( y@xmath148k)/y & @xmath206 & @xmath207 & & & + + & yfsww3 & @xmath208 & @xmath209 & & & @xmath201 + @xmath153 & koralw & @xmath210 & @xmath211 & @xmath204 & @xmath205 & + 161 gev & ( y@xmath148k)/y & @xmath212 & @xmath207 & & & + + & yfsww3 & @xmath213 & @xmath214 & & & @xmath215 + @xmath153 & koralw & @xmath216 & @xmath217 & @xmath218 & @xmath218 & + 500 gev & ( y@xmath148k)/y & @xmath219 & @xmath220 & & & + -like & koralw & @xmath216 & @xmath221 & @xmath218 & @xmath218 & + extrapol . & ( y@xmath148k)/y & @xmath219 & @xmath220 & & & + + & yfsww3 & @xmath222 & @xmath223 & & & @xmath224 + @xmath153 & koralw & @xmath225 & @xmath226 & @xmath227 & @xmath228 & + 500 gev & ( y@xmath148k)/y & @xmath219 & @xmath229 & & & + -like & koralw & @xmath225 & @xmath230 & @xmath227 & @xmath228 & + extrapol . & ( y@xmath148k)/y & @xmath219 & @xmath219 & & & + in table [ tab : yr - xtot - withcuts-161 - 500 ] , we present the analogous ( a)-type test for the integrated cross sections as in table [ tab : yr - xtot - withcuts-200 ] but for two other cms energies : close to the @xmath3-threshold , 161 gev , and at 500 gev , within the range of the future linear collider . as we see , the result of the test is again positive . the integrated reference cross section @xmath189 is again the same from both yfsww3 and koralw to within @xmath231 . in table [ tab : yr - xtot - withcuts-161 - 500 ] , we also show the effect of switching from the extrapolation / reduction procedure of koralw to that of yfsww3 ( which is the source of the largest discrepancy of @xmath232 ) . one can verify that the size of this effect is at the sub - per mille level and appears only in the case with imposed cuts . for the case without any cuts there is no difference in the total cross section , as expected . note that table [ tab : yr - xtot - withcuts-161 - 500 ] shows the case of @xmath233 gev . at @xmath190 gev , there is no difference between these two extrapolation procedures . ( 1600,1480 ) ( 450,1430)(0,0)[cb]*cal05 * ( 1250,1430)(0,0)[cb]*cal25 * ( 850,1430)(0,0)[cb]@xmath234 ( 0,700)(0,0)[lb ] ( 0 , 0)(0,0)[lb ] ( 800,700)(0,0)[lb ] ( -300 , 620)(0,0)[l ] @xmath237 $ ] ( 800 , 0)(0,0)[lb ] ( 1600,1480 ) ( 450,1430)(0,0)[cb]*cal05 * ( 1250,1430)(0,0)[cb]*cal25 * ( 850,1430)(0,0)[cb]@xmath234 ( 0,700)(0,0)[lb ] ( 0 , 0)(0,0)[lb ] ( 800,700)(0,0)[lb ] ( -300 , 620)(0,0)[l ] @xmath238 $ ] ( 800 , 0)(0,0)[lb ] s. jadach _ _ , the monte carlo event generator yfsww3 version 1.16 for w - pair production and decay at lep2/lc energies , preprint cern - th/2001 - 017 , uthep-01 - 0101 , january 2001 ; hep - ph/0103163 ; submitted to comput . phys . commun . j. alberty _ et al . _ , in _ perspectives for new detectors in future supercolliders , roceedings of the workshop of the infn eloisatron project , october 1989 , erice , italy _ , edited by l. cifarelli , r. wigmans , and t. ypsilantis ( world scientific , singapore , 1989 ) , p. 170 . d. bardin _ _ , in _ proceedings of the zeuthen workshop on elementary particle theory - physics at lep200 and beyond , teuplitz , germany _ , edited by t. riemann and j. blmlein ( nucl . suppl . ) , * b37 * , 1994 ) .

suppose that is a linear operator acting on an-dimensional complex hilbert space , and let denote the normalized trace on . set and , and write for the the spectral scale of with respect to . we show that contains full information about , the-numerical range of for each . we then use our previous work on spectral scales to prove several new facts about . for example , we show in theorem 3.4 that the point is a singular point on the boundary of if and only if is an isolated extreme point of . in this case , where is a central projection in in the algebra generated by , and the identity . we show in theorem 3.5 , that is normal if and only if is a polygon for each . finally , it is shown in theorem 5.4 that the boundary of the-numerical range is the finite union of line segments and curved real analytic arcs .

math0107002

the spectral scale was introduced by the present authors and nik weaver in @xcite and further developed by the authors in @xcite . it is defined for any finite set of self - adjoint operators in a finite von neumann algebra . the main theme in @xcite and @xcite is that full spectral information about real linear combinations of such operators is contained in the spectral scale and that much of this information is reflected by the geometry of the spectral scale . in the present paper we view the spectral scale from another perspective and thereby show that full information about the @xmath10numerical range of a finite dimensional operator is also contained in its spectral scale . thus , in this restricted case at least , two rather different sets of data are combined into one three dimensional , compact , convex set . although we restrict consideration in this paper to the finite dimensional world , we shall write this paper using the language of operators ( rather than matrices ) to emphasize how easily many of the concepts generalize to infinite dimensional situations . the notation developed here will be used throughout the rest of the paper . let @xmath0 denote a linear operator on an @xmath1dimensional complex hilbert space @xmath2 and let @xmath3 denote the normalized trace on @xmath4 , the algebra of all linear operators on @xmath2 . write @xmath17 for the subalgebra of @xmath4 generated by @xmath18 and the identity @xmath19 of @xmath4 and set @xmath5 and @xmath20 . the spectral scale was defined in @xcite and @xcite via the map @xmath21 defined by the formula @xmath22 and the spectral scale @xmath23 was defined as @xmath24 it is convenient for the present paper to view the second and third real coordinates as a single complex number . the definition of @xmath21 then becomes @xmath25 and we now define the spectral scale @xmath26 of @xmath0 by the formula @xmath27 thus , we now view @xmath7 as a subset of @xmath28 , rather than as a subset of @xmath29 . as shown in , @xmath30 , where for any self - adjoint subalgebra @xmath31 of @xmath4 , @xmath32 as noted above , the geometry of the spectral scale reflects spectral data for real linear combinations of @xmath33 , i.e. matrix pencil information . information of this sort is widely valued as documented in @xcite the basis for this current work is the observation that @xmath7 also essentially contains the @xmath10-numerical range of @xmath0 for each @xmath34 . since we shall show that several of the currently known properties of the numerical range generalize to the @xmath10numerical range , let us begin by reviewing these properties . recall that the * numerical range * of @xmath0 is by definition @xmath35 we use @xmath36 to denote the spectrum of @xmath0 . the following statements hold . 1 . @xmath37 is a compact , convex subset of @xmath38 . 2 . if @xmath39 and @xmath40 then @xmath41 $ ] . 3 . if @xmath0 is a normal operator , then @xmath37 is the convex hull of its eigenvalues . @xmath37 is a line segment if and only if @xmath42 , where @xmath12 and @xmath43 are complex numbers and @xmath44 is self - adjoint . the boundary of @xmath37 is the union of a finite number of analytic arcs . if @xmath45 and @xmath12 lies in the relative boundary of @xmath37 , then @xmath12 is a reducing eigenvalue of @xmath0 . the boundary of @xmath37 is contained in the real zero set of an algebraic curve . assertion @xmath46 is known as the _ toeplitz hausdorff theorem _ ( see @xcite and @xcite ) . a proof of this and the next 3 assertions can be found in @xcite . the assertion in ( 5 ) was proved independently by agler ( * ? ? ? * theorem 4.1 ) and narcowich ( * ? ? ? * corollary 3.5 ) . finally assertions ( 6 ) and ( 7 ) are due to kippenhahn @xcite . observe that if @xmath37 is a line segment , then it follows from parts ( 2 ) and ( 4 ) of theorem 0.1 that the endpoints of @xmath37 are reducing eigenvalues of @xmath0 . also , @xmath47 if and only if @xmath48 . in fact if @xmath37 is a line segment and @xmath49 is a vector in @xmath2 such that @xmath50 lies on an end point of @xmath37 , then @xmath51 and @xmath52 . to see this , note that by part ( 4 ) of theorem 0.1 , it suffices to show this when @xmath37 is contained in @xmath53 and @xmath54 . in this case we get that @xmath55 and so it has a positive square root . hence , we get @xmath56 and so @xmath57 . the numerical range has a * corner * at @xmath12 if @xmath37 has dimension two and there is more than one tangent line of support for @xmath37 at @xmath12 . the presence of a corner in the numerical range signals the fact that @xmath0 enjoys a special structure . further , it turns out that corners in @xmath37 must be * lineal * in the sense that there are two tangent lines of support which intersect the boundary in line segments of positive length . since we shall show that an analogous results hold for the @xmath10-numerical range , we now discuss this in more detail . recall that a point on the boundary of a convex subset of @xmath58 is said to be * singular * if the the boundary curve is not differentiable at this point . if @xmath37 has dimension two and the boundary of @xmath37 is singular at @xmath12 , then the following statements hold . 1 . @xmath37 has a corner at @xmath12 . 2 . @xmath12 is a reducing eigenvalue for @xmath0 . @xmath12 is lineal . let us begin by presenting a proof of the assertion in ( 1 ) . rotating and translating if necessary , we may assume that @xmath59 , @xmath37 lies in the upper half plane and that the positive imaginary axis intersects the interior of @xmath37 . in this case we may find a convex function @xmath60 defined on an open interval containing 0 whose graph gives a portion of the boundary of @xmath37 which contains 0 . since 0 is a singular point of the boundary , @xmath60 is not differentiable at 0 and since @xmath60 is convex , it follows that @xmath61 thus , the lines though @xmath62 with slopes @xmath63 are tangent to @xmath37 . kippenhahn established assertion ( 2 ) in ( * ? ? ? * satz 13 ) . the proof offered below seems to be new . it is convenient to rotate once more so that @xmath64 and @xmath37 lies in the right half plane so that the corner has the form shown below where both @xmath65 are tangent to @xmath37 at 0 , @xmath66 has positive slope and @xmath67 has negative slope as shown below . .2 in ( 2,2)(-2.5,0 ) ( -1.5,0)(1,0)2 ( -1.3,-1.25)(0,1)2.5 ( -1.3,0)(1,1)1 ( -1.3,0)(1,-1)1 ( .55,-.03)@xmath68 ( -1.33,1.32)@xmath69 ( -.28,1)@xmath66 ( -.28,-1.1)@xmath67 ( -1.34,-.04)@xmath70 ( -1.75,.05)@xmath71 1.8 in since @xmath37 is contained in the right half plane , we get @xmath72 and since there is a unit vector @xmath49 such that @xmath73 , we get @xmath74 by the remark following theorem 0.1 . we may now select @xmath75 so small that @xmath76 is also contained in the right half plane . arguing as above , we get that @xmath77 and so @xmath78 and therefore @xmath79 . hence , @xmath12 must be a reducing eigenvalue for @xmath0 . assertion @xmath80 is due to lancaster ( * ? ? ? * corollary 4 ) . observe that the proof above shows that if @xmath81 is any compact , convex , two dimensional subset of @xmath58 , then the boundary of of @xmath81 is singular at @xmath12 if and only if @xmath81 has a corner at @xmath12 . the @xmath10numerical range of @xmath0 is defined by the formula @xmath82 observe that when @xmath83 we have @xmath84 so that @xmath85 consists of a single point . as the results in this paper make clear it is natural to also include the case where @xmath86 and to define @xmath87 . since we have @xmath88 , this notion is a generalization of the standard numerical range . let us now review the basic properties of the @xmath10-numerical range . if @xmath89 , then the following statements hold . 1 . @xmath9 is a compact , convex subset of @xmath38 . 2 . @xmath90 . we have @xmath91 . if @xmath44 is self - adjoint with eigenvalues @xmath92 , @xmath93 and @xmath94 then @xmath95 $ ] 5 . if @xmath96 denotes the sum of the @xmath10 largest eigenvalues of @xmath15 , then the line @xmath97 is tangent to @xmath9 . the assertion in @xmath46 is due to berger who introduced the @xmath10numerical range in his thesis @xcite . a proof may be found in ( * ? ? ? * problem 167 ) . the assertion in @xmath98 follows from part @xmath46 and a simple calculation . assertion @xmath80 follows from the fact that @xmath99 . the proof of assertion @xmath100 is straight forward . for assertion @xmath101 observe that @xmath102 so that @xmath103 for some real @xmath104 . further , since @xmath96 is the sum of the @xmath10 largest eigenvalues of @xmath15 , if @xmath105 , then @xmath106 . let us now describe our results in more detail . the key to understanding the role played by the @xmath10-numerical range in the spectral scale is the notion of an _ isotrace slice _ of the spectral scale . if @xmath107 , then the * isotrace slice * of @xmath7 at @xmath108 is by definition @xmath109 we prove in theorem 1.3 that if @xmath110 and we define the map @xmath111 from @xmath38 to @xmath28 by @xmath112 , then @xmath111 is an affine map that is a bijection from @xmath9 onto @xmath113 . thus , we may view the @xmath10-numerical range as a subset of the spectral scale . it is also shown in section 1 ( theorem 1.1 ) that the extreme points of @xmath7 lie on the isotrace slices of the form @xmath113 for @xmath114 and so @xmath7 is the convex hull of this finite collection of sets . in section 2 we present some examples ( and pictures ) of various spectral scales . section 3 contains our results on corners in the @xmath10-numerical range . it is shown that several of the known facts about corners on the boundary of @xmath37 generalize to @xmath9 by using the additional structure provided by the spectral scale . for example , in theorem 3.4 , we show that a singularity on the boundary of @xmath9 always occurs at an isolated extreme point of @xmath9 and such points correspond to central projections in the algebra @xmath17 , i.e. reducing subspaces for @xmath0 . we also show that @xmath0 is normal if and only if @xmath9 is a polygon for @xmath115 ( theorem 3.5 ) . sections 4 and 5 are devoted to establishing that the boundary of the @xmath10-numerical range is the finite union of line segments and curved real analytic arcs ( theorem 5.4 ) . in section 4 we review some classical background which is required for our analysis and then use this to derive our results in section 5 . section 6 contains some open question , stated as conjectures . as our results show , the spectral scale provides a new way to study @xmath1-tuples of self - adjoint finite dimensional operators . in the case under study here , when there are just two operators so that @xmath7 is a subset of three dimensional real euclidean space ( or @xmath116 ) , we may actually visualize @xmath7 as shown in the examples and pictures in section 2 . these pictures were created using a matlb program written by jeff duzak as part of an reu research project supervised by the second author . readers can contact the second author for a copy of this program which is quite useful for testing conjectures . the spectral scale has a striking structure in the finite dimensional case under consideration here which we describe in the next proposition . if @xmath49 is an extreme point of @xmath7 , then it lies in an isotrace slice of the form @xmath113 where @xmath114 . since @xmath49 is an extreme point of @xmath7 , it has the form @xmath117 where @xmath118 is a projection in @xmath17 by ( * ? ? ? * theorem 2.3(1 ) ) . if @xmath118 has rank @xmath10 , then @xmath119 and so @xmath120 , where @xmath121 . hence , @xmath122 . for example , if @xmath123 , then extreme points of @xmath7 come from projections of trace @xmath62 , @xmath124 , @xmath125 or @xmath19 . while @xmath126 and @xmath127 are always the single points @xmath62 and @xmath128 , in generic examples @xmath129 and @xmath130 are solid ellipses whose boundaries consist of extreme points of @xmath7 . hence , generically , the boundary of @xmath7 between two successive isotrace slices consists of one dimensional faces . the boundary between @xmath126 and @xmath129 is typically a skewed cone , and , by the symmetry of @xmath7 , the same is true for the boundary of @xmath7 between @xmath130 and @xmath127 . however , as one can see in examples 2.2 and 2.3 below , @xmath7 may have planar faces . a complete description and interpretation of the faces of @xmath7 may be found in @xcite . further , in ( * ? ? ? * ; * ? ? ? * corollary 5.4 ) we showed that @xmath17 is abelian and finite - dimensional if and only if the spectral scale has a finite number of extreme points . thus , in finite dimensions , one can `` see '' that @xmath17 is abelian from the shape of @xmath7 . we now show how the @xmath10-numerical range may be identified with the isotrace slice @xmath113 . this identification depends on a simple convexity result , which we now present . if @xmath131 and @xmath31 is any self - adjoint subalgebra of @xmath4 , then write @xmath132 the extreme points of @xmath133 are precisely the projections of rank @xmath10 . if @xmath118 is a projection of rank @xmath10 , then it an extreme point of @xmath134 and so it is also an extreme point of @xmath133 . for the converse suppose @xmath44 is in @xmath135 , but @xmath44 is not a projection and write @xmath136 for the eigenvalues of @xmath44 . since @xmath137 we get that @xmath138 and since @xmath44 is not a projection we have @xmath139 for at least one index @xmath140 . since @xmath141 , there must also be an index @xmath142 such that @xmath143 . relabeling if necessary , we may assume that @xmath144 . since @xmath145 and @xmath146 we may select @xmath12 and @xmath104 such that @xmath147 now write @xmath148 observe that @xmath149 because @xmath150 . since the diagonal entries of @xmath151 and @xmath152 lie in @xmath153 $ ] and sum to @xmath10 , these matrices are elements of @xmath133 . as @xmath154 there is a real number @xmath108 with @xmath131 , such that @xmath155 . next , note that @xmath156 and therefore @xmath157 . thus @xmath44 is not an extreme point . if @xmath158 and we define the map @xmath111 of @xmath159 into @xmath28 by the formula @xmath160 then @xmath111 is an affine bijection from @xmath9 onto @xmath113 . if @xmath161 , then by part@xmath98 of theorem 0.3 there is a projection @xmath118 with rank @xmath10 such that @xmath162 and the point @xmath163 is in @xmath113 because @xmath164 thus , @xmath165 so , @xmath111 maps @xmath9 into @xmath113 , and @xmath111 is clearly a one - to - one map . now suppose that @xmath166 so that @xmath167 for some @xmath168 . by lemma 1.2 , the krein milman theorem ( * ? ? ? * theorem 2.6.16 ) and ( * ? ? ? * theorem 17.1 ) we have that @xmath44 is a convex combination of projections of rank @xmath10 . since @xmath9 is convex by part @xmath46 of theorem 0.3 and points of the form @xmath169 are in @xmath9 by part @xmath98 , we get that @xmath170 . in this section we describe the spectral scale for four examples . if @xmath171 \dfrac{i}{2}&\dfrac{i}{2 } \endbmatrix,\ ] ] then the nontrivial projections in @xmath17 must have trace @xmath172 and so @xmath173 is the only isotrace slice of @xmath7 that contains nontrivial extreme points . it turns out that @xmath173 is the disk of radius @xmath174 centered at @xmath175 . since @xmath176 the spectral scale in this case is a pair of skewed circular cones joined at their bases as shown below . in this case the extreme points of @xmath7 are @xmath177 and points on the circular boundary of @xmath173 . the one dimensional faces are the line segments joining @xmath62 and @xmath128 to the extreme points on the circle . there are no faces of dimension two . we now present a @xmath178 example where the spectral scale has a `` flat spot '' . write @xmath179 we have that @xmath180)$ ] is a face of @xmath7 by ( * ? ? ? * theorem 2.3 ) ( here if @xmath181 , then @xmath182 = \{a : a^-\le a\le a^+\}$ ] ) . further it follows from the results in @xcite that this face is two dimensional . in fact as can be seen from the figure below , it is diamond shaped . the facial structure of the spectral scale can be quite complex . the next example there are eight faces of dimension two . if we write @xmath183 then the spectral scale is as shown below . although it appears from the figure that the lighter portions between the isotrace slices @xmath184 and @xmath185 are two dimensional faces , in fact they consist of faces of dimension one . this occurred as a result of the matlab shading routine employed to draw this picture . in ( * ? ? ? * example 3.5 ) we showed that in the noncommutative case the spectral scale is not a complete invariant for the operators @xmath15 and @xmath16 by exhibiting two pairs of inequivalent self - adjoint @xmath178 matrices that shared the same spectral scale . in this example each pair generated the full von neumann algebra @xmath4 . in our final example we show that the same phenomenon can occur even if one pair is reducible . write @xmath186 it is clear that for @xmath187 the algebra @xmath17 is the full algebra @xmath4 , while this is not true for @xmath188 because @xmath17 has non - trivial center . on the other hand , the associated spectral scales are equal . this follows from the fact that each of these matrices has the same characteristic polynomial and part @xmath98 of ( * ? ? ? * theorem 3.2 ) . in fact even more is true . write @xmath189 and @xmath190 . a straight forward calculation shows @xmath191 so that even if @xmath17 is irreducible , the associated characteristic polynomial may be reducible . in this section we show that by using the spectral scale we can establish results analogous to theorem 0.1(6 ) and theorem 0.2 for the @xmath10numerical range . let us begin by recalling an idea that proved useful in @xcite . if @xmath49 is an extreme point of a convex set @xmath192 in @xmath193 which is isolated in the set of all extreme points of @xmath192 , then we say that @xmath49 is an * isolated extreme point * of @xmath192 . recall that a face in @xmath7 is said to be a * sharp face * if it is contained in at least two hyperplanes of support ( * ? ? ? * ; * ? ? ? * definition 4.1 ) . we showed in ( * ? ? ? * ; * ? ? ? * lemma 5.2 ) that an isolated extreme point of @xmath7 is always a sharp face . our main result on corners is theorem 3.4 below . the proof of this result requires three technical lemmas which we present next . since the first result is just a slight generalization of lemma 1.2 and uses the notation from that lemma , the proof is somewhat abbreviated . if @xmath44 is an extreme point of @xmath194 , then there is a projection @xmath118 in @xmath17 such that either @xmath195 or @xmath196 where @xmath197 is projection in @xmath17 which is orthogonal to @xmath118 , @xmath198 has dimension one and @xmath199 . fix an extreme point @xmath44 of @xmath194 . if @xmath195 is a projection , then the proof is complete . so , assume that this is not the case , let @xmath200 denote the distinct eigenvalues of @xmath44 and let @xmath201 denote the corresponding eigenprojections . since @xmath44 is not a projection , there is at least one index @xmath140 such that @xmath202 . if there were another index @xmath203 with @xmath143 , then by adding ( resp . subtracting ) and small amount to @xmath204 and subtracting ( resp . adding ) a small amount to @xmath205 we would get two new elements of @xmath206 whose average is @xmath44 so that @xmath44 would not not be an extreme point of this set . hence , @xmath207 and all other indices are 0 or 1 . if @xmath208 did not have dimension one , then it would contain two nonzero orthogonal projections and , arguing as in the previous paragraph , we would again get that @xmath44 is not an extreme point . hence , @xmath44 has the indicated form . in the following lemma it is convenient to return ( briefly ) to the picture of @xmath7 as lying in @xmath29 and then to use some results @xcite which are stated in this framework . us , a corner in @xmath9 corresponds to a corner in @xmath113 . if @xmath81 is a convex subset of @xmath58 with a corner at @xmath12 , then @xmath81 admits an infinite family of tangent lines of support at @xmath12 and these lines all lie in a sector of a disk with maximal angle . we call the tangent lines of support that bound this sector the * sectorial * tangent lines of support . if @xmath10 is an integer with @xmath89 , @xmath113 has dimension two and the point @xmath209 lies at a corner on the isotrace slice @xmath113 , then the following statements hold . 1 . there is a sharp face @xmath210 in @xmath7 of dimension at most one that contains @xmath49 . @xmath211)$ ] , where @xmath212 are central projections . 3 . the points @xmath213 are isolated extreme points of @xmath7 . as @xmath214 has dimension two , there are two distinct sectorial tangent lines of support of @xmath113 in the plane of @xmath113 at the corner @xmath49 . denote these tangent lines by @xmath66 and @xmath67 . each @xmath65 meets b only in the boundary of the isotrace slice @xmath113 and so they are each disjoint from the interior of b. by ( * ? ? ? * corollary 2.4.11 ) , there are distinct planes @xmath215 and @xmath216 such that @xmath217 and each @xmath218 is disjoint from the interior of @xmath7 . since each plane contains @xmath219 , they are planes of support for @xmath7 . thus , if we write @xmath220 , then @xmath210 is a sharp face of @xmath7 . since we are now regarding @xmath7 as a subset of @xmath29 we may use the results in @xcite and @xcite to get that each hyperplane of support is determined by a _ spectral pair _ of the form @xmath221 , where @xmath222 is a real number and @xmath223 is a nonzero vector in @xmath58 . specifically , by ( * ? ? ? * theorem 2.3 ) if @xmath224 is a hyperplane of support for @xmath7 , then there is a spectral pair @xmath221 such that points on@xmath224 satisfy an equation of the form @xmath225 the constant @xmath226 is determined as follows . we write @xmath227 and let @xmath228 and @xmath229 denote the spectral projections of @xmath230 corresponding to the intervals @xmath231 $ ] and @xmath232 . with this , we have @xmath233 observe that @xmath234 is a normal vector for this plane with this , since @xmath215 and @xmath216 are distinct faces of @xmath7 that have nonempty intersection , their normal vectors are linearly independent and so there exist linearly independent spectral pairs @xmath235 and @xmath236 such that each @xmath218 has the equation @xmath237 where @xmath238 . since the point @xmath239 lies in each plane , we get @xmath240 now fix a point @xmath241 on the tangent line @xmath65 and observe that since @xmath113 lies in the plane @xmath242 , we get @xmath243 . since this point also lies in the plane @xmath218 , we get @xmath244 and so , @xmath245 if the vectors @xmath246 and @xmath247 were linearly dependent , then it would follow that the tangent lines @xmath66 and @xmath67 are parallel . since these lines intersect at a corner of the isotrace slice @xmath113 , this is impossible and therefore the @xmath248 s must be linearly independent . applying ( * ? ? ? * ; * ? ? ? * corollary 4.7 ) , we get that @xmath210 has the form @xmath249)$ ] , where @xmath212 are central projections . next , since @xmath210 is a face in @xmath7 , the end points @xmath213 are extreme points of @xmath7 . since the projections @xmath212 are central , the points @xmath213 are isolated extreme points of @xmath7 by ( * ? ? ? * ; * ? ? ? * theorem 5.4(2 ) ) and so assertions @xmath98 and @xmath80 hold . let us now return to our previous notation so that we regard @xmath7 as a subset of @xmath28 . we will use the following to establish @xmath250 in part @xmath80 . if @xmath251)$ ] is any face of @xmath7 that satisfies statements @xmath252 and @xmath80 of lemma 3.2 and @xmath253 , then @xmath254 is an isolated extreme point of @xmath255 . since @xmath256 is a face , its intersection with @xmath255 is a face of @xmath255 , which consists of the single extreme point @xmath254 , by property ( 1 ) and the fact that faces of @xmath7 are transverse to isotrace slices by ( * ? ? ? * ; * ? ? ? * theorem 6.4(1 ) ) . if @xmath257 were not an isolated extreme point of @xmath255 , then there would be a sequence @xmath258 of extreme points in @xmath255 that converges to @xmath254 . since the inverse images of extreme points contain extreme points , each of these points would have the form @xmath259 or @xmath260 , where @xmath261 and @xmath205 are as described in lemma 3.1 . we shall assume that , after passing to a subsequence , @xmath262 and further that @xmath256 is one dimensional . the other cases are handled similarly , but somewhat more easily . since the dimension of @xmath17 is finite , the range of the trace on the projections in @xmath17 is finite and since @xmath263 for each @xmath203 , the @xmath205 s also form a finite set . so , by passing to another subsequence , we may assume that all @xmath264 and @xmath265 . we can now find another subsequence ( which we continue to index with @xmath203 s ) such that @xmath266 converges to @xmath118 and @xmath267 converges to @xmath197 it follows readily that @xmath118 and @xmath197 are orthogonal projections , and @xmath198 has dimension one . since @xmath268 we have @xmath269 and since @xmath270)$ ] is one dimensional , we get that @xmath271 . hence , @xmath272 is a scalar multiple of @xmath14 for @xmath273 by ( * ? ? ? * ; * ? ? ? * lemma 3.3(1 ) ) . since @xmath14 is central , @xmath274 is a multiple of @xmath14 . thus , @xmath275 and so this element is central . but , since @xmath199 this means that both @xmath118 and @xmath197 are central . this is impossible since the central projections are isolated in the set of all projections . hence , @xmath276 must be an isolated extreme point of @xmath255 . with these preparations , we may now present the main theorem of this section . recall that the map @xmath111 from @xmath38 to @xmath28 defined by @xmath277 is an affine map that is a bijection from @xmath9 onto @xmath113 by theorem 1.3 . if @xmath9 has dimension two and if @xmath158 , then the following statements are equivalent . 1 . @xmath12 is a singular point on the boundary of @xmath9 . 2 . @xmath9 has a _ corner _ at @xmath12 . 3 . there is a face @xmath210 of @xmath7 that contains the point @xmath278 which enjoys the properties described below . 1 . @xmath210 is a sharp face in @xmath7 of dimension at most one . 2 . @xmath211)$ ] , where @xmath212 are central projections . the points @xmath213 are isolated extreme points of @xmath7 . there are faces @xmath279 and @xmath280 of dimension two such that @xmath281 and if @xmath282 , then @xmath283 is a line segment for each @xmath140 . in particular , if @xmath66 and @xmath67 denote the sectorial tangent lines of support to @xmath9 at @xmath12 , then their images @xmath284 in @xmath7 intersect the relative interiors of the @xmath285 s . @xmath12 is an isolated extreme point of @xmath9 . as in the proof of @xmath46 in theorem 0.2 and the remark following this theorem , we have that @xmath286 because the boundary of @xmath9 is convex . now suppose that @xmath98 holds . in this case @xmath287 lies at a corner of the isotrace slice @xmath113 and so we may apply lemma 3.2 to get that parts @xmath288 and @xmath289 of part @xmath80 are true . hence , if @xmath290 and @xmath289 hold , then for each @xmath282 , there are faces @xmath291 and @xmath292 in @xmath7 of dimension two such that @xmath293 by ( * ? ? ? * ; * ? ? ? * lemma 6.5(3 ) ) and lemma 3.3 . if @xmath210 has dimension zero so that it is a single point , then we must have @xmath294 and so @xmath250 is true in this case . now suppose that @xmath210 has dimension one , fix @xmath295 in the relative interior of @xmath210 and consider the face @xmath296 . if @xmath297 were a single point , then it would be an extreme point of @xmath7 , which is impossible because it lies in the relative interior of @xmath210 . hence @xmath297 has dimension one . since @xmath297 is contained in @xmath210 we must have @xmath298 . since it is obvious ( and straightforward to prove ) that three distinct faces of dimension two in @xmath7 can not intersect in a face of dimension one , we must have that @xmath291 is either @xmath279 or @xmath280 . hence assertion ( d ) holds in this case . thus , in all cases we get that the faces @xmath285 have dimension two . since such faces are transverse to the isotrace slices of @xmath7 by ( * ? ? ? * ; * ? ? ? * lemma 6.4(1 ) ) , each line @xmath284 intersects @xmath285 in its relative interior . thus , @xmath299 . next , if @xmath80 is true , then it is clear that @xmath12 is an isolated extreme point of @xmath9 and so @xmath300 . if @xmath100 holds , then it is also clear that @xmath46 holds and so these four conditions are equivalent . the operator @xmath0 is normal if and only if @xmath9 is a polygon for @xmath301 . since each @xmath9 is a multiple of the isotrace sliced @xmath113 of @xmath7 , the theorem is equivalent to showing that @xmath7 has a finite number of extreme points if and only if @xmath0 is normal and this is precisely what is asserted in ( * ? ? ? * ; * ? ? ? * corollary 5.6 ) for any two points @xmath302 of @xmath7 with @xmath303 we define the * complex slope * of the segment @xmath304 $ ] to be @xmath305 . this concept will be particularly useful if the segment @xmath306 $ ] is a face @xmath210 of @xmath7 , in which case we call @xmath305 the complex slope of @xmath210 . \(1 ) suppose that @xmath12 lies at a corner of @xmath9 and @xmath65 are the sectorial tangent lines of support for @xmath9 at this point . the faces @xmath279 and @xmath280 described in part ( 3 ) of theorem 3.4 were specifically constructed so that the corresponding line @xmath284 in @xmath307 intersect the relative interiors of the @xmath285 s . if the sharp face @xmath210 has dimension one , then it bounds precisely two faces of dimension two in the boundary of @xmath7 . in this situation , we say that the union of the @xmath285 s is a shelf " of @xmath7 and @xmath210 is the edge " of the shelf . \(2 ) now suppose @xmath308 is a point . in this case @xmath278 is an extreme point of @xmath7 and so it can not be an interior point of any face of @xmath7 . thus , this point is an end point " of any face that contains it . we say that @xmath278 is a mountain peak " in this case . in contrast to the one dimensional case , there are three possibilities for the local geometry of @xmath7 near @xmath278 , which are as follows . \(a ) each face of dimension one that ends at @xmath278 is the edge of a shelf in @xmath7 . \(b ) the point @xmath278 bounds at least one shelf in @xmath7 and at least one face of dimension one that is not the edge of a shelf . \(c ) the point @xmath278 does not bound a shelf . if @xmath17 is abelian so that @xmath7 has a finite number of extreme points , then @xmath278 lies at a mountain peak of type ( a ) . observe that in this case , the faces that intersect the lines @xmath284 do not intersect in a shelf as in remark ( 1 ) above however , their boundaries are edges of shelves formed from one of the @xmath285 s and another two dimensional face of @xmath7 . in most abelian examples the corners of the isotrace slices all lie in shelves and there are no mountain peaks . on the other hand , it is fairly straightforward to construct abelian examples where this type ( a ) phenomenon occurs and it seems possible that there are noncommutative examples of which also display this geometry . examples also show that mountain peaks of type ( b ) can occur . if the boundary of the numerical range is @xmath0 is nonsingular ( i.e. , does not contain any corners ) , then @xmath309 is a mountain peak of type ( c ) . further , it follows from the symmetry of @xmath7 that @xmath128 is also a mountain peak of type ( c ) in this case . we have not been able to construct an example of type ( c ) mountain point at any other points of @xmath7 . \(3 ) if @xmath210 is _ any _ sharp face of dimension one in @xmath7 , then it is the edge of a shelf and if @xmath310 for some @xmath10 , then @xmath161 . to see this , it is convenient to view @xmath7 as a subset of @xmath29 so that we may use the notation developed in @xcite and @xcite . observe that since @xmath210 is a sharp face of dimension one there must be linearly independent spectral pairs @xmath311 such that the corresponding faces @xmath312)\ ] ] have dimension two and contain @xmath210 . further , it follows that the vectors @xmath248 are linearly independent . indeed , if they were linearly dependent , then replacing @xmath313 with a multiple of itself if necessary , we could assume that @xmath314 . ( recall that replacing a spectral pair by a multiple of itself leaves the associated face unchanged ) . in this case since @xmath235 and @xmath315 are linearly independent , we must have that @xmath316 . but then , since these vectors are each normal vectors for @xmath210 we would get that @xmath210 has dimension zero by ( * ? ? ? * ; * ? ? ? * corollary 4.9(2 ) ) , a contradiction . with this we may repeat the remaining portion of the proof of part ( 3 ) of theorem 3.2 verbatim to get that the assertions in part ( 3 ) hold for @xmath210 . \(4 ) now suppose that @xmath317)$ ] is the sharp edge of a shelf in @xmath7 so that @xmath212 are central projections in @xmath17 with @xmath318 . in this case if we write @xmath319 , then we get that @xmath320 for @xmath321 and if we put @xmath322 , then @xmath323 so that @xmath104 is a reducing eigenvalue of @xmath0 . this number is the complex slope " of @xmath210 because @xmath324 we note that if @xmath0 is normal so that every face of dimension one is the sharp edge of a shelf in @xmath7 and the complex slopes of these faces lie in the spectrum of @xmath0 , then these complex slopes fill out the spectrum . a proof of fact this will appear as part of a more general result in a forthcoming paper . \(5 ) if @xmath317)$ ] is the sharp edge of a shelf as in remark ( 4 ) and if the rank of @xmath319 is @xmath325 , then @xmath210 intersects @xmath325 adjacent isotrace slices . thus , in general , the point @xmath278 may be an interior point of the face @xmath210 . however , if the reducing eigenvalues of @xmath0 have multiplicity one , then @xmath210 lies between two adjacent isotrace slices and @xmath278 is one of the end points of @xmath210 and so is an extreme point of @xmath7 . in this section we review the algebraic and analytic facts which can be used to establish the analyticity of the eigenvalues of matrix pencils . ( see 5.3 below for the precise statement of this fact ) . since the material discussed here is classical the presentation is brief . readers may wish to skim this section to grasp the notation introduced here and then read theorem 4.5 , upon which the next section is based . suppose @xmath60 is a complex polynomial of degree @xmath1 of the form @xmath326 since the ring of polynomials @xmath327 $ ] is a unique factorization domain ( * ? ? ? * page 127 , corollary 2 ) and the coefficient of @xmath328 is a nonzero constant there are irreducible polynomials @xmath329 of positive degrees @xmath330 and positive integers @xmath331 such that @xmath332 and @xmath333 we call this factorization the * irreducible decomposition * of @xmath60 in @xmath334 $ ] . recall that if @xmath335 is a polynomial in a single variable of degree @xmath1 and @xmath336 are its roots , then the _ discriminant _ of @xmath118 is by definition @xmath337 thus , @xmath118 has a repeated root if and only if @xmath338 given a polynomial @xmath339 in two variables , we adopt the convention that the roots of @xmath339 are the roots of the polynomial @xmath340 determined by the formula @xmath341 . the discriminant of @xmath60 is the function of @xmath14 defined by @xmath342 the basic properties of @xmath343 are as follows . the following statements hold . 1 . @xmath343 is a polynomial in @xmath69 . 2 . the polynomial @xmath60 has a repeated factor in its irreducible decomposition if and only if @xmath343 is identically zero . this is well known . see ( * ? ? ? * chapter v 10 , corollary , page 138 and proposition 5 , page 139 ) , for example . if @xmath339 is a polynomial of degree @xmath1 such that @xmath344 , where the @xmath345 s are irreducible , then there is a finite set @xmath346 of complex numbers such that if @xmath347 where @xmath348 is held constant , then @xmath349 has @xmath1 distinct roots . since @xmath60 has no repeated factors , its discriminant is not identically zero . since the discriminant is a polynomial in @xmath69 , it has a finite number of roots . let @xmath346 denote the roots of @xmath350 . if @xmath351 , then @xmath352 and so @xmath353 has @xmath1 distinct roots . the roots of @xmath343 are called the * critical points * of @xmath60 . more generally , if @xmath354 is the irreducible decomposition of @xmath60 , then we define the critical points of @xmath60 to be the set of critical points of the * reduced polynomial * @xmath355 . let @xmath339 denote a complex irreducible polynomial of degree @xmath1 as discussed above . an analytic function @xmath12 which satisfies @xmath60 in the sense that @xmath356 for each @xmath14 in the domain of @xmath12 is called an * algebraic analytic function*. such functions were thoroughly analyzed by the first half of the last century . the basic facts of the theory are summarized below . if @xmath60 is an irreducible polynomial of degree @xmath1 and @xmath346 denotes its ( finite ) set of critical points , then the following statements hold . 1 . the equation @xmath357 defines precisely one @xmath1valued function @xmath358 which is analytic on @xmath359 . 2 . if @xmath360 is a closed disk in @xmath361 then the @xmath1 function elements of @xmath210 determine algebraic analytic functions @xmath336 on @xmath360 such that @xmath362 3 . if @xmath363 are closed disks in @xmath361 such that @xmath364 , and @xmath336 are the functions on @xmath365 described in @xmath98 , then each @xmath366 has a unique analytic continuation from @xmath367 to @xmath368 4 . if @xmath369 and @xmath370 encircles a critical point , then when we return to @xmath365 , the same functions are obtained except that they may have been permuted . see ( * ? ? ? * vol ii , chapter 5 ) or @xcite for a complete discussion of algebraic analytic functions . more specifically , part @xmath46 of theorem 4.1 is stated in ( * ? ? ? * , page 121 ) and the remaining parts are proved in ( * ? ? ? * , 15 ) . now suppose @xmath224 is a piecewise linear path consisting of lines joining the critical points and a half - line such as the one shown below . .2 in ( 2,2)(-2.5,0 ) ( -.97,-.03)(1,2).225 ( -.55,.03)(-1,2).195 ( -.55,.03)(1,2).34 ( .245,-.2)(-1,2).455 ( .25,-.2)(1,2).35 ( .9,-.1)(-1,2).3 ( .9,-.1)(1,2).407 ( 1.65,.03)(-1,2).342 ( -.97,-.03)(-1,1)1 ( -1.8,1)(to @xmath371 ) ( .21,-.25)@xmath70 ( .86,-.15)@xmath70 ( -1.01,-.076)@xmath70 ( -.59,-.02)@xmath70 ( 1.6,0)@xmath70 ( -1.03,-.2)@xmath372 ( -.6,-.13)@xmath373 ( .2,-.35)@xmath374 ( 1.6,-.11)@xmath375 ( 0,1)@xmath224 .8 in since @xmath376 is simply connected there are @xmath1 distinct analytic functions @xmath377 on @xmath378 such that @xmath379 by parts @xmath98 and @xmath80 of theorem 5.3 and the monodromy theorem . these results immediately yield yield the following facts about the general case . if @xmath60 is a polynomial of degree @xmath1 such that @xmath380 where each @xmath345 is irreducible with degree @xmath381 , @xmath346 denotes the critical points of @xmath60 and @xmath224 is any path as described above , then there are distinct functions @xmath382 such that : 1 . each @xmath383 is analytic on @xmath384 . we have @xmath385 for each @xmath386 and @xmath387 . 3 . if @xmath388 and @xmath389 , then @xmath390 . we say that functions satisfying property @xmath80 in the theorem above are * completely distinct*. recall that the * pencil * of matrices determined by @xmath15 and @xmath16 is the set of all matrices of the form @xmath391 , for @xmath392 . the theory of matrix pencils has a long and honorable history which includes fundamental contributions by kronecker among others . ( see ( * ? ? ? ii , chapter xii ) , for example ) . in order to analyze the eigenvalues of @xmath393 , write @xmath394 since @xmath15 and @xmath16 are @xmath395 matrices , @xmath60 is a polynomial of total degree @xmath1 . now write write @xmath380 for irreducible decomposition of @xmath60 . if we denote the degree of each @xmath345 by @xmath381 and set @xmath396 , then we may apply theorem 4.4 to get the following result . if @xmath346 denotes the set of critical points of @xmath60 , @xmath224 is a path in @xmath38 as described above , and @xmath397 , then there are @xmath398 completely distinct functions @xmath399 which are analytic on @xmath384 and satisfy @xmath60 . thus , the values of these functions at a point @xmath388 are the @xmath398 distinct eigenvalues of @xmath393 . each of the eigenvalues @xmath400 each has multiplicity @xmath401 . as noted in part 5 of theorem 0.1 , it was shown by agler and narcowich in @xcite and @xcite that if @xmath0 is a compact operator and 0 is in the interior of @xmath37 , then the boundary of @xmath37 is the union of a finite number of analytic arcs . our goal in this section is to derive analogous results for the boundary of the @xmath10-numerical range . we note that the numerical range was also studied by kippenhahn in @xcite , where he showed that the boundary of @xmath37 is an algebraic curve ( * ? ? ? * satz 10 , page 199 ) . the basic idea in the proof of these results , which is the same in all three papers , is as follows . if we first translate so that 0 is in the interior of @xmath37 , then each ray emanating from the origin determines a unique tangent line to the boundary of @xmath37 ; namely the tangent line which is perpendicular to the given ray . it turns out that if @xmath402 denotes the tangent line determined by the ray making the angle @xmath403 with the @xmath68axis , then its distance @xmath404 from the origin is precisely the maximum eigenvalue of @xmath405 . this latter quantity was then analyzed with the characteristic polynomial in the finite dimensional case or by a generalization in the infinite dimensional case . we will show below that this procedure easily generalizes to include the @xmath10numerical range . let us now discuss this in more detail . let @xmath81 denote a compact convex subset of @xmath58 that contains 0 in its interior . if we generalize the notation developed above to apply in this new setting , then we have the following picture . .2 in ( 2,2)(-1.8,0 ) ( -.5,.5)(1,1)(2,0 ) ( -.5,0)(1,0)2 ( 0,-.5)(0,1)2 ( 0,0)(1,2)1 ( .2,1.049)(2,-1)2 ( .15,.07)@xmath403 ( .42,.87)@xmath70 ( .6,.9)@xmath406 ( .7,.3)@xmath81 ( 2.25,0)@xmath402 .8 in thus , @xmath402 is the tangent line to the boundary of @xmath81 which is perpendicular to the ray making the angle @xmath403 with the @xmath68axis , @xmath407 is the intersection point of these lines and @xmath404 is the distance from @xmath407 to the origin . the function @xmath408 is differentiable if and only if @xmath409 further , the points of non differentiability of @xmath404 are in one - to - one correspondence with the line segments in the boundary of @xmath81 . this is basically due to agler @xcite , although it is not explicitly stated there . he observed that if @xmath410 , then @xmath411 ( the formula presented in ( * ? ? ? * ( 4.2 ) ) appears to contain a typographical error since it differs from the one offered above by a minus sign in the second coordinate ) . hence , if @xmath412 , then @xmath404 is differentiable , and if @xmath404 is differentiable , then it follows from @xmath413 that this limit exists . next , agler showed in ( * ? ? ? * lemma 4.2 ) that if @xmath414 where @xmath215 comes from moving counterclockwise around @xmath415 , then @xmath416 hence , if the boundary of @xmath81 contains a line segment with end points @xmath417 , then @xmath418 does not exist and so @xmath404 is not differentiable by the first part of the proof . conversely , if @xmath404 is not differentiable , then @xmath419 does not exist and so @xmath420 since @xmath417 , the line segment joining @xmath215 and @xmath216 lies on the boundary of @xmath81 . now suppose that @xmath421 and @xmath62 is in the interior of @xmath9 . by part@xmath101 of theorem 0.3 , we have that if @xmath96 is the sum of the @xmath10 largest eigenvalues of @xmath422 , then the line @xmath423 is tangent to @xmath9 . if we fix @xmath424 , then applying this argument to @xmath425 we get that the line @xmath426 is tangent to @xmath425 , where @xmath427 is the sum of the @xmath10 largest ( not necessarily distinct ) eigenvalues of @xmath428 if we now rotate back by the angle @xmath403 , we get that the line @xmath402 is at distance @xmath427 from the origin . that is , @xmath429 is tangent to @xmath9 . in other words , @xmath430 in this case . we now record this fact for future reference . if @xmath427 denotes the sum of the @xmath10 largest ( not necessarily distinct ) eigenvalues of @xmath431 , then @xmath430 . now write @xmath432 so that for each fixed pair @xmath433 , @xmath434 is the characteristic polynomial of @xmath435 . thus , the eigenvalues of @xmath431 are precisely the roots of @xmath436 as @xmath437 is a homogeneous polynomial , we may write @xmath438 and set @xmath439 . note that we have the relation @xmath440 also , we have @xmath441 and so we may apply the results described in theorem 4.5 to @xmath60 . we now translate these results to obtain analogous results for the eigenvalues of @xmath431 as follows . if @xmath442 is an eigenvalue for @xmath431 , then we have @xmath443 and so using the relation @xmath444 above , we get that if @xmath445 , then @xmath446 hence , applying theorem 4.5 , we get @xmath447 for some pair @xmath448 . thus , we may write @xmath449 for the eigenvalues of @xmath431 . further , if we select the path @xmath224 in part @xmath80 of theorem 4.5 so that @xmath450 as shown below and @xmath16 are self adjoint , the coefficients of the polynomial @xmath350 are real and so the critical points occur in conjugate pairs ] ( 2,2)(-2.5,0 ) ( -1.5,0)(1,0)3.25 ( -.97,-.01)(1,2).213 ( -.55,.0)(-1,2).208 ( -.55,.0)(5,-2).75 ( .2,-.3)(0,1).6 ( .836,0)(-2,1).6 ( .83,-.02)(1,2).425 ( 1.253,-.75)(0,1)1.6 ( -.95,-.02)(-1,1)1 ( -1.8,1)(to @xmath371 ) ( .16,.28)@xmath70 ( .16,-.343)@xmath70 ( .805,-.04)@xmath70 ( -1.01,-.04)@xmath70 ( -.59,-.04)@xmath70 ( 1.21,.78)@xmath70 ( 1.213,-.8)@xmath70 ( -1.03,-.15)@xmath372 ( -.6,-.15)@xmath373 ( .15,-.44)@xmath374 ( .15,.45)@xmath451 ( 1.21,-.9)@xmath375 ( 1.8,-.06)@xmath452 ( 0,1)@xmath224 .8 in then we get that there is a finite set @xmath453 in @xmath454 , which we may assume contains 0 , such that if @xmath455 and @xmath456 denote successive elements in @xmath453 and we denote the open interval @xmath457 by @xmath458 , then each @xmath459 is real analytic on @xmath458 and the functions @xmath459 are distinct on @xmath458 . the results of this analysis are recorded below . there are positive integers @xmath460 and @xmath331 and a a finite subset @xmath453 of @xmath454 such the following statements hold . 1 . we have @xmath461 . 2 . if @xmath462 , then @xmath431 has precisely @xmath398 distinct eigenvalues , where @xmath463 . if @xmath464 , then @xmath431 has fewer than @xmath398 distinct eigenvalues . 3 . for @xmath462 and @xmath465 , @xmath431 has @xmath381 distinct eigenvalues with multiplicity @xmath466 . there are completely distinct functions @xmath459 for @xmath467 and @xmath468 which are real analytic on @xmath469 and such that @xmath470 is an eigenvalue of multiplicity @xmath471 for @xmath431 . if @xmath472 , then the boundary of the @xmath10numerical range consists of a finite number of line segments and curved real analytic arcs . we use the notation introduced in theorem 4.3 . let @xmath473 denote an open interval in @xmath474 whose end points are adjacent points in @xmath453 . we have then that the functions @xmath459 are real analytic and completely distinct on @xmath473 by part @xmath100 of theorem 5.3 . hence , we may re - index these functions as @xmath475 for @xmath476 and assume that @xmath477 for each @xmath478 . similarly , with additional re - indexing , we may assume that for each fixed @xmath140 the eigenvalue @xmath479 has multiplicity @xmath466 . hence , we may find positive integers @xmath480 and @xmath481 such that @xmath482 and @xmath483 with this we get that @xmath427 , which is the sum of the @xmath10 largest eigenvalues of @xmath431 by theorem 5.2 , is precisely @xmath484 and so the map @xmath485 is real analytic on @xmath473 by theorem 4.3 . since the set @xmath453 is finite , this argument shows that @xmath427 is differentiable on @xmath454 , with the possible exception of a finite subset of @xmath486 . hence , the boundary of @xmath9 contains a finite number of line segments by lemma 5.1 and the remaining boundary points are contained in a finite number of curved real analytic arcs . the spectral scale @xmath487 has a finite number of faces of dimension two . if @xmath210 is a face in @xmath7 of dimension two , then @xmath210 is transverse to the isotrace slices of @xmath7 by ( * ? ? ? * ; * ? ? ? * lemma 6.4(1 ) ) . hence , it must intersect at least one isotrace in a line segment by ( * ? ? ? * ; * ? ? ? * lemma 6.4(1 ) ) . since each isotrace slice contains only a finite number of line segments by theorems 1.3 and 5.4 , @xmath7 can have only a finite number of faces of dimension two . as noted at the beginning of this section , agler , kippenhahn and narcowich each base their analyses on the fact that the tangent line @xmath402 for @xmath37 is determined by the largest eigenvalue of @xmath431 . kippenhahn then continues his analysis in the setting of algebraic geometry . on the other hand , agler uses the theory of several complex variables in his analysis , while narcowich bases his proof on a theorem of nagy . the argument in agler s paper is quite similar to the one presented here . for example , both reduce to a consideration of polynomials without repeated factors and then rely on the discriminant to determine the set @xmath486 . the main difference between the proof above and agler s is that since we are working in finite dimensions , we can base our analysis on the classical results on algebraic analytic functions , rather than delving into the more abstruse world ( from our vantage , at least ) of several complex variables . our contribution consists of the observation that agler s proof may be extended to cover the case where we are working with sums of eigenvalues . nagy s theorem ( ( * ? ? ? * page 376 ) , which is the basis for narcowich s argument may be viewed in this context as an infinite dimensional version of theorem 4.3 . kippenhahn showed that the boundary of the numerical range lies in the zero set of an algebraic curve as follows . he considered @xmath488 , which is a homogeneous polynomial and so defines an algebraic curve in complex projective 3-space . he observed if @xmath404 denotes the maximum eigenvalue of @xmath489 for each fixed @xmath403 , then the formula @xmath490 determines the tangent line @xmath402 described at the beginning of this section . the envelope of these tangents is also an algebraic curve , which is called the * dual curve * to @xmath60 ( * ? ? ? * and proposition 5 , page 253 ) . thus , the boundary of the numerical ranges lies in the zero set of this dual curve . it seems likely that the boundary of the @xmath10-numerical range also lies on an algebraic curve . to show this using kippenhahn s technique requires showing that if @xmath404 denotes the sum of the @xmath10 largest eigenvalues of @xmath431 , then points of the form @xmath491 also lie on an algebraic curve . now it is a standard algebraic that fact that since each eigenvalue of @xmath431 lies on a fixed algebraic curve . this means that for each fixed @xmath403 we may find real numbers @xmath492 and @xmath493 such that point@xmath494 lies on this curve . but it seems possible that we would have @xmath495 for most @xmath403 s . conjecture 6.1 . if @xmath496 and @xmath497 is a corner of @xmath9 , then @xmath498 lies on a shelf in b. charles a. akemann , joel anderson and nik weaver , _ a geometric spectral theory for @xmath1-tuples of self - adjoint operators in finite von neumann algebras _ , j. functional analysis , * 164 * ( 1999 ) , 258 - 292 .

we carry out numerical simulations of vesicle formation based on the density functional theory for block copolymer solutions . it is shown by solving the time evolution equations for concentrations that a polymer vesicle is spontaneously formed from the homogeneous state . the vesicle formation mechanism obtained by our simulation agree with the results of other simulations based on the particle models as well as experiments . by changing parameters such as the volume fraction of polymers or the flory - huggins interaction parameter between the hydrophobic subchains and solvents , we can obtain the spherical micelles , cylindrical micelles or bilayer structures , too . we also show that the morphological transition dynamics of the micellar structures can be reproduced by controlling the flory - huggins interaction parameter .

cond-mat0701187

amphiphilic block copolymers , which consists of hydrophilic subchains and hydrophobic subchains , are known to form micellar structures , such as spherical micelles , cylindrical micelles , and vesicles @xcite . to clarify how these structures are self - organised is a basic problem of the kinetics of micellar systems . several computer simulations have been available to study formation of a vesicle based on the particle models ( brownian dynamics ( bd ) @xcite , dissipative particle dynamics ( dpd ) @xcite , and molecular dynamics ( md ) @xcite ) . the spontaneous vesicle formation process from the homogeneous state observed in these simulations is as follows : the amphiphilic block copolymers aggregate into small spherical micelles rapidly from the homogeneous initial state . the spherical micelles grows to larger micelles by collision ( cylindrical micelles , open disk - like micelles ) . the large disk - like micelles finally close up and form vesicles . the micelle growth process and the closure process are slower than the first spherical micelle formation process . hereafter we call this process the `` mechanism i '' ( see also figure [ vesicle_formation_mechanisms](a ) ) . note that the mechanism i is also supported by monte carlo simulations @xcite or experiments of lipid systems @xcite . strictly speaking , the mechanism obtained by the experiments for lipid systems may be different from one for polymer systems . however , since the dpd simulations for polymer systems and the bd and md simulations for lipid systems give similar results ( mechanism i ) , we believe that the vesicle formation mechanism is common for polymer systems and lipid systems . we also note that the similar mechanism obtained for polymer systems by the experiments of morphological transition of cylindrical micelles to vesicles @xcite . the field theoretical approach , which has been developed to study mesoscale structures of block copolymers @xcite , is considered to be useful for vesicle formation , but most of the works have been limited to simulations in thermal equilibrium @xcite . since these simulations ignore realistic kinetics ( for example , local mass conservation is not satisfied ) , the vesicle formation process observed in these simulations are different from the mechanism i : the first process is similar to the mechanism i. small spherical micelles are formed rapidly . the spherical micelles then grow up to large spherical micelles by the evaporation - condensation like process . the large spherical micelles are energetically unfavourable , and thus the large micelles take the solvents into them to lower the energy . we call this process the `` mechanism ii '' ( see also figure [ vesicle_formation_mechanisms](b ) ) . the dynamical simulations for diblock copolymer solutions were carried by sevink and zvelindovsky @xcite , using the dynamic self consistent field ( scf ) theory . however , in their simulations both subchains are hydrophobic and the resulting structures are so - called onion structures @xcite . the onion structures obtained by simulations are essentially microphase separation structure in the block copolymer rich droplets in the solvents , and qualitatively different from the multilayer vesicle structures which is often called as `` onion structures '' in surfactant solutions . the vesicle structures contain solvents inside them and therefore different from these polymer onion structures , and one should distinguish the onion formation process from the vesicle formation process . we also note that the simulations for onion structures were carried out for weak segregation region and this does not agree with previous equilibrium simulations for vesicles , since vesicles are observed in rather strong segregation region . ( in this work , we use the word `` strong segregation region '' as the region where the minimum and maximum values of the density field is approximately @xmath0 and @xmath1 . one may call such region as the intermediate segregation region . ) it should be noted that the formation processes of the onion structures by simulations @xcite are mainly the lamellar ordering process in droplet like regions . therefore the resulting morphologies ( onions ) are similar to the phase separation patterns in droplets @xcite than vesicles . thus we consider that the onion formation process is not the mechanism i. recently , dynamical simulations of vesicle formation have been done by he and schmid @xcite , using the external potential dynamics ( epd ) @xcite . they obtained vesicles , but their formation process is similar to the mechanism ii . unfortunately the mechanism ii is qualitatively different from the mechanism i , and this means that their result does not agree with particle simulations or experiments ( we will discuss the reason for this difference in section [ comparison_with_epd_simulations ] ) . we expect that the mechanisms observed by different simulation methods should be the same . in the present work , we apply our previous model , the density functional theory for block copolymers @xcite , to the dynamics . we carry out numerical simulations for amphiphilic diblock copolymer solutions in three dimensions . by using the continuous field model , we show , for the first time , that a vesicle is spontaneously formed from a disordered uniform phase . the simulation result is consistent with the mechanism i. we also show that we can simulate the spontaneous formation process of various micellar structures ( such as spherical micelles or cylindrical micelles ) and the morphological transition dynamics . the dynamics of block copolymer systems are well studied by using the dynamic scf theory @xcite as well as the time - dependent ginzburg - landau ( tdgl ) equation @xcite . the dynamic scf simulations are known to be accurate for from weak segregation region to strong segregation region , but they consume memory and need large cpu power . in contrast , the tdgl simulations need less cpu power and enables large scale simulations . however , the free energy functionals @xcite used in the tgdl approach is generally not appropriate for the strong segregation region , since the validity of the ginzburg - landau ( gl ) expansion is guaranteed only for the weak segregation region where the density fluctuation is sufficiently small @xcite . actually maurits and fraaije @xcite showed that the widely used fourth - order gl expansion model is not sufficient for dynamical simulations . to overcome this limitation we need to use free energy functional model such as the flory - huggins - de gennes - lifshitz free energy @xcite , which is not expressed in the gl expansion form . in previous simulations , vesicles are observed for rather strong segregation region in diblock copolymer solutions . thus we can conclude that the use of inappropriate free energy functional models qualitatively affects micellar structures in diblock copolymer solutions . we need to use a free energy model and a dynamic equation which is valid for the strong segregation region and for the macrophase separation . in our previous work @xcite we proposed the free energy functional which is valid for strong segregation region , that is , valid for vesicles @xcite . the free energy functional for the system can be expressed as follows by using the soft - colloid picture @xcite . @xmath2 where @xmath3 and @xmath4 are the entropy functional and the interaction energy functional , and @xmath5 is the temperature . the entropy functional can be decomposed into two parts ; the translational entropy functional @xmath6 and the conformational entropy functional @xmath7 . in the density functional theory , @xmath4 , @xmath6 and @xmath7 are expressed as the functional of the local volume fraction fields . ( the local volume fraction field is equivalent to the local density field , if the segment volume is set to unity . we assume that the segment volume is unity in this work . ) each contributions to the free energy functional can be modelled for ab type diblock copolymer and solvent mixtures by @xcite @xmath8 @xmath9 @xmath10\ ] ] where @xmath11 is the boltzmann constant , @xmath12 is the local volume fraction of the @xmath13 segment at position @xmath14 , @xmath15 is the order parameter field ( @xmath16-field ) defined as @xmath17 @xcite . the coefficients @xmath18 and @xmath19 are constants determined from the block copolymer architecture ( we do nt show the explicit forms of @xmath18 and @xmath19 here ; they can be found in refs @xcite ) , @xmath20 is the block ratio of the block copolymer , @xmath21 is the kuhn length , @xmath22 is the flory - huggins @xmath23 parameter , and @xmath24 is the green function which satisfies @xmath25 \tilde{\mathcal{g}}(\bm{r } - \bm{r } ' ) = \delta(\bm{r } - \bm{r}')$ ] @xcite , where @xmath26 is a cutoff length and is about the size of microphase separation structures . @xmath27 is the lagrange multiplier for the incompressible condition ( @xmath28 ) @xcite . the conformational entropy is expressed in the bilinear form of @xmath16-field @xcite . one significant property of the conformational entropy is that it satisfies the following relation . @xmath29 = \alpha s_{\text{conf}}\left [ \ { \phi_{i}(\bm{r } ) \ } \right]\ ] ] where @xmath30 is arbitrary positive constant . ( it is clear that the conformational entropy in the bilinear form of @xmath16 satisfies eq . ) we can interpret eq as the extensivity of the conformational entropy ; under the current approximation ( using mean field , and neglecting many body correlations ) , the total conformational entropy is sum of the conformational entropy of each polymer chains . since the density profiles of miecellar structures are determined to achieve the low free energy , the conformational entropy which does not satisfy eq may lead unphysical density profile . we note that the lifshitz entropy @xcite and the conformational entropy calculated by the scf theory satisfy eq , while most of previous phenomenological free energy models do not satisfy eq . substituting eqs - into eq , we get @xmath31 \end{split}\ ] ] the difference between the free energy functional and our previous model @xcite is the form of the green function ( in the previous theory , the green function @xmath32 has no cutoff , @xmath33 ) . a cutoff for the green function was first introduced by tang and freed @xcite , without derivation , for the ohta - kawasaki type standard gl free energy @xcite . the ohta - kawasaki type coulomb type long range interaction term may cause unphysical interaction or correlation in non - periodic systems such as block copolymer solutions . the true form of long range interaction can be , in principle , obtained by calculating higher order terms in free energy functional ( the effect of higher order vertex functions ) exactly . while we can calculate the higher order terms numerically by using the scf , it is practically impossible to get analytical form of them and take a summation . to avoid the unphysical correlation without numerically demanding calculations , in the present study , we employ a theory which describes more precisely the properties of the length scale of the order of one polymer chain @xcite ; we modify the coulomb type interaction by introducing tang - freed type cutoff @xcite , instead of calculating higher order terms exactly . it is reasonable to consider that the interaction range ( or the cutoff length @xmath26 ) of the long range interaction can not exceed the characteristic size of the polymer chain . generally it is difficult to determine the cutoff length ( one way to determine it is to use the scf calculation or other microscopic calculations such as the md ) . here we estimate the cutoff length for the simplest case , diblock copolymer melts . for homogeneous ideal state , the cutoff length is considered to be the mean square end to end distance of a polymer chain @xmath34 where @xmath35 is the polymerization index . on the other hand , for microphase separation structures at the strong segregation region , the polymer chains are strongly stretched . the periods of the structures are known to be proportional to @xmath36 @xcite . thus we have the following polymerization index dependence for the cutoff length of diblock copolymer melts . we assume that the micellar structures have the similar cutoff length as the melt case . in this work we use @xmath38 for micellar systems . note that this is a rough estimation and the validity is not guaranteed . for example , the proportional coefficient in eq is ignored here , or the effect of swelling of the hydrophilic subchain in the solution is not taken into account . therefore the validity of this value of @xmath26 should be tested by comparing with the characteristic scale of the resulting phase separation structure . we also note that the dependence of the morphologies to the value of @xmath26 is not so sensitive ( see also appendix [ cutoff_length_dependency ] ) . what important here is that the interaction range is not infinite ( as the original ohta - kawasaki green function ) but finite ( as the modified green function by tang and freed ) . we employ the stochastic dynamic density functional model @xcite for the time evolution equation . the diblock copolymer solutions are expressed as three component systems ( hydrophilic subchain @xmath39 , hydrophobic subchain @xmath40 , and solvent @xmath3 ) . the dynamic equation for multicomponent systems can be expressed as follows . @xmath41 + \xi_{i}(\bm{r},t)\ ] ] where @xmath12 is the concentration of @xmath13-th component ( @xmath42 ) and @xmath43 is a friction coefficient . @xmath44 is the free energy . @xmath45 is the thermal noise which satisfies the fluctuation - dissipation relation . @xmath46 \delta(t - t')\end{aligned}\ ] ] where @xmath47 represents the statistical average and @xmath48 is the constant determined from the degree of coarse graining ( see ref @xcite for detail ) . one can interpret that the temperature of the noise is the effective temperature @xmath49 , instead of the real temperature @xmath5 . we also note that @xmath50 can be generated easily by the algorithm proposed by van vlimmeren and fraaije @xcite . it should be noted here that the free energy functional in eq is generally some kind of effective free energy functional , and is not equal to the free energy functional for the equilibrium state @xcite . in this work we approximate the effective free energy functional as the equilibrium free energy functional . this approximation corresponds to assume that all the polymer chains are fully relaxed , and thus this approximation neglects the viscoelasticity associated with the relaxation of polymer chains . we can interpret the dynamic equation as the tdgl equation . in this case , the onsager coefficients ( or the mobility ) @xmath51 is proportional to @xmath12 . this is most important for the strong segregation region or in the situation that the solute concentration is sufficiently small @xcite . here we note that the tdgl equation with density dependent mobility and the free energy of ideal gases ( the translational entropy of ideal gases ) reduces to the diffusion equation ( or the smoluchowski equation ) @xcite . this suggests that we should use the flory - huggins ( or bragg - williams ) type free energy model once we employed the density dependent mobility . similarly we should use the density dependent mobility if we employ the flory - huggins type free energy model . it is also noted that the tdgl equation with the constant mobility and the flory - huggins type free energy leads unphysical singular behaviour near @xmath52 . for simplicity , we assume that all the segments have the same friction coefficient ( @xmath53 ) . we can rewrite eq by using @xmath54 as follows . @xmath55 \right ] + \xi_{i}(\bm{r},t ) \\ & = \frac{k_{b } t}{2 \zeta } \left [ \psi_{i}(\bm{r } ) \nabla^{2 } \mu_{i}(\bm{r } ) - \mu_{i}(\bm{r } ) \nabla^{2 } \psi_{i}(\bm{r } ) \right ] + \xi_{i}(\bm{r},t ) \\ & = \frac{k_{b } t}{2 \zeta } \left [ \psi_{i}(\bm{r } ) \nabla^{2 } \mu_{i}(\bm{r } ) - \mu_{i}(\bm{r } ) \nabla^{2 } \psi_{i}(\bm{r } ) + \frac{2 \zeta}{k_{b } t } \xi_{i}(\bm{r},t ) \right ] \end{split}\ ] ] where @xmath56 is a kind of chemical potential field . it is noted that @xmath57 is not singular at @xmath58 and thus we can perform stable simulations ( on the other hand , @xmath59 has a singularity and is numerically unstable ) . by introducing rescaled variables defined as @xmath60 and substituting eqs and into eqs , and , we obtain the dynamic equation for rescaled variables @xmath61 and the fluctuation dissipation relation for the rescaled noise field @xmath62 \delta(\tilde{t } - \tilde{t}')\end{aligned}\ ] ] it should be noted here that the magnitude of the noise @xmath63 depends only on @xmath64 . the temperature change corresponds to the change of the time scale ( since @xmath65 ) and the change of the @xmath23 parameter . however , notice that to return from the to rescaled time @xmath66 ( which is used in the actual simulations ) to the real time scale @xmath67 , we have to multiply the factor @xmath68 . we note that we can set @xmath69 instead of introducing rescaled variables ( @xmath66 and @xmath70 ) . this corresponds to setting the effective diffusion coefficient for the monomer to unity ( strictly speaking , the half of the effective diffusion coefficient is set to unity ) . this change can be done without losing generality , as shown above . to return to the real time scale , we have to multiply the factor @xmath71 to the rescaled time . this factor can be estimated from the experimental diffusion data . ( we will estimate the real time scale in section [ comparison_with_experiments ] . ) it should be noted here that eqs and implies that the annealing process ( the temperature change process ) corresponds to the change of the @xmath23 parameter , and the magnitude of the noise is not changed because there are no other parameters related to @xmath5 . the magnitude of the noise is characterized only by @xmath64 in the rescaled units . we solve eq numerically in three dimensions . the chemical potential @xmath57 can be calculated from the free energy functional whereas thermal noise @xmath45 is calculated by the van vlimmeren and fraaije s algorithm @xcite . simulations are started from the homogeneous state ( @xmath72 , where @xmath73 is the spatial average of @xmath12 ) . each step of the time evolution in the simulation is as follows ; 1 . [ simulation_firststep ] calculate the @xmath16-field @xmath15 and the chemical potential field @xmath57 from the density field @xmath12 . 2 . generate the thermal noise @xmath63 . 3 . calculate the lagrange multiplier @xmath27 . [ density_time_evolution ] evolve the density field @xmath12 by time step @xmath74 , using @xmath75 and @xmath63 . return to the step [ simulation_firststep ] . here we describe the numerical scheme for the simulations in detail . as mentioned above , the dynamic density functional equation is reduced to the diffusion equation for the flory - huggins type translational entropy functional . in our free energy , we have the flory - huggins translational entropy term . thus the dynamic equation contains the standard diffusion term ( laplacian term ) . @xmath76 where we set @xmath77 ( @xmath78 ) and @xmath79 . the numerical stability of the diffusion type equation can be improved by the implicit schemes . in the previous work @xcite we employed the alternating direction implicit ( adi ) scheme @xcite for the laplacian term . in this work we employ the adi scheme to improve stability , as the numerical scheme for the equilibrium simulations . we split the each time evolution steps ( time step @xmath80 ) into three substeps ( time step @xmath81 ) . the update scheme for the density field from @xmath82 to @xmath83 is as follows . ( for simplicity , here we use the continuum expression for the position @xmath14 and the differential operators @xmath84 and @xmath85 . in real simulations we use the standard lattice for the position and the center difference operators for the differential operators @xcite ) @xmath86 \phi^{(1)}_{i}(\bm{r } ) = & \ , \phi^{(0)}_{i}(\bm{r } ) + \frac{\delta\tilde{t}}{3 } \bigg [ - 2 \tilde{c}_{i } \frac{\partial^{2}}{\partial x^{2 } } \phi^{(0)}_{i}(\bm{r } ) + \psi^{(0)}_{i}(\bm{r } ) \nabla^{2 } \mu^{(0)}_{i}(\bm{r } ) \\ & \qquad \qquad \qquad - \mu^{(0)}_{i}(\bm{r } ) \nabla^{2 } \psi^{(0)}_{i}(\bm{r } ) + \tilde{\xi}^{(0)}_{i}(\bm{r},\tilde{t } ) \bigg ] \end{split } \\ \label{update_scheme_adi_y } \begin{split } \left [ 1 - \frac{2 \delta\tilde{t}}{3 } \tilde{c}_{i } \frac{\partial^{2}}{\partial y^{2 } } \right ] \phi^{(2)}_{i}(\bm{r } ) = \ , & \phi^{(1)}_{i}(\bm{r } ) + \frac{\delta\tilde{t}}{3 } \bigg [ - 2 \tilde{c}_{i } \frac{\partial^{2}}{\partial y^{2 } } \phi^{(1)}_{i}(\bm{r } ) + \psi^{(1)}_{i}(\bm{r } ) \nabla^{2 } \mu^{(1)}_{i}(\bm{r } ) \\ & \qquad \qquad \qquad - \mu^{(1)}_{i}(\bm{r } ) \nabla^{2 } \psi^{(1)}_{i}(\bm{r } ) + \tilde{\xi}^{(1)}_{i}(\bm{r},\tilde{t } ) \bigg ] \end{split } \\ \label{update_scheme_adi_z } \begin{split } \left [ 1 - \frac{2 \delta\tilde{t}}{3 } \tilde{c}_{i } \frac{\partial^{2}}{\partial z^{2 } } \right ] \phi^{(3)}_{i}(\bm{r } ) = \ , & \phi^{(2)}_{i}(\bm{r } ) + \frac{\delta\tilde{t}}{3 } \bigg [ - 2 \tilde{c}_{i } \frac{\partial^{2}}{\partial z^{2 } } \phi^{(2)}_{i}(\bm{r } ) + \psi^{(2)}_{i}(\bm{r } ) \nabla^{2 } \mu^{(2)}_{i}(\bm{r } ) \\ & \qquad \qquad \qquad - \mu^{(2)}_{i}(\bm{r } ) \nabla^{2 } \psi^{(2)}_{i}(\bm{r } ) + \tilde{\xi}^{(2)}_{i}(\bm{r},\tilde{t } ) \bigg ] \end{split}\end{aligned}\ ] ] in each steps , the chemical potential @xmath87 and @xmath88 is regenerated . the numerical scheme for the calculation of the chemical potential @xmath87 is found in ref @xcite . eqs - can be solved easily by using the numerical scheme for the cyclic tridiagonal matrix @xcite . we need to calculate @xmath89 to calculate the chemical potential @xmath87 . because of the thermal noise and numerical error , the condition @xmath90 is not always satisfied . thus we put @xmath89 as @xmath91 this avoids numerical difficulty associated with negative value of @xmath92 . the noise field @xmath93 is calculated at each adi time steps , by using the scheme described in appendix [ noise_generation_scheme ] ( notice that the size of the time step here is @xmath81 instead of @xmath80 ) . the lagrange multiplier @xmath27 is also updated at each adi time steps by the following scheme . because of the large thermal noise , the incompressible condition is not always satisfied . thus we use roughly approximated and relatively simple scheme to calculate @xmath27 . if we ignore the terms in the chemical potential except for the lagrange multiplier term , the adi update scheme can be approximately written as @xmath94\ ] ] from eq we have @xmath95 \\ \frac{1}{\delta\tilde{t } / 3 } \left [ \sum_{i } \phi^{(n + 1)}_{i}(\bm{r } ) - \sum_{i } \phi^{(n)}_{i}(\bm{r } ) \right ] & \approx \nabla^{2 } p(\bm{r } ) - \nabla \cdot \left [ \left [ 1 - \sum_{i } \phi^{(n)}_{i}(\bm{r } ) \right ] \nabla p(\bm{r } ) \right ] \end{split}\ ] ] using the constraint @xmath96 and assuming @xmath97 , finally we have @xmath98\ ] ] eq and the long range interaction terms ( which contains the green function @xmath24 ) in the chemical potential @xmath57 can be calculated by several numerical methods for partial differential equations . here we calculate them by the direct method , using fftw3 @xcite . parameters for the diblock copolymer are as follows : polymerization index @xmath99 , block ratio @xmath100 , the spatial average of the volume fraction @xmath101 . the cutoff length for the long range interaction should be set appropriately . if there are no solvent diblock copolymers form strongly segregated microphase separation structure ( currently we are interested in the strong segregation region ) . thus from eq we have @xmath102 . here we set @xmath103 . the validity of this value is argued later . ( see also appendix [ cutoff_length_dependency ] . ) parameters for the solvent are as follows : the spatial average of the volume fraction @xmath104 . flory - huggins @xmath23 parameters are @xmath105 ( the @xmath39 monomer is hydrophilic and the @xmath40 monomer is hydrophobic ) and the kuhn length is set to unity ( @xmath106 ) . all the simulations are carried out in three dimensions . the size of the simulation box is @xmath107 and the number of lattice points is @xmath108 . we apply the periodic boundary condition . the time step is set to @xmath109 and the magnitude of noise is @xmath110 . the simulation is carried out up to @xmath111 time steps ( from @xmath112 to @xmath113 ) and it requires about @xmath114 days on a @xmath115 xeon work station . the snapshots of dynamics simulations are shown in figure [ vesicle_formation_snapshots ] . the observed vesicle formation process is different from the mechanism ii , which is observed in the previous thermal equilibrium simulations @xcite and epd simulations @xcite . first , spherical micelles are formed ( figure [ vesicle_formation_snapshots](a ) ) . then the micelles aggregate and become large ( figure [ vesicle_formation_snapshots](b ) ) , and a disk - like micelle ( figure [ vesicle_formation_snapshots](c ) ) appears . the disk - like micelle is closed ( figure [ vesicle_formation_snapshots](d),(e ) ) to form a vesicle ( figure [ vesicle_formation_snapshots](f ) ) . this vesicle formation process agrees with the mechanism i , which is observed in the particle simulations @xcite . to confirm that our vesicle formation is independent of the random seed , next we perform simulations with different random seeds . we peform four simulations here and vesicles are obtained in two simulations . figures [ vesicle_formation_snapshots2 ] and [ vesicle_formation_snapshots3 ] show the snapshots of vesicle formation processes obtained for different random seeds ( all other parameters are the same as the previous simulation ) . the vesicle formation processes in figures [ vesicle_formation_snapshots2 ] and [ vesicle_formation_snapshots3 ] again agree with the mechanism i. while the vesicles are not observed for all the simulations , but we consider that the vesicle formation process obtained by the previous simulation is not artifact . the characteristic size of the phase separation structure should be compared with the cutoff length @xmath26 . the characteristic size of the micellar structure can be found in the two dimensional cross section data or the density correlation function . figure [ cross_section_density_correlation](a ) and ( b ) show the two dimensional cross section of the final structure ( @xmath113 , figure [ vesicle_formation_snapshots](f ) ) . from figure [ cross_section_density_correlation](a ) and ( b ) we can observe that the cutoff length @xmath103 is comparable to the characteristic size of the micelle bilayer structure . figure [ cross_section_density_correlation ] ( c ) shows the radial averaged correlation function of the density field in the fourier space @xmath116 , defined as @xcite @xmath117^{2 } } { \displaystyle \sum_{q - \delta / 2 \le |\bm{q}'| < q + \delta / 2 } 1}\ ] ] where @xmath118 is the fourier transform of the density field and @xmath119 is the width of the shell in the fourier space ( here we set @xmath120 ) . the correlation functions have peaks at @xmath121 . thus the characteristic wave length for the corona ( the @xmath39 subchain rich region ) and the core ( the @xmath40 subchain rich region ) is @xmath122 . while the resolution ( the shell width @xmath119 ) is not fine , the characteristic size of the structure is comparable with the value of the cutoff length , @xmath103 . therefore we consider that the value of @xmath26 used in this simulation is appropriate and valid . ( see also appendix [ cutoff_length_dependency ] ) . figure [ structure_volume_fraction_dependence ] shows various micellar structures obtained from simulations for different volume fractions @xmath123 and @xmath124 . simulations are performed up to @xmath125 time steps ( from @xmath112 to @xmath113 ) . all other parameters are the same as those in figure [ vesicle_formation_snapshots ] . if the volume fraction of the polymer is small ( figure [ structure_volume_fraction_dependence](a),(b ) ) only small micelles ( spherical micelles in [ structure_volume_fraction_dependence](a ) and disk - like micelles in [ structure_volume_fraction_dependence](b ) ) are formed . on the other hand , if the volume fraction of the polymer is large ( figure [ structure_volume_fraction_dependence](c),(d ) ) bilayer structures are formed . vesicles in figure [ vesicle_formation_snapshots ] are observed only for the intermediate volume fraction . in experiments , the mixed solvent ( mixture of organic solvents and water ) is often used to control the morphology of the micellar structures of block copolymers @xcite . if we assume that the mixed solvent is the mixture of common solvent ( @xmath126 and @xmath127 is sufficiently small or negative ) and water , the volume fraction of water mainly affect the interaction between the hydrophobic subchain and the solvent . we change @xmath127 to mimic these experiments . figure [ structure_chi_parameter_dependence ] shows the results of simulations for various @xmath127 ( @xmath128 and @xmath129 ) . we can observe the spherical micelles are formed for low @xmath127 and cylindrical micelles are formed for high @xmath127 . this agree with the experimental results qualitatively . shen and eisenberg @xcite reported that by controlling solvent condition , one can control the morphology of the block copolymer micelles . the control of the morphology is especially important for application use such as the drug delivery system . here we mimic the solvent condition change by changing the @xmath23 parameter between the hydrophobic segment and the solvent , @xmath127 . we change the @xmath23 parameter at @xmath113 ( figure [ vesicle_formation_snapshots](f ) ) , from @xmath130 to @xmath131 or @xmath132 , and perform the simulations up to @xmath133 . the snapshots at @xmath133 are shown in figure [ structural_transition ] . the vesicle is changed into spherical micelles ( figure [ structural_transition](a ) ) or cylindrical micelles ( figure [ structural_transition](b ) ) . the resulting morphologies are consistent with the morphologies obtained by the simulations started from the homogeneous state ( figure [ structure_chi_parameter_dependence](b ) and figure [ structure_chi_parameter_dependence](c ) ) . from the dynamics simulations based on the density functional theory for amphiphilic diblock copolymers , we could observe the time - evolution of spontaneous formation of vesicles . we also observed the micellar structures depending on the volume fraction of polymers . the observed vesicle formation process is just the same as the results of the previous particle model simulations ( mechanism i ) . at the initial stage of time evolution , rapid formation of small spherical micelles is observed . after the small micelles are formed , they aggregate each other by collision and become larger micelles . finally the large disk - like micelles are closed and become vesicles . the late stage process , collision process and the close - up process , is slower than the initial stage , as the mechanism i. there are several important differences between our model and the standard tdgl model for the weak segregation limit . here we mention some essential properties of our model . first , our model can be applied for the strong segregation region , at least qualitatively . this is especially important , since the vesicles are observed in the strong segregation region . second , the interaction between the hydrophilic subchain and solvent should be handled carefully . if the hydrophilic interaction ( @xmath126 ) is too small or both subchains are hydrophobic , the system undergoes macrophase separation and separates into block copolymer rich region and solvent rich region . in such a situation , vesicles are not formed and onion structures @xcite are formed instead . third , we applied rather large thermal noise to the system . recently zhang and wang @xcite have shown that the glass transition temperature and the spinodal line ( stability limit of the disordered phase ) for the microphase separation in diblock copolymer melts are quite close . this implies that microphase separated structures in the block copolymer systems are intrinsically glassy at the strong segregation region . without sufficiently large and realistic thermal noise , the system is completely trapped at intermediate metastable structures ( in most cases , spherical micelles ) and thus thermodynamically stable structures ( vesicles ) will never be achieved . as mentioned above , the magnitude of the noise is determined by the characteristic time scale and the factor @xmath64 is independent of the temperature . thus the noise is important for mesoscopic systems even if the temperature of the system is not high . here we discuss about the thermal noise in detail . intuitively , the large thermal noise is needed to overcome the free energy barrier . in the kinetic pathway of usual macrophase separation processes ( for example , phase separations in homopolymer blends ) , there are not so large free energy barriers . therefore , in most cases , the simulations for such systems work well with small thermal noise or without thermal noise . on the other hand , in the kinetic pathway of vesicle / micelle formation processes , there are various large free energy barriers ( for example , in the collision and coalescence process of spherical micelles ) . however , it is not clear and established how to determine the magnitude of the thermal noise for such systems . as mentioned , the thermal noise in eq is expressed by using the effective temperature @xmath49 , instead of the real temperature @xmath5 and @xmath64 is determined from the degree of coarse graining . intuitively this means that the coarse graining procedure decreases the degree of the freedom of the field . the similar approach can be found in the van vlimmeren and fraaije s theory @xcite . they employed the noise scaling parameter @xmath134 to express the effective degree of freedom in the simulation cell . these two approaches are similar and in fact , we can write the relation between them . @xmath135 where @xmath136 is the lattice vector . from eq we can calculate the noise scaling parameter for our simulation . using the parameters used in section [ results ] , we have @xmath137 . it should be noticed that this value of the noise scaling parameter is much smaller than the values used in most of previous works ( @xmath138 ) by fraaije et al @xcite . however , since the thermal noise play an important role in the vesicle formation process , we can not underestimate @xmath64 . one may consider that the mean field approximation and the free energy functional does not hold for low noise scaling parameter systems . it may be true , but as shown by dean@xcite , the dynamic density functional equation holds even if there are not large number of particles . thus here we believe that the mean field approximation still holds for our systems . to show the necessity of the large thermal noise in the simulations , we perform several simulations with various values of @xmath64 . we expect that the simulations for small boxes are sufficient to see the effect of the magnitude of the thermal noise , since the effect is large even for early stage of the dynamics . thus we perform simulations for small systems ( system size @xmath139 , lattice points @xmath140 ) with different values of @xmath64 , @xmath141 . all the other parameters are set to the previous vesicle formation simulations in section [ results ] ( figure [ vesicle_formation_snapshots ] ) . figure [ thermal_noise_snapshots ] shows the snapshots of the simulations . for small thermal noise case ( figure [ thermal_noise_snapshots](a),(b ) ) , we can observe many micelles with sharp interfaces . for large thermal noise case used in section [ results ] ( figure [ thermal_noise_snapshots](c),(d ) ) we can observe the collision - coalescence type coarsening process of micelles , driven by the thermal noise . for larger thermal noise case ( figure [ thermal_noise_snapshots](e),(f ) ) no structures are observed . ( we consider this is because the thermal noise is too large . ) thus we can conclude that the large ( but not too large ) thermal noise is required . this result is consistent with the proposition of van vlimmeren et al @xcite . they carried out two dimensional simulations for dense amphiphilic triblock copolymer solutions , with the full ( large ) noise and the reduced ( small ) noise and proposed that the real systems should have some intermediate noise ( smaller than the full noise , but larger than the reduced noise ) . it should be noted that simulations with small thermal noise is very difficult to perform for our systems , because they are in the strong segregation region . the simulation with the small noise shows strong lattice anisotropy . this means that there are very sharp interfaces which often leads numerical instability , and thus we should use finer lattices for such systems . on the other hand , we do not observe such strong lattice anisotropy in the simulations with large noise . we believe that the large thermal noise stirs the density fields and stabilizes the simulation . at the end of this section , we note that it is difficult to determine the parameter @xmath64 theoretically . the estimation of the magnitude of the thermal noise is the open problem . next we discuss about the relation between the simulation results and the experiments . we used many parameters for the simulations and they should be related to the real experimental parameters . unfortunately , because the accuracy of our theory is not high and many approximations are involved , it is difficult to calculate the real parameters from our simulation parameters . the most important parameters are the flory - huggins @xmath23 parameters . the @xmath23 parameters are experimentally measured or calculated from the solubility parameters . the @xmath23 parameter between the hydrophilic subchain and the hydrophobic subchain is in most cases sufficiently large to cause the microphase separation . for example , bhargava et al @xcite calculated the @xmath23 parameter between polystyrene ( ps ) , which can be used as the hydrophobic monomer @xcite , and water as @xmath142 . dormidontova @xcite measured the @xmath23 parameter between poly(ethyrene oxide ) ( peo ) and water and reported that the @xmath23 parameter can be expressed as @xmath143 ( @xmath5 is the absolute temperature ) . lam and goldbeck - wood @xcite also measured @xmath144 and obtained @xmath145 . xu et al @xcite measured the interaction between ps and peo and reported @xmath146 . zhu et al @xcite also measured @xmath147 and and obtained @xmath148 . the polymerization index of ps - peo diblock copolymer used in the experiments @xcite is typically @xmath149 , thus we expect that @xmath150 is sufficiently larger than the critical value @xmath151 @xcite . note that the polyelectrolytes such as poly(acrylic acid ) ( paa ) is often used as the hydrophilic subchian @xcite . such polyelectrolytes can be dissolved into water easily , and thus we expect the @xmath23 parameter between the polyelectrolytes and the water is negative . thus we can say the diblock copolymers are in strong segregation region , and the effect of water addition is expected to be large for hydrophobic subchains , but not so large for hydrophilic subchains . thus we consider that our simulation is qualitatively consistent with these experiments . note that , however , we can not compare the @xmath23 parameter used in the simulations and measured by experiments directly because the value of the @xmath23 parameter depends on the definition of the segment , and our free energy functional is not quantitatively accurate as the scf theory @xcite . we also note that most experiments are carried out in the very strong segregation region in which it is quite hard to perform continuum field simulations . while it is difficult to compare our simulations with real experiments , our simulations are expected to give qualitatively correct physical process since our simulations take the crucial physical properties correctly ; the hydrophobic interaction @xmath127 is sufficiently large and the hydrophilic interaction @xmath126 is small . these properties are not well considered in most of previous dynamics simulations . other interesting properties are the size of the vesicles , the content of solvents inside vesicles , or the higher order structures . these properties can not be discussed from our current results , since the system size is not large . we have only one vesicle in the simulation box ( figures [ vesicle_formation_snapshots],[vesicle_formation_snapshots2 ] and [ vesicle_formation_snapshots3 ] ) and there may be the finite size effect ( see also appendix [ finite_size_effect ] ) . nevertheless , we believe our model and simulation results are still valuable . these properties of vesicles should be observed if we perform the simulations for large systems while it is too difficult to perform large scale simulations by the current models , algorithms and cpu power . the characteristic time scales are also interesting and important property . as mentioned above , the real time scale can be estimated from the diffusion data . here we estimate the time scale by using the empirical equation for diffusion coefficient of short polystyrene melts ( at @xmath152 ) by watanabe and kotaka @xcite . @xmath153 where @xmath154 is the molecular weight . we have to determine the number of monomers in one segment used in our simulations . since the polymerization index of typical amphiphilic block copolymers used in experiments @xcite are about @xmath155 , we consider that each segment used in simulations contains about @xmath156 monomer units . in our simulations , we used dimensionless segment size ( @xmath106 ) . thus we also need the size of the segment . the segment size of the styrene monomer is about @xmath157 ( it can be calculated from the experimental data of the radius of gyration@xcite ) . so if we assume the gaussian statistics for monomers in the segment , size of the segment can be expressed as @xmath158 . now we can estimate the characteristic time scale for the system . if we assume that the polymer chains are not entangled and obey the rouse dynamics , the characteristic time scale @xmath159 can be calculated as follows . @xmath160 @xmath161 is the size of the segment and @xmath35 is the number of segments in one polymer chain ( polymerization index ) , and @xmath162 is the diffusion coefficient . @xmath35 and @xmath154 can be related as @xmath163 and thus we have @xmath164 thus we know that the characteristic time scale in our simulations are about @xmath1 microsecond , and the time scale of the vesicle formation is about @xmath156 milliseconds . of course the time scale estimated here is quite rough and can not be compared with real experiments directly , but the time scale of our simulation is considered to be much smaller than one of experiments . the possible reasons are that the accuracy of the dynamic equation used for the simulation is not so good , and that the sizes of the formed vesicles are small since the system size is small . we compare our simulation results with epd simulation results and discuss the reason why the epd simulations @xcite and our df simulations give qualitatively different results ( the mechanism ii and the mechanism i ) . the main difference between the epd simulations and our df simulations are the free energy functional model and the dynamic equation . the epd simulations use the free energy calculated from the scf theory while the df simulations use the free energy functional . it is known that the df theory is less accurate than the scf theory , but gives qualitatively acceptable results @xcite . thus it is difficult to consider that the accuracy of the free energy model affect the observed mechanisms . we expect that the mechanism i is reproduced if we change the free energy functional model from eq to more accurate scf model . other difference is the dynamic equation . the epd model employs nonlocal mobility model , but in the derivation of the dynamic equation , rather rough approximations are involved @xcite . because of these approximations , the local mass conservation is not satisfied in the epd simulations ( unless the system is homogeneous ) . this may affect the dynamic behaviour seriously . the mobility in our dynamic equation is the local type , and thus it is considered to be less accurate than the nonlocal model . but in our dynamic equation , local mass conservation is satisfied exactly . thus we consider that the epd simulations reproduce the mechanism ii because it does not satisfy the local mass conservation . we expect that if the dynamic equation which satisfy the local mass conservation is used with nonlocal mobility and the scf free energy , the mechanism i should be reproduced . we have shown that we can reproduce the vesicle formation mechanism i by the simulation using the dynamic density functional equation and the free energy functional . the simulation results are qualitatively in agreement with other simulations @xcite . this is the first realistic vesicle formation dynamics simulation based on the field theoretical model . to reproduce the mechanism i , we need the free energy functional which can be applied to strong segregation regions , the large thermal noise , and the proper interaction parameters for hydrophilic and hydrophobic interactions . these conditions are not satisfied previous field theoretical dynamic simulations . we have also shown that the formation dynamics of various micellar structures ( spherical micelles , cylindrical micelles , vesicles , bilayers ) and morphological transitions can be simulated by our model . the morphological transitions can be reproduced only by changing the @xmath23 parameter which corresponds to the change of the solvent condition . the merit of using the field theoretical approach is that we can use the flory - huggins interaction parameters , instead of potentials between segments which is required for particle simulations . we have shown that we can mimic the solvent condition control by changing the @xmath23 parameter between the hydrophobic segment and the solvent and have shown that the solvent condition control actually causes morphological transitions . for coarse grained particle simulations , we can not use the microscopic potential directly , and in most cases the phenomenological potential models are used . of course for some particle simulations , such as the dpd , we can use the @xmath23 parameters by mapping them onto the dpd potential parameters . but such method is based on the parameter fitting @xcite and not always justified . thus we consider that especially for many component systems , such as solutions of amphiphiles , the field theoretical approaches have advantages . although the @xmath23 parameters in the current model can not be compared with the experiments directly , our data may help understanding experimental data or the physical mechanism . it will be possible to use the experimentally determined @xmath23 parameters by employing more accurate free energy model such as the scf model @xcite and efficient numerical algorithms . ( we note that recently honda and kawakatsu @xcite proposed the hybrid theory of the df and the scf and reported that accurate and numerically efficient simulations can be performed by the hybrid method . by employing their method , accurate and fast simulations for the dynamics of micelles and vesicles may be possible . ) the quantitative simulations will be the future work . while the efficiency of our simulation method based on the density functional theory ( eqs and ) is larger than the scf simulations , currently it is still comparable to particle simulations . besides , because the late stage of the mechanism i is a very slow process , the current model still need considerable computational costs . to study the late stage dynamics or large scale systems ( for example , many vesicle systems ) , we will need more coarse grained and numerically more efficient computational methods . our current work can be the basis of the further coarse grained field theoretical model . the author thanks t. ohta for reading the manuscript and giving useful suggestions , and for helpful comments and discussions . thanks are also due to h. morita , t. taniguchi , y. masubuchi , h. frusawa , and t. kawakatsu for helpful comments and discussions . the author also thanks anonymous referees who read the original manuscript and gave him useful comments for the improvement of the manuscript . this work is supported by the research fellowships of the japan society for the promotion of science for young scientists . in this section we derive the numerical scheme for generating random noise field which satisfies the fluctuation dissipation relation ( eqs and ) . such a noise field can be expressed as @xcite @xmath165 \\ & = 2 \sqrt{\tilde{\beta}^{-1 } } \nabla \cdot \left [ \psi_{i}(\bm{r } ) \bm{\omega}_{i}(\bm{r},\tilde{t } ) \right ] \end{split}\ ] ] where @xmath166 is the gaussian white noise ( vector ) field which satisfies @xmath167 where @xmath168 is the unit tensor . to generate the noise numerically , we need the descretized version of eq . the descretized noise field @xmath63 is defined only on the lattice points . here we express the position of the lattice point as @xmath169 where @xmath136 ( @xmath170 ) is the lattice vector and @xmath171 is the integer . the rescaled time @xmath66 is also descretized as @xmath172 where @xmath173 is the integer . the laplacian operator is replaced by the standard centre finite difference operator as follows @xmath174\ ] ] where @xmath175 is a descretized field defined on the lattice points . the delta functions are replaced as follows @xmath176 where @xmath177 and @xmath178 . the fluctuation dissipation relation for the second order moment ( eq ) can be rewritten as @xmath179 \delta(\tilde{t } - \tilde{t } ' ) \\ & = - 4 \tilde{\beta}^{-1 } \delta_{ij } \nabla \cdot \left [ \psi^{2}_{i}(\bm{r } ) \nabla \frac{\psi_{i}(\bm{r } ) \delta(\bm{r } - \bm{r}')}{\psi_{i}(\bm{r } ) } \right ] \delta(\tilde{t } - \tilde{t } ' ) \\ & = - 4 \tilde{\beta}^{-1 } \delta_{ij } \left [ \psi_{i}(\bm{r } ) \nabla^{2 } \left [ \psi_{i}(\bm{r } ) \delta(\bm{r } - \bm{r } ' ) \right ] - \left [ \psi_{i}(\bm{r } ) \delta(\bm{r } - \bm{r } ' ) \right ] \nabla^{2 } \psi_{i}(\bm{r } ) \right ] \delta(\tilde{t } - \tilde{t } ' ) \end{split}\ ] ] using eqs - , we get the descretized version of the fluctuation dissipation relation . @xmath180 \\ & - \psi_{i}(\bm{r } ) \psi_{i}(\bm{r } - \bm{h}_{\alpha } ) \left [ \delta_{\bm{r } \bm{r } ' } - \delta_{(\bm{r } - \bm{h}_{\alpha } ) \bm{r } ' } \right ] \bigg ] \delta_{\tilde{t } \tilde{t } ' } \end{split}\ ] ] in this work , we employ the following descretized equation to generate the noise field . @xmath181 \end{split}\ ] ] where @xmath182 is the @xmath30 element of @xmath183 . @xmath183 is the descretized gaussian white noise field which satisfies @xmath184 notice that @xmath183 is defined on the staggered lattice . the position of the staggered lattice point is @xmath185 where @xmath171 is the integer . it is easy to show that the noise field generated by eq satisfies eq . the noise generation scheme is finally written as follows . 1 . generate the gaussian white normal distribution vector noise field @xmath183 on the staggered lattice points . the noise is generated by using the mersenne twister pseudo - random number generator @xcite and the standard box - muller method @xcite . 2 . calculate the noise field @xmath63 from @xmath15 and @xmath183 by using eq . it is well known that the size of the simulation box often affects the simulation results ( the finite size effect ) . in this appendix , we study the finite size effect for our simulations by performing simulations with different box size . figure [ small_boxes ] shows the simulation results for different ( small ) box sizes . all the parameters except for the box size are the same for the previous vesicle formation simulation ( figure [ vesicle_formation_snapshots ] ) . the box sizes are set to @xmath186 ( @xmath187 lattice points ) , @xmath188 ( @xmath189 lattice points ) , @xmath139 ( @xmath190 lattice points ) , and @xmath191 ( @xmath192 lattice points ) . we can observe the spherical micelles ( figure [ small_boxes](a ) ) and the cylindrical micelles ( figure [ small_boxes](b),(c ) ) for small boxes . these structures are much different from the vesicle structures in figure [ small_boxes](d ) or figure [ vesicle_formation_snapshots ] . this means that if the simulation box is too small , we can not obtain vesicles . it also implies that there may be the finite size effect in our vesicle formation simulations ( box size @xmath193 ) , since the size of obtained vesicles are in most cases comparable to the box sizes . but at least we can claim that our vesicle formation process itself is not affected qualitatively since the system size is not so small for the formation of disk like structures ( unlike the case of figure [ small_boxes](a),(b),(c ) ) and we can observe that the small vesicle formation process ( figure [ vesicle_formation_snapshots2](d ) ) is qualitatively same as one for larger vesicles . the cutoff length @xmath26 for the long range interaction term in eq is introduced rather intuitively , and therefore the value of @xmath26 for simulations should be considered carefully . in this section , we perform the simulations with different values of @xmath26 . simulations are performed in two dimensions . parameters are as follows : @xmath99 , @xmath100 @xmath194 . @xmath105 , @xmath106 . the size of the 2 dimensional simulation box is @xmath195 ( @xmath196 lattice points , periodic boundary condition ) . the time step is @xmath109 ( for @xmath197 ) , @xmath198 ( for @xmath199 ) , @xmath200 ( for @xmath201 ) . the magnitude of the noise is @xmath202 . all the simulations are started from the homogeneous state . figure [ lambda_snapshots ] shows the results for various value of the cutoff length ( ( a),(b ) @xmath203 , ( c),(d ) @xmath204 , ( e),(f ) @xmath205 , ( g),(h ) @xmath103 , ( i),(j ) @xmath199 , and ( k),(l ) @xmath201 ) at @xmath206 . we can observe the droplet like structures are observed for @xmath203 . and micellar structures for @xmath207 . the resulting morphologies for @xmath207 are qualitatively same . ( strictly speaking , @xmath204 looks slightly different from others and somehow resembles to @xmath203 . ) as expected , the cutoff length dependence of the morphology is small . thus even if the value of @xmath26 is not accurate , the results are expected not to be affected qualitatively . from this result , we can justify to use the value @xmath103 , which is estimated by the rough argument . 67 natexlab#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , _ _ ( , , ) . , * * , ( ) . , * * , ( ) . , _ _ ( , , ) . , * * , ( ) . , , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , . , * * , ( ) . , , , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , _ _ ( , , ) . , , , , _ _ ( , ) , ed . , in _ _ ( ) , * * , , . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , , , , * * , ( ) . , * * , ( ) . , , , , , * * , ( ) . , , , , , * * , ( ) . , , , , , * * , ( ) . , , , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , , , * * , ( ) . , , , , , , , , , * * , ( ) . , * * , ( ) . _ _ , . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , , . , * * , ( ) , . figure [ vesicle_formation_mechanisms ] : schematic representation of two vesicle formation mechanisms from the initial homogeneous state . black and grey color corresponds to hydrophobic and hydrophilic subchains , respectively . ( a ) mechanism i : first , small micellar structures are formed . the micellar structures grow by collision and become large cylindrical or open disk - like micelles . finally the open disk - like micelle close up to form closed vesicles . ( b ) mechanism ii : the initial stage is similar to the mechanism i. however , the small micellar structures formed in the initial stage grows up to be large spherical micelles . the large spherical micelles then take the solvent inside and thus vesicles are formed . figure [ vesicle_formation_snapshots ] : snapshots of the dynamics simulation for the amphiphilic diblock copolymer solution ( @xmath101 ) . grey surface is the isodensity surface for the hydrophobic subchain , @xmath208 . ( a ) @xmath209 , ( b ) @xmath210 , ( c ) @xmath211 , ( d ) @xmath212 , ( e ) @xmath213 , ( f ) @xmath113 . figure [ vesicle_formation_snapshots2 ] : snapshots of the dynamics simulation for the amphiphilic diblock copolymer solution ( @xmath101 ) . grey surface is the isodensity surface for the hydrophobic subchain , @xmath208 . ( a ) @xmath209 , ( b ) @xmath210 , ( c ) @xmath211 , ( d ) @xmath212 . figure [ vesicle_formation_snapshots3 ] : snapshots of the dynamics simulation for the amphiphilic diblock copolymer solution ( @xmath101 ) . grey surface is the isodensity surface for the hydrophobic subchain , @xmath208 . ( a ) @xmath209 , ( b ) @xmath212 , ( c ) @xmath213 , ( d ) @xmath113 . figure [ structure_volume_fraction_dependence ] : snapshots of the dynamics simulations for amphiphilic diblock copolymer solutions with various volume fractions . grey surface is the isodensity surface for the hydrophobic subchain , @xmath208 at @xmath212 . ( a ) @xmath217 , ( b ) @xmath218 , ( c ) @xmath219 , ( d ) @xmath220 . ( @xmath101 corresponds to figure [ vesicle_formation_snapshots](d ) . ) figure [ structure_chi_parameter_dependence ] : snapshots of the dynamics simulations for amphiphilic diblock copolymer solutions with various @xmath127 ( the @xmath23 parameter between the hydrophobic chain and the solvent ) . grey surface is the isodensity surface for the hydrophobic subchain , @xmath208 at @xmath212 . ( a ) @xmath221 , ( b ) @xmath131 , ( c ) @xmath222 , ( d ) @xmath223 . ( @xmath130 corresponds to figure [ vesicle_formation_snapshots](d ) . ) figure [ structural_transition ] : snapshots of the structural transition dynamics simulations . the @xmath23 parameter @xmath127 is initially set to @xmath130 and at @xmath224 , @xmath127 is set to lower value . grey surface is the isodensity surface for the hydrophobic subchain , @xmath208 at @xmath133 . ( a ) @xmath225 , ( b ) @xmath226 . figure [ small_boxes ] : snapshots of the dynamics simulations for amphiphilic diblock copolymer solutions with various small box sizes at @xmath212 . ( a ) box size @xmath186 ( @xmath187 lattice points ) , ( b ) box size @xmath188 ( @xmath189 lattice points ) , ( c ) box size @xmath139 ( @xmath190 lattice points ) , and ( d ) box size @xmath191 ( @xmath192 lattice points ) . super cells ( size @xmath107 ) is shown . small gray boxes show the real simulation box . figure [ lambda_snapshots ] : snapshots of the dynamics simulations for amphiphilic diblock copolymer solutions with various cutoff length @xmath26 . black color represents @xmath215 . ( a),(b ) @xmath203 , @xmath206 . ( c),(d ) @xmath204 , @xmath206 . ( e),(f ) @xmath205 , @xmath206 . ( g),(h ) @xmath103 , @xmath206 . ( i),(j ) @xmath199 , @xmath206 . ( k),(l ) @xmath201 , @xmath206 .

we present results of n - body / gasdynamical simulations designed to investigate the evolution of x - ray clusters in a flat , low - density ,-dominated cold dark matter ( cdm ) cosmogony . the simulations include self - gravity , pressure gradients and hydrodynamical shocks , but neglect radiative cooling . the density profile of the dark matter component can be fitted accurately by the simple formula originally proposed by navarro , frenk & white to describe the structure of clusters in a cdm universe with . in projection , the shape of the dark matter radial density profile and the corresponding line - of - sight velocity dispersion profile are in very good agreement with the observed profiles for galaxies in the cnoc sample of clusters . this suggests that galaxies are not strongly segregated relative to the dark matter in x - ray luminous clusters . the gas in our simulated clusters is less centrally concentrated than the dark matter , and its radial density profile is well described by the familiar-model . as a result , the average baryon fraction within the virial radius ( ) is only- of the universal value and is lower nearer the center . the total mass and velocity dispersion of our clusters can be accurately inferred ( with uncertainty ) from their x - ray emission - weighted temperature . we generalize kaiser s scalefree scaling relations to arbitrary power spectra and low - density universes and show that simulated clusters generally follow these relations . the agreement between the simulations and the analytical results provides a convincing demonstration of the soundness of our gasdynamical numerical techniques . although our simulated clusters resemble observed clusters in several respects , the slope of the luminosity - temperature relation implied by the scaling relations , and obeyed by the simulations , is in disagreement with observations . this suggests that non - gravitational effects such as preheating or cooling must have played an important role in determining the properties of the observed x - ray emission from galaxy clusters .

astro-ph9708070

galaxy clusters , the largest virialized systems in the universe , are useful cosmological probes . for example , the abundance of massive clusters ( characterized either by mass or x - ray temperature ) depends sensitively on @xmath7 , the cosmological density parameter , and on @xmath8 , the rms amplitude of density fluctuations on the fiducial scale @xmath9 ( white , efstathiou & frenk 1993 , eke , cole & frenk 1996 , viana & liddle 1996 ) . thus , the present - day abundance of clusters and its redshift evolution may be used to place constraints these two fundamental cosmological parameters . similarly , the observed baryon fraction in clusters places strong constraints on the value of @xmath7 ( white 1993 ) . recent applications of these ideas tend to favor low values of @xmath10 ( white & fabian 1995 , evrard 1997 , henry 1997 ) and , for flat models with this @xmath7 , values of @xmath11 which are broadly consistent with the amplitude of the microwave background anisotropies measured by cobe ( smoot 1992 ) . to exploit fully the cosmological information encoded in the cluster population , it is necessary to understand their evolutionary history in some detail . this requires modeling the coupled evolution of the dark matter and gas , which together constitute the dominant contribution to the cluster mass . in its full generality , this problem is best approached by direct simulation and a variety of numerical techniques have now been developed for this purpose . many of the techniques currently in use ( including both eulerian and lagrangian hydrodynamics methods ) have been recently compared by means of a test calculation of the formation of a cluster by hierarchical clustering in which the gas was assumed to be non - radiative ( frenk , in preparation ) . the different simulations resolved the cluster to different degrees , but in the regions resolved by each calculation , they generally gave remarkably similar results for most cluster properties of interest . already the first n - body / gasdynamic simulations showed that in the non - radiative approximation , the x - ray properties of individual clusters formed in flat cold dark matter ( cdm ) cosmologies resemble those of real clusters in many respects ( evrard 1990b ) . subsequent simulations have developed this theme further , generally with qualitatively similar conclusions ( e.g. thomas & couchman 1992 ; kang 1994 ; cen & ostriker 1994 , bryan 1994 ; navarro , frenk & white 1995 , owen & villumsen 1997 ) . yet , it has been clear for some time , that the simulations ( at least in an @xmath1 cdm cosmology ) do not reproduce important systematic trends of the observed cluster _ population _ such as the slope of the relation between x - ray temperature and luminosity . this has led several authors to argue that effects not included in the simulations , such as cooling or preheating of the gas , must have played a role in the evolution of the cluster population ( kaiser 1986 , evrard & henry 1991 , navarro , frenk & white 1995 ) . in particular , navarro , frenk & white ( 1995 ) showed that moderate preheating at high redshift leads to an acceptable luminosity - temperature relation without spoiling the overall agreement with the observed structure of the x - ray gas in individual clusters . useful insights into the evolution and systematic properties of the cluster population may also be obtained by studying scaling relations , an approach developed by kaiser ( 1986 ; see also white & rees 1978 , and white 1982 ) . these authors recognized that since gravity has no preferred scales , cluster properties determined primarily by gravity ( or by other scale - free processes such as pressure gradients or hydrodynamical shocks ) should obey simple scaling relations . kaiser derived these for a population of clusters formed by hierarchical clustering from power - law initial density fluctuations in an einstein - de sitter universe . he concluded that , for most power spectra of interest , the cluster x - ray luminosity function evolves with redshift in the opposite sense to that indicated by the data available at the time ( edge 1990 , gioia 1990 ) . more recent data , however , appear to be consistent with little or no evolution in the cluster x - ray luminosity function out to @xmath12 ( nichol 1997 , rosati 1998 ) . in this paper , we carry out a detailed investigation of the evolution of clusters in a low - density , @xmath13 , cdm universe . we impose the flat geometry required by inflation by setting the cosmological constant @xmath14 . in units of @xmath15 , so that a universe with @xmath16 has a flat geometry . the present value of hubble s constant , @xmath17 , is parameterized by @xmath18 km s@xmath19 mpc@xmath19 . ] we perform a set of n - body - hydrodynamical simulations of cluster formation in this cosmology . we also generalize kaiser s scaling laws to the case of an arbitrary cosmology and generic density fluctuation spectra . cluster evolution in low - density universes has been explored numerically in a few previous papers ( cen & ostriker 1994 ; evrard , metzler & navarro 1996 ) , but none has yet addressed in detail the evolutionary properties of the x - ray emission from individual clusters . our extension of kaiser s scaling laws in based on recent numerical results by navarro , frenk & white ( 1995 , 1996 , 1997 , hereafter nfw95 , nfw96 , and nfw97 , respectively ) , who found that virialized systems formed by hierarchical clustering exhibit a remarkable structural similarity . throughout this paper we make the simplifying assumption that only gravity , pressure gradients and hydrodynamical shocks are important in the evolution of clusters . the plan of this paper is as follows . in 2 we derive generalized scaling laws describing correlations between various cluster properties and their redshift evolution . in 3 , we describe our numerical methods and provide details of the ten clusters which we have resimulated at high resolution . these span a range of formation histories and dynamical states . our main numerical results are presented in 4 where we investigate the structure of the dark matter and gas in our clusters , the accuracy of cluster mass estimates , the evolution of their baryon fraction and the origin of possible deviations from a universal mean baryon fraction . in this section we also carry out a comparison with the generalized scaling laws derived in 2 . in 5 we compare our results with previous numerical work and with observations . a summary of our main conclusions is given in 6 . in an einstein - de sitter ( @xmath1 ) universe with a power - law spectrum of primordial density fluctuations , @xmath20 , the characteristic clustering mass evolves as @xmath21 . since there are no other scales in the problem , the characteristic density of an @xmath22 cluster can only be proportional to the density of the universe , @xmath23 . a characteristic mass and density define , through the virial theorem , a characteristic temperature ( or velocity dispersion ) which scales as @xmath24 . assuming that the x - ray emission from intracluster gas is dominated by bremsstrahlung , the characteristic x - ray luminosity then scales as @xmath25 . the x - ray luminosity of a _ typical _ cluster may thus increase or decrease with redshift , depending on whether @xmath26 is larger or smaller than @xmath27 , respectively . these scaling relations predict the time evolution of a _ typical _ cluster ( ie . a cluster with mass equal to the characteristic clustering mass ) , but they do not describe the evolution of individual clusters or the mass dependence , at fixed redshift , of the x - ray luminosity or other cluster properties . without this information , which depends on the _ internal structure _ of clusters , it is not possible to assess the cosmological significance of observed correlations such as the luminosity - temperature relation or their evolution . the simplest model assumes that : ( i ) the internal structure of clusters of different mass is similar ; ( ii ) all clusters identified at some redshift have the same characteristic density ; and ( iii ) this characteristic density scales like the mean density of the universe , ie . as @xmath28 ( see evrard 1990b , evrard & henry 1991 , nfw95 ) . with these assumptions , and the further hypothesis that the relative distributions of gas and mass are similar in all clusters , it is possible to derive scaling relations between mass , luminosity , and temperature ( velocity dispersion ) . in particular , we have that @xmath29 and , assuming again that bremsstrahlung is the main emission mechanism , @xmath30 , so that @xmath31 thus , in this model , clusters of a given temperature or mass are expected to become more luminous at higher redshifts , a useful prediction that can be tested observationally . we shall come back to this issue in @xmath32 . as mentioned above , this approach has some support from recent n - body work , which shows that indeed the structure of clusters formed through hierarchical clustering exhibit a remarkable similarity ( nfw96 , nfw97 ) . regardless of cluster mass , power spectrum shape , or the value of the cosmological parameters , the density profiles of dark halos formed through hierarchical clustering can be fitted accurately by scaling a simple formula , @xmath33 here @xmath34 is a characteristic dimensionless density and @xmath35 is a scale radius related to the total mass of the system . nfw96 and nfw97 show that these two parameters are correlated , a relation that reflects the different collapse redshift of systems of different mass . these authors also provide the necessary formulae to compute @xmath34 as a function of mass in any hierarchically clustering cosmogony . we note that assuming a common characteristic density for all clusters , as in ( ii ) above , is equivalent to setting @xmath36const . in eq.(3 ) . we show below ( 4.3 ) that the spherical top - hat model provides a useful definition of the mass contained within the virialized region of a system ( eke et al . 1996 ) . the virial mass , @xmath37 , is defined to be the mass contained within the radius , @xmath3 , that encloses a density contrast @xmath38 : @xmath39 . , where @xmath40 is the current value of hubble s constant the term ` overdensity ' will be used to refer to densities in units of the mean background matter density . ] this density contrast depends on the value of @xmath41 and can be approximated by ( lacey & cole 1993 , eke , cole & frenk 1996 ) : @xmath42 this formula is accurate to within @xmath43 for @xmath44 . it is straightforward to show that , with this definition , the ratio between virial and scale radii , which we denote by the ` concentration ' , @xmath45 , is uniquely related to @xmath34 by @xmath46}. \eqno(5)\ ] ] the structure of a halo of mass @xmath37 is then completely specified by a single parameter , which may be chosen to be the characteristic density , @xmath34 , or the concentration , @xmath47 . we can use these results to calculate the expected interdependence of mass , temperature , and x - ray luminosity in various cosmologies . the bolometric x - ray luminosity is given by @xmath48 where @xmath49 is the cooling function and @xmath50 is the gas density distribution . assuming that the gas distribution traces the dark matter , ie . @xmath51 , ( @xmath52 is the gas mass fraction ) and that the gas is isothermal , we find , @xmath53 where @xmath54 is a function only of the concentration , @xmath55 ^ 2}. \eqno(8)\ ] ] note that if gas traces mass the x - ray luminosity converges , since @xmath56 near the center . eq.(7 ) expresses the luminosity in terms of the mass , temperature and concentration of the system . the same dependence of @xmath57 on @xmath58 and @xmath59 ( but different proportionality constants ) is expected even if the gas distribution does not trace mass in detail , so long as the gas density can be written as @xmath60 , with the same dimensionless function @xmath61 for all clusters . in order to derive the luminosity - temperature relation , we need a relationship between mass and temperature or , equivalently , between mass and velocity dispersion . the existence of a characteristic density in the structure of the system implies the existence of a characteristic velocity as well . this is easily seen in the circular velocity profile , @xmath62 , implied by eq.(3 ) , @xmath63 here @xmath64 is the circular velocity at the virial radius , and @xmath65 is the radius in units of @xmath3 . from eq.(9 ) , @xmath66 has a maximum , @xmath67 , at @xmath68 , @xmath69 this maximum circular velocity is independent of the definition of virial radius , and may be taken to represent the characteristic velocity of the system . other measures of the depth of the potential well , such as the velocity dispersion of the dark matter , @xmath70 , are expected to scale with @xmath67 . in the absence of non - gravitational heating or cooling effects , the characteristic temperature of the gas should also scale as @xmath71 , so that @xmath72 } , \eqno(11)\ ] ] or @xmath73 where @xmath74 $ ] and we have used @xmath75 . these relations ( eqs . 7 and 12 ) can be used to construct the luminosity - temperature relation . assuming that the main emission process is bremsstrahlung , @xmath76 , we can write eq.(7 ) as , @xmath77 which , using eq.(12 ) , may be expressed as , @xmath78 these relations generalize the expected dependence between luminosity , mass and temperature to arbitrary values of @xmath41 and arbitrary ( ie . non power - law ) power spectra . comparing eqs . ( 1 ) and ( 2 ) with ( 12 ) and ( 14 ) we see that , except for the functions of @xmath47 , the generalized relations are identical to those derived assuming a common characteristic density for all clusters . indeed , for @xmath1 and @xmath79const . , these equations are identical . eqs.(12 ) and ( 14 ) require the value of the concentration as a function of mass , power spectrum , and cosmological parameters . a simple algorithm to calculate this is given in the appendix of nfw97 . a large n - body simulation of an @xmath80 , @xmath81 , cdm cosmogony was carried out using the ap@xmath82 m code written by couchman ( 1995 ) . the power spectrum used is that of bond & efstathiou ( 1984 ) . this simulation followed @xmath83 particles in a cube of side @xmath84 . gravitational accelerations were computed using a parent mesh of @xmath85 grid cells , and softened with an effective plummer comoving lengthscale of @xmath86 kpc . the simulation was stopped after @xmath87 expansion factors when @xmath8 , the linear theory _ rms _ mass fluctuation on spheres of @xmath9 , was equal to @xmath88 . we identify this epoch with the present , consistent with both the standard cobe normalization and the observed abundance of rich clusters ( white et al 1993 ) . we then applied a ` spherical overdensity ' group - finding algorithm ( lacey & cole 1994 ) on this final configuration in order to identify clumps with a mean interior density contrast of @xmath89 , which corresponds to the ` virial ' radius in the spherical top - hat model for @xmath13 ( see eq.4 ) . the ten most massive clusters were selected from this list and all of their particles within about @xmath90 were identified . no attempt was made to cull the list for the presence of massive neighbors or for the dynamical state of the cluster . some of our selected clusters are in the process of merging , and essentially all of them show , to some degree , signs of recent accretion and departures from equilibrium . figure [ fig : cluspos ] shows , at @xmath91 , the positions of these clusters in projection within the original box . each cluster is surrounded with a circle of radius @xmath92 ( only particles within @xmath3 are shown ) and is labeled in decreasing order of mass . the ten clusters span a factor of @xmath93 in mass , with a mean of @xmath94 , and typically have @xmath95 particles each . for each cluster , all particles identified within @xmath96 were traced back to the initial conditions , where a small box containing all of them was drawn . these particles were then replaced with equal numbers of gas and dark matter particles on a cubic grid which was then perturbed using the waves of the original ap@xmath82 m simulation , together with extra high - frequency waves added to fill out the power spectrum between the nyquist frequencies of the old and new particle grids . the number of ` high - resolution particles ' was varied as a function of the size of the small box and of the mass of each cluster so as to have , at @xmath91 , about the same number of particles in each cluster . all remaining material in the large simulation was coarse - sampled and replaced by approximately @xmath97 dark matter particles of radially increasing mass . we list the relevant numerical parameters in table 1 . the first four columns give : ( 1 ) a label for the run , ( 2 ) the size of the ` high - resolution ' box in comoving @xmath98 , ( 3 ) the number of gas particles ( or dark matter particles , since they are the same ) loaded into this region , and ( 4 ) the mass of each gas particle , which is assigned assuming an overall gas mass fraction of @xmath99 . this is consistent with the universal baryon fraction suggested by the primordial abundance of the light elements , @xmath100 ( copi , schramm & turner 1995 ) . the mass of each dark matter particle is therefore nine times that of a gas particle . ( we note that , since radiative cooling is neglected in the simulations , all our results may be rescaled to arbitrary values of the gas fraction , so long as the gas mass remains a small fraction of the total . ) in order to ensure that hydrodynamical forces do not play a significant role in the evolution of the gas until the collapse of the first resolved structures , we initially assign a very low internal energy to each gas particle , corresponding to a temperature of about @xmath101 k. we use the n - body / smooth particle hydrodynamics ( sph ) code described in detail by navarro & white ( 1993 ) , modified as follows : ( i ) the code has been adapted to run in a @xmath102 universe by the addition of a centrally symmetric acceleration directed outwards , of magnitude proportional to the distance of each particle to the center . ( ii ) the program computes the gravitational accelerations using a grape-3af board . the neighbor lists needed for the sph computations are also retrieved from the grape and processed in the front - end workstation . the implementation of these modifications is straightforward and very similar to those described by steinmetz ( 1996 ) , where the reader may find further details . the effects of radiative cooling , heating by a photoionizing uv background , or star formation can be handled by this code but were neglected in the simulations presented here . the only physical processes included are gravity , pressure gradients , and hydrodynamical shocks . an ideal gas equation of state with @xmath103 is used to relate pressure and internal energy . details and tests of our sph implementation may be found in navarro & white ( 1993 ) . gravitational accelerations are softened using a _ physical _ ( not comoving ) plummer scale - length of @xmath86 kpc . this is less than @xmath104 of the final virial radius of all ten clusters . all simulations were started at @xmath105 and integrated to @xmath91 using individually adjusting timesteps for each particle . typically , timesteps ranged between @xmath106 and @xmath107 yrs , depending on the dynamical situation experienced by each particle . we follow nfw95 , and use the following estimator for the _ bolometric _ x - ray luminosity of a cluster , @xmath108 where @xmath109 is the proton mass , @xmath110 is the mean molecular weight of a primordial plasma , @xmath111 is the mass of a gas particle , and @xmath112 and @xmath113 are the density and temperature at the location of particle @xmath114 , respectively . masses and densities are in cgs units in this formula , which assumes that the main x - ray emission mechanism is bremsstrahlung , ie . @xmath115 erg cm@xmath82 s@xmath19 . the sum is carried out over all gas particles within the virial radius of a cluster . unless otherwise specified , we shall quote luminosities in units of @xmath117 erg s@xmath19 ( @xmath118 erg@xmath119 joules ) . luminosities depend on the value assumed for the gas fraction , @xmath120 , but can be rescaled to other values by @xmath121 . the entropy per particle is defined as @xmath122 , where @xmath123 is expressed in kev and @xmath112 in units of @xmath124mpc@xmath82 . ( the factor @xmath125 is an arbitrary constant . ) in the case of dark matter particles , @xmath112 is the mean dark matter density within a sphere centered on @xmath114 containing the @xmath126 nearest dark matter particles , and @xmath127 , where @xmath128 is the @xmath129 velocity dispersion of the particles . table 2 summarizes the bulk properties of the ten clusters at @xmath91 . all quantities are computed within the virial radius . the majority of clusters have more than @xmath130 dark matter particles within the virial radius , except cl07a and cl08a , which have significantly fewer . this is a result of their physical proximity at @xmath91 ( see figure 1 ) which demands a large high - resolution box in order to enclose all the material that ends up within @xmath131 of the center , and consequently a larger mass per particle if we are limited to @xmath132 , the maximum number we can afford to run . particle plots of the gas and dark matter configurations of three clusters at @xmath91 are shown in figure 2 . the gas appears to be slightly more spherical than the dark matter because it traces the equipotentials of the system , which are significantly rounder than the dark matter density distribution ( evrard 1990a ) . the dark matter retains more small scale structure than the gas . this is presumably a result of the combined effects of ram - pressure stripping of gas as small clumps move in the hot atmosphere of the main cluster , and of numerical limitations which tend to smooth the gas distribution on mass scales smaller than @xmath133 particles . the evolution of all clusters is similar to that described in previous work ( see , eg . , 3.1 of nfw95 ) , and is illustrated in figure 3 for cluster cl01a . a cluster accretes most of its final mass in the form of mergers with smaller clumps which flow along large - scale filaments easily noticeable in this figure . figures 4 and 5 show the evolution of the bulk properties of each cluster : mass , velocity dispersion , temperature , x - ray luminosity , central entropy ( see figure label ) , and ` beta'-parameters . the latter are defined below in eqs.(18 ) and ( 19 ) . all these properties are measured within the current virial radius of the most massive progenitor of the final system . as is clear from figures 4 and 5 , clusters form late , accreting on average half of their final mass since @xmath134 . this is shown quantitatively in table 1 , which lists the ` formation redshift ' , of each cluster , @xmath135 , defined as the epoch when the mass of the most massive progenitor first exceeds one - half of the mass of the system at @xmath91 . the formation redshifts are in good agreement with analytical estimates based on the press - schechter theory ( lacey & cole 1993 ) . note the large scatter in @xmath135 , which varies between @xmath136 and @xmath137 as a result of the intrinsic variety of formation histories of individual clusters . the evolution of our ten clusters is summarized in figure 6 where we plot the average , over all clusters and at each time , of the bulk properties scaled to their value at @xmath91 . as clusters grow increasingly massive , the potential well deepens , and they become hotter and more luminous in x - rays . although the mass increases rapidly with time , other cluster properties evolve more slowly . at @xmath138 clusters are , on average , four times less massive and about half as luminous or as hot as at @xmath91 . cluster masses double after @xmath139 , but their luminosities and temperatures increase only by @xmath140 . the velocity dispersion changes even less since it scales like the square - root of the temperature . therefore , in this cosmogony we expect distant luminous clusters to be almost as bright and as hot as those in the local universe . the evolution is punctuated by mergers during which clusters may brighten temporarily by more than a factor of two ( c.f . figures 4 and 5 ) . the central ` entropy ' of the gas and dark matter ( see 3.4.2 ) also increases steadily , and by roughly similar amounts , as the cluster evolves . this suggests similarity in the evolution of gas and dark matter , conforming to the hypothesis on which we based our derivation of scaling laws in 2 . these results may be compared with the predictions of the scaling laws derived in @xmath141 . the thick dashed lines in figure 6 show the evolution of @xmath59 , @xmath142 , and @xmath57 predicted by eqs.(12 ) and ( 14 ) for a cluster of mass equal to the mean plotted in the upper left panel . all curves are normalized to the values of each quantity at @xmath91 . concentrations are computed using the algorithm given in the appendix of nfw97 . the agreement between the predictions of the scaling laws and the results of the simulations is remarkable , and implies that the gas and dark matter components evolve similarly as clusters grow more massive . the structure of the gas and dark matter within clusters must therefore remain either similar or proportional to one another at all times , an issue we investigate further below . appropriate scaling behaviour is widely regarded as a powerful test of numerical techniques . the agreement between our simulations and the scaling laws provides a convincing validation of the smooth particle hydrodynamics technique we are using here . figures 7 and 8 illustrate the structure of our clusters at @xmath91 . many of them seem to be close to equilibrium , with the exception of cl04a , cl05a , cl08a , and cl09a . departures from equilibrium show up clearly in the ratio of the gas bulk kinetic to thermal energies , shown in the fourth row of figures 7 and 8 . this ratio would be zero , of course , if the gas were in perfect hydrostatic equilibrium , but it is temporarily different from zero as a result of recent accretion . table 2 lists the average values of this ratio within the virial radius . ongoing merger events are also easy to spot in the mean radial velocity profiles , which show significant departures from the zero mean characteristic of a system in virial equilibrium . thus , even in this low - density universe , 4 out of 10 clusters are substantially out of equilibrium , a large fraction which may hamper attempts to determine the cosmological parameters from the fraction of relaxed , virialized clusters ( mohr , evrard & fabricant 1995 ) . in all clusters , and almost regardless of how close to equilibrium they are , the velocity distribution of the dark matter is radially biased , albeit mildly ; @xmath143 , with a weak radial dependence ( see bottom row of figures 7 and 8) . this is more easily seen in the average velocity anisotropy profile shown in the lower right panel of figure 9 . outside a small inner region where the velocity distribution is very nearly isotropic , the magnitude of the radial bias is essentially constant at @xmath144 out to @xmath145 ; beyond that radius the bias increases , peaking at about @xmath146 at the virial radius . this radial behaviour can be approximated as @xmath147 . similar results have been reported in previous work ( eg . , evrard 1990b , cole & lacey 1996 ) . the average gas temperature profile is shown in the upper left panel of figure 9 as a dashed line . to form this average , individual temperatures were scaled to the ` virial temperature ' of each system , defined as @xmath148 as found by evrard ( 1990b ) and nfw95 , the gas near the center is close to isothermal ; its temperature at @xmath149 differs from the central one by less than @xmath150 . beyond that radius , the temperature drops faster , approximately as @xmath151 , reaching about half the central value at the virial radius . the dark matter ` temperature ' ( ie . the square of the velocity dispersion about the mean in each radial shell normalized to @xmath152 ) has a similar behaviour , but drops below the gas temperature near the center . this is a direct consequence of the different density profiles of gas and dark matter in the inner regions . as may be seen in figures 7 and 8 , the gas is less centrally concentrated than the dark matter and must therefore be hotter in order to be in equilibrium within the same gravitational potential . the lower left panel in figure 9 shows the average cumulative x - ray luminosity profile . the density profiles are rather peaked and , as a result , the luminosity is very concentrated ; half the total energy is generated within @xmath153 of the virial radius . thus , most of the emission comes from the region where the gas is nearly isothermal . any significant departures from isothermality observed near the center of x - ray clusters will most likely signal ongoing accretion / merger events or the presence of cooling flows . in the spherical top - hat model , the density of a virialized system is estimated by assuming that , at the time of collapse , the system has an equilibrium radius equal to approximately one - half of its turnaround radius . this virialized density contrast , @xmath38 , can be computed analytically and depends on the values of @xmath41 and @xmath0 at the time of collapse , as given in eq.(4 ) ( see lacey & cole 1993 , and eke et al 1996 for derivations ) . as explained in 2 , we have used this density contrast to guide our choice of the radius , @xmath3 , that encloses the virialized mass of a cluster . how well does this definition work ? we illustrate the situation in the left panels of figure 10 which show the mean radial velocity profiles at three different redshifts , averaged over the ten clusters . solid and dashed lines represent dark matter and gas components , respectively . these panels show three different regions easily identifiable in each system : ( i ) an inner region where the system is close to virial equilibrium , ie . @xmath154v@xmath155 ; ( ii ) a region dominated by infall , @xmath156v@xmath157 ; and ( iii ) an outer region where radial shells are still expanding away from the system , @xmath154v@xmath158 . the virialized region extends out to @xmath159 in all cases . this supports our choice of @xmath3 for characterizing the virialized mass of a system . at @xmath91 , @xmath160 , and @xmath161 this definition of virial radius corresponds to density contrasts of @xmath162 , @xmath163 , and @xmath164 , which correspond to _ overdensities _ of @xmath165 , @xmath166 , and @xmath167 , respectively . our results are therefore consistent with those of crone , evrard & richstone ( 1994 ) who found that , at overdensities of order @xmath168 or higher , clusters were close to hydrostatic equilibrium . figure 10 also shows that alternative definitions of the ` virial ' radius work reasonably well so long as they are referred to some specified density contrast . for example , the radius , @xmath169 , corresponding to density contrast @xmath170 , used by nfw97 ( vertical dotted lines to the left of @xmath171 ) , or the radius where the circular orbit timescale equals the current age of the universe ( vertical dotted lines to the right of @xmath171 ) also describe the virialized region of a cluster relatively well . the second row in figures 7 and 8 shows the density profiles of the gas and dark matter components at @xmath91 , scaled to the mean background density ; the radius is scaled to the virial radius in each case . as discussed by nfw95 ( see also crone et al 1994 ) , these scaled profiles look very similar , regardless of cluster mass . average profiles at @xmath91 , @xmath160 , and @xmath161 are shown in the right - hand panels of figure 10 . error bars indicate the standard deviation in the overdensity at each radius , computed using the ten most massive progenitors present at each redshift . a vertical arrow indicates the value of the gravitational softening . the fits to the dark matter profiles are obtained using eq.(3 ) . this formula clearly describes very well the dark matter profiles over about two decades in radius , from the gravitational softening out to the virial radius . near the center , the dark matter density increases monotonically inwards , and there is no indication that it approaches a well defined central value except for that imposed by the gravitational softening and the finite number of particles . the parameters of the fits are listed in table 3 . these parameters may be compared with the predictions of nfw97 , who found that the characteristic density , @xmath34 ( or , equivalently , the concentration , @xmath47 ) , is directly proportional to the mean density of the universe at the time of assembly of each system . following the procedure outlined in the appendix of their paper we computed the concentrations expected for clusters identified at @xmath91 and at @xmath172 in this particular cosmogony . the results are shown as solid ( @xmath91 ) and dashed ( @xmath172 ) lines in the upper panel of figure 11 . there is no free rescaling allowed in this comparison , so the agreement at @xmath91 ( open squares ) is impressive , especially because the concentrations derived analytically are expected to apply mainly to clusters near equilibrium , and not necessarily to an ensemble of systems chosen at random stages of evolution . according to the analytical procedure , at fixed mass the concentration is expected to _ decrease _ with increasing redshift . the same effect , although slightly more pronounced , is seen in the simulations . at @xmath172 , the predicted concentration is about @xmath173 larger than in the simulated clusters . this discrepancy is most likely the result of numerical limitations , since at @xmath174 clusters are not as well resolved as at @xmath91 . they contain @xmath175 times fewer particles , and the gravitational softening is about @xmath176 of the virial radius , compared with @xmath177 of @xmath3 at @xmath91 . it is therefore likely that the dark matter concentration has been underestimated at @xmath178 because of these effects . we have tested this directly by both increasing and decreasing ( by factors of @xmath179 ) the number of particles in three of the runs ( @xmath180 , @xmath181 , and @xmath182 ) . these extra runs were evolved until @xmath172 , and the dark matter concentration parameter was computed using the same procedure as before . the results are shown with connected starred symbols in the upper panel of figure 11 . as the number of particles increases so does the concentration . the highest resolution runs have @xmath183-@xmath97 particles within the virial radius , and in this case there is little difference between predicted and numerically determined concentrations . the density profile of the gas component differs significantly from the dark matter ( see figure 10 ) . the gas is less centrally concentrated , and , near the center , a well defined , constant density region ( a ` core ' ) is clearly apparent . this core extends beyond the region likely to be compromised by numerical limitations ( ie . the gravitational softening ) , especially at @xmath91 , when the clusters are best resolved . fits using eq.(3 ) give unacceptably large values of @xmath184 , so we decided instead to fit the gas profiles using the @xmath2-model ( cavaliere & fusco - femiano 1976 ) , traditionally used to fit x - ray data : @xmath185 this formula provides a good description of the gas profiles at all redshifts , as may be seen in figure 10 . the typical core radius is about @xmath186 kpc at @xmath91 ( table 3 ) , in good agreement with observations ( eg . jones & forman 1984 ) . on average , @xmath187 of the total x - ray luminosity originates within @xmath188 . clusters are , on average , less massive ( and smaller ) at high - redshift , so their core radii are expected to be smaller as well , by about a factor of two at @xmath178 . this expectation could , in principle , be tested against observations when accurate estimates of core radii at modest - to - high redshifts become available . if the gas and dark matter density profiles are not identical , but clusters evolve approximately as predicted by the scaling laws ( see figure 6 ) , then the profiles must remain proportional to each other , independently of cluster mass or redshift . that this is indeed the case may be seen by examining the fit parameters listed in table 3 . the ratio of the gas core radius to the dark matter scale radius , @xmath189 , remains approximately constant since @xmath190 , as does the ratio of characteristic densities , @xmath191 . why are the radial profiles of the gas and dark matter different but nevertheless remain proportional to each other ? inspection of figures 6 and 10 suggest that the explanation lies in the evolution of the net central entropy of the two components shown in the bottom left panel of figure 6 . by construction , at very early times ( the initial conditions ) the specific entropy of gas and dark matter is the same , but the entropy gain is different during the assembly of the first virialized structures . at later times the entropy gain is such that the ratio between gas and dark matter central entropies remains approximately constant . this difference has been noted and studied by navarro & white ( 1993 ) , pearce , thomas & couchman ( 1994 ) and nfw95 , and occurs as a result of the different behavior of collisional and collisionless fluids during mergers . since gas is stopped in shocks while the dark matter in merging subsystems can mix freely , the gas tends to ` lag behind ' during the assembly of the system , creating an effective phase difference that results in a net transfer of energy ( and entropy ) from the dark matter to the gas . thus in each merger event the gas and dark matter raise their entropies by different amounts . assuming that the central entropy gain of each component per merger event is always @xmath192 and @xmath193 for the gas and dark matter , respectively , the ratio between final entropies will tend to a constant , @xmath194 , as soon as the current entropy exceeds significantly the initial value . since the initial entropy is negligible , soon after the collapse and virialization of the first resolved clumps the gas and dark matter settle to equilibrium configurations that are not identical but which remain proportional to each other at all times . finally , since the gas distribution is effectively determined by that of the dark matter , limited numerical resolution can affect the gas central properties , and in particular estimates of the x - ray luminosity . this is shown in the lower panel of figure 11 , where we show how the x - ray luminosity depends on particle number for the convergence tests described above . x - ray luminosities , @xmath57 , are normalized to @xmath195 , the value expected from the scaling laws described in 2 ( see dashed line in upper left panel of figure 13 ) . clearly , more than about @xmath196-@xmath183 gas particles within the virial radius are needed in order to obtain convergent values of @xmath57 . we shall come back to this issue in 4.6 . as discussed by white et al ( 1993 ) , the mean baryonic mass fraction within the virial radius of a cluster is unlikely to exceed the universal baryon fraction . this is because , by definition , the virial radius separates the virialized region of the system from the region where shells of material are infalling for the first time . measurements of the baryon fraction in clusters can therefore be compared with independent estimates of the universal mean , @xmath197 , from , for example , big bang nucleosynthesis models , in order to constrain the value of the density parameter @xmath7 . this theoretical argument has led to significant interest in measurements of the baryon fraction in clusters ( david , jones & forman 1995 , white & fabian 1995 , loewenstein & mushotzky 1996 , gunn & thomas 1996 , evrard 1997 ) and to renewed theoretical and observational efforts to estimate @xmath198 as accurately as possible ( steigmann & tosi 1995 ; copi 1995 ; rugers & hogan 1996 ; tytler , fan & burles 1996 ; songaila , wampler & cowie 1997 ) . our simulations allow us to quantify possible biases in the cluster baryon fraction relative to the global mean . the cumulative gas ( baryon ) fraction , @xmath199 , for all ten simulated clusters at @xmath91 is shown as a function of radius in figure 12 . this fraction is expressed in units of the universal value ( @xmath200 ) assumed in the simulations . the gas fraction increases with radius because the gas distribution is less centrally concentrated than the dark matter distribution ( cf . @xmath201 ) . at @xmath91 , the gas fraction within @xmath3 is @xmath4-@xmath5 of the universal value but , within one gas core radius ( @xmath175-@xmath153 of the virial radius ; see table 3 ) , it can be as low as @xmath187 of @xmath197 . within @xmath202 , large variations , ranging from @xmath140 to more than @xmath203 of the global mean , are seen from cluster to cluster , reflecting differences in dynamical history . only at @xmath204 mpc ( close to the turnaround radius ) do the baryon fractions converge to the universal mean . because gas and dark matter densities remain approximately proportional at different times ( @xmath205 ) , these results are not expected to depend significantly on redshift . this is illustrated in the bottom panel of figure 12 , where we show , as a function of @xmath206 , the gas fractions within 3 different radii averaged over all ten clusters . to avoid overlaps , error bars have been chosen to represent standard deviations in the mean computed from the ten systems at each redshift , ie . they are @xmath207 times smaller than the scatter in the baryon fraction . to summarize , our simulations indicate that within the virial radius the baryon fraction in clusters provides a measure of the universal mean which is only mildly biased low ( by @xmath208 within @xmath209 ) . a similar result has been reported in most simulations published to date , and appears to be independent of cluster mass and of the values of @xmath7 or @xmath198 ( evrard 1990b ; thomas & couchman 1992 ; kang 1994 ; metzler & evrard 1994 ; nfw95 ; however see anninos & norman 1996 ) . the magnitude of this bias might change if additional physics are included in the simulations . for example , cooling may increase the baryon fraction near the center , while heating by non - gravitational processes acts in the opposite direction . however , even models with extreme cooling are unable to raise the baryon fraction within @xmath3 by more than @xmath150 ( white et al 1993 ) . similarly , extreme preheating models can not reduce it by more than @xmath210 , especially in rich clusters ( metzler & evrard 1994 , nfw95 ) . it is therefore unlikely that the baryon fractions within @xmath3 differ significantly from the universal mean . on the other hand , baryon fractions within @xmath211 can vary significantly from cluster to cluster , depending on the dynamical state of the system . this calls for caution when interpreting observational results based on measurements that probe only the inner regions of clusters and do not extend to @xmath212 ( see , eg . , loewenstein & mushotzky 1996 and evrard 1997 ) . figure 13 shows correlations between the main structural parameters of our simulated clusters : mass , velocity dispersion , x - ray luminosity , and x - ray emission - weighted temperature , as well as the redshift dependence of these correlations . open squares correspond to clusters at @xmath91 , and starred symbols to progenitors identified at @xmath172 . the solid lines in each panel show the scaling relations derived in 2 , at @xmath91 ; dashed lines illustrate the changes in slope and normalization at @xmath172 implied by the scaling laws . specifically , we use eq.(12 ) for the mass - temperature relation , eq.(11 ) for the temperature - velocity dispersion relation , and eq.(14 ) for the luminosity - temperature relation . the zero - point in the temperature - velocity dispersion relation is fixed by the usual ` beta'-parameter , defined by @xmath213 the solid line in the @xmath59-@xmath142 panel in figure 13 assumes @xmath214 . the upper dotted line corresponds to @xmath215 and the lower dotted line to @xmath216 . clearly @xmath214 is a very good approximation to the results of the simulations at all times . this is illustrated further in the right - hand panels of figure 14 , where we show the distribution of @xmath217 values at three different redshifts , @xmath91 , @xmath160 , and @xmath161 . the mean is indistinguishable from unity , and the dispersion is rather small ( @xmath218 ) . since our sample of clusters contains systems in different evolutionary stages , this suggests that unusually large or small observed values of @xmath217 are likely to be caused by systematic effects such as the inclusion of interlopers in the computation of the velocity dispersion , large differences between the galaxy and mass distributions , or temperature measurements affected by non - gravitational effects such as cooling flows . in an analogous way , we set the zero point of the mass - temperature relation through another ` beta'-parameter , @xmath219 thus @xmath220 relates the ` virial temperature ' of the system ( which depends only on the virial mass , see eq . 16 ) to the x - ray emission - weighted temperature , @xmath59 , which is accessible to observation . this parameter is crucial for relating observations to calculations that deal only with masses , such as n - body studies or analytic calculations based on the press - schechter theory ( see , eg . , eke et al 1996 ) . we assume @xmath221 in the scaling relations plotted in the mass - temperature panel of figure 13 . one drawback of this way of relating cluster mass and temperature is its dependence on our somewhat arbitrary definition of virial mass ( 2 ) , which makes no reference to the internal structure of the system . for example , two clusters with the same virial temperature ( or mass ) but different concentrations may have different x - ray temperatures because @xmath59 traces the depth of the potential well near the center , and this depends directly on the concentration . as discussed in @xmath141 , x - ray temperatures are likely related to the characteristic ( maximum ) circular velocity , @xmath67 , of the system rather than to the virial velocity or temperature . using eqs.(11 ) , ( 16 ) and ( 19 ) , we have , @xmath222 \over c}. \eqno(20)\ ] ] as shown in figure 11 , the average concentration of our simulated clusters increases with time , from @xmath223 at @xmath224 to @xmath225 at @xmath91 . thus , from eq.(20 ) we expect @xmath220 to decrease by about @xmath226 during this time interval . this expectation is in excellent agreement with the evolution reported in table 4 : @xmath220 decreases from @xmath227 at @xmath224 to @xmath228 at @xmath91 . this result is shown graphically in the bottom right panel of figure 6 , where the upper thick dashed line is the expected evolution of @xmath220 according to eq.(20 ) ( normalized to unity at @xmath91 ) , which should be compared to the solid line , which shows @xmath220 measured directly from the simulations . although noticeable , these changes in @xmath220 are small ( @xmath229 ) compared with the scatter in individual determinations at any given time ( @xmath230 ) , and therefore they are unlikely to have a large effect on mass determinations based solely on x - ray temperatures ( see , eg . , evrard , metzler & navarro 1996 ) . the values chosen to convert masses into temperatures by eke et al ( 1996 ) ( @xmath231 ) are in good agreement with these results . we normalize the luminosity - temperature relation predicted by eq . ( 14 ) by choosing the proportionality constant so that @xmath232kev@xmath233 erg s@xmath19 at @xmath91 . this relation is shown as a solid line in the top left panel of figure 13 . the dashed curve is the relation expected at @xmath172 using the same proportionality constant . the normalization of the @xmath57-@xmath70 relation follows from this and from the @xmath59-@xmath142 relation discussed above . at @xmath91 , there is good agreement between the analytical and numerical results , but at @xmath172 there are some discrepancies which , as we now discuss , are almost certainly due to numerical limitations . figure 13 shows that , in general , the scaling laws derived in 2 describe quite well the correlations between structural properties of simulated clusters and their evolution . at a fixed temperature , for example , clusters are expected to brighten by almost a factor of two at @xmath178 compared to clusters at @xmath91 . our simulated clusters are only slightly underluminous ( by about @xmath210 ) relative to this expectation . this is not entirely surprising because at @xmath224 poorer numerical resolution results in artificially low central densities , and a significant underestimation of the x - ray luminosity . indeed , according to figure 11 ( top panel ) , at @xmath172 the average concentration of the simulated clusters is @xmath234 , whereas the expected concentration is almost @xmath187 higher . plugging these numbers into the concentration - dependent factors of eq.(14 ) , @xmath235 , we find that this effect alone can lead to underestimates of the total x - ray luminosity of about @xmath236 , consistent with the deviations observed in the @xmath237-@xmath59 panel of figure 13 . further support for this interpretation is provided by the lower panel in figure 11 , which shows that convergent x - ray luminosity estimates require more than about @xmath196-@xmath183 gas particles within the virial radius . most clusters have fewer gas particles than that at @xmath178 , so systematic underestimation of @xmath57 is expected . the same argument helps explain why clusters at @xmath172 are slightly underluminous in the @xmath57-@xmath142 panel . in summary , the scaling laws are in very good agreement with the evolution of the x - ray properties of simulated clusters . we discuss below how these results compare with previous work and with observations . figure 15 compares the luminosity - temperature relation obtained from our simulations with that found in other numerical studies . at @xmath91 , this relation ( open squares ) compares well with that obtained by nfw95 , evrard ( 1990 , labeled e90 ) , and bartelmann & steinmetz ( 1996 , labeled bs96 ) for @xmath238 cdm , once they are all scaled to the same gas mass fraction , @xmath120 . this reflects the similarity of the structure of systems formed through hierarchical clustering ( nfw96 , nfw97 ) . the curves labeled co94 , k94 , and b94 correspond to the work of cen & ostriker ( 1994 ) , kang et al ( 1994 ) , and bryan et al ( 1994 ) , respectively . k94 and b94 considered clusters in an @xmath238 cdm universe , while co94 simulated clusters in an @xmath239 , @xmath240 cdm universe . the slope and normalization of the @xmath57-@xmath59 relation found by these authors are in disagreement with the results of this paper and of nfw95 . one simple explanation of this discrepancy is that the luminosities found by these other authors are severely compromised by numerical resolution . indeed , the spatial resolution in the simulations of co94 , k94 and b94 ( ie . the grid cell size in their eulerian codes ) is @xmath241 mpc , about three times larger than the core radii of our clusters ( see table 3 ) . the core radii in the eulerian simulations are thus largely set by the mesh size , and the corresponding x - ray luminosities are only lower bounds to the actual luminosity . several lines of argument support this conclusion : ( i ) the core radii in co94 , k94 and b94 are independent of cluster mass or temperature . this occurs because all clusters , regardless of mass or physical size , are analyzed in one single simulation with the same grid size , which fixes the core radii . in our work and in nfw95 , the core radius is found to be proportional to the characteristic scale - length of the dark halo , and scales roughly as the virial radius of the system , @xmath242 . ( ii ) the slope of the @xmath57-@xmath59 relation in co94 , k94 and b94 is steeper than expected from the scaling laws . this is also easily explained if the gas core radii just reflect the spatial resolution of the calculation . since cooler clusters are smaller in size , artificially fixed core radii affect cooler clusters more severely ( ie . the core radius is a larger fraction of the virial radius ) , systematically depressing their x - ray luminosity . ( iii ) finally , at a given temperature , the luminosities found by k94 and b94 differ by about a factor @xmath243 , even though they model clusters in exactly the same cosmogony using the same grid size , so at least one of them can not be right . k94 and b94 use different numerical techniques , so the discrepancy is most likely due to the fact that in both cases the luminosity estimates are compromised by numerical resolution , but the _ effective _ spatial resolution is different in the two techniques . our interpretation is similar in spirit to that of anninos & norman ( 1996 ) , who also argued that the x - ray luminosity of clusters in co94 , k94 , and b94 had been severely underestimated . however , we disagree with anninos & norman s conclusion that the x - ray luminosity of x - ray clusters does not converge as the numerical resolution is improved . nfw95 show that provided that adequate mass and spatial resolution are used , the x - ray luminosity is very robust to changes in the numerical parameters ( see figure 13 of nfw95 and the lower panel of figure 11 above ) . indeed , the key to the convergence of the x - ray luminosity lies in resolving the core radius of the gas which , according to table 3 , is @xmath244 . an effective spatial resolution better than @xmath245 of the virial radius , and matching mass resolution ( typically several thousand particles within @xmath246 ) , are thus required for robust estimates of the x - ray luminosity . a very small fraction of the total luminosity comes from within the core because the gas density profile is shallower than @xmath247 , ensuring convergence of the total x - ray luminosity . a similar conclusion has been reached independently by bryan & norman ( 1997 ) . further support for this conclusion comes from the test cluster comparison project mentioned in 1 , in which the same cluster was simulated with different numerical techniques and varying resolution . simulations that are able to resolve @xmath202 give similar x - ray luminosities ( frenk , in preparation ) . it is worth emphasizing that there is nothing intrinsically wrong with the numerical techniques used by co94 , k94 and b94 . as the cluster comparison test demonstrates , sph codes compare well with eulerian codes and give essentially the same results in regions that are adequately resolved by both . figure 16 compares the @xmath57-@xmath59 relation measured for nearby ( @xmath248 , open circles ) and distant ( @xmath249 , filled circles ) clusters with the scaling laws derived in @xmath141 and calibrated as described in @xmath250 . all luminosities are bolometric and have been converted to @xmath251 . solid lines correspond to @xmath91 , and dashed lines to @xmath252 for two different cosmologies . the models have been normalized so that they agree at @xmath91 . the observed @xmath57-@xmath59 relation is steeper than predicted by the scaling laws . although there is good agreement at the bright end , cool ( @xmath253 kev ) clusters are much fainter than expected . this discrepancy was also noticed by nfw95 , who argued that it was indicative of the role played by some non - gravitational mechanism , such as radiative cooling , consumption of gas into galaxies , or pre - heating , in establishing the x - ray properties of clusters . ( see also kaiser 1991 , evrard & henry 1991 , bower 1997 , for similar arguments . ) for example , nfw95 showed that the observed @xmath57-@xmath59 relation can be reproduced if the gas in clusters were preheated to a common entropy before cluster assembly , a conclusion similar to that reached by metzler & evrard ( 1994 ) . the good agreement between simulations and observations for hot ( @xmath254 kev ) is intriguing , and suggests that preheating is unimportant at these very high temperatures . this characteristic preheating entropy is an important clue that may allow us to constrain the mode and timing in which the preheating process operates ( ponman , cannon & navarro , in preparation ) . at the modest redshifts probed by the data ( clusters represented with filled circles span the redshift range @xmath137-@xmath255 , with a mean of @xmath256 , see mushotzky & scharf 1997 ) , there is little evidence for evolution in the slope or zero point of the @xmath57-@xmath59 relation ( see also henry , jiao & gioia 1994 and tsuru 1996 ) . this , in fact , is in good agreement with the weak evolution expected from the scaling laws ( eq.14 ) . the short- and long - dashed lines in figure 16 show the @xmath57-@xmath59 relation predicted by eq.(14 ) at @xmath252 for two different cosmogonies : a low - density and an einstein - de sitter cdm model , respectively . at a given temperature , clusters are expected to be no more than @xmath257 more luminous at @xmath258 than at @xmath91 . such a small difference would be very difficult to disentangle from the large scatter in the observed @xmath57-@xmath59 relation . we conclude that the evolution of the luminosity - temperature relation is consistent with that predicted by the scaling laws . the scatter in the observed @xmath57-@xmath59 relation is about a factor of two to three larger than found in the simulations , although the large errors associated with many temperature measurements quoted in the literature make this a difficult point to assess . provided errors are not the main source of the scatter , and since simulated clusters are at various stages of dynamical evolution , it appears that deviations from equilibrium can not be solely responsible for the observed scatter . further support for this assertion comes from the fact that deviations from the average @xmath57-@xmath59 relation correlate with the strength of the central cooling flow ( fabian et al 1994 ) . figure 17 compares the dark matter projected density and line - of - sight velocity dispersion profiles , averaged over our ten simulated clusters ( solid lines ) , with the corresponding profiles for _ galaxies _ in the clusters studied by the cnoc project ( points with error bars , taken from carlberg , yee & ellingson 1997 ) . the cnoc data are based on @xmath259 redshifts collected in the fields of 16 x - ray luminous clusters at @xmath260 . all cluster profiles ( observed and simulated ) have been rescaled to their individual virial radius prior to averaging . the only free normalization is that of the dark matter density , which has been chosen to match the galaxy number density data . the dark matter and galaxy profiles are remarkably similar thus , the structure of clusters in our low - density cdm model is consistent with the assumption that galaxies are , on average , fair tracers of the mass distribution in clusters . this is a useful result because one of the major uncertainties that plagues the analysis of dynamical data such as those collected by the cnoc group is the extent to which galaxies may be spatially segregated or dynamically ` biased ' relative to the surrounding matter . these ` biases ' are extremely difficult to detect and measure observationally . neglecting them in a virial analysis of the cnoc data leads to an estimate of @xmath261 ( carlberg 1996 ) , consistent with our assumed value of @xmath13 . thus , our results suggest that there is little segregation between mass and galaxies in x - ray luminous clusters , as concluded independently by the cnoc team ( carlberg 1996 ) . it is unclear , however , whether this conclusion is compatible with the large - scale clustering properties of a flat , @xmath13 cdm universe which require galaxies to be significantly antibiased relative to the mass on cluster scales ( jenkins 1997 ) . we have used n - body / gasdynamical simulations to study the structure and evolution of x - ray clusters formed in a low - density cdm universe ( @xmath13 , @xmath14 , @xmath81 , @xmath262 ) . the simulations include gravity , pressure gradients and hydrodynamical shocks , but neglect the effects of radiative cooling or of galaxy formation . a summary of our main conclusions follows . \(1 ) the density profiles of clusters of different mass identified at various redshifts are described accurately by the fitting formula proposed by navarro , frenk & white ( eq . 3 ; see also nfw95 and nfw96 ) . the parameters of the fit are in good agreement with the analytical model proposed by these authors ( nfw97 ) . this formula provides an adequate description of the mass profile over approximately two decades in radius , out to the ` virial ' radius beyond which infall dominates . the extent of this virialized region is consistent with a definition of ` virial radius ' based on the spherical top - hat collapse model . \(2 ) the structure of cluster dark matter halos is in excellent agreement with the distribution and dynamics of galaxies in the clusters analyzed by the cnoc project . this is consistent with the idea that galaxies and dark matter in clusters are not spatially segregated or dynamically biased to a significant degree . \(3 ) the gas density profiles of simulated clusters differ significantly from the dark matter profiles and are better described using the @xmath2-model ( eq.17 ) . however , the gas and dark matter density profiles remain proportional to each other regardless of cluster mass and redshift . numerical estimates of the x - ray luminosity converge quickly when the scale radius of the dark matter and the core radius of the gas are resolved numerically . for an sph simulation , this typically requires @xmath263 particles per cluster and an effective spatial resolution better than about one percent of the virial radius . \(4 ) the structural similarity between the dark and gas components implies that simple scaling laws relate the mass , velocity dispersion , temperature , and x - ray luminosity of galaxy clusters . these scaling laws can be derived using the fact that clusters of a given mass are described by a single free parameter : their characteristic density or concentration . these laws extend the scale - free relations of kaiser ( 1986 ) to universes with @xmath264 and perturbation spectra different from power - laws . the predictions of these scaling laws , as a function of cluster mass and redshift , are in remarkable agreement with the results of the simulations . this provides an impressive validation of the smooth particle hydrodynamics technique . \(5 ) the x - ray luminosity in simulated clusters scales approximately as the square of the temperature , roughly as predicted by the scaling laws . this is a shallower dependence of @xmath57 on @xmath59 than is observed for x - ray clusters . we interpret this disagreement as requiring additional physical processes not included in these simulations ( eg . radiative cooling or preheating ) to account for the x - ray properties of clusters , particularly of low - temperature ( @xmath265 kev ) systems . this relation is expected to evolve only weakly with redshift . at a given temperature clusters are not expected to brighten by more than @xmath257 at @xmath260 , consistent with published measurements . \(6 ) the average baryon fraction within the virial radius is @xmath4-@xmath5 of the universal mean , @xmath197 , and is lower in the inner regions . this result calls for caution when interpreting the baryon fraction measured in clusters in terms of the universal mean . the inclusion of physical processes neglected here , such as radiative cooling , may affect the cluster baryon fraction although such effects are likely to be small . \(7 ) x - ray emission - weighted temperatures can be used to estimate reliably the total mass and velocity dispersion of clusters ( eqs . 18 and 19 ) . these estimators are essentially unbiased and have small scatter , @xmath230 for the mass - temperature , and @xmath266 for the mass - velocity relations . a consequence of this is that semianalytical techniques and n - body simulations can be used to predict the statistical properties of x - ray clusters in different cosmological models without the need for expensive hydrodynamical simulations . physical processes not included in our simulations , such as radiative cooling , galaxy formation , or non - gravitational heating may all have a significant effect on the temperature of the intracluster medium and on the x - ray luminosity of galaxy clusters . we have chosen to neglect them in this study , and to concentrate instead on the simpler ` adiabatic ' evolution of gas within an evolving population of dark matter halos . the failure of ` adiabatic ' clusters to reproduce the observed luminosity - temperature relation indicates that additional physics must be included in the numerical modeling in order to develop a full understanding of the origin and evolution of the x - ray properties of galaxy clusters . we are currently working on these issues . we thank douglas heggie for supporting the grape hardware in edinburgh . hugh couchman kindly made available his excellent ap@xmath82 m code , and ray carlberg provided the cnoc data used in figure 17 . jfn acknowledges useful discussions with matthias steinmetz . vre acknowledges the support of a pparc studentship and a pparc postdoctoral fellowship , and csf a pparc senior research fellowship . this research has been supported by the uk pparc . tsuru t. , koyama k. , hughes j. p. , arimoto n. , kii t. , hattori m. , 1996 , eds watanabe t. , yamashita k. , `` the 11th international colloquium on uv and x - ray spectroscopy of astrophysical and laboratory plasmas '' , in press + tytler d. , fan x .- , burles s. , 1996 , nat , 381 , 207

we study valuation of swing options on commodity markets when the commodity prices are driven by multiple factors . the factors are modeled as diffusion processes driven by a multidimensional lvy process . we set up a valuation model in terms of a dynamic programming problem where the option can be exercised continuously in time . here , the number of swing rights is given by a total volume constraint . we analyze some general properties of the model and study the solution by analyzing the associated hjb - equation . furthermore , we discuss the issues caused by the multi - dimensionality of the commodity price model . the results are illustrated numerically with three explicit examples .

1302.6399

the purpose of this paper is to propose and analyze a model for valuation of a swing option , see , e.g. @xcite , written on multiple commodities when the commodity spot prices are driven by multiple , potentially non - gaussian factors . more precisely , the model is formulated as a dynamic programming problem in continuous time . the holder of the option is contracted an amount of a given commodity that can be purchased for a fixed price during the lifetime of the contract . the purchases can be done ( that is , the option can be exercised ) continuously in time such that contracted rate constraints are fulfilled . this form of contract originates from electricity markets , where they are called flexible load contracts , see , e.g. @xcite . however , this model setting can also fit a traditional swing option with a high number of swing rights and possible exercise times . for example , we can think of a situation in an electricity market where contract is written for a year and holder can exercise on the hour - ahead market . this results into over 8000 possible exercise times , which makes , in particular , monte - carlo methods virtually intractable . during the recent years , there has been a lot of activity on analysis of swing options . being essentially a multi - strike american or bermudan option , a natural way to approach swing options is via an optimal multiple stopping problem . in the recent papers @xcite , the theory of optimal multiple stopping is developed in continuous time using sophisticated martingale theory . to compute option prices numerically , they develop appropriate monte - carlo methodology . other methodology for swing option pricing includes forests of trees @xcite or stochastic meshes @xcite , multi - stage stochastic optimization @xcite , ( quasi-)variational inequalities @xcite and pde approaches @xcite . fundamentally , all of these methods are based on the dynamic programming principle . as the main contribution of this paper , we develop a valuation model for multi - commodity swing options inspired by @xcite . in @xcite , the valuation problem was studied in the case of a single commodity driven by a one - factor gaussian price process . in this paper , we generalize the results of @xcite to cover multiple contracted commodities with prices driven by multiple factors . from applications point of view , this is an important generalization , since there is a substantial body of literature supporting the usage of multi - factor models for commodity prices . moreover , we allow also for non - gaussian factors , which are favored , for example , in electricity price models , see , e.g. @xcite . we model the factors as a multi - dimensional lvy diffusion and the underlying commodity prices are obtained by a linear mapping of the factors . this makes our model more tractable yet keeping it still very flexible as it allows us to take , for example , heat rates and spreads into account in a natural way . our study is also related to @xcite , where a similar model is used to study hedging of swing options . we also refer to @xcite , where swing option pricing is considered under a non - gaussian multi - factor price model . however , the analysis of @xcite is restricted to a modification of the so - called deng model ( see @xcite ) , which is a particular mean - reverting model . in our paper , we set up and analyze a class of models where the underlying factor prices follow a general lvy diffusion . the existing mathematical literature on swing options is mostly concerned with the pricing of a swing option . in addition to pricing , we also address the question of how to exercise a swing option optimally . from the analytical point of view , we identify using the hjb - equation an optimal exercise policy and characterize it in an intuitive way using the notion of marginal lost option value . we also present a numerical analysis of the problem including a numerical scheme based on the finite difference method . the reminder of the paper is organized as follows . in section 2 we propose our model for the valuation of swing options . in section 3 we analyze some general properties of the value function . section 4 is devoted to the derivation of necessary and sufficient conditions for a function to coincide with the value function . we illustrate our results with explicit examples in section 5 , which are solved numerically in section 6 . finally , we conclude in section 7 . as we mentioned in the introduction , the prices of the commodities are driven by multiple factors . throughout the study , the number of commodities is @xmath0 and the number of driving factors is @xmath1 . the factor dynamics @xmath2 are modeled by an @xmath1-dimensional lvy diffusion . to make a precise statement , let @xmath3 be a complete filtered probability space satisfying the usual conditions , where @xmath4 is the filtration generated by @xmath2 . we assume that the factor process @xmath2 are given as a strongly unique solution of the it equation @xmath5 where @xmath6 is an @xmath7-dimensional , potentially correlated , brownian motion satisfying @xmath8 with @xmath9 $ ] for all @xmath10 . furthermore , @xmath11 denotes an @xmath12-dimensional poisson random measure with lvy measure @xmath13 given by the independent poisson processes @xmath14 . here , @xmath15 is the unit measure concentrated on zero and it is finite . the coefficients @xmath16\times{\ensuremath{\mathbb{r}}}^n\rightarrow{\ensuremath{\mathbb{r}}}^n$ ] , @xmath17\times{\ensuremath{\mathbb{r}}}^n\rightarrow{\ensuremath{\mathbb{r}}}^{n\times n_b}$ ] and @xmath18\times{\ensuremath{\mathbb{r}}}^n\times{\ensuremath{\mathbb{r}}}^{n_l}\longrightarrow{\ensuremath{\mathbb{r}}}^{n\times n_l}$ ] are assumed to be sufficiently well behaving lipschitz continuous functions to guarantee that the it equation has a unique strong solution see @xcite , p. 365 the motivation to model the randomness using brownian and finite activity jump noise comes from electricity prices . in this framework , the jump process models the spiky behavior in the prices whereas the brownian motion takes care of the small fluctuations . using the factor dynamics @xmath2 , we define the @xmath0-dimensional price process @xmath19 via the linear transformation @xmath20 where @xmath21 and @xmath22 is a constant @xmath23 matrix with @xmath24 . in other words , there exists constants @xmath25 such that @xmath26 for all @xmath27 , that is , the commodity prices are linear combinations of the driving factors . the component @xmath28 models the time evolution of the price of the @xmath29th commodity and this price is driven by the @xmath1 factors , i.e. the @xmath1-dimensional lvy diffusion @xmath2 given as the solution of the it equation . since the price is linear as a function of the factors it is easy to change the model into a price model for spreads . furthermore , the matrix @xmath22 in can be interpreted as a constant weight between the different factors @xmath2 affecting the price . that allow us to take , for example , heat rates into account in our model . in the definition of the factor dynamics , we assumed that the jump - diffusion @xmath2 and the driving brownian motion and lvy process have all different dimensions . for notational convenience , we assume in what follows that these dimensions are the same , i.e. @xmath30 . we point out that the following analysis holds with obvious modifications also in the case where these dimensions are different . the swing option written on the price process @xmath31 gives the right to purchase the given amount @xmath32 of the commodities @xmath29 over the time period @xmath33 $ ] . in addition to the global constraint @xmath32 , the purchases are also subject to a local constraint @xmath34 which corresponds to the maximal number of swing rights that can be exercised on a given time . since the swing option can be exercised in continuous time , the local constraint is the maximum rate at which the option can be exercised . to formalize this , let @xmath35 be the set of @xmath36-measurable , real - valued processes @xmath37 satisfying the constraints @xmath38 , \ \int_0^t u^i_s ds \leq m^i,\ ] ] for all @xmath27 and @xmath39 $ ] . here , the elements @xmath40 and @xmath41 . the @xmath42-valued process @xmath43 defined as @xmath44 where @xmath27 , keeps track of the amount purchased of commodity @xmath29 up to time @xmath45 . in what follows , we call @xmath43 the total volume and denote the product @xmath46 as @xmath47 . the integral representation for @xmath48 in is well defined due to the local constraint . denote the set @xmath49\times\bigotimes_{i=1}^m[0,m_{i } ] \times { \ensuremath{\mathbb{r}}}^m$ ] and define the affine function @xmath50 as @xmath51 where @xmath52 is an @xmath53 matrix and @xmath54 . define the expected present value of the total exercise payoff @xmath55 given by the rate @xmath56 from time @xmath45 up to the terminal time @xmath57 ( or , the performance functional of @xmath58 ) as @xmath59,\ ] ] where @xmath60 is the constant discount factor . we point out that function @xmath61 is defined explicitly as a function of the factors @xmath2 . this corresponds to that the holder of the contract observes the underlying factor and bases her exercise decisions of this information . furthermore , we remark that this framework covers essentially call- and put - like payoffs , where the strike prices are given by the constant vector @xmath62 . now , the value function @xmath63 is defined as @xmath64 we denote an optimal rate as @xmath65 . we make some remarks on the valuation problem . the dimension of the decision variable @xmath58 is the same as the dimension of the price . that is , we can exercise the option for each price component , which corresponds to different commodities , with a different decision variable . furthermore , we defined the function @xmath66 such that it takes values in @xmath42 . this is done for notational convenience . suppose that we have an @xmath0-dimensional price process but the decision variable @xmath58 is @xmath67-dimensional with @xmath68 . this corresponds to the case where @xmath0 commodities are bundled into @xmath67 baskets and the holder can exercise the option on the baskets . formally this is done by defining the affine function as @xmath69 . this will not affect the form of the value function . in this section we study some general properties of the valuation problem . we split the analysis in two cases , depending on whether @xmath70 or @xmath71 for a given commodity @xmath29 . in the latter case , the limit @xmath72 imposes an effective constraint on the usage of the option in the sense that the amount @xmath72 is dominated by the amount that can be purchased if the option is exercised on full rate over the entire time horizon . this case , i.e. the case when an effective volume constraint is present , is the interesting one from the practical point of view . it is also substantially more difficult to analyze mathematically as we will see later . before considering this case , we study the complementary case when the effective volume constraint is absent . this will give us a point of reference in the other case . we consider first the case where @xmath73 for a given commodity @xmath29 . the total volume constraint for the commodity @xmath29 is now superfluous , since it is possible for the holder to exercise the option at full rate throughout the lifetime of the contract . in the absence of an effective volume constraint for the commodity @xmath29 , an optimal exercise rule is given by the next proposition . [ prop1 ] assume that @xmath73 for a given commodity @xmath29 . then an optimal exercise rate @xmath74 for the commodity @xmath29 reads as @xmath75 for all @xmath76 $ ] . let @xmath77 and @xmath78 $ ] . first , we observe that @xmath79 . furthermore , we find that @xmath80 \\&=\mathbf{e}\left [ \int_{t}^{t}e^{-r(s - t)}a^{i}(p_{t})u_{s}^{i}\mathbf{1}_{\{a^{i}(p_{t})\leq 0\}}ds\right . \left . \vphantom{\int_{t}^{t } } \right| \left . \vphantom{\int_{t}^{t } } z_t = z , \ x_t = x\right ] \\&\quad+\mathbf{e}\left [ \int_{t}^{t}e^{-r(s - t)}a^{i}(p_{t})u_{s}^{i}\mathbf{1}_{\{a^{i}(p_{t})>0\}}ds\right . \vphantom{\int_{t}^{t } } \right| \left . \vphantom{\int_{t}^{t } } z_t = z , \ x_t = x\right ] \\&\quad+\mathbf{e}\left[\int_{t}^{t}e^{-r(s - t)}\sum_{\substack{l=1 \\ l\neq i}}^{m } a^{l}(p_{t})u_{s}^{l}ds\right . \left . \vphantom{\int_{t}^{t } } \right| \left . \vphantom{\int_{t}^{t } } z_t = z , \ x_t = x\right ] \\&\leq \mathbf{e}\left[\int_{t}^{t}e^{-r(s - t)}a^{i}(p_{t}){u_{s}^{i}}^*ds\right . \left . \vphantom{\int_{t}^{t } } \right| \left . \vphantom{\int_{t}^{t } } z_t = z , \ x_t = x\right ] \\&\quad+\mathbf{e}\left[\int_{t}^{t}e^{-r(s - t)}\sum_{\substack{l=1 \\ l\neq i}}^{m } a^{l}(p_{t})u_{s}^{l}ds\right . \left . \vphantom{\int_{t}^{t } } \right| \left . \vphantom{\int_{t}^{t } } z_t = z , \ x_t = x\right ] . \end{split}\ ] ] now , take supremum over all @xmath81 on the left hand side and supremum over @xmath81 , @xmath82 , on the right hand side of . since the functional @xmath61 is linear in @xmath58 , the same inequality still holds and , consequently , the conclusion follows . proposition [ prop1 ] states that in the absence of an effective volume constraint for commodity @xmath29 , it is optimal to exercise the option whenever the payoff @xmath83 is positive , i.e. when @xmath84 . this is a natural result , since the holder does not have to worry of running out of the option over the planning horizon . furthermore , since @xmath34 is a constant vector , we find using proposition [ prop1 ] that the value function does not depend on @xmath85 in the absence of an effective volume constraint for commodity @xmath29 . this yields the following corollary . [ lemma1 ] in the absence of an effective volume constraint for a given commodity @xmath29 , the marginal value @xmath86 . corollary [ lemma1 ] is also a very natural result . indeed , if the holder uses the option on a commodity with no effective volume constraint , the option will not lose value . to close the subsection , we discuss how the dimension of the range of the function @xmath66 affects the value given by . for simplicity , assume that there is no effective volume constraint for any of the commodities and that the function @xmath50 is of the form @xmath87 for @xmath88 . using proposition [ prop1 ] , we know that the optimal exercise rule for the valuation problem specified by the payoff structure is @xmath89 for all @xmath76 $ ] and @xmath90 . formally , we can decrease the dimension of the range of @xmath66 from @xmath0 , for example , as follows . take @xmath91 and define the @xmath92-matrix @xmath93 such that each @xmath94 occurs only once and on exactly one column of @xmath95 and the other elements are zero . in financial terms , this means that the commodities @xmath29 are bundled into @xmath96 pairwise disjoint baskets with weights @xmath94 . then the option gives exercise rights on each of these baskets with separate exercise rates . now , let the function @xmath97 be @xmath98 with @xmath99 such that @xmath100 for all @xmath101 . using the same reasoning as in proposition [ prop1 ] we find that the optimal exercise rule for the valuation problem given by @xmath102 is @xmath103 for all @xmath76 $ ] and @xmath104 . denote the value for @xmath0-dimensional ( @xmath96-dimensional ) problem as @xmath105 ( @xmath106 ) . furthermore , denote the @xmath96-dimensional total volume variable as @xmath107 and assume that all maximal exercise rates coincide : @xmath108 for all @xmath90 and @xmath109 . then , due to the structure of matrix @xmath95 , we find using that @xmath110 \\&\geq \mathbf{e}\left[\int_t^t e^{-r(s - t ) } \sum_{i=1}^{m'}\left(\sum_{\substack{j=1 \\ \hat{q}_{ij}\neq 0}}^m ( \hat{q}_{ij}p^j_s + \hat{k}_j)\bar{u}\mathbf{1}_{h^i_s } \right)ds \right . \vphantom{\int_{t}^{t } } \right| \left . \vphantom{\int_{t}^{t } } \hat{z}_t=\hat{z } , \ x_t = x\right ] \\&= \hat{v}(t,\hat{z},p ) , \end{split}\ ] ] where the events @xmath111 and the cumulative variable @xmath107 is defined analogously to . summarizing , we have shown that by bundling commodities @xmath29 into mutually disjoint baskets and , thus , reducing the dimension of the exercise rate process @xmath58 , we lower the value of the option . this is , again , a natural result , since the bundling of commodities lowers flexibility of option contract in the sense that the holder must exercise the option at the same rate for all commodities in the same basket . this is in contrast to the case with separate commodities , where the exercise rates can be decided individually for each commodity . in this section , we consider the case where @xmath112 , in other words , the case when the total volume constraint is less than the maximal amount of commodity that can be acquired over the lifetime of the option . from the practical point of view , this is the interesting case . it is also substantially more difficult to analyze , since in this case we can not find an optimal exercise policy explicitly as in proposition @xmath113 . instead we find the value function as the solution to the hjb - equation and an optimal exercise policy is obtained as a biproduct . our first task is to write the conditional expectation in such that it depends explicitly on @xmath43 . this will be helpful in the later analysis . to this end , define the process @xmath114 as @xmath115 . then the it formula yields @xmath116 note that since @xmath66 is affine and @xmath117 is linear , we have @xmath118 @xmath119 @xmath120 where @xmath121 , @xmath122 and @xmath123 are constants for all @xmath124 . substitution of , and into yields @xmath125 since @xmath126 where @xmath13 is the lvy measure , we find that can be written as @xmath127 by integrating this from @xmath45 to @xmath57 , we obtain @xmath128ds\nonumber\\ & + \int_{t}^{t}e^{-rs}\sum_{l=1}^{m}a^{l}(p(x_{s}))u_{s}^{l}ds + \int_{t}^{t}e^{-rs}\sum_{i=1}^{n}\sum_{l=1}^{m}z_{s}^{l}c_{li}\sigma_{i}(s , x_{s})dw_{s}\nonumber \\ & + \int_{t}^{t}\sum_{k=1}^{n}\int_{{\ensuremath{\mathbb{r}}}}e^{-rs}\sum_{l=1}^{m}z_{s}^{l}\sum_{v=1}^{m}q_{lv}p^{v}(\gamma^{(k)}(s , x_{s^{-}},\xi^{k}))\tilde{n}^{k}(ds , d\xi^{k}).\end{aligned}\ ] ] consider first the brownian integral @xmath129 . each of the integrands is of the form @xmath130 . by definition of @xmath131 , we know that @xmath132 . since @xmath131 is nondecreasing , it follows that @xmath133 . hence , @xmath134\nonumber\\ & \leq ( \bar{u}^{l}tc_{li})^{2}\mathbf{e}\left[\int_{0}^{t}\sigma^2_{ij}(s , x_{s})ds\right . \left . \vphantom{\int_{0}^{t } } \right| \left . \vphantom{\int_{t}^{t } } z_0=z , \ x_0=x\right]<\infty.\end{aligned}\ ] ] using a martingale representation theorem , see , e.g. @xcite , thrm . 5.3.6 , we conclude that @xmath135 is a martingale with respect to @xmath36 . using the same argument , we find that the process @xmath136 is a also a martingale with respect to @xmath36 . consequently , the conditional expectation with respect to @xmath137 is zero for the last two terms in . by multiplying with @xmath138 on both sides , substituting into @xmath139 and using the martingale properties , we find @xmath140.\end{aligned}\ ] ] using that the measurability of @xmath141 , we can express the value function as @xmath142.\end{aligned}\ ] ] we now have an explicit dependence on @xmath43 in the value function , which will be useful in the proof of the following proposition . we point out that we can assume that we have an effective volume constraint in all commodities @xmath29 , since the complementary case is already covered by proposition [ prop1 ] . [ vznonpositive ] in the presence of an effective volume constraint , i.e. when @xmath143 , the marginal value @xmath144 for all @xmath29 . let @xmath145 be processes giving rise to admissible exercise policies @xmath146 at time @xmath45 . let @xmath147 be the processes giving rise to admissible exercise policies @xmath148 at time @xmath45 . since the exercise policies @xmath149 arising from @xmath150 are admissible and must satisfy the effective volume constraint we have that @xmath151 . also , for an arbitrary admissible @xmath152 on @xmath153 $ ] , define an associated @xmath154 as @xmath155 for all @xmath156 $ ] . with this in mind , we proceed by expressing the marginal value as @xmath157ds\left . \vphantom{\int_{t}^{t } } \right| \left . \vphantom{\int_{t}^{t } } z_t = z , \ x_t = x\right ] \nonumber \\ & & -\:\sup_{u}\mathbf{e}\left[e^{-r(t - t)}\sum_{l=1}^{m}a^{l}(p(x_{t}))z^{l}(t ) \right . \nonumber \\ & & -\ : \int_{t}^{t}\left[-re^{-r(s - t)}\sum_{l=1}^{m}a^{l}(p(x_{s}))z^{l}(s)+\ : e^{-r(s - t)}\sum_{i=1}^{n}\sum_{l=1}^{m}z^{l}(s)c_{li}\alpha_{i}(s , x_{s})\right.\nonumber\\ & & + \:\sum_{k=1}^{n}\int_{{\ensuremath{\mathbb{r}}}}e^{-r(s - t)}\sum_{l=1}^{m}z^{l}(s)\nonumber\\ & & \times\left.\left.\sum_{v=1}^{m}q_{lv}p^{v}(\gamma^{(k)}(s , x_{s^{-}},\xi^{k}))\nu^{k}(d\xi^{k})\right]ds\left . \vphantom{\int_{t}^{t } } \right| \left . \vphantom{\int_{t}^{t } } z_t = z , \ x_t = x\right]\right\}.\end{aligned}\ ] ] by collecting the terms containing @xmath158 in @xmath159 and taking out the @xmath160th term in the supremum over @xmath58 , we obtain @xmath161ds\left . \vphantom{\int_{t}^{t } } \right| \left . \vphantom{\int_{t}^{t } } z_t = z , \ x_t = x\right]\nonumber\\ & & + \sup_{u^{\varepsilon}}i_{1}-\sup_{u}i_{1}\nonumber\\ & & -\sup_{u}\mathbf{e}\left[e^{-r(t - t)}a^{j}(p(x_{t}))z^{j}(t)\right.\nonumber\\ & & -\:\int_{t}^{t}\left[-re^{-r(s - t)}a^{j}(p(x_{s}))z^{j}(s)+e^{-r(s - t)}\sum_{i=1}^{n}z^{j}(s)c_{ji}\alpha_{i}(s , x_{s})\right.\nonumber\\ & & + \:\left.\sum_{k=1}^{n}\int_{{\ensuremath{\mathbb{r}}}}e^{-r(s - t)}z^{j}(s)\sum_{v=1}^{m}q_{lv}p^{v}(\gamma^{(k)}(s , x_{s^{-}},\xi^{k}))\nu^{k}(d\xi^{k})\right.]ds\left . \vphantom{\int_{t}^{t } } \right| \left . \vphantom{\int_{t}^{t } } z_t = z , \ x_t = x\right]\bigg\},\end{aligned}\ ] ] where @xmath162ds\left . l\neq j}}^{m } } \right| \left . l\neq j}}^{m } } z_t = z , \ x_t = x\right].\end{aligned}\ ] ] furthermore , define @xmath163ds\left . \vphantom{\int_t^t } \right| \left . \vphantom{\int_t^t } z_t = z , \ x_t = x\right],\end{aligned}\ ] ] and @xmath164ds\left . \vphantom{\int_t^t } \right| \left . \vphantom{\int_t^t } z_t = z , \ x_t = x\right].\end{aligned}\ ] ] then we can write as @xmath165ds\left . \vphantom{\int_t^t } \right| \left . \vphantom{\int_t^t } z_t = z , \ x_t = x\right]\bigg\}.\end{aligned}\ ] ] since @xmath166 we have that @xmath167 . by there is an injective map between each functional @xmath168 and @xmath169 for arbitrary @xmath158 such that @xmath170 , hence @xmath171 . consequently , @xmath172ds\left . \vphantom{\int_t^t } \right| \left . \vphantom{\int_t^t } z_t = z , \ x_t = x\right].\end{aligned}\ ] ] by applying the it formula to the process @xmath173 and taking conditional expectation with respect to @xmath174 , we find that the right - hand side of is zero . in the case of a one - dimensional decision variable , i.e. an option of one commodity , this states an intuitively obvious result , namely that in the presence of an effective volume constraint , the usage of the option will lower its value . note that if @xmath175 there is an @xmath176 such that @xmath177 and the map @xmath178 is bijective . hence , we obtain the result of corollary @xmath179 . in the previous section , we studied the dynamic programming problem first in the absence of an effective volume constraint for commodity @xmath29 . we showed that in this case the optimal exercise rule can be determined explicitly and that the option does not lose value if used for this commodity . we also considered the problem in the presence of an effective volume constraint and showed that in this case it loses value when used . in this section , we determine an optimal exercise rule in the presence of an effective volume constraint . to this end , we first derive the associated hjb - equation . for the reminder of the paper , we change the notation on the value function . since @xmath180 , from now on we may write @xmath105 explicitly as a function of the factors @xmath2 instead of the price @xmath181 , that is , we write @xmath182 instead of @xmath183 where the domain of @xmath105 is modified accordingly . we derive now the hjb - equation of the problem . to this end , assume that value @xmath105 exists . then the bellman principle of optimality yields @xmath184 , \end{split}\ ] ] for all times @xmath185 . rewrite the equation as @xmath186=0 . \end{split}\ ] ] furthermore , assume that @xmath187 . then we obtain by the it formula @xmath188ds \nonumber \\ & & + \:\int_{t}^{w}e^{-rs}\frac{1}{2}\sum_{i , j}^{n}(\sigma \sigma^{t})_{ij}v_{x^{i}x^{j}}(s , z_{s},x_{s})d\langle w^{i},w^{j}\rangle_s \nonumber \\ & & + \:\int_{t}^{w}e^{-rs}\sum_{l=1}^{m}v_{z^{l}}(s , z_{s},x_{s})u_{s}^{l}ds\nonumber \\ & & + \:\int_{t}^{w}e^{-rs}\sum_{i=1}^{n}\sum_{l=1}^{m}v_{x^{i}}(s , z_{s},x_{s})\sigma_{i}(s , x_s)dw_{s}\nonumber\\ & & + \:\int_{t}^{w}\sum_{k=1}^{n}\int_{{\ensuremath{\mathbb{r}}}}e^{-rs}\left[v(s , z_{s},x_{s}+\gamma^{(k)}(s , x_{s},\xi^{k}))-v(s , z_{s},x_{s})\right]n^{k}(ds , d\xi^{k}).\end{aligned}\ ] ] here , @xmath189 and @xmath190 where @xmath191 is the @xmath192:th element in the matrix @xmath193 . by compensating the poissonian stochastic integral in , we find under suitable @xmath194-assumptions on @xmath195 and @xmath196 , see @xcite , thrm . 5.3.6 , that the brownian and compensated poissonian integrals in are martingales . thus the equation yields @xmath197\nu^{k}(d\xi^{k})\nonumber\\ & & \:+\left.\sum_{l=1}^{m}\left(a^{l}(p(x_{s}))+v_{z^{l}}(s , z_{s},x_{s})\right)u_{s}^{l}\right ] ds\left . \vphantom{\int_t^t } \right| \left . \vphantom{\int_t^t } z_t = z , \ x_t = x\right].\end{aligned}\ ] ] define the integro - differential operator @xmath198 on @xmath199 as @xmath200\nu^{k}(d\xi^{k}),\end{aligned}\ ] ] and rewrite as @xmath201ds \left . \vphantom{\int_t^t } \right| \left . \vphantom{\int_t^t } z_t = z , \ x_t = x\right ] . \end{split}\ ] ] under appropriate conditions on @xmath105 , see , e.g. @xcite , we can pass to the limit @xmath202 and obtain the hjb - equation @xmath203 where the @xmath58 varies over the set of @xmath204-valued functions defined on @xmath33 $ ] satisfying the conditions @xmath205 for all @xmath90 and @xmath39 $ ] . we observe from the equation that the sign of quantity @xmath206 , @xmath90 , determines whether the option should be exercised or not . from economic point of view , this quantity has a natural interpretation . indeed , for a given commodity @xmath207 , the function @xmath208 gives the instantaneous exercise payoff whereas the function @xmath209 measures the marginal lost option value . if the payoff dominates the lost option value for a given point @xmath210 and commodity @xmath207 , the option should exercised at the full rate . that is , for each commodity @xmath207 , the option should exercised according to the rule @xmath211 we also point out that this rule is in line with the case when there is no effective volume constraint . in this case , the marginal lost option value is zero and , consequently , the option is used every time it yields a positive payoff . in particular , we find that the presence of an effective volume constraint postpones the optimal exercise of the option for a given commodity @xmath207 . in this subsection we consider sufficient conditions for a given function to coincide with the value function . these conditions are given by the following verification theorem . [ sufficientcondition ] assume that a function @xmath212 satisfies the following conditions : * @xmath213 , @xmath214 , * @xmath215 for all @xmath216 and @xmath77 , where @xmath198 is defined in , * the processes * * @xmath217 , * * @xmath218\tilde{n}^{k}(ds , d\xi^{k})$ ] , + are martingales with respect to @xmath36 . then @xmath219 dominates the value @xmath105 . in addition , if there exist an admissible @xmath220 such that @xmath221}\nonumber\\ & = & ( \mathcal{l}-r)f(t , z , x)+\sum_{l=1}^{m}(a^{l}(p(x))+f_{z^{l}}(t , z , x))\mathring{u}_{t}^{l}=0,\end{aligned}\ ] ] for all @xmath216 , then @xmath222 and the function @xmath219 coincides with the value @xmath105 . let @xmath77 and @xmath76 $ ] . by applying the it formula to the process @xmath223 , we find in the same way as in that @xmath224ds \nonumber \\ & & + \:\int_{t}^{t}e^{-rs}\frac{1}{2}\sum_{i , j}^{n}(\sigma \sigma^{t})_{ij}(s , x_s)f_{x^{i}x^{j}}(s , z_{s},x_{s})d\langle w^{i},w^{j}\rangle_s \nonumber \\ & & + \:\int_{t}^{t}e^{-rs}\sum_{l=1}^{m}f_{z^{l}}(s , z_{s},x_{s})u_{s}^{l}ds\nonumber \\ & & + \:\int_{t}^{t}e^{-rs}\sum_{i=1}^{n}\sum_{l=1}^{m}f_{x^{i}}(s , z_{s},x_{s})\sigma_{i}(s , x_s)dw_{s}\nonumber\\ & & + \:\int_{t}^{t}\sum_{k=1}^{n}\int_{{\ensuremath{\mathbb{r}}}}e^{-rs}\left[f(s , z_{s},x_{s}+\gamma^{(k)}(s , x_{s},\xi^{k}))-f(s , z_{s},x_{s})\right]n^{k}(ds , d\xi^{k}).\end{aligned}\ ] ] by using the assumption ( i ) , definition of the operator @xmath198 and the equation , we obtain the equality @xmath225\tilde{n}^{k}(ds , d\xi^{k}).\end{aligned}\ ] ] conditioning up to time @xmath45 and the assumption ( iii ) yields @xmath226\nonumber\\ & & + \:\mathbf{e}\left[\int_{t}^{t}e^{-rs}\sum_{l=1}^{m}f_{z^{l}}(s , z_{s},x_{s})u_{s}^{l}ds\right.\left . \vphantom{\int_t^t } \right| \left . \vphantom{\int_t^t } z_t = z , \ x_t = x\right].\end{aligned}\ ] ] by assumption ( ii ) , we get @xmath227.\ ] ] for all @xmath77 . thus , the first claim follows . now , if there exist an admissible @xmath220 such that holds , then we would get equality in , i.e. for all @xmath228 @xmath229.\ ] ] the conclusion follows . in this section we consider three examples . these examples illustrate two main issues . first , we compare two one - factor models ( example 1 and example 2 ) with underlying ornstein - uhlenbeck factor dynamics , where the first model has a single brownian driver whereas the other is driven by a sum of a brownian motion and a compound poisson process . to illustrate the effect of the jumps , the parameters of the factor dynamics are fixed such that the volatilities and the long term means are matched . in the third example , we study a two - factor model with underlying ornstein - uhlenbeck factor dynamics . as we will observe , the boundary conditions in the factor price dimensions are a delicate matter in this case . in all these examples , the aim is to find @xmath230.\ ] ] the boundary conditions in the @xmath231-direction are found by the same arguments as in @xcite . the terminal condition is @xmath232 for all @xmath233 and @xmath231 . this follows directly from the definition of the value function . furthermore , the boundary condition in the @xmath233-direction , i.e. @xmath234 , is @xmath235 for all @xmath45 and @xmath231 . this follows from the fact that when @xmath234 the only exercise rule available in @xmath47 is the trivial one . the conditions and hold for all three examples below . let the factor dynamics @xmath2 be given by @xmath236 and @xmath237 , where @xmath238 . then it is well known that at time @xmath238 , the solution @xmath239 furthermore , @xmath240 with this specification , the value function is given as a solution to the hjb - equation @xmath241=0,\ ] ] with boundary conditions conditions in @xmath231-direction given by @xmath242ds,\ ] ] and @xmath243ds.\ ] ] we remark that this is similar to the example in @xcite , appendix a. however , we consider an arithmetic ou - process whereas in @xcite the dynamics are given by an _ exponential _ ou - process . to illustrate the effect of the jumps in the factor dynamics , we add in this example a compound poisson process to the factor dynamics defined in and match the expectation and volatility with example 1 . more precisely , consider the factor dynamic given by the it equation @xmath244 where the compound poisson process @xmath245 has lvy measure @xmath246 with @xmath247 . this equation can be written as @xmath248 where the compensator @xmath249 with this specification , the value function is given as a solution to the hjb - equation @xmath250f\alpha e^{-\alpha y}dy\nonumber\\ & & \:-rv(t , z , x)+\sup_{u\in \mathcal{u}}[(x - k+v_{z}(t , z , x))u_{t}],\end{aligned}\ ] ] with boundary conditions in @xmath231-direction given by @xmath251ds,\ ] ] and @xmath252ds.\ ] ] the solution @xmath253 to is given by @xmath254 it is easy to compute from the expression above that @xmath255=(x-\frac{f}{\alpha}-\tilde{\mu})e^{-\kappa(s - t)}+\tilde{\mu}+\frac{f}{\alpha},\ ] ] and @xmath256)^2]=\frac{1}{2\kappa}(1-e^{-2\kappa(s - t)})(\tilde{\sigma}^{2}+\frac{2f}{\alpha^{2}}).\ ] ] to match the volatility and the long term mean in example 1 and example 2 , we solve the equations above for @xmath257 and @xmath258 , when the expectation and variance is equal to that in example 1 . it follows that @xmath259 and @xmath260 then the mean and total volatility in example 1 and example 2 will be the same . this is good for comparison reasons , which will be discussed more in the next section . the purpose of this example is illustrate the results when the price is driven by multiple factors . to this end , let @xmath261 be non - negative constants . consider the two - factor model @xmath262 where @xmath263 , @xmath264 and the lvy measure is @xmath265 . here , @xmath266 is the jump frequency and @xmath267 is the parameter of the exponentially distributed jumps . furthermore , @xmath268 @xmath269 @xmath270 in component form we have , @xmath271 and @xmath272 the solutions can be written as @xmath273 and @xmath274 to set up the valuation model , define the price function @xmath275 as @xmath276 . furthermore , let @xmath43 be as in with @xmath277 . the payoff is of call option type , i.e. @xmath278 . then , the value function reads as @xmath279.\ ] ] from proposition [ prop1 ] , we see that in the absence of an effective final volume constraint the optimal exercise policy @xmath280 is given by @xmath281 for all @xmath76 $ ] . hence , it is optimal to use the option whenever the swing yields a positive payoff . this is in line with @xcite , in which no jumps are considered . consider now the case with an effective volume constraint . the value function can be written in the component form as @xmath282.\ ] ] this function is given as the solution to the hjb - equation @xmath283=0.\end{aligned}\ ] ] with boundary conditions in @xmath45- and @xmath233-direction @xmath284 and boundary conditions in @xmath231-direction @xmath285 @xmath286 to solve the problem - , we need to find the boundary conditions - . we assume that we only have positive finite jumps , i.e. @xmath287 and that @xmath288 . that is , the problem is solved in the first quadrant in the @xmath289-plane . the reason for choosing these spatial boundaries is due to the properties of the hjb - equation . in the @xmath290-direction we have diffusion , which requires boundary conditions at both ends however , in the @xmath291-direction we have transport in the positive direction , because the coefficient in front of @xmath292 is negative and that pde is solved backward in time , and therefore no boundary condition is needed at @xmath293 . at @xmath294 the derivative in @xmath291-direction vanishes , thus no boundary condition is needed . in what follows , the calculations rely on the fact that the underlying factor dynamics are ornstein - uhlenbeck processes . by plugging in the processes @xmath295 and @xmath296 given by and , respectively , into the value function and rearranging the terms we obtain @xmath297\right.\nonumber\\ & & + \:\left.\mathbf{e}\left [ \int_{t}^{t}e^{-r(s - t)}\left(\int_{t}^{s}\lambda^1e^{-\beta^1(s - v)}dw_{v } \right.\right.\right . \\ & & + \left.\left.\left.\int_{t}^{s}\int_{{\ensuremath{\mathbb{r}}}}e^{-\beta^2(s - v)}\xi^2\tilde{n}^2(d\xi^2,dv)\right)u_{s}ds\right.\left| \vphantom{\int_{t}^{t } } \right . \left . \vphantom{\int_{t}^{t } } z_t = z , \ x_t = x \right]\right\}\nonumber.\end{aligned}\ ] ] to compute @xmath298 , we plug @xmath299 into . since the volatilities of the processes @xmath295 and @xmath296 are not state dependent , we can , by choosing @xmath300 sufficiently large , expect the trajectories of the process @xmath301 to be decreasing until the maturity since both the processes @xmath302 and @xmath303 tend towards their long term means , @xmath304 and @xmath305 , respectively . then it is optimal to start to exercise the option immediately with maximum rate until @xmath234 since @xmath306 is much larger than the long time expectation . this argument holds since we only consider positive jumps , i.e. @xmath307 . thus , if we start the process @xmath301 in @xmath308 , we can define an optimal control as @xmath309}(s).\ ] ] then we get @xmath310 here , we used the fact that the control @xmath58 as defined in is deterministic . this enables us to use the fubini theorem and the martingale property for the second expectation in . on the contrary , when we start the process @xmath295 at @xmath311 , assumed to be sufficiently small , @xmath295 will increase until maturity . it is thus tempting to wait as long as possible before we use the control , cf . equation ( 4.6 ) in @xcite . however , since @xmath312 can be very large and the jump frequency is state independent , we are unable to draw this conclusion . to deal with this issue , we proceed as follows . first , we assume that the value function is continuous for all @xmath313 and both @xmath311 and @xmath314 are finite . using this we consider the _ deterministic part _ of the process @xmath301 starting at @xmath315 at time @xmath45 , that is , the first expectation in the value function . we observe that the integrand is a continuous function in time and will assume a maximum ( and minimum ) on @xmath316 $ ] . suppose it has its maximum at a time @xmath317 . furthermore , assume that @xmath318 then , due to continuity , the deterministic part will dominate the process on an interval @xmath319 that contains @xmath317 . we then choose a control defined as @xmath320}(s).\ ] ] we substitute this into the expression . this is a deterministic control so , again , we can use the fubini theorem and the martingale property to get rid of the conditional expectations . define @xmath321ds\nonumber.\end{aligned}\ ] ] then @xmath322}j(t_1,t_2)\ ] ] which is found by solving the ( deterministic ) maximization problem : @xmath323}\bar{u}\int_{t_1}^{t_2}e^{-r(s - t)}&\left[x^{1}_{min}e^{-\beta^1(s - t)}+x^{2}e^{-\beta^2(s - t)}+\frac{f}{\kappa\beta^2}(1-e^{-\beta^2(s - t)})\right.\\&+\left.\frac{\mu^1}{\beta^1}(1-e^{-\beta^1(s - t)})-k \right]ds\nonumber,\end{aligned}\ ] ] subject to @xmath324 we solve the limits @xmath325 numerically by using elementary calculus methods . that is , to solve the maximization problem , define @xmath326\nonumber.\end{aligned}\ ] ] this is the integrand in . by differentiating with respect to @xmath327 and @xmath328 , we obtain the first order necessary conditions @xmath329 and @xmath330 furthermore , at the boundary where @xmath331 , we find @xmath332 to conclude , we solve @xmath327 and @xmath328 from these three equations , substitute these into and see which gives the highest value still satisfying the constraint . in this section we will present the numerical solutions of the hjb equations from the three examples in section 5 . all equations have the boundary conditions that the option value is zero when @xmath333 and @xmath234 . in addition we truncate the boundary in infinity , and the truncated boundaries requires boundary conditions . in example 1 we solve with the boundary condition and on the truncated boundary . similarly in example 2 we solve with the boundary condition and on the truncated boundary . and finally in example 3 we solve with the boundary condition and on the truncated boundary . below we specify the model parameters , which are chosen for the purpose of illustration , and the discretization parameters of the numerical scheme . in example 1 and example 2 the hjb equations are pdes defined over the variables @xmath45 , @xmath233 and @xmath231 , while the hjb equation in example 3 is defined over the variables @xmath45 , @xmath233 , @xmath290 and @xmath291 . the equations are solved with finite difference methods ( fdm ) . we use a first order euler scheme in @xmath45-direction . the @xmath233- and @xmath291-directions are handled explicitly with first order upwind schemes , while the @xmath231 or @xmath290-directions are handled implicitly with a second order central difference scheme . we start the time stepping at @xmath333 and go backward in time until @xmath334 . the domain is discretized with a uniform grid in the @xmath45- , @xmath291- and @xmath233-directions whereas in the @xmath231- or @xmath290-direction we use an adaptive grid . the integral term is approximated with numerical integration , more precisely , the rectangle method with second order midpoint approximations . the truncated boundary in the direction of jump , i.e. the @xmath231-direction in example 2 and the @xmath291- direction in example 3 , causes some problems for the approximation of the integral , which is supposed to have upper limits at infinity . this problem is solved by linearly extrapolating the option price outside the truncated domain , and integrating up to a level where we get sufficiently accurate approximation of the integral . in this subsection we study the numerical examples from the previous section . they are all motivated by some swing options traded in the scandinavian electricity market , which are called `` brukstidskontrakt '' . such a contract gives the owner the right to buy a certain amount of electricity for her own selection of hours during 1 year . more precisely , for each of the 8760 hours in 1 year the holder of the contract must choose whether or not to use the contract . in our example we set the portion to 50% , i.e. the holder must choose 4380 hours . in practice the contracts are usually paid in advance and not for each time it is exercised , so the strike price will be @xmath335 in all the examples . we also use @xmath336 , @xmath337 and @xmath338 . the three examples are further specified in the following . * example 1 : * the model is specified by @xmath339 . the truncated domain is defined by @xmath340 . the discretization parameters are @xmath341 . the grid in @xmath342direction is adaptive and consists of 671 grid points , with higher grid point density around @xmath343 and lower density near the truncated boundaries , @xmath344 and @xmath345 . * example 2 : * the model is specified by @xmath346 . with these parameters the model i example 1 and example 2 have thesame mean and volatility . the grid and the truncated domain is as in example 2 . * example 3 : * the model is specified by @xmath347 . the truncated domain is defined by @xmath348 . the discretization parameters are @xmath349 . in the @xmath290-direction we use an adaptive of 1200 grid points , with higher grid point density around @xmath343 and lower density near the truncated boundary in @xmath290-direction , i.e. @xmath350 and @xmath299 . on the truncated boundary condition in @xmath290-direction we need to calculate and solve . with the parameters in our example , @xmath327 and @xmath328 in turn out to be @xmath351 and @xmath352 . the truncated boundary in @xmath291-direction needs no boundary condition due to the nature of the pde . we also tried adaptive grid in the @xmath291-direction , but this did not seem to improve the accuracy . in the following we visualize the numerical solution of these three examples . the two things we are most interested in are the option prices and the trigger prices . the trigger prices are also referred to as exercise curves , and they tell us when to exercise and when to hold . the option price is a function of two variables for each point in time in example 1 and example 2 , and can then be visualized in a 3d plot . however , the option price in example 3 is a function of three variables and is therefore harder to visualize , even for a fixed point in time , but we present it in 3 plots . the trigger prices are in @xcite presented as exercise curves ( see more in figures below ) . these prices are presented similarly for the results of example 1 and example 2 . but for example 3 the trigger price is actually a 3d surface for each point in time . we solve this by projecting the surface down to two different planes . we could also have made 3d surface plots of the trigger price in this example , but we think 2d plots are more instructive . figure [ fig:1 ] and figure [ fig:2 ] show the option price at @xmath353 for example 1 and example 2 respectively . we see that the two plots a quite similar , and that the option price increases with increasing @xmath231-values . this makes sense from an economical point of view , since one would expect that a higher spot price results in a higher option price . mathematically we see it from the value function and the fact that the solution functions and are increasing functions of @xmath231 . furthermore , the option price decreases with increasing @xmath233-values which is in accordance with proposition [ vznonpositive ] , stating that whenever using the option it loses value , which makes economical sense . figure [ fig:3 ] shows the difference in option price between example 1 and example 2 at the same time level . more precisely figure [ fig:3 ] shows the price from example 2 minus the price from example 1 . we see that this difference is negative , which means that the option price is a little higher when we assume an underlying jump process . this is reasonable since the gaussian ou - process in example 1 has a symmetric distribution whereas the non - gaussian ou - process in example 2 has a positively skewed distribution . this positive skewness , which increases value in financial markets , is caused by the fact that we only have positive jumps . [ fig:123 ] in figure [ fig:4 ] we show the exercise curves for example 1 and example 2 at time = 0.5 . the red curve corresponds to example 2 and the black curve corresponds to example 1 . we see that the red curve lies more to the right than the black curve . in figure [ fig:8][fig:10 ] we show the option price of example 3 at @xmath353 . the function is plotted as a function of @xmath290 and @xmath233 for three values of @xmath291 . we see that for each value of @xmath291 the plot looks similar to the plots in figure [ fig:1 ] and figure [ fig:2 ] , only the level of the surfaces changes a little . [ fig:8910 ] in figure [ fig:5][fig:6 ] we visualize the trigger price for @xmath353 in example 3 . figure [ fig:5 ] shows the exercise surface projected down to the price,@xmath233-plane ( where price = @xmath354 ) , for various values of @xmath291 . figure [ fig:6 ] shows the exercise surface projected down to the @xmath355-plane , for various values of @xmath233 . it is worth noting that the slope of the curves in figure [ fig:6 ] is approximately @xmath356 . for example if we study the red curve and remove the point where @xmath294 , and make a linear least squares approximation of it , it will have a slope of @xmath357 . this means that the negative ratio of the mean reversion speeds approximates the slope of the exercise curves in the @xmath355-plane . this is plausible from an economical point of view for the following reason . if the mean reversion speed is smaller for @xmath290 than for @xmath291 , the holder will exploit a deviation from the long term mean earlier for process @xmath296 than @xmath295 by exercising the option . that is , she would require a higher contribution from @xmath290 to the price than from @xmath291 before exercising . this is because a high value of @xmath291 is likely to reduce more quickly and therefore it is beneficial to exercise with a lower value of @xmath291 in relation to @xmath290 . on the contrary , if @xmath290 has a high price it is more likely to stay high longer . in this case , the holder might wait for even higher prices . a similar reasoning can be done for the opposite case when the mean reversion speed is bigger for @xmath290 than for @xmath291 . notice that at @xmath353 and for any value of @xmath233 , the holder should always hold if the values of @xmath290 and @xmath291 are small enough . the reason for this is the following : if the underlying price is small enough , you would expect it to be higher than this the rest of the time , and for @xmath353 it will therefore be beneficial to hold since @xmath358 . however , in figure [ fig:5 ] it can be seen that the exercise curves seems to stay above the line z=0.001 . this is due to numerical error / instability . for finer grids the exercise curves will be closer to @xmath359 at for small values of @xmath360 and @xmath353 . similar effects can be observed for other values of @xmath45 . the solution to this problem is either higher resolution on the grid , which requires more memory on the computer , or more accurate numerical schemes . the curves in figure [ fig:6 ] bends a little in the lower right corner , and this is due to the mentioned instability in the numerical scheme . [ fig:56 ] the analysis of the numerical scheme in example 1 and example 2 is similar to that of @xcite . to analyse example 3 we study the courant - friedrichs - lewy ( cfl ) condition . the cfl number of the hjb equation in example 3 is @xmath361 a necessary condition for convergence is that the cfl number @xmath362 , and in our case with @xmath363 we see that the cfl number @xmath364 . as mentioned we have observed some small instabilities in the numerical solution . we have also tried with larger values of @xmath365 compared to @xmath366 , but the reported discretization parameters seem to give most accurate solutions . in order to establish convergence an implicit scheme should be developed , but this is not done in this work . in the following we will present some evidence that the numerical solution converges to the correct solution of the hjb equation . we attempt to evaluate both the numerical scheme and the calculated boundary conditions using . this is done by trying to see how well they fit for extreme values of @xmath290 , i.e. @xmath367 . we have used a very fine grid to solve hjb equation where @xmath105 is required to be linear in @xmath290-direction at the boundaries . this is similar to @xcite , where it is shown that this type of inaccurate boundary condition gives quite accurate solutions . now this solution can be compared to the values we get from equation . the difference between the numerical solution of the hjb equation and the value calculated by is illustrated in figure [ fig:7 ] . we see that the difference is relatively small compared to the option value . for this example with @xmath353 and @xmath368 it is about 6 orders of magnitude lower than the option value . this is sufficiently small for us to trust the numerical solver . the difference may come from all of the following five sources : assumptions that the control is as described in , numerical inaccuracy / instability of the scheme , truncation in @xmath290 direction , truncation in @xmath291 direction , extrapolation in @xmath291 direction and linear boundary condition on the pde solver . in this paper , we developed and analyzed a valuation model for swing options on multi - commodity markets . the model is formulated as a dynamic programming problem , where the underlying dynamic structure is given by a multi - dimensional lvy diffusion . this process models the price evolution of the commodities . the commodity prices are driven by a multi - dimensional brownian motion and a multi - dimensional compound poisson process . the introduction of the compound poisson process is important since it allows non - gaussian price evolution . this is important , in particular , on electricity markets , see , e.g. @xcite . furthermore , this model allows us to take into account jumps in price processes , which is also important on electricity markets . from a analytical point of view , this study provides a multi - dimensional generalization of the analysis in @xcite . first , we analyze the model in the absence of an effective volume constraint . along the lines of @xcite , we find that in this case the option does not loose value if used . moreover , we prove that in the presence of an effective volume constraint for a given commodity , the usage of the option for this commodity will lower the value of the option . this is a intuitively appealing from the economical point of view . to tackle the problem of finding an optimal exercise rule and the price of the option , we analyze the pricing problem using the bellman principle of optimality and derive the associated hjb - equation . in section 4 , we obtained an optimal exercise rule which states that if the immediate exercise payoff dominated the lost option value for a given commodity , then the option on this commodity should be exercised at a full rate . in particular , we conclude that this optimal exercise rule is a bang - bang rule . we also provide a verification theorem , which states conditions under which a given function coincides with the value function . in addition we illustrate the results with three examples which we study numerically . we set up a straightforward fdm scheme to solve the associated hjb - equations . the numerical experiments seems to give reasonable results from both mathematical and economical points of view . in the last of our examples we have also given evidence for convergence of the numerical solution . benth , f. e. , kallsen , j. and meyer - brandis , t. ( 2007 ) . a non - gaussian ornstein - uhlenbeck process for electricity spot price modeloing abd derivative pricing , _ applied mathematical finance _ , 14 , 153 169 hambly , b. , howison , s. , and kluge t. ( 2009 ) . modelling spikes and pricing swing options in electricity markets , _ quantitative finance _ , 9/8 , 937 949 jaillet , p. , ronn , m. and tompadis , s. ( 2004 ) . valuation of commodity based swing options , _ management science _ , 14/2 , 223 248

a new analysis of the modern precise measured astrophysical factors for the direct capture reaction [ b.s . nara singh _ et al . _ , phys.rev.lett . * 93 * , 262503 ( 2004 ) ; d. bemmerer _ et al . _ , phys.rev.lett . * 97 * , 122502 ( 2006 ) ; f.confortola _ et al . _ , phys.rev.c * 75 * , 065803 ( 2007 ) , t.a.d.brown _ et al . _ , phys.rev . c * 76 * , 055801 ( 2007 ) and a di leva , _ et al._,phys.rev.lett . * 102 * , 232502 ( 2009 ) ] populating to the ground and first excited states of is carried out based on the modified two - body potential approach . new estimates are obtained for theindirectly determined values of the asymptotic normalization constants ( the nuclear vertex constants ) for(g.s . ) and(0.429 mev ) as well as the astrophysical factors at e 90 kev , including=0 . the values of asymptotic normalization constants have been used for getting information about the-particle spectroscopic factors for the mirror ()-pair . pacs : 25.55.-e;26.35.+c;26.65.+t +

0905.2026

the @xmath11 reaction is one of the critical links in the @xmath12 and @xmath13 branches of the @xmath14chain of solar hydrogen burning [ 13 ] . the total capture rate determined by processes of this chain is sensitive to the cross section @xmath15 ( or the astrophysical @xmath0 factor @xmath6 ) for the @xmath16 reaction and predicted neutrino rate varies as @xmath17^{0.8}$ ] @xcite . despite the impressive improvements in our understanding of the @xmath11 reaction made in the past decades ( see refs [ 411 ] and references therein ) , however , some ambiguities connected with both the extrapolation of the measured cross sections for the aforesaid reaction to the solar energy region and the theoretical predictions for @xmath15 ( or @xmath6 ) still exist and they may influence the predictions of the standard solar model @xcite . experimentally , there are two types of data for the @xmath16 reaction at extremely low energies : i ) six measurements based on detecting of @xmath18-rays capture ( see @xcite and references therein ) from which the astrophysical @xmath0 factor @xmath19 extracted by the authors of those works changes within the range 0.47@xmath200.58 @xmath21 and ii ) six measurements based on detecting of @xmath12 ( see @xcite references therein as well as [ 6 - 11 ] ) from which @xmath19 extracted by the authors of these works changes within the range 0.53@xmath220.63 @xmath23 . all of these measured data have a similar energy dependence for the astrophysical @xmath0 factors @xmath6 . nevertheless , the adaptation of the available energy dependencies predicted in @xcite for the extrapolation of each of the measured data to low experimentally inaccessible energy regions , including @xmath8=0 , leads to a value of @xmath19 that differs from others and this difference exceeds the experimental uncertainty . the theoretical calculations of @xmath19 performed within different methods also show considerable spread [ 12,1419 ] and the result depends on a specific model used . for example , the resonating - group method calculations of @xmath19 performed in ref.@xcite show considerable sensitivity to the form of the effective nn interaction used and the estimates have been obtained within the range of 0.312@xmath20 0.841 @xmath23 . the estimation of @xmath19=0.52@xmath24 0.03 @xmath25 @xcite also should be noted . the latter has been obtained from the analysis of the experimental astrophysical @xmath0 factors @xcite , which were performed within the framework of the standard two - body potential model in the assumption that the dominant contribution to the peripheral reaction comes from the surface and external regions of the nucleus @xmath12 @xcite . at this , in @xcite the contribution from the nuclear interior ( @xmath26 , @xmath27=4 fm ) to the amplitude is ignored . in this case , the astrophysical @xmath0 factor is directly expressed in terms of the nuclear vertex constants ( nvc ) for the virtual decays @xmath28 ( or respective the asymptotic normalization coefficients ( anc ) for @xmath29 ) @xcite . as a result , in ref . @xcite , the nvc - values for the virtual decays @xmath30 and @xmath31 were obtained , which were then used for calculations of the astrophysical @xmath0 factors at @xmath32180 kev , including @xmath8=0 . however , the values of the ancs ( or nvcs ) @xmath33 and the @xmath19 obtained in @xcite may not be enough accurate associated both with the aforesaid assumption in respect to the contribution from the nuclear interior ( @xmath26 ) and with a presence of the spread in the experimental data @xcite used for the analysis . as far available values of these ancs obtained in @xcite , they depend noticeably on a specific model used . therefore , determination of precise experimental values of the ancs for @xmath33(g.s . ) and @xmath29(0.429 mev ) is highly desirable since it has direct effects in the correct extrapolation of the @xmath34 astrophysical @xmath0 factor at solar energies . recently , a modified two - body potential approach ( mtbpa ) was proposed in @xcite for the peripheral direct capture @xmath35 reaction , which is based on the idea proposed in paper @xcite that low - energy direct radiative captures of particle @xmath36 by light nuclei @xmath37 proceed mainly in regions well outside the range of the internuclear @xmath38 interactions . one notes that in mtbpa the direct astrophysical @xmath0 factor is expressed in terms of anc for @xmath39 rather than through the spectroscopic factor for the nucleus @xmath40 in the ( @xmath41 ) configuration as it is made within the standard two - body potential method @xcite . in refs.@xcite , mtbpa has been successfully applied to the radiative proton and @xmath9-particle capture by some light nuclei . therefore , it is of great interest to apply the mtbpa for analysis of the @xmath42 reaction . in this work new analysis of the modern precise experimental astrophysical @xmath0 factors for the direct capture @xmath16 reaction at extremely low energies ( @xmath43 90 kev ) [ 6 - 11 ] is performed within the mtbpa @xcite to obtain @xmath3indirectly determined values both of the ancs ( the nvcs ) for @xmath4(g.s . ) and @xmath5(0.429 mev ) , and of @xmath6 at @xmath44 90 kev , including @xmath8=0 . in this work we quantitatively show that the @xmath45 reaction within the aforesaid energy region is mainly peripheral and one can extract ancs for @xmath46 directly from the @xmath47 reaction where the ambiguities inherent for the standard two -body potential model calculation of the @xmath48 reaction , which is connected with the choice of the geometric parameters ( the radius @xmath49 and the diffuseness @xmath36 ) for the woods saxon potential and the spectroscopic factors @xcite , can be reduced in the physically acceptable limit , being within the experimental errors for the @xmath6 . the contents of this paper are as follows . in section 2 the results of the analysis of the precise measured astrophysical @xmath0 factors for the direct radiative capture @xmath45 reaction is presented ( subsections 2.12.3 ) . the conclusion is given in section 3 . in appendix basic formulae of the modified two - body potential approach to the direct radiative capture @xmath16 reaction are given . let us write @xmath50 ( @xmath51 ) for the relative orbital ( total ) angular moment of @xmath52 and @xmath9-particle in nucleus @xmath53 , @xmath54 ( @xmath55 ) for the orbital ( total ) angular moment of the relative motion of the colliding particles in the initial state . for the @xmath56 reaction populating to the ground and first excited ( @xmath57=0.429 mev ; @xmath58=1/2@xmath59 ) states of @xmath60 , the values of @xmath51 are taken to be equal to 3/2 and 1/2 , respectively , the value of @xmath50 is taken to be equal to 1 as well as @xmath61=0 , 2 for the @xmath62-transition and @xmath54=1 for the @xmath63-transition . the basic formulae used for the analysis are presented in appendix . to determine the anc values for @xmath29(g.s ) and @xmath65(0.429 mev ) the recent experimental astrophysical @xmath0 factors , @xmath66 , for the @xmath34 reaction populating to the ground ( @xmath50=1 and @xmath51=3/2 ) and first excited ( @xmath57=0.429 @xmath67 ; @xmath68,@xmath50=1 and @xmath51=1/2 ) states [ 611 ] are reanalyzed based on the relations ( a1)(a7 ) . the experimental data analyzed by us cover the energy ranges @xmath8=92.9168.9 kev [ 79 ] , 420951 kev @xcite , 3271235 kev @xcite and 7011203 kev @xcite for which only the external capture is substantially dominant @xcite . also , in @xcite the experimental astrophysical @xmath0 factors for the the @xmath34 reaction populating to the first and excited states of the @xmath60 nucleus have been separated only for the energies of @xmath8=92.9 , 105.6 and 147.7 kev . whereas , in @xcite the experimental astrophysical @xmath0 factors have been separated for all experimental points of @xmath8 from the aforesaid energy region by means of detecting the prompt @xmath18 ray ( the prompt method ) and by counting the @xmath60 activity ( the activation ) . the woods saxon potential split with a parity ( @xmath69-dependence ) for the spin - orbital term proposed by the authors of refs . [ 3032 ] is used here for the calculations of both bound state radial wave function @xmath70 and scattering wave function @xmath71 . such the choice is based on the following considerations . firstly , this potential form is justified from the microscopic point of view because it makes it possible to take into account the pauli principle between nucleons in @xmath52- and @xmath9-clusters in the ( @xmath72 ) bound state by means of inclusion of deeply bound states forbidden by the pauli exclusion principle . the latter imitates the additional node ( @xmath73 ) arising in the wave functions of @xmath74 relative motion in @xmath12 similarly as the result of the microscopic resonating - group method @xcite . secondly , this potential describes well the phase shifts for @xmath75-scattering in the wide energy range @xcite and reproduces the energies of low - lying states of the @xmath76 nucleus @xcite . we vary the geometric parameters ( radius @xmath49 and diffuseness @xmath36 ) of the adopted woods saxon potential in the physically acceptable ranges ( @xmath49 in 1.621.98 fm and @xmath36 in 0.630.77 fm @xcite ) in respect to the standard values ( @xmath49=1.80 fm and @xmath36=0.70 fm @xcite ) using the procedure of the depth adjusted to fit the binding energies . as an illustration , fig.[fig1 ] shows plots of the @xmath77 dependence on the single - particle anc , @xmath78 for @xmath50= 1 and @xmath51=3/2 and 1/2 only for the two values of energy @xmath8 . the width of the band for these curves is the result of the weak @xmath3residual@xmath79-dependence of the @xmath80 on the parameters @xmath49 and @xmath36 ( up to @xmath81 ) for the @xmath82 @xcite . the same dependence is also observed at other energies . for example , for fig.1 plotted for @xmath8=0.1056 ( 0.1477 ) mev overall uncertainty ( @xmath83 ) of the function @xmath80 in respect to the central values @xmath77 corresponding to the central values of @xmath84 comes to @xmath83=@xmath24 4.5(@xmath244.5)% for the ground state of @xmath12 and @xmath83 = @xmath24 3.4(@xmath242.9)% for the excited state of @xmath12 . as it is seen from here that the @xmath11(0.429 mev ) reaction is slightly more peripheral than the @xmath16(g.s . ) reaction since the binging energy for @xmath12(0.429 mev ) is less than that for @xmath85(g.s ) . the similar dependence of @xmath80 on the @xmath78 values is observed for the other aforesaid energies @xmath8 and the value of @xmath83 is no more than @xmath86 5.0% . it follows from here that the condition ( a2 ) is satisfied for the considered reaction at the energies above mentioned within the uncertainties not exceeding the experimental errors of @xmath66 . it should be noted that values of @xmath83 becomes larger as the energy @xmath8 increases ( for @xmath8 more than 1.3 mev ) . thus , over the energy region 92.9@xmath7e@xmath7 1200 kev the @xmath87-dependence on @xmath88 is exactly sufficiently weak being within the experimental uncertainties for the @xmath66 . such dependence is apparently associated also with the indirect taking into account the pauli principle mentioned above within the nuclear interior in the adopted nuclear @xmath89 potential leading as a whole to reduction of the contribution from the interior part of the radial matrix element into the @xmath87 function , which is typical for peripheral reactions . we also calculated the @xmath90-elastic scattering phase shifts by variation of the parameters @xmath49 and @xmath36 in the same range for the adopted woods saxon potential . as an illustration , the results of the calculations corresponding to the @xmath91 and @xmath92 waves are only presented in fig.[fig2 ] in which the width of the bands corresponds to a change of phase shifts values with respect to variation of values of the @xmath49 and @xmath36 parameters . as it is seen from fig.[fig2 ] , the experimental phase shifts @xcite are well reproduced within uncertainty of about @xmath24 5% . the same results are also obtained for the @xmath93 and @xmath94 waves . this circumstance allows us to test the condition ( a3 ) at the energies of @xmath8= 92.9 , 105.6 and 147.7 kev for which the @xmath16(g.s . ) and @xmath16(0.429 mev ) astrophysical @xmath0 factors were separately measured in @xcite . as an illustration , for the same energies @xmath8 as in fig.[fig1 ] we present in fig.[fig3 ] ( the upper panels ) the results of @xmath95-value calculation given by eq.(a3 ) ( @xmath96=(1@xmath97 3/2 ) and ( 1@xmath97 1/2 ) ) in which instead of the @xmath98 the experimental @xmath0 factors for the @xmath99 reaction populating to the ground and first excited states of @xmath12 were taken . it should be noted that the same dependence occurs for other considered energies . the calculation shows the obtained @xmath95 values also weakly ( up to 5.0 % ) depend on the @xmath78-value . consequently , the @xmath11 reaction within the considered energy ranges is peripheral and a use of parametrization in terms of the ancs given by eq.(a1 ) is adequate to the physics of the reaction under consideration . however , the values of the spectroscopic factors @xmath100 and @xmath101 corresponding to the @xmath102-configuration for @xmath12(g.s . ) and @xmath85(0.429 kev ) , respectively , change strongly about 1.7 times since calculated @xmath103 that vary by 1.75 times ( see , the lower panels in fig.[fig3 ] ) . for each energy @xmath8 experimental point ( @xmath8=92.9 , 105.6 and 147.7 kev ) the values of the ancs are obtained for @xmath104 and @xmath105(0.429 mev ) by using the corresponding experimental astrophysical @xmath0 factor ( @xmath106 and @xmath107 , ( the activation ) ) @xcite in the ratio of the r.h.s . of the relation ( a1 ) instead of the @xmath98 and the central values of @xmath77 corresponding to the adopted values of the parameters @xmath49 and @xmath36 . the results of the ancs , @xmath108 and @xmath109 , for these three energy @xmath8 experimental points are displayed in fig.[fig5 ] @xmath36 and @xmath110 ( filled circle symbols ) and the second and third columns of table [ table1 ] . the uncertainties pointed in this figure correspond to those found from ( a1 ) ( averaged square errors ( a.s.e . ) ) , which include the total experimental errors ( a.s.e . from the statical and systematic uncertainties ) in the corresponding experimental astrophysical @xmath0 factor and the aforesaid uncertainty in the @xmath80 . one should note that the same results for the ancs are obtained when @xmath111 ( @xmath112 and @xmath113 ) @xcite are used in eq.(a5 ) ( in eq.(a6 ) and ( a7 ) ) instead of @xmath6 ( @xmath114 and @xmath115 ) . then in eq.(a6 ) , inserting the averaged means of @xmath116 ( @xmath116=0.666 ) , obtained from the three data , and replacing of the @xmath117 in the l.h.s . of eq.(a5 ) with @xmath111 for the others , the two experimental points of energy @xmath8 ( @xmath8=126.5 and 168.9 kev ) from @xcite , four one @xmath8 ( @xmath8=420.0 , 506.0 , 615.0 and 951.0 kev ) from @xcite , the three one @xmath8 ( @xmath8=93.3 , 106.1 and 170.1 kev ) from @xcite and the ten one @xmath8 ( @xmath8=7011203 kev ) from @xcite can also determine values of ancs , @xmath118 and @xmath119 . the results of the ancs @xmath120 for @xmath121(g.s . ) and @xmath122(0.429 mev ) are displayed in fig.[fig5 ] in which the open cycle and triangle symbols obtained from the analysis of the data of [ 69 ] as well as the filled triangles symbols obtained from the analysis of the data of @xcite(the @xmath60 recoils ) . besides , the results obtained from the data of [ 79 ] are also presented in the second and third columns of table [ table1 ] . the same way the values of the ancs are obtained by using the separated experimental astrophysical @xmath0 factors ( @xmath123 and @xmath124 ) @xcite . the results for these ancs are also presented in figs.[fig5]@xmath125 and @xmath126 both for the activation ( filled star symbols ) and for the prompt method ( filled square symbols ) . as it is seen from figs.[fig5 ] , for each of the independent measured experimental astrophysical @xmath0 factors the ratio in the r.h.s . of the relation ( a4 ) does not practically depend on the energy @xmath8 within the experimental uncertainties , although absolute values of the corresponding experimental astrophysical @xmath0 factors for the reactions under consideration depend noticeably on the energy and change by up to about 1.7 times when @xmath8 changes from 92.6 kev to 1200 kev . this fact allows us to conclude that the energy dependence of the experimental astrophysical @xmath0 factors [ 611 ] is well determined by the calculated function @xmath80 and @xmath127 . hence , the experimental astrophysical @xmath0 factors presented in [ 611 ] can be used as an independent source of reliable information about the ancs for @xmath121(g.s . ) and @xmath128(0.429 mev ) . also , in fig.[fig5 ] and table [ table2 ] the weighted means of the ancs - values and their uncertainties ( the solid line and the band width , respectively ) , derived both separately from each experimental data and from all of the experimental points , are presented . as it is seen also from the first and second ( fifth and sixth ) lines of table [ table2 ] the weighted means of the ancs - values for @xmath121(g.s . ) and @xmath128(0.429 mev ) obtained by the analysis performed separately for the activation , the prompt method and the @xmath60 recoils of the experimental data from the works [ 69 ] ( @xcite ) are in a good agreement with one another . these results are the first ones of the present work . nevertheless , the weighted means @xcite of the anc - values obtained by using separately the experimental data of the works refs.[69 ] and of the works refs.@xcite noticeably differ from one another ( up to 1.13 times for @xmath129(g.s . ) and up to 1.12 times for @xmath122(0.429 mev ) , see the parenthetical figures in table [ table2 ] ) . the main reason of this difference is in the systematical discrepancy observed in absolute values of the experimental astrophysical @xmath0 factors measured by authors of works refs [ 69](the set i ) and of the works ref . @xcite ( the set ii , see fig.[fig6 ] ) . also , the central values of the weighted means for the anc - values for @xmath130(g.s . ) and @xmath131(0.429 mev ) obtained from all of the experimental data [ 611 ] , which is presented in the last line of table [ table2 ] , differ up to 10% more ( 3% less ) than those deduced from the data of [ 69](@xcite ) . as , at present , there is no a reasonable argument to prosecute to some of these experimental data measured by two groups ( [ 6 - 9](the set i ) and @xcite(the set ii ) ) , it seems , it is reasonable to obtain the weighted means of the ancs derived from all these real experimental ancs with upper and lower limits corresponding to the experimental data of the set ii and set i , respectively . this leads to the asymmetric uncertainty for the weighted means of ancs and this is caused with the aforesaid systematical discrepancy observed in absolute values of the experimental data of the sets i and ii ( see the last line of table [ table2 ] ) . in this connection , from our point of view a new precisely measurement of @xmath132 is highly encouraged since it allows one to get an additional information about the ancs . nevertheless , below we will use these ancs for extrapolation of the astrophysical @xmath0 factors at energies down , including @xmath8=0 . the corresponding values of nvcs obtained by using eq.(a8 ) are given in table [ table2 ] . a comparison of the present result and that obtained in paper @xcite shows that the underestimation of the contribution both of the nuclear interior and of the nuclear exterior indeed occurs in @xcite since the contribution of the nuclear interior ( @xmath133 4.0 fm ) to the calculated astrophysical @xmath0 factors and use the experimental data [ 6 - 11 ] more accurate than those in ref.@xcite can influence the extracted values of the ancs . besides , one would also like to note that in reality the values of the ancs , @xmath134 and @xmath135 , should not be equal , as it was assumed in @xcite and the values of @xmath136=14.4 @xmath137 were obtained from the analysis of @xmath132 performed in within the @xmath49-matrix method . the resulting anc ( nvc ) value for @xmath138(g.s . ) obtained by us is in good agreement with the value @xmath118=25.3 @xmath137 ( @xmath139=1.20 fm ) and that for @xmath140(0.429 mev ) differs noticeably from the value @xmath119=22.0 @xmath137 ( @xmath141=1.04 fm ) , which were obtained in @xcite within the ( @xmath142)-channel resonating - group method . also , the results of the present work differ noticeably from the values @xmath118=12.6@xmath24 1.1 @xmath137 and @xmath143 @xmath137(@xmath134=3.55@xmath24 0.15 @xmath144 , @xmath135=2.90@xmath24 0.10 @xmath144 , @xmath139=0.596@xmath24 0.052 fm and @xmath141=0.397@xmath24 0.030 fm ) @xcite as well as those @xmath136=14.4 @xmath137 ( @xmath145=3.79 @xmath144 and @xmath146=0.680 fm ) @xcite . in this connection one would like to note that in @xcite the bound state wave functions , which correspond to the binding energy for @xmath60(g.s . ) in the ( @xmath147)-channel differing noticeably from the experimental ones ( see table i in ref.@xcite ) , and the initial state wave functions were computed with different potentials and , so , these calculations were not self - consistent . since the ancs for @xmath148 are sensitive to the form of the nn potential , it is desirable , firstly , to calculate the wave functions of the bound state using other forms of the nn potential , and , secondly , in order to guarantee the self - consistency , the same forms of the nn potential should be used for such calculation of the initial wave functions . the @xmath3indirectly determined values of the ancs for @xmath150 presented in the last line of table [ table2 ] ) and those for @xmath151 deduced in ref.@xcite can be used for obtaining information on the ratio @xmath152 for the virtual @xmath9 decays of the bound mirror ( @xmath153)-pair , where @xmath154 is the spectroscopic factor for @xmath60 ( @xmath155 ) in the ( @xmath147)((@xmath156))-configuration . for this aim , from @xmath157 ( @xmath158 and @xmath60 ) we form the relation @xmath159 where @xmath160(@xmath161 ) is the ratio of squares of the ancs ( single - particle ancs ) for the bound mirror ( @xmath153)-pair and @xmath51=3/2(1/2 ) for the ground ( first excited ) state of the mirror nuclei . besides , it should be noted that the relation ( [ 18acd ] ) allows one to verify a validity of the approximation ( @xmath162 , i.e. @xmath163 1 ) used in refs.@xcite for the mirror ( @xmath153 ) conjugated @xmath9 decays . for the bound and first excited state of the mirror ( @xmath164)-pair the values of @xmath165 and @xmath166 change by the factor of 1.3 under the variation of the geometric parameters ( @xmath49 and @xmath36 ) of the adopted woods saxon potential @xcite within the aforesaid ranges , while the ratios @xmath167 and @xmath168 for the bound and first excited states of the mirror ( @xmath153)-pair change by only about 1.5% and 6% , respectively . it is seen that the ratios do not depend practically from variation of the free parameters @xmath49 and @xmath36 , which are equal to @xmath169=1.37@xmath24 0.02 and @xmath168=1.40@xmath24 0.09 . they are in good agreement with those calculated in @xcite within the microscopic cluster and two - body potential models ( see table i there ) . the ratios for the ancs are @xmath170=1.83@xmath171 and @xmath172=1.77@xmath173 . from ( [ 18acd ] ) the values of the ratio @xmath174 are equal to @xmath175=1.34@xmath176 and @xmath177=1.26@xmath178 for the ground and the first excited states , respectively . within their uncertainties , these values differ slightly from those of @xmath175=0.995@xmath240.005 and @xmath177=0.990 calculated in ref.@xcite within the microscopic cluster model . one notes that the values of @xmath174 calculated in @xcite are sensitive to the model assumptions ( the choice of the oscillation radius @xmath125 and the form of the effective @xmath179 potential ) and such model dependence may actually influence the mirror symmetry for the @xmath9-particle spectroscopic factors . the mirror symmetry breakup for the @xmath9-particle spectroscopic factors can also be signalled by the results for the ratio of @xmath180 at zero energies for the mirror ( @xmath153)-pair obtained in @xcite within the resonating - group method by using the seven forms for the effective @xmath179 potential . as shown in @xcite , this ratio is sensitive to a form of the effective @xmath179 potential used and changes from 1.0 to 1.18 times in a dependence from the effective @xmath179 potential used . one of the possible reasons of the sensitivity observed in @xcite can apparently be associated with a sensitivity of the ratio @xmath174 to a form of the effective @xmath179 potential used . in a contrast of such model dependence observed in @xcite , the problem of the ambiguity connected with the model @xmath79-dependence for the values of the ratios @xmath181 found by us from eq.([18acd ] ) is reduced to minimum within the experimental uncertainty . it is seen from here that the empirical values of @xmath174 exceed unity both for the ground state and for the first excited state of the mirror ( @xmath153)-pair . this result for @xmath174 is not accidental and can be explained qualitatively by the following consideration . the fact is that the spectroscopic factor @xmath182 ( or @xmath183 ) is determined as a norm of the radial overlap function of the bound state wave functions of the @xmath184 , @xmath9 and @xmath155 ( or @xmath185 , @xmath9 and @xmath186 ) nuclei and is given by eqs.(100 ) and ( 101 ) from ref . the interval of integration ( @xmath187 ) in eq . ( 101 ) can be divided in two parts . in the first integral denoted by @xmath188 for @xmath155 and @xmath189 for @xmath60 , the integration over @xmath190 covers the region 0@xmath191 ( the internal region ) , where nuclear ( @xmath192 or @xmath193 ) interactions are dominate over the coulomb interactions . in the second integral @xmath194 where in the asymptotic region the radial overlap function entering the integrand is replaced by the appropriate whittaker function ( see , for example , eq.(108 ) of ref.@xcite ) , interaction between @xmath36 and @xmath9-particle ( @xmath195 for @xmath196 or @xmath197 for @xmath198 ) is governed by the coulomb forces only ( the external region ) . in ( [ 18ace ] ) , @xmath199 and @xmath200 is the whittaker function . one notes that the magnitudes @xmath188 ( @xmath201 ) and @xmath202 ( @xmath203 ) define the probability of finding @xmath184 ( or @xmath185 ) in the ( @xmath156 ) configuration ( or the ( @xmath147 ) configuration ) at distances of @xmath204 and of @xmath205 , respectively . obviously @xmath206 and @xmath207 . an information about values of @xmath202 and @xmath203 can be obtained from ( [ 18ace ] ) by using the values of the ancs for @xmath208 and @xmath209 recommended in @xcite and in the present work , respectively . for example , for @xmath2104.0 fm ( the surface regions for the mirror ( @xmath211)-pair ) the calculation shows that the ratio @xmath212 is equal to 1.43@xmath213 ( 1.31@xmath214 ) for the ground ( excited ) states of the @xmath155 and @xmath186 nuclei , i.e. the ratio @xmath215 . owing to the principle of equivalency of nuclear interactions between nucleons of the ( @xmath192)-pair in the nucleus @xmath155 and ( @xmath216)-pair in the nucleus @xmath60 @xcite , the values of @xmath188 and @xmath189 should not differ appreciably . if one suggests that @xmath2171 , then the ratio @xmath2181 . the equation ( a1 ) and the weighted means of the ancs obtained for the @xmath5(g.s ) and @xmath5(0.429 mev ) can be used for extrapolating the @xmath219 astrophysical @xmath0 factor for capture to the ground and first excited states as well as the total astrophysical @xmath0 factor at solar energies ( @xmath220 kev ) . we tested again the fulfilment of the condition ( a2 ) in the same way as it is done above for e@xmath221 90 kev and similar results plotted in fig.[fig1 ] are also observed at energies of e@xmath222 kev . the experimental and calculated astrophysical @xmath0 factors for the @xmath11(g.s . ) , @xmath16 ( 0.429 mev ) and @xmath16 ( g.s.+0.429 mev ) reactions are presented in table [ table1 ] and displayed in fig.[fig6]_a _ , _ b _ and _ c _ , respectively . in figs.[fig6]_a _ and _ b _ , the open diamond and triangle symbols ( the filled triangle symbols ) show our results for the @xmath11(g.s . ) and @xmath16(0.429 mev ) reactions ( see table [ table1 ] also ) , which are obtained from the analysis of the total experimental astrophysical @xmath0 factors of [ 79 ] and @xcite(@xcite ) , respectively , by using the corresponding values of the ancs for each energy @xmath8 experimental point presented figs.[fig5]_a _ and _ b _ ( see table [ table1 ] too ) . there the open circle symbols lying along the smooth solid lines are the results of the extrapolation obtained by us in which each of the quoted uncertainties is the uncertainty associated with that for the ancs adopted . all these results are the second ones of the present work . in figs.[fig6]_a _ and _ b _ the experimental data plotted by the filled circle symbols ( filled star and square symbols ) are taken from @xcite ( from @xcite ) . the solid lines present our calculations performed also with the standard values of geometric parameters @xmath49=1.80 fm and @xmath36=0.70 fm . in fig.[fig6]_c _ , the symbols are data of all experiments [ 611 ] and the solid line presents our calculations performed with the standard values of geometric parameters @xmath49=1.80 fm and @xmath36=0.70 fm by using the weighted means of the ancs ( @xmath118 and @xmath119 ) presented in the last line of table [ table2 ] . there the dashed ( dot - dashed ) lines are the results of calculation obtained by using the aforesaid lower ( upper ) limit values of the ancs pointed out in the last line of table [ table2 ] and the standard values of the geometric parameters ( @xmath49 and @xmath36 ) , and the dotted line is the result of ref.@xcite . as it is seen from these figures , the equations ( a1 ) , ( a4 ) and ( a5 ) allow one to perform a correct extrapolation of the corresponding astrophysical @xmath0 factors at solar energies . but , the noticeable systematical underestimation between the results of calculations performed in ref.@xcite in respect to the experimental data occurs . the weighted means of the total astrophysical @xmath0 factor @xmath6 at solar energies ( @xmath8=0 and 23 @xmath223 ) obtained by us are presented in the the last line of table [ table2 ] . as it is seen from table [ table2 ] the weighted means of @xmath19 , deduced by us separately from each the activation and the prompt method of the experimental data from works [ 69 ] and @xcite ( the first and second lines as well as the fifth and sixth lines ) , agree well within their uncertainties with each other and with those recommended in [ 911 ] . but , these weighted means of @xmath19 obtained by us from the independent analysis of the different data ( the set i and the set ii ) differ also noticeably from one another ( about 11% ) and this distinction is mainly associated with the aforesaid difference observed in magnitudes of the corresponding ancs presented in the third and seventh lines of table [ table2 ] . nevertheless , the weighted mean of @xmath224(0)=0.613@xmath225 @xmath226 , obtained by us by using the weighted means of the ancs presented in the last line of table [ table2 ] , within the asymmetric uncertainty , which is caused with the asymmetric uncertainty for the ancs presented in the last line of table [ table2 ] , agrees also with that recommended in [ 911,38 ] . but , it is interesting to note that the central value of it is closer to that given in the third line of table [ table2 ] , than to the central value of the weighted mean given in the seventh line of table [ table2 ] . also , the astrophysical @xmath0 factors , calculated by using the values of the ancs obtained separately from the set i , the set ii and both them ( see table [ table2 ] ) , are fitted independently using a second - order polynomial within three energy intervals ( 0@xmath227 500 kev , 0@xmath227 1000 kev and 0@xmath227 1200 kev ) . the results for the slop @xmath228(0)/@xmath224(0 ) are to be -0.711 mev@xmath229 , -0.734 mev@xmath229 and -0.726 mev@xmath229 in dependence on the aforesaid intervals , respectively , and they do not depend on the values of the ancs used . one notes that they also are in agreement with -0.73 mev@xmath229 @xcite and -0.92@xmath240.18 mev@xmath229 @xcite . it is seen from here that the @xmath6 calculated by us ( the solid lines in fig.[fig6]_a _ , _ b _ and _ c _ ) and that obtained in @xcite have practically the same energy dependence within the aforesaid energy interval but they differ mainly with each other by a normalization . our result for @xmath19 differs noticeably on that recommended in refs.@xcite @xcite and @xcite(@xmath230 40 kevb , 0.52@xmath240.03 kevb and 0.51@xmath240.04 kev b , respectively ) . this circumstance is apparently connected with the underestimation of the contribution from the external part in the amplitude admitted in these works . besides , the result of the present work is noticeably larger than the result of @xmath19=0.516 ( 0.53 ) kev b @xcite(@xcite ) obtained within the standard two - body ( @xmath147 ) potential by using @xmath89 potential deduced by a double - folding procedure . one of the possible reason of this discrepancy can be apparently associated with the assumption admitted in @xcite that a value of the ratio @xmath174 for the bound mirror ( @xmath231)-pair is taken equal to unity ( @xmath232 = @xmath233=1.17 @xcite ) , which in turn results in the observed systematical underestimation of the calculated @xmath6 in respect to the experimental @xmath0 factors ( see the dashed line in fig.[fig6]_c _ ) . but , as shown in subsection 2.3 the values of @xmath174 for the ground and first excited states of the mirror ( @xmath234)-pair are large unity . therefore , the underestimated values of @xmath232 and @xmath233 used in @xcite also result in the underestimated value of @xmath19 for the direct capture @xmath235 reaction . perhaps , the assumption about equal values of the spectroscopic factor ( @xmath236 = @xmath237=1.17 @xcite ) is correct only for the spectroscopic factors @xmath238 since the value of @xmath19 obtained in @xcite and @xcite for the direct capture @xmath239 reaction agree excellently with each other . one notes that in ref.@xcite the analysis of the @xmath239 experimental astrophysical @xmath0 factors @xcite has also been performed within mtbpa and the anc for @xmath208(g.s . ) , which has been deduced there from the results of ref.@xcite , also is in good agreement with that recommended by authors of ref.@xcite . nevertheless , we observe that the value of @xmath19= 0.56 @xmath226 @xcite obtained within the microscopical ( @xmath147)-cluster approach is in agreement with our result within the uncertainty . besides , our result is also in excellent agreement with that of @xmath19= 0.609 @xmath226 @xcite and @xmath19= 0.621 @xmath240 @xcite obtained within the ( @xmath241)-channel of version of the resonating - group method by using the modified wildermuth - tang ( mwt ) and the near - serber exchange mixture forms for the effective nn potential , respectively . it follows from here that the mutual agreement between the results obtained in the present work and works of @xcite , which is based on the common approximation about the cluster @xmath242 structure of the @xmath60 , allows one to draw a conclusion about the dominant contribution of the @xmath242 clusterization to the low - energy @xmath56 cross section both in the absolute normalization and in the energy dependence [ 611 ] . therefore , single - channel @xmath243 approximation for @xmath60 @xcite is quite appropriate for this reaction in the considered energy range . also , it is interesting to note that the ratios of the @xmath3indirectly determined astrophysical @xmath0 factors , @xmath244(0 ) and @xmath245 , for the @xmath246 reaction populating to the ground and first excited states obtained in the present work to those for the mirror @xmath247 reaction populating to the ground and first excited states deduced in ref.@xcite are equal to @xmath248=6.87@xmath249 and @xmath250=6.11@xmath251 , respectively . these values are in a good agreement with those of @xmath248=6.6 and @xmath250=5.9 deduced in ref.@xcite within the microscopic cluster model . this result also confirms directly our estimation for the ratio @xmath252 obtained above since the ancs for @xmath253(g.s ) and @xmath253(0.478 mev ) as well as the ancs for @xmath65(g.s ) and @xmath33(0.429 mev ) determine the astrophysical @xmath0 factors for the @xmath254 and @xmath219 reactions at zero energies and , consequently , the ratios @xmath248 and @xmath250 are proportional to @xmath255 and @xmath172 , respectively . fig . [ fig6]_d _ shows a comparison between the branching ratio @xmath113 obtained in the present work ( the open triangle and square symbols ) and that recommended in refs.@xcite ( the filled square symbols ) , in @xcite ( the filled circle symbols ) and @xcite ( the filled triangle symbols ) . the weighted mean @xmath256 of the @xmath113 recommended by us is equal to @xmath256=0.43@xmath24 0.01 . as it is seen from fig.[fig6]_d _ , the branching ratio obtained in the present work and in @xcite is in a good agreement with that recommended in ref.@xcite although the underestimation occurs for the @xmath132 obtained in ref.@xcite . such a good agreement between two of the experimental data for the @xmath113 can apparently be explained by the fact that there is a reduction factor in @xcite , being overall for the @xmath48(g.s . ) and @xmath48(0.429 mev ) astrophysical @xmath0 factors . the present result for @xmath256 is in an excellent agreement with those of 0.43@xmath24 0.02 @xcite and 0.43 @xcite but is noticeably larger than 0.37 @xcite and 0.32@xmath24 0.01 @xcite . the analysis of the modern experimental astrophysical @xmath0 factors , @xmath111 , for the @xmath16 reaction , which were precisely measured at energies @xmath8=92.9 - 1235 kev [ 611 ] , has been performed within the modified two - body potential approach @xcite . the performed scrupulous analysis shows quantitatively that the @xmath11 reaction within the considered energy ranges is mainly peripheral and the parametrization of the direct astrophysical @xmath0 factors in terms of ancs for the @xmath150 is adequate to the physics of the peripheral reaction under consideration . it is shown that the experimental astrophysical @xmath0 factors of the reaction under consideration [ 611 ] can be used as an independent source of getting the information about the ancs ( or nvcs ) for @xmath257 ( or for the virtual decay @xmath258 ) , and the found ancs can predict the experimental astrophysical @xmath0 factors separated for the @xmath235(g.s . ) and @xmath235(0.429 mev ) reactions at low experimentally acceptable energy regions ( 126.5@xmath7 e@xmath7 1203 kev ) obtained from the total experimental astrophysical @xmath0 factors [ 69,11 ] . the new estimation for the weighted means of the ancs for @xmath259 and nvcs for the virtual decay @xmath258 are obtained . also , the values of ancs were used for getting the information about the @xmath9-particle spectroscopic factors for the mirror ( @xmath260)-pair . the obtained values of the ancs were also used for an extrapolation of astrophysical @xmath0 factors at energies less than 90 kev , including @xmath8=0 . in particular , the weighted mean of the branching ratio @xmath256 ( @xmath256=0.43@xmath24 0.01 ) and the total astrophysical @xmath0 factor @xmath19 ( @xmath224(0)=0.613@xmath261 @xmath226 ) obtained here are in agreement with that deduced in [ 711 ] from the analysis the same experimental asprophysical @xmath0 factors . besides , our result for @xmath19 is in an agreement with that @xmath19=0.56 kev @xcite obtained within the microscopical single - channel ( @xmath147 ) cluster model , @xmath19= 0.609 @xmath262 @xcite and @xmath19= 0.621 @xmath262 @xcite obtained within the ( @xmath263)-channel of version of the resonating - group method , but it is noticeably larger than the result of @xmath19=0.516 ( 0.53 ) kev @xcite(@xcite ) obtained within the standard two - body ( @xmath147 ) potential by using @xmath75 potential deduced by a double - folding procedure . the authors are deeply grateful to s. v. artemov , l.d . blokhintsev and a.m. mukhamedzhanov for discussions and general encouragement . the authors thank also d.bemmerer for providing the experimental results of the updated data analysis . the work has been supported by the academy of science of the republic of uzbekistan ( grant no.fa-f2-f077 ) . according to @xcite , for fixed @xmath50 and @xmath51 we can write the astrophysical @xmath0 factor , @xmath98 , for the peripheral direct capture @xmath11 reaction in the following form @xmath264 here , @xmath265 is the anc for @xmath266 , which determines the amplitude of the tail of the @xmath186 nucleus bound state wave function in the ( @xmath267)-channel and is related to the spectroscopic factor @xmath268 for the ( @xmath147)-configuration with the quantum numbers @xmath50 and @xmath51 in the @xmath60 nucleus by the equation @xmath269 @xcite , and @xmath270 , where @xmath271 is the single - particle astrophysical @xmath0 factor @xcite and @xmath78 ( @xmath272 @xcite ) is the single - particle anc , which determines the amplitude of the tail of the single - particle wave function of the bound ( @xmath273 ) state @xmath274(@xmath275 ) @xcite ) and in turn is itself a function of the geometric parameters ( radius of @xmath49 and diffuseness @xmath36 ) of the woods - saxon potential , i.e. @xmath276@xcite . in order to make the dependence of the @xmath87 function on @xmath88 more explicit , in the radial matrix element @xcite entering in the @xmath277 function , we split the space of interaction in two parts separated by the channel radius @xmath278 : the interior part ( @xmath279 , where nuclear forces between the @xmath280-pair are important , and the exterior part ( @xmath281 ) , where the interaction between the @xmath9-particle and @xmath282 is governed by coulomb force only . the contribution from the exterior part of the radial matrix element into the @xmath87 function does not depend on @xmath88 since for @xmath205 the wave function @xmath283 can be approximated by its asymptotic behavior @xcite . consequently , the parametrization of the astrophysical @xmath0 factor in the form ( a1 ) makes one it possible to fix a contribution from the exterior region ( @xmath281 ) , which is dominant for the peripheral reaction , by a model independent way if the ancs @xmath95 are known . it follows from here that the contribution from the interior part of the radial matrix element into the @xmath87 function , which depends on @xmath88 through the fraction @xmath284 @xcite , must exactly determine the dependence of the @xmath277 function on @xmath88 . in eq . ( a1 ) the ancs @xmath95 and the free parameter @xmath88 are unknown . but , for the peripheral @xmath48 reaction the equation ( a1 ) can be used for determination of the ancs . for this aim , obviously the following additional requirements @xcite @xmath285 and @xmath286 must be fulfilled as a function of the free parameter @xmath88 for each energy @xmath8 experimental point from the range @xmath287 and values of the function of @xmath87 from ( a2 ) . the fulfillment of the relations ( a2 ) and ( a3 ) or their violation within the experimental uncertainty for @xmath288 enables one , firstly , to determine an interval for energies e where the dominance of extra - nuclear capture occurs and , secondly , to obtain the value @xmath289 for @xmath290 using the experimental astrophysical @xmath0 factors @xmath66 , precisely measured by authors of refs . [ 611 ] , instead of @xmath98 , i.e. @xmath291 then , the value @xmath289 can be used for extrapolation of the astrophysical @xmath0 factor @xmath98 to the region of experimental inaccessible energies 0@xmath292 by using the obtained value @xmath289 in ( a1 ) . the total astrophysical @xmath0 factor for the @xmath16(g.s.+0.429 mev ) reaction is given by @xmath293 @xmath294\eqno(a6)\ ] ] @xmath295 \eqno(a7)\ ] ] in which @xmath296 and @xmath115 is a branching ratio . one notes that in the two - body potential model the anc @xmath297 is related to the nvc @xmath298 for the virtual decay @xmath299 by the equation @xcite @xmath300 where @xmath301 is the coulomb parameter for the @xmath302 bound state . in ( a8 ) the combinatorial factor taking into account the nucleon s identity is absorbed in @xmath297 and its numerical value depends on a specific model used to describe wave functions of the @xmath303 , @xmath9 and @xmath60 nuclei @xcite . hence , the proportionality factor in ( a8 ) , which relates nvc s and ancs , depends on the choice of nuclear model @xcite . but , as it is noted in @xcite , the nvc @xmath298 is a more fundamental quantity than the anc @xmath297 since the nvc is determined in a model - independent way as the residue of the partial @xmath0-matrix of the elastic @xmath304-scattering at the pole @xmath305 ( @xmath306 is the binding energy of the bound ( @xmath147 ) state of @xmath60 ) @xcite . therefore , it is also of interest to get an information about values of the nvcs from eqs . ( a4 ) and ( a8 ) . * j. n. bahcall , a. m. serenelli , and s. basu , astrophys . j. * 621 * , l85 ( 2005 ) . j. n. bahcall , w.f . huebner , s. h. lubow , p. d. parker , and r. k. ulrich , rev.mod.phys . * 54 * , 767 ( 1982 ) . j. n. bahcall , and m. h. pinsonneault , rev.mod.phys . * 64 * , 781 ( 1992 ) . e.g. adelberger , s.m.austin , j.n . bahcall , a.b.balantekin , g.bogaert , l.s.brown , l.buchmann , f.e.cecil , a.e.champagne , l. de braeckeleer , ch.a . duba , s.r . elliott , s.j . freedman , m. gai , g.goldring , ch . r. gould , a. gruzinov , w.c . haxton , k.m . heeger , and e.henley , rev . mod . phys . * 70 * , 1265 ( 1998 ) . c.angulo , m.arnould , m.rayet , p.descouvemont , d.baye , c.leclercq-willain , a.coc , s.barhoumi , p.aguer , c.rolfs , r.kunz , j.w . hammer , a.mayer , t. paradellis , s.kossionides , c.chronidou , k.spyrou , s.deglinnocenti , g. fiorentini , b.ricci , s.zavatarelli , c.providencia , h.wolters , j.soares , c.grama , j.rahighi , a.shotter , m.l.rachti , nucl . a * 656 * , 3 ( 1999 ) . b. s. nara singh , m. hass , y. nir - el , and g. haquin , phys.rev.lett . * 93 * , 262503 ( 2004 ) . d. bemmerer , f. confortola , h. costantini , a. formicola , gy . gyrky , r. bonetti , c. broggini , p. corvisiero , z. elekes , zs . flp , g. gervino , a. guglielmetti , c. gustavino , g. imbriani , m. junker , m. laubenstein , a. lemut , b. limata , v.lozza , m. marta , r. menegazzo , p. prati , v. roca , c. rolfs , c. rossi alvarez , e. somorjai , o. straniero , f. strieder , f. terrasi , and h.p . trautvetter ( the luna collaboration ) , phys.rev.lett . * 97 * , 122502 ( 2006 ) ; private communication . gyrky , f. confortola , h. costantini , a. formicola , d. bemmerer , r. bonetti , c. broggini , p. corvisiero , z. elekes , zs . flp , g. gervino , a. guglielmetti , c. gustavino , g. imbriani , m. junker , m. laubenstein , a. lemut , b. limata , v.lozza , m. marta , r. menegazzo , p. prati , v. roca , c. rolfs , c. rossi alvarez , e. somorjai , o. straniero , f. strieder , f. terrasi , and h.p . trautvetter ( the luna collaboration ) , phys.rev . c * 75 * , 035805 ( 2007 ) . f. confortola , d. bemmerer , h. costantini , a. formicola , gy . gyrky , p. bezzon , r. bonetti , c. broggini , p. corvisiero , z. elekes , zs . flp , g. gervino , a. guglielmetti , c. gustavino , g. imbriani , m. junker , m. laubenstein , a. lemut , b. limata , v.lozza , m. marta , r. menegazzo , p. prati , v. roca , c. rolfs , c. rossi alvarez , e. somorjai , o. straniero , f. strieder , f. terrasi , and h.p . trautvetter ( the luna collaboration ) , phys.rev . c * 75 * , 065803 ( 2007 ) . brown , c. bordeanu , r.f . snover , d.w . storm , d. melconian , a.l . sallaska , s.k.l . sjue , s. triambak , phys.rev . c * 76 * , 055801 ( 2007 ) . a di leva , l. gialanella , r. kunz , d. rogalla , d. schrmann , f. strieder , m. de cesare , n. de cesare , a. donofrio , z. flp , g. gyrky , g. imbariani , g. mangano , a. ordine , v. roca , c. rolfs , m. romano , e. somorjai , and f. terrasi , phys.rev.lett . * 102 * , 232502 ( 2009 ) . t. kajino , nucl.phys . a * 460 * , 559 ( 1986 ) . p.descouvement , a. adahchour , c. angulo , a. coc , e. vangioni - flam . atomic data and nuclear data tables . * 88 * , 203 ( 2004 ) . k. langanke , nucl . a * 457 * , 351 ( 1986 ) . a. cst and k. langanke , few - body systems . * 29 * , 121 ( 2000 ) . nollett , phys . c * 63 * , 054002 ( 2001 ) . dubovitchenko , a.v . dzhazairov - kakhramanov , yad . fiz . * 58 * , 635 ( 1995 ) [ phys . at * 58 * , 579 ( 1995 ) ] . * 48 * , 1664 ( 1982 ) ; nucl.phys . a * 419 * , 115 ( 1984 ) . p. mohr , h. abele , r. zwiebel , g. staudt , h. krauss , h. oberhummer , a. denker , j.w . hammer and f. wolf , phys . c * 48 * , 1420 ( 1993 ) . p. mohr , phys . c * 79 * , 065804 ( 2009 ) . igamov , t.m . tursunmuratov , and r. yarmukhamedov , yad.fiz . * 60 * , 1252 ( 1997 ) [ phys . at . nucl . * 60 * , 1126 ( 1997 ) ] . osborne , c. a. barnes , r.w . ravanagh , r.m . kremer , g.j . mathews , j.l . zyskind , p.d . parker , and a.j . howard , nucl.phys.a * 419 * , 115(1984 ) . r.f . christy and i. duck , nucl.phys . * 24 * , 89 ( 1961 ) . igamov , and r. yarmukhamedov , nucl.phys.a * 781 * , 247 ( 2007);nucl.phys.a * 832 * , 346 ( 2010 ) . blokhintsev , i.borbely , e.i . dolinskii , fiz.elem.chastits at . yadra . * 8 * , 1189 ( 1977)[sov . * 8 * , 485 ( 1977 ) ] . blokhintsev and v.o . yeromenko , yad.fiz . * 71 * , 1219 ( 2008 ) [ phys . at . nucl . * 71 * , 1126 ( 2008 ) ] . robertson , p. dyer , r.a . warner , r.c . melin , t.j . bowles , a.b . mcdonald , g.c . ball , w.g . davies , e.d . earle , phys.rev.lett . * 47 * , 1867 ( 1981 ) . kim , m.h . park , b.t . kim , phys.rev . c * 35 * 363 ( 1987 ) . s.b . igamov and r. yarmukhamedov , phys . at . nucl . * 71 * , 1740 ( 2008 ) ] . artemov , s.b . igamov , k.i . tursunmakhatov , and r. yarmukhamedov , izv . ran : seriya fizicheskaya , * 73 * , 176 ( 2009)[bull.ras : physics,*73 * , 165 ( 2009 ) ] . neudatchin , v.i . kukulin , a.n . boyarkina and v.d . korennoy , lett.nuovo cim.*5 * , 834 ( 1972 ) . kukulin , v.g . neudatchin , and yu . f. smirnov , nucl.phys . a * 245 * , 429 ( 1975 ) . kukulin , v.g . neudatchin , i.t . obukhovsky and yu.f . `` clusters as subsystems in light nuclei '' . in clustering phenomena in nuclei , vol.3 , eds . k. wildermuth and p. kramer ( vieweg , braunschweig , 1983)p.1 . dubovichenko and m.a . zhusupov , ivz . nauk kaz.ssr , ser . fiz .- mat . * 4 * , 44 ( 1983 ) ; yad . * 39 * , 1378 ( 1984 ) [ sov . phys.*39 * , 870 ( 1984 ) . ] barnard , c. m. jones , and g.c . phillips , nucl.phys . * 50 * , 629 ( 1964 ) . r. j. spiger , and t. a. tombrello , phys.rev . * 163 * ( 1967 ) 964 . h. walliser , h. kanada , and y.c . tang , nucl.phys . a * 419 * , 133 ( 1984 ) . timofeyuk , p. descouvemont , r.c . johnson , phys.rev . c * 75 * , 034302 ( 2007 ) . r. h. cyburt and b. davids , phys.rev.c * 78 * , 064614(2008 ) . d. kurath and d. j. millener , nucl.phys.a * 238 * , 269 ( 1975 ) . brune , w.h . geist , r.w . kavanagh , k.d . veal , phys.rev.lett . * 83 * , ( 1999 ) 4025 . h.krwinkel , h.w . becker , l. buchmann , j. grres , k.u . kettner , w.e . kieser , r. santo , p. schmalbrock , h.p . trautvetter , a. vlieks , c. rolfs , j.w . hammer , r.e . azuma , w.s . rodney , z.phys . a * 304 * , 307 ( 1982 ) . t. altmeyer , e. kolbe , t. warmann , k. langanke , h.j . assenbaum , z.phys . a * 330 * , 277 ( 1988 ) . u. schrder , a. redder , c. rolfs , r.e . azuma , l. buchmann , c. campbell , j.d . king , t.r . donoghue , phys.lett . b * 192 * , 55 ( 1987 ) . s. a. goncharov , j. dobesh , e. i. dolinskii , a. m. mukhamedzhanov and j. cejpek , yad . fiz.*35 * , 662 ( 1982)[sov . phys . * 35 * , 383 ( 1982 ) ] . a. m. mukhamedzanov , f. m. nunes , phys.rev . c * 72 * , 017602 ( 2005 ) . .[table1]the @xmath3indirectly determined values of the asymptotic normalization constants ( @xmath307 and ( @xmath308 ) for @xmath309 , the experimental astrophysical @xmath0 factors ( @xmath310 and @xmath111 ) and branching ratio ( @xmath113 ) at different energies @xmath8 . [ cols="^,^,^,^,^,^,^ " , ]

lepton flavor violation in and processes is studied in the littlest higgs model with t parity . we consider various asymmetries defined in polarized and decays . correlations among branching ratios and asymmetries are shown in the following lepton flavor violation processes : , , ( = al , ti , au and pb ) , , , , ( = , and ) , ( = , and ) , , , , , , and . it is shown that large parity asymmetries and time - reversal asymmetries are allowed in . for lepton flavor violation processes , sizable asymmetries are possible reflecting characteristic chirality structure of lepton flavor violating interactions in this model .

1012.4385

although the standard model ( sm ) is successful in describing almost all experimental results of high energy physics in terms of gauge theory , the dynamics responsible for the elerctroweak symmetry breaking is unknown . in the minimal version of the higgs sector , one higgs doublet field is introduced as an effective description of the dynamics below some cutoff scale @xmath27 . constraints from electroweak precision measurements indicate that @xmath27 is at least beyond 3 tev . on the other hand , this scale already introduces a fine - tuning problem of the higgs mass term at the @xmath28 % level . this is called a little hierarchy problem @xcite . little higgs models @xcite were proposed as a solution to the little hierarchy problem . in these models , the higgs doublet field appears as pseudo nambu - goldstone ( ng ) bosons of new strong dynamics at the cutoff scale . a remarkable property is that the one - loop quadratic divergence to the renormalization of the higgs mass term is cancelled by a proper choice of global and local symmetries , so that the cutoff scale can be pushed to @xmath29tev without a severe fine - tuning . a simple case is called the littlest higgs model @xcite where the higgs field is realized as a part of the ng bosons associated with the global symmetry breaking of su(5 ) to so(5 ) and the gauge symmetry is [ su(2)@xmath30u(1)]@xmath31 . then the quadratic divergence of the higgs mass term in the sm is cancelled by the extra gauge bosons and the heavy partner of the top quark . subsequent studies @xcite , however , showed that these heavy partners need to be heavier than several tev to satisfy severe constraints imposed by precise electroweak measurements , hence reintroducing some degree of the fine - tuning problem . in the littlest higgs model with t parity ( lht ) @xcite , the model is extended to have a @xmath32 parity so that the heavy gauge bosons assigned to be t - odd particles do not directly couple with a pair of the sm fermions , and the phenomenological constraints are somewhat relaxed . flavor physics provides an interesting possibility to explore the lht . in order to assign t parity , we need to introduce extra fermions ( t - odd fermions ) which are left - handed su(2 ) doublets . then , the flavor transition can arise in the vertices of heavy gauge bosons , t - odd fermions and ordinary sm fermions . there are two 3@xmath303 unitary matrices describing these flavor transitions associated with quark and lepton sectors besides the ordinary flavor mixing matrices in the quark and lepton sectors i.e. the cabibbo - kobayashi - maskawa ( ckm ) @xcite and pontecorvo - maki - nakagawa - sakata ( pmns ) @xcite matrices . these matrices are new sources of flavor changing neutral current ( fcnc ) processes in the quark sector and lepton flavor violating ( lfv ) processes in the charged lepton sector . in the early literature on this subject various observable quantities of fcnc and lfv processes were calculated in the lht and showed that the effects of t - odd partners can be sizable and the correlations among various observable quantities can be different from other new physics model such as supersymmetric ( susy ) models @xcite . it was pointed out later that calculations missed a type of diagrams and the logarithmic dependences on the cutoff scale in the fcnc and lfv amplitudes disappear thanks to new contributions @xcite , thus reducing theoretical ambiguity associated with the physics at the ultraviolet cutoff scale . branching ratios of various fcnc and lfv processes were reevaluated including the new contributions @xcite . the qualitative feature turned out to be similar to the previous calculation though there were sizable changes at the quantitative level . in this paper , we present the results of further studies in tau and muon lfv processes . in addition to branching fractions , we also study observables defined with the help of polarizations of the initial muon and tau lepton @xcite . polarized muon experiments can be done using surface muon beams that are 100% polarized in the opposite direction of the @xmath33 momentum . in fact , a @xmath34 experiment with initial muon polarization is under consideration @xcite . in the tau pair production at @xmath35 colliders , the polarization information can be obtained by taking an angular correlation with tau decays on the opposite side @xcite . the study of polarized tau decays is also being considered for the @xmath7 process at the cern large hadron collider ( lhc ) where we can utilize the tau polarization from @xmath36 decays @xcite . in the following , we consider the processes @xmath2 , @xmath3 , @xmath4 ( @xmath37 al , ti , au and pb ) , @xmath6 , @xmath7 , @xmath8 , @xmath9 ( @xmath38 , @xmath12 and @xmath13 ) , @xmath14 ( @xmath39 , @xmath17 and @xmath18 ) , @xmath19 , @xmath20 , @xmath21 , @xmath22 , @xmath23 , @xmath24 and , @xmath25 . we define a parity asymmetry in two - body decays and two parity and one time - reversal asymmetries in three - body decays . in addition , we can define forward - backward asymmetry and forward - backward - angular asymmetries in the cases of @xmath40 and @xmath21 . we show that the parity asymmetries of two - body decays reflect the characteristic chirality structure of the lfv interactions . for three - body decays , we find that there are useful relations among various asymmetries . we calculate the rates of @xmath1-@xmath41 conversions for different muonic atoms and show that the ratios of the conversion rates can vary within 1 order of magnitude over most of the lht model parameter space . these features , as well as correlations of various branching ratios , can be useful in discriminating different new physics models . this paper is organized as follows : section ii is a brief review of the lht . lfv processes are classified and various asymmetries are defined in sec . numerical results on the various observable quantities are shown in sec . section v is the conclusion . appendix a shows the general formulae of branching ratios and asymmetries for the processes studied . functions for the wilson coefficients are given in appendix b. useful formulae to perform the consistency test of the lht are presented in appendix c. in this section we review the lht in order to fix the notations we use in this paper . we use the lagrangian given in ref . @xcite with corrections discussed in refs . @xcite . gauge and higgs sectors of the littlest higgs model are described as a nonlinear @xmath42 model with the spontaneous global symmetry breaking from su(5 ) to so(5 ) with scalar fields , @xmath43 , which is transformed as @xmath44 representation of the su(5 ) . the vacuum expectation value of the scalar fields , @xmath45 , which breaks the global su(5 ) symmetry is @xmath46 generators of the unbroken so(5 ) symmetry , @xmath47 , and the broken ones of su(5)/so(5 ) , @xmath48 , satisfy the following relations , @xmath49 the lagrangian of the ng bosons and gauge fields is given by @xmath50 + \sum_{i=1,2 } \left ( -\frac{1}{2 } \text{tr } [ w_i{}^{\mu\nu } w_{i\mu \nu } ] -\frac{1}{4 } b_i{}^{\mu\nu } b_{i\mu \nu } \right ) , \label{eqlagnggauge}\end{aligned}\ ] ] where @xmath51 is the decay constant of the nonlinear @xmath42 model and the ng boson field is written as follows , @xmath52 @xmath53 fourteen ng bosons are denoted by @xmath54 , @xmath55 , @xmath56 , @xmath57 , @xmath58 , @xmath59 , and @xmath60 . as explained later , @xmath61 is identified as the sm higgs doublet . @xmath62 is the electroweak symmetry breaking vacuum expectation value , @xmath63 gev . @xmath64 and @xmath65 ( @xmath66 ) are gauge fields of the su(2)@xmath67u(1)@xmath68 , which are subgroups of the su(5 ) . the generators of the gauged subgroups are @xmath69 where @xmath70 ( @xmath711 , 2 , 3 ) are the pauli matrices . the covariant derivative of @xmath43 is defined as @xmath72 where @xmath73 and @xmath74 are gauge coupling constants of su(2)@xmath75u(1)@xmath76 . the lagrangian has a discrete symmetry under the following @xmath32 transformation , which is called t parity @xcite . @xmath77 the t parity eigenstates of the gauge bosons are @xmath78 where @xmath79 ( @xmath80 ) . as for the ng boson fields , we can show from eqs . and that @xmath54 , @xmath55 are t - even particles and the others are t - odd particles . @xmath81 and @xmath82 receive mass of the order of @xmath51 from the gauge symmetry breaking [ su(2)@xmath30u(1)]@xmath83 su(2)@xmath75u(1)@xmath76 by absorbing @xmath56 and @xmath57 . after the electroweak symmetry breaking , @xmath55 are absorbed by the sm gauge fields , @xmath84 and @xmath85 . mass eigenstates of the neutral gauge bosons are defined by @xmath86 where @xmath87 and @xmath88 are the sm @xmath89 boson and photon . @xmath90 is the weak mixing angle determined as @xmath91 . the mixing angle of the t - odd gauge bosons is given by @xmath92 where @xmath93 and @xmath94 . expanding in terms of @xmath95 , we obtain @xmath96 gauge boson masses are given as @xmath97 \nonumber\\&= & m_{w_h}^2 + o(\frac{v^4}{f^2 } ) , \\ m_{a_h}^2 & = & \frac{g^{\prime 2 } f^2}{5(c_h^2-s_h^2 ) } \left [ \left ( 1 - \frac{5}{2 } c_v^2 s_v^2 \right ) c_h^2 - \frac{5g^2}{g^{\prime 2 } } \left ( 1 - \frac{c_v^2 s_v^2}{2 } \right ) s_h^2 \right ] \nonumber\\&= & \frac{g^{\prime 2}f^2}{5 } \left ( 1 - \frac{5v^2}{4f^2 } + o(\frac{v^4}{f^4 } ) \right),\end{aligned}\ ] ] where @xmath98 and @xmath99 . we consider terms up to @xmath100 so that we neglect the difference between @xmath101 and @xmath102 in our analysis . the fermion sector of the lht consists of three families of quark and lepton fields @xmath103 , @xmath104 , @xmath105 , @xmath106 , @xmath107 , @xmath108 , @xmath109 and @xmath110 for each generation , and a set of the `` top - partner '' fermions , @xmath111 and @xmath112 . following the procedure given in ref . @xcite , we assign gauge charges of the fermion fields as summarized in table [ tabcharge ] . the kinetic terms of left - handed fermions are given by @xmath113 where these fermions are introduced as incomplete multiplets of the su(5 ) , @xmath114 and @xmath115 . under the t parity , they transform as @xmath116 thus the sm leptons and the t - odd leptons are given as the t parity eigenstates , @xmath117 respectively . the right - handed sm fermions are introduced in the same manner as in the sm . the right - handed t - odd leptons are introduced as a nonlinear representation of the su(5 ) , @xmath118 where upper three components , @xmath119 and @xmath120 , are heavy enough to be neglected in low energy dynamics . @xmath121 changes the sign with the t parity . in order to construct the gauge invariant kinetic term , we define @xmath122 as @xmath123 which transform linearly under the su(5 ) transformation . thus the kinetic term is given by @xmath124 where @xmath125 and @xmath126 . the quark sector is constructed in the same procedure . for the quark sector , @xmath127 and @xmath128 . @xmath129{|c|cccc| } \hline & \text{su(2)}_1 & \text{su(2)}_2 & y_1 & y_2 \\ \hline q_1 = \left(\begin{array}{c } u_1 \\ d_1 \end{array}\right ) & \mathbf{2 } & \mathbf{1 } & \frac{1}{30 } & \frac{4}{30 } \\ q_2 = \left(\begin{array}{c } u_2 \\ d_2 \end{array}\right ) & \mathbf{1 } & \mathbf{2 } & \frac{4}{30 } & \frac{1}{30 } \\ \hline t'_1 & \mathbf{1 } & \mathbf{1 } & \frac{16}{30 } & \frac{4}{30 } \\ t'_2 & \mathbf{1 } & \mathbf{1 } & \frac{4}{30 } & \frac{16}{30 } \\ \hline \ell_1 = \left(\begin{array}{c } \nu_1 \\ e_1 \end{array}\right ) & \mathbf{2 } & \mathbf{1 } & -\frac{3}{10 } & -\frac{2}{10 } \\ \ell_2 = \left(\begin{array}{c } \nu_2 \\ e_2 \end{array}\right ) & \mathbf{1 } & \mathbf{2 } & -\frac{2}{10 } & -\frac{3}{10 } \\ \hline \end{array } \qquad \begin{array}[t]{|c|cccc| } \hline & \text{su(2)}_1 & \text{su(2)}_2 & y_1 & y_2 \\ \hline u_r & \mathbf{1 } & \mathbf{1 } & \frac{1}{3 } & \frac{1}{3 } \\ d_r & \mathbf{1 } & \mathbf{1 } & -\frac{1}{6 } & -\frac{1}{6 } \\ \hline t'_{1r } & \mathbf{1 } & \mathbf{1 } & \frac{16}{30 } & \frac{4}{30 } \\ t'_{2r } & \mathbf{1 } & \mathbf{1 } & \frac{4}{30 } & \frac{16}{30 } \\ \hline \nu_r & \mathbf{1 } & \mathbf{1 } & 0 & 0 \\ e_r & \mathbf{1 } & \mathbf{1 } & -\frac{1}{2 } & -\frac{1}{2 } \\ \hline \end{array}\ ] ] the yukawa couplings of the t - odd fermions are @xmath130 with generation indices , @xmath131 and @xmath132 . thus their masses are given by @xmath133 where @xmath131 is the generation index and @xmath134 are eigenvalues of the yukawa coupling matrix , @xmath135 . we neglect @xmath136 differences between @xmath137 and @xmath138 in the calculation of lfv processes because these differences only contribute to the higher order corrections in the @xmath95 expansion . for the sm charged leptons , the gauge invariant yukawa coupling terms are written as @xmath139 e_{r}^j + \text{h.c . } , \label{eqdowny}\end{aligned}\ ] ] where @xmath140 and @xmath141 are su(5 ) indices and @xmath131 and @xmath132 are generation indices . the left - handed sm lepton doublets are embedded in su(5 ) multiplets @xmath142 and @xmath142 as @xmath143 where @xmath144 and @xmath145 are su(2)@xmath68 singlet scalar fields whose u(1 ) gauge charges are @xmath146 and @xmath147 , respectively . following ref . @xcite , we use @xmath148 and its t parity conjugate as @xmath144 and @xmath145 , respectively . as explained in ref . @xcite , the mass matrices of the sm fermions and the t - odd fermions are not simultaneously diagonalized in general . consequently , new mixing matrices are introduced in the interaction among the t - odd gauge bosons and fermions . we extract the interaction terms of the t - odd @xmath36 boson and leptons from eq . ( [ eqgaugeint ] ) as @xmath149 where @xmath150 and @xmath151 are new mixing matrices . these two mixing matrices are related to each other by @xmath152 where @xmath153 is the pmns matrix @xcite , because the mass matrices of @xmath154 and @xmath155 are determined by the same yukawa coupling matrix @xmath135 . @xmath150 and @xmath151 also appear in the interaction of the neutral t - odd gauge bosons and leptons . we parameterize @xmath150 as @xmath156 where @xmath157 and @xmath158 . we take @xmath159 and @xmath160 as input parameters . notice that there remain three unremovable phases in @xmath150 after the phase convention of the pmns matrix is fixed , as pointed out in ref . @xcite . for the quark sector , we define new mixing matrices @xmath161 and @xmath162 in the same way as @xmath150 and @xmath151 . we take @xmath161 as a matrix parametrized like eq . and the other mixing matrix , @xmath162 , is expressed by @xmath161 and the ckm matrix @xcite in the following way : @xmath163 in this section , we present formulae for observables in the lfv processes in the lht . in contrast to previous works @xcite , we consider various angular distributions in polarized @xmath1 and @xmath0 decays in addition to the branching ratios and the @xmath164 conversion rates in various muonic atoms . studying the angular distributions is useful to extract information about the chirality structure of the low energy effective lagrangian . the lfv processes that we consider in this paper are the following . * radiative two - body decays : @xmath165 , @xmath166 and @xmath167 . * leptonic three - body decays of @xmath1 and @xmath0 . we classify these decay modes into three types according to the lepton flavor combination in the final state . * * type i : @xmath168 , @xmath169 and @xmath170 . all the three leptons in the final state have the same flavor . * * type ii : @xmath171 and @xmath172 . in these processes , the final state consists of a pair of same flavor leptons with opposite charges and another lepton with different flavor . there are no identical particles in the final state . * * type iii : @xmath173 and @xmath174 . lepton flavor changes by more than one . * semileptonic two - body decay of @xmath0 . we consider @xmath175 and @xmath176 modes , where @xmath10 and @xmath15 stand for a pseudoscalar meson ( @xmath177 , @xmath57 or @xmath13 ) and a vector meson ( @xmath16 , @xmath17 or @xmath178 ) , respectively , as well as corresponding @xmath179 modes . we do not consider @xmath180 modes , where both quark and lepton flavors are violated . * @xmath164 conversion in a muonic atom . in these processes , @xmath4 , we discuss aluminum ( al ) , titanium ( ti ) , gold ( au ) and lead ( pb ) as target atom , @xmath5 , because the conversion rates depend on target nuclei @xcite . in the lht , lfv processes are induced by loop diagrams with the t - odd gauge bosons and leptons in the internal lines . contributions of the t - even particles are negligible because they are suppressed by neutrino masses . since only the left - handed sm ( t - even ) leptons couple to the t - odd gauge bosons , the low energy effective lagrangian for lfv has a restricted chirality structure : the lepton flavor changing occurs in the left - handed lepton sector only . in the following subsections , we present the formulae for the effective lagrangians , branching ratios , and various asymmetries , which can be defined based on angular distributions . except for the @xmath164 conversion , we explicitly show formulae for @xmath0 to @xmath1 transitions . those for @xmath1 to @xmath41 and @xmath0 to @xmath41 are obtained by appropriate replacements of flavor indices . for simplicity , we neglect lepton masses in the final states . in the following , formulae of the low energy effective lagrangians and the observable quantities are shown based on the lfv interactions relevant to the lht . those in the case of the most general low energy effective lagrangians are given in appendix [ appformulae ] . the effective lagrangian for @xmath6 in the lht consists of the following dipole moment type operator . @xmath181 , \label{eq : llg}\end{aligned}\ ] ] where @xmath182 with @xmath183 the functions @xmath184 and @xmath185 are given in appendix [ appwilson ] . the coefficient of the operator with the opposite chirality , @xmath186 , is suppressed by @xmath187 due to the chirality properties in the interaction of the t - odd gauge bosons . for a @xmath188 decay of a polarized @xmath189 , we define a decay angle @xmath190 as the angle between the spin of @xmath189 and the momentum of @xmath33 in the rest frame of @xmath189 . the configuration of spin and momenta are shown in fig . [ fig : pllg ] . the differential branching ratio with respect to @xmath191 is given by @xmath192 we define the polarization asymmetry , which characterizes the chirality structure of the effective interaction in ( [ eq : llg ] ) , as @xmath193 where the integrated branching ratio is @xmath194 @xmath195 is a parity odd quantity since @xmath191 is proportional to the inner product @xmath196 , where @xmath197 and @xmath198 are spin of initial @xmath189 and spatial momentum of final @xmath33 , respectively . by substituting ( [ eq : dbr - tmg ] ) , we obtain @xmath199 the parameter independent result of @xmath195 is a consequence of the special chirality structure of the lht . decay . the dashed arrow denotes the direction of the @xmath189 spin . the solid and the wavy arrows show the directions of momenta of the @xmath33 and the photon , respectively . @xmath190 is defined as the angle between the directions of the @xmath189 spin and the @xmath33 momentum . ] the effective lagrangian for type i leptonic three - body decay , @xmath202 , in the lht is given as @xmath203 , \label{eq : leff - tmmm}\end{aligned}\ ] ] where @xmath204 the functions @xmath205 , @xmath206 and @xmath207 represent contributions of photon penguin , @xmath89 penguin and box diagrams , respectively , and are given in appendix [ appwilson ] . @xmath208 is shown in ( [ eqdipole ] ) . the spin and momentum configuration of a three - body decay of a polarized @xmath189 at rest is determined by two energy variables and two angular variables , as depicted in fig . [ figpl3l ] . the three spatial momenta of the particles in the final state are denoted as @xmath209 , @xmath210 , and @xmath211 . we take @xmath212 axis in the direction of @xmath209 and the decay plane is identified as the @xmath212-@xmath213 plane . the direction of @xmath213 axis is chosen in such a way that the @xmath213 component of @xmath210 has a positive value . we take the @xmath214 axis along @xmath215 . the direction of the polarization vector of the initial particle is parametrized by the polar and the azimuthal angles @xmath190 and @xmath178 , respectively . for the @xmath7 decay , we identify the momentum of @xmath216 as @xmath209 . two @xmath33 s are distinguished by their energies . the momentum of the @xmath33 with larger ( smaller ) energy is identified as @xmath210 ( @xmath211 ) . as for the energy variables , we define @xmath217 where @xmath218 are energies of @xmath33 s with momenta @xmath219 . @xmath220 is larger than @xmath221 by definition . possible ranges of @xmath222 in an approximation of massless final particles are @xmath223 however , the @xmath224 decay amplitude is singular for @xmath225 , where the invariant mass of the pair of @xmath33 ( with @xmath211 ) and @xmath216 is vanishing within the approximation of neglecting the muon mass . the singularity comes from the pole of the photon propagator in the contribution from the dipole moment operator . we introduce a cutoff parameter @xmath226 and take the integration interval of @xmath227 as @xmath228 in order to avoid this collinear singularity . -@xmath213 plane where the @xmath212 axis is taken to be the direction of particle @xmath229 s momentum . the @xmath213 axis is defined so that the @xmath213 component of particle @xmath230 s momentum is positive . the direction of the initial lepton polarization vector @xmath197 is parametrized by the polar and the azimuthal angles @xmath190 and @xmath178 , respectively . ] the @xmath231-dependent branching ratio is calculated by integration over @xmath222 , @xmath191 and @xmath178 as @xmath232 the general formulae of differential branching ratios are shown in appendix [ appformulae ] . we also define the following angular asymmetries : @xmath233 the branching ratio and asymmetries in the lht are expressed as @xmath234 @xmath235 @xmath236 the functions @xmath237 , @xmath238 , and @xmath239 are given in appendix [ appformulaet1 ] . @xmath240 , _ etc . _ are written in terms of the wilson coefficients in ( [ eq : leff - tmmm ] ) as @xmath241 , \quad c_{j3}^{\text{i , lht } } = \text{re } [ ea_r^{\text{lht } } g_{lr}^{\text{i , lht } \ast } ] , \\ c_{j5}^{\text{i , lht } } & = \text{im } [ ea_r^{\text{lht } } g_{ll}^{\text{i , lht } \ast } ] , \quad c_{j6}^{\text{i , lht } } = \text{im } [ ea_r^{\text{lht } } g_{lr}^{\text{i , lht } \ast } ] .\end{aligned}\ ] ] as explained in appendix [ appformulaet1 ] , @xmath242 and @xmath243 are parity odd asymmetries and @xmath244 is a time - reversal asymmetry . the time - reversal asymmetry @xmath244 is induced by the cp violation in the effective lagrangian , since we are considering a theory with cpt invariance . in fact , @xmath245 and @xmath246 are proportional to the jarlskog invariant @xcite defined for @xmath150 : @xmath247 \nonumber\\ & = c_{12}^{\ell } c_{13}^{\ell}{}^2 c_{23}^{\ell } s_{12}^{\ell } s_{13}^{\ell } s_{23}^{\ell } \sin ( \delta_{12}^{\ell } -\delta_{13}^{\ell } + \delta_{23}^{\ell } ) .\end{aligned}\ ] ] therefore , a three - generation mixing is necessary for nonvanishing @xmath244 . notice that two of three complex phases in @xmath150 are removable when t - even neutrino masses are neglected . that is why @xmath248 depends on only one combination of the phases . the effective lagrangian for a type ii trilepton decay @xmath40 has a similar structure to ( [ eq : leff - tmmm ] ) : @xmath249,\end{aligned}\ ] ] where @xmath250 { \notag \\ } & + \sum_{l , m } \bigl [ ( v_{h\ell } ) ^{\ast}_{l3 } ( v_{h\ell } ) _ { l2 } ( v_{h\ell } ) ^{\ast}_{m1 } ( v_{h\ell } ) _ { m1 } { \notag \\ } & \qquad + ( v_{h\ell } ) ^{\ast}_{l3 } ( v_{h\ell } ) _ { l1 } ( v_{h\ell } ) ^{\ast}_{m1 } ( v_{h\ell } ) _ { m2 } \bigr ] b_{(e)}(x_l , x_m ) \biggr ) , \label{eqwilsontypeii}\end{aligned}\ ] ] and @xmath251 @xmath208 is the same as ( [ eqdipole ] ) . kinematical variables are defined in a similar way to the type i case . for the momentum assignment , we identify particles @xmath229 , @xmath230 and @xmath252 in fig . [ figpl3l ] as @xmath33 , @xmath253 and @xmath254 , respectively . a difference from the type i case is that both @xmath255 and @xmath256 are possible in type ii modes because there are no identical particles in the final state . therefore , the range of @xmath257 and @xmath258 are @xmath259 where the cutoff parameter @xmath260 is introduced to avoid the collinear singularity in @xmath261 . formulae for the branching ratio and the angular asymmetries @xmath262 in @xmath171 decay in the lht are obtained as @xmath263 @xmath264 @xmath265 where @xmath266 , \quad c_{j3}^{\text{ii , lht } } = \text{re } [ ea_r^{\text{lht } } g_{lr}^{\text{ii , lht}\ast } ] , \\ c_{j7}^{\text{ii , lht } } & = \text{im } [ ea_r^{\text{lht } } g_{ll}^{\text{ii , lht}\ast } ] , \quad c_{j8}^{\text{ii , lht } } = \text{im } [ ea_r^{\text{lht } } g_{lr}^{\text{ii , lht}\ast } ] .\end{aligned}\ ] ] the functions @xmath267 , @xmath268 and @xmath269 are given in appendix [ appformulaet2 ] . in the type ii decay , another class of asymmetries can be defined . we define the energy asymmetry in @xmath270 as the asymmetry between the partial widths with @xmath271 and @xmath272 in the rest frame of initial @xmath189 : @xmath273 this asymmetry is also called as the forward - backward asymmetry , because @xmath274 ( @xmath275 ) corresponds to @xmath276 ( @xmath277 ) in the rest frame of the @xmath278 pair . furthermore , we define the following asymmetries combining asymmetric integrations over @xmath279 and @xmath280 : @xmath281 in the lht , we obtain @xmath282 where the functions @xmath283 and @xmath284 are given in appendix [ appformulaet2 ] . the type iii leptonic decay modes , such as @xmath24 , are those in which the lepton flavor changes by two . at the one - loop level , only the box diagrams contribute to these decay modes . the effective lagrangian for @xmath24 is @xmath286,\end{aligned}\ ] ] where @xmath287 the kinematics of the type iii decay is treated in a way similar to the type i case . for @xmath24 mode , @xmath254 and one of the @xmath33 s , which has a larger energy in the rest frame of @xmath189 , is identified as particles @xmath229 and @xmath230 in fig . [ figpl3l ] , respectively . we define the angular asymmetries @xmath242 , @xmath243 and @xmath244 by ( [ eq : az - tmmm ] ) , ( [ eq : ax - tmmm ] ) and ( [ eq : ay - tmmm ] ) , respectively . the forward - backward type asymmetries are not defined because there are identical particles in the final state as in the type i case . the branching ratio in the lht is written as @xmath288 where @xmath289 and @xmath290 is given in appendix [ appformulaet1 ] . since the differential width is written in terms of one combination of the wilson coefficient , @xmath291 , the angular asymmetries are determined independently of the input parameters : @xmath292 the effective lagrangian for semileptonic @xmath295 decay processes are written as @xmath296 . \label{eqhlfv}\end{aligned}\ ] ] the wilson coefficients are given as @xmath297 where @xmath298 and @xmath299 are the weak isospin and the electromagnetic charge for quarks , respectively . @xmath208 is given in ( [ eqdipole ] ) . for the two - body decay process @xmath300 , where @xmath10 stands for a neutral pseudoscalar meson @xmath177 , @xmath57 or @xmath13 , the differential branching ratio is written as @xmath301 where the decay angle @xmath190 is defined in the same way as in the @xmath302 case . the effective coupling @xmath303 for @xmath304 and @xmath305 are written in terms of the wilson coefficients in ( [ eqhlfv ] ) as @xmath306 @xmath307 , @xmath308 and @xmath309 are the decay constants of the pseudoscalar mesons that are defined in appendix [ appformulaemeson ] . the effective coupling for @xmath310 is obtained by replacing the subscript @xmath57 with @xmath13 in the expression of @xmath311 . formulae for the branching ratio and the polarization asymmetry @xmath312 are derived from ( [ eq : dbdtheta - tmp ] ) as @xmath313 @xmath312 is purely determined by the chirality structure of the effective lagrangian ( [ eqhlfv ] ) and independent of the values of the input parameters . the semileptonic two - body decay @xmath14 , where @xmath15 denotes a neutral vector meson @xmath16 , @xmath17 , or @xmath178 are also described by the effective lagrangian ( [ eqhlfv ] ) . the branching ratio and the polarization asymmetry @xmath316 , which is defined similarly to @xmath317 and @xmath312 , are written as @xmath318 , \\ a_v(\tau^+ \to \mu^+ v)_{\text{lht } } & = \frac{1}{2b_v^{\text{lht } } } { \notag \\}\times & \left ( { \left|g_{rav}\right|}^2 \left ( 2 - \frac{m_v^2}{m_{\tau}^2 } \right ) + { \left|g_{lv } \right|}^2 \left ( \frac{m_{\tau}^2 -2 m_v^2}{4 m_v^2 } \right ) + \text{re } [ g_{rav } g_{lv}^{\ast } ] \right).\end{aligned}\ ] ] the effective coupling constants for @xmath319 , @xmath17 and @xmath178 are @xmath320 where the decay constants @xmath321 are defined in appendix [ appformulaemeson ] . the effective lagrangian for the coherent @xmath164 conversion processes has the same structure as ( [ eqhlfv ] ) . \label{eq : mueconvlefflht}\end{aligned}\ ] ] the wilson coefficients are obtained by replacing the lepton flavor indices 3(@xmath0 ) and 2(@xmath1 ) with 2(@xmath1 ) and 1(@xmath41 ) in eqs . ( [ eqdipole ] ) , ( [ eq : gllq - tm ] ) and ( [ eq : glrq - tm ] ) . we calculate the @xmath164 conversion rates following the method given in ref . the @xmath164 conversion branching ratio in the lht is written as @xmath323 where @xmath324 is the muon capture rate . @xmath325 , @xmath326 and @xmath327 are the overlap integrals defined in ref . @xcite . for the reader s convenience we quote the values for al , ti , au and pb in appendix [ appformulaemeconv ] . in this section we show numerical results for the lfv observables given in the previous section . we have to specify 19 parameters in order to calculate lfv observables in the lht : the decay constant @xmath51 , six masses of the t - odd fermions @xmath137 and @xmath328 , six mixing angles and six phases in the mixing matrices @xmath150 and @xmath161 . the value of @xmath51 is constrained to @xmath329 gev by the electroweak precision measurements @xcite . throughout the analysis in this paper , we fix @xmath51 as @xmath330 gev . since all the lfv amplitudes are proportional to @xmath331 as shown in sec . [ seclhtlfv ] , the branching ratios scale as @xmath332 . we also assume that the t - odd quarks ( except for the top partner ) are degenerate in mass for simplicity . under this assumption , t - odd particle loops do not induce additional contributions to the quark fcnc observables , and the mixing and the phase parameters in @xmath161 are irrelevant . we fix the t - odd quark mass as 500 gev . the top - partner quark mass is irrelevant for the lfv studied here . there remain nine free parameters in the t - odd lepton sector . we vary these parameters independently within the ranges @xmath333 ( @xmath334 ) , @xmath335 [ @xmath336 and @xmath337 [ @xmath336 . for the type i and type ii trilepton decays , we use following cutoff parameters : @xmath338 for the type iii modes we take @xmath339 since there are no singularity in the differential widths . we evaluate the observables explained in sec . [ seclhtlfv ] for each model parameter set , and check if that set is allowed under current experimental constraints . at present , the upper limits of the branching ratios are available for various lfv processes . we use the values summarized in table [ tabbound ] . .experimental upper bounds of various lfv branching ratios . [ cols="<,<,<",options="header " , ] the following functions are used to describe the wilson coefficients in sec.[seclhtlfv ] : @xmath340 @xmath341 @xmath342 where , @xmath343 we introduce useful formulae to parametrize the various branching ratios and asymmetries in the lht . since the observables are expressed by a small number of parameters , we can obtain implicit relations among these quantities , which can be used for a consistency check of the model . the nonzero elements of the wilson coefficients are only of three types : @xmath208 , @xmath344 and @xmath345 . here we consider the @xmath346 transition . corresponding formulae for the @xmath347 transition are also derived . according to eqs . , , , and , we can derive the following parametrization : @xmath349 , \\ \frac{96\pi}{\alpha } \frac { \text{ba}_z ( \tau^+ \to \mu^+ \mu^+ \mu^- ) } { \text{br}(\tau^+ \to \mu^+ \gamma ) } = & -4 \left ( 4\ln(\delta ) + \frac{49}{3 } \right ) + \frac{1}{6 } { \left|z\right|}^2 -{\left| -z + \delta^{\text{i } } \right|}^2 { \notag \\ } & - 4\text{re}[-z + 2\delta^{\text{i}}],\\ \frac{96\pi}{\alpha } \frac { \text{ba}_x ( \tau^+ \to \mu^+ \mu^+ \mu^- ) } { \text{br}(\tau^+ \to \mu^+ \gamma ) } = & \frac{64}{35 } \left ( -21 + \frac{1}{12 } { \left|z\right|}^2 -\text{re } [ -z + 3\delta^{\text{i } } ] \right ) , \\ \frac{96\pi}{\alpha } \frac { \text{ba}_y ( \tau^+ \to \mu^+ \mu^+ \mu^- ) } { \text{br}(\tau^+ \to \mu^+ \gamma ) } = & \frac{64}{35 } \text{im}[-z + 3\delta^{\text{i}}],\end{aligned}\ ] ] @xmath350 , \\ \frac{96\pi}{\alpha } \frac { \text{ba}_z ( \tau^+ \to \mu^+ e^+ e^- ) } { \text{br}(\tau^+ \to \mu^+ \gamma ) } = & \frac{1}{6 } \left ( 32 ( 3\ln(\delta ) + 5 ) + { \left|z\right|}^2 + { \left| -z + \delta^{\text{ii } } \right|}^2 -8\text{re } [ -2z + \delta^{\text{ii } } ] \right),\\ \frac{96\pi}{\alpha } \frac { \text{ba}_x ( \tau^+ \to \mu^+ e^+ e^- ) } { \text{br}(\tau^+ \to \mu^+ \gamma ) } = & \frac{4\pi}{35 } \left ( -{\left|z\right|}^2 + { \left|-z + \delta^{\text{ii } } \right|}^2 + 16\text{re } [ \delta^{\text{ii } } ] \right ) , \label{eqpropx}\\ \frac{96\pi}{\alpha } \frac { \text{ba}_y ( \tau^+ \to \mu^+ e^+ e^- ) } { \text{br}(\tau^+ \to \mu^+ \gamma ) } = & \frac{48\pi}{35 } \text{im } [ \delta^{\text{ii}}],\end{aligned}\ ] ] @xmath351 ) , \label{eqpropfb}\\ \frac{96\pi}{\alpha } \frac { \text{ba}_{zfb } ( \tau^+ \to \mu^+ e^+ e^- ) } { \text{br}(\tau^+ \to \mu^+ \gamma ) } = & -\frac{1}{8 } \left ( -{\left|z\right|}^2 + { \left| -z + \delta^{\text{ii } } \right| } + 16\text{re } [ \delta^{\text{ii } } ] \right),\label{eqpropzfb}\\ \frac{96\pi}{\alpha } \frac { \text{ba}_{xfb } ( \tau^+ \to \mu^+ e^+ e^- ) } { \text{br}(\tau^+ \to \mu^+ \gamma ) } = & \frac{16}{105 } \left ( 56 + { \left|z\right|}^2 + { \left| -z + \delta^{\text{ii } } \right|}^2 + 16 \text{re } [ -2z + \delta^{\text{ii } } ] \right ) , \\ \frac{96\pi}{\alpha } \frac { \text{ba}_{yfb } ( \tau^+ \to \mu^+ e^+ e^- ) } { \text{br}(\tau^+ \to \mu^+ \gamma ) } = & \frac{64}{35 } \text{im } [ -2z + \delta^{\text{ii } } ] , \end{aligned}\ ] ] where ba@xmath352 ( @xmath353 , @xmath144 , @xmath354 , @xmath355 , @xmath356 , @xmath357 and @xmath358 ) denotes asymmetry , @xmath359 multiplied by the branching ratio . we take the limit @xmath360 except for logarithmically divergent terms in @xmath361 , @xmath362 , @xmath363 and @xmath364 . the proportional relation is found from eqs . , and . furthermore , ten independent observables can be expressed by three complex functions . 99 r. barbieri and a. strumia , phys . b * 462 * ( 1999 ) 144 [ arxiv : hep - ph/9905281 ] . n. arkani - hamed , a. g. cohen and h. georgi , phys . b * 513 * ( 2001 ) 232 [ arxiv : hep - ph/0105239 ] ; n. arkani - hamed , a. g. cohen , t. gregoire and j. g. wacker , jhep * 0208 * ( 2002 ) 020 [ arxiv : hep - ph/0202089 ] ; n. arkani - hamed , a. g. cohen , e. katz , a. e. nelson , t. gregoire and j. g. wacker , jhep * 0208 * ( 2002 ) 021 [ arxiv : hep - ph/0206020 ] . n. arkani - hamed , a. g. cohen , e. katz and a. e. nelson , jhep * 0207 * ( 2002 ) 034 [ arxiv : hep - ph/0206021 ] . c. csaki , j. hubisz , g. d. kribs , p. meade and j. terning , phys . d * 67 * ( 2003 ) 115002 [ arxiv : hep - ph/0211124 ] ; j. l. hewett , f. j. petriello and t. g. rizzo , jhep * 0310 * ( 2003 ) 062 [ arxiv : hep - ph/0211218 ] ; c. csaki , j. hubisz , g. d. kribs , p. meade and j. terning , phys . d * 68 * ( 2003 ) 035009 [ arxiv : hep - ph/0303236 ] . h. c. cheng and i. low , jhep * 0309 * ( 2003 ) 051 [ arxiv : hep - ph/0308199 ] ; h. c. cheng and i. low , jhep * 0408 * ( 2004 ) 061 [ arxiv : hep - ph/0405243 ] . n. cabibbo , phys . lett . * 10 * ( 1963 ) 531 ; m. kobayashi and t. maskawa , prog . * 49 * ( 1973 ) 652 . b. pontecorvo , sov . jetp * 6 * , 429 ( 1957 ) [ zh . eksp . 33 * , 549 ( 1957 ) ] ; sov . jetp * 7 * , 172 ( 1958 ) [ zh . eksp . teor . fiz . * 34 * , 247 ( 1957 ) ] ; sov . jetp * 26 * , 984 ( 1968 ) [ zh . eksp . fiz . * 53 * , 1717 ( 1967 ) ] ; z. maki , m. nakagawa and s. sakata , prog . * 28 * ( 1962 ) 870 . j. hubisz , s. j. lee and g. paz , jhep * 0606 * ( 2006 ) 041 [ arxiv : hep - ph/0512169 ] ; m. blanke , a. j. buras , a. poschenrieder , s. recksiegel , c. tarantino , s. uhlig and a. weiler , phys . lett . b * 646 * ( 2007 ) 253 [ arxiv : hep - ph/0609284 ] . m. blanke , a. j. buras , a. poschenrieder , c. tarantino , s. uhlig and a. weiler , jhep * 0612 * ( 2006 ) 003 [ arxiv : hep - ph/0605214 ] ; m. blanke , a. j. buras , s. recksiegel , c. tarantino and s. uhlig , phys . b * 657 * ( 2007 ) 81 [ arxiv : hep - ph/0703254 ] ; m. blanke , a. j. buras , s. recksiegel , c. tarantino and s. uhlig , jhep * 0706 * ( 2007 ) 082 [ arxiv:0704.3329 [ hep - ph ] ] ; m. blanke , a. j. buras , s. recksiegel _ et al . _ , [ arxiv:0805.4393 [ hep - ph ] ] . m. blanke , a. j. buras , a. poschenrieder , s. recksiegel , c. tarantino , s. uhlig and a. weiler , jhep * 0701 * ( 2007 ) 066 [ arxiv : hep - ph/0610298 ] . s. r. choudhury , a. s. cornell , a. deandrea , n. gaur and a. goyal , phys . d * 75 * ( 2007 ) 055011 [ arxiv : hep - ph/0612327 ] . m. blanke , a. j. buras , b. duling , a. poschenrieder and c. tarantino , jhep * 0705 * ( 2007 ) 013 [ arxiv : hep - ph/0702136 ] . t. goto , y. okada and y. yamamoto , phys . b * 670 * ( 2009 ) 378 [ arxiv:0809.4753 [ hep - ph ] ] . f. del aguila , j. i. illana and m. d. jenkins , jhep * 0901 * ( 2009 ) 080 [ arxiv:0811.2891 [ hep - ph ] ] . m. blanke , a. j. buras , b. duling _ et al . _ , acta phys . * b41 * ( 2010 ) 657 - 683 . [ arxiv:0906.5454 [ hep - ph ] ] . f. del aguila , j. i. illana , m. d. jenkins , jhep * 1009 * ( 2010 ) 040 . [ arxiv:1006.5914 [ hep - ph ] ] . y. okada , k. -i . okumura , y. shimizu , phys . rev . * d61 * ( 2000 ) 094001 . [ hep - ph/9906446 ] . r. kitano , y. okada , phys . * d63 * ( 2001 ) 113003 . [ hep - ph/0012040 ] . s. yamada , talk at the 8th international workshop on tau - lepton physics : http://meg.icepp.s.u-tokyo.ac.jp/docs/talks/s_yamada/tau04.shuei.yamada.ppt . m. giffels , j. kallarackal , m. kramer , b. oleary and a. stahl , phys . * d77 * ( 2008 ) 073010 . [ arxiv:0802.0049 [ hep - ph ] ] . j. hubisz and p. meade , phys . d * 71 * ( 2005 ) 035016 [ arxiv : hep - ph/0411264 ] ; c. r. chen , k. tobe and c. p. yuan , phys . lett . b * 640 * ( 2006 ) 263 [ arxiv : hep - ph/0602211 ] ; a. belyaev , c. r. chen , k. tobe and c. p. yuan , phys . rev . d * 74 * ( 2006 ) 115020 [ arxiv : hep - ph/0609179 ] . r. kitano , m. koike and y. okada , phys . d * 66 * , 096002 ( 2002 ) [ erratum - ibid . d * 76 * , 059902 ( 2007 ) ] [ arxiv : hep - ph/0203110 ] ; v. cirigliano , r. kitano , y. okada and p. tuzon , phys . * d80 * ( 2009 ) 013002 . [ arxiv:0904.0957 [ hep - ph ] ] . c. jarlskog , phys . * 55 * ( 1985 ) 1039 ; c. jarlskog , z. phys . * c29 * ( 1985 ) 491 - 497 . j. hubisz , p. meade , a. noble _ et al . _ , jhep * 0601 * ( 2006 ) 135 . [ hep - ph/0506042 ] ; m. asano , s. matsumoto , n. okada and y. okada , phys . * d75 * ( 2007 ) 063506 . [ hep - ph/0602157 ] . m. l. brooks _ et al . _ [ mega collaboration ] , phys . * 83 * ( 1999 ) 1521 [ arxiv : hep - ex/9905013 ] . u. bellgardt _ et al . _ [ sindrum i collaboration ] , nucl . phys . * 299 * , 1 ( 1988 ) . c. dohmen _ et al . _ [ sindrum ii collaboration ] , phys . b * 317 * , 631 ( 1993 ) . w. h. bertl _ et al . _ [ sindrum ii collaboration ] , eur . j. * c47 * ( 2006 ) 337 - 346 . w. honecker _ et al . _ [ sindrum ii collaboration ] , phys . rev . * 76 * , 200 ( 1996 ) . b. aubert _ et al . _ [ babar collaboration ] , phys . * 104 * ( 2010 ) 021802 . [ arxiv:0908.2381 [ hep - ex ] ] . y. miyazaki _ et al . _ [ belle collaboration ] , phys . b * 660 * ( 2008 ) 154 [ arxiv:0711.2189 [ hep - ex ] ] . b. aubert _ et al . _ [ babar collaboration ] , phys . * 98 * ( 2007 ) 061803 [ arxiv : hep - ex/0610067 ] . y. miyazaki _ et al . _ [ belle collaboration ] , phys . b * 648 * ( 2007 ) 341 [ arxiv : hep - ex/0703009 ] . b. aubert _ et al . _ [ babar collaboration ] , phys . * 103 * ( 2009 ) 021801 . [ arxiv:0904.0339 [ hep - ex ] ] . y. nishio _ et al . _ [ belle collaboration ] , phys . b * 664 * ( 2008 ) 35 [ arxiv:0801.2475 [ hep - ex ] ] . b. aubert _ et al . _ [ babar collaboration ] , phys . * 99 * ( 2007 ) 251803 [ arxiv:0708.3650 [ hep - ex ] ] . b. aubert _ et al . _ [ babar collaboration ] , phys . * 100 * ( 2008 ) 071802 [ arxiv:0711.0980 [ hep - ex ] ] . j. adam _ et al . _ [ meg collaboration ] , nucl . * b834 * ( 2010 ) 1 - 12 . [ arxiv:0908.2594 [ hep - ex ] ] ; r. sawada , proc . ichep2010 ( 2010 ) 263 . y. g. cui _ et al . _ [ comet collaboration ] , kek report 2009 - 10 ; r. m. carey _ et al . _ [ mu2e collaboration ] , fermilab - proposal-0973 . t. aushev , w. bartel , a. bondar _ et al . _ , [ arxiv:1002.5012 [ hep - ex ] ] ; b. oleary _ et al . _ [ superb collaboration ] , [ arxiv:1008.1541 [ hep - ex ] ] . a. brignole and a. rossi , nucl . b * 701 * ( 2004 ) 3 [ arxiv : hep - ph/0404211 ] ; e. arganda , m. j. herrero and j. portoles , jhep * 0806 * ( 2008 ) 079 [ arxiv:0803.2039 [ hep - ph ] ] . t. feldmann , p. kroll and b. stech , phys . d * 58 * , 114006 ( 1998 ) [ arxiv : hep - ph/9802409 ] ; t. feldmann , p. kroll and b. stech , phys . b * 449 * , 339 ( 1999 ) [ arxiv : hep - ph/9812269 ] .

we outline a proof of factorization in exclusive processes , taking into account the presence of soft and collinear modes of arbitrarily low energy , which arise when the external lines of the process are taken on shell . specifically , we examine the process of annihilation through a virtual photon into two light mesons . in an intermediate step , we establish a factorized form that contains a soft function that is free of collinear divergences . in contrast , in soft - collinear effective theory , the low - energy collinear modes factor most straightforwardly into the soft function . we point out that the cancellation of the soft function , which relies on the color - singlet nature of the external hadrons , fails when the soft function contains low - energy collinear modes .

0903.0569

factorization theorems are fundamental to modern calculations in qcd of the amplitudes for hard - scattering exclusive hadronic processes . they allow one to separate contributions to the amplitudes that involve states of high virtuality from those that involve states of low virtuality . the former , short - distance contributions can , by virtue of asymptotic freedom , be calculated in perturbation theory , while the latter , long - distance contributions are parametrized in terms of inherently nonperturbative matrix elements of qcd operators in hadronic states . states of low virtuality can arise from the emission of a soft gluon , whose four - momentum components are all small , or from the emission of a collinear gluon , whose four - momentum is nearly parallel to the four - momentum of a gluon or light quark . in some discussions of factorization that employ soft - collinear effective theory ( scet ) @xcite or diagrammatic methods @xcite , it is assumed that gluons can have no transverse momentum components that are smaller than the qcd scale @xmath1 . that is , gluons can have hard momentum , in which all components are of order the hard - scattering scale @xmath2 , soft momentum , in which all components are of order @xmath3 , or collinear momentum , in which the transverse components are of order @xmath3 and the energy and longitudinal spatial component are much larger than @xmath1 ( usually taken to be of order @xmath2 ) . this assumption is appropriate to the discussion of physical hadrons , in which confinement provides a nonperturbative ir cutoff of order @xmath1 . however , in perturbative matching calculations of short - distance coefficients , one usually takes the external quark and gluon states to be on their mass shells , and , in this situation , soft and collinear gluons of arbitrarily low energy can be emitted . in order to establish the consistency of such calculations , one must prove , to all orders in perturbation theory , that these soft and collinear gluons factor from the hard - scattering process and that the factorized form is identical to the conventional one that is obtained in the presence of an infrared cutoff of order @xmath3 . or greater , then the short - distance coefficients are independent of all infrared modes , even in perturbative calculations in which modes with scales below @xmath3 are present . this was shown to be the case in a one - loop example . however , no general , all - orders proof of that assertion was given . ] in the absence of such a proof , one would have no guarantee in the matching calculation that low - virtuality soft and collinear contributions either cancel or can be absorbed entirely into the standard nonperturbative functions ( distribution functions in inclusive processes and distribution amplitudes in exclusive processes ) . at the one - loop level , gluons of arbitrarily low energy can be treated along with higher - energy soft and collinear gluons , and the conventional proofs of factorization apply . however , as we shall see , the proof of factorization of low - energy gluons becomes more complicated beyond one loop . in multiloop integrals , in the on - shell case , one finds contributions at leading power in the hard - scattering momentum in which collinear gluons of low energy couple to soft gluons . our goal is to construct a proof of factorization that takes this possibility into account . to our knowledge , the existing discussions of factorization , either in the context of scet or diagrammatic methods , have not addressed this possibility . in on - shell perturbative calculations in scet , gluon transverse momenta extend to zero . hence , the possibility of low - energy gluons with momenta collinear to one of the external particles arises . at one - loop level , the soft and collinear contributions can be separated through the use of an additional cutoff @xcite . however , as we have already mentioned , at two - loop level and higher , a low - energy collinear gluon can attach to a soft gluon . the scet action is formulated so that soft gluons can be decoupled from collinear gluons through a field redefinition , but there is no corresponding provision to decouple collinear gluons from soft gluons . therefore , it seems that , in scet , low - energy collinear gluons would be treated most straightforwardly as part of the soft ( or ultrasoft ) contribution . this results in a factorized form in which the soft function contains gluons with both soft and collinear momenta and , hence , contains both soft and collinear divergences . alternatively , one can consider a factorized form in which gluons with collinear momenta are factored completely from the soft function , so that they reside only in jet functions that are associated with the initial- or final - state hadrons . such an alternative factorized form , in which the soft function is free of collinear divergences , has been discussed in the context of factorization for the drell - yan process in refs . @xcite , although the details of the factorization of gluons with collinear momenta from the soft function were not given . this alternative factorized form has also been discussed in connection with resummation of logarithms in , for example , refs . furthermore , it has been discussed in an axial gauge in the context of on - shell quark scattering @xcite . axial gauges are somewhat problematic , in that they introduce unphysical singularities into gluon and ghost propagators . such singularities could potentially spoil contour - deformation arguments that are used to ascertain the leading regions of integration in feynman diagrams @xcite . for this reason , we believe that it is important to construct a proof of factorization in a covariant gauge , such as the feynman gauge , which we employ in the present paper . a factorized form in which the soft function contains no gluons with collinear momenta has several useful features . one is that contributions in which there are two logarithms per loop ( one collinear and one soft ) reside entirely in the jet functions , which have a diagonal color structure , rather than in the soft function , which has a more complicated color structure . here we focus on a feature that is crucial for factorization proofs : a factorized form in which the soft function contains no collinear modes allows one to establish a cancellation of the soft function when it connects to a color - singlet hadron . as we shall explain below , if the soft function contains gluons with collinear momenta , then the cancellation of the soft function fails at leading order in the large momentum scale . in this paper , we outline the proof of factorization at leading order in the hard - scattering momentum for the case of on - shell external partons . for concreteness , we discuss the example of the exclusive production of two light mesons in @xmath0 annihilation . in an intermediate step , the factorized form that we obtain contains a soft function that is free of collinear divergences . this allows us to demonstrate the cancellation of the soft function at leading order in the large momentum scale . our proof makes use of standard all - orders diagrammatic methods for proving factorization @xcite . we find that the factorization of gluons of arbitrarily low energy can be dealt with conveniently by focusing on the factorization of contributions to loop integrals from singular regions , _ i.e. _ , regions that contain the soft and collinear singularities . such singular regions are discussed in refs . @xcite . however , the coupling of low - energy collinear gluons to soft gluons is not discussed in those papers . the remainder of this paper is organized as follows . in sec . [ sec : model - amp ] we describe the model that we use for the production amplitude . section [ sec : leading - regions ] contains a heuristic discussion of the regions of loop momenta that give leading contributions . this discussion is aimed at making contact with previous work on factorization and also sets the stage for a more precise discussion of the singular regions of loop momenta . in sec . [ sec : topology ] , we discuss the diagrammatic topology of the leading contributions and also the topology of the soft and collinear singular contributions . we treat the collinear and soft contributions by making use of collinear and soft approximations that are valid in the singular regions . these are discussed in sec . [ sec : decoupling ] , along with the decoupling relations for the collinear and soft singular contributions . in sec . [ sec : factorization ] , we outline the factorization of the collinear and soft singularities and describe how one arrives at the standard factorized form for the production amplitude . we also outline the proof of factorization in the case in which the relative momentum between the quark and the antiquark in a meson is taken to be nonzero . here , we discuss the difficulty that arises in the cancellation of the soft function if the soft function contains gluons with collinear momenta . finally , in sec . [ sec : summary ] , we summarize our results . let us consider the exclusive production of two light mesons in @xmath0 annihilation through a single virtual photon . we work in the @xmath0 center - of - momentum frame and in the feynman gauge , and we write four - vectors in terms of light - cone components : @xmath4 , with @xmath5 . we take each meson to be moving in the plus ( minus ) direction and to consist of an on - shell quark with momentum @xmath6 ( @xmath7 ) and an on - shell antiquark with momentum @xmath8 ( @xmath9 ) : [ p1p2 ] @xmath10 , \\ p_{1\bar q}&=&\left [ \frac{(1-z_1)q}{\sqrt{2 } } , \frac{\bm{p}_{1\perp}^2}{\sqrt{2}(1-z_1)q } , -\bm{p}_{1\perp } \right ] , \\ p_{2q}&=&\left [ \frac{\bm{p}_{2\perp}^2}{\sqrt{2}z_2q } , \frac{z_2q}{\sqrt{2 } } , \bm{p}_{2\perp } \right ] , \\ p_{2\bar q}&=&\left [ \frac{\bm{p}_{2\perp}^2}{\sqrt{2}(1-z_2)q } , \frac{(1-z_2)q}{\sqrt{2 } } , -\bm{p}_{2\perp } \right],\end{aligned}\ ] ] where @xmath11 and @xmath12 does not lie near the endpoints of its range . the momentum @xmath13 of the meson @xmath14 is given by @xmath15 the large scale @xmath2 is equal to the invariant mass of the virtual photon , up to corrections of relative order @xmath16 . we assume that the components of @xmath17 are all of order @xmath3 . it is useful for subsequent discussions to introduce a dimensionless parameter @xmath18 in order to simplify the initial discussion , we set @xmath19 . we will discuss at the end of the factorization argument the effect of keeping the @xmath17 nonzero . let us now discuss the regions of loop momenta that are leading in powers of the large scale @xmath2 . our analysis will be somewhat heuristic , in that , as we will see , the boundaries between the various momentum types are indistinct . we carry out this analysis in order to make contact with previous discussions of factorization and to set the stage for our proof of factorization . that proof focuses on the soft and collinear singular regions of loop momenta , which _ are _ distinct . suppose that a virtual gluon with momentum @xmath20 attaches to external @xmath21 or @xmath22 lines with momentum @xmath23 and @xmath24 . ( in the remainder of this paper , we call lines that originate in an external @xmath21 or @xmath22 `` outgoing fermion lines . '' ) in the limit in which the components of @xmath20 are all small compared to the largest components of @xmath23 and @xmath24 , the amplitude associated with this process is proportional to @xmath25 because the integral is independent of the scale of @xmath20 , leading contributions arise from arbitrarily small momentum @xmath20 . one can emit an additional virtual gluon of momentum @xmath26 from an outgoing fermion line at a point to the interior of the emission of a gluon with momentum @xmath20 , provided that @xmath27 . such emissions are arranged in a hierarchy along the outgoing fermion lines , according to the virtualities that the emissions produce on the outgoing fermion lines . now let us establish some nomenclature to describe the regions of loop momenta that yield contributions that are leading in powers of the large scale @xmath2 . we call such momentum regions `` leading regions . '' we outline below the construction of an argument to prove that these are the only possible leading regions . we consider hard ( @xmath28 ) , soft ( @xmath29 ) , collinear - to - plus ( @xmath30 ) , and collinear - to - minus ( @xmath31 ) momenta , whose components have the following orders of magnitude : @xmath32 , \\ c^-&\hbox { : } & q\epsilon^-[(\eta^-)^2,1,\bm{\eta}^-_\perp].\end{aligned}\ ] ] we call a line in a feynman diagram that carries momentum of type @xmath33 an `` @xmath33 line . '' the parameters @xmath34 , @xmath35 , and @xmath36 set the energy scales of the momenta . we define the soft region of momentum space by the condition @xmath37 we define the collinear region of momentum space by the conditions @xmath38 in our definitions of momentum regions , the positions of the boundaries between regions are somewhat vague . that is because there is no clear distinction between the @xmath28 , @xmath29 , and @xmath39 regions near the boundaries between regions : when @xmath40 , an @xmath29 momentum is essentially an @xmath28 momentum ; when @xmath41 , a @xmath39 momentum is essentially an @xmath29 momentum . soft singularities occur in the limit @xmath42 , and @xmath39 singularities occur in the limits @xmath43 . hence , we see that , unlike the soft and collinear momentum regions , the soft and collinear singularities _ are _ distinct . there are also singularities that are associated with the scales of the collinear momenta . these appear in the limit @xmath44 . if @xmath45 is finite , these are essentially soft singularities , but they can occur in conjunction with a collinear singularity if @xmath43 . we do not consider gluon loop momenta of the `` glauber '' type @xcite , in which @xmath46 . the reason for this is that , for exclusive processes , the @xmath47 and @xmath48 contours of integration are not pinched in the glauber region , and , hence , one can always deform them out of that region @xcite . if we take @xmath49 to be of order one and @xmath34 and @xmath45 to be of order @xmath50 [ eq . ( [ def - lambda ] ) ] , then the resulting momenta are those that are treated in scet@xmath51 @xcite . soft momenta with @xmath34 of order @xmath52 have been considered in ref . @xcite in the context of two - loop - order contributions to @xmath53-meson decays , and the possibility of leading momentum regions involving momenta of arbitrarily small energy is mentioned in ref . @xcite for the case of massive particles . c .conditions under which a gluon with momentum @xmath20 can attach to a line with momentum @xmath54 . in each table , the left - hand column gives the momentum type of the gluon with momentum @xmath20 , and the top row gives the momentum type of the line with momentum @xmath54 . each entry gives the conditions that must be fulfilled if the attachment is to satisfy our conventions for attachments , as described in the text , and also yield a contribution that is not suppressed by powers of ratios of momentum components . for purposes of power counting , an @xmath28 line behaves as a soft line with @xmath40 . the rules for the attachment if @xmath20 is a @xmath55 momentum are the same as the rules of attachment if @xmath20 is a @xmath39 momentum . as is explained in the text , if @xmath20 is @xmath29 , and the lines to which it attaches have momentum @xmath23 and @xmath24 , then @xmath23 and @xmath24 can not both be @xmath30 or @xmath31 . furthermore , if @xmath20 is @xmath39 , then at least one of @xmath23 and @xmath24 is @xmath39 . [ tab : power - counting ] [ cols="^,^,^,^",options="header " , ] now let us describe the factorization of the contributions from the @xmath30 , @xmath31 , and @xmath29 singular regions . we can determine the momentum assignments that give singular contributions by making use of the power - counting rules that we have outlined in sec . [ sec : leading - regions ] . when we apply these rules to the attachments of gluons with momenta in the singular regions , the symbol @xmath56 and the phrase `` of the same order '' mean that quantities do not differ by an infinite factor , while the phrases `` much less than '' and `` much greater than '' mean that quantities _ do _ differ by an infinite factor . hence , for gluons with momenta in the singular regions , our convention that an allowed attachment of a gluon can not change the essential nature of the momentum of the line to which it attaches has the following meaning : the attaching gluon can not have an energy that is greater by an infinite factor than the energy of the line to which it attaches . the rules in sec . [ sec : leading - regions ] lead to complicated relationships between the allowed momenta of gluons in a given diagrammatic topology . however , there is a general principle , which we have already mentioned , that allows us to organize the discussion : the attachments of gluons to a given line must be ordered so that a given attachment produces a virtuality along the line that is of order or greater than the virtualities that are produced by the attachments that lie to the outside of it . in particular , the virtuality that a @xmath39 , @xmath55 , or @xmath29 singular gluon produces on a @xmath57 , @xmath58 , or @xmath29 line is of order its energy times the energy of the line to which it attaches . the relationships between allowed momenta lead to a hierarchy of scales as the singular limits are approached . consider , for example , the contribution in which an additional soft gluon is attached to the diagram of fig . [ fig:2loopex ] to the same outgoing fermion lines as the other gluons , but to the outside of them . in order for this contribution to be leading , the additional soft gluon must produce a virtuality on the outgoing fermion lines that is of order or less than the virtuality of @xmath59 or @xmath60 . the former condition implies that the energy scale of the additional soft gluon @xmath61 must be of order or less than @xmath62 . since @xmath63 , this implies that @xmath64 . that is , in the collinear singular limit , @xmath61 is infinitesimal with respect to @xmath34 . from such arguments it is clear that an infinite hierarchy of virtualities of various infinitesimal orders appears . however , these orders of virtuality are well separated in the singular limits . that is , the various gluon energy scales differ by infinite factors , as in our example . this property allows us to organize the singular contributions in such a way that we can apply the soft and @xmath39 approximations to obtain the factorized form . in order to carry out the factorization , we need to distinguish two cases for the ordering of the energy scale of a collinear momentum relative to the energy scale of a soft momentum . both of these orderings can yield contributions that are nonvanishing in the limits @xmath42 , @xmath65 . * case 1 : * as @xmath42 , @xmath66 is finite . ( it is easy to see that the contribution in which @xmath66 goes to zero vanishes . see for example , sec . [ sec : two - loop ] . ) in this case , we say that the collinear singular momentum and the soft singular momentum have energies that are of the same order . * case 2 : * as @xmath42 , @xmath67 . in this case we say that the soft singular momentum has energy that is infinitesimal in comparison with the energy of the collinear singular momentum . we will use an iterative procedure to factor gluons at the different levels of the hierarchy of energy scales . it is useful to establish first a general nomenclature to characterize this hierarchy of energy scales . we characterize each level in the hierarchy by the energy scale of the soft singular gluons in that level . we call that energy scale the `` nominal scale '' . we call soft singular and collinear singular gluons that have energies of order this scale nominal - scale gluons . we call collinear singular gluons that have energies that are infinitely larger than the nominal energy scale but infinitely smaller than the next - larger soft - gluon scale `` large - scale '' collinear gluons . the nominal - scale collinear gluons are of the type in case 1 above with respect to the nominal - scale soft gluons . the large - scale collinear gluons are of the type in case 2 above with respect to the nominal - scale soft gluons . let us now describe the factorization of the singular contributions . we make use of an iterative procedure in which gluons of higher energies are factored before gluons of lower energies . as we shall see , this ordering of the factorization procedure is convenient because it allows us to apply the decoupling relations rather straightforwardly to decouple gluons whose attachments lie toward the inside of the feynman diagrams before we decouple gluons whose attachments lie to the outside of the feynman diagrams . we will illustrate the factorization of the large - scale collinear gluons and the nominal - scale soft and collinear gluons for double light - meson production in @xmath0 annihilation by referring to the diagram that is shown in fig . [ fig : fact1 ] . in this diagram , we have suppressed gluons with energies that are much less than the nominal scale . these gluons have attachments that lie to the outside of the attachments of the gluons that are shown explicitly . in the diagram in fig . [ fig : fact1 ] , each gluon represents any finite number of gluons , including zero gluons . for clarity , we have suppressed the antiquark lines in each meson and we have shown explicitly only the attachments of the gluons to the quark line in each meson and only a particular ordering of those attachments . however , we take the diagram in fig . [ fig : fact1 ] to represent a sum of many diagrams , which includes all of the attachments that we specify in the arguments below of the singular gluons to the quark and antiquark in each meson , to other singular gluons , and to the @xmath68 subdiagram . @xmath69 denotes a large - scale @xmath70 singular gluon , @xmath71 denotes a nominal - scale @xmath70 singular gluon , and @xmath72 denotes a nominal - scale @xmath29 singular gluon . [ fig : fact1],height=264 ] we begin with the large - scale @xmath39 gluons that have the largest energy scale , and proceed iteratively through all of the scales of the large - scale @xmath39 gluons . in the first step of the iteration , those are gluons with finite - energy collinear singular momenta . in the subsequent steps , only gluons with infinitesimal collinear singular momenta are present . gluons with relatively infinitesimal energies may attach to a gluon that carries a @xmath39 singular momentum . we still consider that gluon to carry @xmath39 singular momentum . first , we wish to apply the @xmath30 approximation and the @xmath30 decoupling relation ( fig . [ fig : wi ] ) to decouple the large - scale @xmath30 gluons that originate in the @xmath73 subdiagram from the @xmath68 and @xmath74 subdiagrams . in applying the decoupling relation , we need to know the extent of the subdiagram in fig . [ fig : wi ] : eikonal lines appear at the points at which the subdiagram is truncated . we include the attachments of gluons with large - scale @xmath30 momenta to the @xmath74 subdiagram that are allowed by our conventions and by power counting . here , and in the discussions to follow , we consider a @xmath39 gluon to be attached to the @xmath75 subdiagram if and only if its momentum routes through @xmath68 . we include _ all _ of the attachments of gluons with large - scale @xmath30 momenta to @xmath68 . we include the attachments that are allowed by our conventions and by power counting . however , we also include formally attachments to @xmath68 that yield vanishing contributions in the singular limits . ( in subsequent iterations , we include formally , as well , the vanishing attachments of large - scale @xmath30 gluons to points on @xmath31 eikonal lines that lie to the interior of the outermost attachment of a @xmath31 singular gluon . ) in applying the @xmath30 decoupling relation , we do not include attachments of gluons with large - scale @xmath30 momentum to a gluon with nominal - scale @xmath29 momentum : such attachments violate our convention for allowed attachments because they alter the nature of the @xmath29 singular momentum . however , as we mentioned above , gluons with nominal - scale @xmath29 singular momenta can attach to a gluon with large - scale @xmath30 singular momentum without altering the nature of the @xmath30 singular momentum . we carry these attachments along as we attach the gluon with finite @xmath30 singular momentum to other lines in the diagram . we follow this same procedure in discussions below in treating gluons whose energies are infinitesimal with respect to the energy of an @xmath29 or a @xmath39 singular gluon to which they attach . the allowed attachments of gluons with large - scale @xmath30 momenta to a @xmath31 singular line lie to the inside of the attachments of gluons with nominal - scale @xmath29 or @xmath39 momenta . therefore , one might expect that , when the @xmath30 decoupling relation is applied , a @xmath30 eikonal - line contribution would appear at the vertex immediately to the outside of the outermost allowed attachment of a large - scale @xmath30 gluon . in fact , such an eikonal - line contribution vanishes because the propagator on the @xmath31 singular line just to the outside of the outermost allowed attachment of a gluon with large - scale @xmath30 singular momentum is on shell , and , in the case of a gluon line , has physical polarization ( polarization orthogonal to its momentum ) , up to relative corrections of infinitesimal size . therefore , we omit such eikonal - line contributions in applying the decoupling relation . then , the result of applying the decoupling relation is that the gluons with large - scale @xmath30 momenta attach to @xmath30 eikonal lines that attach to the outgoing fermion lines in @xmath73 just to the outside of the @xmath68 subdiagram . next we decouple the gluons that originate in the @xmath74 subdiagram and have large - scale @xmath31 momenta from the @xmath68 and @xmath76 subdiagrams . the procedure follows the same argument as for the gluons with large - scale @xmath30 singular momenta , except for one new ingredient : we must include formally the vanishing attachments of the gluons with large - scale @xmath31 momenta to the @xmath30 eikonal lines from the previous step . note that we need to include only the attachments that lie to the interior of the attachment of the outermost gluon with @xmath30 singular momentum in order to apply the @xmath31 decoupling relation . the result of applying the @xmath31 decoupling relation is that gluons with large - scale @xmath31 momenta attach to @xmath31 eikonal lines that attach to @xmath31 outgoing fermion lines just to the outside of the @xmath68 subdiagram . now , we iterate this procedure for the large - scale @xmath39 gluons at the next - lower energy scale . the result of applying the @xmath39 decoupling relations is that the large - scale @xmath39 gluons attach to @xmath39 eikonal lines that attach to the outgoing fermion lines just to the inside of the @xmath39 eikonal lines from the previous iteration . it is easy to see that , on each outgoing fermion line , the @xmath39 eikonal line from the current iteration can be combined with the @xmath39 eikonal line from the previous iteration into a single @xmath39 eikonal line . on the combined @xmath39 eikonal line , the attachments of @xmath39 gluons with the smaller energy scale lie to the outside of the attachments of gluons with the larger energy scale . ( other orderings yield vanishing contributions . ) we continue iteratively in this fashion until we have factored all of the large - scale @xmath39 gluons . after this decoupling step , the sum of diagrams represented by fig . [ fig : fact1 ] becomes a sum of diagrams represented by fig . [ fig : fact2 ] . . [ fig : fact2],height=264 ] next we factor the nominal - scale @xmath39 gluons . in applying the @xmath30 decoupling relation , we include the allowed attachments of these gluons to the @xmath74 subdiagram and the attachments to the nominal - scale soft gluons . we also include formally the vanishing contributions from the attachments of the nominal - scale gluons to the @xmath68 subdiagram and to the @xmath31 eikonal lines . because of the ordering of virtualities along a line with @xmath30 singular momentum , the outermost attachment to such a line of a gluon with nominal - scale @xmath30 momentum must lie to the outside of the outermost attachment of a gluon with nominal - scale @xmath29 momentum . it is then easy to see that , for every attachment described above of a @xmath30 line to a line with momentum that is not @xmath30 singular , the @xmath30 approximation holds exactly . the @xmath31 propagator that lies to the outside of the outermost allowed attachment of a gluon with nominal - scale @xmath30 momentum to a line with @xmath31 singular momentum is on - shell , and , in the case of a gluon line , has physical polarization , up to relative corrections of infinitesimal size . therefore , when we apply the @xmath30 decoupling relation , no eikonal line appears at the vertex immediately to the outside of this outermost attachment of a gluon with nominal - scale @xmath30 momenta . the result of applying the @xmath30 decoupling relation is that the nominal - scale @xmath30 gluons attach to several @xmath30 eikonal lines . these eikonal lines attach in the following locations : to the outgoing @xmath30 fermion lines just to the outside of @xmath68 , but to the inside of the large - scale @xmath30 eikonal lines ; just to the soft - gluon side of each vertex involving a nominal - scale soft gluon and a @xmath30 singular gluon of the large scale or a larger scale . in a similar fashion , we factor the nominal - scale @xmath31 gluons . the result of applying the @xmath31 decoupling relation is that the @xmath31 singular gluons attach to several @xmath31 eikonal lines . these eikonal lines attach to the following locations : to the outgoing @xmath31 fermion lines just to the outside of @xmath68 , but to the inside of the large - scale @xmath31 eikonal lines ; just to the soft - gluon side of each vertex involving a nominal - scale soft gluon and a @xmath31 singular gluon of the large scale or a larger scale . after this decoupling step , the sum of diagrams represented by fig . [ fig : fact2 ] becomes the sum of diagrams represented by fig . [ fig : fact3 ] . . [ fig : fact3],height=264 ] we now wish to apply the soft decoupling relations to factor the nominal - scale soft gluons . in order to do this , we implement the @xmath77 approximations for the attachments of the soft gluons to the @xmath39 singular lines of the large scale or a larger scale . however , we make a slight modification to the soft approximation by combining the momentum of the nominal - scale soft gluon with the total momentum of the associated nominal - scale @xmath39 eikonal line . then , when we implement the @xmath77 decoupling relations , the nominal - scale @xmath39 eikonal lines are carried along with the nominal - scale soft - gluon attachments . in applying the @xmath78 decoupling relation , we include attachments of nominal - scale soft gluons to the @xmath73 subdiagram , and in applying the @xmath79 decoupling relation , we include attachments of nominal - scale soft gluons to the @xmath74 subdiagram . because we have already factored the attachments of nominal - scale @xmath39 gluons , the @xmath77 approximations hold , up to relative corrections of infinitesimal size . we also include vanishing attachments of the nominal - scale soft gluons to the large - scale eikonal lines @xcite , including only those soft - gluon attachments that lie to the inside of the outermost @xmath30-gluon attachments . the @xmath39 propagator that lies to the outside of the outermost allowed attachment of a nominal - scale soft gluon to a @xmath39 line is on shell , up to relative corrections of infinitesimal size . therefore , when we apply the @xmath77 decoupling relations , no @xmath77 eikonal lines appear at the vertices just to the outside of the outermost allowed attachments . the result of applying the @xmath77 decoupling relations is that soft gluons attach to @xmath77 eikonal lines . these eikonal lines attach to the outgoing @xmath39 fermion lines just to the outside of the nominal - scale @xmath80 eikonal lines and just to the inside of the large - scale @xmath39 eikonal lines . associated with each attachment of a nominal - scale soft gluon to an @xmath77 eikonal line is a @xmath39 eikonal line to which nominal - scale @xmath39 gluons attach . after this decoupling step , the sum of diagrams represented by fig . [ fig : fact3 ] becomes a sum of diagrams represented by fig . [ fig : fact4 ] . . [ fig : fact4],height=264 ] we next factor the nominal - scale @xmath39 gluons from the @xmath77 eikonal lines . in order do this , we include formally the vanishing contributions that arise when one attaches the nominal - scale @xmath39 gluons to all points on the @xmath77 eikonal lines that lie to the inside of the outermost attachment of a nominal - scale soft gluon . we also make use of the following facts : each nominal - scale @xmath39 eikonal line that attaches to an outgoing @xmath39 fermion line is identical to the eikonal line that one would obtain by applying the @xmath39 decoupling relation to the attachments of the nominal - scale @xmath39 gluons to an on - shell fermion line ; each nominal - scale @xmath39 eikonal line that attaches to a nominal - scale gluon is identical to the eikonal line that one would obtain by applying the @xmath39 decoupling relation to the attachments of nominal - scale @xmath39 gluon to an on - shell gluon line . then , applying the @xmath30 decoupling relation , we find that the nominal - scale @xmath30 gluons attach to @xmath30 eikonal lines that attach to the outgoing fermion lines just to the inside of the large - scale @xmath30 eikonal lines . similarly , applying the @xmath31 decoupling relation , we find that the nominal - scale @xmath31 gluons attach to @xmath31 eikonal lines that attach to the outgoing fermion lines just to the inside of the large - scale @xmath31 eikonal lines . this result is represented by the diagram that is shown in fig . [ fig : fact5 ] . the nominal - scale @xmath39 eikonal lines can then be combined with the large - scale @xmath39 eikonal lines . after performing those steps , we arrive at the final factorized form , which is represented by the diagram in fig . [ fig : fact6 ] . . [ fig : fact5],height=264 ] , height=264 ] now we can iterate the procedure that we have given in secs . [ sec : large - c][sec : nom - c - f ] , taking the nominal scale to be the next - smaller soft - gluon scale . in these subsequent iterations , we include formally , in the steps of secs . [ sec : large - c ] and [ sec : nom - c ] , the vanishing contributions from the attachments of the large - scale and nominal - scale @xmath30 and @xmath31 gluons to the soft gluons of higher levels and to the @xmath78 and @xmath79 eikonal lines that are associated with those soft gluons . ( we also include formally the vanishing contributions from the attachments of the large - scale and nominal - scale @xmath30 and @xmath31 gluons to @xmath68 , as in the first iteration . ) proceeding iteratively through all of the soft - gluon scales , we produce new nominal - scale @xmath77 eikonal lines at each step that attach to the outgoing fermion lines just to the outside of the existing @xmath77 eikonal lines . the nominal - scale @xmath39 eikonal lines that attach to the outgoing @xmath39 fermion lines after the steps of sec . [ sec : nom - c ] are situated just to the inside of these nominal - scale @xmath77 eikonal lines . after the further factorization of the nominal - scale @xmath39 gluons that is described in sec . [ sec : nom - c - f ] , the @xmath77 eikonal lines that attach to a given outgoing fermion can be combined into a single @xmath77 eikonal line . the soft gluons of a lower energy scale attach to the outside of the soft gluons of a higher energy scale . this is the only ordering that produces a nonvanishing contribution . following this procedure , we arrive at the standard factorized form for the singular contributions . the @xmath81 subdiagram now attaches only to @xmath78 eikonal lines that attach to the outgoing fermion lines from @xmath73 just outside of @xmath68 and to @xmath79 eikonal lines that attach to the outgoing fermion lines from @xmath74 just outside of @xmath68 . the attachments involve only gluons with @xmath29 singular momenta . all of the @xmath39 singular contributions are contained in the @xmath82 subdiagram , which attaches via @xmath39 singular gluons to @xmath39 eikonal lines that attach to the outgoing fermion lines from @xmath83 just outside of the @xmath77 eikonal lines . this factorized form is illustrated in fig . [ fig : softfa ] . annihilation . after the use of the decoupling relations , gluons with momenta in the @xmath29 singular region attach to @xmath84 eikonal lines and gluons with momenta in the @xmath80 singular regions attach to @xmath80 eikonal lines.[fig : softfa],height=264 ] at this point the @xmath81 subdiagram and associated soft eikonal lines , which we call @xmath85 , have the form of the vacuum - expectation value of a time - ordered product of four eikonal lines : @xmath86[\infty^+,x_{1 q } ] \otimes [ x_{2\bar q},\infty^-][\infty^-,x_{2 q}]\ } \vert 0 \rangle_s,\nonumber\\ \label{s - tilde - me}\end{aligned}\ ] ] where @xmath87=\exp\left[\int_{x}^y igt_a a_\mu^a dx^\mu\right]\ ] ] is the exponentiated line integral ( eikonal line ) running between @xmath88 and @xmath89 , @xmath90 , and @xmath91 . the symbol @xmath92 indicates a direct product of the color factors that are associated with the soft - gluon attachments to meson 1 and the soft - gluon attachments to meson 2 . we note that eikonal - line self - energy subdiagrams , which were absent in our derivation of @xmath85 , vanish for lightlike eikonal lines in the feynman gauge . the subscript on the matrix element indicates that only contributions from the soft singular region are kept . because the @xmath28 and @xmath93 subdiagrams are insensitive to a momentum in the @xmath29 singular region flowing through them , we can ignore the difference between @xmath94 and @xmath95 in eq . ( [ s - tilde - me ] ) . then the @xmath78 eikonal lines cancel . note that this cancellation relies on the color - singlet nature of the external meson . in a similar fashion , we can ignore the difference between @xmath96 and @xmath97 in eq . ( [ s - tilde - me ] ) , and the @xmath79 quark and antiquark eikonal lines cancel . we can make a fierz rearrangement to decouple the color factors of the @xmath98 and @xmath99 subdiagrams from @xmath68 . then , we can write the @xmath100 subdiagrams and their associated eikonal lines , which we call @xmath101 and @xmath102 , as follows : @xmath103 \langle m_i(p_i)\vert \bar{\psi}_\alpha(x^\mp ) t\{[x^\mp,\infty^{\mp } ] [ \infty^\mp , -x^\mp]\}\psi_\beta(-x^\mp)\vert 0\rangle_{c^\pm}.\end{aligned}\ ] ] here , @xmath12 is the fraction of @xmath104 that is carried by the quark in meson @xmath105 , @xmath106 and @xmath107 are dirac indices , and the upper ( lower ) sign in eq . ( [ light - cone - dist2 ] ) corresponds to @xmath108 ( @xmath109 ) . it is understood that the fields @xmath110 and @xmath111 in the matrix element are in a color - singlet state . the subscripts on the matrix elements indicate that only the contributions from the collinear singular regions are kept . there is a partial cancellation of the eikonal lines in @xmath112 and @xmath113 , with the result that the residual eikonal lines run directly from @xmath114 to @xmath115 : @xmath116 \langle m_i(p_i)\vert \bar{\psi}_\alpha(x^\mp ) p[x^\mp ,- x^\mp ] \psi_\beta(-x^\mp)\vert 0\rangle_{c^\pm}.\end{aligned}\ ] ] here , we have written the time - ordered product of the exponentiated line integral as a path - ordered product . because the integrations over @xmath117 and @xmath118 have nonvanishing ranges of support in @xmath68 , @xmath115 and @xmath114 in eq . ( [ light - cone - dist3 ] ) are typically separated by a distance of order @xmath119 . this shows that the @xmath39 singular contributions that have energies much less than @xmath2 cancel , once they have been factored . we have shown that the contributions from @xmath39 singular regions factor from the @xmath75 , @xmath29 , and @xmath28 subdiagrams and are contained entirely in the @xmath120 subdiagrams and that the contributions from the @xmath29 singular region factor from the @xmath83 subdiagrams and cancel . the @xmath120 subdiagrams each have precisely the form of a meson distribution amplitude . hence , we have arrived at the conventional factorized form , except for the following facts : the @xmath120 subdiagrams contain only the infinitesimal @xmath39 singular regions , whereas they are conventionally defined to contain finite regions of integration ; the @xmath68 subdiagram is not yet free of nonperturbative contributions from collinear momenta with transverse components of order @xmath1 or less . cancel and the collinear poles can be factored into @xmath120 . ] next we extend the ranges of integration in the logarithmically ultraviolet divergent integrals in @xmath120 from infinitesimal neighborhoods of the collinear singularities to finite neighborhoods that are defined by an ultraviolet cutoff @xmath121 , which is the factorization scale . in making such an extension , we do not encounter any new singularities in @xmath120 . the soft singularities that do not involve the eikonal lines have already been shown to cancel . there is the possibility that @xmath29 or @xmath57 singularities could arise from the eikonal lines in @xmath120 . however , as we have mentioned , after the cancellation of the quark and antiquark eikonal lines , the remaining segment of eikonal line is finite in length , with length of order @xmath119 . hence , @xmath29 or @xmath57 modes with virtualities much less than @xmath2 can not propagate on these eikonal lines . finally , having extended the momentum ranges in @xmath120 , we redefine @xmath68 to be the factor that , when convolved with @xmath120 , produces the complete production amplitude . this is precisely the conventional definition of the hard subdiagram . since the soft divergences have canceled and the collinear divergences are contained in @xmath120 , @xmath68 is a finite function ( after ultraviolet renormalization ) , and depends only on the scales @xmath2 and @xmath122 and the renormalization scale . therefore , @xmath68 contains only contributions from momenta of order @xmath2 or @xmath122 , _ i.e. _ , from momenta in the perturbative regime . we have now established the conventional factorized form for the production amplitude @xmath123 , which reads @xmath124 where the symbol @xmath92 denotes a convolution over the longitudinal momentum fraction @xmath12 of the corresponding meson and we have suppressed the dirac indices on @xmath120 and @xmath68 . the @xmath125 are now given by eq . ( [ light - cone - dist3 ] ) , but without the subscript @xmath39 on the matrix element . they can be decomposed into a sum of products of dirac - matrix and kinematic factors and the standard light - cone distributions for the mesons . now let us return to the situation in which the @xmath17 in eq . ( [ p1p2 ] ) are nonzero and of order @xmath3 . in this case the outgoing quark and antiquark in each meson are moving in slightly different light - cone directions . therefore , we must define separate singular regions @xmath126 ( @xmath127 ) for the quark direction and @xmath128 ( @xmath129 ) for the antiquark direction in meson 1 ( 2 ) . as we have mentioned , in defining the collinear approximations , we can choose any auxiliary vectors @xmath130 that , for the collinear singular region associated with @xmath23 , satisfy the relation @xmath131 . we choose the lightlike auxiliary vector @xmath132 , which is in the minus direction , for both the @xmath126 and @xmath133 singular regions and the lightlike auxiliary vector @xmath134 , which is in the plus direction , for both the @xmath127 and @xmath135 singular regions . that is , we take the same collinear approximation for the collinear regions associated with the quark and the antiquark in a meson . then the factorization of the collinear singular regions goes through exactly as in the case @xmath136 . for gluons with momenta in the soft singular region , one can still define soft approximations , but the approximations are different for the couplings to lines in the quark and antiquark collinear singular regions . when gluons with momenta in the soft singular region attach to lines with momenta in the @xmath137 ( @xmath138 ) singular region , one can use a unit lightlike vector @xmath139 ( @xmath140 ) that is proportional to @xmath141 ( @xmath142 ) to define the soft approximation . then , the gluons with momenta in the soft singular region still factor . however , in the factored form , the soft eikonal line that attaches to the quark ( antiquark ) line in meson @xmath105 is parametrized by the auxiliary vector @xmath139 ( @xmath140 ) . because @xmath139 and @xmath140 differ by an amount of relative order @xmath50 [ eq . ( [ def - lambda ] ) ] , the quark and antiquark soft eikonal lines in each meson fail to cancel completely . these noncancelling soft contributions violate factorization because they couple one meson to the other in the production amplitude . if we take the approximation @xmath143 , but keep @xmath144 , then the quark and antiquark eikonal lines cancel in meson @xmath105 , but not in meson @xmath145 . however , the remaining soft subdiagram , which attaches only to the quark and antiquark eikonal lines in @xmath146 , can be absorbed into the definition of the @xmath146 subdiagram for meson @xmath145 . , that the configuration in which the soft subdiagram attaches only to the quark and antiquark soft eikonal lines that are associated with @xmath112 ( @xmath113 ) is precisely the configuration that one would obtain by using the soft decoupling relation to factor a soft subdiagram that attaches only to @xmath112 ( @xmath113 ) . here one must use the fact that the contributions in which @xmath30 ( @xmath31 ) collinear gluons with infinitesimal energy attach to the collinear eikonal line in @xmath112 ( @xmath113 ) vanish , owing to the finite length of the eikonal line . ] therefore , we see that we obtain a violation of factorization only if the quark and antiquark eikonal lines fail to cancel in both mesons . hence , the violations of factorization that arise from the soft function are of relative order @xmath52 . in order to express the amplitude in terms of the light - cone distributions @xmath147 in eq . ( [ light - cone - dist2 ] ) , it is necessary to neglect in @xmath68 the minus and transverse components of @xmath6 and @xmath8 and the plus and transverse components of @xmath7 and @xmath9 . in doing so , we make an error of relative order @xmath148 . now let us discuss the cancellation of the soft diagram for the factorized form in which the soft subdiagram contains collinear gluons . as we have mentioned , such a factorized form is the one that would seem to follow most straightforwardly from scet @xcite . suppose that a gluon with momentum @xmath20 attaches to the soft eikonal line that attaches to the quark line in meson 1 . that contribution contains a factor @xmath149 , which is singular in the limit in which @xmath20 becomes collinear to @xmath6 ( @xmath150 ) . on the other hand , for the contribution in which the gluon with momenta @xmath20 attaches to the soft eikonal line that attaches to the antiquark line in meson 1 , there is a factor @xmath151 , which is not singular in the limit in which @xmath20 becomes collinear to @xmath6 . hence , the attachments of the gluon with momentum @xmath20 to the quark and antiquark lines fail to cancel when @xmath20 is in the @xmath126 ( or @xmath128 ) singular region . furthermore , the uncanceled contribution is not suppressed by a power of @xmath2 and is , in fact , divergent . thus , we see that , in the factorized form in which the soft subdiagram contains collinear gluons , the soft subdiagram fails to cancel , and one can not establish the conventional factorized form . subdiagram , then soft approximations in scet and refs . @xcite and the soft decoupling relation involve the replacement of the soft momentum @xmath20 with a collinear momentum @xmath152 in the @xmath73 subdiagram . hence , @xmath153 vanishes when @xmath20 becomes @xmath30 . ] . it might seem that one could recover the cancellation of the soft subdiagram by setting @xmath154 exactly to zero . however , at @xmath136 , the cancellation of the quark and antiquark eikonal lines becomes ill - defined for @xmath20 collinear to the quark and antiquark because of the infinite factors that arise from the eikonal denominators @xmath155 and @xmath156 . for each meson , where @xmath157 lies between @xmath139 and @xmath158 . however , the resulting soft approximation fails to reproduce the collinear divergences that occur if the soft subdiagram contains low - energy gluons that are parallel to the quark or the antiquark . ] we note that this issue also arises in inclusive processes , for example the drell - yan process , in the decoupling of the soft subdiagram from color - singlet hadrons . we have established , to all orders in perturbation theory , factorization of the amplitude for the exclusive production of two light mesons in @xmath0 annihilation through a single virtual photon for the case in which the external mesons are represented by an on - shell quark and an on - shell antiquark . the case of on - shell external particles is important for perturbative matching calculations . the presence of on - shell external particles opens the possibility of soft and collinear momentum modes of arbitrarily low energy . in this situation , low - energy collinear gluons can couple to soft gluons . that coupling leads to additional complications in the factorization proof . nevertheless , we have shown that one can derive the standard factorized form , in which the production amplitude is written as a hard factor convolved with a distribution amplitude for each meson . the hard factor is free of soft and collinear divergences and depends only on the hard - scattering scale @xmath2 , the collinear factorization scale @xmath159 , and an ultraviolet renormalization scale . the meson distribution amplitudes contain all of the collinear divergences and all of the nonperturbative contributions that involve virtualities of order @xmath3 or less . we find that the factorization formula holds up to corrections of relative order @xmath160 . as an intermediate step in the factorization proof , we obtain a form in which the soft subdiagram does not contain gluons with momenta in the collinear singular regions . this form of factorization may be useful in the resummation of soft logarithms , as the contributions with two logarithms per loop are contained entirely in the jet functions , which are diagonal in color . it is essential in establishing the standard factorized form for exclusive processes with on - shell external partons because , as we have shown , the cancellation of the attachment of the soft diagram to a color - singlet hadron fails at leading order in @xmath2 if the soft subdiagram would contain gluons with momenta that are collinear to the constituents of the hadron . this issue also arises in inclusive processes in the decoupling of the soft subdiagram from color - singlet hadrons . in on - shell perturbative calculations in scet , low - energy gluons with momenta collinear to the external particles can appear . at two - loop level and higher , these low - energy collinear gluons can couple to soft gluons . since scet has no provision to decouple the collinear gluons from the soft gluons , it seems that it would be most straightforward in scet to treat the low - energy gluons as part of the soft contribution . in such an approach , the soft subdiagram contains gluons with momenta in both the soft and collinear singular regions . as we have said , the soft subdiagram would fail to cancel in this case , and one would not achieve the standard factorized form . therefore , in the absence of a further factorization argument , there would be no assurance in a matching calculation that the low - virtuality contributions could all be absorbed into the meson distribution amplitudes : some low - virtuality contributions might be associated with a soft function that could not be factored from the meson distribution amplitudes . alternatively , one could abandon the notion that scet should reproduce the contributions of full qcd on a diagram - by - diagram basis and assume that scet is valid only after one sums over all feynman diagrams . furthermore , one could consider the collinear action in scet to apply to all collinear momenta of arbitrarily low energy . then , as is asserted in ref . @xcite , the production amplitude in scet would take the form of a hard - scattering diagram , a @xmath112 light - cone distribution and a @xmath113 light - cone distribution that are convolved with the hard subdiagram , and a soft subdiagram that is free of collinear momenta and that connects to the @xmath112 and @xmath113 light - cone distributions with interactions that are given by the collinear action . that factorized form is the one that we would obtain after the decoupling of the collinear gluons from the soft gluons if we were to extend the ranges of integration in the @xmath81 , @xmath112 , and @xmath113 subdiagrams from the singular regions to finite regions of @xmath29 , @xmath30 and @xmath31 momenta . issues of double counting arise when one extends the ranges of integration . they could be dealt with , for example , by making use of the method of zero - bin subtractions @xcite . once the double - counting issues are resolved , our proof shows that such a form for the amplitude is correct . however , this result does not follow obviously from qcd or from scet . it requires a derivation , such as the one that we have given in this paper . the low - energy contributions that we have discussed involve integrands that are homogeneous in the integration momenta . therefore , one might argue that , if one applies the method of regions @xcite , then such contributions lead to scaleless integrals and vanish . the difficulty in making use of such an argument to prove factorization is that the method of regions extends the range of integration for each region to infinity . there is no proof of the validity of such an extension , and , hence , there is the possibility of double counting . double counting between the soft and collinear subdiagrams is dealt with in scet through the use of zero - bin subtractions @xcite . however , the zero - bin subtractions are formulated rigorously in terms of a hard cutoff . in ref . @xcite , examples of the zero - bin subtraction in dimensional regularization in one - loop perturbation theory are given . to our knowledge , no proof of an all - orders zero - bin subtraction scheme in dimensional regularization has been given . in physical hadrons , gluon momenta are cut off by confinement at a scale of order @xmath161 . in that situation , one does not need to consider the possibility that collinear gluons can attach to soft gluons in order to demonstrate the factorization of nonperturbative contributions , _ i.e. _ , those contributions that involve momentum components of order @xmath161 . however , if one wishes to factor logarithmic contributions up to a scale of order @xmath2 , for example , for the purpose of resummation , then it is again necessary to treat the attachments of collinear gluons to soft gluons along the lines that we have described in this paper . in this paper , we have focused on a specific exclusive process . however , we expect that our method can be generalized straightforwardly to the other exclusive processes and , possibly , to inclusive processes . in the latter case , one must consider glauber - type momenta explicitly , as contributions that arise from such momenta cancel only once one has summed over all possible final - state cuts @xcite . however , it seems plausible that one can implement this cancellation , using standard techniques , independently of the factorization arguments that we have presented here . we thank john collins and george sterman for many useful comments and suggestions . we also thank thomas becher , dave soper , and iain stewart for helpful discussions . we thank in - chol kim for his assistance in preparing the figures in this paper . the work of g.t.b . and x.g.t . was supported in part by the u.s . department of energy , division of high energy physics , under contract no . de - ac02 - 06ch11357 . the research of x.g.t . was also supported by science and engineering research canada . the work of j.l . was supported by the korea ministry of education , science , and technology through the national research foundation under contract no . 2010 - 0000144 . c. w. bauer , s. fleming , d. pirjol , and i. w. stewart , phys . d * 63 * , 114020 ( 2001 ) [ arxiv : hep - ph/0011336 ] . c. w. bauer , d. pirjol , and i. w. stewart , phys . d * 65 * , 054022 ( 2002 ) . [ arxiv : hep - ph/0109045 ] c. w. bauer _ et al . _ , phys . d * 66 * , 014017 ( 2002 ) . [ arxiv : hep - ph/0202088 ] g. t. bodwin , x. garcia i tormo , and j. lee , phys . * 101 * , 102002 ( 2008 ) [ arxiv:0805.3876 [ hep - ph ] ] . a. v. manohar , phys . b * 633 * , 729 ( 2006 ) [ arxiv : hep - ph/0512173 ] . a. v. manohar and i. w. stewart , phys . d * 76 * , 074002 ( 2007 ) [ arxiv : hep - ph/0605001 ] . g. t. bodwin , phys . d * 31 * , 2616 ( 1985 ) [ erratum - ibid . d * 34 * , 3932 ( 1986 ) ] . j. c. collins , d. e. soper , and g. sterman , nucl . b * 261 * , 104 ( 1985 ) . j. c. collins , d. e. soper , and g. sterman , adv . high energy phys . * 5 * , 1 ( 1988 ) . [ arxiv : hep - ph/0409313 ] h. contopanagos , e. laenen , and g. sterman , nucl . b * 484 * , 303 ( 1997 ) [ arxiv : hep - ph/9604313 ] . n. kidonakis , g. oderda , and g. sterman , nucl . b * 525 * , 299 ( 1998 ) [ arxiv : hep - ph/9801268 ] . n. kidonakis , g. oderda , and g. sterman , nucl . b * 531 * , 365 ( 1998 ) [ arxiv : hep - ph/9803241 ] . g. sterman and m. e. tejeda - yeomans , phys . b * 552 * , 48 ( 2003 ) [ arxiv : hep - ph/0210130 ] . a. sen , phys . d * 28 * , 860 ( 1983 ) . j. c. collins , d. e. soper , and g. sterman , nucl . b * 223 * , 381 ( 1983 ) . g. sterman , phys . d * 17 * , 2773 ( 1978 ) . s. b. libby and g. sterman , phys . d * 18 * , 4737 ( 1978 ) . g. t. bodwin , s. j. brodsky , and g. p. lepage , phys . lett . * 47 * , 1799 ( 1981 ) . c. w. bauer , d. pirjol , and i. w. stewart , phys . d * 67 * , 071502 ( 2003 ) . [ arxiv : hep - ph/0211069 ] . m. beneke , g. buchalla , m. neubert , and c. t. sachrajda , nucl . b * 591 * , 313 ( 2000 ) [ arxiv : hep - ph/0006124 ] . v. a. smirnov , phys . b * 465 * , 226 ( 1999 ) . [ arxiv : hep - ph/9907471 ] g. sterman , phys . d * 17 * , 2789 ( 1978 ) . g. grammer , jr . and d. r. yennie , phys . rev . d * 8 * , 4332 ( 1973 ) . j. c. collins and d. e. soper , nucl . b * 193 * , 381 ( 1981 ) [ erratum - ibid . b * 213 * , 545 ( 1983 ) ] . m. beneke and v. a. smirnov , nucl . b * 522 * , 321 ( 1998 ) [ arxiv : hep - ph/9711391 ] . j. c. collins , d. e. soper , and g. sterman , phys . b * 134 * , 263 ( 1984 ) . j. c. collins , d. e. soper , and g. sterman , nucl . b * 308 * , 833 ( 1988 ) .

regulating the stability of microtubule(mt)-kinetochore attachments is fundamental to avoiding mitotic errors and ensure proper chromosome segregation during cell division . while biochemical factors involved in this process have been identified , its mechanics still needs to be better understood . here we introduce and simulate a mechanical model of mt - kinetochore interactions in which the stability of the attachment is ruled by the geometrical conformations of curling mt - protofilaments entangled in kinetochore fibrils . the model allows us to reproduce with good accuracy in vitro experimental measurements of the detachment times of yeast kinetochores from mts under external pulling forces . numerical simulations suggest that geometrical features of mt - protofilaments may play an important role in the switch between stable and unstable attachments . _ key words : _ microtubule ; catch bond ; numerical simulation ; kinetochore ; mitosis . _ _ abbreviations : _ _ mt , microtubule ; pf , protofilament

1407.5484

during mitosis , the cell equally divides into two daughter cells , each getting a copy of the original genetic material . successful division requires that the two identical sister chromatids of mitotic chromosomes attach to the plus - ends of spindle microtubules ( mts ) via their kinetochores @xcite . this process is critical because incorrect attachments lead to mitotic errors giving rise to genetic instabilities , also involved in cancer @xcite . to ensure accurate chromosome segregation , correct mt - kinetochore attachments should remain stable while faulty attachments should be destabilized and corrected @xcite . mts are composed by a number of protofilaments ( pfs ) , typically thirteen , and polymerize by the addition of tubulin dimers in their gtp - bound state . growing mts switch to a shrinkage phase when all or most gtp - bound tubulin is hydrolysed , a process known as catastrophe , while the switch back to a growing state is known as resuce . these two processes constitute the _ dynamic instability _ of mts.@xcite . gtp - hydrolysis also induces a change in conformation of mt protofilaments from a straight to a curved state @xcite which eventually leads to depolymerization , since curved pfs tend to peel from the mt while straight filaments are stable . indeed electron micrographs of microtubules show that individual pfs can be seen curving outwards from the ends @xcite . mechanical measurements of the rigidity of mts show that their young s modulus is two orders of magnitude smaller than the shear modulus @xcite . this implies that tubulin dimers interact strongly along the pf and weakly along the transverse direction . all these structural , mechanical and kinetic aspects have been included in theoretical and computational models that describe with great accuracy the main features of the stability and dynamic instability of mts @xcite . mt - kinetochore attachments vary in different organisms , but all seem to share a common feature @xcite : fibrils extending from the kinetochore either link directly to curling mt - pfs , as in most higher eukaryotes @xcite , or fibrils linking to a ring , known as the @xmath01 complex @xcite , form around the attached mt , as in budding yeast @xcite . once the attachment has been formed , mt depolymerization @xcite provides a force that is strong enough to carry kinetochore - attached loads @xcite , even in the absence of motor proteins . the precise nature of the attachment between kinetochore fibrils and mts has been the object of intense experimental investigation suggesting that binding occurs through the ndc80 complex which interacts with tubulin by weak electrostatic forces @xcite . other experiments show that the ndc80 complex acts like a curvature sensor and binds preferentially to straight mt pfs , typical of polymerizing mts @xcite . on the other hand , electron micrographs illustrate vividly that kinetochore fibrils directly connect to the tips of _ curling _ pfs @xcite . in a recent paper , akiyoshi et al . @xcite have shown that yeast kinetochores form _ catch - bonds _ with mts . _ catch - bonds _ become stronger under a pulling force @xcite , thus providing a possible stabilizing mechanism needed for chromosome segregation . akiyoshi et al . also show that the kinethocore forms _ catch - bonds _ only with depolymerizing mts , whereas it forms standard force - weakening bonds with polymerizing mts @xcite . the authors encompass this information into a simple two - states kinetic model that is able to fit well the experimentally measured detachment times @xcite . the physical mechanism by which a _ catch - bond _ forms remains , however , still unclear . it is easy to understand that electrostatic interactions alone , even if protein complexes act cooperatively @xcite , would not give rise to a _ catch - bond _ : pulling charges apart leads to a weakening of the bond and should therefore increase the detachment rate . thus electrostatic interactions reasonably account for the experimentally observed increase of detachment rate for polymerizing mts under tension , but can not explain at the same time the decreasing detachment rate observed for depolymerizing mts . models of mt - kinetochore interactions are abundant ranging from the classical sleeve model @xcite to curling models @xcite and syntheses of both @xcite as well as incorporation of motor proteins @xcite , but none of them provide insight into the observed _ catch - bond _ behavior . in this paper , we resolve this puzzle by combining existing experimental evidence into a model of mt - kinetochore attachment that can explain the formation of _ catch - bonds_. we consider the interaction of a microtubule with a set of kinetochore fibrils with tips that can directly bind to straight mt - protofilaments and to neighbor fibril tips as suggested in ref . @xcite for the ndc80 complex . similarly , the fibrils can be effectively cross linked by other protein complexes as e.g. the @xmath1 complex in budding yeast or the mis12 complex in ptk@xmath2 cells @xcite . hence , when a polymerizing mt approaches the kinetochore , fibrils attach to its surface due their direct interactions . this attachment is a standard force - weakening bond as expected , but the mutual interactions or cross - linking between fibrils naturally leads to the formation of fibril loops . when mts depolymerize , the direct fibril - mt binding force is strongly suppressed @xcite , but the tips of curling pfs can easily entangle in the fibril loops . this attachment is now a _ catch - bond _ , since it becomes stronger under tension . the mechanism we propose is very general and could also involve other kinetochore proteins , such as cenp - e @xcite , cenp - t @xcite , cenp - f @xcite or the ska complex @xcite , rather than just ndc80 , which is too short to account for the fibrils alone . finally , the entangled organization of the observed in vertebrate kinetochores @xcite , provides another striking example where a velcro - like attachment , such as the one we propose , could naturally take place . we illustrate the formation of a conformational attachment by three dimensional simulations of a single depolymerizing mt interacting with a set of kinetochore fibrils . once we are confident that a conformational attachment is formed , we can reduce the computational complexity of the problem by focusing on a two - dimensional representation of the interaction between a pf and a fibril loop . numerical results of the two dimensional model reproduce with good accuracy the _ catch - bond _ behavior reported experimentally in ref . we also study the stability of the mt attachment and find that it crucially depends on the local conformation of the mt . by changing the intrinsic curvature of the mt - pf , the attachment is destabilized and the _ catch - bond _ disappears . our results suggest that the experimentally observed tension - induced stabilization of mt - kinetochore attachments could be explained by a conformational mechanism although chemical affinities between mts and kinetochore proteins may also play a role . and @xmath3 , corresponding to a gtp bound mt . the sheet will fold with an helicity depending on the number of transversal units and the ratio @xmath4 . d ) clamping one end of the mt while hydrolysing the blocks by letting @xmath5 and allowing transversal bonds to break leads to the ram s horn shape typical of depolymerizing mts . , scaledwidth=70.0% ] we construct a three - dimensional model of a mt starting from a set of wedge - shaped building blocks representing tubulin dimers , as originally suggested in ref . a model similar in spirit to ours has been used in refs . @xcite to simulate mt self - assembly . in our model , each block is composed by eight nodes connected by stiff elastic springs , ensuring that the block behaves as a rigid object , as illustrated in fig . [ fig : block_model]a . diagonal springs prevent shearing and twisting of the block , while springs between nearest neighbouring nodes heavily penalize elongation and compression . the rest - length and stiffness of the springs are given in table [ tab:3dconstants ] . by suitably tuning the geometrical features of the tubulin block , we can induce the desired conformation of the pfs and of the mt . pfs are formed by arranging the blocks linearly so that nodes @xmath61,2,5,6 face nodes @xmath63,4,7,8 and connecting the nodes by elastic springs with rest length @xmath7 and stiffness @xmath8 , as shown in fig . [ fig : block_model]b . when the rest length of the springs on the top face ( i.e @xmath9 ) is equal to the rest length of the springs on the bottom face ( @xmath10 ) the pf is straight ( top of fig . [ fig : block_model]b ) , simulating the gtp - bound state . by changing @xmath10 so that @xmath11 we can induce an intrinsic curvature in the pf , simulating hydrolysis into the gdp - bound state ( as shown in fig . [ fig : block_model]b ) . the mt is generated by laterally aligning @xmath12 pfs and connecting the blocks via springs , with rest length @xmath13 and stiffness @xmath14 , to form a sheet as shown in fig . [ fig : block_model]c ) . a tubular structure is obtained by imposing that the top and bottom faces of the blocks have a trapezoidal shape ( as shown fig . [ fig : block_model]a ) . in this condition , the pf sheet naturally folds into a cylinder , as shown in fig . [ fig : block_model]c . we tune the geometrical parameters of the wedge so that a tubulin sheet consisting of @xmath15 pfs will close into itself ( c.f . table [ tab:3dconstants]a ) . a smaller number of pfs will not yield mt closure , while a larger number of pfs produces an overlapping structure . the helicity of the mt lattice is achieved by shifting the alignment of the blocks by @xmath16 blocks at every turn . while the most common case is @xmath17 and @xmath18 @xcite , here we consider @xmath17 and @xmath19 . when the mt closes up we clamp one end of the mt and hydrolyse or add tubulin units at the other end . the flexural rigidity @xmath20 of individual pfs in the model can be estimated from the theory of elasticity as the product of the area moment of inertia @xmath21 and the elastic modulus @xmath22 : @xmath23 . the area moment of inertia of a rod with trapezoidal cross section of height @xmath24 nm and top and base lengths @xmath25 nm and @xmath26 nm ( values correspond to those in ref . @xcite and ref . @xcite when we consider that the total width and length of a tubulin unit is enhanced by the connecting springs , @xmath7 and @xmath13 , respectively ) is @xmath27 nm@xmath28 , for bending around the median line . the elastic modulus is directly related to the stiffness of the springs by @xmath29 , where the factor 4 comes from the fact that we split the tension between two blocks into four parallel springs , and @xmath30 is the cross section of the block . we chose @xmath8 so that @xmath20 is in the range of @xmath31nm@xmath32 which is in agreement with earlier estimates based on measurements of the bending stiffness of mts @xcite . other estimates of the linear angular - spring stiffness of pfs yields slightly different values @xcite . finally , the particles composing the block are endowed with a hard - core repulsion potential ( for numerical purposes the repulsive part of a lennard - jones potential ) with cutoff @xmath33 nm to avoid inter - penetration of blocks . longitudinal and transversal bond - lengths ( @xmath7 and @xmath13 , respectively ) are chosen smaller than the hard - core cutoff to avoid bond - deformation other than stretching . by allowing for bonds breaking and attachment , we can simulate polymerization and depolymerization processes . the model is implemented and simulated using the lammps software package @xcite . we model the kinetochore - mt interface as an assembly of fibrils @xcite which represent various canidate proteins and complexes ( e.g. ndc80 , cenp - e , cenp - t , cenp - f . ) linked e.g. by the mis12 complex @xcite , the ska complex @xcite or the @xmath1 complex @xcite . here the different proteins at the kinetochore - microtubule interface are treated on equal footing ` coarse grained ' into fibrils and are modelled as bead - spring polymers of length @xmath34 nm @xcite . we consider hard - core repelling beads of radius @xmath35 nm connected by linear springs of rest - length @xmath36 and stiffness @xmath37pn / nm , which makes them practically inextensible . we also assume that fibrils have vanishing bending stiffness compared to the mts . kinetochore fibril bases are arranged randomly on a wide area with a density of @xmath38 fibrils / nm@xmath39 and then extended towards the incident mt - tip . in practice these densities correspond to 10 - 50 fibrils per mt . the density was chosen based on the estimate of twenty ndc80 complexes per captured mt @xcite . a summary of fibril parameters is reported in table [ tab:3dconstants]b . fibril tips can either bind to tubulin blocks in a straight conformation or to other fibril tips . in our model , curling tubulin blocks can not directly bind to fibrils , in agreement with experiments showing that the ndc80 tip binds preferentially with straight tubulin conformations @xcite . this choice is made to illustrate the worst case scenario , but it is expected that a weak attraction with curved tubulin filaments , possibly relevant for other protein complexes ( e.g. cenp - e , cenp - t , ska , cenp - f ) , would not change the results . hence , an incident mt can attach to kinetochore fibrils either directly by binding to straight tubulin units or by locking curling pfs into fibril loops as shown in fig . [ fig : kmt_interface]b ) . in all cases , binding is implemented by a lennard - jones potential @xmath40 \quad r < r^ * \label{eq : lj}\ ] ] where @xmath41 is a cutoff , @xmath42 is the binding strength and @xmath43 controls the width of the potential well . notice that fibrils in our model represent coarse - grained protein complexes so that their mutual interactions may reflect collective properties . interaction with gdp - bound tubulin is modelled as a repulsive potential by setting @xmath44 . the parameters chosen for @xmath43 and @xmath41 are chosen so that inter - penetration between fibrils and blocks are avoided and are reported in table [ tab:3dconstants]c . movie s1 illustrates the mechanism described here . it shows the result of a simulation where a straight mt enters a bundle of kinetochore filaments and binds some on its outer surface . when the mt protofilaments begin to curl and peel off the mt , some become entangled in loops formed by the kinetochore fibrils . [ cols=">,^,^,^,<",options="header " , ] the three - dimensional model of the mt and the mt - kinetochore interface is highly complex and difficult to simulate for long times due to the stiffness of the bonds . therefore only limited results can be obtained by the full 3d model . to obtain relevant statistics for the detachment kinetics , we need to perform some important approximation reducing the problem to two - dimensions . [ fig : kmt_interface]d suggests a way how such a dimensional reduction might take place . the protofilament is modelled as a flexible chain of nodes connected via springs @xcite with longitudinal stiffness @xmath45pn / nm and rest length @xmath46 nm ( approximately one tubulin dimer ) . since the fibrils form loops around the pf tip , the only way to detach save the dissociation of the fibril loop is by sliding the loop off the pf ram s horn . in two dimensions , this process is illustrated in fig . [ fig : kmt_interface]d ) where the cross - section of the fibrils are modelled as an elastic ball . here , we model the kinetics of single protofilament attached to a substrate representing the interaction with neighboring protofilaments @xcite . the chain furthermore has an intrinsic bending angle of @xmath47 @xcite and bending stiffness @xmath20 . hence , the elastic energy of a protofilament with node coordinates @xmath48 is given by @xcite @xmath49 where @xmath50 is the heaviside step - function and @xmath51 are the angles between neighbouring subunits . hence , the elastic force on each node @xmath52 is simply @xmath53 . as in the 3d model , we estimate the bending stiffness of the pf from the flexural rigidity of a microtubule @xmath54nm@xmath39 ( see @xcite and references therein ) and its second moment of inertia @xmath55m@xmath28 @xcite . although the protofilament is modelled as a one - dimensional object , we estimate its model of inertia treating it as a cylinder of radius @xmath56 nm , which yields @xmath57m@xmath58 . by simple calculation we can then estimate @xmath59nm@xmath39 . the interaction between the filament and the substrate is modelled by a spring with stiffness @xmath60pn / nm which breaks when it is extended beyond @xmath61 nm . these parameters ( @xmath16 and @xmath62 ) were fitted to the requirement that depolymerizing mts shrink at the experimentally observed speed , keeping a finite curvature at the tip to capture the fibrils . in fact , as shown in ref . @xcite , only the combination @xmath63 rules the pf dynamics . as shown in fig . [ fig : kmt_interface]d we consider only the cross section of the fibril bundle ( fig . [ fig : kmt_interface]c ) and model it a closed chain of six nodes connected by linear and angular springs with extension @xmath64 , stretching stiffness @xmath65 and bending stiffness @xmath66 . the elastic energy of the chain can then be written in a form equivalent to eq . [ eq : elasticity ] . the interaction between the pf and the fibril is modelled by a repulsive potential in case of a depolymerizing mt and by an attractive force for a polymerizing mt : the repulsive force between for each node of the depolymerizing pf @xmath48 and the fibril @xmath67 according to @xmath68 where @xmath69 and @xmath70 . the attractive interaction with polymerizing pf is modeled according to eq . [ eq : lj ] . the dynamics of the pf - fibril system is governed by the following coupled langevin equations @xmath71 where @xmath72 and @xmath73 are the damping coefficients of the nodes of the pf and the fibril , respectively , and @xmath74 is an externally applied pulling force , acting along the horizontal direction . the thermal noise @xmath75 acting on the fibril is a gaussian random force with average @xmath76 and correlations @xmath77 , where @xmath78 is the temperature and @xmath79 is boltzmann constant . in the following we quantify the amplitude of thermal fluctuations by the parameter @xmath80 . the damping coefficient can be evaluated considering that the drag coefficient of a sphere of radius @xmath81 immersed in a fluid with dynamical viscosity @xmath82 is given by the stokes formula @xmath83 . the dynamical viscosity of water at @xmath84 300k is @xmath85 pa s. assuming that the radii of the fibril and pf are @xmath86 nm and @xmath87 nm , we get @xmath88 kg / s and @xmath89 kg / s . in these conditions , the thermal fluctuations of the pf are overshadowed by the interactions with the strongly fluctuating fibrils and can be safely dropped . this is particularly convenient because it further speeds up the simulations . the mt shrinks or grows by adding or removing tubulin subunits from the tip , but in mammalian cells , mts also depolymerize at the spindle poles which is associated with poleward flux of tubulin @xcite . here , we consider that the pf is able to polymerize ( attach subunits at the tip ) or depolymerize ( lose subunits from the tip ) and switch between these states . the growing / shrinking velocities and switching rates we use are taken from experiments in refs @xcite . we assume that the velocities are dependent on the force on the tip of the protofilament @xcite according to @xmath90 eqs . [ eq : oda ] and [ eq : odb ] are solved by a fourth - order runge - kutta algorithm for the pf and an euler - maruyama algorithm for the fibril . due to numerical constraints , it is not possible to simulate the model over a timescale that is comparable to the experimental one . we thus artificially increase the noise fluctuations and then analyze how the results depend on @xmath91 . in this way , we are able to show that the numerical results approach the experimental ones as @xmath91 tends towards realistic values . the numerical values of the constants used in the simulations are summarized in table [ tab : constants ] + subunit length & @xmath92 & @xmath93 m & length scale + drag coefficient & @xmath72 & @xmath94 kg / s & estimate + bending stiffness & @xmath20 & @xmath95 nm@xmath39 & calculation + stretching stiffness & @xmath96 & @xmath97 n / m & assumption + substrate stiffness & @xmath16 & @xmath98 n / m & fit + breaking length & @xmath62 & 1 nm & fit + bending angle & @xmath47 & 0.18 -0.42 & @xcite + growth velocity & @xmath99 & 0.008- 0.21 @xmath100 m/s & @xcite + shrinking velocity & @xmath101 & 0.2 - 0.3 @xmath100 m/s & @xcite + growth acceleration & @xmath102 & 20pn & @xcite + shrinking stall & @xmath103 & 7pn & @xcite + rescue rate & @xmath104 & 0.045/s & @xcite + catastrophe rate & @xmath105 & 0.058/s & @xcite + + segment length & @xmath64 & @xmath94 m & assumption + drag coefficient & @xmath106 & @xmath107 kg / s & estimate + flexural rigidity & @xmath66 & @xmath108 nm@xmath39 & assumption + stretching stiffness & @xmath96 & @xmath109 n / m & assumption + fluctuation amplitude & @xmath91 & 15 - 35 nm @xmath100s@xmath110 & tried values + we simulate the three dimensional kinetochore - mt interface model presented and discussed in the methods section . in fig . [ fig : kmt_interface]a ) we show a mt whose curling pfs are partly linked into loops formed by kinetochore fibrils . in fig . [ fig : kmt_interface]b ) and movie s2 we illustrate the process by which a curved pf locks into a loop formed by kinetochore fibrils . a polymerizing pf , composed by straight tubulin blocks , reaches into the entangled kinetochore outer plate and fibrils attach to the tubulin blocks ( see fig . [ fig : kmt_interface]b ) left ) . since fibril tips are now close together , they can easily bind with each other ( as suggested in ref . * figure 6c ) and ref . when the pf depolymerizes and starts to curl , fibrils detach from the tubulin due to the reduced affinity , but the ram s horn of the pf can lock mechanically into the loop formed by the connected fibrils ( see fig . [ fig : kmt_interface]b ) right ) . to test the fidelity of out kinetochore - microtubule interface model , we perform a series of simulations to see how many pfs are entangled in the fibril loops . the mt - tip was kept straight and moved into the mass of kinetochore fibrils with a velocity of 1@xmath100m / min @xcite . fibrils attach to the mt surface and bind to each other , especially when they are in proximity of the mt , thus forming loops . as soon as the tip of the longest pf reaches the area of the fibril bases , the incident velocity is set to zero , and the mt is set to depolymerize immediately . in this process curling pfs can hook into the fibril loops . we perform this simulation for various fibril densities and average over 30 realizations in each case . we report the number of entangled pfs as a histogram in fig . [ fig : histogram ] for 10 , 20 and 50 fibrils per mt , or 0.77 , 1.5 and 3.8 fibrils per pf . the average number of entangled pfs is 1.1 , 4.5 and 5.2 for 10,20 and 50 fibrils , respectively . hence even for a relatively small number of fibrils , there is significant and robust entanglement . the geometrical conformations of the pf tips broadly correspond to the two distinct states of the dynamic mt , a growing / polymerizing ( straight ) and a shrinking / depolymerizing ( curved ) state , with differing chemistry and physics of the molecules at the tip . in the growing state , the tip of the mt recruits phosphorylysed gtp - tubulin subunits , while the tubulin subunits in the rest of the mt body are progressively hydrolysed . the main difference for our purposes between gtp and gdp - tubulin is the equilibrium angle between two connected subunits . while a filament made out of gtp units tends to remain straight , a gdp - tubulin filament has a curved equilibrium configuration ( see fig . [ fig : block_model]b ) . as a consequence , the tips of pfs in a growing mt will tend to have rather straight configurations while in the depolymerizing phase , the pfs of the mt form hooks - ram s horns - while losing tubulin subunits . to estimate the load carrying capacity of the mt - kinetochore interface , we consider only the tip of the pfs and assume that the constituting tubulin blocks are uniformly either hydrolysed ( gdp - bound ) or phosphorylated ( gtp - bound ) . in this way , we can discuss detachment from depolymerizing or polymerizing pfs separately and distinguish the underlying detachment mechanisms . in particular , a single pf is first inserted , up to a few blocks , typically ranging from one to three , into a network of kinetochore fibrils and the system is then relaxed for 10s . since we always start from a gtp - tubulin pf , the tips of some fibrils bind to the outer " side of the tubulin blocks ( particles with @xmath61 - 4 , c.f . [ fig : block_model]a ) . when the system is equilibrated , the distal end of the pf is held fixed , while the kinetochore plate , to which fibrils are attached , is moved with constant velocity @xmath111nm / s . in these conditions , we measure the longitudinal component ( parallel to the pulling direction ) of the interaction force between kinetochore and the pf . ( f - p ) . displacing the fibril ends leads at first to no increase of the force as long as the fibril unwraps . upon completion of this straightening process , the force increases rapidly as the fibrils and the pf stretch while overcoming lennard - jones attraction . b ) force - displacement curve of a gdp - tubulin pf in the curved conformation . again , the kinetochore fibrils are pulled with constant velocity and the in - line component of the interaction force is plotted for flexural rigidity @xmath112 pnnm@xmath39 and @xmath113 pnnm@xmath39 . the interaction energy between the fibrils is modelled by a lennard - jones potential with @xmath42(f - f)@xmath114pnnm or @xmath42(f - f)@xmath115pnnm . in the first case the peak corresponds to the bending of the pf while in the second case to the breakdown of the fibril loop . the oscillation of the black curve is due to the discrete nature of the pf and arises from the sliding of the kinetochore fibrils from one block to the next . this oscillation is drowned out by noise when the flexural rigidity is reduced as shown in the red curve.,scaledwidth=80.0% ] for gtp - tubulin pfs , we plot the force - displacement curves in fig . [ fig : x]a for three values of the binding strength @xmath42(p - f ) between fibrils and pfs . the fibril tips are in this case attached only to the first tubulin block of the pf in a way shown in the inset . the curves show the displacement @xmath116 , the difference between the starting and end positions of the kinetochore plate as a function of the in - line component of the interaction force , normalized to the number of attached fibrils . simulations are repeated 30 times to average out the effect of thermal noise . the results show that at short displacements the kinetochore plate feels no restoring force since the fibrils are unwrapped " , i.e. straightened . at this point , attached fibrils finally exert a force on the plate due to the binding with the tubulin dimer , until the bond breaks . the detachment from a gtp - tubulin pf , therefore relies _ solely _ on the attractive interaction between kinetochore fibrils and tubulin blocks . the process is illustrated in movie s3 . when the pf is formed of gdp - tubulin , its end is curved and the fibril - tubulin interaction is turned into a purely repulsive potential . we perform the same simulation as detailed for the gtp - tubulin pf above . the difference is that the pf is inserted more than one tubulin - block into the fibril network , typically two to four blocks , and the relaxation also entails the curling of the pf . we performed simulations for two values of the flexural rigidity and plot the results in fig . [ fig : x]b ( see also movie s4 ) . for the larger value of the stiffness we observe pronounced oscillations of the force - displacement curve due to the sliding of the kinetochore fibrils on the tubulin blocks while uncurling . for the less stiff case , the oscillations are not clearly discernible and the detachment of the kinetochore - mt interface is much smoother . as shown in fig . [ fig : x]b , the peak load of a single depolymerizing pf is larger than 10 pn , suggesting that a mt can on average carry a peak load of up to 50pn , which compares nicely with experimental results @xcite . we also test the effect of the binding interactions between fibrils @xmath42(f - f ) . for @xmath42(f - f)=100 pnnm and @xmath112 pnnm@xmath39 , the fibril loop eventually breaks ( see movie s5 ) but it is still able to carry a load of @xmath117 . hence for @xmath42(f - f)=100 pnnm and @xmath118 pnnm@xmath39 , the fibril loop should be strong enough to sustain a load of 20 pn , as we verified numerically . a feature of the detachment that is not captured in the md simulations is thermally induced / assisted detachment prevalent at low forces . the reason is that thermal phenomena in a system as ours appear on time scales much larger than we are able to simulate in three dimensions . to be able to simulate on statistically relevant time scales we developed and implemented a two - dimensional equivalent of the kinetochore - mt interface as already discussed in a previous section . for a shrinking pf , we assume that the fibrils do not attract to the tubulin molecules that are in curved conformation @xcite but interact via a solely repulsive potential , as discussed in the methods section . to relate the experiments of ref . @xcite with the two - dimensional model , we chose an intrinsic bending angle of @xmath119 , corresponding approximately to the peak of the curvature distribution measured experimentally in @xcite . all the model parameters are reported in table [ tab : constants ] . for the 2d model of a depolymerizing pf together with the experimental data from ref . b ) extrapolation of the diffusion parameter @xmath91 in the experiments from the numerical data . c ) attachment lifetimes ( inverse detachment rates ) of depolymerising pfs to kinetochores at high forces . the numerical data suggests that the lifetime peaks due to deformations and then decreases sharply with applied load . the low - force parts and the peaks can be fitted well with the weibull distribution which yields also a reasonable fit for the experimental data , as shown in panel d ) . , scaledwidth=97.0% ] as described in the previous section , detachments can occur either because of thermal and other non - equilibrium fluctuations that lead the fibril to jump off the pf tip , or because a sufficiently strong pulling force mechanically bends the pf . in agreement with ref . @xcite , low pulling forces increasingly stabilize the pf - fibril attachment . furthermore , our model predicts that for larger forces the attachment should again be destabilized . to compare the model with experiments , we plot in fig . [ fig:2]a the numerical and experimental data for the detachment rate , denoted @xmath120 as in ref.@xcite . the numerical data is plotted for various values of the fluctuation parameter @xmath91 . at small forces the detachment rates is well fitted by exponential decay , @xmath121 . the parameter @xmath122 follows an exponential curve as a function of @xmath91 , as shown in fig . [ fig:2]b . from this plot we can extrapolate the value of @xmath91 in the experiment for our model and arrive at @xmath123 which is a realistic and physical fluctuation parameter for this system . hence , at low forces simulations and experiments fit neatly together . it is , however , unlikely that bond strengthening persists up to larger forces , where elastic deformation of the pf should take place . indeed , simulations for large forces show that the lifetime of the kinetochore - microtubule attachment the inverse of @xmath120 displays a peak at @xmath124 10 pn , and decreases very quickly after that , as shown in fig . [ fig:2]c . the detachment times can be very well fitted at low forces and around the peak with a weibull function @xmath125 where @xmath126 the lifetime at zero force , the amplitude @xmath127 , and the weibull parameters @xmath128 and @xmath96 all depend on a diffusion parameter @xmath91 . the weibull function provides also an alternate fit for the lifetime of attachments with depolymerising pfs in the experimental data as shown in fig . [ fig:2]d , together with the original exponential fit . both fits yield consistent results but it is difficult to decide which one is better . in both cases , the reduced @xmath129 resulting from least - square minimization is of the order of @xmath130 which is a clear indication of over - fitting . our model suggests that measuring the lifetime of depolymerizing mt attachments to kinetochores should result in a peak at larger forces . as mentioned before , detachments at large force are due to a different physical process than the detachments at low forces . in ref . @xcite the destabilization of the kinetochore - mt attachment was attributed to a switch of the mt from the shrinking to the growing state . while we do not discount that possibility , we suggest the existence of an alternate pathway which could be dominant at short time scales : when the pulling force is large enough , the depolymerizing pf is uncurled ( c.f . movie s2 ) . together with the experimental data from ref . @xcite . for high enough binding strengh / fluctuation ratio , the parameter @xmath131 , appearing as the slope here , stays constant and is the same for the numerical and experimental data . b ) extrapolation of the experimentally observed binding strength . , scaledwidth=80.0% ] in the two dimensional model , kinetochore fibrils are pulled with a constant force , while the pf is growing or shrinking . the time - frame for switching between the two states is much larger than the detachment times we are able to simulate and therefore the effects of rescue and catastrophe do not affect our numerical results . to further simplify the numerical simulation and thus results for longer times , we model the detachment from polymerising pfs as a simple escape from a lennard - jones potential with a cutoff and simulate eq . [ eq : odb ] we have tested in some specific cases that the results are indistinguishable from those obtained with the complete model . the result for various values is plotted in fig . [ fig : k4]a reports the value of the attachment lifetime denoted by @xmath132 as in ref . obtaiend in simulations as a function of the binding energy - fluctuation ratio @xmath133 together with the experimental data . numerical simulations and experimental data are well fitted with exponential curves and have the same slope on a logarithmic scale . in particular , we fit the rate as @xmath134 where only the detachment rate at zero force @xmath135 depends on the ratio @xmath133 , while @xmath136 3.8 pn is the same for experimental and numerical data . plotting @xmath137 versus @xmath133 for the numerical result , allows us to extrapolate the value for @xmath133 for the experiment to @xmath138 , as shown in fig . [ fig : k4]b . since we previously estimated @xmath139 for the experiment , we predict the attachment strength of the fibrils to the kinetochore to be @xmath140 . this is of the order of magnitude for usual protein - protein interactions , providing an additional confirmation of the validity of our approach . the main point of cell division is the segregation of replicated chromosomes into two daughter cells . this is chiefly executed by the mitotic spindle , consisting of mts , organized into two astral structures whose centres derive from centrosomes and associated proteins . faithful segregation of the chromosomes requires proper bi - orientation and thus dynamic kinetochore - mt attachments , i.e. it must allow erroneous attachments kinetochores attached to a mt emanating from the wrong pole to be corrected . in other words , kinetochore - mt attachments can be unstable or stable , short - lived or long - lived . for a given value of the diffusion amplitude @xmath91 we can vary @xmath47 and plot the detachment times with respect to constant pulling force , we see that the peak height as well as the offset @xmath141 decrease with @xmath47 , and at a critical value the detachment time at zero force @xmath142 is higher than the peak . decreasing @xmath47 further makes the peak disappear completely and we end up with a detachment curve that is purely exponentially decreasing with applied force . as an example , we plot in fig . s1a the detachment times for @xmath143 nm @xmath100s@xmath110 for set of values of @xmath47 . as @xmath47 decreases , the catch bond peak is reduced and is shifted to smaller forces . at a critical value - for @xmath143 nm @xmath100s@xmath110 it is @xmath144 - the peak detachment time is at the same height as the zero - force detachment time @xmath145 . then , the peak completely disappears for @xmath146 and the detachment curve is a simple exponential decay . these observations allow us to construct a very simple phase diagram " where the transition line corresponds to the condition for which the catch bond disappears . in the stable region the attachment time increases with small applied forces , while it decreases in the the unstable region . an example for @xmath112pnnm@xmath39 and @xmath143 is shown in fig s1b . a stable region for @xmath147 ( right dash ) and an unstable region for @xmath148 , which splits into a weak " part ( left dash ) where @xmath149 and a truly unstable part ( cross dash ) with no peak at all . the same procedure can be repeated for different values of @xmath91 , allowing us to construct a phase diagram for the stability of the attachment which we report in fig . our model for mt - kinetochore attachments captures and explains in easy - to - understand terms several key features of experimentally observed phenomena . most prominently , we could reproduce the detachment times of mts from kinetochores that are pulled with constant force @xcite . using the model we simulate the emergence of the _ catch - bond _ the stabilization by tension of the kinetochore interface with a depolymerizing protofilament , emphasizing the role of the geometrical conformation of the kinetochore - microtubule interface . we identified different mechanisms of kinetochore - microtubule detachment for straight and curled protofilament tips . 1 . kinetochore fibrils are attached to straight protofilaments and detach due to thermal activation . this detachment is facilitated by an applied force , resulting in exponential weakening of the attachment . neighbor fibrils that have attached to each other form loops into which curling protofilaments hook . when no external force is applied the loop can escape by thermal fluctuations , i.e. brownian motion . when the loop and pf - tip are pulled apart , the thermal fluctuations acquire a bias to move preferentially towards the tip of the pf , making an escape from a ram s horn conformation more unlikely , thereby stabilizing the attachment . however , if the pulling force is too strong , then the ram s horn becomes uncurled or the loop breaks , so that the attachment is destabilized again . the third point makes an experimental prediction at higher forces the detachment rate from depolymerizing pfs increases again that could be easily be confirmed experimentally . accordingly , one should extend the experiments plotted in figure 4c of ref . @xcite to measure the detachment rate from depolymerizing mts at higher forces . the two - state model of akiyoshi et . implicitly assume that the detachment rate should continue to decrease with applied force , while our model predicts that it should eventually go up . another prediction that follows from our results is the stabilization of kinetochore - mt attachments via the intrinsic curvature of the mt protofilaments . the simulations show that the higher the pf curvature the longer the lifetime of the attachment . if the intrinsic curvature is small enough then the stabilization , together with the _ catch - bond _ disappears and the attachment becomes weaker with applied force . in our model the duration of kinetochore - mt attachments depends heavily on the flexural rigidity and intrinsic curvature of the pfs . in particular , our model suggests that a variation of the intrinsic curvature @xmath47 can cause a switch between stable / unstable attachments of depolymerizing pfs from kinetochores . this seems reasonable since there is a broad distribution of bending angles found in experiments @xcite . although the intrinsic curvature of a pf in isolation would be solely determined by the angle and stiffness between gtp / gdp - tubulin blocks , it is not unreasonable to assume that other factors influence the mechanical properties of protofilaments and so regulates the stability of the kinetochore - mt interface . our model , therefore , naturally suggests an additional path to vary the type of the attachment from stable to unstable and vice - versa . this mechanical process can of course be seen as complementary to any other accompanying process , like for example the stability regulation by ndc80 @xcite and aurora b @xcite or the switch of the mt state from growing to shrinking @xcite . in particular , the breakdown of the catch - bond could also occur according to two additional pathways : ( i ) the dissociation of fibril loops and ( ii ) the fracture of mt - pfs . a possible hypothesis is that aurora b is responsible for the dissociation of fibril loops as it phosphorylates the terminal ends of the ndc80 complex @xcite . to our best knowledge the influence of chemical or biological agents on the elastic properties of mts and pfs and their tips has not been studied and it will be exciting to see whether future experiments prove our predictions . earlier models for the kinetochore - microtubule interface make use of the curling of pfs to show how force can be produced by the geometry of the protofilament @xcite . these studies focus mainly on the load - bearing capabilities of the pfs and do not go into detail on the mechanism of the coupling to the kinetochore . while our work is building on these insights , we include some very recent experimental results into our model allowing us to explore the kinetochore interface in more detail , and make some specific predictions based on them . in contrast to the biased - diffusion based hill sleeve @xcite or the motor - protein based @xcite models , our model does not investigate the biochemical aspects of the kinetochore - mt interface , but focus instead on its mechanics . the biochemical properties of the interface that are needed , i.e. the fibril - tip attachment strength to neighbor fibril tips and the mt surface , are not readily available and have to be fit . we furthermore investigate slightly different phenomena than solely chromosome transport and breathing for which these models were designed for . our model sheds light instead on the genesis of the kinetochore - mt interface together with the role of the geometrical conformation of the pf tips . an approach that is similar in spirit to ours was recently published in ref . there the authors also model the collective action of kinetochore fibrils on the mts , but the methodology employed is quite different . while in that paper the authors explore the kinetochore - mt interface in a mean - field approximation using rate equations , we simulate directly the movement of each fibril and its interactions with the mt separately and then look at the emergent behavior . tellingly , in both studies the cooperation of kinetochore fibrils is found to be essential to strengthen and stabilize the interface . another difference , of course , is that we explicitly make use of the geometrical details of the protofilament curl and base on it predictions on the emergence of the kinetochore - microtubule interface . our results fit neatly between different facets of the study of the kinetochore - microtubule interface and condense established experimental results into a new kind of model . using molecular and stochastic dynamics simulations , we are able to elucidate and predict some very new aspects of the attachment of kinetochore fibrils to microtubule protofilaments . in this paper we introduce a computational model of the kinetochore - microtubule interface based on the synthesis of different experimental observations showing that fibrils extend from the kinetochore and capture mt tips @xcite , kinetochore proteins form entangled networks @xcite , the essential kinetochore protein complex ndc80 interacts in different ways with curved and straight protofilaments @xcite and ndc80 complexes bind to each other @xcite . hence , we assume that individual kinetochore fibrils are composed by different essential protein - complexes and can be treated as building blocks for the kinetochore - mt interface . we then employ a three - dimensional mt model treating the tubulin hetero - dimers as wedge shaped blocks that can have two configuration corresponding to gtp and gdp - bound tubulin . the tubulin blocks organize into protofilaments with intrinsic curvature ( gdp - tubulin ) or without ( gtp - tubulin ) . the kinetochore fibrils are modelled as bead - spring polymers that can attach to the top - side of straight protofilaments . the fibril tips also attach to each other forming effective loops into which curling protofilament - tips can hook . the model provides a clear illustration on how the conformation of mt - pfs influence the stability of their attachments with the kinetochore that is quite general and robust . beside its application to the experiments reported in ref . @xcite , our model represents a general computational tool that could be useful to guide future experimental investigations of mts - kinetochore interactions . we thank zoe budrikis and alessandro sellerio for useful discussions and comments . zb and sz are supported by the european research council through the advanced grant 2011 sizeffects . work in the laboratory of hm is funded by feder through the operational competitiveness programme compete and by national funds through fct - fundaco para a cincia e a tecnologia under the project fcomp-01 - 0124-feder-015941 ( ptdc / sau - onc/112917/2009 ) , the human frontier science program and the 7th framework program grant precise from the european research council . 55 [ 1]#1 abbas , t. , m. a. keaton , and a. dutta . 2013 . genomic instability in cancer . _ cold spring harb . _ 5:a012914 . akiyoshi , b. , k. k. sarangapani , a. f. powers , c. r. nelson , s. l. reichow , h. arellano - santoyo , t. gonen , j. a. ranish , c. l. asbury , and s. biggins . 2010 . tension directly stabilizes reconstituted kinetochore microtubule attachments . _ nature_. 468:576579 . , m. , b. medina , a. santamaria , j. zou , c. plasberg - hill , a. madhumalar , u. jayachandran , p. m. redli , j. rappsilber , e. a. nigg , and a. a. jeyaprakash . 2014 . structural basis for microtubule recognition by the human kinetochore ska complex _ 5:2964 . alushin , g. m. , v. h. ramey , s. pasqualato , d. a. ball , n. grigorieff , a. musacchio , and e. nogales . the ndc80 kinetochore complex forms oligomeric arrays along microtubules . _ nature_. 467:805810 . biggins , s. , and a. w. murray . the budding yeast protein kinase ipl1/aurora allows the absence of tension to activate the spindle checkpoint . _ genes dev . _ 15:31183129 . cheeseman , i. m. , and a. desai . molecular architecture of the kinetochore microtubule interface . nat rev mol cell biol_. 9:3346 . cheng , s. , a. aggarwal , and m. j. stevens . self - assembly of artificial microtubules . _ soft matter_. 8:56665678 . cheng , s. , and m. j. stevens . self - assembly of chiral tubules . _ soft matter_. 10:510518 . chrtien , d. , i. jnosi , j .- c . taveau , and h. flyvbjerg . microtubule s conformational cap . _ cell structure and function_. 24:299303 . ciferri , c. , s. pasqualato , e. screpanti , g. varetti , s. santaguida , g. dosreis , a. maiolica , j. polka , j. g. deluca , p. d. wulf , m. salek , j. rappsilber , c. a. moores , e. d. salmon , and a. musacchio . 2008 . implications for kinetochore microtubule attachment from the structure of an engineered ndc80 complex . _ cell_. 133:427439 . cimini , d. , x. wan , c. b. hirel , and e. d. salmon . aurora kinase promotes turnover of kinetochore microtubules to reduce chromosome segregation errors . _ 16:17111718 . civelekoglu - scholey , g. , d. j. sharp , a. mogilner , and j. m. scholey . 2006 . model of chromosome motility in drosophila embryos : adaptation of a general mechanism for rapid mitosis . _ biophys . j. _ 90:3966 . deluca , j. g. , w. e. gall , c. ciferri , d. cimini , a. musacchio , and e. d. salmon . kinetochore microtubule dynamics and attachment stability are regulated by hec1 . _ cell_. 127:969982 . dong , y. , k. j. van den beldt , x. meng , a. khodjakov , and b. f. mcewen . the outer plate in vertebrate kinetochores is a flexible network with multiple microtubule interactions . cell biol_. 9:516522 . efremov , a. k. , e. l. grishchuk , j. r. mcintosh , and f. i. ataullakhanov . 2007 . in search of an optimal ring to couple microtubule depolymerization to processive chromosome motions . _ pnas . _ 104:1901719022 . grishchuk , e. l. , and j. r. mcintosh . 2006 . microtubule depolymerization can drive poleward chromosome motion in fission yeast . _ embo j. _ 25:48884896 . grishchuk , e. l. , m. i. molodtsov , f. i. ataullakhanov , and j. r. mcintosh . force production by disassembling microtubules . _ nature_. 438:384388 . gudimchuk , n. , b. vitre , y. kim , a. kiyatkin , d. w. cleveland , f. i. ataullakhanov , and e. l. grishchuk . kinetochore kinesin cenp - e is a processive bi - directional tracker of dynamic microtubule tips . . hill , t. l. 1985 . theoretical problems related to the attachment of microtubules to kinetochores . _ pnas_. 82:44044408 . hunyadi , v. , d. chrtien , h. flyvbjerg , and i. m. jnosi . why is the microtubule lattice helical ? _ cell_. 99:117128 . hunyadi , v. , and i. m. jnosi . metastability of microtubules induced by competing internal forces . 92:30923097 . jnosi , i. m. , d. chrtien , and h. flyvbjerg . 1998 . modeling elastic properties of microtubule tips and walls . _ european biophysics journal_. 27:501513 . jnosi , i. m. , d. chrtien , and h. flyvbjerg . structural microtubule cap : stability , catastrophe , rescue and third state . _ biophysical journal_. 83:13171330 . joglekar , a. p. , and a. j. hunt . 2002 . a simple , mechanistic model for directional instability during mitotic chromosome movement . _ biophys . j. _ 83:42 . kis , a. , s. kasas , b. babi , a. j. kulik , w. benot , g. a. d. briggs , c. schnenberger , s. catsicas , and l. forr . nanomechanics of microtubules . _ 89:248101 . koshland , d. e. , t. j. mitchison , and m. w. kirschner . polewards chromosome movement driven by microtubule depolymerization in vitro . _ nature_. 331:499504 . lampson , m. a. , and i. m. cheeseman . 2011 . sensing centromere tension : aurora b and the regulation of kinetochore function . _ trends cell biol . _ 21:133128 . lawrimore , j. , k. s. bloom , and e. d. salmon . 2011 . point centromeres contain more than a single centromere - specific cse4 ( cenp - a ) nucleosome . _ j. cell biol . _ 195:573582 . marshall , b. t. , m. long , j. w. piper , t. yago , r. p. mcever , and c. zhu . 2003 . direct observation of catch bonds involving cell - adhesion molecules . _ 423:190193 . mcintosh , j. r. , e. l. grishchuk , m. k. morphew , a. k. efremov , k. zhudenkov , v. a. volkov , i. m. cheeseman , a. desai , d. n. mastronarde , and f. i. ataullakhanov . fibrils connect microtubule tips with kinetochores : a mechanism to couple tubulin dynamics to chromosome motion . _ cell_. 135:135322 . mcintosh , j. r. , e. otoole , k. zhudenkov , m. morhpew , c. schwartz , f. i. ataullakhanov , and e. l. grishchuk . 2013 . conserved and divergent features of kinetochores and spindle microtubule ends from five species . _ j. cell biol . _ 200:459474 . mcintosh , j. r. , v. volkov , f. i. ataullakhanov , and e. l. grishchuk . tubulin depolymerization may be an ancient biological motor . _ j. cell sci . _ 123:34253434 . mickey , b. , and j. howard . rigidity of microtubules is increased by stabilizing agents . _ j. cell biol . _ 130:909 917 . mitchison , t. j. 1989 . polewards microtubule flux in the mitotic spindle : evidence from photoactivation of fluorescence . _ j. cell biol . _ 109:637652 . mitchison , t. j. , and m. kirschner . dynamic instability of microtubule growth . _ nature_. 312:237242 . molodtsov , m. i. , e. a. ermakova , e. e. shnol , e. l. grishchuk , j. r. mcintosh , and f. i. ataullakhanov . 2005 . a molecular - mechanical model of the microtubule . _ biophys . j. _ 88:31673179 . molodtsov , m. i. , e. l. grishchuk , a. k. efremov , j. r. mcintosh , and f. i. ataullakhanov . force production by depolymerizing microtubules : a theoretical study . _ pnas_. 102:43534358 . mller - reichert , t. , d. chrtien , f. severin , and a. a. hyman . structural changes at microtubule ends accompanying gtp hydrolysis : information from a slowly hydrolyzable analogue of gtp , guanylyl ( alpha , beta)methylenediphosphonate . _ pnas_. 95:36613666 . plimpton , s. 1995 . fast parallel algorithms for short - range molecular dynamics . _ j. comp . _ 117:119 . rusan , n. m. , c. j. fagerstrom , a. c. yvon , and p. wadsworth . 2001 . cell cycle - dependent changes in microtubule dynamics in living cells expressing green fluorescent protein - alpha tubulin . _ cell_. 12:971 . schmidt , j. c. , h. arthanari , a. boeszoemenyi , n. m. dashkevich , e. m. wilson - kubalek , n. monnier , m. marcus , m. oberer , r. a. milligan , m. bathe , g. wagner , e. l. grishchuk , and i. m. cheeseman . 2012 . the kinetochore - bound ska1 complex tracks depolymerizing microtubules and binds to curved protofilaments . cell_. 23:968980 . shtylla , b. , and d. chowdhury . 2013 . a theoretical model for attachment lifetimes of kinetochore - microtubules : mechano - kinetic catch - bond mechanism for error - correction . _ arxiv_. : 1301.5692v2 . shtylla , b. , and j. p. keener . 2011 . a mathematical model for force generation at the kinetochore - microtubule interface . _ siam j. appl . _ 71:18211848 . suzuki , a. , t. hori , t. nishino , j. usukura , a. miyagi , k. morikawa , and t. fukagawa . spindle microtubules generate tension - dependent changes in the distribution of inner kinetochore proteins . _ j. cell biol . _ 193:12540 . tanaka , k. , e. kitamura , y. kitamura , and t. u. tanaka . 2007 . molecular mechanisms of microtubule - dependent kinetochore transport toward spindle poles . _ j. cell biol . _ 178:269281 . tanaka , t. u. , n. rachidi , c. janke , g. pereira , m. galova , e. schiebel , m. j. stark , and k. nasmyth . evidence that the ipl1-sli15 ( aurora kinase - incenp ) complex promotes chromosome bi - orientation by altering kinetochore - spindle pole connections . _ cell_. 108:317329 . thomas , w. e. , v. vogel , and e. sokurenko . 2008 . biophysics of catch bonds . _ 37:399419 . tirnauer , j. s. , j. c. canman , e. d. salmon , and t. j. mitchison . targets to kinetochores with attached , polymerizing microtubules . 13:43084316 . vanburen , v. , l. cassimeris , and d. j. odde . mechanochemical model of microtubule structure and self - assembly kinetics . . j. _ 89:29112926 . volkov , v. a. , a. v. zaytsev , n. gudimchuk , p. m. grissom , a. l. gintsburg , f. i. ataullakhanov , j. r. mcintosh , and e. l. grishchuk . long tethers provide high - force coupling of the dam1 ring to shortening microtubules . _ pnas_. 110:77087713 . wang , h. w. , and e. nogales . 2005 . nucleotide - dependent bending flexibility of tubulin regulates microtubule assembly . _ nature_. 435:911915 . westermann , s. , a. avila - sakar , h. w. wang , h. niederstrasser , j. wong , d. g. drubin , e. nogales , and g. barnes . 2005 . formation of a dynamic kinetochore microtubule interface through assembly of the dam1 ring complex . cell_. 17:277290 . westermann , s. , h. w. wang , a. avila - sakar , d. g. drubin , e. nogales , and g. barnes . the dam1 kinetochore ring complex moves processively on depolymerizing microtubule ends . _ nature_. 440:565569 . zapperi , s. , and l. mahadevan . dynamic instability of a growing adsorbed polymorphic filament . _ biophys . j. _ 101:267275 . zaytsev , a. v. , f. i. ataullakhanov , and e. l. grishchuk . highly transient molecular interactions underlie the stability of kinetochore microtubule attachment during cell division . _ cell . and mol . _ 6:393405 . kinetochore molecules are assumed to form uniform units represented as fibrils , here shown in red . the tips of the fibrils , dark red , can sense and attach to the outer surface of an incident microtubule ( blue ) if the mt straight , while fibril - tips do not attach directly to mt - tips that curl . the kinetochore fibril - tips also attach to each other when brought into close contact , a circumstance that is especially likely when the fibrils attach to the mt as neighbours . when fibril tips attach to each other , loops are formed into which curling protofilaments of a depolymerising mt can hook and thus exert force on the kinetochore . this movie shows a single protofilament interacting with a bunch of kinetochore fibrils from two different angles . the incident straight protofilament captures some kinetochore fibril - tips . when hydrolysed and curling , the protofilament - tip hooks into the loops formed by the fibrils and exerts a force - the loop is visibly pulled out of the mass of the other fibrils , while the fibril bases , as well as the protofilament base , are held static and are not allowed to move . depolymerisation of the protofilament increases the force exerted on the kinetochore - the fibril loop stretches further - until the interface fractures . this movie shows the interface of a straight pf with for simplicity only one attached kinetochore fibril . after initial thermalization , the fibril base is moved with constant velocity , resulting in the unwrapping of the fibril during which the pulling force is not transmitted onto the kinetochore - protofilament interface . after the unwrapping is complete , the fibril is pulled off the pf , after enough force built up to overcome the potential attaching the fibril tip to the pf . we use the same setup as in movie s3 , with the difference that the bases of the kinetochore fibrils are not held fixed , but each base is attached to every other by linear springs , to emulate the kinetochore body . at first no external force acts on the fibrils and the curling of the protofilament transports the fibril bundle a short distance until external forces are switched . then the kinetochore fibril bases are pulled with constant velocity which results in the uncurling of the depolymerizing protofilament and eventual detachment . we use a similar setup as in movie s4 , with the following differences : we reduce the number of fibrils to two and the binding energy between the fibril tips by 50% . when the kinetochore fibril bases are pulled with constant velocity , the mt starts to uncurl slightly , but then the force on the fibril tips becomes too high and they detach from each other .

we argue that features of hadron production in relativistic nuclear collisions , mainly at cern - sps energies , may be explained by the existence of three forms of matter : hadronic matter , quarkyonic matter , and a quark - gluon plasma . we suggest that these meet at a triple point in the qcd phase diagram . some of the features explained , both qualitatively and semi - quantitatively , include the curve for the decoupling of chemical equilibrium , along with the non - monotonic behavior of strange particle multiplicity ratios at center of mass energies near gev . if the transition(s ) between the three phases are merely crossover(s ) , the triple point is only approximate . , , , , , , , , , , , and dense quark matter , chiral symmetry breaking , large expansion

0911.4806

the sps heavy ion program at cern resulted in some of the first experimental data on heavy ion collisions at ultrarelativistic energies , see , e.g. , @xcite . a summary of these data and implications for the possible formation of a new state of matter were announced in a cern press release @xcite . in this paper we consider some generic features discovered in heavy ion experiments at the sps . this gives us a general overview of how the collisions of heavy ions evolve in going from low energies , as studied at the sis ( gsi ) and the ags ( bnl ) , to higher energies , at rhic ( bnl ) and soon at the lhc ( cern ) @xcite . in particular , we concentrate on hadron abundances in heavy ion collisions . these have been widely studied using resonance gas models . by assuming that the observed particle yields are generated at a common surface at which all particles decouple , values of the baryon chemical potential , @xmath2 , and temperature , @xmath3 , on this surface , can be extracted . fitting these two parameters , @xmath2 and @xmath3 , together with the volume parameter gives values for the particle abundances which are in close agreement with experiment @xcite . the resulting values of @xmath2 and @xmath3 are shown in fig . [ energy_t_mu ] as functions of center - of - mass energy per nucleon pair . we note that , near 10 gev center of mass energy , the temperature saturates with increasing beam energy , reaching an asymptotic value of about 160 mev , while the baryon chemical potential decreases smoothly . plotting these temperature - chemical potential pairs for all available energies results in a phase diagram - like picture as is illustrated in fig . [ line ] . in the @xmath2 region from 800 to 400 mev , as @xmath3 increases from 50 to 150 mev , the experimental points rise approximately linearly . in contrast , below @xmath4 mev , the temperature is approximately constant , @xmath5 mev . the highest collision energies studied to date at rhic are those for which @xmath6 mev . also shown on this plot are lines of fixed energy per particle and fixed entropy density per @xmath7 ; also shown is a line of hadron percolation ( see below ) . these experimental results can be compared to phase transition points computed on the lattice @xcite . numerical simulations in lattice qcd can be performed at nonzero temperature , and small values of @xmath2 without running into problems of principle . at @xmath8 , these simulations indicate that there is no true phase transition from hadronic matter to a quark - gluon plasma , but rather a very rapid rise in the energy density at a temperature @xmath9 which lines in @xmath10 mev within the systematic errors . further , studies using the lattice technique imply that @xmath9 decreases very little as @xmath2 increases , at least for moderate values of @xmath2 . [ cols= " < , > " , ] with the parametrizations of @xmath3 and @xmath2 from fig . [ energy_t_mu ] one can compute the energy dependence of the production yields of various hadrons relative to pions , shown in fig . [ fig_k2pi ] . important for our purposes is the observation that there are peaks in the abundances of strange to non - strange particles at center of mass energies near 10 gev . in particular , the @xmath11 and @xmath12 ratios exhibit rather pronounced maxima there . we further note that in the region near 10 gev , there is also a minimum in the chemical freeze - out volume @xcite obtained from the statistical model fit to particle yields @xcite , as well as in the volume obtained from the hanbury - brown and twiss ( hbt ) radii of the fireball @xcite . the energy dependence of the volume parameters is shown in fig . [ size ] . these experimental observations have long resisted interpretation in terms of a transition between hadronic matter and a quark - gluon plasma . the general structures observed in the data are well reproduced only by the most recent model calculations @xcite . there , it is argued that these structures arise due to the interplay between the limit in hadronic temperature ( see fig . [ energy_t_mu ] ) due to the qcd phase transition and the rapid decrease of @xmath2 with increasing energy , thereby establishing a connection between hadron gas and quark - gluon plasma . the possible existence of a critical endpoint is , however , not relevant for these considerations . the above described structures seem puzzling if the corresponding energies would probe a critical endpoint in the qcd phase diagram @xcite . near a critical point , lighter particles , such as pions , should be affected more than heavier particles , such as kaons ; hbt radii should also increase . both of these features are not easily linked to the trends in the data . we will discuss the relationship between the above statistical model descriptions of the transition to both the quark - gluon plasma and quarkyonic matter , the triple point where three phases of matter coexist , and the underlying contribution to the spectrum of strange particles below , and argue that generic features of these curves may be explained in this context . in the following we show that by considering quarkyonic matter , which was recently proposed @xcite , the two regimes observed in the phase diagram and described above can be understood as arising from a triple point where hadronic matter , the quark - gluon plasma , and quarkyonic matter all coexist . this triple point is located where the temperature is reaching its limiting value and , hence , is naturally also situated in the vicinity of the peaks in the observed hadron production ratios . a sketch of a possible phase diagram for qcd is shown in fig . [ quarkyonic ] . there are hadrons in the lower , left - hand corner of this phase diagram , at low temperatures and @xmath2 . there are two , qualitatively distinct , phase boundaries by which one can leave hadronic matter . the first , is to increase the temperature at low @xmath2 until it is beyond @xmath9 . this is the usual transition from a meson - dominated phase , is similar in this regime to that at large-@xmath2 ( @xmath130.12 @xmath14 ) @xcite . ] to a quark - gluon plasma . this phase boundary is probed by collisions at high sps energies , and by collisions at rhic and the lhc . the second way is to increase @xmath2 at low temperatures , @xmath15 , going from hadronic matter to quarkyonic matter . we suggest that this phase boundary is studied by heavy ion collisions at moderate and low energies , such as those at the ags , sis , and at low energies at the sps , and in the future at fair and nica @xcite . at a special value of the baryon chemical potential and temperature , there is a triple point where hadronic matter , the quark - gluon plasma , and quarkyonic matter all coexist . from experiment , [ line ] , we estimate that this occurs for @xmath16 this point is presumably near where the linear and the flat temperature regime in fig . [ line ] intersect . we argue in the following how this arises from a triple point . in thermodynamics a triple point is the point in a phase diagram where three lines of first order phase transitions meet . a common example is where a gas , liquid , and solid coexist at a given value of the pressure and temperature . since there are only first order phase transitions , no correlation length diverges at the triple point . for example , in the phase diagram of water , the phases of vapor , water , and ice all coexist at the triple point . there is also a critical point in the phase diagram of water , but it is situated far from the triple point , at much higher temperature and pressure . the properties of strongly interacting matter at large density are characterized by several order parameters . one is the thermal wilson or polyakov loop , which measures the degree of deconfinement reached . this is strictly an order parameter in theories without quarks , or in the limit of a large number of colors , @xmath17 , if the number of flavors , @xmath18 , is kept fixed . the second is the chiral condensate as an order parameter for chiral symmetry breaking . chiral symmetry is an exact symmetry when there are two ( or more ) flavors of massless quarks . the last is the density of baryons , which is an order parameter even in the large @xmath1 limit , when @xmath18 grows with @xmath1 @xcite . hadronic matter is confined and exhibits chiral symmetry breaking . it is technically difficult to define confinement for finite @xmath1 for a finite number of quark flavors , since the potential that separates quarks is never linear at large distances . this argument has a precise meaning at zero @xmath18 or infinite @xmath1 , or for zero temperature . nevertheless , there should be a well defined region of low baryon density and low temperature where the physical degrees of freedom are mesons . this phase is also to a good approximation free of baryons since their densities , @xmath19 for typical values of @xmath2 and @xmath3 not too close to the phase boundary . the quark - gluon plasma is deconfined with restored chiral symmetry , and has nonzero baryon number density when @xmath20 . it is composed of quarks and gluons , although we note that lattice simulations indicate that the transition to a deconfined state is rapid , but not discontinuous @xcite . this means that , for a range of temperatures above @xmath9 , there is a `` semi '' quark - gluon plasma , in which the theory is only partially deconfined @xcite . we neglect here the effects of the semi quark - gluon plasma , since lattice simulations @xcite indicate that the energy density rises quickly to values close to the ideal gas value near @xmath9 , and this is the main quantity which will concern us . this is unlike the pressure , which does not approach the ideal gas value until several times @xmath9 , and for which the semi quark - gluon plasma is important . quarkyonic matter is ( approximately ) confined , but has a large baryon number density , and also a large energy density . whether chiral symmetry is restored in quarkyonic matter is not yet fully understood . even at very high densities , there could be residual chiral symmetry breaking from pairing effects near the fermi surface . for the present discussion it does not matter when and how chiral symmetry is restored in the quarkyonic phase . we remark that studies of the sakai - sugimoto model at nonzero quark density serve as one realization of quarkyonic matter @xcite . at the outset we concede that , in the strict thermodynamic sense , the qcd phase diagram might or might not have a true triple point . after all , the deconfining transition at low @xmath2 , and nonzero temperature , appears not to be of first order , but a rapid crossover . if the deconfinement transition remained a crossover for all @xmath2 values , then the triple point would not be a true point , since it would not connect matter separated by a true first order phase transition . it might happen that there is a second order critical end point along the deconfinement line , in which case the triple point might truly reflect three different phases connected by first order phase transitions . we do suggest that there is a true triple point in the limit of an infinite number of colors @xcite . in this limit , the deconfinement transition is of first order @xcite , and the quarkyonic transition may exist @xcite . thus the behavior for qcd may be reminiscent of that for a large number of colors , and exhibit an _ approximate _ triple point . for the present discussion , it is not important whether the triple point is exact , or only approximate . what is important is that , in going from hadronic matter to either the quark - gluon plasma , or quarkyonic matter , there is a large increase in the number of degrees of freedom . hadronic matter is dominated by goldstone bosons . in qcd , the hadronic phase has three types of pions , and a relatively small amount of kaons ; for @xmath18 flavors , there are @xmath21 goldstone bosons in the hadronic phase . these goldstone bosons dominate bulk properties of the system for temperatures and quark chemical potentials , @xmath22 much smaller than @xmath23 . as one gets close to a transition temperature , massive degrees of freedom become important , eventually becoming so numerous that a transition to a new phase of matter is induced . as is well known , there are many more degrees of freedom in the quark - gluon plasma : for @xmath1 colors , there are @xmath24 bosonic and @xmath25 fermionic , or @xmath26 bosonic and 24(36 ) fermionic degrees of freedom in qcd with 2(3 ) favours . we note that for the pressure and energy density , ideal fermions contribute @xmath27 of a boson . while quarkyonic matter is confined , the principal point of ref . @xcite is that the energy density , or equivalently the number of degrees of freedom , can be counted as for deconfined quarks . while near the fermi surface the degrees of freedom are confined baryons , most of the energy density is due to quarks , deep in the fermi sea . this is a coarse description of what is surely a much more complicated reality . if we assume that chiral symmetry remains broken in the quarkyonic phase , quarkyonic matter then has @xmath28 bosonic , and @xmath29 fermionic , degrees of freedom . the number of fermionic degrees of freedom is half that of the quark - gluon plasma , since in quarkyonic matter , only quarks , but not anti - quarks , contribute . in qcd , there are 3(8 ) bosonic degrees of freedom , plus 12(18 ) fermionic degrees of freedom for 2(3 ) flavours . the number of degrees of freedom is smaller for quarkyonic matter than for the quark - gluon plasma , but significantly larger than the number of goldstone degrees of freedom of hadronic matter . thus , we argue , that while there may be no true phase transitions from either quarkyonic matter , or a quark - gluon plasma , to hadronic matter , there is a rapid decrease in the number of degrees of freedom and so in the energy density . this rapid decrease could well cause the matter to decouple , and so define , experimentally , the surfaces for chemical equilibrium . this is approximately true for the transition from the quark - gluon plasma to the hadronic phase , as observed at rhic energies @xcite . at rhic energies , chemical freeze - out was shown @xcite to take place very close ( within less than about 10 mev ) to the phase boundary , driven by the rapid density change across the phase transition . further it was argued that freeze - out ends when the system is fully hadronized , i.e. at low density in the hadronic phase . were this not the case @xcite , one would also expect different freeze - out parameters for each hadron species due to widely different hadronic cross sections . this is not observed . we believe this argument to be generic @xcite : to ensure simultaneous ( within a very small interval in temperature and chemical potential ) freeze - out of all hadrons , the freeze - out curve has to be very close to a line with a rapid density change . an immediate consequence of this would be that the chemical freeze - out curve delineates phase boundaries , not only for small values of @xmath2 but everywhere . but what provides the phase boundary for large values of @xmath2 , where the deconfinement transition seems far away , at least if one follows the guidance from lattice qcd calculations ? as already indicated above we believe that the transition from hadronic to quarkyonic matter provides the missing link . across the quarkyonic line , we would expect that the transition takes place in a range of baryon chemical potentials of order @xmath30 mev in width , for @xmath31 mev as a typical baryonic mass scale in qcd and for @xmath32 mev . this width is parametrically of order @xmath33 which accounts for its anomalously small size compared to typical hadronic energy scales . in this section , we explore a very simple model of the quarkyonic transition . this model only counts the number of degrees of freedom of baryonic resonances , and ignores effects due to the strong nucleonic interactions . it assumes that the resonance spectrum `` turns on '' in a very narrow window of @xmath2 , as suggested by the large @xmath1 arguments of the previous section . interaction effects should not therefore change the position of the phase boundary . nevertheless in realistic computations interactions should be taken into account , and a realistic spectrum of baryons should be used . these modifications will be discussed in the next section . resonance formation is the dominant feature for mesonic interactions , and the most detailed model of hadron dynamics , the dual resonance model @xcite , in fact describes all scattering amplitudes in terms of resonance poles in the different kinematic channels . the number of states of mass @xmath34 , the degeneracy @xmath35 , is found to increase as @xmath36 where @xmath37 gev@xmath38 is the universal regge resonance slope and @xmath39 a positive constant @xcite . a basic result in the study of interacting systems is that if the interactions are resonance - dominated , the system can be replaced by an ideal gas of all possible resonances @xcite . the partition function determining the thermodynamics of an ideal resonance gas @xcite becomes @xmath40 } , \label{2}\ ] ] where @xmath41 . it is seen that this partition function has a singular point at @xmath42 mev , indicating that the system can not exist at higher temperatures . previous work assuming self - similar resonance formation , the so - called statistical bootstrap model @xcite , had also led to an exponentially increasing level density , and for some time it was assumed that @xmath43 was the ultimate temperature of matter . subsequently it was noted @xcite that @xmath43 marks a critical point , with a possible new state of matter at @xmath44 , which presumably is the quark - gluon plasma . an alternative approach is based on the intrinsic size of hadrons @xcite . with increasing temperature , the hadron density increases , and - assuming again a mesonic system - the individual constituents will overlap more and more . at a certain density , the system will percolate , i.e. , form a connected network spanning the entire system . the spanning cluster consists of overlapping mesons , so that it ceases to be meaningful to speak of the existence of individual mesons within this cluster . the density of mesons in the cluster is at the percolation point approximately @xmath45 where @xmath46 and @xmath47 fm . we can now ask for the temperature at which an ideal resonance gas , with all resonances having size @xmath48 , attains this density . it is found to be @xcite @xmath49 so that such geometric percolation considerations lead to a limit of hadronic matter very much like that obtained from resonance dynamics . the `` mesonic '' arguments used up to now continue to be valid also in the presence of baryons , as long as the baryon density is well below the point of dense packing ; we will elaborate on this below . as a result , we conclude that resonance formation or percolation lead to a temperature limit @xmath43 approximately independent of the baryon density with baryochemical potential @xmath50 @xcite ; we ignore this here for simplicity . ] . our `` phase diagram '' thus is so far a straight horizontal line @xmath51 const . in the @xmath52 plane , as shown in fig . [ diagram ] . the nature of the limit depends on the conceptual basis . an ideal resonance gas with an exponentially growing mass spectrum results in a genuine thermal critical line , corresponding to continuous transitions ; the associated critical exponents can be determined in terms of the space dimension @xmath53 and the coefficient @xmath39 in eq . ( [ 1 ] ) @xcite . percolation is in general a geometric critical phenomenon , with singular behavior and corresponding critical exponents for cluster variables . it does not imply singular behavior of the partition function and could thus from a thermal point of view correspond to a rapid cross - over @xcite . we now turn to the other extreme , dense baryonic matter at low temperature . for baryochemical potential @xmath54 , the contribution of baryons / antibaryons and baryonic resonances is relatively small , but with increasing baryon density , they form an ever larger fraction of the species present in the medium , and beyond some baryon density , they become the dominant constituents . finally , at vanishing temperature , the medium consists essentially of nucleons . for vanishing or low baryon number density , when the interactions are resonance dominated , the system could be described as an ideal gas of all possible resonance species . at high baryon density , however , the dominant interaction is non - resonant . nuclear forces are short - range and strongly attractive at distances of about 1 fm ; but for distances around 0.5 fm , they become strongly repulsive . the former is what makes nuclei , the latter ( together with coulomb and fermi repulsion ) prevents them from collapsing . the repulsion between a proton and a neutron shows a purely baryonic `` hard - core '' effect and is connected neither to coulomb repulsion nor to pauli blocking of nucleons . as a consequence , the volume of a nucleus grows linearly with the sum of its protons and neutrons . with increasing baryon density , the conceptual basis of a resonance gas thus becomes less and less correct , so that eventually one should encounter a regime of quite different nature . at high baryon density , the most striking effect is the onset of a `` jamming '' of nucleons : the mobility of baryons in the medium becomes strongly restricted by the presence of other baryons , leading to a jammed state @xcite , as shown in fig . the inverse mobility @xmath55 of a nucleon here plays the role of an order parameter : up to a certain density , it is zero , and beyond this point , it remains finite . baryonic matter thus becomes again a medium of extensive hadrons of radius @xmath56 , but these now contain a hard core of a smaller radius @xmath57 . the overlap of such hadrons in percolation studies is thus restricted ; nevertheless , the percolation onset can still be determined @xcite , and it is found @xcite that the density of a spanning cluster now becomes @xmath58 assuming @xmath59 . with @xmath60 fm , this leads at @xmath61 to a limit of about 5.5 times standard nuclear density . requiring the baryon density ( baryons minus antibaryons ) in an ideal resonance gas to attain this limit as function of @xmath3 and @xmath50 then defines a critical curve based on baryon percolation . in the simplest model , @xmath62 becomes the limiting baryochemical potential @xmath61 . the general curve is included in fig . [ diagram ] @xcite . in the case of hard core percolation , a connection to thermodynamic critical behaviour has also been discussed @xcite . if a system with hard core repulsion between its constituents is in addition subject to a density - dependent negative background potential , first order critical behaviour can appear , ending in a second order critical point specified by the background potential strength and the hard core volume . the interpretation of the situation illustrated in fig . [ diagram ] allows different interesting possibilities . in ref . @xcite it is assumed that the state outside the hadronic matter region is a deconfined quark - gluon plasma . it is , however , also conceivable that below the meson percolation / resonance curve confined mesonic states survive , while baryons enter into the new phase . such quarkyonic matter @xcite is dealt with in detail in this work . one can get some insight into the nature of the transition in the various regions of @xmath2 and @xmath3 by plotting the entropy density inferred from resonance model descriptions as a function of center of mass energy of the collision , as shown in fig . [ baryon ] . for low energies , below the hypothetical critical point , the matter is baryonic , consistent with a transition from hadronic matter to quarkyonic matter . for higher energies , it is largely mesonic matter , and consistent with a transition from hadronic matter to a quark - gluon plasma . turning this into @xmath2 , one goes from a region dominated by baryons at decoupling , when @xmath63 mev , to one dominated by mesons at decoupling , for @xmath64 mev @xcite . as a simple model that embodies some of the features discussed above , we suggest that , for small values of the chemical potential , @xmath65 , the transition between a hadronic phase , and the quark - gluon plasma , is controlled by a single hagedorn temperature below , implies asymptotically @xmath66 . note , however , that , from the the presently - known hadron spectrum up to 2 gev , the effective @xmath43 for mesons appears to be larger than for baryons @xcite . ] for _ mesons _ , @xmath67 @xcite . assuming that this transition is controlled entirely by mesons , we obtain a line which is independent of @xmath2 . of course we do not believe that this behavior is exact , but it seems to be not a bad approximation in qcd . numerical simulations of lattice qcd imply that @xmath9 decreases very slowly , by only about @xmath68 , for @xmath2 from 0 to 400 mev @xcite . the @xmath2 independence of @xmath67 is also in accord with arguments at large @xmath1 and small @xmath18 , which imply that the critical temperature is independent of the baryon chemical potential ( as we discuss below , this is true as long as @xmath69 is of order one , and does not grow with @xmath1 . ) we suggest that this horizontal line intersects with a second line , which is controlled by a hagedorn temperature for _ baryons _ , @xmath70 . if there is such a hagedorn temperature , the density of states of baryons grows like @xmath71 we assume , as is typical for a hagedorn spectrum , that this balances against the usual boltzmann factor , @xmath72 . then for a given value of @xmath2 , there is a phase transition at a `` quarkyonic '' temperature @xmath73 , which is @xmath2-dependent . in the plane of @xmath2 and temperature , this dependence is just a straight line : @xmath74 we have made a gross approximation in this formula , which is represented by the parameter @xmath75 . the hagedorn mass spectrum in eq . ( [ baryonic_hagedorn ] ) is only valid asymptotically , as @xmath76 . thus strictly speaking , the transition temperature from a hagedorn spectrum is independent of @xmath2 . ( for this reason , in string models the hagedorn temperature is common to all particles , determined only by a single parameter , which is the string tension @xcite . ) instead , in eq . ( [ quark_temp ] ) we introduce a new parameter , @xmath75 , by hand . this is meant to represent a _ finite _ mass scale at which a hagedorn spectrum appears . here @xmath75 is entirely a phenomenological parameter , meant to illustrate how the transition temperature @xmath73 to quarkyonic matter might depend upon @xmath2 . clearly @xmath77 can not be less than the mass of the lightest baryon ; it could well be much larger . as one decreases @xmath2 , eventually there will be a temperature at which this line crosses that for deconfinement . we assume that , when this happens , the line for the quarkyonic transition ends and that the transition to a quark - gluon plasma , which has a much larger energy density , dominates . the point at which these two lines cross defines the position of a triple point : @xmath78 we stress that our approximations are _ very _ crude , and are only meant to illustrate how a triple point _ might _ arise . the transition temperature line of eq . ( [ quark_temp ] ) intersects the axis of @xmath79 when @xmath80 . this formula does not apply at arbitrarily low temperatures , however . a quarkyonic phase is defined to be one in which both the baryon and energy densities are large , but at small temperatures , when @xmath2 is close to the nucleon mass , one probes not quarkyonic , but dilute nuclear matter . at large @xmath1 , though , the region in which nuclear matter is dilute is a narrow window in @xmath2 @xcite . this suggests that @xmath81 is near the nucleon mass ; fitting to fig . [ line ] gives @xmath82 gev . the value of the baryonic hagedorn temperature can be read off from where @xmath83 intersects the axis of @xmath8 . again from fig . [ line ] , this gives @xmath84 mev . these values for mesonic and baryonic hagedorn temperatures should only be taken as illustrative . even at @xmath8 , experiment gives us the results at chemical freeze - out . this value is certainly lower than the temperature for the true transition ( or crossover ) , and is lower still than that for the hagedorn temperature . one might , however , expect that these values are close to one another . this is indicated by results from the lattice in pure gauge theories @xcite . the limit of a large number of colors shows that the introduction of the parameter @xmath75 is not quite as contrived as might first appear . in the limit of large @xmath1 and small @xmath18 the transition from a hadronic to a quarkyonic phase is a straight line along @xmath85 , where @xmath86 is the mass of the lightest baryon ( up to small effects from nuclear binding ) @xcite . this is just the usual mass threshold for a chemical potential . in the limit in which both @xmath1 and @xmath18 are large , one can not speak of deconfinement rigorously , and there is only a phase transition for the condensation of baryons , which is a straight line in the @xmath87 plane @xcite . this is because the density of states , for even the lowest baryon multiplet , grows exponentially . note that this is analogous , but not identical , to a hagedorn temperature , since the exponential growth is for the lightest multiplet , and not for asymptotically large masses . there are several effects which will act to modify this naive prediction , however . first , even in the hadronic phase , baryons interact strongly with the numerous mesons . this will modify the baryon mass , and so shift the threshold at which they condense . second , baryon baryon interactions are strong . in ordinary nuclear matter it is well known that baryons have a large hard core repulsion between them , and this surely persists when both @xmath1 and @xmath18 are large . such a hard core repulsion between baryons will act to cut off the singularity in the free energy , which otherwise would be generated by an exponential growth in the degeneracy of states . of course in qcd the degeneracy of the lowest baryons does not grow exponentially . but @xmath75 can then be viewed as a way of characterizing when the growth of baryonic states starts to take off . for example , this could be estimated more accurately in resonance gas models . consider , alternately , the result of @xcite : @xmath88 in this equation , @xmath89 is the number of approximately degenerate baryonic states . extrapolating the formula for large @xmath1 and @xmath18 down to small values , for three colors and two flavors , @xmath90 mev ; for three flavors , @xmath91 mev for @xmath92 mev @xcite . in qcd we can also estimate @xmath89 directly . including nucleons and the @xmath93 resonance , @xmath94 ; including all strange baryons , @xmath95 . further , by including higher resonances , in eq . ( [ largenf ] ) we should take not the nucleon mass for @xmath96 , but some heavier state , which can then be used to define @xmath75 . by fitting the data in fig . [ line ] , with @xmath97 gev , one finds that @xmath98 , which is not too far from the extrapolation from large @xmath1 . of course , the precise tradeoff between the increasing masses of various states and their abundances is a tricky issue . it is not clear how much to include of the flavor and spin excitations of the lowest mass nucleon states , and hence the uncertainty . this may be addressed more directly within statistical model computations . a natural question is what happens to the two phase transition lines beyond the triple point . consider first the transition between the hadronic phase and the quark - gluon plasma , to the right of the triple point at approximately constant temperature , with @xmath99 . at large @xmath1 , this line is of first order , and remains a boundary for a true phase transition . the lattice qcd results show that the rapid rise in the energy density is relatively independent of @xmath2 ; thus we suggest that this line delineates an approximate phase transition for @xmath99 . at large @xmath1 , when @xmath100 , eventually deconfinement is washed out by the quarks , and there is a critical endpoint for deconfinement . ( this is the value of chemical potential where the debye screening length becomes less than the confinement size scale . ) in qcd , since there is nt a first order transition to deconfinement , we expect that eventually the large increase in the energy density , seen in a narrow region in temperature , is just washed out by the contribution of dense quarks . it is also possible to consider continuing the phase boundary for the quarkyonic phase at chemical potentials below the triple point , that is , for @xmath101 . one might imagine that there is then a line for the quarkyonic transition _ above _ that for deconfinement , with @xmath102 when @xmath101 . even at large @xmath1 , such a line of quarkyonic `` transitions '' can only reflect the properties of some metastable state in the ( semi-)quark - gluon plasma . numerical simulations on the lattice do find that like the energy density , quark number susceptibilities approach the ideal gas values very near @xmath9 , by @xmath103 above @xmath9 @xcite . thus perhaps this change in the quark number susceptibilities reflects the remnants of the quarkyonic `` transition '' in the deconfined phase . we conclude by discussing the relationship to the chiral phase transition . it is possible that the triple point coincides with a critical end point for a line of first order ( chiral ) transition @xcite . however , as we noted above , the experimentally observed properties of the triple point do not seem indicative of a critical point . it is possible that qcd matter behaves similar to water , with a critical end point for the chiral transition which is distinct from the triple point . if so , it probably exists in the quarkyonic phase . it is also possible that there is no well defined chiral transition ; _ i.e. _ , that because of the nonzero quark masses , the chiral transition is only a crossover . there would then be no critical end point for the chiral phase transition . we have already discussed and shown in fig . [ fig_k2pi ] that there are abrupt changes in the abundances of various ratios of strange to non - strange particles at @xmath104 around 10 gev . is in logarithmic scale so the variation at higher energies is indeed quite slow . ] the reason for such behaviour may be linked with the appearance of the quarkyonic phase . along the quarkyonic line , the temperature changes substantially . the fraction of strange particles should increase as the temperature increases . along the quark - gluon plasma line , the temperature is constant and we expect the strange quark relative abundance to be roughly unchanged . when these two boundaries meet , we would expect a change in the strange quark density near the triple point . this is most easily seen approaching the triple point along the quarkyonic curve since as one approaches the deconfinement transition , there should be a rapid increase in the energy density , favoring a higher relative abundance of strange quarks . the strange quark relative abundance increases rapidly as one approaches the triple point , but then slowly decreases beyond it due to the decreasing @xmath2 at the almost constant temperature . some of the strange to non - strange particle ratios are very sensitive to small variations in @xmath3 and/or @xmath2 as demonstrated in fig . [ wroblewski]-left showing the @xmath11 ratio as contour lines in the @xmath105 plane @xcite . if in the region of the triple point the freeze - out would happen at somewhat higher temperatures then especially this ratio will increase . [ wroblewski]-left illustrates that in the statistical model the @xmath11 ratio can never exceed the value of @xmath106 . while different strange to non - strange particle ratios exhibit different trends , the relative strangeness content quantified by the wrblewski factor , determines the relative abundance of the initially produced strange to light quarks multiplicities . ] , similarly as @xmath11 , exhibits a well pronounced peak as seen in fig . [ wroblewski]-right . the peak in the strangeness abundance naturally arises in the statistical model due to the presence of the phase boundaries between qgp - quarkyonic matter and hadronic matter , for the reasons stated above . it is nevertheless an indirect measure of the singularity associated with a triple point . if the triple point region is somewhat spread out , we might expect that the peaks in various particle ratios might not appear at the same point . if the critical point region is very narrow , there should be approximately discontinuous behavior at the critical point , but this does not necessarily imply a maximum in ratios of strange to non - strange particles near to the critical point . the relationship between strangeness abundance , the triple point , and experimental data is certainly worth more detailed and precise experimental and theoretical study . so far chiral symmetry played little role in our argument ; the hadron resonance gas description assumes no explicit modification on the hadron masses in a hot and dense medium . in principle , however , it would be conceivable to anticipate a substantial change in the hadron spectrum depending on whether chiral symmetry is ( partially ) restored or not . there are in fact several theoretical and experimental indications that chiral symmetry is affected in a medium @xcite . for instance , the leading - order of the virial expansion suggests that the chiral condensate receives , at normal nuclear density , a @xmath107-@xmath108 reduction , which is shifted back by around @xmath68 by higher order corrections in the in - medium chiral perturbation theory @xcite . of course , the chiral condensate itself is not a direct experimental observable , but useful information is available from the spectroscopy of deeply - bound pionic atoms and the experimentally deduced in - medium pion decay constant at normal nuclear density is reduced by 36% compared to its vacuum value @xcite . although chiral perturbation theory gives a fairly model independent statement on chiral properties at finite temperature and baryon density , forming a productive research area together with experimental measurements , its validity is strictly limited to low - energy regimes . as one tries to go beyond low - energy regimes to explore the phase diagram of strongly interacting qcd matter , it is extremely difficult to make any statement in a model - independent way . it is notable that the in - medium condensate strongly relies on the pion mass coming from two - pion exchange correlations with virtual @xmath109 excitations which stabilizes the dropping of the condensate for physical pion mass with increasing density @xcite . particularly , a deviation from the result in the linear density approximation is remarkable for symmetric nuclear matter . this might indicate that the chiral symmetry restoration would take place at much higher density as compared to the critical density given in the mean - field models . this also could suggest that in - medium correlations might weaken a phase transition and eventually a first - order phase transition might disappear from the phase diagram . in the context of the chiral phase transition the most frequently used models are the nambu jona - lasinio ( njl ) model and the quark meson ( qm ) model , which are sometimes improved by the introduction of partial gauge degrees of freedom , namely the polyakov loop , and promoted to the pnjl and pqm @xcite models , respectively . crucial points in this sort of model treatment are that a description in terms of quasi - quarks is assumed and the effect of the confinement is totally neglected . such chiral quark models as well as another non - perturbative approach using the schwinger - dyson equation @xcite favor a first - order chiral phase transition at high density . this suggests a termination point of the first - order phase boundary , which defines a critical endpoint ( or often called the qcd critical point ) . results from finite - density lattice simulations are far from conclusive yet and thus the existence of the critical endpoint is still under extensive dispute . in a description in terms of quarks , the driving force to induce a chiral transition is the density contribution to the pressure . therefore , a chiral phase transition in this region of low-@xmath3 and high-@xmath2 is always acompanied by a significant jump in the quark number density . so , if the correct degrees of freedom are quarks rather than baryons , the quarkyonic transition is naturally close to the chiral phase transition @xcite . in these kinds of models there is a general tendency that the critical endpoint is found not far from the triple point region . this is because the quarkyonic transition boundary tends to stay along the chiral phase transition where the quark number density jumps discontinuously or increases rapidly . one must , however , bear in mind that the above - mentioned model indications on the critical endpoint are strongly dependent on neglected effects . these include the unknown model parameters and their dependence on @xmath3 and @xmath2 in the njl and qm models , other possibilities of ground states such as the color superconducting phase , inhomogeneous states like the crystalline color superconducting phase and chiral density wave @xcite , an unconventional pattern of chiral symmetry breaking @xcite , etc . of course , in addition to them , the baryon degrees of freedom may change the whole picture completely . let us consider how the quark model results are affected by the baryons . to this end , we must consider how the baryon belongs to the representation of chiral symmetry . there are two possible assignments ; one is just the same as the quark field which is called a naive assignment , and another is the so - called mirror assignment @xcite . the important point is that one can construct a mass term which is chiral invariant in the case of the mirror assignment . this means that the baryons need not be lighter associated with chiral restoration in this case , so there is no jump in the baryon number density across the chiral phase transition if any . thus the baryon number density need not necessarily exhibit a clear indication of either quarkyonic or chiral transitions . on the other hand , if the assignment is naive , the situation becomes more or less similar to what we have seen for the quark model studies at the qualitative level . finally we shall comment on what the chiral model suggests for the triple point . in the pnjl model at small chemical potential ( @xmath110 ) , the behavior of the polyakov loop as a function of @xmath111 and @xmath3 has a deviation from that of the chiral condensate . such a possibility of unlocking of the deconfinement and chiral transitions was already pointed out in the first paper on the pnjl model @xcite . together with the fact that the quark number density has a strong correlation with the chiral condensate in the quark - based model , this observation of separate deconfinement and chiral crossovers may well suggest that there appears a triple point region on the phase diagram . at least within the uncertainty of the model which can easily move the critical endpoint , it seems that the appearance of the triple point is a robust feature of the model output . it is notable that the anomaly matching condition may well imply that the chiral phase transition takes place later than the deconfinement @xcite , but strictly speaking , because of the violation of lorentz symmetry in the presence of matter , the same argument as in vacuum can not be directly applied to constrain the ordering of the phase transitions @xcite . while the arguments that chiral symmetry should be approximately restored in the high baryon density region are strong , the arguments that it is completely restored are less so . it might turn out for example that chiral symmetry remains broken in the quarkyonic phase due to non perturbative effects at the fermi surface . such effects would be proportional to powers of @xmath112 , and would be small but nevertheless non - negligible . presumably these effects would disappear when confinement also disappears . for example , effects at the fermi surface might make the chiral condensate of order @xmath113 . while this would be small compared to the baryon number density , @xmath114 , and might be ignored for many purposes , its magnitude would be parametrically unchanged from its value in the confined phase . although we have very little to say which is strongly compelling about the nature of chiral symmetry breaking and its relation to quarkyonic matter , the questions that arise are of fundamental interest for our understanding of the nature of mass generation in qcd . as such , these issues must eventually be understood in an absolutely compelling and simple way . the possible relation between the chiral and deconfining phase transitions , discussed above , can be more transparent when referring to properties of an effective lagrangian @xcite . consider the interaction of the ( renormalized ) polyakov loop , @xmath115 , which is the trace of the renomalized wilson line , @xmath116 , @xmath117 . the polyakov loop couples to the chiral field , @xmath118 , as @xmath119 this term is chirally invariant . it is not invariant under the global @xmath120 symmetry of the pure glue theory , under which @xmath121 , but this symmetry is broken by the presence of quarks . while the polyakov loop @xmath115 is dimensionless , for the purposes of power counting , let us assume that like ordinary scalar fields , it has dimensions of mass . this implies that the coupling @xmath122 has dimensions of mass . it is the dominant coupling of the polyakov loop to quarks . the observation of ref . @xcite is that the sign of @xmath122 controls how the chiral and deconfining transitions are related . assume that chiral symmetry is broken in the vacuum , so the expectation value of @xmath123 is nonzero . if @xmath122 is positive , @xmath124 is positive , so that this term resists the polyakov loop from developing an expectation value until chiral symmetry is restored . that is , positive @xmath122 links the deconfining and chiral symmetry phase transitions together . conversely , if @xmath122 is negative , the transitions tend to repel one another . clearly a special point occurs when @xmath122 vanishes . at this point , it is natural for the deconfining and chiral phase transitions to split apart from one another . we suggest , then , that the chiral phase transition _ may _ split from the deconfining line at the triple point . this assumes that the triple point also coincides with the critical endpoint for the chiral phase transition @xcite . about the triple point , it is then natural to ask what the next leading term is . there are many such terms . one involves the mass matrix of the chiral fields , @xmath125 , which is proportional to the current quark masses : @xmath126 like @xmath122 , it has dimensions of mass . it is chirally suppressed , however , and so is less important except when @xmath122 is small . assuming that @xmath122 vanishes at the triple point , the coupling @xmath127 dominates in the region where @xmath128 . in this region , the coupling is reversed from the usual expectation : in particular , the coupling to the heavier goldstone bosons , such as kaons , is _ larger _ than that to the lighter goldstone bosons , the pions . the phenomenological implications of this term are interesting for experiment . about the point where @xmath122 vanishes , even though they are heavier , the coupling of kaons to the polyakov loop is larger , by a factor of @xmath129 . this might explain the enhancement of strangeness observed at the triple point . in this work , we have presented an interpretation of the experimental data on particle production obtained in heavy ion collisions form sis up to rhic energies in the context of a new structure of the qcd phase diagram with quarkyonic matter . we have shown that by considering quarkyonic matter , the two regimes of chemical freeze - out with meson and baryon dominance observed phenomenologically can be understood as arising from a triple point where hadronic matter , the quark - gluon plasma , and quarkyonic matter all coexist . this triple point is located where the freeze - out temperature is reaching its limiting value and were different strange to non - strange particle ratios exhibit non - monotonic behavior . we have presented a set of qualitative and semi - quantitative arguments that the observed statistical properties of experimental data are naturally explained when assuming the existence of quarkyonic matter and of a triple point in the qcd phase diagram . we have also discussed in the context of different models the possible role of the chiral symmetry restoration and the interplay between triple and critical points . our findings and interpretation can be justified and/or verified in the near future with more data expected from ultrarelativistic heavy ion collisions . new data are soon to be available from the rhic low energy runs and from the cern na61 experiment which aim at a scan of energy and system size dependence in the vicinity of the triple point discussed here . while these experiments will , most likely , pierce into the triple point region coming from higher energies , where the phase transition is a crossover between hadronic matter and quark plasma , the dedicated exploration of the phase border between the hadronic and the quarkyonic matter will be a task for the future experiments with highly compressed baryonic matter such as cbm@fair darmstadt and nica@jinr dubna . if chemical equilibration will be substantiated with more precise data at those energies our quarkyonic phase boundary argument will gain a dramatic support as a viable explanation for equilibration . we gratefully acknowledge insightful comments from jean - paul blaizot and christof wetterich . the research of d. blaschke is supported by the polish ministry of science and higher education ( mnisw ) under grants no . n n 202 0953 33 and no . n n 202 2318 37 , and by the russian fund for fundamental investigations under grant no . 08 - 02 - 01003-a . the research of r. pisarski and l. mclerran is supported under doe contract no . de - ac02 - 98ch10886 . r. pisarski and k. redlich thank the alexander von humboldt foundation ( avh ) for their support ; k. redlich also thanks the polish ministry of science ( mnisw ) for their support . k. fukushima is supported by japanese mext grant no . 20740134 . the work of c. sasaki was supported in part by the dfg cluster of excellence `` origin and structure of the universe '' . h. j. specht , nucl . a * 698 * , 341 ( 2002 ) . u. w. heinz and m. jacob , arxiv : nucl - th/0002042 . j. stachel , nucl . a * 654 * , 119c ( 1999 ) . i. arsene _ et al . _ [ brahms collaboration ] , nucl . phys . a * 757 * , 1 ( 2005 ) . b. b. back _ et al . _ [ phobos collaboration ] , nucl . phys . a * 757 * , 28 ( 2005 ) . j. adams _ et al . _ [ star collaboration ] , nucl . phys . a * 757 * , 102 ( 2005 ) . k. adcox _ et al . _ [ phenix collaboration ] , nucl . a * 757 * , 184 ( 2005 ) . p. braun - munzinger and j. stachel , nature * 448 * , 302 ( 2007 ) . p. braun - munzinger , j. stachel , j. p. wessels , and n. xu , phys . lett . b * 344 * , 43 ( 1995 ) . j. cleymans and k. redlich , phys . c * 60 * , 054908 ( 1999 ) . u. w. heinz , nucl . a * 661 * , 140 ( 1999 ) . j. letessier and j. rafelski , int . j. mod . e * 9 * , 107 ( 2000 ) . p. braun - munzinger , d. magestro , k. redlich , and j. stachel , phys . b * 518 * , 41 ( 2001 ) . w. florkowski , w. broniowski , and m. michalec , acta phys . b * 33 * , 761 ( 2002 ) . f. becattini , j. manninen , and m. gadzicki , phys . c * 73 * , 044905 ( 2006 ) . f. becattini , et al . c * 64 * , 024901 ( 2001 ) . j. cleymans , h. oeschler , k. redlich , and s. wheaton , eur . j. a * 29 * , 119 ( 2006 ) a. andronic , p. braun - munzinger , and j. stachel , nucl . phys . a * 772 * , 167 ( 2006 ) . braun - munzinger , k. redlich , and j. stachel , in quark - gluon plasma 3 , eds . r.c . hwa and x.n . wang , ( world scientific publishing , 2004 ) [ nucl - th/0304013 ] . j. cleymans and k. redlich , phys . lett . * 81 * , 5284 ( 1998 ) . c. detar and u. m. heller , eur . j. a * 41 * , 405 ( 2009 ) . o. kaczmarek , f. karsch , p. petreczky , and f. zantow , ; p. petreczky and k. petrov , ; o. kaczmarek , f. karsch , f. zantow , and p. petreczky , [ erratum- ] ; m. doring , s. ejiri , o. kaczmarek , f. karsch , and e. laermann , ; y. aoki , z. fodor , s. d. katz , and k. k. szabo , ; k. hubner , f. karsch , o. kaczmarek , and o. vogt , ; s. gupta , k. hbner , and o. kaczmarek , ; m. cheng _ et al . _ , ; a. bazavov _ et al . v.d . toneev and a.s.parvan , j. phys . g * 31 * , 583 ( 2005 ) . d. adamova _ et al . _ [ ceres collaboration ] , phys . lett . * 90 * , 022301 ( 2003 ) . h. oeschler , et al . , j. phys . g * 32 * , s223 ( 2006 ) . m. gazdzicki and m. i. gorenstein , acta phys . b * 30 * , 2705 ( 1999 ) . m. a. stephanov , k. rajagopal , and e. v. shuryak , phys . * 81 * , 4816 ( 1998 ) ; phys . d * 60 * , 114028 ( 1999 ) . l. mclerran and r. d. pisarski , nucl . a * 796 * , 83 ( 2007 ) . y. hidaka , l. d. mclerran , and r. d. pisarski , nucl . a * 808 * , 117 ( 2008 ) . l. mclerran , k. redlich , and c. sasaki , phys . a * 824 * , 86 ( 2009 ) . k. fukushima , phys . d * 77 * , 114028 ( 2008 ) [ erratum - ibid . d * 78 * , 039902 ( 2008 ) ] . l. y. glozman and r. f. wagenbrunn , phys . d * 77 * , 054027 ( 2008 ) . c. hhne , nucl . a * 830 * 369c ( 2009 ) . r. d. pisarski , ; a. dumitru and r. d. pisarski , ; ; a. dumitru , y. hatta , j. lenaghan , k. orginos , and r. d. pisarski , ; a. dumitru , j. lenaghan , and r. d. pisarski , ; r. d. pisarski , ; y. hidaka and r. d. pisarski ; . k. fukushima , ; . y. hatta and k. fukushima , ; e. megias , e. ruiz arriola , and l. l. salcedo , ; ; ; ; [ arxiv:0903.1060 ] ; a. dumitru , r. d. pisarski , and d. zschiesche , ; c. ratti , m. a. thaler , and w. weise , ; s. k. ghosh , t. k. mukherjee , m. g. mustafa , and r. ray , ; h. hansen , w. m. alberico , a. beraudo , a. molinari , m. nardi , and c. ratti , ; s. mukherjee , m. g. mustafa , and r. ray , ; s. roessner , c. ratti , and w. weise , ; k. fukushima and y. hidaka , ; c. sasaki , b. friman , and k. redlich , ; s. k. ghosh , t. k. mukherjee , m. g. mustafa , and r. ray , ; d. blaschke , m. buballa , a. e. radzhabov , and m. k. volkov , yad . * 71 * , 2012 ( 2008 ) ; s. roessner , t. hell , c. ratti , and w. weise , nucl . a * 814 * , 118 ( 2008 ) ; p. costa , c. a. de sousa , m. c. ruivo , and h. hansen europhys . lett . * 86 * 31001 ( 2009 ) ; y. sakai , k. kashiwa , h. kouno , and m. yahiro , ; d. gomez dumm , d. b. blaschke , a. g. grunfeld , and n. n. scoccola , ; p. costa , m. c. ruivo , c. a. de sousa , h. hansen , and w. m. alberico ; k. dusling , c. ratti , and i. zahed , [ arxiv:0807.2879 ] ; t. hell , s. roessner , m. cristoforetti , and w. weise , ; k. y. kim , s. j. sin and i. zahed , arxiv : hep - th/0608046 ; n. horigome and y. tanii , jhep * 0701 * , 072 ( 2007 ) ; a. parnachev and d. a. sahakyan , nucl . phys . b * 768 * , 177 ( 2007 ) ; s. nakamura , y. seo , s. j. sin and k. p. yogendran , j. korean phys . soc . * 52 * , 1734 ( 2008 ) ; s. kobayashi , d. mateos , s. matsuura , r. c. myers and r. m. thomson , jhep * 0702 * , 016 ( 2007 ) ; h. hata , t. sakai , s. sugimoto and s. yamato , prog . phys . 117 ( 2007 ) , 1157 ; d. yamada , jhep * 0810 * , 020 ( 2008 ) ; o. bergman , g. lifschytz and m. lippert , jhep * 0711 * , 056 ( 2007 ) ; m. rozali , h. h. shieh , m. van raamsdonk and j. wu , jhep * 0801 * , 053 ( 2008 ) ; k. ghoroku , m. ishihara and a. nakamura , phys . rev . d * 76 * , 124006 ( 2007 ) ; a. parnachev , jhep * 0802 * , 062 ( 2008 ) ; a. karch and a. obannon , jhep * 0711 * , 074 ( 2007 ) ; d. mateos , s. matsuura , r. c. myers and r. m. thomson , jhep * 0711 * , 085 ( 2007 ) ; m. rozali , h. h. shieh , m. van raamsdonk and j. wu , jhep * 0801 * , 053 ( 2008 ) ; k. y. kim , s. j. sin and i. zahed , jhep * 0809 * , 001 ( 2008 ) ; o. bergman , g. lifschytz and m. lippert , phys . d * 79 * , 105024 ( 2009 ) ; k. hashimoto , prog . * 121 * , 241 ( 2009 ) ; k. y. kim and i. zahed , jhep * 0812 * , 075 ( 2008 ) ; m. kulaxizi and a. parnachev , nucl . phys . b * 815 * , 125 ( 2009 ) . m. teper , pos * lattice2008 * , 022 ( 2008 ) . p. braun - munzinger , j. stachel , and c. wetterich , phys . b * 596 * , 61 ( 2004 ) . j. knoll , nucl . a * 821 * 235 ( 2009 ) . g. veneziano , nuovo cim . a * 57 * , 190 ( 1968 ) . p. g. o. freund and j. l. rosner , phys . * 68 * , 765 ( 1992 ) ; w. broniowski , w. florkowski and l. y. glozman , phys . d * 70 * , 117503 ( 2004 ) . w. broniowski and w. florkowski , phys . b * 490 * , 223 ( 2000 ) ; see also j. stachel , talk at the sqm2009 , buzios , brazil , sep . 2009 . r. s. hayano and t. hatsuda , arxiv:0812.1702 [ nucl - ex ] . n. kaiser , p. de homont and w. weise , phys . c * 77 * , 025204 ( 2008 ) ; n. kaiser and w. weise , phys . lett . b * 671 * , 25 ( 2009 ) . k. suzuki _ et al . _ , phys . lett . * 92 * , 072302 ( 2004 ) . b .- j . schaefer , j. pawlowski and j. wambach , phys . d * 76 * ( 2007 ) 074023 ; b .- j . schaefer , m. wagner and j. wambach , arxiv:0910.5628 . a. barducci , r. casalbuoni , s. de curtis , r. gatto , and g. pettini , phys . b * 231 * , 463 ( 1989 ) ; a. barducci , r. casalbuoni , s. de curtis , r. gatto , and g. pettini , phys . d * 41 * , 1610 ( 1990 ) ; y. taniguchi and y. yoshida , phys . d * 55 * , 2283 ( 1997 ) ; m. harada and a. shibata , phys . d * 59 * , 014010 ( 1999 ) ; c. d. roberts and s. m. schmidt , prog . part . nucl . * 45 * , s1 ( 2000 ) ; r. alkofer and l. von smekal , phys . rept . * 353 * , 281 ( 2001 ) . e. nakano and t. tatsumi , phys . d * 71 * , 114006 ( 2005 ) . d. nickel , phys . d * 80 * , 074025 ( 2009 ) . c. e. detar and t. kunihiro , phys . d * 39 * , 2805 ( 1989 ) . g. baym , in _ quark matter formation and heavy ion collisions , proc . bielefeld workshop _ , edited by m. jacob and h. satz ( world scientific , singapore , 1982 ) . h. itoyama and a. h. mueller , nucl . b * 218 * , 349 ( 1983 ) ; r. d. pisarski , phys . lett . * 76 * , 3084 ( 1996 ) ; r. d. pisarski , t. l. trueman and m. h. g. tytgat , phys . d * 56 * , 7077 ( 1997 ) . a. mocsy , f. sannino and k. tuominen , ; ; f. sannino and k. tuominen , .

the human brain is organized in functional modules . such an organization presents a basic conundrum : modules ought to be sufficiently independent to guarantee functional specialization and sufficiently connected to bind multiple processors for efficient information transfer . it is commonly accepted that small - world architecture of short lengths and large local clustering may solve this problem . however , there is intrinsic tension between shortcuts generating small - worlds and the persistence of modularity ; a global property unrelated to local clustering . here , we present a possible solution to this puzzle . we first show that a modified percolation theory can define a set of hierarchically organized modules made of strong links in functional brain networks . these modules are `` large - world '' self - similar structures and , therefore , are far from being small - world . however , incorporating weaker ties to the network converts it into a small - world preserving an underlying backbone of well - defined modules . remarkably , weak ties are precisely organized as predicted by theory maximizing information transfer with minimal wiring cost . this trade - off architecture is reminiscent of the `` strength of weak ties '' crucial concept of social networks . such a design suggests a natural solution to the paradox of efficient information flow in the highly modular structure of the brain .

1102.0604

one of the main findings in neuroscience is the modular organization of the brain which in turn implies the parallel nature of brain computations @xcite . for example , in the visual modality , more than thirty visual areas analyze simultaneously distinct features of the visual scene : motion , color , orientation , space , form , luminance and contrast among others @xcite . these features , as well as information from different sensory modalities , have to be integrated , as one of the main aspects of perception is its unitary nature @xcite . this leads to a basic conundrum of brain networks : modular processors have to be sufficiently isolated to achieve independent computations , but also globally connected to be integrated in coherent functions @xcite . a current view is that small - world networks provide a solution to this puzzle since they combine high local clustering and short path length @xcite . this view has been fueled by the systematic finding of small - world topology in a wide range of human brain networks derived from structural @xcite , functional @xcite , and diffusion tensor mri @xcite . small - world topology has also been identified at the cellular - network scale in functional cortical neuronal circuits in mammals @xcite and even in the nervous system of the nematode _ caenorhabditis elegans _ @xcite . moreover , small - world property seems to be relevant for brain function since it is affected by disease @xcite , normal aging and by pharmacological blockade of dopamine neurotransmission @xcite . while brain networks show small - world properties , several experimental studies have also shown that they are hierarchical , fractal and highly modular @xcite . as there is an intrinsic tension between modular and small - world organization , the main aim of this study is to reconcile these ubiquitous and seemingly contradictory topological properties . indeed , traditional models of small - world networks can not fully capture the coexistence of highly modular structure with broad global integration . first , clustering is a purely local quantity which can be assessed inspecting the immediate neighborhood of a node @xcite . on the contrary , modularity is a global property of the network , determined by the existence of strongly connected groups of nodes that are only loosely connected to the rest of the network @xcite . in fact , it is easy to construct modular and unclustered networks or , reciprocally , clustered networks without modules . second , the short distances of a small - world may be incompatible with strong modularity which typically presents the properties of a `` large - world '' @xcite characterized by long distances which effectively suppress diffusion and free flow in the system @xcite . while a clustered network preserves its clustering coefficient when a small fraction of shortcuts are added ( converting it into a small - world ) @xcite , the persistence of modules is not equally robust . as we show below , shrinking the network diameter may quickly destroy the modules . hence , the concept of small - world may not be entirely sufficient to explain the modular and integration features of functional brain networks on its own . we propose that a solution to modularity and broad integration can be achieved by a network in which strong links form large - world fractal modules , in agreement with @xcite , which are short - cutted by weaker links establishing a small - word network . a modified percolation theory @xcite can identify a sequence of critical values of connectivity thresholds forming a hierarchy of modules which progressively merge together . this proposal is inspired by a fundamental notion of sociology termed by granovetter as `` the strength of weak ties '' @xcite . according to this theory , strong ties ( close friends ) clump together forming modules . an acquaintance ( weak tie ) becomes a crucial bridge ( a shortcut ) between the two densely knit clumps ( modules ) of close friends @xcite . interestingly , this theme also emerges in theoretical models of large - scale cognitive architecture . several theories suggest integration mechanisms based on dynamic binding @xcite or on a workspace system @xcite . for instance , the workspace model @xcite proposes that a flexible routing system with dynamic and comparably weaker connections transiently connects modules with very strong connections carved by long - term learning mechanisms . we capitalize on a well - known dual - task paradigm , the psychological refractory period @xcite . a total of 16 subjects responded with the right hand to a visual stimulus and with the left hand to an auditory stimulus ( see si appendix ) . the temporal gap between the auditory and visual stimuli varied in four stimulus onset asynchrony , soa= 0 , 300 , 900 and 1200 ms . the sequence of activated regions which unfolds during the execution of the task has been reported in a previous study @xcite . the network analysis relies on the time - resolved bold - fmri response based on the phase signal obtained for each voxel of data @xcite . we first compute the phase of the bold signal for each voxel with methods developed previously @xcite . for each subject and each soa task , we obtain the phase signal of the @xmath0-th voxel of activity , @xmath1 , over @xmath2 trials performed for a particular soa value and subject . we use these signals to construct the network topology of brain voxels based on the equal - time cross - correlation matrix , @xmath3 , where a network link indicates a high cross - correlation in the phase activity of the two voxels ( see si appendix ) . the accuracy of the calculated @xmath3 values was estimated through a boot strapping analysis . the 95% confidence interval becomes more narrow for higher @xmath3 values , e.g. , for @xmath4 it is ( 0.9744 , 0.9760 ) . the corresponding standard deviation is of the order of 0.003 . thus , we typically distinguish between values that differ by 0.005 ( see si appendix and fig . s1 ) . to construct the network , we link two voxels if their cross - correlation @xmath3 is larger than a predefined threshold value @xmath5 @xcite . the resulting network for a given @xmath5 is a representation of functional relations among voxels for a specific subject and soa . we obtain 64 cross - correlation networks resulting from the four soa values presented to the 16 subjects . graph analyses of brain correlations relies on a threshold @xcite which is problematic since small - world like properties are sensitive to even a small proportion of variation in the connections . the present analysis may be seen as an attempt to solve this problem . the thresholding procedure can be naturally mapped to a percolation process ( defined in the @xmath6 space of interactions @xmath3 ) ; a model to describe geometrical phase transitions of connected clusters in random graphs , see chapters 2 and 3 in @xcite and @xcite . in general , the size of the largest component of connected nodes in a percolation process remains very small for large @xmath5 . the crucial concept is that the largest connected component increases rapidly through a critical phase transition at @xmath7 , in which a single incipient cluster dominates and spans the system @xcite . a unique connected component is expected to appear if the links in the network are occupied at random without correlations . however , when we apply the percolation analysis to the functional brain network , a more complex picture emerges revealing a hierarchy of clusters arising from the non - trivial correlations in brain activity . for each participant , we calculate the size of the largest connected component as a function of @xmath5 . we find that the largest cluster size increases progressively with a series of sharp jumps ( fig . [ perco]a , soa=900 ms , all participants , other stimuli are similar ) . this suggests a multiplicity of percolation transitions where percolating modules subsequently merge as @xmath5 decreases rather than following the typical uncorrelated percolation process with a single spanning cluster . each of these jumps defines a percolation transition focused on groups of nodes which are highly correlated , constituting well - defined modules . figure [ perco]b shows the detailed behavior of the jumps in a typical individual ( subject labeled # 1 in our dataset @xcite , soa=900 ms ) . at high values of @xmath5 , three large clusters are formed localized to the medial occipital cortex ( red ) , the lateral occipital cortex ( orange ) and the anterior cingulate ( green ) . at a lower @xmath8 , the orange and red clusters merge as revealed by the first jump in the percolation picture . as @xmath5 continues to decrease this mechanism of cluster formation and absorption repeats , defining a hierarchical process as depicted in the top panel of fig . [ perco]b . this analysis further reveals the existence of `` stubborn '' clusters . for instance , the anterior cingulate cluster ( green ) , known to be involved in cognitive control @xcite and which hence can not commit to a specific functional module , remains detached from the other clusters down to low @xmath5 values . even at the lower values of @xmath5 , when a massive region of the cortex including motor , visual and auditory regions has formed a single incipient cluster ( red , @xmath9 ) , two new clusters emerge ; one involving subcortical structures including the thalamus and striatum ( cyan ) and the other involving the left frontal cortex ( purple ) . this mechanism reveals the iteration of a process by which modules form at a given @xmath5 value and merged by comparably weaker links . this process is recursive . the weak links of a given transition become the strong links of the next transition , in a hierarchical fashion . here , we focus our analysis on the first jump in the size of the largest connected component , for instance , @xmath10 in fig . [ perco]b . we consider the three largest modules at @xmath7 with at least 1,000 voxels each . this analysis results in a total of 192 modules among all participants and stimuli which are pooled together for the next study . an example of an identified module in the medial occipital cortex of subject # 1 and soa=900 ms is shown in fig . [ perco]c in the network representation and in fig . [ perco]d in real space . the topography of the modules reflects coherent patterns across the subjects and stimuli as analyzed in si appendix ( see fig . s2 ) . to determine the structure of the modules we investigate the scaling of the `` mass '' of each module ( the total number of voxels in the module , @xmath11 ) as a function of three length - scales defined for each module : _ ( i ) _ the maximum path length , @xmath12 , _ ( ii ) _ the average path length between two nodes , @xmath13 , and _ ( iii ) _ the maximum euclidean distance among any two nodes in the cluster , @xmath14 . the path length , @xmath15 , is the distance in network space defined as the number of links along the shortest path between two nodes . the maximum @xmath12 is the largest shortest path in the network . figure [ fractal]a indicates power - law scaling for these quantities @xcite . for instance : @xmath16 defines the euclidean hausdorff fractal dimension , @xmath17 . the scaling with @xmath12 and @xmath18 is consistent with eq . ( [ df ] ) as seen in fig . [ fractal]a . the exponent @xmath19 quantifies how densely the volume of the brain is covered by a specific module . next , we investigate the network properties of each module , applying renormalization group ( rg ) analysis for complex networks @xcite . this technique allows one to observe the network at different scales transforming it into successively simpler copies of itself , which can be used to detect characteristics which are difficult to identify at a specific scale of observation . we use this technique to characterize sub - modular structure within each brain module @xcite . we consider each module identified at @xmath7 separately . we then tile it with the minimum number of boxes or sub - modules , @xmath20 , of a given box diameter @xmath21 @xcite , i.e. , every pair of nodes in a box has shortest path length smaller than @xmath21 . notice that the calculations are performed in network space , where path lengths are defined across the network links without the need for an embedding dimension . covering the network with minimal @xmath20 sub - modules represents an optimization problem which is solved using standard box - covering algorithms , such as the maximum excluded mass burning algorithm , memb , which was introduced in @xcite to describe the self similarity of complex networks ranging from the www , biological and technical networks ( see si appendix and fig . [ fractal]b describing memb ; the code can be downloaded from @xcite ) . the requirement to minimize the number of boxes is important since the resulting boxes are characterized by the proximity between all their nodes and minimization of the links connecting the boxes @xcite . thus , the box - covering algorithm detects boxes / submodules that also tend to maximize modularity . the repetitive application of box - covering at different values of @xmath21 is a rg transformation @xcite that yields a different partition of the brain modules in submodules of varying sizes ( fig . [ fractal]b ) . figure [ fractal]c shows the scaling of @xmath20 versus @xmath21 averaged over all the modules for all individuals and stimuli . this property is quantified in the power - law relation @xcite : @xmath22 where @xmath23 is the box fractal dimension @xcite which characterizes the self - similarity between different scales of the module where smaller - scale boxes behave in a similar way as the original network . the resulting @xmath24 averaged over all the modules is @xmath25 . the rg analysis reveals that the module topology does not have many redundant links and it represents the quantitative statement that the brain modules are `` large - worlds '' . however this analysis is not sufficient to precisely characterize the topology of the modules . for example , a two - dimensional complex network architecture and a simple two - dimensional lattice are compatible with the scaling analysis and the value of the exponents described in the previous section . to identify the network architecture of the modules we follow established analysis @xcite based on the study of the degree distribution of the modules , @xmath26 , and the degree - degree correlation @xmath27 . the form of @xmath26 distinguishes between a euclidean lattice ( delta function ) , an erdos - renyi network ( poisson ) @xcite , or a scale - free network ( power - law ) @xcite . we find that the brain modules have a broad degree distribution @xcite with an approximate power - law @xmath28 . the statistical analysis provides strong evidence for a power law form and rules out exponential decay ( see si appendix ) . in fig . s4 we present a number of @xmath26 curves for different modules , along with their best fittings . in the si appendix we describe the calculation method that takes into account all the clusters and finally yields an average exponent @xmath29 . an ` average ' curve for the distribution is shown in fig . [ fractal]d , where the exponent @xmath30 is not a direct fit to this curve , but instead represents the result of the accurate calculation . this result indicates that the modules have a scale - free fractal structure , different from a simple two - dimensional lattice , where @xmath26 should be a narrow function . the embedding of scale - free networks in a finite - dimension real space constitutes a problem which has attracted recent attention @xcite . scale - free networks may arise from a 2-dimensional lattice with added dense connectivity locally , where the weights and connectivity are inversely proportional to the euclidean distance on the lattice . to investigate this possibility we study the correlation function of the phases of the voxels as a function of euclidean distance in real space : @xmath31 versus @xmath32 . this function can be interpreted as the correlation between two spins with orientation determined by the phase @xmath33 of the voxel at location @xmath34 ( average is over all pairs at distance @xmath35 ) . we find ( see si appendix and fig . s3 ) that @xmath36 decays algebraically with distance . thus , our results indicate that modules are scale - free networks which can be embedded in a lattice with an added long - range connectivity . how can fractal modularity emerge in light of the scale - free property , which is usually associated with small - worlds @xcite ? in a previous study @xcite , we introduced a model to account for the simultaneous emergence of scale - free , fractality and modularity in real networks by a multiplicative process in the growth of the number of links , nodes and distances in the network . the dynamic follows the inverse of the rg transformation @xcite where the hubs acquire new connections by linking preferentially with less connected nodes rather than other hubs . this kind of `` repulsion between hubs '' @xcite creates a disassortative structure with hubs spreading uniformly in the network and not crumpling in the core as in scale - free models @xcite . hubs are buried deep into the modules , while low degree nodes are the intermodule connectors @xcite . a signature of such mechanism can be found by following hubs degree during the renormalization procedure . at scale @xmath21 , the degree of a hub @xmath37 changes to the degree of its box @xmath38 , through the relation @xmath39 . the dependence of the scaling factor @xmath40 on @xmath21 defines the renormalized degree exponent @xmath41 by @xmath42 @xcite . scaling theory defines precise relations between the exponents for fractal networks @xcite , through @xmath43 . for the case of brain modules analyzed here ( fig . [ fractal]e ) , we find @xmath44 . using the values of @xmath24 and @xmath41 for the brain clusters , the prediction is @xmath45 , which is close to the calculated value of @xmath29 from fig . [ fractal]d . the previous analysis reveals the mechanism of formation of a scale - free network , but it does not assure a fractal topology . fractality can be determined from the study of the degree - degree correlation through the distribution , @xmath27 to find a link between nodes with @xmath46 degree . this correlation characterizes the hub - hub repulsion through scaling exponents @xmath47 and @xmath48 ( see si appendix and fig . s5 ) @xcite . in a fractal , they satisfy @xmath49 . a direct measurement of these exponents yields @xmath50 and @xmath51 ( fig . s5 ) . using the measured values of @xmath47 and @xmath41 , we predict @xmath52 , which is close to the observed exponent . taken together , these results indicate a scale - free fractal morphology of brain modules . such structure is in agreement with previous results of the anatomical connectivity of the brain @xcite and functional brain networks @xcite . standard modularity decomposition methods @xcite based on maximization of the modularity factor @xmath53 as defined in @xcite are particularly suitable to uncover the submodular structure . for example , the girvan - newman method @xcite yields a value of @xmath54 for the brain clusters , indicating a strong modular substructure . the box covering algorithm benefits from detecting submodules ( the boxes ) at different scales . then , we can study the hierarchical character of modularity @xcite , and detect whether modularity is a feature of the network that remains scale - invariant ( see si appendix and fig . s6 for a comparison of the submodular structure obtained using girvan - newman and box covering ) . the minimization of @xmath20 guarantees a network partition with the largest number of intramodule links and the fewest intermodule links . therefore , the box covering algorithm maximizes the following modularity factor @xcite : @xmath55 which is a variation of the modularity factor , @xmath53 , defined in @xcite . here , @xmath56 and @xmath57 represent the intra and intermodular links in a submodule @xmath0 , respectively . large values of @xmath58 ( i.e. @xmath59 ) correspond to high modularity @xcite . we make the whole modularization method available at @xcite . figure [ fractal]f shows the scaling of @xmath60 averaged over all modules at percolation revealing a monotonic increase with a lack of a characteristic value of @xmath21 . indeed , the data can be fitted with a power - law form @xcite : @xmath61 which is detected through the modularity exponent , @xmath62 . we study the networks for all the subjects and stimuli and find @xmath63 ( fig . [ fractal]f ) . the lack of a characteristic length - scale expressed in eq . ( [ modular ] ) implies that submodules are organized within larger modules such that the inter - connections between those submodules repeat the basic modular character of the entire brain network . the value of @xmath62 reveals a considerable modularity in the system as it is visually apparent in the sample of fig . [ box]a , left panel , where different colors identify the submodules of size @xmath64 in a typical fractal module . for comparison , a randomly rewired network ( fig . [ box]a , right and central panels ) shows no modularity and has @xmath65 . scaling analysis indicates that @xmath62 is related to @xmath66 , which defines the outbound exponent @xmath67 characterizing the number of intermodular links for a submodule @xcite ( @xmath67 is related to the rent exponent in integrated circuits @xcite ) . from eq . ( [ modular ] ) , we find : @xmath68 , which indicates that the strongest possible modular structure has @xmath69 ( @xmath70 ) @xcite . such a high modularity induces very slow diffusive processes ( subdiffusion ) for a random walk in the network @xcite . comparing eq . ( [ modular ] ) with ( [ db ] ) , we find @xmath70 , which quantifies the maximum degree of modularity in the brain modules . an important consequence of eqs . ( [ df ] ) and ( [ db ] ) is that the network determined by the strong links above the first @xmath7-jump lacks the logarithmic scaling characteristic of small - worlds and random networks @xcite : @xmath71 a fractal network poses much larger distances than those appearing in small - worlds @xcite : a distance @xmath72 observed in fig . [ fractal]a ( red curve ) would require an enormous small - world network @xmath73 , rather than @xmath74 , as observed for fractal networks in fig . [ fractal]a . the structural differences between a modular fractal network and a small - world ( and a random network ) are starkly revealed in the panels of fig . we rewire the fractal module on the left panel by randomly reconnecting a fraction @xmath75 of the links while keeping the degree of each node intact @xcite . figure [ box]b quantifies the transition from fractal ( @xmath76 ) to small - world ( @xmath77 ) and eventually to random networks ( @xmath78 ) , illustrated in fig . [ box]a : we plot @xmath79 , the clustering coefficient @xmath80 and @xmath81 for a typical @xmath64 as we rewire @xmath75 links in the network . as we create a tiny fraction @xmath82 of short - cuts , the topology turns into a collapsed network with no trace of modularity left , while @xmath83 still remains quite high ( fig . [ box]b ) . the rewired networks present the exponential behavior of small - worlds @xcite , and also random networks as @xmath84 increases , obtained from eq . ( [ sw1 ] ) : @xmath85 where @xmath11 is averaged over all the modules ( fig . [ box]c ) . the characteristic size is very small and progressively shrinks to @xmath86 when @xmath87 . the hallmark of small - worlds and random networks , exponential scaling eq . ( [ small ] ) , is incompatible with the hallmark of fractal large - worlds , power - law scaling eq . ( [ db ] ) . similarly , while we find a broad domain where short network distances coexist with high clustering forming a small - world behavior , modularity does not show such a robust behavior to the addition of shortcuts . figure [ box]b suggests that modularity and small - world can not coexist at the same level of connectivity strength . next , we set out to investigate how the small - world emerges . when we extend the percolation analysis lowering further the threshold @xmath5 below @xmath7 , weaker ties are incorporated to the network connecting the self - similar modules through short - cuts . a typical scenario is depicted in fig . [ optimal]a showing the three largest percolation modules identified just before the first percolation jump in the subject # 1 shown in fig . [ perco]b at @xmath88 . for this connectivity strength , the modules are separated and show submodular fractal structure indicated in the colored boxes obtained with box covering . when we lower the threshold to @xmath89 , fig . [ optimal]b , the modules are now connected and a global incipient component starts to appears . a second global percolation - like transition appears in the system when the mass of the largest component occupies half of the activated area ( see e.g. fig . [ perco ] ) . for different individuals , global percolation occurs in the interval @xmath90 $ ] as indicated in the inset of fig . [ perco]a . our goal is to investigate whether the weak links shortcut the network in an optimal manner . when the cumulative probability distribution to find a euclidean distance between two connected nodes , @xmath91 , larger than @xmath35 follows a power - law : @xmath92 statistical physics makes precise predictions about optimization schemes for global function as a function of the short - cut exponent @xmath93 and @xmath19 @xcite . specifically , there are three critical values for @xmath93 as shown schematically in fig . [ optimal]c . if @xmath93 is too large then shortcuts will not be sufficiently long and the network will behave as fractal , equal to the underlying structure . below a critical value determined by @xmath94 @xcite , shortcuts are sufficient to convert the network in a small world . within this regime there are two significant optimization values : _ ( i ) _ wiring cost minimization with full routing information . this considers a network of dimension @xmath19 , over which short - cuts are added to optimize communication , with a wiring cost constraint proportional to the total shortcut length . it is also assumed that coordinates of the network are known , i.e. it is the shortest path that it is being minimized . under these circumstances , the optimal distribution of shortcuts is @xmath95 @xcite . this precise scaling is found in the us airport network @xcite where a cost limitation applies to maximize profits . _ ( ii ) _ decentralized greedy searches with only local information . this corresponds to the classic milgram s `` small - world experiment '' of decentralized search in social networks @xcite , where a person has knowledge of local links and of the final destination but not of the intermediate routes . under these circumstances , which also apply to routing packets in the internet , the problem corresponds to a greedy search , rather than to optimization of the minimal path . the optimal relation for greedy routing is @xmath96 @xcite . hence , the analysis of @xmath97 provides information both on the topology of the resulting network and on which transport procedure is optimized . this distribution reveals power - law behavior eq . ( [ alpha ] ) with @xmath98 when averaged over the modules below @xmath7 ( fig . [ optimal]d ) . given the value obtained in eq . ( [ df ] ) , @xmath99 , this implies that the network composed of strong and weak links is small - world ( @xmath100 ) @xcite and optimizes wiring cost with full knowledge of routing information ( @xmath101 ) @xcite . the existence of modular organization which become small - world when short - cut by weaker ties is reminiscent of the structure found to bind dissimilar communities in social networks . granovetter s work in social sciences @xcite proposes the existence of weak ties to cohese well - defined social groups into a large - scale social structure . the observation of such an organization in brain networks suggests that it may be a ubiquitous natural solution to the puzzle of information flow in highly modular structures . over the last decades , wire length minimization arguments have been used successfully to explain the architectural organization of brain circuitry @xcite . minimizing wire length is in fact of paramount importance , since about 60% of the cortical volume is taken up by wire ( axons and dendrites ) @xcite . this turns out to optimize conduction rate , posing a strict packing limitation of the amount of wire in cortical circuits @xcite . our finding of a distribution of weak links which minimizes wiring cost is hence in line with a previous literature , consistently showing that neural circuit design is under pressure to minimize wiring length . however , some important nuances of the specific optimization procedure ought to be considered . first , we specifically showed that at the mesoscopic scale , short - cut distribution optimizes wiring cost while maintaining network proximity . this is consistent with the organization of large - scale neural networks in which total wiring can in fact be decreased by about 32% ( in 95 primate cortical areas ) and up to 48% in the global neuronal network of the nematode caenorhabditis elegans @xcite . this extra wiring cost comes from long - range connections which achieve network benefits of shortening the distance between pairs of nodes @xcite . our results are in agreement with this observation , suggesting that simultaneous optimization of network properties and wiring cost might be a relevant principle of brain architecture . in simple words , this topology does not minimize the total wire per - se , simply to connect all the nodes ; instead it minimizes the amount of wire required to achieve the goal of shrinking the network to a small - world . a second intriguing aspect of our results , which is not usually highlighted , is that this minimization assumes that broadcasting and routing information are known to each node . how this may be achieved what aspects of the neural code convey its own routing information remains an open question in neuroscience . bold fmri is an indirect measure of brain activity which relies on multiple vascular and biophysical factors which couple the neural response to the haemodynamic signal @xcite . even if in fmri research it is always assumed that haemodynamic signals reflect metabolic demand generated by local neuronal activity , recent studies have shown reliable haemodynamic signals that entrains to task structure independently of standard neural predictors of haemodynamics @xcite . hence , our results , as any other fmri analysis , have to be taken cautiously and may partly reflect the underlying structure of vascular motives . specifically , the human cortical vascular system has a large number of arterial anastomoses which show a seemingly looking fractal structure in the mm to cm range @xcite . precise measurements of fractality have been reported at the micrometer scales in volumes of the order of a few mm@xmath102 @xcite , which corresponds to approximately a voxel volume , where branching structure of microcapilarities then generates fractals . hence , it is possible that the fractal organization of brain modules is inherited from the vascular system itself . although we can not readily test the influence of the vascular system at a large scale , it is still possible to address this concern at a microscopic scale , by discarding neighboring correlations . neighboring voxels are expected to carry some shared signal due to spatial autocorrelations from the microvascular network . to assure that our results do not rely on neighbouring correlations which might be particularly spurious , we coarse - grained the original fmri signal by doubling the lattice spacing , reducing the number of voxels by a factor of 8 and repeat the calculations . the results are consistent with the percolation picture of fractal modules , albeit with an expected lower @xmath7 . such a renormalized @xmath7 is expected from renormalization theory to change under coarse - graining , while the main results on long - range links , such as the value of the exponents , are insensitive to this type of coarse - graining . we also investigate whether the map of fractal dimension @xmath24 reflects a meaningful organization based on known facts of functional properties of the cortex and the specific task which subjects are performing . we found a topographical organization of fractality in the human brain ( fig . the right portion of the anterior cingulate , sma and the right ppc regions involved in routing of information and cognitive control @xcite , which are expected to have a more complex functional organization , are the clusters with higher fractal dimension . the left - right asymmetry is interesting since , in this specific task , the left hand response is queued for a few hundred millisencods and has to be temporally connected to working memory and inhibitory circuits . while not fully conclusive , this analysis suggests a functional role of the network architectures described here . another similar concern is that the recovered brain modules may be a manifestation of the fractal structure of the underlying three - dimensional vortex grid or of the cortex . however , since the dimensions of the grid ( @xmath103 ) and of the cortex ( @xmath104)@xcite are both sufficiently different from 1.9 and the connectivity distribution of the modules is much broader than the typical euclidean fractal cortex ( which should be narrow around @xmath105 ) or a 3d lattice ( @xmath106 ) , we may safely assume that these objects have their own structure . moreover , we also observed modules with similar fractal dimension in subcortical structures suggesting that these results do not simply reflect anatomical properties of the cortical mantle . a hierarchical modular organization of the brain composed of modules within modules has been invoked in @xcite to describe the brain structure . the present results support these previous findings , while , at the same time , provide a new view by integrating the results with the ( non - critical ) properties of small - worlds and placing self - similarity in the framework of scaling theory , universality and renormalization groups @xcite . in this framework , brain modules are characterized by a set of novel scaling exponents , the septuplet : @xmath107 , and the scaling relations @xmath108 , relating fractality with modularity , @xmath101 , relating global integration with modularity , @xmath43 , relating scale - free with fractality , and @xmath109 , relating degree correlations with fractality . one advantage of this formalism is that the different brain topologies can be classified into universality classes under rg @xcite according to the septuplet @xmath110 . universality applies to the critical exponents but not to quantities like @xmath111 which are sensitive to the microscopic details of the different experimental situations @xcite . in this framework , ( non - critical ) small - worlds are obtained in the limit @xmath112 . a path for future research will be to test the universality of the septuplet of exponents under different activities covering other areas of the brain , e.g. , the resting - state correlation structure @xcite . in conclusion , we propose a formal solution to the problem of information transfer in the highly modular structure of the brain . the answer is inspired by a classic finding in sociology : the strength of weak ties @xcite . the present work provides a general insight into the physical mechanisms of network information processing at large . it builds up on an example of considerable relevance to natural science , the organization of the brain , to establish a concrete solution to a broad problem in network science . the results can be readily applied to other systems where the coexistence of modular specialization and global integration is crucial ranging from metabolic , protein and genetic networks to social networks and the internet . lkg and ham thank the nsf-0827508 emerging frontiers program for financial support . ms is supported by a human frontiers science program fellowship . we thank d. bansal , s. dehaene , s. havlin , and h.d . rozenfeld for valuable discussions . dehaene s , naccache l ( 2001 ) towards a cognitive neuroscience of consciousness : basic evidence and a workspace framework . _ cognition _ * 79:*1 - 37 . meunier d , lambiotte r , bullmore et ( 2010 ) modular and hierarchically modular organization of brain networks . _ front . neurosci . _ * 4:*200 . bassett ds , et al . ( 2010 ) efficient physical embedding of topologically complex information processing networks in brains and computer circuits . _ plos comput . biol . _ * 6:*e1000748 . felleman dj , van essen dc ( 1991 ) distributed hierarchical processing in the primate cerebral cortex . _ . cortex _ * 1:*1 - 47 . treisman a ( 1996 ) the binding problem . _ current opinion in neurobiology _ * 6:*171 - 178 . tononi g , sporns o , edelman gm ( 1994 ) a measure for brain complexity : relating functional segregation and integration in the nervous system . usa _ * 91:*5033 - 5037 . sporns o , chialvo dr , kaiser m , hilgetag cc ( 2004 ) organization , development and function of complex brain networks . _ trends cognit . * 8:*418 - 425 . watts d , strogatz s ( 1998 ) collective dynamics of small - world networks . _ nature _ * 393:*440 - 442 . bassett ds , bullmore et ( 2006 ) small - world brain networks . _ neuroscientist _ * 12:*512 - 523 . he y , chen zj , evans ac ( 2007 ) small - world anatomical networks in the human brain revealed by cortical thickness from mri . . cortex _ * 17:*2407 - 2419 . eguiluz vm , chialvo dr , cecchi ga , baliki m , apkarian av ( 2005 ) scale - free brain functional networks . lett . _ * 94:*018102 . achard s , salvador r , whitcher b , suckling j , bullmore et ( 2006 ) a resilient , low - frequency , small - world human brain functional network with highly connected association cortical hubs . _ j. neurosci . _ * 26:*63 - 72 . achard s , bullmore et ( 2007 ) efficiency and cost of economical brain functional networks . _ plos comput . _ * 3:*e17 . hagmann p , et al . ( 2007 ) mapping human whole - brain structural networks with diffusion mri . _ plos one _ * 2:*e597 . song s , sjostrom pj , reigl m , nelson s , chklovskii db ( 2005 ) highly nonrandom features of synaptic connectivity in local cortical circuits . _ plos biol . _ * 3:*e68 . yu s , huang d , singer w , nikolic d ( 2008 ) a small world of neuronal synchrony . . cortex _ * 18:*2891 - 2901 . stam cj , jones bf , nolte g , breakspear m , scheltens p ( 2007 ) small - world networks and functional connectivity in alzheimer s disease . . cortex _ * 17:*92 - 99 . ravasz e , somera al , mongru da , oltvai zn , barabsi a - l ( 2002 ) hierarchical organization of modularity in metabolic networks . _ science _ * 297:*1551 - 1555 . palla g , derenyi i , farkas i , vicsek t ( 2005 ) uncovering the overlapping community structure of complex networks in nature and society . _ nature _ * 435:*814 - 818 . girvan m , newman mej ( 2002 ) community structure in social and biological networks . * 99:*7821 - 7826 . fortunato s ( 2010 ) community detection in graphs . rep . _ * 486:*75 - 174 . song c , havlin s , makse ha ( 2005 ) self - similarity of complex networks . _ nature _ * 433:*392 - 395 . song c , havlin s , makse ha ( 2006 ) origins of fractality in the growth of complex networks . _ nature phys . _ * 2:*275 - 281 . goh ki , salvi g , kahng b , kim d ( 2006 ) skeleton and fractal scaling in complex networks _ phys . _ * 96:*018701 . radicchi f , ramasco jj , barrat a , fortunato s ( 2008 ) complex networks renormalization : flows and fixed points . lett . _ * 101:*148701 . rozenfeld hd , song c , makse ha ( 2010 ) the small world - fractal transition in complex networks : a renormalization group approach . * 104:*025701 . gallos lk , song c , havlin s , makse ha ( 2007 ) scaling theory of transport in complex biological networks . usa _ * 104:*7746 - 7751 . galvao v , et al . ( 2010 ) modularity map of the human cell differentiation network . usa _ * 107:*5750 - 5755 . bunde a , havlin s , editors ( 1996 ) _ fractals and disordered systems _ 2nd edition ( springer - verlag , heidelberg ) . bollobs b ( 1985 ) _ random graphs _ ( academic press , london ) . granovetter ms ( 1973 ) the strength of weak ties . _ american journal of sociology _ * 78:*1360 - 1380 . onnela j - p , et al . ( 2007 ) structure and tie strengths in mobile communication networks . _ usa _ * 104:*7332 - 7336 . tononi g , edelman gm ( 1998 ) consciousness and complexity . _ science _ * 282:*1846 - 1851 . baars bj ( 1997 ) in the theater of consciousness : the workspace of the mind . ( oxford university press , usa ) . pashler h ( 1994 ) dual - task interference in simple tasks : data and theory . _ psychological bulletin _ * 116:*220 - 220 . sigman m , dehaene s ( 2008 ) brain mechanisms of serial and parallel processing during dual - task performance . _ j. neurosci . _ * 28:*7585 - 7598 . sigman m , jobert a , dehaene s ( 2007 ) parsing a sequence of brain activations of psychological times using fmri . _ neuroimage _ * 35:*655 - 668 . salvador r , et al . ( 2005 ) neurophysiological architecture of functional magnetic resonance images of human brain . . cortex _ * 15:*1332 - 1342 . stanley he ( 1971 ) _ introduction to phase transitions and critical phenomena_. ( oxford university press , oxford ) . the entire experimental dataset and modularization and fractal codes are available at http://lev.ccny.cuny.edu/@xmath113hmakse/brain.html zylberberg a , dehaene s , roelfsema pr and sigman m ( 2011 ) the human turing machine : a neural framework for mental programs . _ trends in cognitive sciences _ * 15 : * 293 - 300 . duncan , j ( 2010 ) the multiple - demand ( md ) system of the primate brain : mental programs for intelligent behaviour . _ trends in cognitive sciences _ * 14:*172 - 179 . song c , gallos lk , havlin s , makse ha ( 2007 ) how to calculate the fractal dimension of a complex network : the box covering algorithm . _ j. stat . _ p03006 . barabsi a - l , albert r ( 1999 ) emergence of scaling in random networks . _ science _ * 286:*509 - 512 . de arcangelis l , herrmann hj ( 2010 ) learning as a phenomenon occurring in a critical state . usa _ * 107:*3977 - 3981 . li d , li g , kosmidis k , stanley he , bunde a , havlin s ( 2011 ) percolation of spatially constraint networks . lett . _ * 93:*68004 . daqing l , kosmidis k , bunde a , havlin s ( 2011 ) dimension of spatially embedded networks . _ nature phys . _ * 7 : * 481484 . gallos lk , song c , makse ha ( 2008 ) scaling of degree correlations and its influence on diffusion in scale - free networks , _ phys . lett . _ * 100:*248701 . kleinberg j ( 2000 ) navigation in a small world . _ nature _ * 406:*845 . li g , reis sds , moreira aa , havlin s , stanley he , andrade jr js ( 2010 ) towards design principles for optimal transport networks . lett . _ * 104:*018701 . bianconi g , pin p , marsili m ( 2009 ) assessing the relevance of node features for network structure . usa _ * 106:*11433 - 11438 . cowey a ( 1979 ) cortical maps and visual perception : the grindley memorial lecture . _ the quarterly journal of experimental psychology _ , * 31:*1 - 17 . linsker r ( 1986 ) from basic network principles to neural architecture : emergence of spatial - opponent cells . _ proceedings of the national academy of sciences _ * 83:*7508 . mitchison g ( 1991 ) neuronal branching patterns and the economy of cortical wiring . _ proceedings of the royal society of london b _ * 245:*151 . cherniak c ( 1995 ) neural component placement . _ trends in neuroscience _ * 18:*522 - 527 . chklovskii db ( 2000 ) optimal sizes of dendritic and axonal arbors in a topographic projection . _ journal of neurophysiology _ * 83:*2113 . chklovskii db , schikorski , t and stevens , cf ( 2002 ) wiring optimization in cortical circuits _ neuron _ * 34 : * 341347 kaiser m , hilgetag cc ( 2006 ) nonoptimal component placement , but short processing paths due to long - distance projections in neural systems . _ plos comput . biol . _ * 2:*e95 . logothetis nk , pauls j , augath m , trinath t , and oeltermann a ( 2001 ) neurophysiological investigation of the basis of the fmri signal . _ nature _ * 412:*150 - 157 . sirotin yb and das a ( 2009 ) anticipatory haemodynamic signals in sensory cortex not predicted by local neuronal activity . _ nature _ * 457:*475 - 479 . duvernoy hm , delon s and vannson jl ( 1981 ) anticipcortical blood vessels of the human brain . _ brain research bulletin _ * 7:*519 - 579 . cassot f , lauwers f , fouard c , prohaska s , and lauwers - cances v ( 2006 ) a novel three - dimensional computer - assisted method for a quantitative study of microvascular networks of the human cerebral cortex . _ microcirculation _ * 13:*1 - 18 . risser l , plourabou f , steyer a , cloetens p , le duc g , and fonta c ( 2006 ) from homogeneous to fractal normal and tumorous microvascular networks in the brain . _ journal of cerebral blood flow & metabolism _ * 27:*293 - 303 . kiselev vg , hahn kr , auer dp ( 2003 ) is the brain cortex a fractal ? _ neuroimage _ * 20:*1765 - 1774 . stanley he ( 1999 ) scaling , universality , and renormalization : three pillars of modern critical phenomena . _ _ * 71:*s358-s366 . raichle me , et al . ( 2001 ) a default mode of brain function . usa _ * 98:*676 - 682 . * supporting information * a total of 16 participants ( 7 women and 9 men , mean age , 23 , ranging from 20 to 28 ) were asked to perform two tasks with the instruction that they had to respond accurately and fast to each of them . the first task was a visual task of comparing a given number ( target t1 ) to a fixed reference , and , second , an auditory task of judging the pitch of an auditory tone ( target t2 ) [ 36 ] . the two stimuli are presented with a stimulus onset asynchrony ( soa ) , i.e. , the delay in the onset of t1 and t2 , varying from : soa=0 , 300 , 900 and 1200 ms . in the number - comparison task , a number varying randomly among four values ( 28 , 37 , 53 , 62 ) was flashed on a computer screen and subjects had to respond , with a key press using the right hand , whether the number was larger or smaller than 45 . in the auditory task , subjects had to respond whether the tone was high or low frequency with a key press using the left hand . full details and preliminary statistical analysis of this experiment have been reported in [ 36 ] . the study is part of a larger neuroimaging research program headed by denis le bihan and approved by the comit consultatif pour la protection des personnes dans la recherche biomdicale , hpital de bictre ( le kremlin - bictre , france ) . subjects performed a total of 160 trials ( 40 for each soa value ) with a 12 s inter - trial interval [ 37 ] . the 160 trials were performed in five blocks of 384 s with a resting time of @xmath113 5 min between blocks . for each trial , we recorded whole - brain fmri images at a sampling time , tr = 1.5 s producing 8 fmri images between two consecutive trials . from these images we computed the phase and amplitude of the hemodynamic response of each trial as explained in [ 37 ] the experiments were performed on a 3 t fmri system ( bruker ) . functional images sensitive to blood oxygenation level dependent ( bold ) contrast were obtained with a t2@xmath114-weighted gradient echoplanar imaging sequence [ repetition time ( tr ) = 1.5 s ; echo time = 40 ms ; angle = 90@xmath115 ; field of view ( fov ) = 192 @xmath116 256 mm ; matrix = 64 @xmath116 64 ] . the whole brain was acquired in 24 slices with a slice thickness of 5 mm . high - resolution images ( three - dimensional gradient echo inversion - recovery sequence , inversion time = 700 mm ; fov = 192 @xmath116 256 @xmath116 256 mm ; matrix = 256 @xmath116 128 @xmath116 256 ; slice thickness = 1 mm ) were also acquired . to estimate the periodicity and phase of the event - related bold response , the data from each subject were submitted to a first - level model in which the signal from each trial ( 8 trs of 1.5 s ) was fitted with three regressors : a constant , a sine , and a cosine function at the above period . to facilitate intersubject averaging across possible differences in anatomical localization , the regression weights of the sines and cosines were stereotactically transformed to the standardized coordinate space of talairach and tournoux ( [ montreal neurological institute ] mni 152 average brain ) to spatially normalize for individual differences in brain morphology . normalized images had a resolution of 8 mm@xmath102 . normalized phase images were transformed with the inverse tangent function to yield a phase lag expressed in radians for each voxel @xmath0 and each trial @xmath117 over @xmath2 trials : @xmath118 $ ] [ 37 ] , indicating phase lags in the interval @xmath119$]s . we calculate cross - correlations between different brain areas based on these phases [ 11 , 12 , 38 ] . we determine the equal - time cross - correlation matrix @xmath120 with elements @xmath3 measuring the cross - correlation between the phase activity @xmath121 of the @xmath0-th and @xmath122-th voxel over @xmath2 trials for each subject and soa condition : @xmath123 by construction , the elements satisfy @xmath124 , where @xmath125 corresponds to perfect correlations , @xmath126 corresponds to perfect anticorrelations , and @xmath127 describes a pair of uncorrelated voxels . the entire experimental dataset is available in [ 40 ] . for our analysis , we create a mask where we keep voxels which were activated in more than 75% of the cases , i.e. , in at least 48 instances out of the 64 total cases considered . the obtained number of activated voxels is @xmath128 , varying slightly for different individuals and stimuli . the ` activated or functional map ' exhibits phases consistently falling within the expected response latency for a task - induced activation [ 36 ] . as expected for an experiment involving visual and auditory stimuli and bi - manual responses , the responsive regions included bilateral visual occipito - temporal cortices , bilateral auditory cortices , motor , premotor and cerebellar cortices , and a large - scale bilateral parieto - frontal structure , see si - section `` spatial projection of the modules '' below . in the present analysis , we do not explore the differences in networks between different conditions . rather , we consider them as independent experiments , generating a total of 64 different networks , one for each condition of temporal gap and subject . the use of fmri neighboring voxels can be expected to carry some shared signal due to spatial autocorrelations ( vascular , subject motion or scanner noise ) , which could give rise to spurious correlations over short distance . to test for this effect , we double the lattice spacing , reducing the voxels by a factor of 8 and repeat the calculations . the results are consistent with the percolation picture of fig . 1 , albeit with a lower @xmath7 , while the main results on long - range links are insensitive to this type of artifacts . in order to estimate the accuracy of the correlation calculations , we performed a nonparametric boot strap analysis . we consider the set of the 40 trials per subject and soa value . we perform the boot strap analysis for each possible pair of voxels . the correlation between two voxels for each of those trials serves as our original sample of 40 correlation values . we then draw 10000 re - samples from this sample with substitution . the arithmetic mean is calculated for each re - sample . calculating the average value of all these means gives the boot strap estimate for the mean correlation . the 95% boot strap confidence interval is calculated by the distribution of the 10000 mean values at the 0.05 and 0.95 points of the distribution , respectively . the above process yields the confidence interval for the correlation value between two voxels . a different pair of voxels may have very different value of correlation , so in fig . [ si_bs ] we present the 95% bootstrap confidence interval as a function of the average value of correlation . the interval becomes smaller , i.e. the accuracy of the calculation increases , for larger @xmath5 values . considering the networks of fig . 4a and b , for example , the intervals for @xmath89 and @xmath88 correspond to ( see inset ) [ 0.9744 , 0.9760 ] and [ 0.9795 , 0.981 ] , respectively . the complex network representation reveals functional links between brain areas , but can not directly reveal spatial correlations . since voxels are embedded in real space , we also study the topological features of modules in three dimensions , where now voxels assume their known positions in the brain and links between them are transferred from the corresponding network , i.e. , they are assigned according to the degree of correlation between any two voxels , eq . ( [ c ] ) , which is independent of the voxels proximity in real space . the above procedure yields a different spatial projection of the modules for each subject ; an example for subject # 1 and soa=900 ms in the medial occipital cortex is shown in fig . we study each of these percolation modules separately and find that they all carry statistically similar patterns . the topography of the identified modules reflects coherent patterns across different subjects , as shown next . s2a shows a medial sagital view of the largest four percolation modules for all the participants under stimulus soa=0 . in virtually all subjects we observe a module covering the anterior cingluate ( ac ) region , a module covering the medial part of the posterior parietal cortex ( ppc ) and a module covering the medial part of posterior occipital cortex ( area v1/v2 ) , along the calcarine fissure . we measure the likelihood that a voxel appears in the largest percolation module among all the participants in fig . s2a by counting , for each voxel , the number of individuals for which it was included in one of the first four percolation modules . the spatial distribution of the first percolation modules averaged over all the subjects depicted in figs . s2b and s2c shows that modules in the three main modes , v1/v2 , ac and ppc , are ubiquitously present in percolation modules and , to a lesser extent , voxels in the motor cortex ( along the central sulcus ) are slightly more predominantly on the left hemisphere . the correlation networks obtained from each subject yield modules with consistent topographic projections . for a given percolation module , the detection of submodules or boxes follows from the application of the box - covering algorithm for self - similar networks [ 22,43 ] . the algorithm can be downloaded at [ 40 ] . in box covering we assign every node to a box or submodule , by finding the minimum possible number of boxes , @xmath129 , that cover the network and whose diameter ( defined as the maximum distance between any two nodes in this box ) is smaller than @xmath21 . we implement the maximum excluded mass burning ( memb ) algorithm from [ 43 ] for box covering . the algorithm uses the basic idea of box optimization , where we require that each box should cover the maximum possible number of nodes , and works as follows : we first locate the optimal ` central ' nodes which will act as the origins for the boxes . this is done by first calculating the number of nodes ( called the mass ) within a distance @xmath130 from each node . we use , @xmath131 . the node that yields the largest mass is marked as a center . then we mark all the nodes in the box of this center node as ` tagged ' . we repeat the process of calculating the mass of the boxes starting from all non - center nodes , and we identify a second center according to the largest remaining mass , while nodes in the corresponding box are ` tagged ' , and so on . when all nodes are either centers or ` tagged ' we have identified the minimum number of centers that can cover the network at the given @xmath130 value . starting from these centers as box origins , we then simultaneously burn the boxes from each origin until the entire network is covered , i.e. each node is assigned to one box ( we call this process burning since it is similar to burning algorithms developed to investigate clustering statistics in percolation theory [ 29,30 ] ) . in fig . 2a we show how box - covering works for a simple network at different @xmath21 values . rg is then the iterative application of this covering at different @xmath21 . connections between voxels are determined according to the value of the correlation between the two voxels , as described above . this value may also depend on the physical ( euclidean ) distance between the two voxels , since areas that are close to each other should interact stronger . we studied the correlation function , @xmath36 of the phases of the voxels : @xmath132 where @xmath33 denotes the phase of voxel @xmath0 . the distance @xmath35 is the euclidean distance between the two voxels 1 and 2 and the average is taken over all pairs at distance @xmath35 . this function can be interpreted as the correlation between two spins with orientation determined by the phases @xmath33 of the voxels . we notice that this correlation function is usually studied in ising - like spin models . we find that @xmath36 decays algebraically with distance , as shown in fig . [ si_correl ] , and follows a power law form , @xmath133 . the value of the exponent @xmath134 was calculated through standard ols regression . notice that this function does not go to 0 asymptotically , but reaches a value of 0.1 , which represents the average correlation ( notice that in the definition of the correlation , the average value was not subtracted ) . this indicates that long - range correlations remain strong even at large distances . further analysis is required to elaborate on this point , which is currently outside the scope of our present study . in fig . 2d of the main text we show an aggregate average of the degree distributions for all clusters . this curve exhibits the general trends of the @xmath26 distribution , demonstrating for example the heavy tail , but it can not be used for a direct determination of the exponent @xmath30 . in our work we studied the properties of 192 network clusters , as described in the main text . the calculation of the scaling exponents was done separately for each network . the resulting set of 192 values was then analyzed through non - parametric boot strap analysis , in order to get the average value of the exponent and the corresponding confidence intervals . as an example , in fig . [ si_pk ] we show the degree distributions for 9 different clusters . in the plots , it is clear that there is always a plateau at small @xmath37 values , while in many cases there is an asymptotic exponential cutoff . we fitted these distributions assuming that a power law describes the data within a given interval only . for this , we used a generalized power - law form @xmath135 where @xmath136 and @xmath137 are the boundaries of the fitting interval and the hurwitz @xmath138 function is given by @xmath139 . we used the maximum likelihood method , following e.g. clauset et al , siam review , 51 , 661 ( 2009 ) . the fit was done in an interval where the lower boundary was @xmath136 . for a given @xmath136 value we were fixing the upper boundary to @xmath140 , where @xmath141 is a parameter . we calculated the slopes in successive intervals by continuously increasing @xmath136 and varying the value of @xmath141 from 4 to 30 . in this way , we sampled a large number of possible intervals . for each one of them we calculated the maximum likelihood estimator through the numerical solution of @xmath142 \right)\ ] ] where @xmath143 are all the degrees that fall within the fitting interval and @xmath144 is the total number of nodes with degrees in this interval . the optimum interval was determined through the kolmogorov - smirnov test . for the goodness - of - fit test , we used the monte - carlo method described in clauset et al . for each possible fitting interval we generated 10000 synthetic random distributions following the best - fit power law . we then calculated the value of the kolmogorov - smirnov ( ks ) test for each one of them and measured the fraction @xmath145 of realizations where the real data ks value was smaller than the synthetic sk value . we accepted the power - law hypothesis when this ratio was larger than @xmath146 . the average ratio over all clusters that were retained was @xmath147 . in this way , it is possible that we could accept more than one exponents for a given cluster at different intervals . in all these cases , the different @xmath30 values were very close to each other and we considered the final exponent to be the average of the individual exponents . standard boot strap analysis on the resulting set of the individual cluster values yielded the exponent @xmath148 , with a 95% confidence interval @xmath149 $ ] . the same analysis was performed to test for a possible exponential description of the data . we scanned the same intervals as for the case of power - law and we used the maximum likelihood method to determine the optimum exponential fitting to the form : @xmath150 we again used ks statistics to determine the optimum fitting intervals and also the goodness - of - fit . in all the cases where the power - law was accepted , the exponential fitting gave an average ratio of @xmath151 , which rules out the possibility of an exponential distribution . the structure of a fractal network can be characterized by a set of scaling exponents . they define the scaling of many important system properties . some of these properties and the corresponding exponents are as follows : \a . the degree distribution : @xmath28 , where @xmath30 is the degree exponent [ 44 ] . the scaling of the mass with size : @xmath152 , which defines the fractal exponent @xmath24 [ 22 ] . the degree - degree distribution @xmath153 , where @xmath48 is the degree - degree exponent , and can be measured through @xmath154 , which is the integration of @xmath27 over @xmath155 [ 48 ] . the probability that modules are connected through their hubs , @xmath156 defines the hub - hub exponent @xmath47 [ 23 ] . the scaling of the degree of the modules with the size of the modules : @xmath157 , which defines the @xmath41 exponent [ 22 ] . the scaling of the modular factor as defined in eq . ( 3 ) : @xmath158 , through the modularity exponent @xmath62 [ 27,28 ] . scaling theory then defines precise relations between the exponents valid for fractal scale - free networks : \g . @xmath43 [ 22 ] , \h . @xmath49 [ 48 ] , and \i . @xmath159 [ 27,28 ] . we have measured directly all the exponents ( see fig . 2 and fig . [ si_matrix ] ) for the brain modules and find : @xmath148 , @xmath50 , @xmath160 , @xmath44 , @xmath51 , @xmath161 . using these values in the known scaling relations above ( g ) and ( h ) , we predict @xmath45 and @xmath162 , which are reasonably close to the calculated exponents @xmath163 and @xmath164 from the direct measures . this set of results gives support to a scale - free fractal morphology of the brain modules . notice that a euclidean 2d lattice would be obtained in the limit @xmath165 , @xmath41=0 , @xmath166 . in the main text of the paper we have described our modularity analysis of the brain clusters according to the memb technique . the modular properties of the same clusters can be also analyzed through techniques that partition a network according to maximization of modularity . we employed the girvan - newman method [ 20 ] , which locates the point where the modularity measure , @xmath53 , is maximum . the definition of @xmath53 according to [ 20 ] is : @xmath167 where @xmath168 is the number of modules , @xmath169 is the number of links in the network , @xmath170 is the number of links within the module @xmath0 , and @xmath171 is the sum of the degrees in this module . a value of @xmath172 corresponds to a completely random configuration or to the case of one module only . for the brain clusters we found an average modularity value of @xmath173 . this is an indication of strong modularity within each cluster . a direct comparison between memb and the girvan - newman method shows that they result in quite similar partitions . we calculated that 92% of the total links belong within a given module in both methods . a visual comparison is shown in fig . [ si_newman ] . the maximization of modularity verifies the modular character of the clusters . the use of the memb , though , provides us with the extra advantage of modifying the scale at which we observe the modules to determine whether the modular structure is scale - invariant , i.e. , if it is composed of modules inside modules .

let be a group . we define the _ coprime graph of subgroups _ of , denoted by , is a graph whose vertex set is the set of all proper subgroups of , and two distinct vertices are adjacent if and only if the order of the corresponding subgroups are coprime . in this paper , we study some connections between algebraic properties of a group and graph theoretic properties of its coprime graph . [ [ keywords ] ] keywords : + + + + + + + + + coprime graph , finite groups , connectedness , independence number , clique number , planar . [ [ mathematics - subject - classification ] ] mathematics subject classification : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + 05c25 , 20k27 , 05c69 , 05c10 .

1510.00129

graph theory provide tools to study the algebraic properties of algebraic structures . in particular , there are several graphs associated with groups to study some specific properties of groups , for instance , intersection graph of subgroups of groups , prime graph of groups , non - commuting graphs of groups and permutability graph of subgroups of groups ( see @xcite , @xcite , @xcite , @xcite and the references therein ) . in @xcite , sattanathan and kala defined the order prime graph of a group @xmath0 , which is a graph having the set of all elements of @xmath0 as its vertices , and two distinct vertices are adjacent if and only if the orders of the corresponding subgroups are coprime . they have studied some properties of this graph . this graph was further investigated by xuanlong ma et al @xcite and hamid reza dorbidi @xcite . they called the order prime graph of a group as the coprime graph of a group . the relation of coprimeness of the orders of the subgroups of a group plays a significant role in the determination of the structural properties of that group . also the relation of coprimeness is not transitive on the set of all proper subgroups of a group . in this paper , we define the _ coprime graph of subgroups _ of @xmath0 , denoted by @xmath1 . it is a graph having all the proper subgroups of @xmath0 as its vertices , and two distinct vertices @xmath2 and @xmath3 are adjacent if and only if @xmath4 and @xmath5 are coprime . now we recall some basic definitions and notations of graph theory . we use the standard terminology of graphs ( e.g. , see @xcite ) . let @xmath0 be a simple graph . @xmath0 is said to be @xmath6-partite if the vertex set of @xmath0 can be partitioned to @xmath6 sets such that no two vertices in same partitions are adjacent . a complete @xmath6-partite graph , denoted by @xmath7 , is a @xmath6-partite graph having partition sizes @xmath8 such that every vertex in each partition is adjacent with all the vertices in the remaining partitions . in particular , @xmath9 is called a _ star_. a graph whose edge set is empty is called a _ null _ graph or _ totally disconnected _ graph . @xmath10 denotes the complete graph on @xmath11 vertices . @xmath12 and @xmath13 respectively denotes the path and cycle with @xmath11 edges . we denote the degree of a vertex @xmath14 in @xmath0 by @xmath15 . a graph is said to be _ connected _ if any two vertices of it can be joined by a path . the _ diameter _ of a connected graph is the maximum of the length of the shortest path between any pair of vertices . a _ tree _ is a connected graph with out cycles . @xmath0 is said to be @xmath2-_free _ if @xmath0 has no subgraph isomorphic to @xmath2 . the girth of @xmath0 , denoted by @xmath16 , is the length of its shortest cycle , if it exist ; other wise @xmath17 . an _ independent set _ of @xmath0 is a subset of @xmath18 having no two vertices are adjacent . the _ independence number _ of @xmath0 , denoted by @xmath19 , is the cardinality of the largest independent set . a _ clique _ of @xmath0 is a complete subgraph of @xmath0 . the clique number @xmath20 of @xmath0 is the cardinality of a largest clique in @xmath0 . the _ chromatic number _ @xmath21 of @xmath0 is the smallest number of colours needed to colour the vertices of @xmath0 such that no two adjacent vertices gets the same colour . g is said to be _ weakly @xmath22-perfect _ if @xmath23 . a graph is said to be _ planar _ , if it can be drawn in the plane , so that no two lines intersect except at the vertices ; otherwise it is said to be _ a graph is called _ unicyclic _ , if it contains exactly one cycle . for any integer @xmath24 , @xmath25 denotes the set of all prime divisors of @xmath11 . if @xmath0 is a finite group , then @xmath26 is denoted by @xmath27 . the number of sylow p - subgroups of a group g is denoted by @xmath28 . we denote the order of an element @xmath29 by @xmath30 . moreover , through out this paper , @xmath31 , @xmath32 , @xmath33 , @xmath34 denotes the distinct primes . since the only groups having no proper subgroups are the trivial group , and the groups of prime order , it follows that , we can define @xmath1 only when the group @xmath0 is neither the trivial group nor the group of prime order . so , unless otherwise mentioned , throughout this paper we consider only groups other than the trivial group , and the groups of prime order . we use only elementary methods . in section 2 , we classify all the finite groups whose coprime graph of subgroups are one of totally disconnected , bipartite , connected , complete , complete bipartite , tree , star or path , and show that the coprime graph of subgroups of a finite group can not be a cycle . for a finite group @xmath0 , we obtain the independence number , clique number , chromatic number , diameter , girth of @xmath1 , and show that @xmath1 is weakly @xmath22-perfect . moreover , we obtain the degrees of vertices of @xmath35 , and we show that every simple graph is an induced subgraph of @xmath35 , for some @xmath11 . in section 3 , we classify all the finite groups whose coprime graph of subgroups of groups are one of planar , @xmath36-free , @xmath37-free , @xmath38-free , @xmath39-free , unicyclic . [ order prime graph of subgroups 14 ] let @xmath40 and @xmath41 be two groups . if @xmath42 , then @xmath43 . let @xmath44 be a group isomorphism . define a map @xmath45 by @xmath46 , for every @xmath47 . since a group isomorphism preserves the order of subgroups , so it follows that @xmath48 is a graph isomorphism . * remark : * the converse of the above theorem [ order prime graph of subgroups 14 ] , is not true , for if @xmath49 and @xmath50 , then the number of proper subgroups of @xmath40 is four and their orders are @xmath31 , @xmath51 , @xmath52 , @xmath53 ; the number of proper subgroups of @xmath41 is four and their orders are 4 , 4 , 4 , 2 . here @xmath54 , but @xmath55 . [ order prime graph of subgroups of groups 4 ] let @xmath0 be a group of order @xmath56 , where @xmath57 s are distinct primes , @xmath58 . then 1 . @xmath1 is @xmath6-partite ; 2 . @xmath59 , where for each @xmath60 , 2 , @xmath61 , @xmath6 , @xmath62 is the set of all proper subgroups of @xmath0 whose order is divisible by @xmath57 ; 3 . @xmath63 ; in particular , @xmath1 is weakly @xmath22-perfect . let @xmath64 be the set of all proper subgroups of @xmath0 whose order is divisible by @xmath65 . for each @xmath66 , 3 , @xmath61 , @xmath67 , let @xmath68 . then clearly the collection @xmath69 forms a partition of the vertex set of @xmath1 . also no two vertices in a same partition are adjacent in @xmath1 . moreover , @xmath6 is the minimal number such that a @xmath6-partition of the vertex set of @xmath1 is having this property , since @xmath70 . it follows that @xmath1 is @xmath6-partite . now for each @xmath60 , 2 , @xmath61 , @xmath6 , let @xmath62 be the set of all proper subgroups of @xmath0 whose order is divisible by @xmath57 . clearly each @xmath62 is a maximal independent set of @xmath1 . thus @xmath59 . for each @xmath71 , @xmath0 has a subgroup of order @xmath57 . then the set having one subgroup from each of these orders forms a clique in @xmath1 . since @xmath1 is @xmath6-partite , it follows that @xmath72 . obviously , @xmath73 . weakly @xmath22-perfectness of @xmath1 follows from the definition . the next result is an immediate consequence of theorem [ order prime graph of subgroups of groups 4 ] ( 1 ) . [ c1 ] let @xmath0 be a group with @xmath70 . then 1 . @xmath1 is totally disconnected if and only if @xmath74 ; 2 . @xmath1 is bipartite if and only if @xmath75 . [ order prime graph of subgroups of groups 6 ] let @xmath0 be a finite group . then 1 . @xmath1 is complete bipartite if and only if @xmath0 is isomorphic to one of @xmath76 , @xmath77 , @xmath78 or @xmath79 ; 2 . the following are equivalent : 1 . @xmath80 or @xmath77 ; 2 . @xmath1 is a tree ; 3 . @xmath1 is a star . 3 . the following are equivalent : 1 . @xmath80 ; 2 . @xmath1 is complete ; 3 . @xmath1 is a path . in view of part ( 2 ) of corollary [ c1 ] , to prove parts ( 1 ) , ( 2 ) , and @xmath81 of ( 3 ) of this theorem , it is enough to consider the groups of order @xmath82 and @xmath83 . if @xmath84 , then by corollary [ c1](1 ) , @xmath1 is totally disconnected and so it is neither complete bipartite nor a tree . let @xmath85 with @xmath86 . then @xmath87 or @xmath77 . if @xmath87 , then @xmath88 and so @xmath1 is a path . if @xmath89 , then @xmath0 has an unique subgroup of order @xmath32 , and @xmath32 subgroups of order @xmath31 ; also these are the only proper subgroups of @xmath0 . it follows that @xmath90 and so @xmath1 is complete bipartite , star ; but not a path . let @xmath91 . suppose @xmath0 is abelian , then @xmath0 has a subgroup of order @xmath92 and so @xmath1 is disconnected . it follows that @xmath1 is neither complete bipartite nor a tree . now assume that @xmath0 is non - abelian . here we use the classification of groups of order @xmath93 given in @xcite . * case 1 : * @xmath94 : * case 1a : * @xmath95 . by sylow s theorem , it is easy to see that there is no non - abelian group in this case . * case 1b : * @xmath96 but @xmath97 . in this case , there are two non - abelian groups . the first group is @xmath98 . here @xmath40 has an unique subgroup of order @xmath32 ; unique subgroup of order @xmath92 ; @xmath32 subgroups of order @xmath51 and unique subgroups of order @xmath31 ; also these are the only proper subgroups of @xmath40 . therefore , @xmath99 . so @xmath100 is disconnected . hence @xmath100 is neither complete bipartite nor a tree . the second group is @xmath101 . here @xmath41 has a subgroup @xmath102 of order @xmath92 and so @xmath103 is disconnected . hence @xmath103 is neither complete bipartite nor a tree . * case 1c : * @xmath104 . in this case , we have both groups @xmath40 and @xmath41 from case 1b together with the group @xmath105 . here @xmath106 has unique subgroup of order @xmath32 ; unique subgroup of order @xmath92 ; @xmath32 subgroups of order @xmath51 , and @xmath32 subgroups of order @xmath31 . also these are the only proper subgroups of @xmath106 . therefore , @xmath107 and so @xmath108 is disconnected . hence @xmath108 is neither complete bipartite nor a tree . * case 2 : * @xmath109 * case 2a : * @xmath110 . then there is no non - abelian subgroups . * case 2b : * @xmath111 . in this case , we have two groups . the first one is @xmath112 . here @xmath113 has a unique subgroup of order @xmath51 ; unique subgroup of order @xmath31 ; @xmath31 subgroups of order @xmath92 ; @xmath51 subgroups of order @xmath32 . also these are the only proper subgroups of @xmath113 . therefore , @xmath114 and so @xmath115 is disconnected . hence @xmath115 is neither complete bipartite nor a tree . next we have the family of groups @xmath116 . there are @xmath117 isomorphism types in this family ( one for @xmath118 and one for each pair @xmath119 in @xmath120 . we will refer to all of these groups as @xmath121 of order @xmath93 . here @xmath121 has a subgroup @xmath102 of order @xmath92 and so @xmath122 is disconnected . hence @xmath123 is neither complete bipartite nor a tree . * case 2c : * @xmath124 . in this case , we have only one subgroup of order @xmath93 , given by @xmath125 , where @xmath126 has order @xmath32 in @xmath127 . here @xmath128 does not have a subgroup of order @xmath92 . but @xmath128 has an unique subgroup of order @xmath51 ; @xmath129 subgroups of order @xmath31 ; @xmath51 subgroups of order @xmath32 , also these are the only proper subgroups of @xmath128 . hence @xmath130 , which is complete bipartite ; but which is not a tree . note that if @xmath131 , the cases 1 and 2 are not mutually exclusive . up to isomorphism , there are three non - abelian groups of order 12 : @xmath132 , @xmath133 and @xmath79 . in case 1b , we already dealt with the group @xmath134 . it is easy to see that @xmath135 and @xmath136 . these three graphs are not trees and the only complete bipartite graph is @xmath137 . if @xmath138 , @xmath139 , then let @xmath140 , @xmath141 be elements in @xmath0 of order @xmath31 , @xmath32 respectively . here @xmath142 is a proper subgroup of @xmath0 whose order is divisible by @xmath31 and @xmath32 . therefore , @xmath1 is disconnected . hence @xmath1 is neither complete bipartite nor a tree . if @xmath143 , @xmath144 , @xmath145 , then @xmath0 has a subgroup with prime index , since @xmath0 is solvable and so @xmath1 is disconnected . hence @xmath1 is neither complete bipartite nor a tree . combining all the above cases together , the proof of parts ( 1 ) , ( 2 ) , and @xmath81 of ( 3 ) of this theorem follows . now , we prove @xmath146 of part ( 3 ) : clearly @xmath147 . so assume that @xmath1 is complete . then by theorem [ order prime graph of subgroups of groups 4 ] , each partition @xmath148 , @xmath149 of @xmath1 must contain exactly one subgroup of distinct prime order and so these subgroups are normal in @xmath0 . if @xmath150 , then @xmath0 must contain a subgroup whose order is a product of @xmath6 distinct primes , so this subgroup is an isolated vertex in @xmath1 , which is not possible . hence @xmath151 and so by part ( 1 ) of this theorem , it turns out that @xmath80 . this completes the proof . [ order prime graph of subgroups of groups 8 ] if @xmath0 is a finite group , then @xmath152 , for @xmath153 . suppose @xmath1 is the cycle @xmath154 of length @xmath11 . since @xmath155 , so without loss of generality , we assume that , @xmath156 , @xmath157 and @xmath158 or @xmath82 . if @xmath158 , then @xmath159 , @xmath160 , @xmath161 are adjacent and so @xmath1 is complete , which is not possible , by theorem [ order prime graph of subgroups of groups 6](3 ) . if @xmath162 , then @xmath163 implies that @xmath164 or @xmath33 . if @xmath165 , then @xmath159 , @xmath160 , @xmath166 are adjacent , which is not possible . so we have @xmath164 . then @xmath167 and so @xmath159 and @xmath166 are adjacent in @xmath1 . it follows that @xmath168 and @xmath143 , @xmath144 , @xmath169 . now we check the existence of such a group . if @xmath170 , then @xmath0 has at least five proper subgroups , which is not possible . if @xmath171 , then @xmath91 or @xmath92 . in this case , we have shown in the proof of theorem [ order prime graph of subgroups of groups 6 ] that @xmath1 can not be a cycle . this completes the proof . [ order prime graph of subgroups of groups 3 ] let @xmath0 be a finite group . then @xmath1 is connected if and only if @xmath0 does not have a proper subgroup @xmath2 with @xmath172 . in this case , @xmath173 . suppose @xmath0 has a subgroup , say @xmath2 with @xmath172 . then @xmath4 is not relatively prime to any other subgroups of @xmath0 . therefore , @xmath1 is disconnected . conversely , assume that @xmath0 does not have a subgroup @xmath2 with @xmath172 . suppose @xmath1 is complete , then @xmath1 is connected and @xmath174 . now assume that @xmath1 is not complete . let @xmath2 and @xmath3 be two non - adjacent vertices in @xmath1 . then by assumption , @xmath175 and @xmath176 , and so there exist @xmath57 , @xmath177 such that @xmath178 and @xmath179 . if @xmath180 , then there is a path @xmath181 , where @xmath159 is a subgroup of @xmath0 of order @xmath57 . if @xmath182 , then there is a path @xmath183 , where @xmath159 , @xmath160 are subgroups of @xmath0 of orders @xmath57 , @xmath184 respectively . it follows that @xmath1 is connected and @xmath185 . note that @xmath186 , @xmath187 and @xmath188 , so it shows that the diameter of @xmath1 takes all the values in @xmath189 . this completes the proof . from the above theorem , for a given finite group @xmath0 , if @xmath1 is disconnected , then @xmath0 has a proper subgroup @xmath2 with @xmath172 . it turns out that such a subgroup @xmath2 will be an isolated vertex of @xmath1 . as a consequence , we have the following result . let @xmath0 be a finite group . if @xmath1 is disconnected , then @xmath190 , where @xmath191 is a connected component of @xmath1 , and @xmath33 is the number of proper subgroups @xmath2 of @xmath0 with @xmath172 . [ order prime graph of subgroups 9 ] if @xmath0 is a finite group , then @xmath192 . let @xmath0 be a group of order @xmath193 , where @xmath57 s are distinct primes and @xmath58 . if @xmath194 , then any three subgroups of @xmath0 of distinct prime orders are mutually adjacent in @xmath1 and so @xmath1 contains @xmath195 as a subgraph . it follows that @xmath196 . if @xmath197 , then by corollary [ c1](2 ) , @xmath1 is bipartite and so @xmath1 can not contain an odd cycle . now we consider the following cases : * case a : * @xmath143 , @xmath144 , @xmath198 . here @xmath0 has subgroups of orders @xmath31 , @xmath51 , @xmath32 , @xmath199 , let them be @xmath159 , @xmath160 , @xmath161 , @xmath166 respectively . then @xmath200 contains the cycle @xmath201 and so @xmath202 . * case b : * @xmath138 , @xmath203 . suppose sylow @xmath32-subgroup of @xmath0 is not unique , then @xmath0 has at least two sylow @xmath32-subgroup , let them be @xmath159 , @xmath160 and @xmath0 has subgroups of order @xmath31 , @xmath51 , let them be @xmath161 , @xmath166 respectively . then @xmath1 contains the cycle @xmath201 and so @xmath202 . suppose sylow @xmath32-subgroup of @xmath0 is unique , then in the bipartition of the vertex set of @xmath1 , one partition contains only the sylow @xmath32-subgroup of @xmath0 and another partition contains the remaining subgroups of @xmath0 . it follows that @xmath1 does not contains a cycle , so @xmath204 is @xmath205 . * case c : * @xmath85 . by theorem [ order prime graph of subgroups of groups 6](3 ) , @xmath1 is a path and so @xmath204 is @xmath205 . * case d : * @xmath84 . by corollary [ c1](1 ) , @xmath1 is totally disconnected and so @xmath204 is @xmath205 . proof follows by combining all the above cases together . [ order prime graph of subgroups 13 ] let @xmath206 , where @xmath57 s are distinct primes and @xmath58 . if @xmath2 is a proper subgroup of @xmath207 of order @xmath208 , then @xmath209 . it is well known that for every divisor @xmath210 of @xmath11 , @xmath207 has a unique subgroup of order @xmath210 . let @xmath3 be a subgroup of @xmath207 which is adjacent with @xmath2 in @xmath35 . then @xmath211 and @xmath212 , with @xmath213 , @xmath214 , @xmath61 , @xmath215 . but for each @xmath216 , the power of @xmath184 can be chosen in @xmath217 ways . it follows that , such a subgroup @xmath3 can be chosen in @xmath218 ways . excluding the trivial subgroup , we have @xmath219 subgroups in @xmath207 which are adjacent with @xmath2 in @xmath35 . this completes the proof . [ order prime graph of subgroups 15 ] if @xmath191 is a simple graph on @xmath220 vertices , then there exist @xmath221 such that @xmath191 is an induced subgraph of @xmath222 . let @xmath11 be the number of maximal independent sets of @xmath191 . now assign @xmath11 distinct primes for each of these maximal independent sets . let @xmath14 be a fixed vertex of @xmath191 . if @xmath14 belongs to @xmath223 maximal independent sets of @xmath191 , then label to @xmath14 , the product of primes which are assigned to these @xmath223 maximal independent sets . similarly we can label the other vertices of @xmath191 . if all these labeling are distinct , then keep them as it is . otherwise , in order to make the labeling distinct , we relabel the vertices by using different powers of these primes . now let @xmath224 be the least common multiple of the labels assigned to vertices of @xmath191 . again relabel each of these labels by subgroup of @xmath225 whose order is the same label . then it turns out that @xmath191 is an induced subgraph of @xmath225 . hence the proof . now we illustrate theorem [ order prime graph of subgroups 15 ] in the following example . consider the graph @xmath191 as shown in figure [ fig : f3 ] . ] here @xmath226 , @xmath227 are the only maximal independent subsets of @xmath191 . first we assign prime @xmath57 to @xmath228 for each @xmath60 , 2 . here @xmath229 , so label @xmath230 by @xmath65 ; @xmath231 , so label @xmath232 by @xmath233 ; @xmath234 , so label @xmath235 by @xmath65 ; @xmath236 , so label @xmath237 by @xmath65 ; @xmath238 , so label @xmath239 by @xmath233 . since @xmath230 , @xmath235 , @xmath237 have the same label , and @xmath232 , @xmath239 have the same label , so we relabel them in the following way : label @xmath230 , @xmath235 , @xmath237 by @xmath65 , @xmath240 , @xmath241 respectively , and @xmath232 , @xmath239 by @xmath233 , @xmath242 respectively . let @xmath243 . again relabel each of these labels by subgroup of @xmath244 whose order is the same label . it follows that @xmath191 is the induced subgraph of @xmath244 . the famous kuratowski s theorem ( see ( * ? ? ? * theorem 11.13 ) ) characterized planar graphs as those graphs which does not contain subgraphs homeomorphic to either @xmath245 or @xmath246 . we repeatedly use this theorem to show the non - planarity of the coprime graph of subgroups of groups . the explicit plane embedding will be given when the coprime graph of subgroups of a group is planar . 1 . @xmath1 is planar if and only if @xmath0 is one of a @xmath31-group , @xmath247 , @xmath248 , where @xmath249 is a @xmath31-group , @xmath250 , @xmath251 , @xmath252 , @xmath133 , @xmath253 , @xmath254 , @xmath255 , @xmath256 , @xmath257 with @xmath0 has unique cyclic sylow @xmath32-subgroup , @xmath258 , @xmath259 or @xmath260 ; 2 . @xmath1 is @xmath36-free if and only if @xmath0 is one of a @xmath31-group , @xmath247 , @xmath248 , where @xmath249 is a @xmath31-group , @xmath250 , @xmath251 , @xmath133 , @xmath261 or @xmath258 ; 3 . @xmath1 is @xmath262-free if and only if @xmath0 is one of @xmath31-group , @xmath248 , @xmath247 , where @xmath249 is a @xmath31-group , @xmath250 , @xmath251 , @xmath133 or @xmath258 ; 4 . @xmath1 is @xmath37-free if and only if @xmath0 is one of @xmath31-group , @xmath263 , @xmath261 , @xmath264 , @xmath265 , @xmath266 or @xmath258 ; 5 . @xmath1 is @xmath38-free if and only if @xmath0 is one of a @xmath31-group , @xmath267 @xmath268 , @xmath261 or @xmath258 ; 6 . @xmath1 is @xmath39-free if and only if @xmath0 is either a @xmath31-group or @xmath76 ; 7 . @xmath1 is unicyclic if and only if @xmath0 is either @xmath261 or @xmath258 . let @xmath270 . let @xmath159 , @xmath160 , @xmath161 , @xmath166 be subgroups of @xmath0 of order @xmath31 , @xmath32 , @xmath33 , @xmath34 respectively and let @xmath271 be subgroup of @xmath0 , where @xmath140 , @xmath141 are elements of @xmath0 of order @xmath32 , @xmath34 respectively . then @xmath1 contains @xmath37 as a subgraph with bipartition @xmath272 and @xmath273 . we divide the rest of the proof into two cases : * case 1 : * let one of @xmath144 , @xmath274 , @xmath275 or @xmath276 , without loss of generality we assume that @xmath139 . then @xmath1 contains @xmath246 as a subgraph with bipartition @xmath277 , @xmath160 , @xmath278 and @xmath279 , @xmath280 , @xmath281 , where order of @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 , @xmath282 are @xmath31 , @xmath51 , @xmath52 , @xmath32 , @xmath33 , @xmath34 respectively . * case 2 : * let each @xmath144 , @xmath274 , @xmath275 and @xmath283 . suppose the sylow subgroups of @xmath0 is not unique , without loss of generality , we assume that the sylow @xmath31-subgroup is not unique . here @xmath0 has at least three subgroups , say @xmath159 , @xmath160 , @xmath161 of order @xmath82 , and subgroups @xmath166 , @xmath280 , @xmath282 of order @xmath32 , @xmath33 , @xmath34 respectively . then @xmath1 contains @xmath246 as a subgraph with bipartition @xmath277 , @xmath160 , @xmath278 and @xmath279 , @xmath280 , @xmath281 . * subase 2a : * @xmath0 is cyclic . if @xmath269 , then let @xmath284 , @xmath60 , 2 , @xmath61 , 14 be subgroups of @xmath0 of orders @xmath31 , @xmath32 , @xmath33 , @xmath34 , @xmath92 , @xmath285 , @xmath286 , @xmath287 , @xmath288 , @xmath289 , @xmath290 , @xmath291 , @xmath292 , @xmath293 respectively . then @xmath1 is planar and the corresponding plane embedding is shown in figure [ fig : f4 ] ; also @xmath1 contains @xmath36 and @xmath37 . if @xmath294 , then @xmath1 contains @xmath246 as a subgraph with bipartition @xmath277 , @xmath160 , @xmath278 and @xmath279 , @xmath280 , @xmath281 , where @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 , @xmath282 are subgroups of @xmath0 of order @xmath31 , @xmath51 , @xmath92 , @xmath33 , @xmath34 , @xmath289 respectively . if @xmath295 , @xmath296 or @xmath297 , then @xmath0 has @xmath298 as a subgroup and so @xmath1 contains @xmath246 . * subcase 2b : * @xmath0 is non - cyclic abelian . here @xmath0 has atleast three subgroups of any one of prime order @xmath31 , @xmath32 , @xmath33 or @xmath34 , without loss of generality , say @xmath31 . then @xmath1 contains @xmath246 as a subgraph with bipartition @xmath277 , @xmath160 , @xmath278 and @xmath279 , @xmath280 , @xmath281 , where @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 , @xmath282 are subgroups of @xmath0 of order @xmath31 , @xmath31 , @xmath31 , @xmath32 , @xmath33 , @xmath34 respectively . 1 . @xmath1 is planar if and only if @xmath0 is isomorphic to either @xmath258 or @xmath259 ; 2 . @xmath1 contains @xmath39 ; 3 . the following are equivalent : 1 . @xmath299 ; 2 . @xmath1 is @xmath37-free ; 3 . @xmath1 is @xmath36-free ; 4 . @xmath1 is @xmath38-free ; 5 . @xmath1 is @xmath262-free ; 6 . @xmath1 is unicyclic . * case 1 : * let one of @xmath144 , @xmath274 or @xmath301 , without loss of generality we assume that @xmath139 . then @xmath1 contains @xmath246 as a subgraph with bipartition @xmath277 , @xmath160 , @xmath278 and @xmath279 , @xmath280 , @xmath281 , where @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 are subgroups of @xmath0 of order @xmath31 , @xmath51 , @xmath52 , @xmath32 , @xmath33 respectively and @xmath302 , @xmath140 , @xmath141 are elements of @xmath0 of order @xmath32 , @xmath33 respectively . moreover , @xmath1 contains @xmath37 as a subgraph with bipartition @xmath277 , @xmath160 , @xmath161 , @xmath303 and @xmath304 . * case 2 : * let each @xmath144 , @xmath274 and @xmath305 . suppose a sylow subgroup of @xmath0 is not unique , without loss of generality , we assume that sylow @xmath31-subgroup is not unique . then @xmath0 has at least three subgroups , say @xmath159 , @xmath160 , @xmath161 of order @xmath82 , and subgroups @xmath166 , @xmath280 of order @xmath32 , @xmath33 , respectively , and @xmath302 , @xmath140 , @xmath141 are elements of @xmath0 of order @xmath32 , @xmath33 respectively . then @xmath1 contains @xmath246 as a subgraph with bipartition @xmath277 , @xmath160 , @xmath278 and @xmath279 , @xmath280 , @xmath281 . also @xmath1 contains @xmath37 as a subgraph with bipartition @xmath277 , @xmath160 , @xmath161 , @xmath303 and @xmath304 . 1 . let @xmath0 be cyclic . if @xmath299 , then let @xmath284 , @xmath60 , 2 , @xmath61 , 6 be subgroups of @xmath0 of orders @xmath31 , @xmath32 , @xmath33 , @xmath92 , @xmath285 , @xmath287 respectively . then @xmath1 is planar and the corresponding plane embedding is shown in figure [ fig : f1 ] . + also @xmath1 is unicyclic ; it does not contain @xmath262 , @xmath38 as subgraphs ; but it contains @xmath39 as a subgraph . if @xmath306 , then let @xmath284 , @xmath60 , 2 , @xmath61 , 10 be subgroups of @xmath0 of orders @xmath31 , @xmath51 , @xmath32 , @xmath33 , @xmath92 , @xmath285 , @xmath287 , @xmath93 , @xmath307 , @xmath290 respectively . then @xmath1 is planar and the corresponding plane embedding is shown in figure [ fig : f5 ] . + also @xmath1 is not unicyclic ; it contains @xmath36 as a subgraph and does not contain @xmath37 . if @xmath308 or @xmath309 , then @xmath1 contains a subdivision of @xmath245 as shown in figure [ fig : f6 ] with vertices @xmath284 , @xmath60 , @xmath61 , 7 of order @xmath31 , @xmath51 , @xmath32 , @xmath199 , @xmath33 , @xmath285 , @xmath287 respectively . also @xmath1 contains @xmath37 as a subgraph with bipartition @xmath277 , @xmath160 , @xmath161 , @xmath310 and @xmath311 ; it contains @xmath36 as a subgraph with bipartition @xmath277 , @xmath312 and @xmath313 . 2 . let @xmath0 be non - cyclic abelian . here @xmath0 has atleast three subgroups of any one of prime order @xmath31 , @xmath32 , @xmath33 or @xmath34 , let it be @xmath31 . then @xmath1 contains @xmath246 as a subgraph with bipartition @xmath277 , @xmath160 , @xmath278 and @xmath279 , @xmath280 , @xmath281 , where @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 , @xmath282 are subgroups of @xmath0 of order @xmath31 , @xmath31 , @xmath31 , @xmath32 , @xmath33 , @xmath34 respectively . also @xmath1 contains @xmath37 as a subgraph with bipartition @xmath277 , @xmath160 , @xmath161 , @xmath314 and @xmath304 . * case 1 : * @xmath315 and @xmath144 , @xmath318 . let @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 , @xmath282 be subgroups of @xmath0 of order @xmath31 , @xmath51 , @xmath52 , @xmath32 , @xmath199 , @xmath319 respectively . then @xmath1 contains @xmath246 as a subgraph with bipartition @xmath320 and @xmath321 . if @xmath322 , then @xmath0 has subgroups of order @xmath31 , @xmath51 , @xmath52 , @xmath53 , @xmath32 , let them be @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 respectively . then @xmath1 contains @xmath37 as a subgraph with bipartition @xmath323 and @xmath311 . let @xmath324 . suppose sylow @xmath31-subgroup and sylow @xmath32-subgroup of @xmath0 are cyclic and are unique , then @xmath0 is cyclic and so @xmath1 is bipartite with one partition contains all subgroups whose orders has @xmath31 as a divisor and another partition contains remaining proper subgroups of @xmath0 . thus @xmath325 . it follows that @xmath1 is @xmath37-free . if sylow @xmath31-subgroup or sylow @xmath32-subgroup of @xmath0 is not cyclic , without loss of generality , we assume that sylow @xmath31-subgroup is not cyclic , then @xmath0 has subgroups of order @xmath31 , @xmath51 , @xmath52 , @xmath326 , @xmath32 , for some @xmath327 , let them be @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 respectively . then @xmath1 contains @xmath37 as a subgraph with bipartition @xmath323 and @xmath311 . if sylow @xmath31-subgroup or sylow @xmath32-subgroup is not unique , without loss of generality , we assume that sylow @xmath31-subgroup is not unique , then @xmath0 has subgroups of order @xmath31 , @xmath51 , @xmath52 , @xmath52 , @xmath32 , let them be @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 respectively . then @xmath1 contains @xmath37 as a subgraph with bipartition @xmath323 and @xmath311 . * case 2 * : @xmath315 , @xmath328 . let @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 , @xmath282 be subgroups of @xmath0 of order @xmath31 , @xmath51 , @xmath52 , @xmath53 , @xmath199 , @xmath32 respectively . then @xmath1 contains @xmath37 as a subgraph with bipartition @xmath323 and @xmath311 ; also @xmath1 contains @xmath36 as a subgraph with bipartition @xmath320 and @xmath329 . if sylow @xmath32-subgroup of @xmath0 is not unique , then @xmath0 has subgroups of order @xmath31 , @xmath51 , @xmath52 , @xmath199 , @xmath199 , @xmath32 , let them be @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 , @xmath282 respectively . then @xmath1 contains @xmath246 as a subgraph with bipartition @xmath320 and @xmath321 . if sylow @xmath32-subgroup of @xmath0 is unique and is isomorphic to @xmath330 , then @xmath0 has subgroups of order @xmath31 , @xmath51 , @xmath52 , @xmath199 , @xmath32 , @xmath32 , let them be @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 , @xmath282 respectively . then @xmath1 contains @xmath246 as a subgraph with bipartition @xmath320 and @xmath321 . if sylow @xmath32-subgroup of @xmath0 is unique and cyclic , then @xmath0 has an unique subgroups of order @xmath32 , @xmath199 , let them be @xmath159 , @xmath160 respectively . then @xmath1 is planar , since @xmath1 is bipartite with one partition contains all the subgroups whose order has @xmath31 as a divisor and another partition contains @xmath159 , @xmath160 . let @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 be subgroups of @xmath0 of order @xmath31 , @xmath51 , @xmath52 , @xmath32 , @xmath199 respectively . here @xmath1 contains @xmath333 as a subgraph with bipartition @xmath320 and @xmath334 . suppose sylow @xmath31-subgroup , and sylow @xmath32-subgroup of @xmath0 are cyclic , and are unique , then @xmath0 is cyclic and so @xmath1 is bipartite with one partition contains all subgroups whose order has @xmath31 as a divisor and another partition contains remaining subgroups of @xmath0 . thus @xmath335 . it follows that @xmath1 is @xmath37-free . if sylow @xmath31-subgroup of @xmath0 is not unique , then @xmath0 has subgroups of order @xmath52 , @xmath52 , @xmath51 , @xmath31 , @xmath199 , say @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 respectively . then @xmath1 contains @xmath37 as a subgraph with bipartition @xmath323 and @xmath311 . if sylow @xmath31-subgroup of @xmath0 is not cyclic , then @xmath0 has subgroups of order @xmath52 , @xmath326 , @xmath51 , @xmath31 , @xmath199 , for some @xmath327 , say @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 respectively . then @xmath1 contains @xmath37 as a subgraph with bipartition @xmath323 and @xmath311 . suppose sylow @xmath32-subgroup of @xmath0 is isomorphic to @xmath330 , then @xmath0 has subgroups of order @xmath52 , @xmath51 , @xmath31 , @xmath32 , @xmath32 , @xmath32 , @xmath199 , say @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 , @xmath282 , @xmath336 respectively . then @xmath1 contains @xmath246 as a subgraph with bipartition @xmath320 and @xmath321 and @xmath1 contains @xmath37 as a subgraph with bipartition @xmath337 and @xmath338 . suppose sylow @xmath32-subgroup of @xmath0 is not unique , then @xmath0 has subgroups of order @xmath52 , @xmath51 , @xmath31 , @xmath32 , @xmath199 , @xmath199 , @xmath199 , say @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 , @xmath282 , @xmath336 respectively . then @xmath1 contains @xmath246 as a subgraph with bipartition @xmath320 and @xmath321 and it contains @xmath37 as a subgraph with bipartition @xmath337 and @xmath338 . if sylow @xmath32-subgroup of @xmath0 is unique and it is cyclic , then @xmath0 has an unique subgroup of order @xmath32 , @xmath199 , let them be @xmath159 , @xmath160 respectively . then @xmath1 is planar , since @xmath1 is bipartite with one partition contains all the subgroups whose order has @xmath31 as a divisor and another partition contains @xmath159 , @xmath160 . it follows that @xmath1 is planar if and only if @xmath0 has a unique cyclic sylow @xmath32-subgroup . in @xcite myron owen tripp showed that up to isomophism , there are 15 such groups of order @xmath331 exist . 1 . @xmath1 is planar if and only if @xmath0 is isomorphic to one of @xmath253 , @xmath254 , @xmath255 or @xmath256 ; 2 . @xmath1 contains @xmath262 ; 3 . the following are equivalent : 1 . @xmath340 ; 2 . @xmath1 is @xmath37-free ; 3 . @xmath1 is @xmath36-free ; 4 . @xmath1 is @xmath38-free ; 5 . @xmath1 is unicyclic . * case 1 : * @xmath0 is non - abelian . we use the classification of groups of order @xmath339 given in @xcite . with out loss of generality , we assume that @xmath341 . let @xmath249 and @xmath342 denote a sylow @xmath31-subgroup and sylow @xmath32-subgroup of @xmath0 respectively . by sylow s theorem , @xmath343 . but @xmath344 is not possible , since @xmath345 . if @xmath346 , then @xmath347 , this implies that @xmath348 , which is true only when @xmath349 . when @xmath350 , then @xmath351 . if @xmath352 , then @xmath1 is planar , since it is bipartite with one partition contains all the subgroups of @xmath0 whose order has @xmath32 as a divisor , and another partition contains the subgroups of @xmath0 order @xmath31 , @xmath51 ; also @xmath0 has @xmath51 sylow @xmath32-subgroups . it follows that @xmath1 contains @xmath36 and @xmath37 as subgraphs and so @xmath1 is not unicyclic . if @xmath353 , then @xmath1 is planar , since it is bipartite with one partition contains all the subgroups whose order has @xmath32 as a divisor , and another partition contains the subgroups of @xmath0 order @xmath31 , @xmath51 ; also @xmath0 has @xmath51 sylow @xmath32-subgroups . it follows that @xmath1 contains @xmath36 and @xmath37 as a subgraph and so @xmath1 is not unicyclic . if @xmath354 , where @xmath355 has order @xmath199 in @xmath127 , then @xmath1 contains @xmath246 as a subgraph with bipartition @xmath277 , @xmath160 , @xmath278 and @xmath279 , @xmath280 , @xmath281 , where @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 , @xmath282 are subgroups of @xmath0 of order @xmath31 , @xmath31 , @xmath31 , @xmath199 , @xmath199 , @xmath32 respectively . also @xmath0 has @xmath51 sylow @xmath32-subgroups , so @xmath1 contains @xmath37 as a subgraph . if @xmath356 , then @xmath1 contains @xmath246 as a subgraph with bipartition @xmath277 , @xmath160 , @xmath278 and @xmath279 , @xmath280 , @xmath281 , where @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 , @xmath282 are subgroups of @xmath0 of order @xmath31 , @xmath31 , @xmath31 , @xmath32 , @xmath32 , @xmath32 respectively . also @xmath0 has @xmath51 sylow @xmath32-subgroups , so @xmath1 contains @xmath37 as a subgraph . @xmath357 . here @xmath0 has unique subgroups of order 3 and 9 respectively ; the order of remaining proper subgroups have 2 as their common divisor . therefore , @xmath1 is planar , since it is bipartite with one partition contains subgroups of order 3 , 9 , and another partition contains remaining proper subgroups of @xmath0 . since there are 18 subgroups of @xmath256 has order 2 , so @xmath1 contains @xmath36 and @xmath37 as a subgraph . 2 . @xmath358 . let @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 , @xmath282 , @xmath336 be subgroups of @xmath0 of order 3 , 3 , 9 , 2 , 2 , 2 , 2 respectively . then @xmath1 contains @xmath246 as a subgraph with bipartition @xmath320 and @xmath321 . also @xmath1 contains @xmath37 as a subgraph with bipartition @xmath337 and @xmath338 . 3 . @xmath359 . let @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 , @xmath282 , @xmath336 be subgroups of @xmath0 of order 3 , 3 , 9 , 2 , 2 , 2 , 4 respectively . then @xmath1 contains @xmath246 as a subgraph with bipartition @xmath320 and @xmath279 , @xmath280 , @xmath281 . also @xmath1 contains @xmath37 as a subgraph with bipartition @xmath337 and @xmath338 . 4 . @xmath360 . let @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 , @xmath282 , @xmath336 be subgroups of @xmath0 of order 3 , 3 , 9 , 2 , 2 , 2 , 2 respectively . here @xmath1 contains @xmath246 as a subgraph with bipartition @xmath320 and @xmath279 , @xmath280 , @xmath281 . also @xmath1 contains @xmath37 as a subgraph with bipartition @xmath337 and @xmath338 . 5 . @xmath361 , then @xmath1 is planar , since it is bipartite with one partition contains all the subgroups whose order has @xmath31 as a divisor , and another partition contains the subgroups of order @xmath32 , @xmath199 . it follows that @xmath1 contains @xmath36 as a subgraph . also @xmath0 has nine subgroups of order 4 . so @xmath1 contains @xmath37 as a subgraph with bipartition @xmath323 and @xmath311 , where @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 are subgroups of @xmath0 of order 4 , 4 , 4 , 2 , 3 respectively . let @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 , @xmath282 , @xmath336 be subgroups of @xmath0 of order 3 , 3 , 9 , 2 , 4 , 4 , 4 respectively . then @xmath1 contains @xmath246 as a subgraph with bipartition @xmath320 and @xmath279 , @xmath280 , @xmath281 . also @xmath1 contains @xmath37 as a subgraph with bipartition @xmath337 and @xmath338 . @xmath363 , where @xmath355 has order @xmath364 in @xmath365 . let @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 , @xmath282 , @xmath336 be subgroups of @xmath0 of order 3 , 3 , 3 , 9 , 2 , 4 , 4 respectively . here @xmath1 contains @xmath246 as a subgraph with bipartition @xmath320 and @xmath366 , @xmath282 , @xmath367 . also @xmath1 contains @xmath37 as a subgraph with bipartition @xmath323 and @xmath311 . let @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 , @xmath282 , @xmath336 be subgroups of @xmath0 of order 3 , 3 , 3 , 9 , 2 , 2 , 2 respectively . here @xmath1 contains @xmath246 as a subgraph with bipartition @xmath320 and @xmath366 , @xmath282 , @xmath367 . also @xmath1 contains @xmath37 as a subgraph with bipartition @xmath323 and @xmath311 . let @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 , @xmath282 , @xmath336 be subgroups of @xmath0 of order 3 , 3 , 9 , 2 , 2 , 2 , 4 respectively . here @xmath1 contains @xmath246 as a subgraph with bipartition @xmath320 and @xmath279 , @xmath280 , @xmath281 . also @xmath1 contains @xmath37 as a subgraph with bipartition @xmath337 and @xmath311 . if @xmath371 , then @xmath1 is planar , since @xmath1 is bipartite with one partition contains all the subgroups whose order has @xmath31 as a divisor , and another partition contains the subgroups of order @xmath32 , @xmath199 . it follows that @xmath1 contains @xmath36 as a subgraph . also @xmath1 contains @xmath37 as a subgraph with bipartition @xmath323 and @xmath311 , where @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 are subgroups of @xmath0 of order @xmath32 , @xmath32 , @xmath32 , @xmath199 , @xmath31 respectively . if @xmath372 , then @xmath1 contains @xmath246 as a subgraph with bipartition @xmath277 , @xmath160 , @xmath278 and @xmath279 , @xmath280 , @xmath281 , where @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 , @xmath282 , @xmath336 are subgroups of @xmath0 of order @xmath31 , @xmath31 , @xmath31 , @xmath32 , @xmath32 , @xmath32 , @xmath199 respectively . also @xmath1 contains @xmath37 as a subgraph with bipartition @xmath337 and @xmath338 . 1 . @xmath1 is planar if and only if @xmath0 is one of @xmath374 , @xmath247 , where @xmath249 is a @xmath31-group , @xmath250 , @xmath251 , @xmath252 or @xmath133 ; 2 . the following are equivalent : 1 . @xmath0 is one of @xmath374 , @xmath247 , where @xmath249 is a @xmath31-group , @xmath250 , @xmath251 or @xmath133 ; 2 . @xmath1 is @xmath36-free ; 3 . @xmath1 is @xmath262-free . @xmath1 is @xmath37-free if and only if @xmath0 is one of @xmath263 or @xmath266 . @xmath1 is @xmath38-free if and only if @xmath0 either @xmath76 or @xmath375 ; 5 . @xmath1 is @xmath39-free if and only if @xmath80 ; 6 . @xmath1 is acyclic . suppose sylow @xmath32-subgroup of @xmath0 is not unique , then @xmath0 has subgroups of order @xmath31 , @xmath51 , @xmath52 , @xmath32 , @xmath32 , @xmath32 , let them be @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 , @xmath282 respectively . here @xmath1 contains @xmath246 as a subgraph with bipartition @xmath320 and @xmath321 . suppose sylow @xmath32-subgroup is unique , then @xmath376 or @xmath377 , where @xmath249 is a sylow @xmath31-subgroup of @xmath0 and so @xmath1 is bipartite with one partition contains a subgroup whose order is @xmath32 and another partition contains remaining proper subgroups of @xmath0 . hence @xmath1 is planar and @xmath262-free . now we check the @xmath37-freeness of @xmath1 . if @xmath322 , then @xmath0 has subgroups of order @xmath31 , @xmath51 , @xmath52 , @xmath53 , @xmath32 , let them be @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 respectively . then @xmath1 contains @xmath37 as a subgraph with bipartition @xmath323 and @xmath311 . let @xmath324 . suppose sylow @xmath31-subgroup of @xmath0 is not unique , then @xmath0 has subgroups of order @xmath31 , @xmath51 , @xmath52 , @xmath52 , @xmath32 , let them be @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 respectively . then @xmath1 contains @xmath37 as a subgraph with bipartition @xmath323 and @xmath311 . suppose sylow @xmath31-subgroup of @xmath0 is unique and cyclic , then @xmath0 has a unique subgroups of order @xmath31 , @xmath51 , @xmath52 , let them be @xmath159 , @xmath160 , @xmath161 , and so @xmath1 is a bipartite graph with one partition contains @xmath159 , @xmath160 , @xmath161 and another partition contains remaining proper subgroups of @xmath0 . hence @xmath1 is @xmath37-free . but @xmath1 contains @xmath38 as a subgraph with bipartition @xmath320 and @xmath304 , where @xmath166 is a subgroup of order @xmath32 . suppose @xmath0 is cyclic , then @xmath0 has only four proper subgroups and their orders are @xmath51 , @xmath31 , @xmath32 , @xmath92 respectively . therefore , @xmath380 , which is planar and @xmath39-free . if @xmath0 is non - cyclic abelian , then @xmath0 has @xmath129 subgroups of order @xmath31 ; unique subgroup of order @xmath51 ; unique subgroup of order @xmath32 ; @xmath129 subgroups of order @xmath92 . also these are the only proper subgroups of @xmath0 . it follows that @xmath381 and hence @xmath1 is planar ; it contains @xmath37 as a subgraph and it is @xmath262-free . 1 . from the structure of @xmath100 , it follows that @xmath100 is planar , @xmath262-free and it contains @xmath37 . 2 . note that @xmath41 has unique subgroup of order @xmath32 . @xmath103 is planar , since @xmath103 is bipartite with one partition contains all the subgroups whose order has @xmath31 as a divisor , and another partition contains the subgroup of order @xmath32 . so @xmath103 contains @xmath37 as a subgraph with bipartition @xmath323 and @xmath311 , where @xmath159 , @xmath160 , @xmath161 , @xmath166 , @xmath280 are subgroups of @xmath0 of order @xmath31 , @xmath31 , @xmath31 , @xmath51 , @xmath32 respectively . clearly @xmath103 is @xmath262-free . 3 . from the structure of @xmath108 , it follows that @xmath108 is planar , @xmath262-free and it contains @xmath37 . 4 . from the structure of @xmath115 , it follows that @xmath115 is planar , and it contains @xmath36 , @xmath37 as a subgraph . @xmath121 has at least three subgroups of order @xmath31 , and at least three subgroups of order @xmath32 . so @xmath123 contains @xmath246 as a subgraph with bipartition @xmath320 and @xmath321 , where @xmath284 , @xmath60 , 2 , 3 and @xmath382 , @xmath383 , 5 , 6 are subgroups of @xmath0 of order @xmath31 , @xmath31 , @xmath31 , @xmath32 , @xmath32 , @xmath32 respectively . also @xmath123 contains @xmath37 as a subgraph with bipartition @xmath384 and @xmath304 , where @xmath336 is a subgroup of @xmath0 of order @xmath51 . 6 . from the structure of @xmath385 , it follows that @xmath385 contains @xmath246 as a subgraph , so it is non - planar ; also it contains @xmath37 as a subgraph . 7 . from the structure of @xmath386 , it follows that @xmath386 is planar , @xmath262-free and it contains @xmath37 as a subgraph . also from the structure of @xmath137 , it follows that @xmath137 contains @xmath246 and @xmath37 as a subgraph . * case 3 : * let @xmath387 . then @xmath0 has a unique subgroup of order @xmath32 , let it be @xmath2 ; it has @xmath32 subgroups of order @xmath31 ; also these are the only proper subgroups of @xmath0 . it follows that @xmath90 , which is planar , acyclic and @xmath262-free ; it is @xmath37-free if and only if @xmath388 ; it contains @xmath38 . if @xmath0 is a @xmath31-group , then order of every subgroup of @xmath0 is power of @xmath31 and so no two subgroups are adjacent in @xmath1 . thus @xmath1 is totally disconnected and this completes the proof .

a calculation of the bulk viscosity for the massive gross - neveu model at zero fermion chemical potential is presented in the large- limit . this model resembles qcd in many important aspects : it is asymptotically free , has a dynamically generated mass gap , and for zero bare fermion mass it is scale invariant at the classical level ( broken through the trace anomaly at the quantum level ) . for our purposes , the introduction of a bare fermion mass is necessary to break the integrability of the model , and thus to be able to study momentum transport . the main motivation is , by decreasing the bare mass , to analyze whether there is a correlation between the maximum in the trace anomaly and a possible maximum in the bulk viscosity , as recently conjectured . after numerical analysis , i find that there is no direct correlation between these two quantities : the bulk viscosity of the model is a monotonously decreasing function of the temperature . i also comment on the sum rule for the spectral density in the bulk channel , as well as on implications of this analysis for other systems .

1009.2741

transport coefficients are essential inputs to describe the space - time evolution of systems not far from equilibrium . during the last years there has been a very active effort to analyze them from both the theoretical and phenomenological points of view in the context of heavy - ion collisions , condensed matter physics , astrophysics and cosmology . the calculation of transport coefficients in quantum field theory at intermediate and strong coupling is still a challenge from both the analytical and the numerical points of view . due to their intrinsic non - perturbative nature , even in weakly interacting theories a resummation of an infinite number of diagrams is needed in order to obtain the leading - order result . in the strongly coupled regime , the most prominent method available is the ads / cft correspondence , although it is only applicable to a limited class of field theories . on the other hand , lattice simulations are still not accurate enough regarding the calculation of spectral densities , and the introduction of a finite quark chemical potential makes things even more difficult because of the sign problem . it was recently conjectured , based on a sum rule for the spectral density of the trace of the energy - momentum tensor in yang - mills theory @xcite , that a maximum of the trace anomaly near the critical temperature might drive a maximum for the bulk viscosity near that temperature . the corresponding sum rule was later corrected in @xcite , and the ansatz for the spectral density used to extract the bulk viscosity questioned @xcite . since the trace anomaly measures the breaking of scale invariance in a system , and the bulk viscosity @xmath1 essentially represents the difficulty for a system to relax back to equilibrium after a scale transformation , it seems in principle reasonable to think that @xmath1 would be maximum when the breaking of scale invariance is maximum . in heavy - ion phenomenology , bulk viscosity has usually been neglected because it is expected to be much smaller than the shear viscosity even at temperatures not very high @xcite . however , as suggested by the analysis of @xcite , non - perturbative phenomena responsible for the main contribution to the trace anomaly near @xmath2 could also produce a significant increase in the bulk viscosity . in this paper i will present an explicit calculation in the massive gross - neveu model in @xmath3 dimensions , where the correlation between trace anomaly and bulk viscosity can be accurately tested . i will not try to give an estimation for the absolute value of @xmath1 in qcd near the phase transition though ; as we will see this model is not suitable for that purpose . there are several works analyzing this issue employing different approaches ( see for instance @xcite and references therein ) , but still the order of magnitude of the bulk viscosity near the crossover temperature is uncertain . in 1 + 1 dimensions , transverse flow of momentum is not possible , and the bulk viscosity is the only viscous coefficient present to linear order in gradients . in this paper i will analyze only finite temperature effects , considering a vanishing fermion chemical potential , thus the thermal conductivity will be zero in this case . therefore , the only constitutive equation relevant for us is . ] @xmath4 with @xmath5 the non - equilibrium expectation value for the spatial component of the energy - momentum tensor , @xmath6 is the pressure in equilibrium , and @xmath7 is the fluid velocity . the bulk viscosity can be in principle calculated perturbatively in field theory @xcite : @xmath8 where @xmath9 is the spectral density corresponding to the thermal propagator @xmath10 . here though , i will use a kinetic theory approach , which should be equivalent to the diagrammatic one in the perturbative ( and dilute ) regime @xcite . this paper is organized as follows . in sections [ sec : mgnvac ] and [ sec : mgntemp ] , i review well known properties of the massive gross - neveu model at zero and finite temperature , and i prove the breaking of integrability in the large-@xmath0 limit when a mass term for the fermion field is explicitly introduced . then in section [ sec : kinetictheory ] , the calculation of the bulk viscosity within kinetic theory is presented . in section [ sec : discussion ] , i comment on sum rules and implications of the previous analysis for other systems . finally in [ sec : conclusions ] i summarize the main conclusions . there is also the appendix [ app : fermions ] , where the result of factorization for fermion loops in @xmath3 dimensions is derived , and appendix [ app : amplitudes ] where the reader can find some details on the calculation of the inelastic scattering amplitude . let s consider the gross - neveu model @xcite with an explicit bare mass for the fermion field : @xmath11 since we are interested in studying the large @xmath0 limit of the model , in order for the perturbative expansion in powers of @xmath12 to be sensible , the bare coupling constant must be re - scaled , @xmath13 , with @xmath14 being constant as @xmath15 . also , it is convenient to introduce an auxiliary field @xmath16 to properly classify the different feynman diagrams according to their topologies and power counting in @xmath12 @xcite : @xmath17 clearly , the introduction of this field does not affect the dynamics of the system because its equation of motion is simply @xmath18 . in terms of the auxiliary field , the discrete chiral symmetry then corresponds to the simultaneous transformations @xmath19 and @xmath20 . in @xmath3 space - time dimensions and in the large-@xmath0 limit , this model shares many important features with massless qcd in @xmath21 dimensions : it is renormalizable , asymptotically free , classically scale invariant ( for zero bare fermion mass ) , it has a dynamically generated mass gap which manifests as a peak in the trace anomaly , and in vacuum undergoes an spontaneous breaking of the discrete `` chiral '' symmetry which is spontaneously broken in vacuum down to @xmath22 . ] @xmath19 . as we will see in the next subsection , the introduction of this bare mass @xmath23 is a simple way of allowing the system to relax back to thermodynamic equilibrium after a small perturbation in the distribution of momenta . in addition , the bare mass also suppresses the density of kink - anti - kink configurations in the thermodynamic limit and makes the mean - field @xmath12 expansion well defined @xcite . to leading order in the large-@xmath0 limit , only one counter - term is necessary to renormalize all the divergences , @xmath24 , which essentially amounts to a renormalization of the coupling constant . the effective potential for the classical field @xmath25 is obtained using standard techniques and renormalized imposing the condition @xmath26 @xcite , with @xmath27 the renormalization scale . this fixes the counter - term to be @xmath28\ , \ ] ] with @xmath29 an ultraviolet cutoff . then , the leading - order renormalized effective potential is @xmath30\ .\ ] ] this is a mexican - hat potential ( tilted by the mass @xmath23 ) with a non - zero mass gap @xmath31 determined by the condition @xmath32=0\ .\ ] ] if we define @xmath33 , we can now use ( [ eq : gapeq0 ] ) to write the effective potential in a scale - independent form : @xmath34\ .\end{aligned}\ ] ] as shown in fig . [ fig : potential ] , for @xmath35 the discrete chiral symmetry is spontaneously broken by choosing as vacuum one of the two minima . for small enough values of @xmath23 , the potential still has two minima , whereas for larger @xmath23 one disappears and the other one becomes deeper ( i use the units @xmath36 ) . for different values of @xmath23.,width=302 ] the effective potential ( [ eq : renpot ] ) satisfies the renormalization - group equation @xmath37v_\mathrm{eff}^\mathrm{r}(\sigma_\mathrm{c},g,\sigma_0)=0\ , \end{aligned}\ ] ] which implies @xmath38 i.e. , the theory is asymptotically free . although the running coupling constant becomes arbitrarily large at low energies , the interaction between fermions is also suppressed by powers of @xmath12 , thus in the large-@xmath0 limit we are still able to probe the low - energy regime of the theory . the leading - order contribution to the self - energy of the @xmath16 field corresponds to the diagram depicted in fig . [ fig : self - energy ] . in euclidean space , the expression for the ( renormalized ) @xmath16 propagator in vacuum is @xmath39^{-1}=g^2n\frac{m}{m_0}+\frac{g^2n}{2\pi}\beta(p^2)\ln\left[\frac{\beta(p^2)+1}{\beta(p^2)-1}\right]\ , \end{aligned}\ ] ] with @xmath40 a phase - space factor , and @xmath41 . field to leading order , @xmath42.,width=226 ] in the next subsection , i show how the first term in ( [ eq : sigmapvacuum ] ) breaks the integrability of the model in the large-@xmath0 limit . the gross - neveu model ( without the bare mass ) is an integrable quantum field theory @xcite , this implies the existence of an infinite number of conserved charges and @xmath3 dimensions the factorization of the @xmath43-matrix in terms of binary collisions , so inelastic processes have vanishing scattering amplitude . since in @xmath3 dimensions binary collisions can not modify the distribution of momenta , integrability then prevents momentum transport in this system . consequently , the bulk viscosity of the gross - neveu model is infinite . after including the bare mass in the model , this factorization in terms of binary collisions no longer happens , and hence it renders the bulk viscosity finite . limit to the inelastic process @xmath44.,width=453 ] to see this , consider the leading - order diagrams corresponding to the inelastic process @xmath45 in fig . [ fig : leadinginel ] . as it was shown in @xcite , the fermion loop of fig . [ fig : leadinginel](g ) factorizes into tree diagrams corresponding to all the possible ways of cutting it ( figs . [ fig : diaggcut ] and [ fig : diaggcut1 ] ) . one particular cut is depicted in fig . [ fig : diaggcut1 ] . from the result ( [ eq : deff ] ) derived in appendix [ app : fermions ] , it is easy to see that the four - point amplitude and the factor @xmath46 in fig . [ fig : diaggcut1 ] cancel out giving a @xmath47 factor . hence , the diagram of fig . [ fig : diaggcut1 ] exactly cancels ( when @xmath35 ) the one of fig . [ fig : leadinginel](a ) , and the same for the rest of diagrams . if we now introduce the mass @xmath23 , from ( [ eq : sigmapvacuum ] ) we see that this cancellation can not happen , so the total inelastic amplitude is now @xmath48 to leading order in @xmath12 . this proves the non - integrability of the massive gross - neveu model in the large-@xmath0 limit . ( g ) expressed in terms of different cuts according to eq . ( [ eq : fermioncut]).,width=415 ] the thermodynamic properties of this model have been studied in detail in many papers , see for instance @xcite and references therein . in this section , i am simply going to review leading order - results in the mean - field approximation , which are relevant for the later analysis of the bulk viscosity . the leading - order renormalized effective potential at finite temperature is @xmath49- \frac{2 n t}{\pi}\int\limits_0^\infty\mathrm{d}k\ \ln\left(1+\mathrm{e}^{-\sqrt{k^2+g^2 \sigma_\mathrm{c}^2}/t}\right)\ .\ ] ] the thermal mass gap is defined by @xmath50 where ( [ eq : gapeq0 ] ) has been used , @xmath51 , and @xmath52 is the fermi - dirac distribution function . in fig . [ fig : thermalmass ] , i plot the fermion mass gap as a function of the temperature for different values of @xmath23 . for the case @xmath35 , the mass gap vanishes at the temperature @xmath53 , indicating restoration of the discrete chiral symmetry . this is however an artifact of the mean - field approximation ; the chiral symmetry is actually immediately restored at @xmath54 due to kink - anti - kink configurations is taken first @xcite . ] . nevertheless , as mentioned above , the introduction of a finite bare mass suppresses these kink - anti - kink configurations in the thermodynamic limit , and therefore we can approach in the mean - field approximation the curve @xmath35 as much as we wish provided we keep @xmath23 finite and thermodynamic limits are taken first keeping @xmath23 finite , and afterwards we study the limit @xmath55 . ] . .,width=302 ] the pressure is immediately obtained from the effective potential : @xmath56\notag\\ & + \frac{2 n t}{\pi}\int\limits_0^\infty\mathrm{d}k\ \ln\left(1+\mathrm{e}^{-\sqrt{k^2+m(t)^2}/t}\right)\ , \end{aligned}\ ] ] and the `` bag pressure '' is @xmath57 entropy , energy density , specific heat , speed of sound , and trace anomaly are calculated from the pressure using the thermodynamic relations @xmath58 these are plotted in figs . [ fig : pressure ] , [ fig : entropy ] , [ fig : energydens ] , [ fig : specificheat ] , [ fig : speedsound ] , [ fig : traceanom ] . we see that the trace anomaly has a very pronounced peak right at @xmath2 for @xmath59 , which will allow us to study the possible correlation with the bulk viscosity . for @xmath59 , above @xmath2 the pressure corresponds to an ideal gas of massless fermions : @xmath60 with @xmath61 . . the color code is the same as in fig . [ fig : thermalmass].,width=302 ] , @xmath62 has a discontinuity at @xmath63.,width=302 ] , @xmath64 has a discontinuity at @xmath63.,width=302 ] in order to calculate dynamical quantities , it is convenient to shift @xmath65 , so tadpole diagrams vanish and have not to be taken into account . the @xmath16 propagator to leading order and at finite temperature , calculated from the diagram of fig . [ fig : self - energy ] in the imaginary - time formalism and continued to real frequencies , is @xmath66^{-1}=\frac{m}{m}+\frac{1}{2\pi}\left\{[\theta(-s)+\theta(s-4 m^2)]\beta(s)\ln\left|\frac{\beta(s)+1}{\beta(s)-1}\right|\right.\notag\\ & \left.+\theta(s)\theta(4m^2-s)2b(s)\arctan\left(\frac{1}{b(s)}\right)\right\}+\pv\int\limits_{-\infty}^\infty\frac{\mathrm{d}k}{\pi}\ \frac{n_\mathrm{f}(e_k)}{e_k}\frac{s\beta(s)^2(2pk - s)}{(2pk - s)^2 - 4e_k^2\omega^2 } \notag\\ & -\frac{\mathrm{i}}{2}\sgn(\omega)\beta(s)\left\{\theta(s-4m^2)[1-n_\mathrm{f}(\epsilon_+)-n_\mathrm{f}(\epsilon_-)]+ \theta(-s)|n_\mathrm{f}(\epsilon_+)-n_\mathrm{f}(\epsilon_-)|\right\}\ , \end{aligned}\ ] ] where @xmath67 , @xmath68 , @xmath69 , and @xmath70 . the first term in ( [ eq : sigmapt ] ) is responsible of breaking the integrability of the model also at finite temperature , which follows from the result ( [ eq : fermioncut ] ) in appendix [ app : fermions ] in the same way as for the vacuum case analyzed in the previous section . we realize that the breaking of integrability is now controlled by the factor @xmath71 , instead of @xmath72 . interestingly , for @xmath73 , the limit of @xmath71 as @xmath55 is not zero , but a ( temperature - dependent ) constant . thus , the scattering amplitude for inelastic processes is always finite when @xmath59 for @xmath73 . the massive gross - neveu model is a non - confining theory and , as we have seen in the previous sections , the interaction between the fundamental fermions is suppressed by powers of @xmath12 , hence in principle it seems reasonable to adopt a kinetic theory treatment to analyze the transport properties of this system in the large-@xmath0 limit . alternatively , one could formally work out the resummation of an infinite series of ladder and chain diagrams contributing the spectral density of the energy - momentum tensor when the external frequency goes to zero . however , it is known that this resummation leads to solving an integral equation which coincides with the boltzmann equation in the effective kinetic theory describing thermal excitations in the system @xcite . i consider that the kinetic theory approach is simpler , and i will employ it for the calculation of the bulk viscosity in this paper dimensions only in the limit @xmath15 , where the long - time tail in the energy - momentum tensor correlator @xmath74 becomes negligible @xcite . otherwise , for @xmath0 finite , the bulk viscosity of the massive gross - neveu model would be infinite . ] . i am going to follow essentially the previous works @xcite so , although i try to keep the discussion self - contained , the reader is referred to these papers for additional details . in order to obtain the bulk viscosity , we need to determine the statistical average of the energy - momentum tensor of the system in a cell of fluid for a small departure from equilibrium . in kinetic theory , this average is @xcite @xmath75 where @xmath76 is the non - equilibrium distribution function , @xmath77 is a collective index denoting the fermionic or anti - fermionic character and the flavor of the corresponding particle species , @xmath78 , and @xmath79 is the canonical momentum ( the underline emphasizes that it is on - shell ) . the boltzmann - uehling - uhlenbeck equation determines the space - time evolution of distribution functions for dilute systems due to the change in the number of particles of type @xmath77 produced by collisions in the fluid . in 1 + 1 dimensions it reads @xmath80\ , \end{aligned}\ ] ] with @xmath81 a column vector containing the distribution functions for every type of particle . considering only the leading - order elastic and inelastic processes in the large-@xmath0 expansion , which in our case are @xmath82 or @xmath83 , and @xmath84 or @xmath85 respectively , the collision term is given by @xmath86\,\frac{\mathrm{d}p_1}{2\pi e_1}=\sum\limits_{b , c , d}\,\int\limits_{2,3,4}\frac{\mathrm{d}p_1}{2\pi}\frac{\mathrm{d}p_2}{2\pi}\ , \mathrm{d}\mathnormal{\gamma}_{1 2\rightarrow 3 4}^{a , b;c , d } l\ , s_{(1 ) 2\rightarrow 3 4}^{a , b;c , d}\ , [ f^c_3 f^d_4 ( 1-f^a_1)(1-f^b_2)\notag\\ & -f^a_1 f^b_2 ( 1-f^c_3)(1-f^d_4)]+\sum\limits_{b , c , d , e , f}\bigg\{\int\limits_{2,3,4,5,6}\frac{\mathrm{d}p_5}{2\pi}\frac{\mathrm{d}p_6}{2\pi}\ , \mathrm{d}\mathnormal{\gamma}_{5 6\rightarrow 1 2 3 4}^{e , f;a , b , c , d } l\ , s_{5 6\rightarrow ( 1 ) 2 3 4}^{e , f;a , b , c , d}\ , \notag\\ & \times [ f^e_5 f^f_6 ( 1-f^a_1)(1-f^b_2)(1-f^c_3)(1-f^d_4)-f^a_1 f^b_2 f^c_3 f^d_4 ( 1-f^e_5)(1-f^f_6)]\notag\\ & + \int\limits_{2,3,4,5,6}\frac{\mathrm{d}p_1}{2\pi}\frac{\mathrm{d}p_2}{2\pi}\ , \mathrm{d}\mathnormal{\gamma}_{1 2\rightarrow 3 4 5 6}^{a , b;c , d , e , f } l\ , s_{(1 ) 2\rightarrow 3 4 5 6}^{a , b;c , d , e , f}\ , [ f^c_3 f^d_4 f^e_5 f^f_6 ( 1-f^a_1)(1-f^b_2)\notag\\ & -f^a_1 f^b_2 ( 1-f^c_3)(1-f^d_4)(1-f^e_5)(1-f^f_6)]+\int\limits_{2,3,4,5,6}\frac{\mathrm{d}p_1}{2\pi}\frac{\mathrm{d}p_2}{2\pi}\frac{\mathrm{d}p_3}{2\pi}\ , \mathrm{d}\mathnormal{\gamma}_{1 2 3\rightarrow 4 5 6}^{a , b , c;d , e , f } l^2\ , s_{(1 ) 2 3\rightarrow 4 5 6}^{a , b , c;d , e , f}\notag\\ & \times [ f^d_4 f^e_5 f^f_6 ( 1-f^a_1)(1-f^b_2)(1-f^c_3)-f^a_1 f^b_2 f^c_3 ( 1-f^d_4)(1-f^e_5)(1-f^f_6)]\,\bigg\}\ , \label{eq : collision1}\end{aligned}\ ] ] where the sum over indices runs over all the possible configurations of fermion , anti - fermion , and flavor states . the symmetry factors ( to be specified later ) @xmath87 , @xmath88 , @xmath89 , and @xmath90 avoid the double counting from relabeling of momenta for identical particles ( except for the particle denoted as ` 1 ' ) in the integral and considering equivalent processes after summing over all the fermion types . the transition rate for an arbitrary process @xmath91 is given in terms of the scattering amplitude @xmath92 by @xcite @xmath93 \left[\prod\limits_\beta\frac{\mathrm{d}p_\beta}{(2\pi ) 2 e_\beta}\right]\ , |\mathcal{m}(\alpha\rightarrow\beta)|^2 ( 2\pi)^2\delta^{(2)}(\sum_\alpha\underline{p}{}_\alpha-\sum_\beta\underline{p}{}_\beta)\ , \end{aligned}\ ] ] where @xmath94 is the size of the system ( although we consider the limit @xmath95 ) , and @xmath96 the number of particles in the initial state . in order to obtain an expression for the bulk viscosity we need to solve ( [ eq : boltzmann ] ) for small departures from equilibrium . to do it , we first write @xmath97 where @xmath98 is small , and the fermion or anti - fermion distribution function at equilibrium for zero chemical potential is @xmath99 with @xmath100 the local temperature , and @xmath101 the velocity of the corresponding fluid cell . expanding the left - hand side of ( [ eq : boltzmann ] ) in the local rest frame ( @xmath102 ) to linear order in spatial derivatives , we obtain , @xmath103 . also , from the conservation of the energy - momentum tensor , @xmath104 , applied to leading order to the perfect fluid , @xmath105 , we derive the relations in the local rest frame ( @xmath106 ) : @xmath107 , @xmath108 . ] @xmath109\beta\left[\left(e_k - t\frac{m}{e_k}\frac{\mathrm{d}m}{\mathrm{d}t}\right ) c_\mathrm{s}^2-\frac{k^2}{e_k}\right]\frac{\partial u^1}{\partial x}\ , \end{aligned}\ ] ] where @xmath110 is the speed of sound in the fluid . consequently , the deviation from equilibrium can be written in the form @xmath111\mathcal{b}^a_k\,\frac{\partial u^1}{\partial x } , \end{aligned}\ ] ] with @xmath112 some dimensionless function to be determined by solving the integral equation obtained after the previous linearization of both sides of ( [ eq : boltzmann ] ) : @xmath113\ , |\mathcal{m}_{a , b}^{c , d}(p_1,p_2;p_3,p_4)|^2\notag\\ & \times s_{(1 ) 2\rightarrow 3 4}^{a b ; c d}(2\pi)^2\delta^{(2)}(\underline{p}{}_1+\underline{p}{}_2-\underline{p}{}_3-\underline{p}{}_4)n_\mathrm{f,2}(1-n_\mathrm{f,3 } ) ( 1-n_\mathrm{f,4})(\mathcal{b}^a_1+\mathcal{b}^b_2-\mathcal{b}^c_3-\mathcal{b}^d_4)\notag\\ & + \sum\limits_{b , c , d , e , f}\,\int\limits_{-\infty}^\infty\left[\prod\limits_{i=2}^6\frac{\mathrm{d}p_i}{(2\pi)2e_i}\right ] \bigg\{\,|\mathcal{m}_{e , f}^{a , b , c , d}(p_5,p_6;p_1,p_2,p_3,p_4)|^2 s_{5 6\rightarrow ( 1 ) 2 3 4}^{e f ; a b c d}\notag\\ & \times(2\pi)^2\delta^{(2)}(\underline{p}{}_5+\underline{p}{}_6-\underline{p}{}_1-\underline{p}{}_2-\underline{p}{}_3-\underline{p}{}_4)n_\mathrm{f,2 } n_\mathrm{f,3}n_\mathrm{f,4}(1-n_\mathrm{f,5})(1-n_\mathrm{f,6})\notag\\ & \times(\mathcal{b}^a_1+\mathcal{b}^b_2+\mathcal{b}^c_3+\mathcal{b}^d_4-\mathcal{b}^e_5-\mathcal{b}^f_6)+|\mathcal{m}_{a , b}^{c , d , e , f } ( p_1,p_2;p_3,p_4,p_5,p_6)|^2 s_{(1 ) 2\rightarrow 3 4 5 6}^{a b ; c d e f } ( 2\pi)^2\notag\\ & \times\delta^{(2)}(\underline{p}{}_1+\underline{p}{}_2-\underline{p}{}_3-\underline{p}{}_4-\underline{p}{}_5-\underline{p}{}_6)n_\mathrm{f,2 } ( 1-n_\mathrm{f,3})(1-n_\mathrm{f,4})(1-n_\mathrm{f,5})(1-n_\mathrm{f,6})\notag\\ & \times(\mathcal{b}^a_1+\mathcal{b}^b_2-\mathcal{b}^c_3-\mathcal{b}^d_4-\mathcal{b}^e_5-\mathcal{b}^f_6)+|\mathcal{m}_{a , b , c}^{d , e , f } ( p_1,p_2,p_3;p_4,p_5,p_6)|^2 s_{(1 ) 2 3\rightarrow 4 5 6}^{a b c ; d e f}\notag\\ & \times(2\pi)^2\delta^{(2)}(\underline{p}{}_1+\underline{p}{}_2+\underline{p}{}_3-\underline{p}{}_4-\underline{p}{}_5-\underline{p}{}_6)n_\mathrm{f,2 } n_\mathrm{f,3}(1-n_\mathrm{f,4})(1-n_\mathrm{f,5})(1-n_\mathrm{f,6})\notag\\ & \times(\mathcal{b}^a_1+\mathcal{b}^b_2+\mathcal{b}^c_3-\mathcal{b}^d_4-\mathcal{b}^e_5-\mathcal{b}^f_6)\bigg\}\bigg\}\ .\label{eq : integral1}\end{aligned}\ ] ] we can interpret the right - hand side of ( [ eq : integral1 ] ) as the action of a linear operator @xmath114 over a function in the space of solutions of the transport equation , and we split this operator into two terms , @xmath115 , corresponding to elastic and inelastic processes respectively . at this point , an important simplification is in order . since the source term in ( [ eq : source ] ) is invariant under charge conjugation , and the theory is symmetric under @xmath116-flavor rotations ( this symmetry can not be broken in @xmath3 dimensions @xcite ) , then the departures from equilibrium are the same for all the particle types , i.e. , @xmath117 . furthermore , the @xmath118-function in the elastic @xmath119 term of the collision integral implies in @xmath3 dimensions that the final set of momenta are the same as the initial , i.e. , @xmath119 elastic collisions in @xmath3 dimensions can not relax back to equilibrium a perturbation in the distribution of momenta . thus , @xmath120 and therefore , to leading order in the large-@xmath0 expansion , @xmath121 . once we know @xmath122 , from ( [ eq : deviation ] ) , ( [ eq : const11 ] ) , and ( [ eq : emtensor ] ) it is straightforward to obtain the bulk viscosity:\mathcal{b}^a(\underline{k})\ , \end{aligned}\ ] ] imposed to make the decomposition ( [ eq : fdecomp ] ) unique , is also used here @xcite . ] @xmath123 \left[k^2-c_\mathrm{s}^2\left(e_k^2-tm\frac{\mathrm{d}m}{\mathrm{d}t}\right)\right ] \mathcal{b}^a(\underline{k})\ .\label{eq : bulk1}\end{aligned}\ ] ] the linearized version of the boltzmann equation ( [ eq : integral1 ] ) can be written as @xmath124 denotes the source term . defining the scalar product of two square - integrable functions as @xmath125\,\chi^a(\underline{k})\psi^a(\underline{k})\ , \end{aligned}\ ] ] then the bulk viscosity is given by @xmath126 as shown in @xcite , in order to calculate numerically this expectation value for the inverse of the collision operator , it is optimal to do it variationally . if we define the functional @xmath127\equiv\langle\chi|\mathcal{s}\rangle-\frac{1}{2}\langle\chi|\hat{\mathscr{c}}|\chi\rangle\ , \end{aligned}\ ] ] then the solution of ( [ eq : boltzmannkets ] ) corresponds to a maximum in this functional , @xmath128 . hence , the bulk viscosity is proportional to this maximum : @xmath129 we now expand the solution for the boltzmann equation in terms of a given set of @xmath130 linearly independent functions : @xmath131 then @xmath132=\sum\limits_{i=1}^n b_i s_i-\frac{1}{2}\sum\limits_{i , j=1}^n b_i c_{ij } b_j\ , \end{aligned}\ ] ] where @xmath133 maximizing ( [ eq : qcoeff ] ) with respect to the set of coefficients @xmath134 , implies @xmath135 ( a tilde denotes matrices ) , and therefore @xmath136 it is important to notice , from ( [ eq : integral1 ] ) , that the collision operator has one zero - mode @xmath137 corresponding to energy conservation , i.e. , @xmath138.in addition to @xmath139 , the elastic collision operator also has the zero - mode @xmath140 corresponding to the conservation of the total number of particles in these type of processes . however , this is not a zero - mode of the inelastic part of the collision integral and therefore we do not have to worry about it when calculating @xmath141 . the presence of this other zero - mode , though , implies that the bulk viscosity is dominated by inelastic processes at very low temperatures due to fermi - dirac factors : @xmath142 on the other hand , at temperatures close to @xmath2 , the fermi - dirac factors are @xmath143 and both @xmath144 and @xmath145 processes are equally important . ] therefore , in order to be able to invert the collision matrix , it is necessary to calculate it in the vector space orthogonal to this zero - mode . this does not affect the result for the bulk viscosity because the source term is orthogonal to this zero - mode , @xmath146 . besides the simplifications already commented in the previous paragraphs , in order to obtain an explicit expression for the matrix element of the collision operator in the large-@xmath0 limit , it is obvious ( cf . appendix [ app : amplitudes ] ) that the dominant scattering processes are those for which three different flavors participate . moreover , the symmetry factors have to be specified , see table [ tab : symmetryf ] . .symmetry factors corresponding to relabeling the momenta of identical particles under the collision integral after summing over all particle types . here @xmath147 denote only flavor ( processes with three different flavors dominate ) . the momenta for each particular process are initially labeled according to the order indicated in the title of columns 2 and 3 . the particle labeled with ` ( 1 ) ' is `` distinguishable '' . some cells are empty because the corresponding process has already been taken into account by another symmetry factor . the processes obtained by permutation of the three flavors , although omitted in the table , have the same symmetry factors and also have to be taken into account . [ cols="^,^",options="header " , ] finally , @xmath148\,\big\{(2\pi)^2\delta^{(2 ) } ( \underline{p}{}_1+\underline{p}{}_2-\underline{p}{}_3-\underline{p}{}_4-\underline{p}{}_5-\underline{p}{}_6 ) n_\mathrm{f,1}n_\mathrm{f,2}\notag\\ & \times(1-n_\mathrm{f,3})(1-n_\mathrm{f,4})(1-n_\mathrm{f,5})(1-n_\mathrm{f,6})\left[|\mathcal{m}_{12\rightarrow \bar{3}456}|^2+\frac{3}{2}\,|\mathcal{m}_{\bar{1}2\rightarrow \bar{3}\bar{4}56}|^2\right]\notag\\ & \times [ \phi_i(p_1)+\phi_i(p_2)-\phi_i(p_3)-\phi_i(p_4)-\phi_i(p_5)-\phi_i(p_6)]\notag\\ & \times [ \phi_j(p_1)+\phi_j(p_2)-\phi_j(p_3)-\phi_j(p_4)-\phi_j(p_5)-\phi_j(p_6)]\notag\\ & + ( 2\pi)^2\delta^{(2 ) } ( \underline{p}{}_1+\underline{p}{}_2+\underline{p}{}_3-\underline{p}{}_4-\underline{p}{}_5-\underline{p}{}_6 ) n_\mathrm{f,1}n_\mathrm{f,2}n_\mathrm{f,3}(1-n_\mathrm{f,4})(1-n_\mathrm{f,5})(1-n_\mathrm{f,6})\notag\\ & \times \left[\frac{1}{6}\,|\mathcal{m}_{123\rightarrow 456}|^2+\frac{3}{2}\,|\mathcal{m}_{\bar{1}23\rightarrow \bar{4}56}|^2\right ] [ \phi_i(p_1)+\phi_i(p_2)+\phi_i(p_3)-\phi_i(p_4)-\phi_i(p_5)-\phi_i(p_6)]\notag\\ & \times [ \phi_j(p_1)+\phi_j(p_2)+\phi_j(p_3)-\phi_j(p_4)-\phi_j(p_5)-\phi_j(p_6)]\big\}\ , \end{aligned}\ ] ] where @xmath149 and @xmath150 are the inelastic amplitudes of fermion - fermion and anti - fermion - fermion scattering respectively . the amplitude squared @xmath151 of anti - fermion - anti - fermion scattering , as well as its corresponding symmetry factor , are equal to the fermion - fermion ones by charge conjugation , and have already been included in ( [ eq : cmatrix ] ) . it is evident from ( [ eq : cmatrix ] ) that the collision matrix is symmetric and positive semidefinite ( positive definite in the space orthogonal to its only zero - mode ) . note also that @xmath152 , and since @xmath153 ( cf . appendix [ app : amplitudes ] ) , therefore @xmath154 . a particularly convenient set of functions , which becomes a basis when @xmath155 , is @xcite @xmath156 with the thermal average @xmath157 for @xmath158 , @xmath159 for @xmath160 , and interpolating between these two behaviors for intermediate temperatures . this set of functions automatically incorporates the required asymptotic behavior for the solution of the boltzmann equation in the bulk channel : @xmath161 for @xmath162 , and @xmath163 for @xmath164 . in fig . [ fig : bulkv ] , i plot the numerical result of a variational computation of the bulk viscosity in the massive gross - neveu model with @xmath165 using @xmath166 basis functions . it is not difficult to realize that @xmath1 increases exponentially at low temperatures , like @xmath167 ) , due to the fermi - dirac factors present in @xmath168 as well as in @xmath169 ( cf . footnote [ ft : lowtinel ] ) . this behavior is analogous to the case of @xmath1 for @xmath170 in @xmath21 dimensions @xcite . the numerical error corresponding to considering only @xmath166 basis functions ( including the error from the numerical evaluation of integrals ) is estimated to be of order @xmath171 for temperatures around @xmath2 , increasing as we go down in temperatures , being @xmath172 for @xmath173 , which indicates that the basis ( [ eq : basis ] ) is not the best choice at those temperatures . by considering @xmath174 , the precision can be improved to @xmath175 at @xmath173 . however , due to the exponential growth of @xmath1 at low @xmath176 and since the result shown corresponds to a lower bound , the qualitative behavior with temperature is not expected to change significantly . , calculated with @xmath166 basis functions . the continuous line simply joins the data points.,width=377 ] from the numerical result we clearly see that there is no maximum in the bulk viscosity near @xmath2 , it is a monotonously decreasing function of the temperature . by reducing further the value of @xmath23 , we would eventually reconstruct ( continuously ) a discontinuity for @xmath1 at @xmath2 . for infinitesimally small @xmath23 , above @xmath177 the bulk viscosity would be arbitrarily small . going down in temperatures , it would increase very sharply right at @xmath2 ( with an arbitrarily large value for qcd . ] ) , and it would continue increasing exponentially at very low temperatures . this is shown in the plot of fig . [ fig : bulkv0 ] . , calculated with @xmath166 basis functions . the continuous line simply joins the data points.,width=415 ] in the paper @xcite ( section iv ) , the authors obtained a sum rule for the spectral density of the two - point function involving the trace of the energy - momentum tensor in pure yang - mills theory . the derivation is essentially based on the asymptotically - free character of the theory explicitly breaks scale symmetry , it is eventually restored at high energies because @xmath23 appears in the lagrangian multiplied by @xmath178 . ] . in this subsection i am going to comment on an interesting aspect of the massive gross - neveu model concerning its sum rule in the bulk channel . naively , since the massive gross - neveu model is asymptotically free , we can follow the analysis in @xcite and convince ourselves that the version of the sum rule for this particular system is simply @xmath179 interestingly , we notice that this sum rule does not depend explicitly on the mass parameter @xmath23 . since @xmath180 and @xmath181 , the lhs of ( [ eq : sumrule ] ) is @xmath182 . on the other hand , the order of @xmath183 depends on the frequency . for frequencies of order @xmath184 , a diagram contributing to @xmath183 at leading order is the one of fig . [ fig : spectraldensn ] , which is @xmath182 . hence , the contribution to the integral in ( [ eq : sumrule ] ) from this region of frequencies is @xmath182 , consistent with the lhs of the sum rule . at leading - order for @xmath184 . the crosses denote insertions of the trace of the energy - momentum tensor . the fermion propagator is not dressed.,width=113 ] if we now consider lower frequencies , @xmath185 , then a resummation of diagrams is necessary to obtain the leading - order spectral density . this is essentially because of the presence of pinching singularities when the external frequency becomes of the order of the fermion width @xmath186 ( cf . [ fig : fermionwidth ] ) @xcite , or smaller . these singularities correspond to the product of retarded and advanced fermion propagators sharing approximately the same momentum : @xmath187 consequently , the set of ladder diagrams depicted in fig . [ fig : ladder ] all contribute at leading order , @xmath188 , to the spectral density in this range of frequencies . thus , their contribution to the integral is again consistent with the lhs of the sum rule . at leading order for frequencies @xmath185 . double lines denote dressing of the fermion propagator according to the diagram of fig . [ fig : fermionwidth ] . the presence of pinching singularities is crucial in this regime of low frequencies.,width=226 ] for even lower frequencies , we know from the analysis of the boltzmann equation in the previous section that @xmath144 and @xmath145 processes will eventually dominate the spectral density . since the bulk viscosity is of order @xmath189 , for frequencies close to zero @xmath190 , and therefore , in order for it to be consistent with the lhs of the sum rule , this region of frequencies must be @xmath191 . now , in the previous section we saw that for temperatures @xmath192 , if we reduce the value of @xmath23 , the bulk viscosity can become arbitrarily large . however , the lhs of ( [ eq : sumrule ] ) remains finite as @xmath55 , this implies that the region where @xmath183 is of order @xmath189 has a width @xmath193 with @xmath194 and therefore the bulk viscosity does not contribute to the sum rule in this regime of temperatures for any value of @xmath23 . from the analysis of the previous sections for the massive gross - neveu model we can already extract some conclusions for other similar systems . consider for instance the non - linear @xmath16-model in @xmath3 dimensions @xcite . this model also shares with massless qcd the features of asymptotic freedom , dynamical generation of a mass gap , and classical scale invariance broken by the trace anomaly . the lagrangian of the model is @xmath195 since the @xmath116 symmetry can not be broken in @xmath3 dimensions , there is no phase transition in this system at finite temperature . it again is convenient to introduce an auxiliary field @xmath196 in order to analyze diagrammatically the large-@xmath0 limit : @xmath197 this model is also integrable @xcite . to leading order in the @xmath12 expansion , the inelastic diagrams have the same topology as the ones in fig . [ fig : leadinginel ] . integrability is proven in an analogous way to the gross - neveu model using the factorization of scalar loops in @xmath3 dimensions . and it can be broken for instance by introducing a term @xmath198 in the lagrangian , so the correlation between the bulk viscosity and the trace anomaly can be studied by making @xmath199 arbitrarily small . the thermodynamic properties of the system have been studied for instance in the works @xcite . in fig . [ fig : thermonlsm ] i plot the thermal mass gap , speed of sound , and trace anomaly . there are no discontinuities for these quantities in the limit @xmath200 . we then realize that the qualitative behavior of the bulk viscosity as we decrease @xmath199 ( restoring integrability ) is the one depicted in fig . [ fig : bulkvnlsm ] . for this model , integrability is always restored as @xmath200 and therefore bulk viscosity diverges . one could also consider the sum rule , which presumably is identical to the case of the gross - neveu model , and similarly the bulk viscosity would not contribute to it . -model in @xmath3 dimensions to leading order in the @xmath12 expansion . @xmath201 denotes the finite - temperature part.,width=604 ] -model in @xmath3 dimensions in the large-@xmath0 limit.,width=415 ] pure yang - mills theory in @xmath21 dimensions in the large-@xmath202 limit is also very similar to the massive gross - neveu model regarding the bulk viscosity . for low energies and temperatures , interaction between glueballs is suppressed by powers of @xmath202 , being the 3-point scattering amplitude @xmath203 @xcite . in this case the bulk viscosity is dominated by inelastic processes , due to the presence of a zero - mode in the collision integral corresponding to particle - number conservation @xcite . this implies an exponential growth of the bulk viscosity as the temperature decreases , @xmath204 , with @xmath205 the mass of the lightest glueball . regarding the sum rule derived in @xcite for this theory , at temperatures below @xmath2 , the region of frequencies where inelastic processes dominate becomes very narrow to be consistent with the sum rule . this implies for instance that extracting transport coefficients on the lattice at temperatures below @xmath2 becomes much more difficult as @xmath202 increases @xcite . in this regime of temperatures , a kinetic theory approach instead is more suitable in the large-@xmath202 limit to obtain transport coefficients . on the other hand , massless @xmath206 qcd is qualitatively different from the previous models . at very low temperatures , the dynamics is dominated by goldstone bosons and there is no exponential growth in the bulk viscosity as the temperature decreases . in addition , since this system undergoes a second - order phase transition in @xmath21 dimensions , the bulk viscosity would diverge at the critical temperature @xcite . we have seen that the massive gross - neveu model is non - integrable in the large-@xmath0 limit , which allows the study of momentum transport in this system . we found that there is no direct correlation between the trace anomaly and the bulk viscosity in general , i.e. , a peak in the former does not necessarily imply a peak in the latter . this was already obtained by s. jeon in @xcite where he analyzed the bulk viscosity of massive @xmath170 theory in @xmath21 dimensions , but since it is not an asymptotically - free theory and the scale symmetry is explicitly broken in that case , it remained to analyze whether a qcd - like theory could be qualitatively different due to the anomaly ( idea originally motivated by the paper @xcite ) . the use of this simple model in @xmath3 dimensions also avoids interference from critical phenomena present in higher dimensions ( for instance , the bulk viscosity would diverge near a second - order phase transition ) . in addition , it is a useful model to study sum rules and transport coefficients both at zero and finite fermion density in the large-@xmath0 limit . for instance , regarding sum rules , after a first superficial analysis , we saw that below @xmath2 the bulk viscosity would not contribute to the sum rule . this implies in general that it is not necessarily possible to extract bulk viscosity from sum rules . further work in these directions is in progress . i thank harmen warringa for proposing to analyze the bulk viscosity of the non - linear sigma model in 1 + 1 dimensions , and for helpful discussions . i also want to thank useful discussions with antonio dobado , kenji fukushima , angel gomez nicola , elena gubankova , yoshimasa hidaka , xu - guang huang , and toru kojo . this work has been sponsored by the helmholtz international center for fair . i acknowledge partial financial support from the spanish research projects fis2008 - 01323 , fpa2008 - 00592 , and ucm - bsch gr58/08 910309 . in @xmath3 dimensions , the lorentz group consists only of boots and therefore fermions have no spin . a two - dimensional representation of the dirac algebra is for instance @xmath208 with @xmath209 the pauli matrices , and @xmath210 . the general solution of the dirac equation @xmath211 can be written as a linear combination of plane waves @xmath212 , corresponding to fermions and anti - fermions respectively , where @xmath213 and @xmath214 . the spinors are normalized as @xmath215 and they also verify @xmath216 in the rest of this appendix , i will derive the finite - temperature version of the result previously obtained in @xcite concerning the factorization of fermion loops in @xmath3 dimensions . consider the momentum integral and matsubara sum corresponding to the fermion loop in fig . [ fig : fermion - loop ] at finite temperature with @xmath217 ( which is finite in @xmath3 dimensions ) : @xmath218 with @xmath219 , @xmath220 ( @xmath221 ) , @xmath222 , @xmath223 ( @xmath224 ) . external legs corresponding to the @xmath16 field . the different momenta satisfy @xmath225 , with @xmath226 , and @xmath227.,width=188 ] we first perform the matsubara sum using the result @xcite @xmath228 the integrand in ( [ eq : fermionl ] ) has poles at @xmath229 , and no cuts . thus , @xmath230 where @xmath231 , and @xmath232 is on - shell , i.e. , @xmath233 . if we now make the change of variables @xmath234 then the integrand in ( [ eq : ln1 ] ) becomes rational : @xmath235 where @xmath236 and @xmath237 mean making the substitution @xmath238 in @xmath239 and @xmath240 according to ( [ eq : changev ] ) , explicitly @xmath241 and @xmath242 are the solutions of the equation @xmath243 after partial fraction decomposition , we have give a zero contribution to the integral ( it can be easily seen from the fact that the integral must finite ) , so we ignore them . ] @xmath244 where obviously @xmath245 and explicitly , @xmath246\notag\\ & & \times\frac{m-(\tilde{\ksl}^s_{(ij\pm)}+\qsl_{ji})}{(\mathrm{i}2smq_{ji,2}+q_{ji}^2)(l_{ji}^{\pm , s}-l_{ji}^{\mp , s})}\ , \end{aligned}\ ] ] where @xmath247 means making the substitution @xmath248 in @xmath237 . from eq . ( [ eq : zeros ] ) it is evident that the momentum @xmath249 is also on - shell . and using the relation ( [ eq : spinorprod ] ) , rotated to euclidean space , we can rewrite this expression as @xmath250\notag\\ & \equiv&\frac{\hat{t}_{ji}^{\pm , s}}{\mp 4s\pi q_{ji}^2\sqrt{1 + 4m^2/q_{ji}^2}}\ .\label{eq : tree}\end{aligned}\ ] ] also note that the momenta @xmath251 with @xmath252 instead , are not on - shell . now , since @xmath253 and @xmath254 are both on - shell , this implies , and @xmath255 . thus , @xmath256 ( in two dimensions there are only two linearly - independent vectors ) , so @xmath257 ( after looking at ( [ eq : tildek ] ) and ( [ eq : zeros ] ) ) , and therefore @xmath258 . on the other hand , @xmath259 , so @xmath257 , and @xmath260 . ] @xmath261 then , from ( [ eq : tree ] ) and ( [ eq : condk ] ) we get @xmath262 furthermore , from the property @xmath263 , in a way analogous to the previous case it is not difficult to obtain @xmath264 then , working out the integral in ( [ eq : partialf ] ) and using the previous symmetry relations , after a tedious ( although straightforward ) simplification , one arrives to the final result @xmath265 -2\int\limits_{-\infty}^\infty\mathrm{d}k\,\frac{n_\mathrm{f}(e_k)}{e_k}\,\frac{q_{ji}^2 + 2kq_{ji,1}}{(q_{ji}^2 + 2kq_{ji,1})^2 + 4e_k^2q_{ji,2}^2}\right\}\ , \end{aligned}\ ] ] with @xmath266 . the result of @xcite is then obtained by particularizing for @xmath267 and performing the rotation to minkowski space . we can represent the result of ( [ eq : fermioncut ] ) diagrammatically as in fig . [ fig : fermioncut ] , with ) is apparently gone , but it is actually there because the external momenta of the trees are equal at both sides , and therefore @xmath268[\bar{u}(p)\tilde{b}u(q)]=[\bar{u}_a(q)a_{ab}u_b(p)][\bar{u}_c(p)b_{cd}u_d(q)]\notag\\ & = u_d(q)\bar{u}_a(q)a_{ab}u_b(p)\bar{u}_c(p)b_{cd}=\tr\{u(q)\bar{u}(q)\tilde{a}u(p)\bar{u}(p)\tilde{b}\}\ , \label{eq : tracespinors}\end{aligned}\ ] ] for any momenta @xmath269 and any matrices @xmath270 . ] @xmath271 -\int\limits_{-\infty}^\infty\mathrm{d}k\,\frac{n_\mathrm{f}(e_k)}{2\pi e_k}\,\frac{q_{ji}^2 + 2kq_{ji,1}}{(q_{ji}^2 + 2kq_{ji,1})^2 + 4e_k^2q_{ji,2}^2}\ .\end{aligned}\ ] ] dimensions into two tree graphs , as implied by the result ( [ eq : fermioncut ] ) . the momenta @xmath272 and @xmath273 are on - shell.,width=415 ] as we saw in section [ sec : kinetictheory ] , in @xmath3 dimensions only @xmath144 and @xmath145 scattering amplitudes contribute to the bulk viscosity at leading order in the @xmath12 expansion . among them , in the large-@xmath0 limit , processes involving three different flavors dominate because of an overall factor @xmath274 obtained after summing over all flavor types . by following the discussion in section [ sec : integrability ] , we realize that the diagram of fig . [ fig : leadinginel](g ) , after being cut in all possible ways , cancels the contribution from the diagrams [ fig : leadinginel](a f ) , except for a term proportional to the integrability - breaking parameter , i.e. , @xmath71 . and this also happens in the case of diagrams contributing to the @xmath144 amplitude , which are of the same order as the @xmath145 ones and have the same topology . consequently , the fermion - fermion inelastic and three - fermion elastic scattering amplitudes are given by the sum of the diagrams of fig . [ fig : fermion - fermion ] and [ fig : fermion - fermion - fermion ] respectively times the integrability - breaking factor with a minus sign : @xmath275\ , \\ & \mathrm{i}\mathcal{m}_\mathrm{f\,f\,f\rightarrow f\,f\,f}=-\mathrm{i}\frac{m}{m(t)}\left[\mathcal{m}_\mathrm{f\ , f\,f\rightarrow f\,f\,f}^{(a)}+\mathcal{m}_\mathrm{f\ , f\ , f\rightarrow f\,f\,f}^{(b)}+\mathcal{m}_\mathrm{f\ , f\,f\rightarrow f\,f\,f}^{(c)}+\mathcal{m}_\mathrm{f\ , f\,f\rightarrow f\,f\,f}^{(d)}+\mathcal{m}_\mathrm{f\ , f\,f\rightarrow f\,f\,f}^{(f)}\right]\ , \end{aligned}\ ] ] and similarly for the processes involving more anti - fermions , @xmath276 , and @xmath277 . ] . in the massive gross - neveu model . here @xmath278 , @xmath279 , and @xmath280 denote arbitrary flavors . the diagrams corresponding to @xmath281 are obtained inverting the fermionic flow ( but not the momentum ) in the line which joins @xmath282 and @xmath283.,width=453 ] in the massive gross - neveu model . here @xmath278 , @xmath279 , and @xmath280 denote arbitrary flavors . the diagrams corresponding to @xmath284 are obtained inverting the fermionic flow ( but not the momentum ) in the line which joins @xmath282 and @xmath285.,width=453 ] for instance , the explicit contribution from the diagram [ fig : fermion - fermion](a ) is @xmath286 and analogously for the rest of diagrams . we already see from ( [ eq : ma ] ) , since @xmath287 , that the amplitude squared is @xmath288 ( and the same for the other amplitudes ) . it is not difficult to realize that the momenta in the argument of the retarded fermion propagators can not be on - shell , being each @xmath289 on - shell , therefore we can make the substitution @xmath290 in our calculations without worrying about singularities . after summing all the amplitudes and squaring them , we can simplify the expression using ( [ eq : spinorprod ] ) , the property @xmath291 , and the relations for traces in dirac space @xmath292\ , \\ \tr\left\{(\psl+m)(\ksl+m)(\qsl+m)(\lsl+m)\right\}&=&2[m^2(m^2+l\cdot(p+k+q)+p\cdot(k+q)+k\cdot q)\notag\\ & & + ( l\cdot p)(k\cdot q)-(l\cdot k)(p\cdot q)+(l\cdot q)(p\cdot k)]\ .\end{aligned}\ ] ] since the final expressions for the squared amplitudes after simplification are still very long and do not explicitly provide further significant information , i avoid to write them down here .

i describe some recent developments in the understanding of gluon scattering amplitudes in super - yang - mills theory in the large- limit . these amplitudes can be computed to high orders in the weak coupling expansion , and also now at strong coupling using the ads / cft correspondence . they hold the promise of being solvable to all orders in the gauge coupling , with the help of techniques based on integrability . they are intimately related to expectation values for polygonal wilson loops composed of light - like segments . slac pub13167 address = stanford linear accelerator center , stanford university , stanford , ca 94309 , usa

0803.2475

in this talk i would like to describe some remarkable progress that has been made in the past few years in understanding the structure of gauge boson scattering amplitudes in a particular gauge theory , @xmath0 super - yang - mills theory . while this theory differs in many details from the electroweak and qcd theories whose radiative corrections were the subject of this symposium , there are many common issues , particularly associated with infrared structure . indeed , the understanding of infrared divergences in qcd acquired over the last few decades has proved extremely useful in unraveling some of the structure of @xmath0 super - yang - mills theory . @xmath0 super - yang - mills theory is the most supersymmetric theory possible without gravity . in the free theory , starting from the helicity @xmath2 massless gauge boson ( `` gluon '' ) state , the four supercharges can be used to lower the helicity by @xmath3 units , until the helicity @xmath4 gluon state is reached . if one had more supercharges , one would need spin @xmath5 states , and it is not known how to quantize such theories in a unitary way without including at least spin 2 gravitons . along the way from the helicity @xmath2 to the helicity @xmath4 gluon state , one passes through the 4 massless ( majorana ) spin @xmath6 gluinos , and 6 real ( or 3 complex ) massless spin @xmath7 scalars . in this maximally supersymmetric yang - mills theory ( msym ) , all the massless states are in the adjoint representation of the gauge group , which we will take to be @xmath8 . the interactions are all uniquely specified by the choice of gauge group , and one dimensionless gauge coupling @xmath9 . the theory is an exactly scale - invariant , conformal field theory ; that is , the beta function vanishes identically for all values of the coupling @xcite . here we will consider the t hooft limit of msym , in which the number of colors @xmath10 , with the t hooft parameter @xmath11 held fixed @xcite . in this limit , only planar feynman diagrams contribute . also , the anti - de sitter space / conformal field theory ( ads / cft ) duality @xcite suggests that for @xmath10 the weak - coupling perturbation series in @xmath12 might have some very special properties . the reason is that , according to ads / cft , the strongly - coupled ( large @xmath12 ) limit of the four - dimensional conformal gauge theory has an equivalent description in terms of a weakly - coupled string theory . the intuition is that the perturbative series should know about this simple strong - coupling limit , and organize itself accordingly @xcite . sketches how events such as gluon scattering look in the ads / cft duality @xcite . five - dimensional anti - de sitter space , ads@xmath13 , contains , besides the usual four - dimensional space - time @xmath14 , an additional radial variable @xmath15 , which corresponds to a resolution scale in the four - dimensional theory . large values of @xmath15 correspond to the ultraviolet ( uv ) region ; small values to the infrared ( ir ) . the figure shows a `` big '' glueball state in the ir , and a `` small '' glueball state in the uv . the arrows represent the motion of plane - wave single gluon states in @xmath14 for @xmath16 scattering at @xmath17 . we ll discuss the motion in @xmath15 later . the radius of curvature of ads@xmath13 is proportional to @xmath18 . large @xmath12 means that the space - time is only weakly curved , which makes it much simpler to study the string theory ; higher excitations of the string can usually be neglected . the ads / cft duality is a weak / strong duality . quantities that can be computed at weak coupling in one picture have a strong - coupling description in the other picture . this property makes ads / cft both powerful and difficult to check explicitly although there is certainly convincing evidence in its favor . there are a few quantities that are known ( modulo a few assumptions ) to all orders in @xmath12 ; that is , for which one can interpolate all the way from weak to strong coupling . notable among these is the cusp ( or soft ) anomalous dimension @xmath19 . the qcd version of this quantity crops up a lot in soft - gluon resummation . beisert , eden and staudacher @xcite have given an all - orders proposal for @xmath19 , based on integrability , plus a number of other properties . their proposal is consistent with the first four loops in the weak - coupling expansion @xcite , and also agrees @xcite with the first three terms in the strong - coupling expansion @xcite . in this talk i would like to discuss the evidence for another proposal @xcite , namely that gluon - gluon scattering @xmath16 in msym , for any scattering angle @xmath20 can be fully specified by just three functions of @xmath12 , independent of @xmath20 . one of these three functions is already `` known '' , because it is just @xmath19 . this proposal has received some confirmation at strong coupling , through the work of alday and maldacena @xcite . it was motivated by the structure of ir divergences in gauge theory . in a conformal field theory , scale invariance implies that the interactions never shut off , so that a scattering process can not really be defined . while strictly speaking this is true , we are able to get around it in practice by regulating the theory in the ir . we ll use dimensional regularization with @xmath21 and @xmath22 ( actually a version of it that preserves all the supersymmetry @xcite ) . the regulator breaks the conformal invariance , but we can recover it by performing a laurent expansion around @xmath23 , up to and including the @xmath24 terms . at one loop , there are two types of ir divergences : _ soft_-gluon exchange , in which the virtual gluon energy @xmath25 ; and _ collinear _ regions , in which the gluon s transverse momentum ( with respect to a massless external line ) @xmath26 . the soft and collinear regions each produce a @xmath27 pole , resulting in a @xmath28 leading behavior for on - shell amplitudes at one loop . at @xmath29 loops , the leading behavior is @xmath30 , coming from multiple soft - gluon exchange that is arranged hierarchically , so that the outermost gluons are softer and more collinear than the innermost ones . in fact , all the pole terms for @xmath29-loop amplitudes are predictable in planar gauge theory , thanks to decades of work on the soft / collinear factorization and exponentiation of amplitudes , and of quark and gluon form factors , in qcd @xcite . for both qcd and msym , in the planar limit the pole terms are given in terms of three quantities ( in the notation of refs . @xcite ) : * the beta function @xmath31 ( but of course this vanishes in msym ) , * the cusp anomalous dimension @xmath19 , * a `` collinear '' anomalous dimension @xmath32 . the cusp anomalous dimension gets its name because it appears @xcite in the renormalization group equation for the expectation value of a wilson line @xmath33 for two semi - infinite straight lines , joined at a kink or cusp : ( + ( g ) g ) w(,g ) = - 2 _ k ( ) ^2 + ( ^0 ) , [ cuspeq ] where @xmath34 as the two straight lines become light - like , @xmath35 . the cusp anomalous dimension also controls @xcite the universal ( flavor independent ) large - spin limit of anomalous dimensions @xmath36 of leading - twist operators with spin @xmath37 , such as the quark operators @xmath38 : _ j = 12 _ k ( ) j + ( j^0 ) , j . [ gammajcusp ] finally , through a mellin transform of , @xmath19 appears in the large @xmath39 limit of the dglap kernel for evolving the parton distributions , p_aa(x ) = 12 _ k ( ) ( 1-x)_+ + , x 1 . [ splitcusp ] thus , in the study of qcd at colliders it is an important quantity for resumming the effects of soft gluon emission . softcollfact.eps the general infrared structure of massless gauge amplitudes can be exposed @xcite by factoring off soft singularities , which arise from long - distance gluon exchange , and collinear singularities , which are also at long distances , but only out along the axis of a hard parton . this space - time picture is shown in . defining @xmath40 to be the full amplitude @xmath41 divided by the tree amplitude @xmath42 , the factorization formula reads , _ n = s(\{k_i } , , ) _ i=1^n j_i(k_i , , ) h_n(\{k_i } , ) , [ sjh ] where @xmath43 is the factorization scale , and @xmath44 is the hard remainder function , and is finite as @xmath45 . the soft function @xmath46 only sees the classical color charge of the @xmath47 particle . in general it is a complicated matrix acting on the possible color configurations for @xmath44 , because soft gluons can attach to any pair of external partons . the jet function @xmath48 is color - diagonal , but depends on the @xmath47 spin . terms that are color - diagonal and spin - independent can be moved arbitrarily between @xmath46 and @xmath48 . softcollfactplanar.eps in the large-@xmath1 planar limit , the picture simplifies , to that shown in . here @xmath49 represents the coefficient of a particular color structure , @xmath50 $ ] . now soft gluons can only connect adjacent external partons ; and indeed there is no mixing of different color structures at large @xmath1 . because of the color - triviality of the planar limit , one can absorb the entire soft function @xmath46 into jet functions , or break up the right - hand side of into @xmath51 wedges . each wedge represents the square root of the sudakov form factor , the amplitude @xmath52}$ ] for a color - singlet state `` 1 '' to decay to a pair of partons , say gluons . hence the planar version of is _ _ i=1^n ^1/2 h_n(\{k_i},,_s ) . [ sjhplanar ] the only dependence of the singular terms on the kinematics is through the momentum scale , @xmath53 , entering the @xmath47 sudakov form factor . factorization also implies that the sudakov form factor obeys a differential equation in the momentum scale @xcite , ^[1gg](q^2/^2,_s , ) = 12 . [ suddiff ] here @xmath54 is a pure counterterm , or series of @xmath27 poles . by analogy with the @xmath55-dimensional @xmath56-function , @xmath57 , the single poles ( related to @xmath58 ) determine @xmath59 completely . the function @xmath60 is finite as @xmath45 , but contains all the @xmath61 dependence ; it will generate a single pole in @xmath62}$ ] upon integrating with respect to @xmath61 . the functions @xmath59 and @xmath60 obey renormalization group equations , ( + ) k = - ( + ) g = - _ k ( ) . [ kgdeq ] the collinear anomalous dimension @xmath32 arises as a constant of integration for the differential equation for @xmath60 . solving the differential equations for @xmath59 , @xmath60 and the sudakov form factor is particularly easy in a conformal theory because the four - dimensional coupling does not run . doing this , and inserting the form - factor solution into for the @xmath51-point amplitude , we obtain @xcite , _ n ( ) & = & 1 + _ l=1^a^l m_n^(l ) ( ) + & = & h_n(\{k_i } ) , [ mplanarsing ] where a ( 4e^-)^ = 8 ^ 2 ( 4e^-)^[adef ] is the loop expansion parameter in the t hooft limit , and @xmath63 and @xmath64 are the @xmath65-loop coefficients of @xmath66 and @xmath67 . the argument of the exponential in looks very much like the one - loop amplitude , but with @xmath68 replaced by @xmath69 , denoted by @xmath70 . thus we are motivated to rewrite as _ n ( ) = , [ mplanarnew ] where @xmath71 collects three series of constants . two of these are identified with the previous quantities as , f_0^(l ) = 14 _ k^(l ) , f_1^(l ) = l2 _ 0^(l ) , [ f01def ] while the third quantity , @xmath72 , is related to the consistency of under collinear limits @xcite . the surprise in planar msym is that in some cases the hard remainder function @xmath73 defined through is actually a constant , independent of the kinematics . this result , which has been tested perturbatively for @xmath74 through three loops @xcite , and for @xmath75 at two loops @xcite , is a conjecture beyond that : _ n = . [ bdsansatz ] the dependence of the finite part of the logarithm of the amplitude is predicted to all orders by , in terms of the cusp anomalous dimension . the prediction for four - gluon scattering is _ 4^finite = , [ bdsfiniteterms ] where @xmath76 , @xmath77 . as we shall discuss in section [ strongsection ] , this formula was confirmed at strong coupling by alday and maldacena @xcite using the ads / cft correspondence @xcite . in contrast , even at two loops there does not appear to be any comparably simple formula for the finite parts of four - gluon scattering amplitudes in qcd , or for the subleading - in-@xmath1 terms in msym @xcite . instead of a constant , as in , one finds that @xmath78 in is given by a complicated combination of polylogarithms involving the dimensionless ratio @xmath79 . on the other hand , is reminiscent of the observation @xcite that finite terms can also exponentiate in qcd , in _ e.g. _ the drell - yan cross section near partonic threshold . the evidence in favor of was collected from explicit computations of the multi - loop scattering amplitudes . the amplitudes were constructed by evaluating ( generalized ) unitarity cuts and matching them to compact representations in terms of a relatively small number of multi - loop integrals , which turn out to have rather interesting properties . ordinary unitarity relates discontinuities ( cuts ) in a given channel to products of lower - loop amplitudes , summed over the possible intermediate states in that channel . generalized unitarity allows the lower - loop amplitudes to be further sliced , all the way down to tree amplitudes . shows an ordinary three - particle cut for the four - gluon amplitude . the information in this cut can be extracted more easily by further cutting the one - loop five - point amplitude on the right - hand side of the cut , decomposing it into the product of a four - point tree and a five - point tree ; as illustrated , there are three inequivalent ways to do this . if one finds a representation of the amplitude that reproduces all the generalized cuts ( in @xmath55 dimensions ) , then that representation is correct . onetofourloops.eps shows the integrals that enter the four - gluon scattering amplitude in planar msym , from one to four loops @xcite , along with their numerator factors . an overall factor of @xmath80 is omitted from the rescaled amplitude @xmath81 , and only one permutation of each integral is shown . at one and two loops , only scalar integrals appear ; that is , the numerator factors in the integrand depend only on the external momentum invariants . at three loops , there are two integrals , the scalar triple ladder integral and the `` tennis - court '' integral shown at the top right of . the latter integral marks the first appearance of a loop - momentum factor in the numerator , of the form @xmath82 , as dictated by the `` rung rule '' @xcite . the rung - rule correctly describes all integral topologies that can be reduced to trees by a sequence of two - particle cuts . at four loops , the last two integrals in have no two - particle cuts , and are somewhat more work to determine . at five loops ( not shown ) there are a total of 34 distinct integrals @xcite . still , it is remarkable that so few integrals are required to describe the amplitude . twoloopdual.eps in fact , the integrals that appear in the four - point amplitude through five loops are all _ pseudo - conformal_. to describe what this means @xcite , first consider taking all the external legs off shell , @xmath83 , in order to be able to perform the integral without dimensional regularization , in @xmath84 . next define dual momentum or sector variables @xmath85 , such that the original momentum variables @xmath86 are differences of the @xmath85 , with @xmath87 . similarly define an @xmath85 associated with each loop , such that @xmath88 is equal to the momentum flowing through the propagator that separates @xmath85 from @xmath89 . illustrates the dual diagram ( in blue ) associated with the planar double box integral ( in orange ) which appears in the two - loop msym amplitude . the dual propagators ( denominator factors ) are shown as solid blue lines , while dashed blue lines correspond to numerator factors in the integrand . the integral is given by i^(2)(\{k_i } ) & = & s^2 t [ doublebox ] + & = & ( x_13 ^ 2)^2 x_24 ^ 2 , [ doubleboxdual ] using @xmath90 , @xmath91 , and so forth . under an inversion , @xmath92 , we have x_ij^2 , 1 cm d^4 x_5 , 1 cm d^4 x_6 , [ inversionsym ] and it is easy to see that is left invariant . in general , an integral is invariant under inversion if there is a net of zero ( four ) lines emerging from each external ( internal ) @xmath85 vertex , where `` net '' means solid lines minus dashed lines . every integral is automatically invariant under translations of the dual variables , @xmath93 , and under lorentz transformations . because these transformations , together with inversions , generate the conformal group , invariance under inversion suffices to guarantee dual conformal invariance for the integral . now we can define a pseudo - conformal integral to be one which is finite in @xmath84 , after all the @xmath94 are taken off - shell , is dual conformal invariant , and possesses a smooth @xmath95 limit . the last condition ensures that the integral does not become infinite or vanish as we return to the on - shell limit . dual conformal symmetry arose in the context of multi - loop ladder integrals @xcite , and in two dimensions in the theory of ( planar ) reggeon interactions @xcite . its relevance for the structure of msym amplitudes was first pointed out by drummond , henn , smirnov and sokatchev @xcite , based on the structure of the amplitudes through three loops , and the rung - rule contributions at four loops . the four- and five - loop four - gluon amplitudes can be organized as well , according to the two principles : * only pseudo - conformal integrals appear . * the pseudo - conformal integrals appear only with weight @xmath96 . originally it appeared that two integrals at four loops @xcite and 25 integrals at five loops @xcite were pseudo - conformal but did _ not _ appear in the amplitude . however , it was later pointed out that those integrals were not actually finite in @xmath84 @xcite . recently , some intuition into the signs @xmath96 has been given by considering the singularity structure of the various integrals more carefully @xcite . once the structure of the amplitude is known in terms of basic integrals , the next task is to evaluate those integrals , analytically if possible , otherwise numerically . for example , to test at three loops , we first expand it out to third order , obtaining the iterative relation , m_n^(3 ) ( ) = - 13 ^3 + m_n^(1 ) ( ) m_n^(2 ) ( ) + f^(3 ) ( ) m_n^(1)(3 ) + c^(3 ) + ( ) . [ bds3 ] to test this relation at order @xmath97 for @xmath74 @xcite , we need the following integrals : * the one - loop box integral through @xmath98 because it has @xmath28 poles , and appears cubed in . * the planar double box integral @xcite in through @xmath99 because @xmath100 appears in multiplied by @xmath101 . * the triple ladder @xcite and tennis - court @xcite integrals through @xmath97 . mellin - barnes techniques ( see _ e.g. _ ref . @xcite ) are very useful in this regard . inserting the results into , and using identities among weight 6 harmonic polylogarithms @xcite , the relation ( [ bds3 ] ) was verified , and three of the four constants at three loops could be extracted : f_0^(3 ) & = & 115 ( _ 2)^2 , f_1^(3 ) = 6 _ 5 + 5 _ 2 _ 3 , f_2^(3 ) = c_1 _ 6 + c_2 ( _ 3)^2 , [ f3loop ] + c^(3 ) & = & ( 341216 + 29 c_1 ) _ 6 + ( -179 + 29 c_2 ) ( _ 3)^2 . [ c3loop ] the first two of these constants control infrared divergences . the value of @xmath102 confirms a result for the three - loop cusp anomalous dimension in planar msym , which was first obtained @xcite by applying the principle of `` maximal transcendentality '' to the corresponding result in qcd @xcite . the value of @xmath103 gives the three - loop collinear anomalous dimension , which was found to agree ( applying the same principle ) with the qcd result @xcite . the constants @xmath104 and @xmath105 are inseparable using only the four - gluon amplitude ; either the five - gluon amplitude or a collinear analysis would be required to separate them . the numbers @xmath106 and @xmath107 are expected to be rational . a similar analysis can be performed at four loops @xcite , except that the integrals become less tractable analytically . fortunately , there are methods available for automating the construction of mellin - barnes representations @xcite , the extraction of @xmath27 poles , and the setting up of numerical integration over multiple contours for the mellin inversion @xcite . before describing the four - loop results , let us turn to some very interesting developments that have taken place , based on integrability . in large-@xmath1 gauge theory , a preferred role is played by local `` single - trace operators '' . in the case of msym , one subsector of such operators is provided by products of the 3 complex scalar fields , @xmath108 , @xmath109 . the operator @xmath110 $ ] is a so - called bps operator , and is unrenormalized to all orders in @xmath12 . a set of operators with more interesting renormalization properties are close to bps @xcite , and contain @xmath111 fields as well as @xmath112 , for example , . [ singletracesu2 ] as shown in ( a ) , this set of operators can be mapped to a one - dimensional , periodic spin chain , in which @xmath112 ( @xmath111 ) is mapped to spin up ( spin down ) , corresponding to a finite - dimensional ( spin 1/2 ) representation of @xmath113 spin symmetry . the anomalous dimensions of the set of operators ( [ singletracesu2 ] ) are found by diagonalizing the dilatation operator , which can be mapped to a hamiltonian for the spin chain . in the large-@xmath1 limit , this hamiltonian is local , because non - local interactions correspond to non - planar diagrams . for example , as shown in ( b ) , a one - loop contribution from a four - scalar interaction can only affect color - adjacent @xmath108 fields ( spins ) . ( the range of the interactions does increase with the number of loops . ) minahan and zarembo @xcite showed that the one - loop hamiltonian was _ integrable _ ; that is , the system possesses * infinitely many conserved charges , * a spectrum of quasi - particles ( spin waves , or magnons ) , * magnon scattering via a @xmath114 @xmath46 matrix obeying the yang - baxter equation , * solutions for the anomalous dimensions ( energies ) via a bethe ansatz . integrable structures in qcd had been identified previously @xcite . in planar msym , however , the integrability appears to persist to all orders in @xmath12 ; indeed , it is known to be present at strong coupling , from the form of the classical sigma model on target space ads@xmath115 @xcite . there is a rich literature of extensions of the one - loop results of ref . @xcite to higher loops , even all loop orders , and to more general sectors of planar msym , which i can only touch on here @xcite . the sector most relevant to gluon scattering amplitudes is not the spin 1/2 @xmath113 sector ( [ singletracesu2 ] ) , but that in which the @xmath111 fields are replaced by covariant derivatives @xmath116 acting in the @xmath117 ( light - cone ) direction , . [ singletracesl2 ] these derivatives act as an infinite - dimensional representation of the noncompact version of @xmath113 , namely @xmath118 . within this sector , the cusp anomalous dimension can be found by taking the limit of a small number of fields ( spin chain length ) @xmath29 , and a large number of derivatives @xmath37 , to get the operator o_j = , j. [ cuspscalar ] by the universality of the cusp anomalous dimension , it does not matter which leading - twist large @xmath37 operator is used ; they all have the behavior ( [ gammajcusp ] ) at large @xmath37 . in brief , and omitting many subtleties , the bethe - ansatz solution consists of taking the eigenstates of the hamiltonian to be multi - magnon states , with phase - shifts induced by repeated @xmath114 scatterings . the periodicity of the wave function on the closed chain leads to the bethe condition , which depends on the chain length @xmath29 . in the limit @xmath119 , the bethe condition becomes an integral equation , which depends on the form of the @xmath114 magnon @xmath46 matrix @xcite . this @xmath46 matrix is _ almost _ fixed by the symmetries , but an overall phase , the _ dressing factor _ , is not so easily deduced . finally , there is a potential _ wrapping problem _ in extrapolating to the cusp anomalous dimension : the bethe ansatz is only rigorously valid when the interaction range ( the number of loops ) is smaller than the chain periodicity @xmath29 . however , even though the cusp anomalous dimension has @xmath120 , it has been argued that its universality leads it to appear within large-@xmath29 sectors , and renders it immune to the wrapping problem @xcite . eden and staudacher @xcite derived an integral equation for the all - orders behavior of the cusp anomalous dimension from an all - loop bethe ansatz @xcite , by assuming that the dressing factor did not play a role perturbatively . this equation agreed with the known one- , two- , and three - loop coefficients of @xmath19 , and made the four - loop prediction , f_0^(4 ) |_es = 14 _ k^(4 ) |_es = - 732520 ^6 + ( _ 3)^2 = - 26.4048255 , [ esprediction ] motivating the computation of the four - loop four - gluon scattering amplitude , and the numerical extraction of @xmath121 from it . the result found @xcite , f_0^(4 ) = -29.335 0.052 , [ f04bcdks ] and later with much improved precision @xcite , f_0^(4 ) = -29.29473 0.00005 , [ f04csv ] was consistent , not with , but with a version in which the sign of the @xmath122 term was flipped , f_0^(4 ) |_bes = 14 _ k^(4 ) remarkably , the latter value was predicted , simultaneously with ref . @xcite , by beisert , eden and staudacher ( bes ) @xcite , based on a modified integral equation taking into account a new proposal for the dressing factor , with nontrivial effects beginning at four loops . the proposed dressing factor was deduced by using its properties at strong - coupling , where it had been known to be nontrivial @xcite . perhaps even more remarkably , the only effect of including the dressing - factor term on the weak - coupling expansion of the integral equation , is to make the substitution @xmath123 , which affects only the signs of the odd - zeta terms in the perturbative expansion . at five loops , this sign - flip is f_0^(5 ) |_es & = & ( 887/56700 ) ^8 - 2 _ 2 ( _ 3)^2 - 10 _ 3 _ 5 = 131.21 [ es5loops ] + f_0^(5 ) |_bes & = & ( 887/56700 ) ^8 + 2 _ 2 ( _ 3)^2 + 10 _ 3 _ 5 = 165.65 , [ bes5loops ] which also agrees with interpolation - based estimates @xcite . the bes integral equation was solved numerically @xcite , and later expanded analytically to all orders in the strong - coupling ( @xmath124 ) expansion @xcite . its strong - coupling behavior is consistent with the known first three terms in this expansion @xcite . this concordance , plus the agreement with the first four loops at weak coupling , strongly suggests that the bes equation is an exact solution for the cusp anomalous dimension , valid for arbitrary @xmath12 . the next quantity appearing in the planar msym gluon scattering amplitudes , @xmath32 , which controls single poles in the argument of the exponential in , is not quite as well known . the first four loop coefficients are known , the fourth numerically @xcite , g_0 ( ) = -_3 ( 8 ^ 2)^2 + 23 ( 6 _ 5 + 5 _ 2_3 ) ( 8 ^ 2)^3 - ( 77.56 0.02 ) ( 8 ^ 2)^4 + , [ g04loops ] and one coefficient is now known in the strong - coupling expansion @xcite . a pad approximant incorporating this data has been constructed @xcite . clearly , it would be of great interest if an integral equation could be found governing @xmath32 for all values of the coupling . finding a cleaner operator interpretation for this quantity may be quite useful in this respect . now let us return to the picture of gluon scattering at strong coupling developed by alday and maldacena @xcite . is another view of the ads space sketched in , showing also a pair of incoming open string states prior to a hard scattering . the ends of the open strings are anchored on a @xmath55 brane , which serves as an infrared regulator and is located at a small value of the ads radial variable , @xmath125 . the short - distance ( uv ) nature of the hard scattering forces part of the string to penetrate to large values of @xmath126 . gluons correspond to this part of the string , and the rest of the string can be thought of as the color string a gluon has to drag along with it , which is particularly important at strong coupling . because the string has to stretch a long way , the scattering is semi - classical @xcite . this regime is similar to very high - energy , fixed - angle scattering in string theory in flat space - time , which was studied long ago @xcite . evaluated on the classical solution , for the case of color - ordered scattering with gluons 1 and 3 incoming , 2 and 4 outgoing , the string world - sheet action is imaginary . the euclidean action , or area , is real , and is logarithmically divergent , leading to a large exponential suppression @xcite , _ 4 ~~ ~ , [ expsuppress ] where @xmath126 . the coupling - constant dependence in originates from the formula for the radius of curvature of ads , @xmath127 , which enters the world - sheet action . from the string point of view , the suppression can be attributed to a tunnelling suppression factor . from the point of view of a four - dimensional collider physicist , it is a typical sudakov suppression factor @xcite : the probability for a pair of gluons to make it all the way into and back out of the scattering without radiating at all is exponentially small especially at strong coupling , @xmath128 with a double log in the exponential . ybdycusp.eps to make contact with the perturbative results , alday and maldacena constructed a dimensionally - regularized version of ads@xmath115 , instead of using the @xmath55 brane location @xmath125 as a regulator . they also introduced @xmath129-dual variables @xmath130 in place of the usual four - dimensional coordinates . the @xmath129-duality transformation is a kind of fourier transform , so the @xmath130 are like momentum variables . indeed , the asymptotic boundary value for the world - sheet , which resides at @xmath131 in the dimensionally - regularized setup , is a polygon constructed from light - like segments in @xmath130 , with the corners @xmath132 satisfying y_i+1 ^ - y_i^= k_i^ , [ yboundary ] where @xmath133 is the momentum of the @xmath47 gluon . from , we see that the @xmath134 coincide precisely with the dual variables @xmath85 introduced in section [ pseudoconfsection ] to discuss dual conformal invariance ! ( a ) shows the light - like quadrilateral boundary satisfying for the case of @xmath114 gluon scattering at @xmath17 in the 1 - 2 plane , with @xmath135 and @xmath136 incoming . the vertical direction is the ( dualized ) time direction . planarexp.eps near each corner of the polygonal boundary , the solution must look like a cusp solution , previously constructed by kruczenski @xcite , in which @xmath15 behaves like @xmath137 } = \sqrt{(2+\e)y^+y^-}$ ] for some spatial coordinate @xmath138 , and light - cone coordinates @xmath139 . this hyperboloid is shown in ( b ) . the classical action ( area ) for this solution has a divergence regulated by @xmath68 , i s_cl = - s^e_cl - r^2_ads _ 0 dy^+ dy^- ( y^+y^-)^1+/2 ~ - 1 ^2 2 ~ - 1 ^ 2 _ k ( ) 2 . [ cuspaction ] the coefficient of the leading divergence is just the strong - coupling limit of the cusp anomalous dimension @xcite , _ k ( ) ~ , . [ cuspstrong ] illustrates the situation heuristically . the singular part of the planar amplitude can be broken up into sudakov wedges , as in and . the overlap of soft and collinear divergences corresponds to regions between two hard lines , _ e.g. _ @xmath86 and @xmath140 . thus each wedge is associated with a single divergent cusp @xcite , of the form shown in ( b ) . the full classical solution , for arbitrary scattering angle , was found by alday and maldacena @xcite . its action gives a strong - coupling amplitude of the form , _ 4 & = & , + -s^e_cl & = & - 1 ^2 + & & - 1 2 ( 1 - 2 ) + & & + 8 + ( ) , [ amse ] where @xmath141 . this expression can be compared with the strong - coupling extrapolation of the ansatz ( [ bdsansatz ] ) @xcite . the @xmath28 poles agree , using the strong - coupling value for @xmath19 from . the @xmath27 poles give the strong - coupling limit of the collinear anomalous dimension @xmath32 , g_0 ( ) ~ , . [ g0strong ] the finite part of @xmath142 has a dependence on @xmath143 and @xmath144 which is precisely as predicted by . the dual momentum variables @xmath145 play a prominent role in the strong - coupling computation of alday and maldacena , which is essentially the same as computing a wilson loop vacuum expectation value at strong coupling . inspired by this connection , there has been a sequence of recent wilson - loop computations for loops corresponding to the dual - momentum boundary conditions for an @xmath51-point amplitude , namely polygons composed of @xmath51-light - like segments , with corners obeying . the first of these computations was by drummond , korchemsky and sokatchev @xcite , for the one - loop expectation value of a quadrilateral ( @xmath74 ) wilson loop . up to constants of the kinematics , attributable to a different regulator ( in the uv ) than the one used for the amplitudes ( in the ir ) , the expectation value agreed , surprisingly , with the one - loop four - gluon amplitude , normalized by the tree amplitude , _ i.e. _ . next , brandhuber , heslop and travaglini @xcite showed that the same statement is actually true for the @xmath51-gon wilson loop for any @xmath51 , compared with the normalized one - loop amplitude @xcite for the so - called maximally - helicity - violating ( mhv ) configuration of gluon helicities ( two negative and @xmath146 positive ) . the wilson - loop computation knows nothing about the polarizations of the external gluons . it is manifestly symmetric under cyclic permutations and reflections of the polygon . for @xmath74 and 5 , a ward identity for @xmath0 supersymmetry shows that all helicity configurations in msym are equivalent , and that the normalized amplitudes have the same manifest symmetries as the polygonal wilson loop @xcite . however , beyond @xmath75 there are non - mhv configurations which do not have these symmetries . how does the wilson loop know it is `` supposed to '' match the mhv amplitude alone ? drummond , henn , korchemsky and sokatchev ( dhks ) then repeated the wilson - loop computation in msym at two loops , first for the @xmath74 case @xcite and then for the @xmath75 case @xcite . again the results matched the full two - loop msym scattering amplitudes @xcite , up to constants of the kinematics . furthermore , dhks first proposed @xcite and then proved @xcite an anomalous dual conformal ward identity for wilson loops , in which the anomaly arises from uv divergences proportional to @xmath19 . the solution to the ward identity is unique for @xmath74 and 5 . beyond @xmath75 , there are multiple solutions , due to the existence of nontrivial conformally - invariant cross ratios . for example , for @xmath147 the quantity @xmath148 is invariant under the inversion ( [ inversionsym ] ) , and there are two other such cross ratios . ( the appearance of @xmath149 in a cross ratio is forbidden by the on - shell constraint @xmath150 . ) dhks also showed that the amplitude ansatz ( [ bdsansatz ] ) obeys the anomalous dual conformal ward identity . given that the ansatz was known to be correct for @xmath74 and 5 @xcite , and the uniqueness of the ward identity solution for these cases , this result could explain why the amplitude should match the wilson loop in these cases . however , it was not clear what should happen for larger @xmath51 . indeed , alday and maldacena @xcite gave an argument , based on approximating a euclidean rectangular loop by a zig - zag configuration composed of many light - like segments , that the ansatz ( [ bdsansatz ] ) should fail at strong coupling for sufficiently large @xmath51 . dhks @xcite found that the hexagonal wilson loop could _ not _ be described at two loops by the ansatz ( [ bdsansatz ] ) . this result left open the question , however , of whether the ansatz failed to describe mhv amplitudes beyond @xmath75 , or whether the relation between amplitudes and wilson loops failed beyond two loops ( or both ) . the high - energy limits of the ansatz ( [ bdsansatz ] ) have been examined for consistency with expected regge behavior . for @xmath74 and 5 , the ansatz appears to have consistent behavior in all such limits @xcite however , there appears to be a difficulty with the ansatz for the six - gluon amplitude starting at two loops @xcite . very recently , a computation of the `` parity even '' part of the six - gluon mhv amplitude @xcite has revealed directly that the ansatz ( [ bdsansatz ] ) does fail for @xmath147 . however , a numerical comparison @xcite with the corresponding hexagonal wilson loop @xcite shows that the mhv - amplitude - wilson - loop equivalence is still intact at two loops and @xmath147 . this result means that the scattering amplitude also obeys the dual conformal ward identity . on the other hand , the solution to the ward identity is not unique for @xmath147 . hence some other principle , as yet unidentified , is needed to explain why mhv amplitudes are equivalent to wilson loops in msym . we have seen that gluon scattering amplitudes in planar @xmath0 super - yang - mills theory have some remarkable properties . it appears that the exact forms of the four - gluon and five - gluon amplitudes are given by the ansatz ( [ bdsansatz ] ) , which depends only on four different functions of the large-@xmath1 coupling parameter @xmath12 : @xmath151 , @xmath152 , @xmath153 and @xmath154 . because an exact solution for one of the four functions @xmath151 , the cusp anomalous dimension seems to be in hand @xcite , perhaps one can say that these cases are `` @xmath155 solved '' . the fixed dependence of the ansatz ( [ bdsansatz ] ) on the scattering angle(s ) is apparently related to the uniqueness of solutions to a dual conformal ward identity for @xmath74 and 5 @xcite , and an equivalence between ( mhv ) amplitudes and wilson lines @xcite . although the ansatz ( [ bdsansatz ] ) fails for the mhv six - gluon amplitude @xcite at two loops , the equivalence remains valid @xcite . there are still many open questions . are there simple(r ) ads / operator interpretations of the other three functions ? can one find integral equations for them , based on integrability ? what is the precise relation between integrability and dual conformal invariance ? do non - mhv amplitudes obey any simple patterns , or bear any relation to wilson loop expectation values ? from the structure of the one - loop amplitudes , _ e.g. _ for six gluons @xcite , any such relations must be considerably more intricate . what happens in other conformal theories ? finally , we can hope that some of these advances may eventually help to shed light on scattering amplitudes in other gauge theories , particularly qcd , whose understanding as exemplified by the other talks at this symposium is vital to the search for new physics at the large hadron collider . 0.4 cm * acknowledgments * + -0.2 cm i am grateful to stefano catani , massimiliano grazzini and the other organizers of radcor2007 , for the opportunity to present this talk , and for putting together a wonderful conference . i also thank babis anastasiou , zvi bern , michal czakon , david kosower , radu roiban , volodya smirnov , mark spradlin , cristian vergu and anastasia volovich for collaboration on some of the topics reviewed here . m. f. sohnius and p. c. west , phys . b * 100 * , 245 ( 1981 ) ; + s. mandelstam , nucl . b * 213 * , 149 ( 1983 ) ; + p. s. howe , k. s. stelle and p. c. west , phys . b * 124 * , 55 ( 1983 ) ; + p. s. howe , k. s. stelle and p. k. townsend , nucl . b * 236 * , 125 ( 1984 ) . g. t hooft , nucl . b * 72 * , 461 ( 1974 ) . j. m. maldacena , adv . * 2 * , 231 ( 1998 ) [ int . j. theor . * 38 * , 1113 ( 1999 ) ] [ hep - th/9711200 ] ; + s. s. gubser , i. r. klebanov and a. m. polyakov , phys . b * 428 * , 105 ( 1998 ) [ hep - th/9802109 ] ; + e. witten , adv . theor . * 2 * , 253 ( 1998 ) [ hep - th/9802150 ] ; + o. aharony , s. s. gubser , j. m. maldacena , h. ooguri and y. oz , phys . rept . * 323 * , 183 ( 2000 ) [ hep - th/9905111 ] . c. anastasiou , z. bern , l. j. dixon and d. a. kosower , phys . rev . lett . * 91 * , 251602 ( 2003 ) [ hep - th/0309040 ] . l. f. alday and j. m. maldacena , jhep * 0706 * , 064 ( 2007 ) [ 0705.0303 [ hep - th ] ] . n. beisert , b. eden and m. staudacher , j. stat . mech . * 0701 * , p021 ( 2007 ) [ hep - th/0610251 ] . z. bern , m. czakon , l. j. dixon , d. a. kosower and v. a. smirnov , phys . d * 75 * , 085010 ( 2007 ) [ hep - th/0610248 ] . f. cachazo , m. spradlin and a. volovich , phys . d * 75 * , 105011 ( 2007 ) [ hep - th/0612309 ] . m. k. benna , s. benvenuti , i. r. klebanov and a. scardicchio , phys . lett . * 98 * , 131603 ( 2007 ) [ hep - th/0611135 ] ; + a. v. kotikov and l. n. lipatov , nucl . b * 769 * , 217 ( 2007 ) [ hep - th/0611204 ] . b. basso , g. p. korchemsky and j. kotaski , 0708.3933 [ hep - th ] . s. s. gubser , i. r. klebanov and a. m. polyakov , nucl . b * 636 * , 99 ( 2002 ) [ hep - th/0204051 ] . s. frolov and a. a. tseytlin , jhep * 0206 * , 007 ( 2002 ) [ hep - th/0204226 ] . r. roiban , a. tirziu and a. a. tseytlin , jhep * 0707 * , 056 ( 2007 ) [ 0704.3638 [ hep - th ] ] ; + r. roiban and a. a. tseytlin , jhep * 0711 * , 016 ( 2007 ) [ 0709.0681 [ hep - th ] ] . z. bern , l. j. dixon and v. a. smirnov , phys . d * 72 * , 085001 ( 2005 ) [ hep - th/0505205 ] . z. bern and d. a. kosower , nucl . b * 379 * , 451 ( 1992 ) ; + z. bern , a. de freitas , l. dixon and h. l. wong , phys . d * 66 * , 085002 ( 2002 ) [ hep - ph/0202271 ] . v. v. sudakov , sov . jetp * 3 * , 65 ( 1956 ) [ zh . fiz . * 30 * , 87 ( 1956 ) ] ; + r. akhoury , phys . d * 19 * , 1250 ( 1979 ) ; + a. h. mueller , phys . d * 20 * , 2037 ( 1979 ) ; + j. c. collins , phys . d * 22 * , 1478 ( 1980 ) ; in _ perturbative qcd _ , ed . a. h. mueller , advanced series on directions in high energy physics , vol . 5 ( world scientific , singapore , 1989 ) [ hep - ph/0312336 ] ; + a. sen , phys . rev . d * 24 * , 3281 ( 1981 ) ; + g. sterman , nucl . b * 281 * , 310 ( 1987 ) . a. sen , phys . * d28 * , 860 ( 1983 ) ; + j. botts and g. sterman , nucl . b * 325 * , 62 ( 1989 ) ; + n. kidonakis , g. oderda and g. sterman , nucl . * b525 * , 299 ( 1998 ) [ hep - ph/9801268 ] ; nucl . * b531 * , 365 ( 1998 ) [ hep - ph/9803241 ] . g. p. korchemsky , mod . lett . a * 4 * , 1257 ( 1989 ) ; + g. p. korchemsky and g. marchesini , nucl . b * 406 * , 225 ( 1993 ) [ hep - ph/9210281 ] . l. magnea and g. sterman , phys . d * 42 * , 4222 ( 1990 ) . s. catani , phys . b * 427 * , 161 ( 1998 ) [ hep - ph/9802439 ] . g. sterman and m. e. tejeda - yeomans , phys . b * 552 * , 48 ( 2003 ) [ hep - ph/0210130 ] . a. m. polyakov , nucl . b * 164 * ( 1980 ) 171 . s. v. ivanov , g. p. korchemsky and a. v. radyushkin , yad . fiz . * 44 * , 230 ( 1986 ) [ sov . j. nucl . * 44 * , 145 ( 1986 ) ] . g. p. korchemsky and a. v. radyushkin , nucl . * b283 * , 342 ( 1987 ) . f. cachazo , m. spradlin and a. volovich , phys . d * 74 * , 045020 ( 2006 ) [ hep - th/0602228 ] ; + z. bern , m. czakon , d. a. kosower , r. roiban and v. a. smirnov , phys . * 97 * , 181601 ( 2006 ) [ hep - th/0604074 ] . t. o. eynck , e. laenen and l. magnea , jhep * 0306 * , 057 ( 2003 ) [ hep - ph/0305179 ] . r. j. eden , p. v. landshoff , d. i. olive , j. c. polkinghorne , _ the analytic s matrix _ ( cambridge university press , 1966 ) . z. bern , l. j. dixon , d. c. dunbar and d. a. kosower , nucl . b * 425 * , 217 ( 1994 ) [ hep - ph/9403226 ] . z. bern , l. j. dixon , d. c. dunbar and d. a. kosower , nucl . b * 435 * , 59 ( 1995 ) [ hep - ph/9409265 ] . z. bern , l. j. dixon and d. a. kosower , nucl . b * 513 * , 3 ( 1998 ) [ hep - ph/9708239 ] . z. bern , l. j. dixon and d. a. kosower , jhep * 0001 * , 027 ( 2000 ) [ hep - ph/0001001 ] ; jhep * 0408 * , 012 ( 2004 ) [ hep - ph/0404293 ] . r. britto , f. cachazo and b. feng , nucl . b * 725 * , 275 ( 2005 ) [ hep - th/0412103 ] . z. bern , j. s. rozowsky and b. yan , phys . b * 401 * , 273 ( 1997 ) [ hep - ph/9702424 ] . z. bern , j. j. m. carrasco , h. johansson and d. a. kosower , phys . d * 76 * , 125020 ( 2007 ) [ 0705.1864 [ hep - th ] ] . j. m. drummond , j. henn , v. a. smirnov and e. sokatchev , jhep * 0701 * , 064 ( 2007 ) [ hep - th/0607160 ] . d. j. broadhurst , phys . b * 307 * , 132 ( 1993 ) . l. n. lipatov , nucl . b * 548 * , 328 ( 1999 ) [ hep - ph/9812336 ] . j. m. drummond , g. p. korchemsky and e. sokatchev , nucl . b * 795 * , 385 ( 2008 ) [ 0707.0243 [ hep - th ] ] . f. cachazo and d. skinner , 0801.4574 [ hep - th ] . v. a. smirnov , phys . b * 460 * , 397 ( 1999 ) [ hep - ph/9905323 ] . v. a. smirnov , phys . b * 567 * , 193 ( 2003 ) [ hep - ph/0305142 ] . v. a. smirnov , _ evaluating feynman integrals _ , springer tracts in modern physics , * 211 * ( springer , berlin , heidelberg , 2004 ) . e. remiddi and j. a. m. vermaseren , int j. mod . phys . a * 15 * , 725 ( 2000 ) [ hep - ph/9905237 ] . a. v. kotikov , l. n. lipatov , a. i. onishchenko and v. n. velizhanin , phys . b * 595 * , 521 ( 2004 ) [ erratum - ibid . b * 632 * , 754 ( 2006 ) ] [ hep - th/0404092 ] . s. moch , j. a. m. vermaseren and a. vogt , nucl . b * 688 * , 101 ( 2004 ) [ hep - ph/0403192 ] ; nucl . b * 691 * , 129 ( 2004 ) [ hep - ph/0404111 ] . s. moch , j. a. m. vermaseren and a. vogt , phys . b * 625 * , 245 ( 2005 ) [ hep - ph/0508055 ] . f. cachazo , m. spradlin and a. volovich , phys . d * 76 * , 106004 ( 2007 ) [ 0707.1903 [ hep - th ] ] . j. gluza , k. kajda and t. riemann , comput . commun . * 177 * , 879 ( 2007 ) [ 0704.2423 [ hep - ph ] ] . c. anastasiou and a. daleo , jhep * 0610 * , 031 ( 2006 ) [ hep - ph/0511176 ] . m. czakon , comput . commun . * 175 * , 559 ( 2006 ) [ hep - ph/0511200 ] . d. e. berenstein , j. m. maldacena and h. s. nastase , jhep * 0204 * , 013 ( 2002 ) [ hep - th/0202021 ] . j. a. minahan and k. zarembo , jhep * 0303 * , 013 ( 2003 ) [ hep - th/0212208 ] . l. n. lipatov , hep - th/9311037 ; jetp lett . * 59 * , 596 ( 1994 ) [ pisma zh . fiz . * 59 * , 571 ( 1994 ) ] ; + l. d. faddeev and g. p. korchemsky , phys . b * 342 * , 311 ( 1995 ) [ hep - th/9404173 ] . v. m. braun , s. e. derkachov and a. n. manashov , phys . * 81 * , 2020 ( 1998 ) [ hep - ph/9805225 ] ; + v. m. braun , s. e. derkachov , g. p. korchemsky and a. n. manashov , nucl . b * 553 * , 355 ( 1999 ) [ hep - ph/9902375 ] ; + a. v. belitsky , phys . lett . b * 453 * , 59 ( 1999 ) [ hep - ph/9902361 ] . a. v. belitsky , a. s. gorsky and g. p. korchemsky , nucl . b * 667 * , 3 ( 2003 ) [ hep - th/0304028 ] . i. bena , j. polchinski and r. roiban , phys . d * 69 * , 046002 ( 2004 ) [ hep - th/0305116 ] . n. beisert , c. kristjansen and m. staudacher , nucl . b * 664 * , 131 ( 2003 ) [ hep - th/0303060 ] ; + n. beisert , nucl . b * 676 * , 3 ( 2004 ) [ hep - th/0307015 ] ; jhep * 0309 * , 062 ( 2003 ) [ hep - th/0308074 ] ; + n. beisert and m. staudacher , nucl . b * 670 * , 439 ( 2003 ) [ hep - th/0307042 ] ; + m. staudacher , jhep * 0505 * , 054 ( 2005 ) [ hep - th/0412188 ] . n. beisert and m. staudacher , nucl . b * 727 * , 1 ( 2005 ) [ hep - th/0504190 ] . n. beisert , bulg . j. phys . * 33s1 * , 371 ( 2006 ) [ hep - th/0511013 ] ; n. beisert , hep - th/0511082 . b. eden and m. staudacher , j. stat . mech . * 0611 * , p014 ( 2006 ) [ hep - th/0603157 ] . a. v. kotikov , l. n. lipatov , a. rej , m. staudacher and v. n. velizhanin , j. stat . mech . * 0710 * , p10003 ( 2007 ) [ 0704.3586 [ hep - th ] ] . g. arutyunov , s. frolov and m. staudacher , jhep * 0410 * , 016 ( 2004 ) [ hep - th/0406256 ] ; + r. hernndez and e. lpez , jhep * 0607 * , 004 ( 2006 ) [ hep - th/0603204 ] ; + r. a. janik , phys . d * 73 * , 086006 ( 2006 ) [ hep - th/0603038 ] ; + l. freyhult and c. kristjansen , phys . b * 638 * , 258 ( 2006 ) [ hep - th/0604069 ] ; + n. beisert , r. hernndez and e. lpez , jhep * 0611 * , 070 ( 2006 ) [ hep - th/0609044 ] . d. j. gross and p. f. mende , phys . b * 197 * , 129 ( 1987 ) . m. kruczenski , jhep * 0212 * , 024 ( 2002 ) [ hep - th/0210115 ] . e. i. buchbinder , phys . b * 654 * , 46 ( 2007 ) [ 0706.2015 [ hep - th ] ] . a. brandhuber , p. heslop and g. travaglini , nucl . b * 794 * , 231 ( 2008 ) [ 0707.1153 [ hep - th ] ] . m. t. grisaru , h. n. pendleton and p. van nieuwenhuizen , phys . d * 15 * , 996 ( 1977 ) ; + m. t. grisaru and h. n. pendleton , nucl . b * 124 * , 81 ( 1977 ) ; + z. bern , l. j. dixon , d. c. dunbar and d. a. kosower , phys . b * 394 * , 105 ( 1997 ) [ hep - th/9611127 ] . j. m. drummond , j. henn , g. p. korchemsky and e. sokatchev , nucl . b * 795 * , 52 ( 2008 ) [ 0709.2368 [ hep - th ] ] . j. m. drummond , j. henn , g. p. korchemsky and e. sokatchev , 0712.1223 [ hep - th ] . l. f. alday and j. maldacena , jhep * 0711 * , 068 ( 2007 ) [ 0710.1060 [ hep - th ] ] . j. m. drummond , j. henn , g. p. korchemsky and e. sokatchev , 0712.4138 [ hep - th ] . h. j. schnitzer , hep - th/0701217 ; 0706.0917 [ hep - th ] ; + s. g. naculich and h. j. schnitzer , nucl . b * 794 * , 189 ( 2008 ) [ 0708.3069 [ hep - th ] ] ; + v. del duca and e. w. n. glover , 0802.4445 [ hep - th ] . r. c. brower , h. nastase , h. j. schnitzer and c .- i . tan , 0801.3891 [ hep - th ] . j. bartels , l. n. lipatov and a. sabio vera , 0802.2065 [ hep - th ] . z. bern , l. j. dixon , d. a. kosower , r. roiban , m. spradlin , c. vergu and a. volovich , 0803.1465 [ hep - th ] . j. m. drummond , j. henn , g. p. korchemsky and e. sokatchev , 0803.1466 [ hep - th ] .

we present the detailed theoretical study of a heterostructure comprising of two coupled ferromagnetic superconducting layers . our model may be also applicable to the layered superconductors with alternating interlayer coupling in a parallel magnetic field . it is demonstrated that such systems exhibit a competition between the nonuniform larkin - ovchinnikov - fulde - ferrel ( fflo ) state and the superconducting state where the sign of the superconducting order parameter is opposite in adjacent layers . we determine the complete temperature - field phase diagram . in the case of low interlayer coupling we obtain a new phase inserted within the fflo phase and located close to the usual tricritical point , whereas for strong interlayer coupling the bilayer in the state reveals a very high paramagnetic limit and the phenomenon of field - induced superconductivity .

cond-mat0509599

the question of coexistence of singlet superconductivity and magnetism has been adressed for many years . it was found that the superconducting order parameter is destroyed by a magnetic field both via the orbital effect @xcite and the paramagnetic effect.@xcite in the usual case of an isotropic three - dimensional ( 3d ) superconductor under an external magnetic field , the orbital effect prevails and leads to the well - known temperature - field phase diagram of conventional type i or ii superconductors.@xcite in contrast , superconductivity is essentially suppressed by the paramagnetic effect in presence of a ferromagnetic exchange interaction . this is also true for quasi - two - dimensional ( 2d ) superconductors under in - plane magnetic field and for heavy fermions materials wherein the orbital effect is partially quenched . in the whole paper , the magnetism is characterized by an internal exchange field @xmath1 ( given in energy units ) which may arise either from an externally applied magnetic field or from ferromagnetic ordering . note that ferromagnetism must be weak in order to avoid complete suppression of superconductivity . this is realized in rare - earth metals or actinides in which the indirect exchange interaction leads to curie temperatures of a few degrees . superconductors with internal homogeneous exchange field @xmath1 exhibit a very special behaviour . according to chandrasekhar@xcite and clogston,@xcite at zero temperature uniform superconductivity should be destroyed when the polarization energy of the free electron gas exceeds the energy gain due to cooper pairing in the bcs ground state . this criterion gives the exchange field @xmath2 where the superconductor should undergo a first - order transition to the normal state , @xmath3 being the zero temperature superconducting gap . larkin and ovchinnikov @xcite and fulde and ferrell @xcite ( fflo ) predicted the existence of a nonuniform superconducting state with higher critical exchange field @xmath4 and second - order transition to the normal state . this prediction was made for 3d superconductors . in quasi-2d superconductors the critical exchange field of the fflo state is even higher , namely @xmath5,@xcite while in quasi - one - dimensional systems there is no paramagnetic limit at all.@xcite the appearance of the modulated fflo state is related to the pairing of electrons with opposite spins which do not have the opposite momenta anymore due to the zeeman splitting . from now on we focus on the 2d case for which a generic temperature - exchange field phase diagram has been established.@xcite at low field and temperature , the ground state is characterized by a uniform superconducting order parameter . a tricritical point , located at @xmath6 and @xmath7 , is the meeting point of three transition lines separating the normal metal , the uniform and the nonuniform superconductors . at @xmath8 , the ( low - field ) uniform superconductor is separated from the ( high - field ) normal metal by a narrow fflo nonuniform superconducting phase . in contrast , at @xmath9 , the system undergoes merely a second - order phase transition from the uniform superconductor to the normal metal when increasing the exchange field . the nonuniform fflo state is settled in a small region of the phase diagram and is very sensitive to impurities,@xcite making it difficult to observe experimentally . nevertheless , several evidences of the fflo state have been obtained recently in organic superconductors @xcite and in heavy fermions compounds , see martin _ _ et al.__@xcite and references therein . in the context of organic and high-@xmath10 superconductors , layered systems made of conducting atomic planes have been extensively studied.@xcite in order to investigate the interplay of superconductivity and magnetism in such anisotropic systems,@xcite andreev _ et al . _ considered a periodic array of alternating ferromagnetic and superconducting 2d planes.@xcite solving the corresponding gorkov equations , these authors established the existence of a @xmath0 state wherein each f layer separates superconducting planes with opposite order parameter . this is relevant for the ruthenocuprate compound rusr@xmath11gdcu@xmath11o@xmath12 which comprises cuo@xmath11 superconducting planes and ruo@xmath11 magnetic planes.@xcite a related system is an isolated f / s / f trilayer which exhibits the so - called superconducting spin - valve effect . namely , its critical temperature is higher in the antiparallel ( ap ) orientation of the layers magnetizations than in the parallel ( p ) orientation both for thick layers@xcite and atomic size layers.@xcite surprisingly , in the atomic thickness limit , the superconducting gap at zero temperature is higher for p orientation of the magnetizations.@xcite hence one expects a transition from ap to p orientation by cooling the system below a finite crossing temperature . the recent progress in molecular beam epitaxy@xcite enables to fabricate such f / s / f trilayer with atomic thicknesses . in this paper , we consider a periodic array of sf bilayers . each bilayer is made of two atomic planes coupled by single electron tunneling . both exchange fields and bcs superconducting pairing are present in each sf plane . the possibility of @xmath13 phase difference between the planes inside each bilayer is also taken into account . in the whole paper , we assume that the coupling @xmath14 between successive bilayers is considerably weaker than the intra - bilayer coupling @xmath15 . our study is performed within the framework of the bcs theory of s - wave superconductivity . solving exactly the gorkov equations in the limit @xmath16 , we first derive the critical temperature and the superconducting gap both for parallel ( p ) and antiparallel ( ap ) orientation of the magnetizations . we show that the critical temperature is higher for the ap orientation than for the p orientation whereas it is the opposite for the zero temperature gap . we also investigate the interlayer josephson current in the small coupling limit : the current increases as a function of the exchange field for ap orientation whereas it is field - independent for the p orientation . furthermore , we find that for low exchange fields and high temperatures , the ground state corresponds to identical superconducting order parameters on adjacent layers . for high enough fields and/or low enough temperatures , the @xmath0 phase ground state is favoured and compete with the fflo state . for the p orientation , the full temperature - exchange field phase diagram is constructed in the two limits of extremely low and high coupling between the planes . as expected , for perturbative coupling between two sf planes , the phase diagram is very close to the quasi-2d superconductor s phase diagram . nevertheless an important change arises . indeed a new @xmath17phase is inserted inside the usual fflo phase close to the tricritical point . for higher tunneling coupling @xmath18 , this @xmath17 phase is pushed to low temperatures @xmath19 and high fields @xmath20 . in this unusual superconducting phase , the zeeman splitting is compensated by the bonding / antibonding energy splitting due to single - electron tunneling between the planes.@xcite as a result , field - induced superconductivity and enhanced paramagnetic limit are realized in this simple model . these new phenomena are encountered due to the introduction of an additional discrete degree of freedom , here the layer index @xmath21 . the layer index acts as a pseudo - spin and thus enlarges the usual spin - space for singlet pairing . this idea was introduced by kulic and hofmann@xcite in the context of two - bands superconductivity for which the pseudo - spin was the band index . nevertheless , these authors did not investigate the presently studied @xmath0 state . the outline of the paper is the following . in sec.ii , we present the model , derive the corresponding gorkov equations and give their exact solutions . in sec.iii , we investigate the critical temperature , the gap and the interlayer josephson current in the small exchange field regime for which there are only uniform superconducting phases . in the last two sections the temperature - exchange field phase diagram of the bilayer is studied thoroughly . in sec.iv , we first construct a ginzburg - landau functional to determine the transitions between the different phases in the low interlayer coupling limit . sec.v is devoted to the opposite limit of strong interlayer coupling . in conclusion , we discuss the conditions for the observation of field - induced superconductivity . we consider a superconducting ferromagnetic bilayer ( see fig.1 ) constituted of two superconducting atomic layers , labeled as @xmath22 and @xmath23 . in the whole article , we assume @xmath24 where @xmath25 is the interlayer coupling energy and @xmath26 the fermi energy . as a consequence , cooper pairs are localized within each plane.@xcite each layer @xmath21 supports a superconducting singlet bcs coupling with the energy gap @xmath27 and an internal exchange field @xmath28 . the hamiltonian of the system can be written as @xmath29 + h_{t}\text { , } \label{hamiltonien}%\ ] ] where @xmath30 is the attractive bcs interaction constant and @xmath31 is the two - dimensional coordinate within each layer . for the layer @xmath21 the kinetic and zeeman parts of the hamiltonian are written together as @xmath32 in which summation over repeated spin indexes @xmath33 and @xmath34 is implied . creation ( resp . annihilation ) operator of an electron with spin @xmath33 and two - dimensional momentum @xmath35 in the layer @xmath21 is denoted @xmath36 ( resp . @xmath37 ) . the exchange fields @xmath28 are assumed to be either equal ( @xmath38 ) or opposite ( @xmath39 ) . as a consequence the matrix @xmath40 is spin - diagonal , and the zeeman effect manifests itself in breaking the spin degeneracy of the electronic energy levels according to @xmath41 \text{,}%\ ] ] where @xmath42 . the s - wave singlet superconductivity is represented by the standard mean - field hamiltonian @xmath43 \text{,}%\ ] ] and the layers are coupled together by the hopping hamiltonian @xmath44 \text { .}%\ ] ] in order to investigate the occurence of modulated superconducting phases ( fflo ) , we choose the following spatial dependence for the superconducting order parameter @xmath45 where @xmath46 is the fflo modulation wave vector and @xmath47 the superconducting phase difference between the layers . the above model can be solved exactly using the green functions @xmath48 where @xmath21 and @xmath49 are the layer s indexes . the brackets mean statistical averaging over grand - canonical distribution.@xcite we obtain the following gorkov equations in the fourier representation:@xmath50 where @xmath51 are the fermionic matsubara frequencies . in quasi-2d superconductors @xcite the maximal fflo modulation amplitude is of the order of @xmath52 , @xmath53 being the typical superconducting coherence length . this means that with a good approximation we can consider @xmath54 , @xmath55 being the fermi velocity vector in the plane . solving the gorkov equations ( [ equationgorkov ] ) yields the anomalous gorkov green function for the @xmath22 sf layer @xmath56 where @xmath57 with @xmath58 and @xmath59 similar equation holds for @xmath60 note that in the case where a lattice made of such sf / sf bilayers is considered , the generalized anomalous green function is obtained by replacing @xmath61 by @xmath62 in eq.([fcroix ] ) , @xmath63 being the projection of the momentum @xmath35 along the @xmath64 axis and @xmath65 the period of the lattice . as a consequence , a finite inter - bilayer coupling @xmath66 introduces an anisotropy in the dispersion relation which leads to a broadening of the electronic excitation levels . in the absence of tunneling @xmath67 we retrieve from eq.([fcroix ] ) the anomalous green function of a quasi-2d superconductor with the exchange field @xmath68 @xmath69 although the dependence on momentum has been removed for simplicity , notice that @xmath70 and @xmath71 the set of basic equations ( [ equationgorkov ] ) must be completed by the self - consistency equation @xmath72 close to the critical temperature @xmath10 of the second - order phase transition , the order parameters @xmath27 are small and eq.([self ] ) can be written as @xmath73 where @xmath74 is the critical temperature for the 2d superconducting single layer in the absence of exchange field , namely for @xmath75 . at zero temperature , it is convenient to write eq.([self ] ) as @xmath76 where @xmath77 is the superconducting order parameter at @xmath78 in the absence of exchange field and interlayer coupling . in this section , we investigate phases with uniform superconductivity within each layer . we obtain the critical temperature of the second - order superconducting ( @xmath79 ) to normal metal ( @xmath80 ) phase transition and the order parameter @xmath81 as a function of the temperature , the exchange field @xmath1 and the interlayer coupling @xmath25 . we consider both parallel ( p ) and antiparallel ( ap ) orientations of the magnetizations , the superconducting phase difference being either @xmath82 or @xmath13 . we also calculate the josephson interlayer current when the bilayer is connected to external superconducting leads . most of these results are obtained in the perturbative limit of small coupling between the layers @xmath83 . the field is also assumed to be sufficiently small to prevent the occurence of a spatial modulation of the superconductivity within the planes . study of nonuniform phases and strong coupling @xmath84 are respectively postponed to sec . iv and sec . v. we consider the second - order phase transition between the normal metal and the uniform bcs superconductor . thus the order parameters @xmath85 and @xmath86 are small and the anomalous gorkov green function ( [ fcroix ] ) can be linearized in the following form @xmath87 where @xmath88 . similar equation may be found for @xmath89 . we first consider _ the parallel ( p ) orientation of the magnetizations , _ namely _ @xmath90_. the first possibility is @xmath82 wherein the layers have the same superconducting order parameters @xmath91 . in this situation the anomalous green function obtained from eq.([fcroixlinearisee ] ) is denoted @xmath92 . the identity @xmath93 and eq.([selftnonnul ] ) yield the following implicit equation for the critical temperature @xmath94 @xmath95 where @xmath96 denotes the euler digamma function . therefore the interlayer coupling disappears from the self - consistency equation and eq.([tcp0 ] ) is identical to that for the 2d monolayer in a uniform exchange field : the bilayer is equivalent to a single layer in the neighborhood of superconducting to normal state transition.@xcite the critical temperature of the bilayer decreases when the exchange field @xmath1 increases . the equation ( [ tcp0 ] ) describes the second - order phase transition between the normal metal and the uniform superconductor which is realised only for fields smaller than the tricritical one @xmath6 . for larger fields , superconductivity becomes nonuniform . a second possibility is _ the p orientation with @xmath13 phase difference between the layers_. now the anomalous gorkov green function is denoted @xmath97 . then @xmath98 , \ ] ] and the self - consistency relation eq.([selftnonnul ] ) yield a critical temperature @xmath99 given by @xmath100 from this expression one may notice that superconductivity in the @xmath0 state is destroyed by a combination of two effective exchange fields @xmath101 . in the small interlayer coupling limit @xmath102 , eq.([tcppi ] ) becomes @xmath103 where the function @xmath104 is defined and represented in appendix a. in the regime of low magnetic fields , namely for @xmath105 , the factor @xmath106 is positive and thus the critical temperature is smaller in the @xmath0 superconducting state than in the @xmath107 state . however the situation may be inverted if @xmath108 . moreover , along the critical line , the value @xmath109 corresponds to the tricritical point , @xmath110 and @xmath111 , where fflo nonuniform states appear . as a consequence one expects competition between the @xmath0 superconducting phase and fflo phases in the neighborhood of the tricritical point . this competition will be detailed in sec . iv . let us focus on the case of _ ap orientation @xmath112_. following the same procedure as previously , the equations for the critical temperatures @xmath113 are obtained . in the limit @xmath83 , it reads @xmath114 for @xmath82 , and @xmath115 for @xmath13 . from eqs.([tcap0petitt],[tcappi ] ) the critical temperature is clearly higher in the @xmath107 phase than in the @xmath0 phase . therefore the @xmath107 phase is the more stable in this region of the @xmath116 phase diagram , _ i.e. _ in the vicinity of the critical temperature and for low fields @xmath117 in conclusion , the bilayer is always in the @xmath107 superconducting state for temperatures close to the critical temperature , whatever the relative orientation of magnetizations is . a spin - valve effect is also present : the critical temperature is higher for the ap orientation than for the p orientation of the magnetizations . for @xmath67 and low fields @xmath118 it is well - known that the zero temperature gap @xmath119 is field - independent.@xcite for small interlayer coupling @xmath120 , the anomalous green function ( [ fcroix ] ) may be expanded to the second order in @xmath25 as @xmath121 where the full nonlinear dependence on @xmath122 is kept in @xmath123,@xmath124 and @xmath125 then self - consistency relation ( [ selft=0 ] ) becomes @xmath126 where @xmath127 and @xmath82 or @xmath13 . using the preceding equation in _ the p orientation _ we obtain @xmath128 , either for @xmath107 or @xmath0 phase difference . as a result , the superconducting gap @xmath129 is not affected by a small interlayer coupling , at least at the order of @xmath61 . the superconducting condensation energy gain has also been calculated and the zero state found to be more stable than @xmath0 state . _ for the ap orientation _ , the superconducting gap @xmath130 is given by @xmath131 \text { , } \label{gapapzero}%\ ] ] for _ zero phase difference . _ the unphysical divergence at @xmath132 is removed by terms of higher order in @xmath25 . expression ( [ gapapzero ] ) is the main result of this paragraph and reduces to @xmath133 in the small field regime @xmath134__. _ _ therefore in the @xmath82 state and for ap orientation , the order parameter is suppressed by the exchange field in the small coupling limit . this is surprising because ap orientation was expected to weaken the effective exchange field and thus enhance superconducting properties . nevertheless such a decrease of the superconducting order parameter has already been found in a ballistic atomic - scaled f / s / f trilayer.@xcite _ for the ap orientation and _ @xmath135 _ phase difference , _ the gap @xmath136 is field and coupling independent . moreover the energy of the @xmath13 state does not depend on the relative orientation of the magnetizations . to summarize , the lowest energy corresponds to the @xmath137 phase . the @xmath138 and @xmath139 phases are degenerate with a somewhat higher energy than the @xmath137 phase . although we have not performed the energy calculation in the case where the magnetizations are antiparallel and the phase difference is @xmath107 , we believe that the highest energy corresponds to the @xmath140 phase since its order parameter is the smallest one . we now extend our investigation of the superconducting gap to finite temperatures . in order to determine the gap @xmath81 as a function of temperature @xmath141 , exchange field @xmath1 and coupling @xmath25 , we analyse numerically the self - consistency relation ( [ selft=0 ] ) using the exact anomalous gorkov green function ( [ fcroix ] ) . the result is shown schematically in fig.2 . _ for the p orientation and @xmath82 _ , the superconducting gap @xmath142 is the same as the gap @xmath143 of a single layer whereas _ for @xmath13 _ the superconducting gap @xmath144 is lowered by finite interlayer coupling . _ for the ap orientation and @xmath82 _ , the gap is smaller than @xmath143 for @xmath145 and larger for @xmath146 where the inversion temperature @xmath147 depends only on the exchange field in the small interlayer coupling limit ( see fig.2 inset ) . this phenomenon has been called inversion of the proximity effect.@xcite moreover , the gap @xmath148 is larger than @xmath149 for all temperatures . according to these results , one may suggest several experiments . first we consider a bilayer with magnetizations pinned in the ap mutual orientation . by lowering the temperature , a @xmath107-@xmath0 transition is expected at some temperature @xmath150 . in the small interlayer coupling limit , this temperature @xmath151 is a function of the exchange field only ( see fig.2 inset ) . in contrast , the @xmath107 state is more favorable energetically for all temperatures in the case of magnetizations pinned in the p orientation . as another illustration we consider samples where the relative orientation of magnetizations is free . then the orientation is chosen by the system to minimize its energy . cooling such a bilayer will result in a switching from the ap orientation to the p orientation at the inversion temperature @xmath152 . the same prediction was made recently in a ballistic f / s / f trilayer.@xcite here we consider that the sf / sf bilayer is connected to superconducting electrodes . in this set - up , one may impose an arbitrary superconducting phase difference @xmath47 between the sf layers , and thus a non dissipative josephson current flows through the bilayer in the direction perpendicular to the planes . this interlayer josephson current is evaluated here in the tunneling limit @xmath153 and at zero temperature . within the green functions formalism , the general formula for the interlayer josephson current is @xmath154 where @xmath155 is the two - dimensional density of state per spin direction and unit surface . solving exactly the gorkov equations ( [ equationgorkov ] ) leads to @xmath156 the function @xmath157 is obtained from eq.([g21 ] ) by permuting the layer indexes @xmath158 the corresponding anharmonic current - phase relationship is given by@xmath159 in the tunneling limit @xmath160 the interlayer josephson current becomes sinusoidal as a function of the phase difference , @xmath161 where @xmath162 . the second harmonic @xmath163 has also been evaluated and is smaller than the first one by a factor @xmath164 . the preceding equation ( [ courant1 ] ) yields the current - phase relation both for parallel ( p ) and antiparallel ( ap ) orientation of magnetizations . _ in the parallel case _ the critical current does not depend on the field as already reported in other systems since @xmath165.@xcite _ for the antiparallel orientation _ and to the lowest order in @xmath25 , the current - phase relation reads @xmath166 where @xmath167 @xmath168 . therefore the critical current increases with the exchange field @xmath1 and even diverges for @xmath169 of course this divergence is unphysical and should disappear if all orders in @xmath25 were taken into account . in fig.3 , the critical current is shown as a function of the exchange field both for p and ap orientations . recently , the issue of the josephson coupling between two clean sf layers through an insulating layer was considered using eilenberger equations @xcite or bogoliubov - de gennes formalism.@xcite similar results as ours were obtained : the critical current increases with @xmath1 only if three conditions are met : low temperature , very weak coupling between the sf layers and ap orientation . otherwise the presence of an exchange interaction suppresses the josephson current . using usadel equations , krivoruchko demonstrated that this statement holds in the diffusive regime for which the divergence for @xmath170 is remplaced by a regular peak.@xcite from now on , we consider the sf / sf bilayer only for the parallel ( p ) orientation . hence the results obtained in the next sections may be also applied to a superconducting bilayer in an external in - plane magnetic field . the present section is devoted to the weak coupling regime @xmath171 in contrast to the low field restriction of sec . iii , regions of the phase diagram with @xmath172 are also investigated here . then competition between the fflo and @xmath0 phases is expected to take place . particular attention is paid to the vicinity of the tricritical point given by @xmath173 and @xmath174@xcite in order to examine this narrow region of the @xmath175 plane , we construct a ginzburg - landau ( gl ) functional from the gorkov equations used in the previous sections . in the past buzdin and kachkachi @xcite derived a generalized ginzburg - landau ( gl ) functional for a single sf layer that describes the fflo superconducting state near the tricritical point . here we extend this functional to a sf / sf bilayer for which it is possible to have not only fflo modulation within the planes but also @xmath13 superconducting phase difference between the planes . for @xmath82 , the physics of the bilayer is independent of the coupling and thus the buzdin - kachkachi gl functional is retrieved . in contrast for @xmath13 , we obtain a free energy functional which depends on the interlayer coupling @xmath25 and leads to the presence of a superconducting @xmath0 phase . the free energy of the sf / sf bilayer in a uniform superconducting state with @xmath82 ( @xmath176 state ) is given by ( see details in appendix b ) @xmath177 with @xmath178 where @xmath179 and @xmath180 are respectively the reduced exchange field and the reduced order parameter . here we retrieve the well - known case of a single sf layer . for reduced exchange fields lower than the tricritical one @xmath181 , the transition between the superconducting and the normal states is a second - order one since @xmath182 . the critical line is given by the equation @xmath183 for higher fields , the transition becomes a first - order one because @xmath184 and @xmath185 . as for any first - order transition , two conditions must be fullfilled . on one hand , the free energy ( [ freezero ] ) is minimized , ( @xmath186 , and on the other hand the free energies of the superconducting and normal phases are equal , @xmath187 hence in the @xmath175 plane the equation for this first - order line is @xmath188 @xmath189 , and the jump of the superconducting gap at the transition is given by @xmath190 . it is well - known that this scenario is not realized because it is replaced by a transition between normal metal and nonuniform superconductivity.@xcite nevertheless , this first - order transition provides a useful energy scale @xmath191 and a reference line in the @xmath175 plane that will be used to construct a universal phase diagram , namely a @xmath25-independent phase diagram valid for any weakly coupled sf / sf bilayers , see sec . the sf / sf bilayer may support opposite order parameters on the layers , superconductivity being still uniform within each sf plane . in this so - called @xmath192 state , the free energy of the bilayer depends on the reduced interlayer coupling @xmath193 according to @xmath194 for low reduced fields @xmath195 @xmath196 . hence the uniform superconducting phase with @xmath82 is more stable than the @xmath0 phase , as already found in the sec . interestingly for higher reduced fields @xmath197 this @xmath0 phase is in competition with fflo nonuniform superconducting phases having either @xmath82 ( @xmath198 ) or @xmath13 ( @xmath199 ) . according to buzdin and kachkachi the order parameter @xmath200 leads to the lowest energy.@xcite for the @xmath198 phase , the corresponding free energy reads @xmath201 whereas for the @xmath199 phase , the free energy depends on the interlayer coupling @xmath25 in the following manner @xmath202 the notation @xmath203 is introduced in appendix b. now we proceed to analyse the above free energies in order to determine the critical line between the normal and the superconducting states . we will also describe the nature of the various superconducting states and what kinds of transitions are encountered . we focus on the vicinity of the tricritical point . then @xmath204 is a linear function of the exchange field with @xmath205 , whereas @xmath185 is nearly field and temperature independent . for high reduced fields @xmath206 , it appears that the @xmath176 and the @xmath199 never lead to the highest critical temperature . then we emphasize the competition between the two remaining phases , namely @xmath192 and @xmath207 the minimization of the gl functional ( [ fflozero1 ] ) leads to a modulation wave vector given by @xmath208 in the limit @xmath209 , _ i.e. _ near the critical line . the @xmath192 phase is more stable than the @xmath198 phase under the energy condition @xmath210 or equivalently for @xmath211 therefore the uniform @xmath0 phase is `` inserted '' within the usual fflo superconducting state . the upper and lower values of @xmath212 between which this new @xmath0 phase is stable depend on the particular value of the coupling @xmath25 . it is convenient to define a dimensionless generalized coordinate @xmath213 @xmath214 that quantifies the `` distance '' from the tricritical point along the s / n transition line . indeed @xmath215 at the tricritical point and the @xmath0 phase is settled in the region @xmath216 as shown on fig . 4 , going along the critical line from low to high fields , one expects the sequence of superconducting states : uniform in the planes with @xmath82 for @xmath217 , fflo modulation along the planes with @xmath82 for @xmath218 then uniform @xmath0 state for @xmath219 and finally fflo modulation along the planes with @xmath82 for @xmath220 . in all cases , the sign of the @xmath221 coefficient in the gl free energy is always positive and thus transitions between these superconducting states and the normal metal are second - order ones . we now construct the phase diagram around the tricritical point . because each value of the coupling leads to different transition lines , we introduce the following mapping of the thermodynamic variables @xmath223 in order to obtain a universal phase diagram valid in the small coupling regime . this mapping makes use of the energy scale @xmath191 and of the function @xmath224 related to the first - order transition between the normal state and the uniform superconducting state , see sec . iv.a . then the free energies for the uniform superconducting phases eqs.([freezero],[freepi ] ) become @xmath225 and @xmath226 where @xmath227 first it is straightforward to minimize @xmath228 and @xmath229 with respect to @xmath230 . then replacing the reduced gap @xmath230 by its equilibrium value , one obtains the equilibrium energies @xmath231 and @xmath232 of each superconducting phase . these energies are functions of both field and temperature via the dimensionless thermodynamical variables @xmath233 and @xmath234 . using the same scaling eq.([scaling ] ) the free energies of the @xmath198 phase , @xmath235 and of the @xmath199 phase , @xmath236 are also obtained for the order parameter @xmath237 . the equilibrium energies @xmath238 and @xmath239 of the modulated phases are obtained after minimization of eqs.([energyfflozero],[energyfflopi ] ) with respect to @xmath230 and @xmath240 .then eq.([energyfflozero ] ) enables to study the second - order phase transition between the normal metallic state and the nonuniform fflo state@xmath241 under the assumption of second - order phase transition , it is sufficient to consider the gl free energy up to the @xmath242 order . then the free energy ( [ energyfflozero ] ) is minimal for @xmath208 . for this particular modulation , the critical fflo / n line is given by @xmath243 we now consider the transition lines between the various superconducting states obtained in the previous paragraph , in particular the @xmath198/@xmath192 , the @xmath198/@xmath176 and the @xmath244 transitions . let us focus on the transition between the uniform phases @xmath176 and @xmath192 . solving @xmath245 , we obtain the critical line @xmath246 which corresponds to a first order @xmath247 phase transition . however , this transition is not realized ( see fig.4 ) because the transition to the nonuniform superconducting state occurs before . the @xmath248 transition line is obtained in a similar way . the equation @xmath249 has the solution @xmath250 . at this value of @xmath233 , the system undergoes a first order phase transition from the uniform state to the modulated fflo state . adding higher harmonics to the order parameter @xmath251 gives a more accurate evaluation , namely @xmath252 . finally , the @xmath253 transition line is obtained from @xmath254 and shown in fig.4 in the @xmath255 plane . this transition is a first - order one . using the mapping @xmath256 of the thermodynamical variables , we have obtained a universal phase diagram fig.4 of all weakly coupled sf / sf bilayers in the vicinity of the tricritical point . an important feature of this phase diagram is the presence of a superconducting @xmath257-phase . as an example , the phase diagram has been redrawn in the @xmath175 plane on fig.5 for a particular value of the coupling @xmath25 . now we consider the ballistic sf / sf bilayer in the regime of strong interlayer coupling limit @xmath84 and low temperature . a very unusual @xmath257-superconducting state is found between a lower @xmath258 and an upper @xmath259 critical exchange field , and below a maximal temperature of the order of @xmath260 . therefore field - induced superconductivity is obtained above @xmath261 within the bcs theory of superconductivity . the underlying physical mechanism is the compensation of the zeeman splitting by the energy splitting between bonding and antibonding electronic states of the bilayer , see fig.6.@xcite thus the new zero temperature paramagnetic limit @xmath259 may be tuned far above the usual one @xcite @xmath262 merely by increasing the interlayer coupling . this compensation also occurs for small coupling , but the @xmath257-superconducting state is then less energetically favorable than the usual @xmath107-superconducting phase as demonstrated in sec.iv . therefore the @xmath175 phase diagrams are topologically distinct in the opposite limits of small ( sec.iv ) and strong ( sec.v ) coupling . we first analyse the second - order superconducting / normal phase transition in sec.v.a . then the first - order transition between uniform superconductivity and the normal state is discussed in sec.v.b . here we study the second - order phase transition between the @xmath0 superconducting state and the normal metal state , as a function of the field . we start from the the linearized anomalous green function ( [ fcroix ] ) for arbitrary coupling @xmath25 and @xmath0 superconducting phase difference , @xmath263 .\left [ t^{2}% -(i\omega+h+\xi)^{2}\right ] } \text{. } \label{fcroixgrandt}%\ ] ] from this equation and the self - consistency relation ( [ self ] ) , the critical exchange field @xmath1 is shown to satisfy @xmath264 where @xmath265 , and @xmath266 is the critical exchange field for the second - order superconducting phase transition in a two - dimensional monolayer . one must then find the value of @xmath267 which maximizes the critical field @xmath268 . if the @xmath0 phase is assumed to be uniform inside each plane , namely if @xmath269 , eq.([hcritfflo1 ] ) merely reduces to @xmath270 the lower and upper critical fields are respectively given by @xmath271 , in the limit @xmath272 . thus at zero temperature and strong enough coupling , the superconductivity destruction follows a very special scenario . at low fields , superconductivity is first suppressed as usual at the paramagnetic limit @xmath273 leading to the normal metal phase . then further increase of the field leads to a normal to superconducting phase transition at the lower critical field . this superconducting @xmath0 phase is finally suppressed at the upper critical field . this is a new paramagnetic limit which may be tuned far above the usual one merely by choosing the coupling @xmath25 greater than @xmath274 . thorough analysis of eq.([hcritfflo1 ] ) shows that the upper critical field is even increased by an in - plane modulation in analogy with the case of the two - dimensional fflo phase.@xcite the upper critical field is maximal for the choice @xmath275 , and then eq.([hcritfflo1 ] ) reduces to @xmath276 that gives the upper and lower fields @xmath277 in the @xmath278 limit . note that the period of the modulated order parameter @xmath279 is larger than the corresponding period in the two - dimensional fflo phase which coincides with the ballistic coherence length @xmath280 @xcite furthermore one may derive the full temperature - field phase diagram using eqs.([selftnonnul],[fcroixgrandt ] ) and the result is shown in fig.7 . when the temperature is increased , the lower critical field increases whereas the upper one decreases . along the upper ( resp . lower ) critical line the fflo modulation is lost at some temperature @xmath281 ( resp . @xmath282 ) . for higher temperatures a uniform @xmath0 phase ( @xmath192 ) is recovered and the temperature dependence of the critical field is given by @xmath283 , \label{hcriticaltemperature}%\ ] ] where @xmath284 is the digamma function and @xmath285 , @xmath286 being the euler constant . finally the lower and upper critical lines merge at field @xmath287 and temperature @xmath288 in the limit @xmath289 . therefore the field - induced @xmath0 superconductivity is confined to temperatures lower than @xmath290 . the structure of these @xmath192 and the @xmath199 phases is reminiscent of the corresponding @xmath176 and the @xmath198 phases although the former are shifted to higher fields and lower temperatures than the later . above results were obtained for relatively strong coupling . for lower coupling @xmath291 , the @xmath192 and the @xmath199 phases merge continuously into the usual @xmath82 phases as shown in fig.8 , and finally disappear for @xmath25 slightly smaller than @xmath274 . from an experimental point of view , one might choose a system with intermediate coupling @xmath25 small enough to settle the @xmath0 phase island in an available range of temperatures but also large enough to separate the @xmath0 phase island from the usual superconducting phases with @xmath82 . in the general sf multilayer case the inter - bilayer coupling constant @xmath66 needs to be sufficiently high to prevent from superconductivity destruction by 2d fluctuations but also sufficiently low to preserve the effect of field - induced superconductivity.@xcite in the following we investigate the first - order @xmath292 transition to determine whether it is more or less favorable than the above studied second - order transition . the zero - temperature superconducting order parameter @xmath293 is calculated from the self - consistency equation ( [ selft=0 ] ) for p orientation of magnetizations and @xmath13 phase difference . at zero temperature , the difference between the energy @xmath294 of the superconducting state and the energy @xmath295 of the normal metal state is given by @xcite@xmath296 d\delta\label{energieself}%\ ] ] in the limit @xmath297 we retrieve the well - known case of the single sf layer.@xcite then the self - consistency relation ( [ selft=0 ] ) admits two branches of solutions . the lower branch @xmath298 , labelled ( 2 ) in the inset of fig.9 , corresponds to a positive energy cost @xmath299 . thus this superconducting solution is never realized . the actual superconducting gap is given by the upper horizontal branch @xmath300 , ( 1 ) in the inset of fig.9 , which corresponds to the energy difference @xmath301 hence the superconducting phase is settled for low fields @xmath302 with a field - independent order parameter @xmath303 . for higher fields @xmath304 , the system is in the normal phase @xmath305 . finally the zero temperature gap exbibits a jump at @xmath306 which reveals the first - order transition from the uniform superconducting phase to the normal phase . in the opposite limit of strong interlayer coupling , we have obtained in sec.v.a . that field - induced superconductivity with @xmath13 phase difference occurs for fields close to @xmath25 and at low temperatures . from the self - consistency equation ( [ selft=0 ] ) one obtains several possible solutions for the zero - temperature superconducting gap @xmath307 as a function of the exchange field @xmath1 , see fig.9 . for relatively low fields @xmath308 and for high fields @xmath309 , the bilayer is in the normal phase @xmath310 the limiting fields @xmath311 are solutions of @xmath312 for intermediate fields ranging between @xmath313 and @xmath314 there are three superconducting branches . two of them , ( 2 ) and ( 2 ) are never realized owing to their energy cost @xmath315 the third branch ( 1 ) requires more detailed analysis . namely , it is given by the equation @xmath316 and the corresponding energy cost is @xmath317 \nonumber\\ & + \frac{\pi(h - t)^{2}}{2}. \label{energiemoyen}%\end{aligned}\ ] ] analysis of these equations reveals that @xmath299 is negative for @xmath318 where @xmath319 and @xmath320 hence the sf bilayer undergoes first - order transition at @xmath321 and @xmath322 . this scenario is quite similar than the one for @xmath67 , but with a smaller order parameter jump at the transition . moreover there are two first - order transitions , respectively at @xmath323 and @xmath324 instead of one at @xmath325 . in order to generalize the above gap calculations to finite temperatures and determine the first - order s / n transition line , we have solved numerically together the self - consistency equation ( [ selftnonnul ] ) and the condition @xmath326 . the result is given in the inset of fig.7 . collecting results from sec.v.a and b. we obtain the full @xmath175 phase diagram for the field - induced @xmath0 superconductivity . note that this @xmath0 superconductivity reproduces the structure of the phase diagram in quasi-2d superconductors @xcite although it is shifted to higher fields and lower temperatures . in this paper we have studied a periodic array of sf / sf bilayers in the limit of small coupling between the different bilayers . the corresponding gorkov equations have been solved exactly , taking into account both in - plane fflo modulation and arbitrary superconducting phase difference between sf layers . the superconducting state with zero phase difference is always settled in the low field regime , @xmath327 for parallel ( p ) orientation of the magnetizations . for antiparallel ( ap ) orientation , the @xmath0 state predominates at low temperatures over the @xmath107 state which is settled in the neighborhood of the critical line . consequently if the system is pinned in the antiparallel orientation , we predict a transition from the usual @xmath82 superconducting state to the @xmath0 state by cooling . while the critical temperature is higher for the ap orientation , the zero temperature order parameter is larger for the p orientation . this results in a crossing temperature @xmath152 below which the p orientation is more suitable for superconductivity . this temperature has been calculated as a function of the exchange field . in an experiment where the magnetizations might be easily reversed , one therefore expects a transition from the ap to the p orientation by cooling the system below this crossing temperature . in the low interlayer coupling limit , a ginzburg - landau functional has been derived from the exact expression of the anomalous gorkov green function . as a main result , we have obtained a @xmath0 superconducting state located in the vicinity of the tricritical point @xmath328 . details of the bilayer phase diagram are obtained in this framework , including the first - order transition lines between superconducting phases . since increasing the interlayer coupling enlarges the @xmath0 phase region , experimental observation of such details of the phase diagram requires the use of sf layers with large enough interlayer coupling , namely @xmath329 . finally the case of even stronger interlayer coupling , namely @xmath330 has been also investigated . it appears that at low temperatures the @xmath0 superconducting state is settled for exchange fields of the order of @xmath25 , which are well above the chandrasekhar - clogston paramagnetic limit . thus this new paramagnetic limit may be tuned by varying the interlayer coupling . in the present article we have reported the detailed structure of the phase diagram in this regime of high magnetic field . the first - order @xmath292 transition line is also derived . we expect that our results may be applicable to compounds like bi@xmath11sr@xmath11cacu@xmath11o@xmath12 under a magnetic field . indeed such perovskite superconductors comprise tightly coupled superconducting cuo planes separated by bio layers . however observing the field - induced superconductivity in a reasonable range of magnetic field requires relatively low critical temperatures which are realized in the heavily doped or underdoped regimes . finally the latter effect is solely related to the compensation of the energy shift in the two layers systems by the zeeman splitting . so it should be quite general and might appear also in two band superconductors or in weakly coupled superconducting grains . note that the inhomogeneous superconductivity has been obtained in the absence of magnetic field in two - bands superconductors.@xcite however since the @xmath0 state was not considered in this latter work no field - induced superconductivity had been noticed . bulaevskii @xcite studied thoroughly josephson coupling in periodic layered structures with one sf plane as unit cell . here we have demonstrated that systems with several sf planes as unit cell exhibit qualitatively new phenomena like field - induced superconductivity . the simplest case , two planes per unit cell , has been studied here . it may be regarded as a basic approach to understand the properties of more complex ferromagnetic superconducting compounds or artificial heterojunctions . we thank m. daumens , m. faure , m. houzet and m. kulic for useful discussions and comments . this work was supported , in part , by esf pi - shift program . we define the function @xmath332 by @xmath333 , \ ] ] and for any integrer @xmath334 the function @xmath331 is given by@xmath335 the variations of @xmath336 , @xmath337 , @xmath338 , with @xmath212 are represented in fig.10 . one can notice that in the vicinity of the tricritical point , _ i.e. _ @xmath339 , the functions @xmath332 and @xmath340 are negative and of the order of unity . @xmath341 cancels exactly at @xmath342 and becomes negative in the domain @xmath343 , which is studied sec.iv . this part of the appendix refers to sec . iv of the paper . in the ginzburg - landau theory , the free energy is expanded in terms of the gap @xmath122 , _ i.e. _ the order parameter , assuming the temperature close to @xmath10 . originaly it was introduced as a phenomenological theory for superconductivity before the bcs theory . here we derive the ginzburg - landau free energy from the full microscopic knowledge of our model in order to analyze the vicinity of the tricritical point . to do this , we consider the simplest case where the fflo gap modulation is exponential , namely @xmath344 , @xmath345 being the in - plane modulation wave vector . it is known that this modulation structure is not realized to the benefit of the cosine modulation discussed in the article s body . however , in sec.ii , the gorkov green functions of the sf / sf bilayer were derived for a modulated order parameter @xmath344 and @xmath82 or @xmath346 this modulation structure is then convenient to calculate the coefficients of the generalized gl functional because the exact expression of the anomalous green function ( see eq.([fcroix ] ) ) is valid for this gap modulation structure , whereas it is unknown with the cosine structure . we first expand the exact anomalous green function ( [ fcroix ] ) and the self - consistency relation in powers of the gap @xmath122 and the fflo wave vector @xmath347 then this self - consistency relation is interpreted as the stationarity condition for the ginzburg - landau free energy , which allows ( by identification ) to determine the coefficient of every term of the gl functional . in the @xmath82 case , the expansion of the anomalous green function reads @xmath348 where @xmath349 . we first consider the case of uniform superconductivity , _ i.e. _ @xmath350 . after integration over @xmath351 , we obtain:@xmath352 with @xmath353 . note that the interlayer coupling @xmath25 has disappeared in eq.([gna ] ) . we are now able to write down the self - consistency equation ( [ selftnonnul ] ) as an expansion in powers of @xmath122 @xmath354 where the functions @xmath355 are those defined in appendix a , and @xmath180 . this self - consistency relation may be interpreted as the stationnary condition @xmath356 for the ginzburg - landau free energy with uniform order parameter within each superconducting plane and @xmath82 phase difference between the planes . close to the tricritical point , @xmath357 is small and it is enough to retain only the first term in this infinite expansion as @xmath358 where @xmath179 . by identification with eq.([minim ] ) we obtain the gl free energy for the @xmath176 phase as a function of the variational parameter @xmath357 and the thermodynamical variable @xmath359 : @xmath360 \left\vert \tilde{\delta}\right\vert ^{2}+b_{1}% k_{3}(\tilde{h})\frac{\left\vert \tilde{\delta}\right\vert ^{4}}{2}-b_{2}% k_{5}(\tilde{h})\frac{\left\vert \tilde{\delta}\right\vert ^{6}}{3}\nonumber\\ & = \left [ \ln\frac{t}{t_{c0}}-k_{1}(\tilde{h})\right ] \left\vert \tilde{\delta}\right\vert ^{2}+\frac{k_{3}(\tilde{h})}{4}\left\vert \tilde{\delta}\right\vert ^{4}-\frac{k_{5}(\tilde{h})}{8}\left\vert \tilde{\delta}\right\vert ^{6 } \label{energyzero}%\end{aligned}\ ] ] which corresponds to eq.([freezero ] ) . note that in this usual @xmath107 state the same coefficients have been already reported in ref.@xcite the same procedure may be followed when the phase difference is @xmath0 . the anomalous green function is then @xmath361 and leads to the self - consistency relation which contains explicitely the coupling @xmath25 , via the normalized coupling @xmath193:@xmath362 \left\vert \tilde{\delta } \right\vert ^{2n}=0\text{.}%\ ] ] from the latter expression we deduce that the coefficients of the gl free energy for the @xmath13 state can be directly obtained using the coefficient of @xmath363 in which we replace @xmath364 by @xmath365 . finally the free energy of the @xmath192 state is @xmath366 \left\vert \tilde{\delta}\right\vert ^{2}\nonumber\\ & + \frac{k_{3}(\tilde{h}+\tilde{t})+k_{3}(\tilde{h}-\tilde{t})}{8}\left\vert \tilde{\delta}\right\vert ^{4}\nonumber\\ & -\frac{k_{5}(\tilde{h}+\tilde{t})+k_{5}(\tilde{h}-\tilde{t})}{16}\left\vert \tilde{\delta}\right\vert ^{6 } \label{energypi}%\end{aligned}\ ] ] which yields eq.([freepi ] ) in the small interlayer coupling limit @xmath367 . we have developped in a similar way the gl free energy in the case where the order parameter is modulated within each superconducting plane . using the expressions ( [ fzeroannexe ] ) and ( [ fpiannexe ] ) for the anomalous green function of the bilayer with @xmath344 fflo modulation respectively in the @xmath82 and @xmath13 cases , one obtains the expansion of the self - consistency equation in powers of @xmath122 and of the fflo wave vector @xmath46 . finally , after averaging over all possible orientations of the fflo modulation vector , the self - consistency equation reads:@xmath368 for @xmath82 . the coefficients @xmath369 are symmetric with respect to the expansion indexes @xmath370 and @xmath371@xmath372 and related to the coefficients @xmath373 by @xmath374 . from eq.([yoyo ] ) the gl free energy can be constructed using the method described in the previous paragraph for uniform phases . we retrieve all the coefficients already obtained by buzdin and kachkachi,@xcite including the coefficients of the gradient terms of the generalized functional . to derive the free energy of the @xmath198 phase , we have therefore used the bk functional with the cosine modulation which is effectively realized in each superconducting layer . as a result , it reads @xmath375 where @xmath203 . in the @xmath13 state the free energy has been derived from the bk functional in which the replacement @xmath376 has been done in order to obtain the modified coefficients . finally the free energy of the @xmath199 phase can be written as@xmath377 in the article body more convenient forms of eqs.([energyfflozeroannexe],[energyfflopiannexe ] ) involving the reduced quantities @xmath233 , @xmath230 and @xmath234 are used in order to derive the universal phase diagram . 99 v. l. ginzburg , zh . eksp . . fiz . * 31 * , 202 ( 1956 ) [ sov . jetp * 4 * , 153 ( 1957 ) ] . d. saint - james , d. sarma and e. j. thomas , _ type ii superconductivity _ ( pergamon press , new york , 1969 ) . a. a. abrikosov , _ fundamentals _ _ of the theory of metals _ ( north holland , netherlands , 1988 ) . b. s. chandrasekhar , appl . . lett . * 1 * , 7 ( 1962 ) . a. m. clogston , phys . * 9 * , 266 ( 1962 ) . a. i. larkin and y. n. ovchinnikov , zh . . fiz . * 47 * , 1136 ( 1964 ) [ sov . jetp * 20 * , 762 ( 1965 ) ] . p. fulde and r. a. ferrell , phys . rev . * 135 * , a550 ( 1964 ) . bulaevskii , zh . . fiz . * 64 * , 2241 ( 1973 ) [ sov . jetp * 37 * , 1133 ( 1973 ) ] . buzdin and v.v . tugushev , zh . . fiz . * 85 * , 735 ( 1983 ) [ sov . jetp * 58 * , 428 ( 1983 ) ] . aslamazov , zh . * 55 * , 1477 ( 1968 ) [ sov . jetp * 28 * , 773 ( 1969 ) ] . buzdin and s. v. polonskii , zh . . fiz . * 93 * , 747 ( 1987 ) [ sov . jetp * 66 * , 422 ( 1987 ) ] . et al . _ , nature * 410 * , 908 ( 2001 ) . a. m. tanatar , t. ishiguro , h. tanaka and h. kobayashi , phys . b * 66 * , 134503 ( 2002 ) . c. martin _ et al . _ , b * 71 * , 020503(r ) ( 2005 ) l.n . bulaevskii and m.v . zyskin , phys . b * 42 * , 10230 ( 1990 ) . a. i. buzdin , rev . phys . * 77 * , 935 ( 2005 ) . andreev , a. i. buzdin , and r. m. osgood , iii phys . b * 43 * , 10124 ( 1991 ) . a. c. mclaughlin , w. zhou , j. p. attfield , a. n. fitch , and j. l. tallon , phys . b * 60 * , 7512 ( 1999 ) . t. nachtrab , d. koelle , r. kleiner , c. bernhard and c.t . lin , phys . 92 , 117001 ( 2004 ) a. i. buzdin , a. v. vedyayev and n. v. ryzhanova , europhys . lett . * 48 * , 686 ( 1999 ) . i. baladie and a. buzdin , phys . b * 67 * , 014523 ( 2003 ) . a. buzdin and m. daumens , europhys . lett . * 64 * , 510 - 516 ( 2003 ) . s. tollis , m. daumens and a. buzdin , phys . b * 71 * , 024510 ( 2005 ) . v. apinyan and r. melin , eur . j. b * 25 * , 373 - 389 ( 2002 ) . h. jirari , r. melin , and n. stefanakis , eur . j. b * 31 * , 125 - 139 ( 2003 ) . i. bozovic , ieee . supercond . * 11 * , 2886 ( 2001 ) . a. buzdin , s. tollis , j. cayssol , cond - mat/0505726 ( 2005 ) . to be published in prl . m. l. kulic and u. hofmann , solid state communications * 77 * , 9 , 717 ( 1991 ) . abrikosov , l. p. gorkov , and i. e. dzyaloshinski , _ methods of quantum field theory in statistical physics _ ( dover , new york ) ( 1963 ) . m. houzet and a. buzdin , phys . b * 63 * , 184521 ( 2001 ) . f. s. bergeret , a. f. volkov , k. b. efetov , phys . lett * 86 * , 3140 ( 2001 ) . l. xiaowei _ et al . _ , phys . rev . b * 65 * , 134507 ( 2002 ) . v. n. krivoruchko and e. a. koshina , phys . rev . b * 64 * , 172511 ( 2001 ) . a. i. buzdin , h. kachkachi , phys . lett . a * 225 * , 341 ( 1997 ) . l. i. glazman and a. e. koshelev , zh . * 97 * , 1371 ( 1990 ) [ sov . jetp * 70 * , 774 ( 1990 ) ] . fig.1 : sf / sf bilayer . the interlayer coupling constant is denoted @xmath25 . the exchange fields @xmath28 can be either equal ( parallel orientation ) or opposite ( antiparallel orientation ) . the superconducting phase difference between @xmath85 and @xmath86 can be either @xmath107 ( @xmath378 ) or @xmath0 ( @xmath379 ) . fig.2 : schematic representation of the superconducting gap as a function of temperature for p orientation and @xmath82 ( thicker solid line ) , ap orientation and @xmath82 ( intermediate thickness line ) , and ap orientation and @xmath13 ( thiner line ) . all curves are given for the same value of the exchange field that is smaller @xmath10 ( in energy units ) . the inversion of the proximity effect occurs at the temperature @xmath152 and the transition from @xmath107 state to @xmath0 state in the ap orientation at the temperature @xmath380 . temperatures @xmath152 and @xmath151 as a function of the field are shown in the inset . fig.4 : universal phase diagram for weakly coupled bilayers , in @xmath255 coordinates . the critical line ( solid line ) corresponds either to a @xmath292 or to a @xmath382 transition depending on @xmath234 . the transition between the nonuniform @xmath198 superconducting phase and uniform @xmath176 ( resp.@xmath383 ) phase is represented with dash - dotted ( resp . dashed ) line . fig.5 : phase diagram in @xmath175 coordinates , for @xmath384 . only the neighborhood of the tricritical point is represented . the lines have the same meaning than in fig.4 . note that the @xmath0 phase is settled in a very narrow region of the phase diagram . fig.6 : excitation spectrum . usual singlet pairing ( thin line circles ) between opposite - spin electrons occupying the same orbital is affected by zeeman effect . in contrast , @xmath0 coupling ( thick line ) between two electrons occupying a bonding and an antibonding orbitals may lead to the cancellation of the zeeman splitting . fig.7 : phase diagram for @xmath385 . thick ( resp . thin ) solid lines represents second - order transition between @xmath386 ( resp . @xmath387 ) and normal metal phase ( @xmath80 ) for @xmath82 and @xmath0 . we expect the @xmath388 transition lines ( not calculated ) to be in the vicinity of the ( virtual ) first order @xmath389 lines ( dash - dotted ) .

we propose in this paper to study the energy- , thermal- and performance - aware resource management in heterogeneous datacenters . witnessing the continuous development of heterogeneity in datacenters , we are confronted with their different behaviors in terms of performance , power consumption and thermal dissipation : indeed , heterogeneity at server level lies both in the computing infrastructure ( computing power , electrical power consumption ) and in the heat removal systems ( different enclosure , fans , thermal sinks ) . also the physical locations of the servers become important with heterogeneity since some servers can ( over)heat others . while many studies address independently these parameters ( most of the time performance and power or energy ) , we show in this paper the necessity to tackle all these aspects for an optimal resource management of the computing resources . this leads to improved energy usage in a heterogeneous datacenter including the cooling of the computer rooms . we build our approach on the concept of heat distribution matrix to handle the mutual influence of the servers , in heterogeneous environments , which is novel in this context . we propose a heuristic to solve the server placement problem and we design a generic greedy framework for the online scheduling problem . we derive several single - objective heuristics ( for performance , energy , cooling ) and a novel fuzzy - based priority mechanism to handle their tradeoffs . finally , we show results using extensive simulations fed with actual measurements on heterogeneous servers . datacenter heterogeneity ; online scheduling ; server placement ; cooling ; multi - objective optimization .

1410.3104

the last years have witnessed the development of heterogeneity in clusters and datacenters . two main reasons have led to this situation today . the first one is due to the maintenance and evolution of the components in the datacenters : different generations of computers are commonly seen in production datacenters since the owners are not changing everything at each update . the second reason is driven by the idea that heterogeneity might be the key to achieving energy - proportional computing @xcite , especially for high - performance computing applications . many recent studies alert dramatically on the energy consumption of the datacenters . for instance , koomey s report @xcite claims that today s datacenters are consuming nearly 2% of the global energy , and up to half of that is spent on cooling - related activities @xcite . this results generally in very poor power usage effectiveness ( pue ) . in this paper , we study the multi - objective resource management problem for heterogeneous datacenters . besides the performance criterion , we also consider the energy consumption of the servers and their thermal impact on the datacenter cooling . the aim of our work is to optimize these objectives and to explore their tradeoffs . in particular , the energy consumption is partly due to the cooling efficiency in the datacenter @xcite , which is related to both the physical placement of the servers and the scheduling strategies when jobs dynamically enter and leave the system . the latter also affects the performance and the energy consumed by the servers . server placement in a computer room has been relatively less studied , especially its impact on the cooling efficiency . the reason for this lack of attention is mainly due to the fact that , when servers are homogeneous , their relative positions have no impact on the performance and computing energy . however , server placement can have an impact on the cooling infrastructure . the main observation is that one server might contribute to the temperature raise at the inlets of the other servers , due to the recirculation of heat in a datacenter . such mutual influence can be modeled by a heat distribution matrix among the servers . if one wants to keep the inlet temperature under a given threshold , the supplied air temperature has to be decreased accordingly by the cooling system , which in turn increases its energy consumption . in the presence of heterogeneous servers with different power consumptions and hence heat dissipation , the problem of find the optimal placement becomes complicated and , to the best of our knowledge , has not been studied . since it is not feasible to change dynamically the positions of the servers in a datacenter , we focus on static placement to minimize the cooling cost induced by different configurations . with a given server placement , the traditional problem of job scheduling in the heterogeneous environment remains . many previous work ( e.g. , @xcite ) considered only the performance criterion and hence focused on the jobs execution times . in order to address the power consumption issue in datacenters , however , application scheduling must employ a multi - objective approach by considering performance , energy and cooling together . to account for the fact that a scheduler has no future knowledge ( jobs arrive over time ) , we need an online scheduling strategy . instead of designing different independent algorithms , we design a greedy online scheduling framework that can be adapted easily by redefining the cost function , from a single objective to two or more objectives . to tackle the energy - performance tradeoff , we further introduce a fuzzy - based priority approach , which allows to explore the potential improvement in one objective while relaxing the other objective up to an acceptable range . this approach can be extended to incorporate more than two objectives in the framework . its principle is not limited to the case at hand and can potentially be applied to other multi - objective optimization problems . the main contributions of this paper are the following : * a static server placement heuristic to reduce the cooling cost for the servers in a datacenter . * a greedy scheduling framework and several cost functions to tackle single - objective scheduling ( for performance , energy , and cooling ) . * a fuzzy - based priority approach to handle the tradeoff between two conflicting objectives , and its extension to multi - objective optimization . these proposals are supported by extensive simulations conducted using real hardware specifications and software benchmarks , as well as experimentally verified cooling model and heat distribution matrix @xcite . specifically for the hardware , a server system with high packing density and integrated cooling support is chosen for the experiments , which we believe represents well an emerging class of highly integrated energy - efficient servers . the results demonstrate the flexibility of our scheduling framework and confirm the effectiveness of the fuzzy - based approach for exploring the energy - performance tradeoff in heterogeneous datacenter environments . our static server placement heuristic is also shown to provide much improved thermal distribution leading to significant reduction in cooling cost . the rest of this paper is organized as follows . section [ sec : statement ] formally states the system model and the scheduling problem . section [ sec : gmp ] describes our greedy server placement heuristic . section [ sec : scheduling ] presents the job scheduling framework , various cost functions and the fuzzy - based priority approach . the simulation results are shown in section [ sec : evaluation ] . section [ sec : related ] reviews some related work , and section [ sec : conclusion ] summarizes the paper and addresses future directions . motivated by the placement of physical servers and the scheduling of high - performance computing ( hpc ) applications in heterogeneous datacenters , we consider the following system model : a set @xmath0 of @xmath1 servers ( or machines ) needs to be placed inside a computer room ( or datacenter ) with a set of @xmath1 rack slots , denoted by @xmath2 . each server @xmath3 consists of @xmath4 processors of the same type ( possibly on different boards ) , but the type and the number of processors may vary for different servers , rendering the system heterogeneous . each server consumes a _ base power _ @xmath5 to support the basic operations of the infrastructure backbone , such as monitoring , networking and cooling ( for instance fans ) . a set @xmath6 of @xmath7 jobs arrive at the system over time , and they need to be assigned in an online manner to the servers . each job @xmath8 has a release time @xmath9 and a processor requirement @xmath10 that must be granted in order to run on any server . to execute job @xmath11 on server @xmath12 incurs a processing time @xmath13 and a power consumption @xmath14 , both of which are server - dependent and become known upon the job s arrival by prior profiling of the applications . in particular , the profiled application power consumption is assumed to include the leakage power . we study two orthogonal problems that deal with the placements of hardware and software , respectively . we call the two problems _ static server placement _ and _ online job scheduling_. the former concerns the positioning of physical servers in the datacenter , which as explained in section [ sec : gmp ] will have an impact on the cooling energy in heterogeneous environment . the latter concerns the dynamic assignment of workloads to the servers , which will impact energy ( due to both computing and cooling ) as well as performance . for the first problem of static server placement , each server needs to be physically and statically placed in advance to one of the available rack slots in the datacenter . in particular , we are looking for a mapping @xmath15 from rack slots to servers so that each slot @xmath16 is filled with a server @xmath17 . the objective is to minimize the cooling cost . more details about this problem will be described in section [ sec : gmp ] . given a particular server placement , an online scheduling strategy is then required to assign the jobs to the servers for execution . specifically , each arrived job @xmath18 must be assigned irrevocably to a server with at least @xmath10 idle processors , and without any knowledge of the future arriving jobs . once the job has been assigned , no preemption or migration is allowed , which is typically assumed for hpc applications since they tend to incur a significant cost in terms of data reallocation . at any time @xmath19 , the total computing power of server @xmath12 is the sum of its base power and the power consumed for executing all jobs assigned to it , i.e. , @xmath20 where @xmath21 is a binary variable that takes value @xmath22 if job @xmath11 is running on server @xmath12 at time @xmath19 and @xmath23 otherwise . in order to optimize performance , we do not allow processor sharing among the jobs . thus , each server at any time can only host a subset of the jobs whose total processor requirements are no more than the server s total number of available processors , i.e. , @xmath24 for all @xmath25 at all time @xmath19 . to characterize the cost of cooling , we consider a standard datacenter layout , where server racks are organized in rows with alternating cold and hot aisles . the computer room air conditioning ( crac ) unit supplies cool air to the cold aisles through raised floor vents . each server @xmath26 in the racks is oriented such that it draws cool air with temperature @xmath27 from the inlet and dissipates hot air with temperature @xmath28 to the outlet . assuming that the computing power consumed by a server is completely transformed into heat , the relationship between the power consumption and the inlet / outlet temperature of server @xmath12 at any time @xmath19 can be characterized by @xcite : @xmath29 where @xmath30 , with @xmath31 denoting the air density ( in @xmath32 ) , @xmath33 the airflow rate of server @xmath12 ( in @xmath34 ) , and @xmath35 the air heat capacity ( in @xmath36 ) . due to complex airflow patterns , typical datacenters experience the so - called _ heat recirculation _ phenomenon , where the hot air from the server outlets recirculates in the room and is mixed with the supplied cool air from the crac , causing the temperature at the server inlets to be higher than that of the supplied air . prior studies @xcite have characterized this phenomenon with a _ heat distribution matrix _ @xmath37 by assuming a fixed airflow pattern in the room and conservation of energy as described by equation ( [ eq : energyconserve ] ) . we adopt this approach here . let each element @xmath38 represent the temperature increase at the inlet of server @xmath12 per unit of power consumed by server @xmath39 . represents the temperature increase for the server at slot @xmath40 due to the power consumption by the server at slot @xmath16 . for convenience , we simply assume that the servers are renamed such that server @xmath12 is placed in slot @xmath40 for all @xmath41 . ] combining the heat contributions from all servers , the inlet temperature of server @xmath12 at time @xmath19 is given by the following equation : @xmath42 where @xmath43 denotes the supplied air temperature at time @xmath19 , which should be adjusted to prevent the inlet temperature of any server from going beyond a _ redline temperature _ @xmath44 ; otherwise , the electronic components may not work reliably or are at risk of being damaged . hence , the supplied air temperature should be set at most to @xmath45 the cooling cost is specified as @xmath46 where cop is the _ coefficient of performance _ , defined as the ratio of the amount of heat to be removed to the energy that needs to be consumed in order to perform the cooling @xcite . this coefficient characterizes the efficiency of the crac unit , and is an increasing ( usually non - linear ) function of the supplied air temperature . intuitively , it means that the crac unit needs to work harder and thus consumes more energy in order to provide cooler air to the computer room . we consider the following _ bi - objective optimization problem _ : optimizing the performance of the jobs and minimizing the energy consumption of the datacenter , due to both computing and cooling . for performance , we use the average response time of the jobs as the metric , and it is defined as @xmath47 where @xmath48 and @xmath9 denote the completion time and release time of job @xmath11 , respectively . the energy consumption comes from two sources : computing and cooling . the one due to computing is given by the total computing power of all servers integrated over time , i.e. , @xmath49 where @xmath50 $ ] denotes the interval of interest , during which all jobs arrive and complete their executions . this computing energy can be further divided into two parts , namely , the _ static _ part due to the base power consumption , i.e. , @xmath51 and the _ dynamic _ part due to the power consumed for executing the jobs , i.e. , @xmath52 where @xmath53 if job @xmath11 is assigned to server @xmath12 and 0 otherwise . the energy spent on cooling is the total cooling power integrated over time , i.e. , @xmath54 and as with computing energy , cooling energy can also be broken into a static part and a dynamic part . specifically , the static part is the cooling energy that will be spent during interval @xmath55 $ ] even if no job arrives , i.e. , @xmath56 and the dynamic part is the difference between the total cooling energy and the static one , i.e. , @xmath57 in this paper , we assume that all servers are turned on all the time to sustain the servers infrastructure backbone , so the static energy due to both computing and cooling is independent of the workload and the job scheduling strategy . on the other hand , the total dynamic energy given by @xmath58 is closely related to job scheduling , and it will be the focus of this study . due to the heterogeneity of the servers in the datacenter , different job scheduling strategies may result in very different job response time , computing energy and cooling cost . while a specific scheduling strategy may optimize one objective , these different objectives can be conflicting with each other , making the optimization difficult . in section [ sec : scheduling ] , we will propose and evaluate online scheduling algorithms to address both performance and energy as well as to deal with their tradeoffs . in this section , we consider the problem of static server placement . we first motivate the study from the perspective of cooling in heterogeneous datacenters . we then formulate the problem and present a greedy heuristic . the literature contains extensive studies on virtual machine placement ( e.g. , @xcite ) for datacenters , but the placement of physical servers has received little attention . there are two main reasons . first , many traditional datacenters are homogeneous , so different placements of identical servers do not make a difference . second , traditional metrics such as job performance and energy consumption ( due to computing ) are independent of the servers relative positions , so they are unaffected by the different placement configurations . as far as the cooling cost is concerned for heterogeneous datacenters , however , the placement of the physical servers will have an impact . in particular , the studies in @xcite have shown that the heat recirculation phenomenon in typical datacenters exhibits the following properties : ( 1 ) . different rack positions tend to behave differently in terms of heat recirculation . typically , servers located at the upper parts of the racks inhale " more recirculated hot air while servers located at the lower parts contribute " more hot air to recirculate in the room . ( 2 ) . in a closed computer room with fixed locations of all major objects and without moving objects , the airflow pattern that characterizes the heat recirculation among different rack positions is relatively stable . while the first property suggests that the heat distribution matrix tends to be highly asymmetric , the second property assures that the matrix does not change significantly with different workloads in the servers or different positions of the servers . in the next section , we will rely on workload placement ( or job scheduling ) techniques to manage the cooling cost together with other objectives . here , we focus on arranging the positions of the servers with different power profiles . the goal is to reduce the maximum inlet temperature of the servers so as to minimize the cooling cost under a given load condition . to illustrate the effectiveness of this approach , consider a simple datacenter with two servers , two rack slots , and the following heat distribution matrix : @xmath59 suppose the two servers consume an average power of @xmath60w and @xmath61w , respectively . by placing the first server in slot 1 and the second server in slot 2 , their inlet temperatures increase by @xmath62 and @xmath63 respectively according to equation ( [ eq : tin ] ) . by simply swapping the positions of the two servers , their temperature increases will now become @xmath64 and @xmath65 . the @xmath66 difference in the maximum inlet temperature of these two configurations directly determines the temperature of the supplied air by equation ( [ eq : supplytemp ] ) , and therefore impacts the cooling cost . for instance , consider a redline temperature of @xmath67 and the following cop model for a water - chilled crac unit in an hp datacenter @xcite : @xmath68 according to equations ( [ eq : supplytemp ] ) and ( [ eq : coolingcost ] ) , the cooling costs for the two placement configurations are @xmath69w and @xmath70w , respectively . the impact will be more significant with a lower redline temperature or a more skewed heat distribution matrix , or when the servers are consuming more power . the problem will also become more challenging when there is a large number of servers / positions , since exhaustive search will no longer be possible . the next subsection considers this general case and proposes a heuristic algorithm for the problem . to reduce the cooling cost , we should minimize the maximum temperature increase at the inlet of any server in the datacenter . as we have seen previously , this is determined by both the heat - distribution matrix and the power consumption profile of all servers . while the former is relatively stable and can be measured using a sensor - based approach @xcite , the latter essentially depends on the servers workloads , which can vary with time . to cope with this uncertainty , we characterize the power consumption of each server statically using the average power it consumes when executing historical workloads . this provides a reasonable estimation on the server s typical power consumption during runtime . we call this static value the _ reference power _ , and use it to determine the placement of the servers . let @xmath71 denote the reference power of server @xmath3 . the _ static server placement _ problem can then be formulated as follows : find a mapping @xmath72 from rack slots to servers , so as to @xmath73 where @xmath74^t$ ] . finding the optimal placement turns out to be a np - hard problem for arbitrary heat - distribution matrix and reference power vector . appendix provides the np - hardness proof . given the hardness result , we design a heuristic algorithm for the static server placement problem based on a greedy allocation strategy . algorithm [ alg : gsp ] presents the pseudocode of our greedy server placement ( gsp ) heuristic . first , gsp sorts the servers in descending order of reference powers ( line 1 ) . since the servers that consume more power on average will have larger contributions to the temperature increases at all inlets , they are placed first to have more flexibility in the slot selection and so to avoid high peak temperature . let @xmath75 denote the existing temperature increase at the inlet of slot @xmath76 , and it is initially set to zero for all inlets ( line 2 ) . let @xmath77 denote the maximum temperature increase if the next server @xmath3 is placed in slot @xmath16 , i.e. , @xmath78 server @xmath12 will be placed in one of the remaining slots @xmath79 that minimizes the maximum temperature increase , i.e. , @xmath80 . the temperature increase at all inlets will then be updated and the filled slot @xmath81 will be removed from the available set @xmath82 ( lines 12 and 13 ) . the algorithm iterates over all servers and terminates after the last one is placed . for the complexity of the algorithm , sorting and initialization takes @xmath83 time . in the iteration , placing each server incurs @xmath84 time as all remaining slots are examined to determine the maximum temperature increase at all inlets . therefore , the overall complexity is @xmath85 . this is reasonable even for a large number of servers , since the process is performed relatively infrequently : new placement of the servers is only necessary if there are significant alteration to the datacenter layout or when some servers are removed and new ones are introduced . once the servers have been placed in a datacenter , they will start operation by executing the applications or jobs . in practice , jobs are submitted by different users over time , so each job must be assigned to a server without knowing future job arrivals . this section considers online job scheduling under a given server placement to optimize performance and energy , and to deal with their tradeoffs . all online scheduling algorithms described in this section fall under a greedy scheduling framework ( gsf ) , which is evoked whenever a new job arrives or an existing job completes execution . algorithm [ alg : gsf ] presents the pseudocode of this framework . the variable @xmath86 shown in the pseudocode represents the cost of assigning job @xmath11 to server @xmath12 . specifically , @xmath86 can be a single - objective cost function of job response time , energy consumption , etc . ( see section [ sec : single ] ) , or it can be a composite cost function of two or more objectives ( see section [ sec : multi ] ) . for each newly arrived job @xmath11 , among the servers that have sufficiently available processors to host it , the server with the minimum cost in terms of @xmath86 will be chosen for assigning the job ( lines 2 - 9 ) . this makes the scheduling framework greedy . if no server has enough processors to host it , the job will be put in a waiting queue @xmath87 in shortest job first ( sjf ) order @xcite , which is known to optimize the average response time ( line 12 ) . note that although the processing times of the jobs are server - dependent , their relative sizes are assumed to be consistent on different servers , i.e. , a fast server is fast for all jobs . hence , sjf can be realized by using any server as the reference for comparing the jobs processing times . when a job completes execution on a server and therefore releases the occupied processors , the waiting jobs in the queue will be tested in sequence to see if they can be assigned to this server ( lines 16 - 18 ) . whenever a job is assigned or a running job completes execution , the number of available processors on the server will be updated ( lines 10 , 15 , 19 ) . under this greedy scheduling framework , the assignment of each job takes @xmath88 time , so the overall complexity is @xmath89 for assigning @xmath7 jobs . the next two subsections will describe heuristic algorithms that minimize different single- and multi - objective cost functions depending on the optimization criteria . single - objective scheduling considers one optimization criterion when deciding where to assign each job . in this subsection , we will present several single - objective scheduling heuristics . some of them will also be used as the base algorithms for designing the more complex multi - objective scheduling heuristics in the next subsection . first , the following describes some single - objective heuristics proposed in the literature @xcite . * _ uniform _ : assign each job randomly to a server according to the uniform distribution . * _ minhr _ : assign each job to a server that contributes minimally to the heat recirculation in the room . the cost function is defined as @xmath90 * _ coolestinlet _ : assign each job to a server with the lowest temperature at its inlet . the cost function is defined as @xmath91 where @xmath27 denotes the current temperature at the inlet of server @xmath12 . note that , in @xcite , these heuristics were applied in the offline setting , where the information of all jobs is available to the scheduler . here , they are cast as online heuristics . while the aim of _ uniform _ is to balance the workload on all servers , _ minhr _ and _ coolestinlet _ attempt to minimize the overall heat recirculation and to achieve a uniform temperature distribution , respectively . however , these heuristics were proposed for the homogeneous datacenter environments , and therefore do not consider job - specific characteristics . the following heuristics take job - dependent information into account by minimizing the performance , energy consumption , and temperature , respectively . * _ perf - aware _ : assign job @xmath11 to a server that renders the minimum response time . the cost function is defined as @xmath92 where @xmath93 denotes the execution time of job @xmath11 on server @xmath94 . * _ energy - aware _ : assign job @xmath11 to a server that incurs the minimum dynamic energy consumption due to both computing and cooling . the cost function is defined as @xmath95 where @xmath96 is the total dynamic energy defined in equation ( [ eq : totaldynamic ] ) , and it is evaluated based on the currently running jobs and with job @xmath11 assigned to server @xmath12 , i.e. , @xmath53 . * _ thermal - aware _ : assign job @xmath11 to a server that minimizes the maximum inlet temperature . the cost function is defined as @xmath97 where @xmath98 denotes the current temperature at the inlet of server @xmath39 , and @xmath14 denotes the power consumption of job @xmath11 on server @xmath12 . except for _ uniform _ , all heuristics above break the tie by randomly selecting a server with the best cost function . the difference between _ coolestinlet _ and _ thermal - aware _ is that the former considers the current inlet temperature before the job is assigned , whereas the latter considers the resulting temperature if the job is assigned to the server . note that all of these heuristics make greedy decisions locally for each arriving job , so they are not guaranteed to provide the optimal global cost . scheduling jobs to optimize two or more objectives usually require exploring the tradeoff between the conflicting goals . in this subsection , we propose a novel _ fuzzy - based priority _ approach to handle such a tradeoff . [ [ fuzzy - based - priority - for - bi - objective - scheduling ] ] fuzzy - based priority for bi - objective scheduling + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we first consider optimizing two objectives , for which we define the following composite cost function : @xmath99 in this case , the objectives @xmath100 and @xmath101 are considered one after another by first selecting all servers that offer the best performance in terms of @xmath100 , and then selecting among this subset any server that offers the best performance in terms of @xmath101 . to avoid depriving the second objective altogether , a _ fuzzy factor _ @xmath102 , where @xmath103 $ ] , is used to relax the selection criterion for the first objective up to a predefined margin ( in percentage ) . the purpose is to explore any potential improvement for @xmath101 while maintaining the performance for @xmath100 within a user - defined range of acceptance . the approach will be particularly effective if a small compromise in @xmath100 can lead to a large improvement in @xmath101 . setting @xmath104 indicates the high importance of @xmath100 that should not be compromised at all , while setting @xmath105 suggests that @xmath100 does not matter in the optimization . varying @xmath102 in between gives the user a flexible and intuitive way to specify the tradeoff between the two objectives . to implement the fuzzy - based priority approach in the online greedy scheduling framework ( gsf ) as shown in algorithm [ alg : gsf ] , the cost function for the first objective @xmath100 needs to be normalized between 0 and 1 in order to take the fuzzy factor into account , i.e. , @xmath106 where @xmath107 and @xmath108 denote the minimum and maximum costs in terms of objective @xmath100 among all available servers to assign job @xmath11 , and they can be easily collected by a linear scan of the available servers . the implementation then relies on the following rule for comparing the relative cost of assignment on any two servers . * _ fuzzy - based priority rule ( for two objectives ) _ * : the costs incurred by assigning job @xmath11 to any two servers @xmath109 and @xmath110 satisfy @xmath111 if and only if one of the following conditions holds : * @xmath112 , or * @xmath113 and @xmath114 and @xmath115 , or * @xmath116 and @xmath117 , or * @xmath118 , or * @xmath119 and @xmath115 . this rule can be applied to optimize any two objectives , as long as they have well - defined cost functions , such as the ones given in section [ sec : single ] . the value of the fuzzy factor as well as the priority depend on the relative importance of the two objectives to optimize , which can be determined by the user or the system administrator . [ [ extension - to - multi - objective - scheduling ] ] extension to multi - objective scheduling + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the fuzzy - based priority approach can be extended to include more than two objectives . as in the bi - objective case , we can optimize a sequence of objectives one after another , while using a ( possibly different ) fuzzy factor to specify the acceptable range for each objective . the following illustrates this method with a composite cost function consisting of @xmath120 objectives : @xmath121 in this case , the servers that are ranked among the top @xmath122 percent in terms of objective @xmath123 will be selected first . then , within this subset , the ones that fall into the top @xmath124 percent in terms of objective @xmath125 will be further selected . this process continues until the @xmath126-th objective is considered . finally , a server that survives the first @xmath127 rounds of selection and has the best performance in terms of the last objective @xmath128 will be chosen as the final winner . again , the order of the priorities and the values of the fuzzy factors should be determined by the relative importance of different objectives to optimize . [ [ comparison - with - other - approaches ] ] comparison with other approaches + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we now comment on the similarities and differences of the fuzzy - based priority approach in comparison with a few other multi - objective optimization approaches commonly found in the literature . figure [ fig : schema ] illustrates the basic principles of these approaches using bi - objective scheduling as an example . section [ sec : related ] describes some related work on the applications of these approaches in multi - objective scheduling . _ simple priority_. this is a special case of the fuzzy - based priority approach with fuzzy factor @xmath129 . it is usually applied in settings where strict priorities are imposed on different objectives . this approach provides better result for the first objective , but may lead to much worse performance for the second one . in contrast , the fuzzy - based priority approach is more effective in settings with soft ( or non - strict ) priorities , especially if an objective with slightly lower priority can be significantly improved with just a little compromise for a high - priority objective . _ pareto frontier_. this approach returns a set of _ nondominated _ solutions to the user instead of only one solution . it is widely applied in offline settings to quantify the tradeoffs among different objectives . in the context of online scheduling , however , multiple solutions are hard to maintain over time , and one of the intermediate solutions must be selected on - the - fly in order to decide where each job should be assigned . ( 3 ) . _ constraint optimization_. this approach optimizes one objective subject to certain constraints imposed on the other(s ) . it is commonly applied in environments with strict or clearly - defined requirements , e.g. , job deadline or energy budget . instead of using an absolute value as the constraint , the fuzzy - based priority approach specifies the constraint as a relative threshold , i.e. , fuzzy factor , in terms of percentage . weighted sum_. this approach transforms multiple objectives into a single one by optimizing a weighted combination . although priorities are not explicitly specified , it uses weights to indicate the relative importance of the objectives . as different objectives can have different units , they are often normalized in order to be combined . however , it may not be intuitive to set the values of the weights , e.g. , for time and energy . compared to simple priority and constraint optimization , fuzzy - based priority is particularly suitable for scheduling hpc applications in datacenters , where no strict constraints or priority are normally imposed on job performance or energy consumption . compared to weighted sum , fuzzy - based priority provides an intuitive alternative to describing the tradeoffs while specifying soft preference of the user on the priority of the objectives . setting an appropriate fuzzy factor encodes such preference in an online manner . as shown in figure [ fig : schema ] , the solution returned by fuzzy - based priority ( and other approaches ) when scheduling an individual job actually lies on the pareto frontier . in this section , we will evaluate the proposed online scheduling heuristics with the fuzzy - based priority approach and the greedy heuristic for server placement . the evaluations are performed by simulation using the data center workload and resource management simulator ( dcworms ) @xcite . [ [ datacenter - configuration ] ] datacenter configuration + + + + + + + + + + + + + + + + + + + + + + + + we simulate a datacenter with 50 servers and the same configuration as the one considered in @xcite , which essentially determines the heat distribution matrix . specifically , the datacenter consists of two rows of racks in a typical cold aisle and hot aisle layout . the cool air is supplied by the crac unit from the cold aisle between the two rows . each row has five racks and each rack contains five servers . the server platform used in the simulation is based on christmann s _ resource efficient cluster server ( recs ) _ unit @xcite , which is a multi - node computer system consisting of 18 processors . the datacenter consists of 900 processors in total . the recs platform is chosen because it represents an emerging class of high - density and energy - efficient servers with built - in power and temperature sensors and integrated cooling support . table [ tab : parameters ] shows the parameters used in the simulation , whose values are based on real measurements in a recs unit . from the first three parameters , the heat distribution matrix @xmath37 used in this paper is derived by adopting the same airflow pattern as the one measured in @xcite . the coefficient of performance ( cop ) is based on the one in an hp datacenter @xcite as shown by equation ( [ eq : hp_cop ] ) . .[tab : parameters ] values of the parameters used in the simulation . [ cols="^,^",options="header " , ] each job is randomly selected from one of these benchmarks and the number of processors it requires is randomly generated from 1 to 8 with uniform distribution . following the definition in @xcite , the system load @xmath130 is defined to be @xmath131}{\sum_{j=1}^{m}l_j } \ , \ ] ] where @xmath132 is the arrival rate ( in # jobs per hour ) , @xmath133 $ ] is the average sequential execution time of the jobs on all processor types ( roughly 4.5 hours ) and @xmath134 is the total number of processors , which is 900 in the simulation . jobs arrive according to the poisson process , and the arrival rate @xmath132 is increased from 20 to 200 with a fixed arrival duration of 8 hours . the total number of jobs ranges from 160 to 1600 , and the system load is between 0.1 and 1 . this section presents the simulation results . first , we evaluate the performance of various online scheduling heuristics with a fixed placement for the servers . we then study the impact of different placement configurations on the performance of the scheduling heuristics . all results are obtained by carrying out the experiments 10 times and taking the average . we first evaluate the online scheduling heuristics for a single objective . the results are used as references for exploring the energy - performance tradeoff in the next subsection . in both cases , the server placement is fixed with each type of processor occupying 10 contiguous server slots over two racks , according to the order specified in table [ tab : processors ] . six heuristics presented in section [ sec : single ] are evaluated , namely , _ uniform _ , _ minhr _ , _ coolestinlet _ , _ perf - aware _ , _ energy - aware _ and _ thermal - aware_. figure [ fig : single ] presents the results of these heuristics . as we can see in figure [ fig : single : meanrt ] , _ perf - aware _ has significantly better average job response time compared to the other heuristics , especially under light system loads . this is because all jobs in _ perf - aware _ are assigned to high - performance ( faster ) processors before slower ones whenever possible . for the same reason , _ perf - aware _ also has better makespan ( completion time of the last finished job ) and processor utilization ( ratio between the utilized processor cycles and all processor cycles during the simulation period ) , as shown in figures [ fig : single : makespan ] and [ fig : single : utilization ] . note that the processor utilizations remain under 70% even when the system load reaches 1 . this is partly due to the fragmented processors in some servers that can not be utilized because a ready job simply has higher processor requirement . figure [ fig : single : totalenergy ] compares the total ( dynamic ) energy consumption of the scheduling heuristics , and figures [ fig : single : computingenergy ] and [ fig : single : coolingenergy ] show the energy consumed for computing and cooling , separately . for all heuristics , the energy consumption increases with the system load or the total number of jobs in the arrival interval . _ energy - aware _ consumes less total energy compared to the other heuristics , since jobs are assigned to processors with better energy efficiency . the improvement is more significant in terms of computing energy . for the cooling part , _ minhr _ and _ thermal - aware _ consumes roughly the same energy as _ energy - aware _ , since they are designed to minimize the heat recirculation and the maximum inlet temperature , which in turn increases the supplied temperature in the room and hence directly impacts the cooling cost . figure [ fig : supply_temp ] shows the average supply temperature of the different scheduling heuristics in the simulation period . indeed , _ thermal - aware _ and _ minhr _ are better than _ energy - aware _ in terms of the average supply temperature by up to 1.3@xmath135 and 1.6@xmath135 , respectively . as the system load increases further and hence the processor utilization becomes higher , the performance of all heuristics tend to converge , since all servers are roughly equally loaded under all heuristics . in particular for _ energy - aware _ , some jobs are forced to be assigned to the high - performance servers since the energy - efficient ones are all occupied , resulting in improved average job response time . we now evaluate the effectiveness of the fuzzy - based priority approach for exploring the energy - performance tradeoff in online scheduling . to this end , we consider the composite cost function @xmath136 that optimizes the energy consumption followed by the job response time . figure [ fig : two1 ] shows the results of minimizing @xmath137 when the fuzzy factor @xmath102 is increased from @xmath23 to @xmath22 at three different system loads ( @xmath138 , @xmath139 and @xmath140 ) . the values of both objectives are plotted as a function of @xmath102 , with energy consumption shown on the left @xmath101 axis and average response time on the right . in addition , the figure also shows the results when @xmath141 and @xmath142 , denoting the cases where the scheduling decision is based solely on the first objective ( energy ) and the second objective ( response time ) . the two cases are equivalent to the single - objective heuristics _ energy - aware _ and _ perf - aware _ , respectively . as we can see , the average response time improves with increased fuzzy factor at the expense of the energy consumption under all system loads . however , the improvement can be significant even before major compromise in energy consumption is observed . for instance , at medium load ( @xmath143 ) , the response time is reduced by about 1000 when @xmath102 reaches 0.6 without much increase in the energy consumption . similar results can also be observed at light load and heavy load . the fuzzy - based priority approach can take advantage of such characteristics by setting suitable fuzzy factors in order to achieve desirable energy - performance tradeoff in the online setting . figure [ fig : two2 ] shows the energy - performance tradeoff curve for @xmath144 obtained by varying the fuzzy factor from 0 to 1 . the results of the six single - objective heuristics are also shown in the figure under the respective load . we can see that _ minhr _ and _ thermal - aware _ lie around the curve ( or even slightly to the left of the curve in the case of _ minhr _ ) , indicating that they achieve fairly efficient tradeoffs between job response time and energy consumption . on the other hand , _ uniform _ and _ coolestinlet _ are completely dominated by the curve , which suggests that they provide less attractive tradeoff results . figure [ fig : two3 ] plots the tradeoff curves achieved by optimizing the heat recirculation and the maximum inlet temperature followed by the job response time , i.e. , with cost functions @xmath145 and @xmath146 . the results under three different system loads are shown alongside the ones for @xmath137 . the curves indicate that the two heuristics are able to provide better tradeoffs in the medium to high energy range ( e.g. , between 150 and 220 for _ minhr _ at @xmath143 ) while the tradeoff remains efficient for the cost function @xmath137 when the energy consumption is close to the minimum . the results demonstrate the flexibility of the fuzzy - based priority approach in exploring the energy - performance tradeoff in online scheduling . the approach can be potentially applied to other multi - objective optimization problems . we now study the impact of server placement on the performance of the online scheduling heuristics . besides the simple location - based placement used in the previous evaluations , which we call loc , we generate three additional placements for the servers . one is based on our gsp heuristic and the other two are based on its variations . we call the three placement configurations gsp1 , gsp2 and gsp3 , respectively . the two variants ( gsp2 and gsp3 ) are obtained in a similar fashion as gsp1 . in particular , in gsp2 the servers are sorted in ascending order of reference power instead of descending order , and in gsp3 each server is assigned to a remaining rack slot that maximizes the maximum inlet temperature instead of minimizing it . apparently , these two heuristics are counter - intuitive and are expected to provide undesirable configurations . the purpose of including them is to demonstrate the impact of different server placements on a scheduling algorithm s performance , especially on the cooling cost . figure [ fig : serverplacement_example ] shows the inlet temperature distribution of the 50 servers under the four placement configurations . in all cases , each processor is loaded with the average power consumption of the benchmarks shown in table [ tab : benchmarks ] . as we can see , gsp1 has better thermal balance than the other configurations . specifically , it improves loc by about @xmath147 in terms of the maximum inlet temperature and improves gsp2 and gsp3 by over @xmath148 and @xmath149 , respectively . figures [ fig : sp_perf ] and [ fig : sp_energy ] show the performance of _ perf - aware _ and _ energy - aware _ under the four server placements at different system loads . in both heuristics , job response time and computing energy are not affected by different configurations . however , gsp1 has reduced cooling energy compared to the other configurations . this is particularly evident under heavy system load , where all servers are almost fully and equally loaded , thus their power consumption ratios match closely those of the average values used in the server placement heuristic . under light system load , however , the servers could experience unbalanced loads , which causes their power consumption ratios to deviate from those of the average values . as a result , the advantage of gsp1 becomes smaller or even diminishes , but since the overall energy consumption is small in this case , the impact of server placement is not significant . quite similar effect on the cooling energy can be observed for _ thermal - aware _ and _ minhr _ as shown in figures [ fig : sp_thermal ] and [ fig : sp_minhr ] . notice that , for these two heuristics , different server placements also lead to a tradeoff between job response time and computing energy . to further investigate the tradeoff efficiency , figure [ fig : tradeoff_mix ] shows the energy - performance tradeoff curves for three heuristics with cost functions @xmath137 , @xmath150 and @xmath151 at load @xmath152 under different server placements . we can see that , although the tradeoff remains , in all cases gsp1 provides the best cooling energy and hence improves the overall tradeoff efficiency . note that _ minhr _ and _ perf - aware _ behave exactly the same under gsp1 , since servers with faster processors and hence more power consumptions are placed in the slots with less heat recirculation . therefore , the same performance and energy are observed for @xmath150 regardless of the fuzzy factor , as shown in figure [ fig : tradeoff_mix : minhr ] . the results confirm that strategic server placement indeed improves the thermal balance in a heterogeneous datacenter , which helps reduce the cooling cost . this is achieved with little impact on the job response time and computing energy , or the tradeoff between them . in this section , we review some related work in the literature on multi - objective scheduling and thermal - aware scheduling for datacenters . [ [ multi - objective - scheduling ] ] multi - objective scheduling + + + + + + + + + + + + + + + + + + + + + + + + + + scheduling with multiple conflicting objectives has attracted much attention in many optimization problems . section [ sec : multi ] described a few commonly used approaches . the following reviews some applications of these approaches in various problem domains . _ simple priority . _ this is a simple priority - based approach to optimize multiple objectives in sequence . assayad et al . @xcite introduced a bi - criteria compromise function to set priorities between makespan and reliability for scheduling real - time applications . to minimize carbon emission and to maximize profit , two - step policies were proposed by garg et al . @xcite to map applications to heterogeneous datacenters based on the relative priority of the two objectives . du et al . @xcite proposed heuristics to optimize the qos for interactive services before considering energy consumption on multicore processors with dvfs ( dynamic voltage & frequency scaling ) capability . _ pareto frontier . _ this approach is often used in the offline setting to generate a set of nondominated solutions . durillo et al . @xcite applied this technique to tradeoff makespan and energy consumption for heterogeneous servers . torabi et al . @xcite used particle swarm optimization to approximate the pareto frontier for the unrelated machine scheduling problem with uncertainties in the inputs . gao et al . @xcite utilizes ant colony optimization to obtain the pareto frontier for resource wastage and power consumption in virtual machine placement . evolutionary algorithms were employed in @xcite to obtain a set of alternative solutions for scheduling scientific workloads in the grid environment . _ constraint optimization . _ this approach optimizes one objective subject to constraint(s ) on the other(s ) . rizvandi et al . @xcite applied it to minimize the energy consumption subject to the makespan achieved in an initial schedule . a mixed integer programming model was used by petrucci et al . @xcite to reduce the power consumption of virtualized servers subject to qos requirements . fard et al . @xcite developed a double strategy to minimize the euclidean distance between the generated solutions to a set of user - specified constraints in a four - objective optimization problem . the authors in @xcite applied @xmath153-constraint method to cloud scheduling , which optimizes each objective in turn with upper bounds specified for the others . weighted combination . _ this approach combines multiple objectives into a single one . lee and zomaya @xcite used dvfs to tradeoff makespan with energy consumption by considering a weighted sum of the two objectives . the same technique was used by the authors of @xcite in an online manner to minimize a combined objective of job response time and energy . a similar approach was taken by sheikh and ahmad @xcite , who considered an additional objective of peak temperature in a multicore system , and hence optimizing three objectives at the same time . instead of summation , some work ( e.g. , @xcite ) also used energy - delay product as a metric for scheduling applications in heterogeneous multicore systems . compared to these approaches , our fuzzy - based priority approach provides a rather flexible solution to handling two or more conflicting objectives . although multi - objective scheduling with fuzzy " or good enough " solutions @xcite are known in the pareto - based approach , our fuzzy method is novel when ( soft ) priorities exist between different objectives . the principle can be potentially applied to other multi - objective optimization problems . [ [ thermal - aware - scheduling ] ] thermal - aware scheduling + + + + + + + + + + + + + + + + + + + + + + + + as cooling energy constitutes a signifcant fraction of the total energy consumption in today s large - scale datacenter , thermal - aware scheduling for this environment has been the focus of many research in recent years . wang et al . @xcite considered thermal - aware workload placement in datacenters to reduce the server temperatures characterized by an rc - model , while minimizing the job response time . they proposed simple heuristics that allocate hot " jobs to cool " computing nodes , as well as backfilling techniques for scheduling parallel applications . in their study , the thermal map of the data center is assumed to be available through ambient and on - board sensors . moore et al . @xcite first introduced the concept of heat recirculation effect and proposed workload placement algorithms , including _ minhr _ , to reduce the recirculation of heat and the cooling cost in a datacenter . a prediction tool called weatherman @xcite was used to predict the data center thermal map using machine learning techniques . the authors showed that the tool accurately predicts the heat distribution of the datacenter without the need of static thermal configuration , and a scheduling algorithm based on weatherman achieves similar performance as _ minhr_. tang et al . @xcite also studied the problem of minimizing the cooling cost in datacenters with heat recirculation consideration . based on an abstract heat flow model , they characterized the thermal behavior of datacenters via a heat distribution matrix . the model was validated by computational fluid dynamics ( cfd ) simulations in @xcite . they proposed offline scheduling solutions by using genetic algorithms and quadratic programming , which were evaluated using the heat distribution matrix captured for a small - scale datacenter . the same matrix is used in this paper for evaluating our online scheduling heuristics . instead of minimizing only the cooling cost , pakbaznia and pedram @xcite considered minimizing the total energy of a datacenter from both computation and cooling . they showed that performing consolidation to turn off idle servers together with job scheduling to account for the heat recirculation can significantly reduce the total power usage . banerjee et al . @xcite further considered cooling - aware scheduling workload placement by exploring the dynamic cooling behavior of the crac unit in a datacenter . while the above results considered only the energy consumption of a datacenter , the following also takes application performance into consideration . mukherjee et al . @xcite considered a similar problem as in @xcite but further took the temporal dimension of the job placements into account . they formulated the problem as a non - linear program and proposed both offline and online heuristics to minimize the total energy subject to the deadline constraint for the jobs . sansottera and cremonesi @xcite considered a datacenter environment hosting web services , and presented heuristics to minimize the total energy subject to service response time constraints . kaplan et al . @xcite studied the dual optimization of cooling and communication costs for hpc applications in a datacenter . they proposed a heuristic algorithm that achieves a good tradeoff between the two objectives , and subject to reliabity constraint specified by the processor junction temperature . in contrast to the previous work , which focused on either offline scheduling or homogeneous datacenters , we studied the problem of online scheduling for heterogeneous datacenters with both energy and performance considerations , as well as their tradeoffs . furthermore , we considered static server placement to balance the thermal distribution in the presence of nonuniform heat distribution matrix . in our previous work @xcite , we have applied this concept to the arrangement of computing nodes in a smaller scale problem ( at the server level ) . to the best of our knowledge , no prior work has considered this problem for heterogeneous datacenters . in this paper , we have considered the energy - efficient and thermal - aware placements for both servers and workloads in heterogeneous datacenters . for the static server placement problem , we have shown that it is np - hard and presented a greedy heuristic . to schedule the workloads , we have presented a greedy scheduling framework , which can be applied in an online manner with any well - defined cost function . moreover , a novel fuzzy - based priority approach was proposed to simultaneously optimize two or more conflicting objectives . simulations were conducted for a heterogeneous datacenter with heat recirculation effect . the results demonstrated the effectiveness of the proposed approaches for exploring the energy - performance tradeoff with cooling consideration . our static server placement heuristic was also shown to provide better thermal balance , which directly leads to reductions in cooling cost . for future work , other resource management techniques , such as dvfs or server consolidation , can be applied to achieve better energy and thermal efficiency . in this context , the tradeoff between the computing energy and cooling energy can be explored , possibly with the fuzzy - based priority approach . for the static server placement problem , it will be useful to design better heuristic solutions or good approximation algorithms , and to consider large datacenters with more rack slots than servers , which will provide additional space for optimization . finally , we considered server placement and job scheduling separately in this paper ; it may be helpful to consider the two aspects jointly to achieve further energy savings . this research was funded by the european commission under contract 288701 through the project coolemall . andrew , m. lin , and a. wierman . optimality , fairness , and robustness in speed scaling designs .. in _ proceedings of acm sigmetrics _ , pp . 37 - 48 , 2010 . i. assayad , a. girault , and h. kalla . a bi - criteria scheduling heuristic for distributed embedded systems under reliability and real - time constraints . in _ proceedings of the ieee / ifip international conference on dependable systems and networks ( dsn ) , page 347 , 2004 . a. banerjee , t. mukherjee , g. varsamopoulos , and s. k. gupta . integrating cooling awareness with thermal aware workload placement for hpc data centers . , 1(2):134 - 150 , 2011 . j. barbosa , and b. moreira . dynamic job scheduling on heterogeneous clusters . in _ proceedings of the ieee international symposium on parallel and distributed computing ( ispdc ) _ , pp . 3 - 10 , 2009 . barroso , and u. holzle . the case for energy - proportional computing . , 40(12):33 - 37 , 2007 . d. borgetto , h. casanova , g. da costa , j .- m , pierson , energy - aware service allocation . , 28(5):769 - 779 , 2012 . j. cong , and b. yuan . energy - efficient scheduling on heterogeneous multi - core architectures . in _ proceedings of the international symposium on low power electronics and design ( islped ) _ , pp . 345 - 350 , 2012 . description for resource efficient computing system ( recs ) . , 2009 . g. da costa . heterogeneity : the key to achieve power - proportional computing . in _ proceedings of the ieee / acm international symposium on cluster , cloud and grid computing ( ccgrid ) _ , pp . 656 - 662 , 2013 . g. da costa , t. zilio , m. jarus , and a. oleksiak . energy - and heat - aware hpc benchmarks . in _ proceedings of the international conference on cloud and green computing ( cgc ) 435 - 442 , 2013 . downey . a parallel workload model and its implications for processor allocation . in _ proceedings of the acm symposium on high - performance parallel and distributed computing ( hpdc ) _ , pp . 112 - 123 , 1997 . z. du , h. sun , y. he , y. he , d.a . bader , and h. zhang . energy - efficient scheduling for best - effort interactive services to achieve high response quality . in _ proceedings of the ieee international parallel & distributed processing symposium ( ipdps ) _ , pp . 637 - 648 , 2013 . durillo , v. nae , and r. prodan . multi - objective workflow scheduling : an analysis of the energy efficiency and makespan tradeoff . in _ proceedings of the ieee / acm international symposium on cluster , cloud and grid computing ( ccgrid ) _ , pp . 203 - 210 , 2013 . fard , r. prodan , j. barrionuevo , and t. fahringer . a multi - objective approach for workflow scheduling in heterogeneous environments . in _ proceedings of the ieee / acm international symposium on cluster , cloud and grid computing ( ccgrid ) _ , pp . 300 - 309 , 2012 . y. gao , h. guan , z. qi , y. hou , and l. liu . a multi - objective ant colony system algorithm for virtual machine placement in cloud computing . , 79(8):1230 - 1242 , 2013 . garey and d.s . computers and intractability : a guide to the theory of np - completeness . , 1979 . r. garg and a.k . reference point based multi - objective optimization to workflow grid scheduling . , 3(1):80 - 99 , 2012 . garg , c. yeo , a. anandasivam , and r. buyya . environment - conscious scheduling of hpc applications on distributed cloud - oriented data centers . , 71(6):732 - 749 , 2011 . l. grandinetti , o. pisacane , and m. sheikhalishahi . an approximate @xmath153-constraint method for a multi - objective job scheduling in the cloud . , 29(8):1901 - 1908 , 2013 . f. kaplan , j. meng , and a.k . coskun . optimizing communication and cooling costs in hpc data centers via intelligent job allocation . in _ proceedings of the international green computing conference ( igcc ) _ , pp . 1 - 10 , 2013 . j. koomey . worldwide electricity used in data centers . , vol . 3 , 034008 , 2008 . k. kurowski , a. oleksiak , w. piatek , t. piontek , a. przybyszewski , and j. weglarz . dcworms a tool for simulation of energy efficiency in distributed computing infrastructures . , 39:135 - 151 , 2013 . y. lee and a.y . energy conscious scheduling for distributed computing systems under different operating conditions . , 22(8):1374 - 1381 , 2011 . j. moore , j. s. chase , and p. ranganathan . weatherman : automated , online and predictive thermal mapping and management for data centers . in _ proceedings of the international conference on autonomic computing ( icac ) 155 - 164 , 2006 . j. moore , j. chase , p. ranganathan , and r. sharma . making scheduling cool " : temperature - aware workload placement in data centers . in _ proceedings of the annual conference on usenix annual technical conference ( atec ) _ , pp . 5 - 5 , 2005 . t. mukherjee , a. banerjee , g. varsamopoulos , s. gupta , and s. rungta . spatio - temporal thermal - aware job scheduling to minimize energy consumption in virtualized heterogeneous data centers . , 53(17):2888 - 2904 , 2009 . e. pakbaznia and m. pedram . minimizing data center cooling and server power costs . in _ proceedings of the international symposium on low power electronics and design ( islped ) _ , pp . 145 - 150 , 2009 . passmark@xmath154 results . v. petrucci , e. v. carrera , o. loques , j. c. b. leite , and d. mosse . optimized management of power and performance for virtualized heterogeneous server clusters . in _ proceedings of the ieee / acm international symposium on cluster , cloud and grid computing ( ccgrid ) _ , pp . 23 - 32 , 2011 . v. petrucci , o. loques , and d. mosse . lucky scheduling for energy - efficient heterogeneous multi - core systems . in _ proceedings of the usenix conference on power - aware computing and systems ( hotpower ) _ , pp . 7 - 7 , 2012 . rizvandi , j. taheri , a.y . zomaya and y. lee . linear combinations of dvfs - enabled processor frequencies to modify the energy - aware scheduling algorithms . in _ proceedings of the ieee / acm international symposium on cluster , cloud and grid computing ( ccgrid ) _ , pp . 388 - 397 , 2010 . a. sansottera , and p. cremonesi . cooling - aware workload placement with performance constraints . , 68(11):1232 - 1246 , 2011 . r. sawyer . calculating total power requirements for data centers . , 2004 . sheikh and i. ahmad . simultaneous optimization of performance , energy and temperature for dag scheduling in multi - core processors . in _ proceedings of the international green computing conference ( igcc ) _ , pp . 1 - 6 , 2012 . a. silberschatz , p.b . galvin , and g. gagne . operating systems concepts . , 2005 . h. sun , y. cao , and w .- j . non - clairvoyant speed scaling for batched parallel jobs on multiprocessors . in _ proceedings of the acm conference on computing frontiers _ , pp . 99 - 108 , 2009 . h. sun , p. stolf , j .- pierson , and g. da costa . multi - objective scheduling for heterogeneous server systems with machine placement . in _ proceedings of the ieee / acm international symposium on cluster , cloud and grid computing ( ccgrid ) _ , 2014 . q. tang , s. gupta , and g. varsamopoulos . energy - efficient thermal - aware task scheduling for homogeneous high - performance computing data centers : a cyber - physical approach . , 19(11):14581472 , 2008 . q. tang , t. mukherjee , s. gupta , and p. cayton . sensor - based fast thermal evaluation model for energy efficient high - performance datacenters . in _ proceedings of the international conference on intelligent sensing and information processing _ , pp . 203208 , 2006 . h. topcuoglu , h , s. hariri , and m .- y . performance - effective and low - complexity task scheduling for heterogeneous computing . , 13(3):260 - 274 , 2002 . torabi , n. sahebjamnia , s.a . mansouri , m.a . a particle swarm optimization for a fuzzy multi - objective unrelated parallel machines scheduling problem . , 13(12):4750 - 4762 , 2013 . l. wang , g. von laszewski , j. dayal , x. he , a. j. younge , and t. r. furlani . towards thermal aware workload scheduling in a data center . in _ proceedings of the international symposium on pervasive systems , algorithms , and networks ( ispan ) _ , pp . 116 - 122 , 2009 . l. wang , s.u . khan , and j. dayal . thermal aware workload placement with task - temperature profiles in a data center . , 61(3):780 - 803 , 2012 . j. xu and j. a. b. fortes . multi - objective virtual machine placement in virtualized data center environments . in _ proceedings of the ieee / acm international conference on green computing and communications & international conference on cyber , physical and social computing ( greencom - cpscom ) _ , pp . 179 - 188 , 2010 . j. yu , m. kirley , and r. buyya . multi - objective planning for workflow execution on grids . in _ proceedings of the ieee / acm international conference on grid computing _ , pp . 10 - 17 , 2007 . f. zhang , j. cao , k. li , s.u . khan , and k. hwang . multi - objective scheduling of many tasks in cloud platforms . the static server placement problem is np - hard . we reduce the 3-partition problem to the static server placement problem . in 3-partition , a finite set @xmath155 of @xmath156 positive integers is given , and the sum of the integers is @xmath157 . the question is whether @xmath158 can be partitioned into @xmath159 disjoint subsets such that the sum of the numbers in each subset is equal to @xmath160 . the problem is known to be np - hard even if every integer in @xmath158 is strictly between @xmath161 and @xmath162 , so each subset must contain exactly three numbers @xcite . given an instance @xmath158 of the 3-partition problem , where each integer @xmath163 satisfies @xmath164 , we construct an instance of the static server placement problem as follows . let @xmath165 , and assign @xmath166 the heat - distribution matrix @xmath37 is specified by setting @xmath167 and setting all the other elements to zero . suppose @xmath168 is an optimal mapping for the server placement instance constructed above . the temperature increase at the inlet of slot @xmath16 , where server @xmath169 is placed , is given by @xmath170 the maximum temperature increase at any inlet is therefore @xmath171 this leads to the conclusion that the server placement instance has a maximum temperature increase of @xmath160 if and only if @xmath158 can be partitioned into @xmath159 disjoint subsets , where the sum of the numbers in each subset is also @xmath160 .

first - principles calculations combining density - functional theory and continuum solvation models enable realistic theoretical modeling and design of electrochemical systems . when a reaction proceeds in such systems , the number of electrons in the portion of the system treated quantum mechanically changes continuously , with a balancing charge appearing in the continuum electrolyte . a grand - canonical ensemble of electrons at a chemical potential set by the electrode potential is therefore the ideal description of such systems that directly mimics the experimental condition . we present two distinct algorithms , a self - consistent field method ( gc - scf ) and a direct variational free energy minimization method using auxiliary hamiltonians ( gc - auxh ) , to solve the kohn - sham equations of electronic density - functional theory directly in the grand canonical ensemble at fixed potential . both methods substantially improve performance compared to a sequence of conventional fixed - number calculations targeting the desired potential , with the gc - auxh method additionally exhibiting reliable and smooth exponential convergence of the grand free energy . finally , we apply grand - canonical dft to the under - potential deposition of copper on platinum from chloride - containing electrolytes and show that chloride desorption , not partial copper monolayer formation , is responsible for the second voltammetric peak . density - functional theory ( dft ) enables theoretical elucidation of reaction mechanisms at complex catalyst surfaces , making it now possible to design efficient heterogeneous catalysts for various industrial applications from first principles , for example for high - temperature gas - phase transformation of hydrocarbons to a variety of valuable chemical products. the extension of this predictive power to electrocatalysis would be highly valuable for an even broader class of technological problems , including a cornerstone of future technology for renewable energy : converting solar energy to chemical fuels by electrochemical water splitting and carbon dioxide reduction. accurately describing electrochemical phenomena , however , presents two additional challenges . first , the electrolyte , typically consisting of ions in a liquid solvent , strongly affects the energetics of structures and reactions at the interface . treating liquids directly in dft requires expensive molecular dynamics to sample the thermodynamic phase space of atomic configurations . historically , a number of continuum solvation models that empirically capture liquid effects have enabled theoretical design of liquid - phase catalysts. more recently , empirical solvation models suitable for solid - liquid interfaces, joint density - functional theory ( jdft ) for efficiently treating liquids with atomic - scale structure, and minimally - empirical solvation models derived from jdft, have made great strides towards reliable yet efficient treatment of electrochemical systems . second , electrons can flow in and out of the electrode as electrochemical reactions proceed . changes in electronic charge of electrode surfaces and adsorbates can be especially important because the electrolyte stabilizes charged configurations with a counter charge from the ionic response . for example , reduction of formic acid on platinum at experimentally relevant potentials is dominated by formate ions rather than neutral molecules at the surface. proton adsorption on stepped and polycrystalline surfaces involves displacing oxidatively - adsorbed water at relevant potentials , resulting in non - integer charge transfers and an anomalous ph dependence deviating from the nernst equation. accounting for the electrolyte response using our solvation models, and adjusting the electron number to match experimentally relevant electrode potentials , realistic predictions of electrochemical reaction mechanisms have now become possible. in particular , application of this methodology to the reduction of co on cu(111 ) predicts onset potentials for methane and ethene formation with 0.05 v accuracy in comparison to experiment , for a wide range of ph varying from 1 to 12. however , conventional dft software and algorithms are optimized for solving the quantum - mechanical problem at fixed electron number , requiring extra work ( both manual and computational ) to calculate properties for a specified electrode potential . this paper introduces algorithms for grand canonical dft , where electron number adjusts automatically to target a specified electron chemical potential ( related to electrode potential ) , thereby enabling efficient and intuitive first - principles treatment of electrochemical phenomena . section [ sec : theory ] summarizes the theoretical background of first - principles electrochemistry using jdft and continuum solvation models , and sets up the fundamental basis of grand - canonical dft . then , section [ sec : algorithms ] introduces the modifications necessary to make two distinct classes of dft algorithms , the self - consistent field ( gc - scf ) method and the variational free energy minimization using auxiliary hamiltonians ( gc - auxh ) , directly converge the grand free energy of electrons at fixed potential . sections [ sec : resultsscf ] and [ sec : resultsaux ] establish the algorithm parameter(s ) that optimize the iterative convergence of the gc - scf and gc - auxh methods respectively , while section [ sec : algocompare ] compares the performance of these algorithms for a number of prototypical electrochemical systems . finally , section [ sec : upd ] demonstrates the utility of grand canonical dft by solving an electrochemical mystery : the identity of the second voltammetric peak in the under - potential deposition ( upd ) of copper on platinum in chloride - containing electrolytes .

1701.04490

the exact helmholtz free energy @xmath0 of a system of interacting electrons in an external potential @xmath1 at a finite temperature @xmath2 satisfies the hohenberg - kohn - mermin variational theorem@xcite @xmath3 + \int d{\mathbf{r } } v({\mathbf{r } } ) n({\mathbf{r } } ) \right ) , \label{eqn : dft - hkm}\ ] ] where @xmath4 is a universal functional that depends only on the electron density @xmath5 ( and temperature ) , and not on the external potential . however , constructing approximations for this unknown universal functional that accurately capture the energies and geometries of chemical bonds in terms of the density alone is extremely challenging , partly because the quantum mechanics of the electrons is completely implicit in @xmath6 $ ] ( dropping the @xmath2 labels here onward for notational simplicity ; all the functionals below depend on temperature ) . most practical approximations in electronic density - functional theory follow the kohn - sham approach@xcite that includes the exact free energy of a non - interacting system of electrons with the same density @xmath5 . the universal functional is typically split as @xmath7 = a{\ensuremath{_{\textrm{ni}}}}[n ] + \underbrace{\int d{\mathbf{r } } \int d{\mathbf{r } } ' \frac{n({\mathbf{r } } ) n({\mathbf{r}}')}{2|{\mathbf{r}}-{\mathbf{r}}'|}}_{e_h[n ] } + e{\ensuremath{_{\textrm{xc}}}}[n ] , \label{eqn : dft - ks}\ ] ] where @xmath8 $ ] is the non - interacting free energy ( which we describe in detail below ) , the second ` hartree ' term @xmath9 $ ] is the mean - field coulomb interaction between electrons ( using atomic units @xmath10 throughout ) , and the final ` exchange - correlation ' term @xmath11 $ ] captures the remainder which is not known and must hence be approximated . the exact free energy of non - interacting electrons is @xmath12 = \min _ { \mbox{\scriptsize$\begin{array}{c } \{\psi_i({\mathbf{r}}),f_i\}\\ \rightarrow n({\mathbf{r } } ) \end{array}$ } } \sum_i \bigg ( \underbrace{\frac{f_i}{2 } \int d{\mathbf{r } } |\nabla\psi_i({\mathbf{r}})|^2}{\ensuremath{_{\textrm{kinetic}}}}\\ - t\underbrace{s(f_i)}{\ensuremath{_{\textrm{entropy } } } } \bigg ) , \label{eqn : dft - ni}\ ] ] which includes the kinetic energy and entropy contributions of a set of orthonormal single - particle orbitals @xmath13 with occupation factors @xmath14 $ ] . above , the single - particle entropy function is @xmath15 . note that we let the orbital index @xmath16 include spin degrees of freedom as well , and therefore do not introduce factors of 2 for spin degeneracy . for notational convenience , we also let @xmath16 include bloch wave - vectors in the brillouin zone ( in addition to spin and band indices ) for periodic systems . the minimization in ( [ eqn : dft - ni ] ) is constrained such that the density of the non - interacting system @xmath17 the density of the real interacting system . performing this minimization with lagrange multipliers @xmath18 ( the kohn - sham potential ) to enforce the density constraint and @xmath19 ( kohn - sham eigenvalues ) to enforce orbital normalization constraints , results in a set of single - particle schrdinger - like equations @xmath20 for stationarity with respect to @xmath13 , and the fermi occupation condition @xmath21 for stationarity with respect to @xmath22 . here , the electron chemical potential @xmath23 appears as a lagrange multiplier to enforce the electron number constraint , and is chosen so that @xmath24 , the number of electrons in the system . optimizing the total free energy functional ( [ eqn : dft - hkm ] ) with @xmath6 $ ] implemented by ( [ eqn : dft - ks],[eqn : dft - ni ] ) then yields the stationarity condition with respect to electron density , @xmath25({\mathbf{r } } ) = v({\mathbf{r } } ) + \frac{\delta}{\delta n({\mathbf{r } } ) } \left ( e_h[n ] + e{\ensuremath{_{\textrm{xc}}}}[n ] \right ) . \label{eqn : v_ks}\ ] ] conventional density - functional theory calculations then amount to self - consistently solving the kohn - sham equations ( [ eqn : schrodingerks ] ) along with ( [ eqn : v_ks ] ) , coupled via the electron density constraint ( [ eqn : densityconstraint ] ) . the kohn - sham potential is arbitrary up to an overall additive constant : changing this constant introduces a rigid shift in the eigenvalues @xmath26 and the electron chemical potential @xmath23 , but does not affect the occupations @xmath22 , electron density or free energy . for finite systems ( of any charge ) and for neutral systems that are infinitely periodic in one or two directions , this arbitrariness can be eliminated by requiring that the potential vanishes infinitely far away from the system , giving meaning to the absolute values of @xmath23 and @xmath26 as being referenced to ` zero at infinity ' . however , for materials that are infinite in all three dimensions , such as periodic crystalline solids , there is no analogous natural choice for the zero of potential , making the absolute reference for @xmath23 and @xmath26 meaningless . more importantly , for systems that are periodic in one , two or all three directions , the net charge per unit cell must be zero , otherwise the energy per unit cell becomes infinite . in terms of a finite system size @xmath27 and then taking the limit @xmath28 , the energy per unit cell of a system with net charge per unit cell diverges @xmath29 for one periodic direction , @xmath30 for two periodic directions and @xmath31 for three periodic directions . therefore , the number of electrons per unit cell is physically constrained to keep the unit cell neutral in systems with _ any _ periodicity , and only finite systems ( like molecules and ions ) have number of electrons as a degree of freedom . for describing electrochemical systems and electrocatalytic mechanisms , we are typically interested in adsorbed species in a solid - electrolyte interface which exchange electrons with the solid ( electrode ) . in these systems , the solid surface , which we would describe in a density - functional calculation as a slab periodic in two directions , does have a net charge per unit cell that depends on the electrode potential . in contrast to the discussion at the end of the previous section , this is physically possible ( i.e. has a finite energy ) because the electrolyte contains mobile ions that respond by locally increasing the concentration of ions of opposite charge near the surface , thereby neutralizing the unit cell . ( the charge per unit areas of the electrode and electrolyte are equal and opposite . ) next to an electrolyte , the charge of a partially - periodic system ( one - dimensional ` wires ' or two - dimensional ` slabs ' ) is no longer constrained , allowing the number of electrons to vary . now , the absolute reference for the electron chemical potential @xmath23 does become physically meaningful , and @xmath23 now corresponds to the electrode potential that controls the number of electrons in the electrode . however , treating electrochemical systems using electronic density - functional theory alone is extremely challenging for a variety of reasons . first , treating liquids requires a statistical average over a large number of atomic configurations to integrate over thermodynamic phase space . for this , techniques such as molecular dynamics typically require calculations of at least @xmath32 atomic configurations ( instead of just one for a solid ) . second , such calculations require a large number of liquid molecules to minimize finite size errors in the molecular dynamics . for example , for electrolytes with a realistic ionic concentration of 0.1 m , there is _ on average _ one ion for a few hundred solvent molecules . making statistical errors in the ion number manageable in such calculations therefore requires @xmath33 solvent molecules with @xmath34 electrons , contrasted with a typical @xmath35 atoms with @xmath36 electrons in the electrode slab + adsorbate of interest . combined , these factors make density - functional molecular dynamics simulations of electrochemical systems prohibitively expensive computationally , in additional to being difficult to set up and analyze . a viable alternative to the above direct approach is to employ joint density - functional theory@xcite ( jdft ) , a variational theorem akin to the hohenberg - kohn theorem that makes it possible to describe the free energy of a solvated system in terms of the electron density @xmath5 for the solute and in terms of a set of nuclear densities @xmath37 ( where @xmath38 indexes nuclear species ) of the solvent ( or electrolyte ) . specifically , the equilibrium free energy of the combined solute and solvent systems minimizes @xmath39 \\ + \int d{\mathbf{r } } v({\mathbf{r } } ) n({\mathbf{r } } ) + \sum_\alpha \int d{\mathbf{r } } v_\alpha({\mathbf{r } } ) n_\alpha({\mathbf{r } } ) \bigg ) , \label{eqn : jdft - theorem}\end{gathered}\ ] ] where @xmath1 is the external electron potential , @xmath40 is the external potential on the liquid nuclei and @xmath41 is a universal functional independent of these external potentials . separating out the hohenberg - kohn electronic density functional @xmath42 $ ] for the solute , the total free energy is @xmath43 = \underbrace{a{\ensuremath{_{\textrm{hkm}}}}[n]}{\ensuremath{_{\textrm{electronic } } } } + \underbrace{\tilde{a}{\ensuremath{_{\textrm{diel}}}}[n,\{n_\alpha\}]}{\ensuremath{_{\textrm{solvation}}}}. \label{eqn : jdft}\ ] ] in practice , the functional @xmath44 , much like @xmath45 , is unknown and needs to be approximated . importantly , the liquid is now described directly in terms of its average density rather than individual configurations , and the expensive quantum - mechanical theory of the electrons is restricted to the solute alone , thereby addressing both the sampling and system - size problems that make density - functional molecular dynamics prohibitively expensive . typically , we are interested in a situation where the external potentials on the liquid are zero , and the liquid only interacts with itself , and with the electrons and nuclei of the solute . in this situation , we can perform the optimization over liquid densities and define the implicit functional @xmath46 = a{\ensuremath{_{\textrm{hkm}}}}[n ] + a{\ensuremath{_{\textrm{diel}}}}[n]$ ] where we define @xmath47\equiv \min_{\{n_\alpha({\mathbf{r}})\ } } \tilde{a}_x[n({\mathbf{r } } ) , \{n_\alpha({\mathbf{r}})\}]$ ] for @xmath48 = jdft and @xmath48 = diel . we will work with these reduced functionals below for simplicity . the framework of joint density - functional theory encompasses an entire hierarchy of solvation theories . further separating the solvation term @xmath44 into a classical density functional for the liquid@xcite and an electron - liquid interaction functional , unlocks the full potential of jdft to describe atomic - scale structure in the liquid without statistical sampling . starting from ` full - jdft ' and performing perturbation theory for linear - response of the liquid results in the non - empirical salsa solvation model@xcite which continues to capture the atomic - scale nonlocality in the liquid response and introduces no fit parameters for the electric response of the solvent . at the simplest end of the jdft hierarchy , are continuum solvation models that neglect the nonlocality of the liquid response and replace it by that of an empirically - determined dielectric cavity ( optionally with debye screening due to electrolytes ) . this includes our recent candle solvation model@xcite that builds on the stability of salsa for highly polar systems , and earlier solvation models suitable for molecules and less - polar systems such as our glssa13 model@xcite ( or its equivalent , vaspsol@xcite ) and the comparable self - consistent continuum solvation ( sccs ) model.@xcite even the traditional quantum - chemistry finite - system solvation models such as the pcm series@xcite and the smx series@xcite can be mapped on to this class of solvation models . for simplicity , we will work here with this simplest class of continuum solvation models . the theoretical considerations and algorithms in this work focus primarily the electronic density - functional theory component , are largely agnostic to the internals of the solvation model and therefore straightforwardly generalize up the jdft hierarchy to the more detailed and complex solvation theories . essentially , all the simple electron - density - based continuum solvation models@xcite can be summarized abstractly as@xcite @xmath49 = \int d{\mathbf{r } } \rho{\ensuremath{_{\textrm{el}}}}({\mathbf{r } } ) \frac{(\hat{k}^{-1}-\hat{\chi})^{-1}-\hat{k}}{2 } \rho{\ensuremath{_{\textrm{el}}}}({\mathbf{r } } ) + a{\ensuremath{_{\textrm{cav}}}}[s ] . \label{eqn : adielpcm}\ ] ] here , the first term is the electrostatic solute - solvent interaction energy given by the difference between the solvent - screened coulomb interaction and the bare coulomb interaction @xmath50 of the total solute charge density , @xmath51 , where @xmath52 is the solvent nuclear charge density . the screened coulomb interaction term above is expressed in terms of the bare coulomb interaction @xmath50 and the solvent susceptibility operator @xmath53 , defined for the local - response models by @xmath54 here , @xmath55 is the bulk dielectric constant of the solvent and @xmath56 is the inverse debye screening length in vacuum of a set of ionic species of charge @xmath57 with concentrations @xmath58 ( the net inverse debye screening length in the presence of the background dielectric is @xmath59 ) . the cavity shape function @xmath60 modulates the dielectric and ion response , and varies from zero in the solute region ( no solvent present ) to unity in the solvent at full bulk density . finally , the second term of ( [ eqn : adielpcm ] ) , @xmath61 , empirically captures effects beyond mean - field electrostatics , such as the free energy of cavity formation in the liquid and dispersion interactions between the solute and solvent.@xcite different solvation models at this level of the hierarchy only differ in the details of how @xmath60 is determined from the electron density ( or even from atomic positions and fit radii for the pcm@xcite and smx@xcite solvation models ) , and in the details of @xmath61 . in practice , evaluating the first term of ( [ eqn : adielpcm ] ) requires the calculation of @xmath62 , which is the net electrostatic potential including screening by the solvent ( and electrolyte ) . the second part of the first term in ( [ eqn : adielpcm ] ) represents the electrostatic interactions between electrons and the nuclei in vacuum which cancels corresponding terms in the vacuum dft functional ( specifically the hartree , ewald and long - range part of the local pseudopotential terms ) , so that all long - range terms that contribute to the net free energy are present in the first term of ( [ eqn : adielpcm ] ) . finally , substituting ( [ eqn : pcmchi ] ) and @xmath63 into the definition of @xmath64 results in the linearized poisson - boltzmann ( modified helmholtz ) equation @xmath65 without ionic screening ( second term of ( [ eqn : linearpbe ] ) above ) , the absolute reference for @xmath64 is undetermined and does not contribute to the bound charge charge induced in the liquid , @xmath66 ( given by the first term of ( [ eqn : pcmchi ] ) alone ) , because @xmath67 . however , with ionic screening , a constant shift of @xmath64 does affect the second terms of ( [ eqn : pcmchi ] ) and ( [ eqn : linearpbe ] ) , producing a charge response in the liquid and making the absolute reference of the electrostatic potential @xmath64 meaningful . in particular , integrating ( [ eqn : linearpbe ] ) over space yields @xmath68 from ( [ eqn : pcmchi ] ) , the left hand side is the negative of the total bound charge in the liquid , while the right hand side is the total charge of the solute system . therefore , the continuum electrolyte automatically compensates for any net charge in the solute and makes the complete system neutral . as a side effect , the absolute reference of the electrostatic potential is meaningful and automatically corresponds to ` zero at infinity ' . this happens because the greens function of ( [ eqn : linearpbe ] ) is @xmath69 in the bulk liquid with finite @xmath70 ( instead of @xmath71 ) , causing @xmath64 to exponentially decay to zero outside the solute region ( where @xmath72 ) . ref . gives a detailed version of the above discussion including rigorous proofs of the absoluteness of the reference for @xmath64 , and numerical details and algorithms for efficiently solving ( [ eqn : linearpbe ] ) with periodic boundary conditions in plane - wave basis dft calculations . the result that the electrolyte neutralizes charges in the solute is true more generally , even for solvation models with nonlinear@xcite and/or nonlocal response.@xcite however , as ref . describes in detail , for the general nonlinear case , unit cell neutralization is not automatic and must be imposed using a lagrange multiplier constraint ; this lagrange multiplier then fixes the absolute value of @xmath64 such that it is zero at infinity . the key conclusion of this discussion is that continuum electrolytes present two advantages . first , automatic charge neutralization implies we can perform well - defined calculations with net charge per unit cell in the solute . second , meaningful absolute reference values for potentials implies that the electron chemical potential @xmath23 ( which determines the electron occupations in ( [ eqn : fermioccupations ] ) ) is also referenced to zero at infinity . consequently , @xmath23 is related directly to the potential @xmath73 of the electrode providing the electron reservoir in experiments . this potential is typically referenced to the standard hydrogen electrode ( she ) , so that @xmath74 where @xmath75 is the absolute position of the standard hydrogen electrode relative to the vacuum level ( i.e. the zero at infinity reference ) . calculations can employ either the experimental estimate @xmath76 ev to relate the absolute and relative potential scales,@xcite or a theoretical calibration based on the calculated and measured potentials of zero charge of solvated metal surfaces.@xcite the latter approach has the advantage of minimizing systematic errors in the solvation model since they cancel between the calculations used for calibration and prediction . for example , with the candle solvation model we use below , the calibrated @xmath77 ev.@xcite using joint density - functional theory of continuum solvation models to treat electrolytes as described above , we can calculate the helmholtz free energy and electrochemical potential ( @xmath23 , and hence @xmath73 ) of specific microscopic configurations of adsorbates on electrode surfaces at a fixed number of electrons @xmath78 in the solute subsystem ( electrode + adsorbates ) . however , in electrochemical systems , @xmath78 is an artificial constraint because electrons can freely exchange between the electrode and an external circuit . instead , experiments set the electrode potential @xmath73 , and hence the electron chemical potential @xmath23 , and @xmath78 adjusts accordingly as a dependent variable . thermodynamically , this corresponds to switching the electrons from the finite - temperature , fixed - number canonical ensemble to the finite - temperature , fixed - potential grand - canonical ensemble . correspondingly , the relevant free energy minimized at equilibrium is the grand free energy @xmath79 , instead of the helmholtz free energy @xmath0 . the most straightforward approach to fixed - potential dft for electrochemistry is to repeat conventional fixed - charge dft calculations at various electron numbers @xmath78 to reach a target chemical potential @xmath23 . using a steepest descent / secant method to optimize @xmath78 results in convergence of @xmath23 typically within 10 iterations.@xcite however this approach is inefficient since it requires multiple dft calculations to calculate the grand free energy and charge of a single configuration , typically taking 3x the time of a single fixed - charge calculation ( not 10x because subsequent calculations have a better starting point and converge quicker).@xcite a more efficient approach would be to directly optimize the kohn - sham functional in the grand - canonical ensemble . for a solvated system , this corresponds to a small modification of the minimization problem ( [ eqn : jdft - theorem ] ) with @xmath41 given by ( [ eqn : jdft ] ) and with @xmath45 evaluated using the kohn - sham approach ( [ eqn : dft - ks],[eqn : dft - ni ] ) . the only differences are that the lagrange multiplier term @xmath80 that enforced the electron number constraint is replaced by @xmath81 which implements the legendre transform of the helmholtz free energy to the grand free energy , and that the fixed-@xmath78 constraint is removed . the electron occupation factors are still fermi functions ( [ eqn : fermioccupations ] ) , but @xmath23 is specified as an input ( instead of being adjusted to match a specified @xmath78 ) . operationally , this amounts again to solving the kohn - sham eigenvalue problem ( [ eqn : schrodingerks ] ) self - consistently with @xmath25({\mathbf{r } } ) = v({\mathbf{r } } ) + \frac{\delta}{\delta n({\mathbf{r } } ) } \left ( e_h[n ] + e{\ensuremath{_{\textrm{xc}}}}[n ] + a{\ensuremath{_{\textrm{diel}}}}[n]\right ) , \label{eqn : v_ks_sol}\ ] ] where there is now a single extra contribution to the potential due to the solvent ( electrolyte ) , and with the aforementioned changes to the lagrange multipliers and constraints . this conceptually simple modification , however , presents numerical challenges to the algorithms commonly used for solving the kohn - sham problem , such as the self - consistent field ( scf ) method , which solves the kohn - sham eigenvalue problem from an input density ( or kohn - sham potential ) , adjusting this density ( or potential ) iteratively until self - consistency is achieved . a common instability for metallic systems in the scf method is ` charge sloshing ' : the electron density oscillates spatially between iterations instead of converging . switching to the grand - canonical ensemble can significantly exacerbate this problem : electrons can now additionally slosh between the system and the electron reservoir . here , we present modifications to the scf method and an alternate algorithm that allow reliable and efficient convergence for _ grand - canonical _ kohn - sham dft . given electron density @xmath82 , the kohn - sham equations ( [ eqn : schrodingerks ] ) with potential @xmath83 given by ( [ eqn : v_ks_sol ] ) define orbitals @xmath84 and eigenvalues @xmath85 . at a given electron chemical potential @xmath23 , these in turn define the occupations @xmath86 given by ( [ eqn : fermioccupations ] ) and a new electron density @xmath87 given by ( [ eqn : densityconstraint ] ) . the self - consistent field ( scf ) method attempts to find @xmath88 such that @xmath89 . there are two algorithmic ingredients to this method : solution of the kohn - sham eigenvalue equations and optimization of the electron density . the eigenvalue equations remain unchanged between conventional and fixed - potential calculations . to solve , these we use the standard davidson algorithm.@xcite the difficulty in fixed - potential calculations arise in the electron density optimization , which we discuss below . a robust and frequently - used algorithm for charge - density optimization is pulay mixing@xcite with kerker preconditioning.@xcite briefly , pulay mixing assumes that the residual @xmath90 \equiv n{\ensuremath{_{\textrm{out}}}}({\mathbf{r } } ) - n{\ensuremath{_{\textrm{in}}}}({\mathbf{r}})$ ] is approximately linear in the input electron density @xmath91 , and calculates the optimum input electron density as a linear combination of previous iterations , @xmath92 minimization of the norm of the corresponding residual @xmath93 \hat{m } r[n^{(j)}{\ensuremath{_{\textrm{in}}}}({\mathbf{r}})]\ ] ] with the constraint @xmath94 yields a set of linear equations determining the coefficients @xmath95 , where @xmath96 is the metric for defining the norm of the residual . finally , the next input density is obtained by mixing the optimum input density with its corresponding output density @xmath97,\ ] ] where @xmath50 is the kerker preconditioning operator . the metric @xmath96 and preconditioner @xmath50 serve to balance the influence of different components of the electron density to the optimization procedure . these are usually defined in reciprocal space , expanding @xmath98 , where @xmath99 are reciprocal lattice vectors . in reciprocal space , the hartree potential in ( [ eqn : v_ks_sol ] ) takes the form @xmath100 , causing small @xmath101 ( long - wavelength ) variations of the electron density to produce larger changes in the kohn - sham potential than large @xmath101 ones . uncompensated , this makes the electron density optimization unstable against long - wavelength perturbations ( the charge - sloshing instability ) . the kerker preconditioner @xmath102 mitigates this problem by suppressing the contribution of the problematic small @xmath101 components in determining the next input density . the metric @xmath103 enhances the contribution of small @xmath101 components in the residual norm , thereby prioritizing the optimization of these components when determining @xmath104 . the wavevectors @xmath105 and @xmath106 , which control the balance between short and long - wavelength components , and the prefactor @xmath0 , which sets the maximum fraction of @xmath107 that contributes to the next @xmath108 , can be adjusted to optimize the convergence of the scf method . see ref . for a detailed discussion of the pulay - kerker scf approach and its performance for conventional fixed electron number ( canonical ) dft calculations . the @xmath109 component of the electron density , which equals @xmath110 , where @xmath111 is the unit cell volume , remains fixed in canonical dft calculations and is therefore excluded from the metric and the preconditioner . this is no longer true in fixed - potential dft , where the number of electrons changes . with the above prescriptions , the kerker preconditioner ( [ eqn : kerker ] ) @xmath112 as @xmath113 which will prevent the electron number from changing between scf iterations . likewise , the @xmath113 divergence of the pulay metric ( [ eqn : metric ] ) cause the residual norm to become undefined when the electron number changes between iterations . in order to generalize the pulay - kerker scf approach for fixed - potential dft , we therefore need to fix the @xmath113 behavior of both the preconditioner and the metric . the need for the preconditioner and the metric arose from the reciprocal space coulomb operator @xmath114 in the hartree potential . in a uniform electrolyte with debye screening , the reciprocal space coulomb operator instead takes the form @xmath115 ( from ( [ eqn : linearpbe ] ) in reciprocal space with @xmath116 ) . since the electrolyte is responsible for fixing the indeterminacy of the @xmath109 component of the potential , a reasonable _ ansatz _ for extending pulay - kerker to fixed potential calculations is replacing @xmath117 with @xmath118 , where @xmath119 in section [ sec : resultsscf ] below , we show that setting @xmath120 and @xmath121 indeed makes the grand canonical self - consistent field ( gc - scf ) method function efficiently , with optimum convergence for @xmath122 given by ( [ eqn : qkappaopt ] ) . an alternate approach to solving the kohn - sham equations is to directly minimize the total ( free-)energy functional ( [ eqn : dft - hkm ] ) , with @xmath45 given by ( [ eqn : dft - ks],[eqn : dft - ni ] ) , in terms of the kohn - sham orbitals as independent variables . for joint density - functional theory , this amounts to @xmath123 + e{\ensuremath{_{\textrm{xc}}}}[n ] + a{\ensuremath{_{\textrm{diel}}}}[n ] + \int d{\mathbf{r } } v({\mathbf{r } } ) n({\mathbf{r } } ) \right ] , \label{eqn : jdft - direct}\end{gathered}\ ] ] where @xmath5 is now derived from @xmath124 and @xmath125 as given by ( [ eqn : densityconstraint ] ) . in the above minimization , the orbitals @xmath124 must be orthonormal , the occupation factors must satisfy @xmath126 , and optionally , @xmath127 for the canonical fixed electron number case . for insulators at @xmath128 , the occupations @xmath22 are known in advance . the constrained optimization over orthonormal @xmath124 is most efficiently carried out using a preconditioned conjugate - gradients ( cg ) algorithm on unconstrained orbitals using the analytically - continued energy functional approach.@xcite briefly , this approach evaluates the energy functional ( [ eqn : jdft - direct ] ) on a set of orthonormal orbitals , which are a functional of the unconstrained orbitals used for minimization . ( see ref . for further details . ) the general case of metallic systems and/or finite @xmath2 additionally requires the optimization of @xmath125 , which is challenging for non - linear optimization algorithms because of the inequality constraints @xmath126 . one possibility is to update the fillings from the kohn - sham eigenvalues using ( [ eqn : fermioccupations ] ) after every few steps of the cg algorithm,@xcite but this hinders the convergence of cg because the functional effectively changes each time the fillings are altered . the ` ensemble dft ' approach@xcite rectifies this convergence issue by optimizing the occupation factors at fixed orbitals in an inner loop , and performing the optimization of orbitals in an outer loop using the cg method , but this increases the computational cost compared to the case of the insulators . the scf approach is typically much more computational efficient than these variants of the direct variational minimization algorithm with variable occupations.@xcite an alternate strategy for direct variational minimization with variable occupations is to introduce an auxiliary subspace hamiltonian matrix @xmath129 as an independent variable,@xcite and setting the occupations @xmath130 in terms of the eigenvalues @xmath131 of @xmath129 instead of the kohn - sham eigenvalues @xmath132 . ( here , @xmath133 is the fermi function given by ( [ eqn : fermioccupations ] ) , and the electron chemical potential @xmath23 is chosen so as to satisfy the electron number constraint @xmath134 . ) this eliminates the problematic inequality constraints on the occupations , and upon minimization , the auxiliary subspace hamiltonian @xmath135 approaches the true subspace hamiltonian , @xmath136 , where @xmath137 is the kohn - sham hamiltonian . by choosing the undetermined unitary rotations of the orbitals @xmath124 to diagonalize @xmath129 , ref . further shows that the gradient of the free energy with respect to the independent variables simplifies to @xmath138 and @xmath139 these gradients are used to perform line minimization along a search direction in the space of independent variables . with every update of the independent variables , the orbitals are re - orthonormalized and the unitary rotations of the orbitals are updated to keep the auxiliary hamiltonian diagonal . in the preconditioned cg algorithm , the next search direction is obtained as a linear combination of the current search direction and the preconditioned gradients given by @xmath140 and @xmath141 where @xmath142 and @xmath143 are preconditioners . the role of preconditioning is to balance the weight of different directions in the minimization space in the explored search directions , and ideally the preconditioner equals the inverse of the hessian ( which is difficult to compute exactly ) . for the orbital directions , the standard preconditioner @xmath142 resembles the inverse of the dominant kinetic energy operator in the kohn - sham hamiltonian.@xcite for the auxiliary hamiltonian direction , the preconditioner removes the fermi function derivatives and finite difference factors from ( [ eqn : hauxgradient ] ) in order to more equitably weight all components of @xmath129 . the preconditioning factor @xmath143 controls the overall contribution of the @xmath129 components relative to that of the orbital components , and is adjusted to achieve optimum convergence . in this auxiliary hamiltonian ( auxh ) approach , the orbitals and occupations are continuously and simultaneously optimized to minimize the total free energy , resulting in better convergence and computational efficiency comparable to the fixed - occupations insulator case , and competitive with the scf method even for metallic systems . see ref . for further details on the algorithms and performance comparisons for conventional fixed electron number calculations . now , for fixed potential calculations , we set the occupations to fermi functions of the auxiliary hamiltonian eigenvalues at a specified @xmath23 , instead of selecting @xmath23 based on the electron number constraint . correspondingly , the second term of the auxiliary hamiltonian gradient ( [ eqn : hauxgradient ] ) , which arises from this constraint , drops out , and the algorithm requires no further modification . the convergence rate of this algorithm , however , is sensitive to the preconditioning factor @xmath143 and we propose a modified heuristic to update @xmath143 automatically and continuously . at the end of each line minimization , the derivative of the optimized free energy @xmath144 with respect to @xmath143 can be evaluated from the overlap between the auxiliary hamiltonian gradient and search direction.@xcite the line - minimized energy is optimum for @xmath145 . therefore , if variation of @xmath144 with respect to @xmath143 is convex , we should increase @xmath143 if we find @xmath146 and vice versa . however convexity is often lost if @xmath143 is initialized at too high a value . therefore , our heuristic tries to zero @xmath147 while limiting the contribution of the auxiliary hamiltonian gradient . in particular , we update @xmath148 \label{eqn : kadjust}\ ] ] at the end of each line minimization , where @xmath149 to saturate the factor by which @xmath143 can change in one iteration , @xmath150 is the overlap of the @xmath129 components of the gradient and preconditioned gradient , and @xmath151 is the total overlap of the gradient and preconditioned gradients ( orbital + @xmath129 ) . finally , we reset the conjugate - gradient algorithm ( i.e. set the search direction to the negative of the preconditioned gradient direction ) after @xmath143 has increased or decreased by a factor greater than @xmath152 . we do this because dynamically changing the preconditioner technically invalidates the strict orthogonality of the cg search direction with previous directions . section [ sec : resultsaux ] below shows that this heuristic exceeds the convergence obtained with fixed @xmath143 , while section [ sec : algocompare ] shows that the grand - canonical auxiliary hamiltonian ( gc - auxh ) algorithm consistently outperforms gc - scf for fixed - potential calculations . we implement all algorithms and perform all calculations using the open - source plane - wave density - functional theory software , jdftx.@xcite all calculations in this work employ the pbe@xcite exchange - correlation functional with gbrv ultrasoft pseudopotentials@xcite at a kinetic energy cutoff of 20 hartrees for kohn - sham orbitals and 100 hartrees for the charge density . the metal surface calculations use inversion - symmetric slabs of at least five layers , with at least 15 @xmath153 vacuum separation and truncated coulomb potentials@xcite to minimize interactions with periodic images . for brillouin zone integration , we use a fermi smearing of @xmath154 hartrees and a monkhorst - pack @xmath155-point mesh along the periodic directions with the number of @xmath155-points chosen such that the effective supercell is larger than 30 in each direction . we use the candle solvation model to describe the effect of liquid water and debye screening due to 1 m electrolyte , which we showed recently to most accurately capture the solvation of highly - charged negative and positive solutes.@xcite we emphasize that the methods and algorithms described above do not rely on specific choices for the pseudopotential , exchange - correlation functional , @xmath155-mesh or solvation model ; we keep these computational parameters constant here for consistency . . upper panel shows the convergence of grand free energy @xmath156 on a logarithmic scale , and lower panel shows that of electron number @xmath78 . best convergence is obtained with @xmath157 , the inverse debye - screening length , which we use as the default value henceforth . results shown here are for a five - layer ( 5ml ) cu(111 ) slab solvated in 1 m candle aqueous electrolyte , with potential fixed to 1v she ( @xmath158 ) , starting from a converged neutral calculation of the same slab in vacuum . [ fig : scf - qkappa ] ] the self - consistent field ( gc - scf ) algorithm summarized in section [ sec : algoscf ] depends on several parameters that control its iterative convergence . all these parameters are common to the conventional fixed - charge version ( scf ) and the fixed - potential variant ( gc - scf ) introduced here , except for the low - frequency cutoff wavevector @xmath122 that is necessary to allow the net electron number to change in the fixed - potential case . figure [ fig : scf - qkappa ] compares the dependence of gc - scf convergence on @xmath122 for a prototypical calculation of an electrochemical system : a cu(111 ) surface treated using a five - layer inversion - symmetric slab surrounded by 1 m aqueous non - adsorbing electrolyte treated using the candle solvation model,@xcite at a fixed potential of @xmath158 ( 1v she ) . ( the remaining gc - scf parameters are set to their default values which we discuss below . ) the best convergence is obtained with @xmath159 which corresponds to the debye screening length of the electrolyte ( [ eqn : qkappaopt ] ) . for small @xmath122 , including the conventional case of @xmath160 , the number of electrons does not respond sufficiently quickly , stalling at about 0.1 electrons from the converged value , correspondingly with the free energy stalling at about 0.1 @xmath161 ( @xmath162 ev ) away from the converged value . larger @xmath122 causes the electron number to change too rapidly , hindering convergence and eventually leading to a divergence as seen for the case of @xmath163 . an issue remains in the convergence independent of @xmath122 : after initial convergence , the free energy oscillates at the @xmath164 level , while the electron number oscillates at the @xmath165 level . . good convergence is observed near the typical recommended value @xmath166 ( @xmath167 @xmath168 ) . convergence is relatively insensitive to @xmath105 near this value , but becomes unstable for small @xmath105 approaching the low - frequency cutoff @xmath122 . system and remaining details are identical to figure [ fig : scf - qkappa ] . [ fig : scf - qkerker ] ] keeping @xmath122 at this optimum value given by ( [ eqn : qkappaopt ] ) , we next examine the dependence of gc - scf convergence on the remaining algorithm parameters for the same example system . figure [ fig : scf - qkerker ] shows the dependence on the kerker mixing wavevector @xmath105 , which helps stabilize the gc - scf algorithm against long - wavelength charge oscillations . optimal convergence is obtained for the typical recommended value@xcite of @xmath169 ( @xmath167 @xmath168 ) . as expected , convergence is relatively insensitive to the exact choice of @xmath105 , as long as @xmath105 does not become so small that convergence is ruined by charge sloshing . notice that the final convergence beyond the @xmath164 and @xmath165 electron level remains an issue that is not resolved for any choice of @xmath105 . . convergence is relatively insensitive to @xmath106 , and we set @xmath170 henceforth . system and remaining details are identical to figure [ fig : scf - qkappa ] . [ fig : scf - qmetric ] ] next , figure [ fig : scf - qmetric ] shows the variation of gc - scf convergence with the wavevector @xmath106 controlling the reciprocal - space metric used by the pulay algorithm . the convergence is entirely insensitive to this choice , and we henceforth set @xmath170 ( the recommended value@xcite ) . again , the final convergence issue remains unaffected by the choice of @xmath106 . . convergence is relatively insensitive to @xmath0 , except that it slows down for small @xmath0 we henceforth set @xmath171 which nominally exhibits the best convergence . system and remaining details are identical to figure [ fig : scf - qkappa ] . [ fig : scf - mixfraction ] ] finally , figure [ fig : scf - mixfraction ] compares the dependence of gc - scf convergence on the maximum kerker mixing fraction @xmath0 , which effectively controls what fraction of the new electron density is mixed into the current value . we find nominally best convergence for @xmath171 , but the performance of other values is not much worse . smaller values of @xmath0 lead to greater stability initially , but marginally slower convergence later on , while larger values of @xmath0 lead to greater oscillations initially , but faster convergence later on . regardless , as before , good convergence is obtained until the free energy reaches the @xmath164 level , but continues to oscillate at that level beyond that point . , and with vacuum spacing ranging from 15 @xmath172 to 35 @xmath172 with the 5ml slab . convergence slows only marginally with increasing system size , with either vacuum spacing or layer count . each calculations is for solvated cu(111 ) charged to 1v she in candle electrolyte , starting from the state of the corresponding converged neutral vacuum calculation . [ fig : scf - size ] ] this final convergence issue may not affect practical calculations where relevant energy differences are at the @xmath173 level or higher . however , smooth exponential convergence to the final answer is desirable as this makes it easier to determine when the target accuracy has been reached . unfortunately , no combination of gc - scf parameters achieves uniformly smooth convergence for the fixed - potential case . on the other hand , with our @xmath122 modification , the gc - scf algorithm at least converges to the @xmath164 level independent of system size ( figure [ fig : scf - size ] ) , with the number of cycles required for convergence remaining mostly unchanged with increasing number of cu(111 ) layers , and with increasing thickness of the solvent region . , the preconditioning scale factor for subspace rotations generated by the auxiliary hamiltonian . near - optimal convergence is obtained when @xmath143 is automatically adjusted using the heuristic given by ( [ eqn : kadjust ] ) , which we use by default henceforth . calculations are for solvated 5ml cu(111 ) at 1v she , starting from the corresponding neutral vacuum calculation , exactly as in figures [ fig : scf - qkappa]-[fig : scf - mixfraction ] . note the smooth exponential convergence ( without oscillations in electron number and free energy ) here , in contrast to the gc - scf case . [ fig : aux - k ] ] the grand - canonical auxiliary hamiltonian ( gc - auxh ) approach discussed in section [ sec : algoaux ] , directly minimizes the total free energy of the system without assuming any models for physical properties of the system ( such as dielectric response models that are built into the scf mixing schemes ) . this algorithm contains a single parameter @xmath143 , which weights the relative contributions of the kohn - sham orbital and subspace hamiltonian degrees of freedom in the conjugate - gradients search direction for free energy minimization . figure [ fig : aux - k ] shows the dependence of iterative convergence of the gc - auxh algorithm on this preconditioning parameter @xmath143 for the same cu(111 ) test problem considered above . if the preconditioning factor @xmath143 is held fixed , the rate of convergence is sensitive to the choice of @xmath143 , with the optimum choice being @xmath174 for this system . with the preconditioner auto - adjusted using the heuristic given by ( [ eqn : kadjust ] ) , we find that indeed the convergence picks up from that of the sub - optimal @xmath175 towards that of the optimal value . more importantly , we observe smooth exponential convergence of both the free energy and the electron number , in contrast to our experiences with the gc - scf method . for the gc - scf method ) . convergence is invariant with vacuum spacing , and slows down only marginally with number of layers . [ fig : aux - size ] ] figure [ fig : aux - size ] further shows that this smooth convergence sustains with changing system size . in particular , the convergence is virtually unchanged with the thickness of the solvent regions , but slows down slightly with increasing number of copper layers in the surface slabs . having analyzed and optimized the convergence of the gc - scf and gc - auxh methods , we now compare the performance of these algorithms for a few different cases . in this comparison , we also include the present state of the art : the ` loop ' method which uses a secant method to adjust the number of electrons in an outer loop to match the specified electron chemical potential.@xcite for a fair comparison , we use the fixed - charge scf method in the inner loops , because it achieves the fastest convergence ; this algorithm works equally well with the auxh method in the inner loop , but is then marginally slower . also , we now compare the wall time between algorithms , because there is no straightforward correspondence between gc - scf cycles and conjugate - gradient iterations of the gc - auxh method . free energy accuracy in half the wall time of the loop method , but the gc - auxh method is the clear winner with smooth exponential convergence . calculations here are for the 5ml cu(111 ) slab at 1v she , as before . timings are measured on a single 32-core nersc cori node in all cases . [ fig : compare - cu5 ] ] first , figure [ fig : compare - cu5 ] compares the performance of the algorithms for the 5-layer cu(111 ) slab used in all the tests so far . the spikes in the free energy seen in the loop method are the points where the electron number changes in the outer loop and a new scf convergence at fixed charge begins . both the gc - scf and gc - auxh methods are quite competitive , cutting the time to convergence within @xmath164 in half compared to the loop method . given the smooth convergence beyond @xmath164 however , the gc - auxh method is preferable over gc - scf for fixed potential calculations . , but for a 5ml pt(111 ) slab at 0v she ( @xmath176 ) . platinum is neutral at @xmath177v she , so this corresponds to negatively charging the slab , which activates candle s asymmetry correction . additionally , platinum has @xmath178 bands crossing the fermi level in contrast to copper which has occupied @xmath178 bands . this calculation therefore explores a more complicated charge vs. potential landscape , causing deviations from exponential convergence , but the relative performance of the algorithms remains similar . [ fig : compare - pt5 ] ] next , we compare these algorithms for more complex text cases . figure [ fig : compare - pt5 ] compares the convergence for a five - layer pt(111 ) slab fixed to a potential of @xmath176 ( 0v she ) . at this potential , the surface of pt(111 ) charges negatively , and the candle solvation model brings the cavity closer to the electrons to capture the more effective solvation of negative charges by liquid water . additionally , the dielectric response of platinum is more complex than copper due to the partially filled @xmath178 shell . both of these factors make this system harder to converge than the previous test case , and therefore the convergence of the gc - auxh method is no longer clearly a single exponential . despite this , both the direct grand - canonical methods converge faster than the loop method , with the gc - auxh method eking out an advantage in final convergence as before . and [ fig : compare - pt5 ] , but for a 5ml pt(111 ) slab decorated with a @xmath179 partial monolayer of adsorbed chloride anions ( 1/3 coverage ) at 0v she . the increased complexity slows down the convergence of the gc - scf method , which also slows down the loop over fixed - charge scf calculations , while the gc - auxh method continues to exhibit rapid near - exponential convergence with no charge oscillations , beating the loop method by a factor of 4 in time . all subsequent calculations of complex metal surfaces with adsorbates therefore use the gc - auxh method . [ fig : compare - pt5_cl ] ] finally , figure [ fig : compare - pt5_cl ] compares the convergence for chloride anions adsorbed at one - third monolayer coverage , in a @xmath179 supercell of a five - layer pt(111 ) slab . despite the increased complexity , the direct grand - canonical methods exhibit the best convergence , edging out the loop method by a factor of four in wall time now , again with smoothest convergence for the gc - auxh method . due to the systematic convergence advantage of the gc - auxh method , we use it for all remaining calculations in this work and recommend it as the default general purpose algorithm for converging electrochemical calculations . in the theory section and all calculations so far , we used fermi smearing where the electron occupations are given by the fermi function ( [ eqn : fermioccupations ] ) . practical @xmath155-point meshes typically require the use of a temperature @xmath2 substantially higher than room temperature ( we used 0.01 @xmath161 which is approximately ten times higher ) , which could result in inaccurate free energies . such errors can be reduced substantially by changing the functional form of the occupations and the electronic entropy , with the caveat that the smearing width @xmath2 no longer corresponds to an electron temperature . common modifications include gaussian smearing , where the fermi functions are replaced with error functions , and cold smearing,@xcite where the functional form is chosen to cancel the lowest order variation of the free energy with @xmath2 . ( see ref . for details . ) , the gc - auxh method with cold smearing converges marginally slower than gaussian or fermi smearing , but still faster than when using any of the smearing methods with the smaller width of 0.001 @xmath161 . [ fig : smearing ] ] figure [ fig : smearing ] compares the performance of the preferred gc - auxh method for various smearing methods . the insets show the variation of the converged free energy and electron number with smearing width @xmath2 . gaussian smearing reduces the coefficient of the quadratic @xmath2 dependence compared to fermi smearing , while cold smearing cancels the quadratic dependence altogether , by design . the variation of electron number with smearing width is also reduced by cold smearing , but to a lesser extent . notice that the use of cold smearing marginally slows down the iterative convergence of the gc - auxh method . however , using cold smearing at a high width of 0.01 @xmath161 is still faster than using any smearing method with the lower width 0.001 @xmath161 that is close to room temperature ( because of the far denser brillouin zone sampling required for smaller widths ) . therefore , it is still advantageous to use cold smearing at elevated smearing widths , and so we use cold smearing with a width of 0.01 @xmath161 for the final demonstration below . the application of sufficiently negative ( reductive ) potentials on an electrode immersed in a solution containing metal ions , reduces those ions and results in bulk electro - deposition of metal on the surface . additionally , for many pairs of metals , a single monolayer of one metal deposits on a surface of the other at an under potential , that is , at a potential less favorable than for bulk deposition . this phenomenon of under - potential deposition ( upd ) has several technological applications since it enables precise synthesis of heterogeneous metal interfaces . it also serves as an archetype for fundamental studies of electrochemical processes ( see ref . for an extensive review ) , which makes it a perfect example for demonstrating our grand - canonical density - functional theory method . the basic reason for underpotential deposition is that the heterogeneous binding between the two metals is stronger than the homogeneous binding of the depositing metal to itself . indeed , metal pairs that exhibit underpotential deposition also display analogous phenomena in vapor adsorption.@xcite . however , the process in solution is far more complicated and highly sensitive to the composition of the solution because of competing adsorbates,@xcite as well as to the structure of the electrode surface.@xcite the upd of copper on pt(111 ) in the presence of chloride anions is particularly interesting and the subject of considerable debate in the literature . voltammetry for this system@xcite exhibits two well - separated under - potential peaks , as shown in the background of figure [ fig : upd ] . certain leed and in situ x - ray scattering studies of this system@xcite find evidence of a @xmath180 bilayer of copper and chloride ions co - adsorbed on the surface at potentials between the two peaks , suggesting that one peak corresponds to a formation of a partial layer , and the second peak , to the formation of the full monolayer . in contrast , other studies@xcite do not find this signature and propose that the additional peak arises from adsorption and desorption of chloride ions alone . to address this debate , we perform grand - canonical density functional theory calculations of various configurations of copper and chlorine adsorbed on a 5-layer pt(111 ) slab in @xmath179 and @xmath180 supercells . we determine the most stable configurations at each potential(@xmath23 ) and the potentials at which transitions between configurations occur by comparing their grand free energies . the relevant grand free energy of a configuration @xmath38 containing @xmath181 platinum atoms in the slab with @xmath182 copper and @xmath183 chlorine atoms adsorbed at the surface within the calculation cell is @xmath184 normalized by the number of surface platinum atoms @xmath185 in order to correctly compare energies of calculations in different supercells . ( @xmath186 for the unit cell , 6 for the @xmath179 supercell and 8 for the @xmath180 supercell , accounting for the top and bottom surfaces in the inversion - symmetric setup . ) since no light atoms are present , we safely neglect changes in vibrational contributions to the free energy between adsorbate configurations . above , @xmath187 is the free energy of adsorbate configuration @xmath38 calculated by fixed - potential dft , which is grand canonical with respect to the electrons at chemical potential @xmath23 ( related to the electrode potential by ( [ eqn : mucalibration ] ) as discussed at the end of section [ sec : theoryjdft ] ) . then , ( [ eqn : freeenergyupd ] ) above calculates the free energy @xmath188 which is additionally grand canonical with respect to all relevant atoms with chemical potentials @xmath189 , @xmath190 and @xmath191 . several conventions are possible in defining the electron - grand - canonical free energy @xmath156 , depending on what electron number we subtract : change from neutral value , total electron number , or number of valence electrons in pseudopotential dft calculations . the atom chemical potentials would then respectively correspond to neutral atoms , bare nuclei or pseudo - nuclei ( nuclei + core electrons in pseudopotential ) . the full grand canonical free energy @xmath192 does not depend on this choice . in our jdftx implementation , we choose the last option above ( number of valence electrons and correspondingly atom chemical potentials of the pseudo - nuclei ) . the bulk of the platinum electrode sets the pt chemical potential , @xmath193 , where @xmath194 is the dft energy of a bulk fcc pt calculation with a single atom in the unit cell , and @xmath195 is the number of valence electrons in that calculation . the second term here implements the electron counting convention discussed above . next , copper ions in solution set @xmath190 , but directly calculating the free energy of such ions using solvation models is error - prone.@xcite so , instead , we use the dft calculated energy , @xmath196 , of a bulk fcc cu calculation ( containing @xmath197 valence electrons ) , and relate it to the free energy of the ion via the experimentally - determined standard reduction potential @xmath198 v she.@xcite this yields @xmath199,\end{gathered}\ ] ] where the second term accounts for the change from cu(s ) to @xmath200 ions , and the final term accounts for change in ionic concentration from the standard value of 1 mol / liter to the current value of [ cu@xmath201 ( in mol / liter ) . similarly , chlorine ions in solution set @xmath191 , but to minimize dft errors , we connect to the dft calculated energy , @xmath202 , of an isolated chlorine atom ( containing @xmath203 valence electrons ) , via the experimentally - determined atomization energy @xmath204 kj / mol,@xcite gas - phase entropy @xmath205 j / mol - k,@xcite and reduction potential @xmath206 v she.@xcite specifically , @xmath207,\end{gathered}\ ] ] where the second term accounts for the change from atomic to gas - phase chlorine , the third term for the change to chloride ions , and the final term for the change in chloride ion concentration to [ cl@xmath208 ( in mol / liter ) . mol / liter cu@xmath209 ions and 10@xmath210 mol / liter cl@xmath211 ions . calculated free energies for various adsorbate configurations as a function of electrode potential are shown as calculated by explicit fixed - potential solvated calculations in ( a ) , and from vacuum calculations in ( b ) , with the experimental voltammogram@xcite shown for comparison . solid red lines indicate no copper , purple dotted lines indicate @xmath180 partial copper monolayer and blue dot - dash lines indicate full copper monolayer . thickness of the lines indicate cl coverage , ranging from no cl ( thinnest ) , @xmath180 ( 1/4 ) coverage ( intermediate ) to @xmath179 ( 1/3 ) coverage ( thickest ) . prominent configurations are sketched in ( c - f ) . vacuum calculations in ( b ) predict only a single voltammetric peak in disagreement with experiment . explicit potential - dependent solvated calculations in ( a ) predict two peaks in qualitative agreement with experiment ( @xmath212 v accuracy ) , with the second peak due to chloride desorption.@xcite the partial cu monolayer ( d ) proposed by some@xcite is not predicted to be the most stable configuration at any relevant potential . [ fig : upd ] ] figure [ fig : upd](a ) shows the calculated grand free energies as a function of electrode potential for a number of cu and cl adsorbate configurations on the surface of pt(111 ) . at high potentials , the most stable ( lowest free energy ) configuration is 1/4 cl coverage ( figure [ fig : upd](f ) ) , which transitions to a clean pt surface ( figure [ fig : upd](e ) ) at a potential of 0.55 v she . upon further lowering the potential , the stable configuration transitions to a full monolayer of copper with 1/3 cl coverage ( figure [ fig : upd](c ) ) at a potential of 0.46 v experimentally , the two voltammogram peaks are at approximately @xmath213 and @xmath214 v she , averaging over the forward and reverse direction sweeps . therefore chlorine desorption and full - copper - monolayer formation are plausible explanations@xcite of the two peaks , with our first - principles predictions reproducing well the peak spacing ( 0.09 ev versus 0.12 ev in experiment ) , and placing the absolute locations of the peaks to within 0.07 ev . the partial 2x2 monolayer of copper ( figure [ fig : upd](d ) ) proposed by others@xcite as the reason for the second peak is not the most stable configuration in our calculations at any potential , lying a singinificant 0.3 ev above the other phases at relevant potentials . for comparison , figure [ fig : upd](b ) shows the analogous results that would be obtained using only conventional vacuum calculations . in the above formalism , this corresponds to assuming @xmath215 , where @xmath216 is the helmholtz energy from a neutral vacuum dft calculation of configuration @xmath38 containing @xmath217 valence electrons . this approximation results in a single transition directly from the cl - covered pt surface to the one with a copper monolayer , predicting a single voltammogram peak in disagreement with experiment . accurate predictions for electrochemical systems therefore require treating charged configurations stabilized by the electrolyte at relevant electron potentials , now easily accomplished with the methods and algorithms introduced in this work . this work introduces algorithms for directly converging dft calculations in the grand - canonical ensemble of electrons , where the number of electrons adjusts to maintain the system at constant electron chemical potential , while ionic response in a continuum solvation model of electrolyte keeps the system neutral . we show that , with appropriate modifications , grand canonical versions of both the self - consistent field ( gc - scf ) method as well as direct free - energy minimization with auxiliary hamiltonians ( gc - auxh ) method are able to rapidly converge the grand free energy of electrons . this substantially improves upon the current state of the art of running an outer loop over conventional fixed - charge dft calculations . with detailed tests of the convergence of all these algorithms , we show that the gc - auxh method is the most suitable default choice exhibiting smooth exponential convergence to the minimum . grand - canonical dft directly mimics the experimental condition in electrochemical systems , where electrode potential sets the chemical potential of electrons , and the number of electrons at the electrode surface ( including adsorbates in the electrochemical interface ) changes continuously in response . describing this change in charge at the surface plays an important role in accurately modeling several electrochemical phenomena.@xcite here , we showcase the new algorithms by analyzing the under - potential deposition ( upd ) of copper on platinum in an electrolyte containing chloride ions . we resolve an old debate about the identity of a second under - potential peak , showing that partial copper monolayers are not plausible and that the second peak is due to desorption of chloride ions . we expect the new methods presented here to substantially advance the realistic treatment of electrochemical phenomena in first principles calculations . rs and wag acknowledge support from the joint center for artificial photosynthesis ( jcap ) , a doe energy innovation hub , supported through the office of science of the u.s . department of energy under award number de - sc0004993 . rs and taa acknowledge support from the energy materials center at cornell ( emc@xmath218 ) , an energy frontier research center funded by the u.s . department of energy , office of science , office of basic energy sciences under award number de - sc0001086 . calculations in this work used the national energy research scientific computing center ( nersc ) , a doe office of science user facility supported by the office of science of the u.s . department of energy under contract no . de - ac02 - 05ch11231 . we thank kendra letchworth - weaver , kathleen schwarz , yuan ping , hai xiao , tao cheng , robert smith nielsen and jason goodpaster for insightful discussions . 49ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop `` , '' ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) , ed . , @noop _ _ ( , ) @noop _ _ ,

in this paper we apply to gravitational waves ( gw ) from the merger phase of binary systems a recently derived frequentist methodology to calculate analytically the error for a maximum likelihood estimate ( mle ) of physical parameters . we use expansions of the covariance and the bias of a mle estimate in terms of inverse powers of snrs where the square root of the first order in the covariance expansion is the cramer rao lower bound ( crlb ) . we evaluate the expansions , for the first time , for gw signals in noises of gw interferometers . the examples are limited to a single , optimally oriented , interferometer . we also compare the error estimates using the first two orders of the expansions with existing numerical monte carlo simulations . the first two orders of the covariance allows to get error predictions closer to what is observed in numerical simulations than the crlb . the methodology also predicts a necessary snr to approximate the error with the crlb and provides new insight on the relationship between waveform properties , snr , dimension of the parameter space and estimation errors . for example the timing match filtering can achieve the crlb only if the snr is larger than the kurtosis of the gravitational wave spectrum and the necessary snr is much larger if other physical parameters are also unknown .

0912.0065

the ground - based gravitational waves detectors ligo , virgo , and geo 600 ( @xcite , @xcite , @xcite ) are rapidly improving in sensitivity . by 2015 , advanced versions of these detectors should be taking data with a design sensitivity approximately 10 times greater than the previous generation , and the probed volume will grow by a factor of about a thousand . such improvements in detector sensitivity mean that the first gravitational - wave signature of a compact - binary coalescence ( cbc ) could be detected in the next few years ( see for example @xcite ) . among the expected signals , a special role is covered by inspiralling compact binaries . this follows from the ability to model the phase and amplitude of the signals quite accurately and consequently to maximize the signal - to - noise ratio ( snr ) by using matched filtering techniques . matched filters also provide a maxim likelihood estimation of the waveform parameters such as component masses or time of coalescence . the choice of the mles as reference estimators is also motivated by the fact that if an unbiased estimator that attain the crlb exists , is the mle@xcite . + + the existing gw frequentist literature ( @xcite .. @xcite ) evaluates the mle accuracy in two ways : ( a ) analytically by calculating the so called fisher information matrix ( fim ) or equivalently the cramer rao lower bound ( crlb ) which is the square root of the diagonal elements of the inverse fim , and ( b ) numerically by performing monte carlo simulations . + the fim was derived analytically in ( @xcite,@xcite,@xcite ) using newtonian waveforms , extended to second - order post - newtonian ( @xcite,@xcite ) and recently revisited up to 3.5pn ( @xcite,@xcite ) . in ( @xcite),(@xcite ) the authors calculate the crlb for the three _ standard _ binary systems ( nns , nbh , bbh ) , and show how the errors change when the successive different pn orders are taken into account . they consider initial ligo , advanced ligo and virgo noises . they also considers pn corrections to the _ amplitude_. + monte carlo simulations were performed , for example in ( @xcite , @xcite ) , for the lowest pn orders , where it is also suggested that the inclusion of the higher pn orders would be computationally expensive . more recent monte carlo simulations with 3.5 pn waveforms are described in ( @xcite ) . + we did not try to compare the uncertainties derived here to other existing papers ( especially from the 90 ) since different parameter sets , noise spectra ( even for the same antenna ) and pn terms were used . for example in ( @xcite ) a comparison between the crlb and other bounds is done for a waveform at the 0th pn order . this work also uses different conventions on the waveform spectrum than more recent literature . in ( @xcite ) phasing is extended to the 1.5pn order . the spin parameters are taken in account but the noise spectrum for ligo is different than the currently used design noise curves . in ( @xcite ) , ( @xcite ) , ( @xcite ) the 2pn wave was used . in the work ( @xcite ) interesting observations are made about the fluctuation of the parameters variance with the pn orders , analyzing both the case of spin and spinless systems . the fluctuations of the variance in the spinless case is also stressed in ( @xcite ) . + + the crlb is a convenient tool to approximate the accuracies in large snrs and to obtain error bounds for unbiased estimators . unfortunately , for low snrs ( below 20 ) where the first detections might emerge , the crlb can grossly underestimate the errors @xcite,@xcite,@xcite,@xcite,@xcite . the reason is that with non linear data models and ( or ) with non gaussian noises the crlb depends only on the curvature of the likelihood function around the true value of the parameter . + + in this paper we apply a recently derived analytical tool to better predict an mle accuracy and to establish necessary conditions on the signal - to - noise ratio ( snr ) for the mle error to attain the crlb . explicitly , within the frequentist framework , for arbitrary probability distribution of the noise , expansions of the bias and the covariance of a mle in inverse powers of the snr are discussed . the first order of the expansion of the variance is the inverse of the fim . by requiring that the second order covariance is smaller , or much smaller , than the first order this approach predicts a necessary snr to approximate the error with the crlb . the higher order of the expansions are determinant in the low snr regime where the inverse fim underestimates the error . we compared our the errors computed using the first two orders of the expansions to the monte carlo simulations in ( @xcite ) . we observed the first two orders of the expansions get error predictions closer than the crlb to what is observed in the numerical simulations . in ( @xcite ) the simulations are related to the fim to estabish ranges of snr where the crlb describe the error . our expansions predict the same snr range of validity for the crlb . + the expansions are sensitive to the side lobes of the likelihood function because they make use of higher order derivatives than the second one ( which is only sensitive to the curvature of the main lobe ) . the methodology also provides new insight on the relationship between waveform properties , snr , dimension of the parameter space and estimation errors . for example the timing match filtering accuracy achieves the crlb only if the snr is larger than the kurtosis of the gravitational wave spectrum and the necessary snr is much larger if the other physical parameters are unknown . more specifically the mle of the arrival time for ns - ns binary signals might require an snr equal to 2 with the time as the only parameter or 15 when all the other parameters are unknown . these results are important to understand the domain of validity of recent papers like @xcite that defines @xmath0 confidence regions in direction reconstruction with time triangulation . the regions discussed in @xcite for snr smaller than 10 are based on timing mles , with the arrival time being the only unknown parameter , and the time uncertainties quantified by the crlb . + we also note that @xcite , using a formalism introduced in @xcite,@xcite , describes a consistency criterion , different from the condition derived in this paper , for the validity of the crlb that , if applied to a 4pp compact binary signal computed with a 2pn expansion and @xmath1 , requires an snr of at least 10 . at the time of the writing of this paper we established with m.vallisneri that the equation ( [ varmatrixsimplified ] ) of this paper becomes , in the one parameter case and colored gaussian noise , equivalent to equation ( 60 ) in ( @xcite ) or ( a35 ) in ( @xcite ) . a comparison for the gaussian noise and multi parameter case is object of current work , while a comparison for arbitrary noise is not possible because ( @xcite ) and ( @xcite ) use from the beginning of their derivations gaussian noises . the explicit calculations shown here for different gws are also not performed in ( @xcite ) and ( @xcite ) . + + in section [ expansions ] we present the explicit expressions of the expansions of the bias and the covariance matrix for arbitrary noise and size of the parameter space . in section [ match ] we explain how the expansion can be evaluated for signals in additive colored gaussian noise . in section [ waveformsection ] we describe the post - newtonian inspiral waveform used for the examples , describe the parameter space and the initial and advanced ligo noises . in section [ onedim ] we study the one - dimensional parameter space results when only one parameter at a time is considered unknown . in section [ fullparspace ] we present the results for full parameter space with the initial and advanced ligo noises . we also compare our results with published results from monte carlo simulations . in section [ conclusions ] we present some conclusions and in the appendix we describe the derivation of the expansions as well as the relationship of this method with the statistics literature . in this section we present the first two orders of the expansions in inverse powers of the snr for the bias and the covariance matrix . the details of the derivation are provided in appendix a. given an experimental data vector @xmath2 , where @xmath3 is the dimension , we assume that the data are described by a probability density @xmath4 that depends on a d - dimensional parameter vector @xmath5 . according to ( @xcite ) , we suppose that the mle @xmath6 of @xmath7 is given by a stationary point of the likelihood function @xmath8 with respect to the components of @xmath7 @xmath9 if we introduce the notations @xmath10 @xmath11\ ] ] where @xmath12 is the fisher information matrix @xmath13=e[l_{a}l_{b}]$ ] ( @xmath14 $ ] is the expected value ) , the first two orders of the bias for the mle of the @xmath15 component of the parameter vector @xmath7 are given by @xmath16 @xmath17 + \frac{i^{ma}i^{bd}i^{ce}}{8}[v_{abcde } + 4v_{ac , bde } + 8v_{de , abc } + 4v_{abce , d } + 4v_{abc , d , e}\nonumber\\ & + & 8v_{ab , cd , e}]+ \frac{i^{ma}i^{bc}i^{df}i^{eg}}{4}\bigg[(2v_{afed}v_{gb , c } + 2v_{bedf}v_{ac , g } + 4v_{abed}v_{gf , c } ) + ( v_{afed}v_{gcb } + \nonumber\\ & + & 2v_{abed}v_{gcf } + 2v_{dbeg}v_{acf } ) + ( 2v_{aed}v_{gb , fc } + 4v_{acf}v_{dg , eb } + 4v_{bed}v_{ac , gf } + 2v_{fcb}v_{ag , ed } ) + \nonumber\\ & + & ( 4v_{afe , g}v_{db , c } + 4v_{afe , c}v_{db , g } + 4v_{dbe , g}v_{af , c } ) + ( 2v_{abe , g}v_{cdf } + 4v_{dbe , g}v_{acf } + 4v_{abe , f}v_{cfg } + \nonumber\\ & + & 2v_{dge , b}v_{acf } ) + ( 4v_{ag , fc}v_{ed , b } + 4v_{ed , fc}v_{ag , b } + 4v_{ag , ed}v_{fc , b } ) \nonumber \\ & + & \left.(4v_{acg}v_{ef , b , d } + 2v_{cde}v_{ab , f , g } ) + \frac{2}{3}v_{abde}v_{c , f , g}\right ] \nonumber \\ & + & \frac{i^{ma}i^{bc}i^{de}i^{fg}i^{ti}}{8}[v_{adf}(v_{ebc}v_{gti } + 2v_{etc}v_{gbi } + 4v_{gbe}v_{tci } + 8v_{gbt}v_{eci } + 2v_{ebc}v_{gt , i } \nonumber \\ & + & 4v_{etc}v_{gb , i } + 2v_{gti}v_{eb , c } + 4v_{gtc}v_{eb , i } + 8v_{gbt}v_{ce , i } + 8v_{gbt}v_{ci , e } + 8v_{gbe}v_{ct , i } + 8v_{cte}v_{gb , i } \nonumber \\ & + & 4v_{cti}v_{gb , e } + 4v_{gt , i}v_{eb , c } + 4v_{eb , i}v_{gt , c } + 8v_{gt , b}v_{ic , e } + 8v_{gt , e}v_{ic , b } + 4v_{bet}v_{g , c , i } ) \nonumber \\ & + & v_{dci}(8v_{bgt}v_{ae , f } + 4v_{bgf}v_{ae , t } + 8v_{ae , t}v_{bg , f } + 8v_{ae , f}v_{bg , t } + 8v_{af , b}v_{ge , t})]\end{aligned}\ ] ] were we assumed the einstein convention to sum over repeated indices . for the covariance matrix the first order is the inverse of the fisher information matrix while the second order is given in by ( for simplicity we provide the diagonal terms ) : @xmath18\end{aligned}\ ] ] for this analysis we assume that the output of an optimally oriented gw interferometer has the form : @xmath19 where @xmath20 is the signal , which depends on the parameters vector @xmath21 , and @xmath22 a stationary gaussian noise with zero mean . the probability distribution can be written : @xmath23\,\omega(t - t_1)\,[x(t_1)-h(t_1,\underline{\theta})]\,{\mathrm{d}}t { \mathrm{d}}t_1}\right\}\ ] ] the first and second derivative of the log - likelihood give : @xmath24\,{\mathrm{d}}t { \mathrm{d}}t_1}\\ l_{ab } & \equiv & \frac{\partial l_a}{\partial \theta_b } = \int{\left[h_{ab}(t,\underline{\theta})\,\omega(t - t_1)\,[x(t_1)-h(t_1,\underline{\theta } ) ] - h_a(t,\underline{\theta})\,\omega(t - t_1)\,h_b(t_1,\underline{\theta } ) \right]{\mathrm{d}}t { \mathrm{d}}t_1}\end{aligned}\ ] ] @xmath25 = -e[l_{ab}]&= & \int{h_a(t,\underline{\theta})\,\omega(t - t_1)\,h_b(t_1,\underline{\theta})\,{\mathrm{d}}t { \mathrm{d}}t_1 } = \\ & = & \int{{\mathrm{d}}f { \mathrm{d}}f ' \,h_a(f ) \,h_b(f ' ) \,\omega(-f ,- f')}\end{aligned}\ ] ] it s easy to verify that @xmath26 where we have introduced , @xmath27,the _ one sided power spectral density _ defined as the fourier transform of the noise auto - correlation : @xmath28.\\ s_h(f)&\equiv & \int{{\mathrm{d}}t\ , e^{-2\pi i f t } r(t)}.\label{onesidednoise}\end{aligned}\ ] ] notice that the sign convention in ( [ sign ] ) follows from the implicit assumption that r(t ) is the fourier transform of @xmath29 $ ] . in the literature another convention with the minus sign is also found corresponding to defining r(t ) as the fourier transform of @xmath30 $ ] . using the relation @xmath31 , we can finally write the fim : @xmath32 = \langle h_a(f)\,,\,h_b(f)\rangle\ ] ] where @xmath33 are the derivatives of the fourier transform of the signal with respect to the a - th parameter . we have introduced a mean in the frequency space : @xmath34\ ] ] where the range of integration depends on the antenna properties and the theoretical model for the binary system . the snr corresponds to the optimal filter : @xmath35 we can express in the same way all the quantities we need in order to calculate the second order variance , like scalar products of @xmath36 derivatives . @xmath37 @xmath38 @xmath39 if one uses these functions , the form for the second order variance , eq ( [ varmatrix ] ) can be further simplified : @xmath40 we apply the general results of the theory to the case of a binary system in the inspiral phase . starting from the the 3.5pn phasing formula ( @xcite ) we write the fourier transform of the chirp signal : @xmath41\ ] ] where @xmath42 is the implicit solution of the 3.5pn phasing formula , using the _ stationary phase approximation _ ( spa ) ( @xcite , @xcite ) . the final result is : @xmath43 the phase is given by : @xmath44 where @xmath45 and @xmath46 are the arrival time and the arrival phase . the function @xmath47 can be defined either in terms of the total mass of the binary system , @xmath48 , or in terms of the chirp mass , @xmath49 : @xmath50 where @xmath51 is the symmetrized mass ratio @xmath52 the amplitude is a function of the chirp mass , the effective distance , and the orientation of the source : @xmath53 the coefficients @xmath54 s with @xmath55 ( the meaning of each terms being the @xmath56 pn contribution ) are given by : @xmath57\right)\end{aligned}\ ] ] @xmath58 where @xmath59 , @xmath60 . @xmath61 is the euler constant , and @xmath62 , with @xmath63 the _ last stable orbit _ frequency for a test mass in a scharzschild space - time of mass m : @xmath64 which will be also used as upper cutoff for the integrals ( [ meanfreq ] ) .. ( [ vabcd ] ) . given the waveform ( [ waveform ] ) one can easily calculate the fisher information matrix , and its inverse , the crlb . ( [ waveform ] ) contains five unknown parameters , @xmath65 ( the total mass @xmath66 could be used instead of the chirp mass ) , so that one should calculate a five dimensional square matrix . it was already observed by ( eg @xcite ) that the errors in the distance , and consequently the amplitude @xmath67 , are uncorrelated with the errors in other parameters , ie that the fisher information is block diagonal . we observed that this is also the case for the waveform ( [ waveform ] ) we use here for both the fim and the second order covariance matrix . we can therefore work in a four dimensional parameter space @xmath68 . however it is worth observing that this is not an obvious point since in general the amplitude estimation can be coupled with other other parameters if they enter in the signal in specific ways ( see ch.3 of @xcite ) . + it is also worthwhile noticing that the snr and the amplitude @xmath67 are related like follows : @xmath69 we perform the calculations using analytical forms of the design initial and advanced ligo noise spectrum ( [ iniligo]).the initial one sided power spectral density of ligo can be written for @xmath70 ( @xmath71 ) : @xmath72 , \ ] ] where the lower frequency cutoff value is @xmath73 , @xmath74 , @xmath75 , and @xmath76 . the advanced ligo one sided psd has the following expression , for @xmath70 ( @xmath77 ) : @xmath78 , \ ] ] where the lower frequency cutoff value is @xmath79 , @xmath74 , @xmath80 , and @xmath81 . they are plotted in fig . [ ligonoise ] [ cols="^,^ " , ] in this paper we applied a recently derived statistical methodology to gravitational waves generated by the inspiral phase of binary mergers and for noise spectral densities of gravitational wave interferometers . explicitly we computed the first two orders of mle expansions of bias and covariance matrix to evaluate mle uncertainties . we also compared the improved error estimate with existing numerical estimates . the value of the second order of the variance expansions allows to get error predictions closer to what is observed in numerical simulations than the inverse of the fim . the condition where the second order covariance is negligible with respect to the first order predicts correctly the necessary snr to approximate the error with the crlb and provides new insight on the relationship between waveform properties snrs and estimation errors . future applications include imr waveforms , network of detectors and source location estimation . mz would like to thank the national science foundation for the support through the awards nsf0855567 and nsf0919034 , while sv would like to thank the universite pierre - et - marie - curie and embry riddle aeronautical university for logistic and financial support . analytic expressions for the moments of a ( mle ) are often difficult to obtain given a non - linear data model . however , it is known from ( @xcite ) that likelihood expansions can be used to obtain approximate expressions for the moments of a mle in terms of series expansions in inverse powers of the sample size ( @xmath82 ) . these expansions are valid in small samples or snr where the mle may have significant bias and may not attain minimum variance . an expressions for the second covariance equivalent to the one used here , was first given in a longer form ( about a factor 2 ) , and with the prescription of isolating the different asymptotic orders inside the tensors , in ( @xcite ) . the expression presented here were first derived in an unpublished mit technical report ( @xcite ) by two of the authors of this paper , where ( a ) the bartlett identities ( @xcite,@xcite ) were used to simplify the expression of the second order variance , and derive the second order bias . ( b ) the prescription on the tensors no longer needed to be implemented . the final form of the second order covariance has already been published without proof , in two papers ( @xcite , @xcite ) , were the first and third author of this paper are coauthors , involving mle of source and environment parameters that use acoustic propagation within shallow water ocean wave guide . in this section , we derive the second order bias for a multivariate mle and we introduce a chain rule that allows the derivation of the of the second order covariance matrix from the second order bias . the explanation follows closely @xcite . the derivation of the bias is performed in two steps : first we derive the expansions in the non physical scenario of n statistically independent identical measurements and then set n=1 for the case of interest of this paper . + the derivation follows the approach of ( @xcite ) for likelihood expansions , ( @xcite ) for the large sample approximations and ( @xcite , @xcite ) for the notation of the observed likelihood expansions . for the small sample case of interest , @xmath83 orders generated from the likelihood expansion may contain different powers of @xmath84 , and the contributions to a single power of @xmath84 may have to be collected from different @xmath83 orders . ( @xcite ) extending the work of ( @xcite ) , confronted the equivalent problem of separating the powers in @xmath82 within the expectations of products of linear forms and arrived at expressions for the second order bias of the mle by embedding the derivation in a discrete probability scheme . some applications of their results are given in ( @xcite ) , ( @xcite ) , ( @xcite),(@xcite),(@xcite ) . we present here an independent derivation for the second order bias that is valid for general discrete or continuous random variables . + let us consider a set of @xmath82 independent and identically distributed experimental data vectors @xmath85 , where @xmath3 is the dimension of every vector . we assume that these data are described by a probability density @xmath86 that depends on a d - dimensional parameter vector @xmath5 . according to ( @xcite ) , we suppose that the mle @xmath6 of @xmath7 is given by a stationary point of the likelihood function @xmath87 with respect to the components of @xmath7 @xmath88 the first step in deriving the likelihood expansion , if @xmath89 can be expanded as a taylor series in the components of the observed error @xmath90 , consists of writing @xmath89 as where @xmath92 for @xmath93 are the components of the observed error . we will use the notation @xmath94 \nonumber \end{aligned}\ ] ] where @xmath95 and @xmath12 is the information matrix @xmath13=e[l_{a}l_{b}]$ ] . introducing @xmath96 as the inverse of the matrix whose elements are given by @xmath97 , equation ( [ cippo ] ) can be rearranged to solve for @xmath98 finally we can iterate equation ( [ 2 ] ) with respect to the components of the observed error and expand @xmath96 in terms of the information matrix inverse @xmath100 ( @xcite , page 149 ) @xmath101_{ab}&=&j^{ab}=\label{barnone}[(1-i^{-1}(i - j))^{-1}i^{-1}]^{ab}=\nonumber \\ & = & { i^{ab}}+{i^{at}i^{bu}h_{tu}}+{i^{at}i^{bu}i^{vw}h_{tv}h_{uw}}+\nonumber\\ & + & { i^{at}i^{bu}i^{vw}i^{cd}h_{tv}h_{wc}h_{du}}+ ... \,\,\,\,\,\,\,\,\,\end{aligned}\ ] ] from ( [ 2 ] ) and ( [ barnone ] ) the terms that contribute to each @xmath83 order of the observed error can be obtained . however , in order to isolate the @xmath83 orders necessary to compute the second order bias we have chosen only a finite number of terms within equations ( [ 2 ] ) and ( [ barnone ] ) . this choice can be made by @xmath102 observing that @xmath103 is proportional to @xmath82 @xmath104 = n \ , e[\frac{\partial^s \,ln ( p(\underline{x}_i,\underline{\vartheta } ) ) } { \partial\vartheta_{a_1}\partial\vartheta_{a_2} .. \partial\vartheta_{a_s } } ] \end{aligned}\ ] ] where the value of @xmath105 is irrelevant because all the vector data have the same distribution , and @xmath106 using the large sample approximation expressed by ( @xcite , page 221 ) @xmath107\sim n^{\frac{1}{2}},\end{aligned}\ ] ] and proved for discrete and continuous random variables in the next paragraph . equation ( [ 5 ] ) indicates that the expectation of a product of @xmath108 generic factors @xmath109 is a polynomial of integer powers of @xmath82 , where the highest power of @xmath82 is the largest integer less than or equal to @xmath110 ( we use the notation @xmath111 ) . for example , @xmath112 is proportional to @xmath82 . + the proof of the large sample approximation ( [ 5 ] ) is an extension to continuous random variables of the analysis performed in ( @xcite , page 36 ) , for discrete random variables . + to prove equation ( [ 5 ] ) we show that the quantities @xmath113\end{aligned}\ ] ] obtained as the expectation of products of @xmath114 functions @xmath115 are polynomials in integer powers of @xmath82 of order less than or equal to @xmath116 . + here , @xmath117 . the subscripts @xmath118 that appear in @xmath119 represent collections of indexes as introduced in equation ( [ 5 ] ) . + the factors @xmath120 appearing in the expectation can be seen as the first @xmath114 terms of a succession @xmath119 , where we choose arbitrary @xmath119 for @xmath121 . using such a succession we can define the quantity @xmath122\end{aligned}\ ] ] where we assume that all the expectations exist . where the quantity @xmath124 can be equal to 0 or to 1 to indicate that the quantity @xmath125 is present or not in the product inside the expectation . the summation over all @xmath126 is intended to account for all possible configurations of the indexes ( choosing some equal to zero and the others equal to one ) with the constraint that the number of factors within the expectation is equal to @xmath127 . we can now group together terms @xmath128 for different @xmath105 and define @xmath129 , where @xmath130 . in this manner we can also factorize the expectations in ( [ 100 ] ) as expectations for different data vectors . by making use of the statistical independence of the different data vectors , and defining @xmath131 $ ] , ( [ 100 ] ) can be rewritten as follows : we observe that , when the expectations contain only one factor , for @xmath133 , we have + @xmath134=e[h_{1,\underline{\alpha}_{j}}]=\lambda(a_{1,\underline{a}_j})=0 $ ] for any @xmath135 . as a consequence , the taylor expansion of @xmath136 in @xmath137 can be written as @xmath138 . + eventually , it is the absence of @xmath139 in this expansion that allows us to explain the properties of the polynomials in equation ( [ esempio ] ) and finally the large sample approximation ( [ 5 ] ) . to accomplish this last step we explain how the polynomials @xmath140 are related to the coefficients @xmath141 . let us consider the contribution to @xmath140 that originates from the product of @xmath142 factors @xmath143 with the constraints @xmath144 and @xmath145 . the number of ways these products can be formed for an assigned set of @xmath146 is proportional to @xmath147 . moreover , the contributions to @xmath140 are formed by an integer number of factors less than or equal to @xmath116 because @xmath148 . this property limits the highest power in @xmath82 contained in @xmath140 with the largest integer number smaller than or equal to @xmath116 . equation ( [ esempio ] ) is then proved as is the large sample approximation ( [ 5 ] ) . + each @xmath83 order @xmath114 of the @xmath149 component of the observed error is denoted by @xmath150 $ ] where the index is given by @xmath151 and @xmath127 is a natural number . for example , the @xmath83 order for @xmath152 is given by @xmath153=\frac{1}{2}\upsilon^{rst}l_{s}l_{t}+i^{rs}i^{tu}h_{st}l_{u } \end{aligned}\ ] ] where we have adopted the lifted indexes notation @xmath154 . the @xmath83 orders of the bias are then given as the expectation of the @xmath83 orders of the observed error @xmath155=e [ ( \widehat{\vartheta } -\vartheta ) ^r[m]].\ ] ] the @xmath83 order @xmath156 $ ] contains different powers of @xmath84 as we discuss in this paragraph . + it follows from equations ( [ 2 ] ) to ( [ 5 ] ) that @xmath156 $ ] is the sum of terms having the structure + @xmath157 $ ] where @xmath158,@xmath159 , and @xmath108 are the numbers of factors in the three products satisfying @xmath160 . different powers of @xmath84 can be generated because @xmath161 $ ] can contain all the integer powers of @xmath82 equal to or less than @xmath162 . however from the fact that no power higher than @xmath162 can be generated follows that an asymptotic order @xmath114 will never generate powers in the sample size @xmath163 with @xmath164 smaller than @xmath114 . it still remains to prove which is the largest power in @xmath84 contained in the asymptotic order @xmath114 . we show that in the following . since the largest range of powers of @xmath84 is allowed when @xmath165 , we study these terms to evaluate the range of powers of @xmath84 contained in @xmath156 $ ] . the structure of equation ( [ 2 ] ) implies that its iteration with respect to the observed error components generates an equal number of @xmath96 and @xmath166 ( we recall that @xmath167 ) . similarly , the number of @xmath168 generated from the expansion of @xmath96 given in equation ( [ barnone ] ) is equal to the number of @xmath169 plus 1 . these two observations imply that the terms where @xmath165 also verify @xmath170 . as a consequence , the powers of @xmath84 that @xmath156 $ ] contains can range from @xmath171 , up to @xmath172 . + the analysis above implies that in order to compute the contributions to @xmath84 and to @xmath173 in the bias expansion that we denote with @xmath174 $ ] and @xmath175 $ ] , it is necessary to obtain the first three @xmath83 orders of the bias , @xmath176 $ ] , @xmath177 $ ] , and @xmath178 $ ] . in the explicit expressions below , to condense the notation , we introduce the quantities @xmath179 , so that @xmath180&=&\label{11}i^{ri , kl}\upsilon _ { ik , l}+\frac{1}{2}i^{rj , lp}\upsilon _ { jlp}\\ { \tilde{b } } [ \widehat{\vartheta } ^r , \frac{3}{2}]&=&\frac{1}{6}(i^{r\alpha , s\beta , t\gamma , u\delta } \upsilon _ { \alpha\beta\gamma\delta}+\label{12}3i^{r\alpha , s\beta , \gamma\delta , tg , uv } \upsilon _ { \alpha\beta\gamma}\upsilon _ { \delta , g , v } \upsilon _ { s , t , u}+ i^{r\alpha , s\beta , u\gamma , tv } \upsilon _ { \alpha\beta\gamma}\upsilon _ { uv , s , t}+\\ & + & \frac{1}{2}i^{rs , t\alpha , u\beta , v\gamma}\upsilon _ { \alpha\beta\gamma}\upsilon _ { st , u , v } + + \frac{1}{2}i^{rs , tu , vw}\upsilon _ { stv , u , w } + i^{rs , tu , vw}\upsilon _ { st , vu , w } + 2i^{rs , tw}\upsilon _ { st , w } \nonumber\\ & + & i^{r\alpha,\beta\gamma}\upsilon _ { \alpha\beta\gamma}\nonumber\\ { \tilde{b } } [ \widehat{\vartheta } ^r , 2]&=&\label{13 } ( i^{ra , bq , cd , tp}+ \frac{1}{2}i^{rd , ta , bq , cp } ) \upsilon_{\{abc\}\{dt\}\{q\}\{p\}}+\frac{1}{2 } ( i^{ra , bs}\upsilon ^{cpq } + i^{bp , cq}\upsilon ^{rsa } ) \upsilon_{\{abc\ } \{s\}\{p\}\{q\ } } \nonumber \\ & + & ( \frac{1}{2}i^{ap , tq}\upsilon ^{rbg}+i^{ra , tq}\upsilon ^{bpg}+ \frac{1}{2}i^{ra , bg}\upsilon ^{tpq}+i^{bg , tq}\upsilon ^{rpa})\upsilon_{\{ab\}\{gt\}\{p\}\{q\}}+\frac{1}{2 } [ \upsilon ^{rsz}\upsilon ^{pqt } \nonumber\\ & + & \upsilon ^{rzqs}i^{tp}+\frac{1}{3}\upsilon ^{pzqt}i^{rs}+(i^{dz , eq , pt}\upsilon ^{rcs } + i^{rp , dz , et}\upsilon ^{sqc}+2i^{dq , es , pt}\upsilon ^{rzc})\upsilon_{cde } ] \upsilon_{\{sp\}\{z\}\{q\}\{t\ } } \nonumber \\ & + & i^{ra , st , bc , de}\upsilon_{\{ab\}\{cd\}\{et\}\{s\}}+\frac{1}{6}i^{rj , kl , mp , qz}\upsilon_{\{jlpz\}\{k\}\{m\}\{q\}}+[\frac{1}{24}\upsilon^{r \alpha \beta \gamma \delta}+\nonumber\\ & + & \frac{1}{6}\upsilon^{r \alpha a } i^{\beta b , g\delta,\gamma d } \upsilon _ { abgd}+\frac{1}{4}\upsilon^{r a \gamma \delta } i^{\beta c,,b \alpha } \upsilon_{a b c}+ \frac{1}{8 } \upsilon^{r v a } i^{w \alpha , z \beta , b \gamma , g \delta } \upsilon_{v w z}\upsilon_{abg}+\nonumber\\ & + & \frac{1}{2}\upsilon^{r \alpha v } i^{w\beta } \upsilon_{v w z}\upsilon^{z \gamma \delta}]\upsilon_{\{\alpha\}\{\beta\}\{\gamma\}\{\delta\}}. \label{biastworough}\end{aligned}\ ] ] the first order of the bias @xmath174 $ ] can than be obtained by substituting the explicit expressions for the tensors in @xmath176 $ ] . the second order @xmath175 $ ] takes contributions from @xmath177 $ ] and @xmath178 $ ] . however , while @xmath181 $ ] generates only @xmath182 contributions , @xmath183 $ ] generates @xmath182 and @xmath184 contributions . consequently , to collect all and only the contributions to @xmath182 , it is necessary to extract the @xmath182 component from @xmath178 $ ] and add it to @xmath177 $ ] . the extraction can be done by introducing into @xmath178 $ ] only the highest powers of @xmath82 of the tensors . starting from ( [ biastwo ] ) the form for the second order covariance matrix can be obtained using the following theorem : + theorem : _ let @xmath185 be an invertible and differentiable function and let @xmath186 be a set of parameters dependent on @xmath7 . let @xmath187 and @xmath188 be the biases relative to the estimation of the components of @xmath7 and @xmath189 which can be computed as power series of @xmath84 , as explained in section 2 . the terms of the expansion in powers of @xmath84 for @xmath188 can be obtained from the terms of the expansion for @xmath187 by replacing the derivatives with respect to the components of @xmath7 , with derivatives with respect to the components of @xmath189 . the explicit dependence on @xmath189 can be removed at the end of a derivation by means of the jacobian matrix j defined by @xmath190 . _ + to prove the chain - rule theorem given in section [ expansions ] , we analyze how the derivation of the expansion in powers of @xmath84 for @xmath191 , which is described there , changes if we are interested in the expansion for @xmath192 where @xmath193 is a vector of parameters in an invertible relationship with @xmath194 : @xmath195 . the starting point for the derivation of the expansion of @xmath192 is the stationarity condition @xmath196 that can be obtained from equation ( [ stazionarieta ] ) by replacing only the derivatives @xmath197 with @xmath198 , because @xmath199 implies @xmath200 . the subsequent steps can then be obtained from equations ( [ cippo ] ) up to ( [ biastwo ] ) by replacing the derivatives in the same way . since @xmath201 , where the components of the jacobian matrix @xmath202 behave like constants in the expectations , the substitution of the derivatives can also be done only in the final result ( for example , in the orders given in equations ( [ biastworough ] ) and ( [ biastwo ] ) ) . the expectations contained in the expansion of @xmath203 in powers of @xmath84 can also be computed before the substitution of the derivatives if the likelihood function dependence on the parameters is expressed in terms of auxiliary functions . examples of auxiliary functions are @xmath204 for a general parametric dependence and @xmath205 @xmath206 for a scalar gaussian distribution . by means of these auxiliary functions , the derivatives @xmath207 and @xmath208 become @xmath209 and @xmath210 . as a consequence the orders of the expansion for @xmath211 can be found from the orders for the expansion of @xmath203 implementing , in the result of the expectations , the substitutions @xmath212 the converse of the chain - rule , in which higher moments of the observed error are used to compute lower moments , is not possible . we can observe , for example , that the expansion of a general moment of an mle does not always contain all the powers of @xmath84 . the lowest order present in the expansion of the m order moment is given by the largest integer number smaller than or equal to @xmath116 ( we use the notation @xmath213 ) : @xmath214= \frac{1}{n^{int(\frac{m}{2})}}+higher\,powers\ , o\!f\ , \frac{1}{n}.\end{aligned}\ ] ] the consequence of this observation is that only the bias and the error correlation matrix may contain first order terms . for this reason an inverse chain - rule would have to generate non - zero orders of lower moments expansions from the corresponding orders of the higher moments expansions that are zero for powers of @xmath84 lower than @xmath215 . let us consider how the chain - rule makes it possible to compute the expansion of the error correlation matrix @xmath216 defined by @xmath217 $ ] and the covariance matrix . using the invariance property of the mle , @xmath218 where @xmath219 is the transpose of a vector and @xmath220 is a matrix whose components are the bias of the product of two components of @xmath7 . once @xmath221 and the bias are known , then the covariance matrix @xmath222 can also be computed by means of @xmath223 to compute the right hand side of equation ( [ 74 ] ) we express it in terms of the components , obtaining @xmath224 it is important to realize that knowledge of @xmath225 is sufficient because the expansion of @xmath226 can be derived from it by means of the theorem given above . in fact , if we choose @xmath227 as a new set of parameters , @xmath226 becomes @xmath228 . however , is necessary to insure that the relationship between @xmath7 and @xmath189 is invertible . this condition holds if both @xmath229 where @xmath230 and the sign of the components of @xmath7 is known , for example , by means of constraints on the data model . + in a scalar estimation scenario , equation ( [ 74 ] ) becomes simply @xmath231=b(\label{61}\widehat{{\vartheta } ^2})-2\vartheta b(\widehat{{\vartheta } } ) \end{aligned}\ ] ] and the variance @xmath232 . in this case @xmath233 can be derived from @xmath234 if we choose @xmath235 as the new parameter and the jacobian matrix becomes @xmath236 because @xmath237 . + a useful simplification of the algebra of the chain - rule in the derivation of second order moments is described in the following two paragraphs . the chain - rule and the subsequent conversion of the derivatives require the substitutions @xmath238 } { \partial \vartheta_{i_1 } .. \partial \vartheta_{i_m } } \rightarrow \frac { \partial^m [ . ] } { \partial \xi_{i_1 } .. \partial \xi_{i_m } } \rightarrow j_{i_1 j_1}(\underline{\vartheta } ) \frac { \partial } { \partial \label{convder } \vartheta_{j_1 } } j_{i_2 j_2}(\underline{\vartheta } ) \frac { \partial } { \partial \vartheta_{j_2 } } .. j_{i_m j_m}(\underline{\vartheta } ) \frac { \partial } { \partial \vartheta_{j_m } } [ .]\end{aligned}\ ] ] from the right hand side of ( [ convder ] ) it is clear that the derivatives will also be applied to the jacobian matrix , thereby generating @xmath239 terms for every derivative of order @xmath114 . however , it can be observed that the terms contributing to the final result are only a small subset . in fact , among all the terms generated from the conversion of the derivatives in the bias expansion , only those where a first derivative of the jacobian matrix appears must be considered . for example , we can consider the term @xmath240 that comes from equation ( [ biastwo ] ) , which must be used to derive the second order @xmath216 . in this case , we need to consider only the 3 terms in which one of the derivatives represented by @xmath241 operates on a jacobian matrix and @xmath242 operate on the likelihood function , plus the 3 terms where the role of @xmath243 and @xmath244 are inverted . in general , the total number of relevant terms among those generated in every derivative is equal to or less than the order of the derivative @xmath114 . + the detailed analysis of equation ( [ 21 ] ) reveals that the terms generated in @xmath245 can be divided into three groups : ( a ) the terms where no derivative of the jacobian matrix appears are equal to @xmath246 ( we show in example 1 that @xmath246 cancels with @xmath247 after its introduction in equation ( [ 21 ] ) ) ; ( b ) the terms where only one derivative of the jacobian matrix appears give the error correlation matrix ; and ( c ) the terms that contain higher derivatives of the jacobian matrix or more than one first derivative of the jacobian matrix in the same term , summed together , give zero . + + to clarify the use of the chain - rule and the algebraic issues discussed above , we present two examples . in example 1 we use the first order term of the bias in a general vector estimate to derive the first order term of @xmath216 . it is useful to recall that the expansion of @xmath216 and the expansion of @xmath222 can be different only from the second order on , so this example also describes a derivation of the first order of the covariance matrix @xmath222 . following the same approach , the second order term of the error correlation matrix expansion can be derived from equation ( [ biastwo ] ) and the second order covariance matrix can also be derived if we also use ( [ biasone ] ) and ( [ conversion ] ) . + in example 2 we illustrate the way the chain - rule can still be used , if the available expression of the bias expansion is explicitly computed for a particular data model . in particular , we derive the second order mean square error and the second order variance from the second order bias in two scalar gaussian models . in example 2 we also illustrate the simplification introduced above for the algebra involved in the conversion of the derivatives . + + example 1 + using the bartlett identities ( @xcite,@xcite ) , we rewrite the first order bias , given by equation ( @xmath248 ) , as @xmath249=-\frac{1}{2}i^{rj}i^{lp}\label{trichecone}(\upsilon _ { j , l , p}+\upsilon _ { j , lp})\end{aligned}\ ] ] from equation ( [ trichecone ] ) , @xmath250 can be computed by means of the chain - rule by replacing the derivatives with respect to the components of @xmath7 with derivatives with respect to the components of @xmath189 , where @xmath189 is given in equation ( [ variab ] ) and using the corresponding jacobian matrix . the tensors in equation ( [ trichecone ] ) become @xmath251=j_{lr}i_{rs}(\vartheta)(j^t)_{sp}=\\ & = & j_{lr}e [ \frac{\partial^2 l } { \partial \vartheta_i \partial \vartheta_j}]\label{tricheco2}(j^t)_{sp}\\ { i^{lp}}(\xi)&=&{((j^{-1})^t)_{lr}}{i^{rs}}(\vartheta)(j^{-1})_{sp}\\ \upsilon _ { j , l , p}\label{tricheco3}(\xi)&=&j_{j\alpha}j_{l\beta}j_{p\gamma}\upsilon _ { \alpha,\beta,\gamma}(\vartheta)\\ \upsilon _ { j , lp}(\xi)&=&\label{tricheco4}j_{j\alpha}j_{l\beta}(j_{p\gamma}\upsilon _ { \alpha,\beta,\gamma}(\vartheta)+ \frac{\partial j_{p \gamma}}{\partial \beta } \upsilon _ { \alpha,\gamma}(\vartheta))\end{aligned}\ ] ] where we have specified in the bracket beside the tensors the dependence on the parameter sets . inserting these expressions in equations ( [ trichecone ] ) and observing that equation ( [ 21 ] ) can be expressed in the form @xmath252 the first order term of the error correlation matrix can then be obtained as @xmath253=\frac{1}{2 } i^{\alpha \beta}\frac{\partial ( j^{-1})_{\alpha i}}{\partial \vartheta_{\beta}}=\frac{1}{2 } i^{\alpha \beta}\gamma_{\alpha \beta}=i^{ij}\end{aligned}\ ] ] where we have introduced the tensor @xmath254 + example 2 + in this example we determine the second order variance and mean square error from the second order bias for two cases of parameter dependence for the scalar gaussian density @xmath255 as a direct application of equation ( [ simplechain ] ) . + in the case where c does not depend on the parameters ( @xmath256 ) the second order bias can be derived using the scalar version of equation ( [ biastwo ] ) and of the tensors @xmath257 defined in equation ( [ upsilondef ] ) . for this parameter dependence , the @xmath83 order for @xmath258 ( equation ( [ 12 ] ) ) is zero , and the second order bias can be directly obtained also from equation ( [ 13 ] ) the result is @xmath259=\frac{c^2}{(\dot{\mu})^8}[\frac{5}{4}\dddot{\mu}\ddot{\mu}(\dot{\mu})^2-\frac{1}{8}\ddddot{\mu}(\dot{\mu})^3-\frac{15}{8}\dot{\mu}(\ddot{\mu})^3]$ ] . applying the chain - rule , the second order mean square error for the estimation of @xmath260 becomes @xmath261=\frac{15{\ddot{\mu}}^2 } { 4{\dot{\mu}}^6 } -\frac{\dddot{\mu}}{{\dot{\mu}}^5 } $ ] , where the full conversions of the derivatives are given by @xmath262 . by means of these conversions and the expression of the second order bias , it can be observed that among the 18 terms that are in principle generated by the chain - rule , only 6 contribute to the second order mean square error . + + in the following we show that the expansion in the inverse of the sample size is equivalent to an expansion in @xmath263 . the derivation of the asymptotic orders in @xmath84 would be the same for an expansion in any quantity if ( [ upsilondef ] ) and ( [ 5 ] ) can be derived for a certain quantity @xmath61 instead of the sample size n. in this section we illustrate indeed that this is the case where the signal to noise ratio for a set of scalar data distributed according to a deterministic gaussian pdf takes the role of the sample size . the probability distribution and parameter dependent part of the likelihood function are given by : @xmath264 we also define the signal to noise ratio for this example as @xmath265 following standard practice for scalar deterministic gaussian data . we can obtain : from ( [ gaussian_expect_la ] ) it becomes obvious that @xmath268 is proportional to not only the sample size , @xmath82 , but also to the @xmath61 . the term inside the square brackets is simply a sum over normalized derivatives of the mean and contains information about the shape of the signal . in the above equations we use : @xmath271 & = & 2^p \mu_{\alpha_1 } \ldots \mu_{\alpha_p } \frac{1}{\sigma^{2p } } e[\sum_{i = 1}^{n}(x_i - \mu(\vartheta))^p ] \nonumber \\ & = & 2^p ( \frac{\mu_{\alpha_1}}{\mu } ) \ldots ( \frac{\mu_{\alpha_p}}{\mu } ) \frac{\mu^p}{\sigma^{2p } } \sigma^p \frac{(2\frac{p}{2})!}{2^{\frac{p}{2}}(\frac{p}{2})!}n^{\frac{p}{2 } } \nonumber \\ & = & ( \frac{\mu^2}{\sigma^2})^{\frac{p}{2 } } \ldots ( \frac{\mu_{\alpha_1}}{\mu } ) \ldots ( \frac{\mu_{\alpha_p}}{\mu})n^{\frac{p}{2 } } \nonumber \end{aligned}\ ] ] note that for deterministic gaussian data , the non - integer asymptotic orders of the bias are zero and the integer orders are equal to @xmath274 etc . this is sufficient to prove that the orders in @xmath275 of the bias expansion are also orders in @xmath276 . this also holds for the variance expansion . a similar , although longer proof can be written for the snr definition provided in [ snrtoa ] . abbott et al . , rep . phys * 72 * , 076901 ( 2009 ) h. grote , class . quantum grav . * 25 * , 114043 ( 2008 ) f. acernese et al . , class . quantum grav . * 23 * , 563 ( 2006 ) v. kalogera , k. belczynski , c. kim richard w. oshaughnessy b. willems , phys . rept . * 442 * , 75 - 108 ( 2006 ) abramowitz and stegun _ handbook of mathematical functions _ dover ( 1965 ) o.e . barndoff - nielsen and d.r . cox , int . , * 52 * , 309 - 326 ( 1984 ) o.e . barndoff - nielsen and d.r . asymptotic techniques for use in statistics _ , chapman - hall , london ( 1989 ) o.e . barndoff - nielsen and d.r . cox_inference and asymptotic _ , chapman - hall , london , ( 1994 ) o.e . barndoff - nielsen , d.r.cox and n. reid , _ the role of differential geometry in statistics _ ( 1986 ) o.e . barndoff - nielsen and p.a . mc cullagh , biometrika , 80 . 321 - 8 ( 1993 ) o.e . barndoff - nielsen , w.s . kendall , m.n.m . lieshout , _ stocastic geometry : likelihood and computation _ boca raton , fla . : chapman and hall / crc , c ( 1999 ) m.s . bartlett , biometrika , 40 , 12 - 19.(1953 ) m.s . bartlett , biometrika , 40 , 306 - 317.(1953 ) k.o.bowman and l.r.shenton , commun . theory meth . 27 ( 11 ) 2743 - 2760 ( 1998 ) k.o.bowman , l.r.shenton , ann.inst.statist . 44 , no 781 - 798 ( 1992 ) k.o . bowman , l.r . shenton , commun stat - simul c : ( 6 ) 697 - 710 ( 1983 ) haldane and smith , biometrika , 43 , 96 - 103 ( 1956 ) p.j . hanlon , r. stanley , j.r . stembridge , contemporary mathematics , vol 138 , ( 1992 ) d.n lawley , biometrika , 43 , 295 - 303 , ( 1956 ) p. mccullagh , biometrika , 71 , 461 - 76.(1984 ) p. mccullagh , _ tensor methods in statistics _ , chapman and hall ( 1987 ) p. mccullagh , d.r . cox , ann stat 14 , 1419 - 1430 ( 1986 ) pierce d.a . and peters d. , biometrika , 81 , 1 , pages . 1 - 10 ( 1994 ) rao c. , _ linear statistical inference and its application _ , wiley , new york , ( 1966 ) riordan j. , _ combinatorial identities _ , wiley ( 1968 ) shenton l.r . , k.o.bowman , _ higher moments of a maximum - likelihood estimate _ j.roy.stat.soc . ( 1963 ) shenton l.r . , k.o.bowman , _ maximum likelihood estimation in small samples _ ( 1977 ) shenton l.r . , k.o.bowman,_the dark side of asymptotics and moment series _ , sankhya , ( 1986 ) shenton l.r . , k.o.bowman,_the approximate distribution of four moment statistics from type iii distributions_,commun . theory meth.1511 - 1579 ( 1990 ) shenton l.r . , k.o.bowman,_asymptotic kurtotis for maximum likelihood estimators _ , commun . theory meth . 28 ( 11 ) 2641 - 2754 ( 1999 ) shenton l.r . , bowman k.o . and sheehan , _ sampling moments associated with univariate distributions _ stanley r. , _ enumerative combinatorics _ , cambridge press,(1993 ) m. vallisneri , arxiv : gr - qc/0703086 ( 2007 ) m. vallisneri , phys . d 77 , 042001 ( 2008 ) c. cutler , and e. flanagan phys . d 49,6 2658 ( 1994 ) l.s.finn phys . d 49,6 2658 ( 1992 ) a.thode , m. zanolin , e. naftali , i. ingram , p. ratilal and n.c . makris , j. acoust . * 112 * ( 5 ) ( 2001 ) m. zanolin , i. ingram , a.thode and n.c . makris , j. acoust . * 116 * ( 4 ) ( 2004 ) m. zanonlin , n. makris , mit tech . report ( 2001 ) e. naftali and n.c . makris , j. acoust . * 110 * ( 4 ) ( 2001 ) waterman r.p . , lindsay b.g . _ a simple and accurate method for approximate conditional inference applied to exponential family models _ j.r.statist . soc.b 58,no 1 , pages 177 - 188 ( 1996 ) wlodzinierz bryc,_the normal distribution : characterization with applications_,springer verlag ( 1995 ) . l. blanchet , g. faye , b.r . iyer and b. joguet , phys . d * 65 * , 061501(r ) ( 2002 ) t. damour , b.r . iyer , and b.s . sathyaprakash , phys . d * 63 * , 044023 ( 2001 ) t. damour , b.r . iyer , and b.s . sathyaprakash , phys . d * 66 * , 027502 ( 2002 ) k.g arun , b. r. iyer , b.s sathyaprakash and p.a . sundararajan , phys . d * 71 * , 084008 - 1 ( 2005 ) k.g arun , b. r. iyer , b.s sathyaprakash and p.a . sundararajan , phys . d * 72 * , 069903 ( e ) ( 2005 ) r. balasubramanian , b.s . sathyaprakash and s.v . dhurandhar , phys . d * 53 * , 3033 ( 1996 ) r. balasubramanian and s.v . dhurandhar , phys . d * 57 * , 3408 ( 1998 ) l.s . finn and d.f . chernoff , phys . d * 47*,2198 ( 1993 ) p. jaranowski and a. krolak , phys . d * 47 * 1723 ( 1994 ) d. nicholson , a. vecchio , phys . d * 57 * , 4588 ( 1998 ) c. cutler , e. e. flanagan , phys . d * 49 * , 2658 ( 1994 ) a.willsky , j.shapiro ee6432 mit technical note , sept . 1999 e. poisson , c.m . will , phys . d * 52 * , 848 ( 1995 ) a. krlak , k.d . kokkotas and g. schfer , s.fairhrst new j.phys . 11 123006 ( 2009 ) d * 52 * , 2089 ( 1995 ) p. ajith and s. bose , arxiv 0901.4936v4 ( 2009 ) t. cokelaer , class . quantum grav . * 25 * , 184007 ( 2008 )

we give a definition of ` coherent tangent bundles ' , which is an intrinsic formulation of wave fronts . in our application of coherent tangent bundles for wave fronts , the first fundamental forms and the third fundamental forms are considered as induced metrics of certain homomorphisms between vector bundles . they satisfy the completely same conditions , and so can reverse roles with each other . for a given wave front of a-manifold , there are two gauss - bonnet formulas . by exchanging the roles of the fundamental forms , we get two new additional gauss - bonnet formulas for the third fundamental form . surprisingly , these are different from those for the first fundamental form , and using these four formulas , we get several new results on the topology and geometry of wave fronts .

0910.3456

in this paper , we give a definition of _ coherent tangent bundles _ , which is an intrinsic formulation of wave fronts . the advantage of this formulation is that the first fundamental forms and the third fundamental forms satisfy exactly the same conditions for wave fronts in space forms , and they can reverse roles with each other . then the two gauss - bonnet formulas induce two more gauss - bonnet formulas by exchanging the first and third fundamental forms . these turn out to be different from those for the first fundamental form , and using these four formulas , we get new results on fronts . in section [ sec : coherent ] , we generalize the definitions of _ coherent tangent bundles _ and _ singular curvature _ , which were given for @xmath0-dimensional coherent tangent bundles ( or fronts ) in @xcite and @xcite , to fronts of general dimension . a coherent tangent bundle on an @xmath1-manifold @xmath2 induces a positive semi - definite metric on @xmath2 ( called the first fundamental form ) , and can be regarded as a generalization of riemannian manifolds . using the concept of coherent tangent bundles , we can give a unified treatment of hypersurfaces and @xmath3-maps between the same dimensional manifolds . the points where the metric degenerates are called _ singular points_. we define a notion of @xmath4 ( singular ) points , which is a generalization of that of fronts and of @xmath3-morin maps at the same time . we define _ singular principal curvatures _ for each @xmath5-point of a given coherent tangent bundle , which are @xmath6-tuples of real numbers . one of them diverges to @xmath7 at @xmath8-points as shown in theorem [ thm : infty ] . this is a generalization of the result in @xcite for @xmath0-dimensional fronts . moreover , as shown in section [ sec : gauss - bonnet ] , our intrinsic setting enables us to introduce the singular curvature on @xmath5-morin singular points . when @xmath9 , as an application of the two intrinsic gauss - bonnet formulas , we give a new proof of quine s formula ( cf . proposition [ prop : quine ] ) , and also get a new formula ( cf . proposition [ prop : id ] ) for the total singular curvature of generic smooth maps between @xmath0-manifolds . furthermore , in section [ sec : gauss - bonnet ] , we also give several results on compact @xmath0-dimensional fronts , as an application of our new gauss - bonnet formulas for the third fundamental form . in particular , we show that the total negative gaussian curvature @xmath10 of a closed immersed surface @xmath11 in @xmath12 is equal to the signed sum of total geodesic curvature on each connected component of the singular set of its gauss map , see theorem [ thm : c ] . the deepest applications are given for surfaces of bounded gaussian curvature . for example , the euler characteristic of a closed wave front with gaussian curvature @xmath13 ( @xmath14 ) in a @xmath15-dimensional flat torus vanishes ( cf . theorem [ thm : g ] ) . let @xmath2 be an oriented @xmath1-manifold ( @xmath16 ) and @xmath17 a vector bundle of rank @xmath1 over @xmath2 . before defining coherent tangent bundles , we define @xmath4-points for each vector bundle homomorphism from the tangent bundle of @xmath18to @xmath17 . we assume that @xmath17 is orientable , namely there exists a non - vanishing section @xmath19 , where @xmath20 is the determinant line bundle of the dual bundle @xmath21 . this assumption does not create any restriction for the local theory . in fact , for each point @xmath22 , there exists a section @xmath23 of @xmath24 defined on a neighborhood @xmath25 of @xmath26 . now we fix a bundle homomorphism @xmath27 where @xmath28 is the tangent bundle of @xmath2 . a point @xmath22 is called a _ @xmath29-singular point _ if @xmath30 is not a bijection , where @xmath31 is the fiber of @xmath17 at @xmath26 . we denote by @xmath32 the set of @xmath29-singular points on @xmath2 . on the other hand , a point @xmath33 is called a _ @xmath29-regular point_. we now fix a @xmath29-singular point @xmath26 , and take a positively oriented local coordinate neighborhood @xmath34 of @xmath26 . then it holds that @xmath35 where @xmath36 is a @xmath3-function . we call the function @xmath37 the _ @xmath38-jacobian function_. the set of @xmath29-singular points on @xmath25 is expressed as @xmath39 we now set @xmath40 since the sign of @xmath41 does not depend on the choice of the positively oriented local coordinate system , the above definitions of @xmath42 make sense . a @xmath29-singular point @xmath26 @xmath43 is called _ non - degenerate _ if @xmath44 does not vanish at @xmath26 . on a neighborhood of a non - degenerate @xmath29-singular point , the @xmath29-singular set consists of an @xmath6-submanifold in @xmath2 , called the _ @xmath29-singular submanifold_. if @xmath26 is a non - degenerate @xmath29-singular point , the rank of @xmath45 is @xmath46 . the direction of the kernel of @xmath45 is called the _ null direction_. let @xmath47 be _ a null vector field _ , namely , smooth ( non - vanishing ) vector field along the @xmath29-singular submanifold @xmath48 , which gives the null direction at each point in @xmath48 . [ def : a2-point ] a non - degenerate @xmath29-singular point @xmath22 is called an _ @xmath5-point _ , or an _ @xmath5-point of @xmath29 _ , if the null direction @xmath49 is transversal to the @xmath38-singular submanifold . we set @xmath50 where @xmath51 is a vector field on a neighborhood @xmath25 of @xmath26 which gives a null vector field on @xmath52 . then @xmath26 is an @xmath5-point if and only if the function @xmath53 does not vanish at @xmath26 ( see ( * ? ? ? * theorem 2.4 ) ) . [ def : a3-point ] a non - degenerate @xmath38-singular point @xmath22 is called an _ @xmath8-point _ or an _ @xmath8-point of @xmath38 _ if it is not an @xmath5-point of @xmath38 , but @xmath54 does not vanish . similarly , we can define @xmath55-points inductively . we set @xmath56 if @xmath57 on @xmath58 , then we can define a subset of the submanifold @xmath59 by @xmath60 where we set @xmath61 and @xmath62 . if a non - degenerate @xmath38-singular point @xmath26 is not an @xmath4-point of @xmath38 , but @xmath63 is well - defined inductively and is nonvanishing at @xmath26 , then it is called an _ @xmath55-point of _ @xmath38 . here an @xmath64-point means a @xmath38-regular point . ( see [ 17 , section 2 ] for details . ) as a generalization of the @xmath0-dimensional case in @xcite and @xcite , we give a general setting for intrinsic fronts : let @xmath2 be an oriented @xmath1-manifold ( @xmath16 ) . a _ coherent tangent bundle _ over @xmath2 is a @xmath65-tuple @xmath66 , where @xmath17 is a vector bundle of rank @xmath1 over @xmath2 with an inner product @xmath67 , @xmath68 is a metric connection on @xmath69 , @xmath70 is a bundle homomorphism which satisfies @xmath71)=0\ ] ] for vector fields @xmath72 and @xmath73 on @xmath2 . in this setting , the pull - back of the metric @xmath74 is called the _ @xmath29-metric _ , which is a positive semidefinite symmetric tensor on @xmath2 . it should be remarked that as in , @xmath75)\ ] ] gives a skew - symmetric tensor on @xmath2 , called the _ torsion tensor _ of @xmath68 with respect to @xmath29 . if @xmath76 is a coherent tangent bundle , then the pull back of the connection @xmath68 by @xmath29 coincides with the levi - civita connection . in this sense , a coherent tangent bundle can be considered as a generalization of riemannian manifold . in fact , as an application , a duality of conformally flat manifolds with admissible singularities is given in terms of coherent tangent bundles in @xcite . a coherent tangent bundle @xmath76 is called _ co - orientable _ if the vector bundle @xmath17 is orientable , namely , there exists a globally defined smooth section @xmath23 of @xmath20 such that @xmath77 for any orthonormal frame @xmath78 on @xmath17 . the form @xmath23 is determined uniquely up to a @xmath79-ambiguity . co - orientation _ of the coherent tangent bundle @xmath17 is a choice of @xmath23 . an orthonormal frame @xmath78 is called _ positive _ with respect to the co - orientation @xmath23 if @xmath80 . we give here typical examples of coherent tangent bundles : [ ex : map ] let @xmath2 be an oriented @xmath1-manifold and @xmath81 an oriented riemannian @xmath1-manifold . a @xmath3-map @xmath82 induces a coherent tangent bundle over @xmath2 as follows : let @xmath83 be the pull - back of the tangent bundle @xmath84 by @xmath85 . then @xmath86 induces a positive definite metric @xmath67 on @xmath87 , and the restriction @xmath68 of the levi - civita connection of @xmath86 gives a connection on @xmath17 which is compatible with respect to the metric @xmath67 . we set @xmath88 , which gives the structure of the coherent tangent bundle on @xmath2 . a necessary and sufficient condition for a given coherent tangent bundle over an @xmath1-manifold to be realized as a smooth map into the @xmath1-dimensional space form is given in @xcite . [ ex : front ] let @xmath89 be an @xmath90-dimensional riemannian manifold . a @xmath3-map @xmath91 is called a _ frontal _ if for each @xmath22 , there exists a neighborhood @xmath25 of @xmath26 and a unit vector field @xmath92 along @xmath85 defined on @xmath25 such that @xmath93 holds for any vector field @xmath72 on @xmath25 ( that is , @xmath92 is a unit normal vector field ) , and the map @xmath94 is a @xmath95-map , where @xmath96 is the unit tangent bundle of @xmath97 . moreover , if @xmath92 can be taken to be an immersion as a smooth section of @xmath96 for each @xmath22 , @xmath85 is called a _ front _ or a _ wave front_. we remark that @xmath85 is a front if and only if @xmath85 has a lift @xmath98 as a legendrian immersion , where @xmath99 is a projectified cotangent bundle on @xmath97 with the canonical contact structure . the subbundle @xmath87 which consists of the vectors in the pull - back bundle @xmath100 perpendicular to @xmath92 gives a coherent tangent bundle . in fact , @xmath101 gives a bundle homomorphism . let @xmath102 be the levi - civita connection on @xmath97 . then by taking the tangential part of @xmath102 , it induces a connection @xmath68 on @xmath87 satisfying . let @xmath67 be a metric on @xmath87 induced from the riemannian metric on @xmath97 , then @xmath68 is a metric connection on @xmath87 . thus we get a coherent tangent bundle @xmath103 . since the unit tangent bundle can be canonically identified with unit cotangent bundle , the map @xmath94 can be identified with @xmath104 . a frontal @xmath85 is called _ co - orientable _ if there is a unit normal vector field @xmath92 globally defined on @xmath2 . when @xmath97 is orientable , the coherent tangent bundle is co - orientable if and only if so is @xmath85 . from now on , we assume that @xmath76 is a co - orientable coherent tangent bundle , and fix a co - orientation @xmath23 on the coherent tangent bundle . ( if @xmath17 is not co - orientable , one can take a double cover @xmath105 such that the pull - back of @xmath17 by @xmath106 is a co - orientable coherent tangent bundle over @xmath107 . ) [ def : volume ] the _ signed @xmath29-volume form _ @xmath108 and the ( _ unsigned _ ) _ @xmath29-volume form _ @xmath109 are defined as @xmath110 where @xmath34 is a local coordinate system of @xmath2 compatible with the orientation of @xmath2 , and @xmath111 is the @xmath29-jacobian function on @xmath25 given in . both @xmath108 and @xmath109 are independent of the choice of a positively oriented local coordinate system @xmath34 , and give two globally defined @xmath1-forms on @xmath2 . ( @xmath108 is @xmath3-differentiable , but @xmath109 is only continuous . ) when @xmath2 has no @xmath29-singular points , the two forms coincide up to sign . then the two sets @xmath112 as in are written as @xmath113 the @xmath29-singular set @xmath32 coincides with the boundary @xmath114 . when @xmath9 and @xmath115 comes from a front in @xmath15-manifold as in example [ ex : front ] ( resp . a map into @xmath0-manifold as in example [ ex : map ] ) , an @xmath5-point corresponds to a cuspidal edge ( resp . a fold ) @xcite , see fig . [ fig : folds ] ) . though cuspidal cross caps in surfaces in @xmath12 are not singular points of a front , they are also @xmath5-points ( see ( * ? ? ? * remark 1.6 ) ) . in this way , our definition of @xmath5-points are wider than the original definition of @xmath5-points on fronts or morin maps . however , if @xmath115 is associated to a wave front , @xmath5-points really corresponds to cuspidal edges on its realization as fronts . on the other hand , an @xmath8-point corresponds to a swallowtail ( resp . a cusp ) when @xmath9 and @xmath115 comes from a front in a @xmath15-manifold ( resp . a map into a @xmath0-manifold ) . this fact was shown in @xcite ( see fig . [ fig : cusps ] ) . when @xmath29 is induced from a smooth map between @xmath0-manifolds ( resp . a front in a @xmath15-manifold ) , it is well - known that @xmath5-points and @xmath8-points are generic singular points of @xmath29 ( cf . however , if one consider deformations of @xmath5 or @xmath8-points , an additional three types of singular points of type @xmath116 ( lips ) , @xmath117 ( beaks ) and of @xmath118 ( butterfly ) also generically appear . so we mention here an intrinsic characterization of these three singular points . let @xmath115 be a coherent tangent bundle over a @xmath0-manifold @xmath11 . [ def : lips ] a @xmath29-singular point @xmath119 is called a _ lips of @xmath29 _ if it satisfies the following conditions : the rank of @xmath29 at @xmath26 is @xmath120 , the exterior derivative @xmath44 vanishes at @xmath26 , the hessian matrix of the function @xmath41 at @xmath26 is positive definite . [ def : beaks ] a @xmath29-singular point @xmath119 is called a _ beaks of @xmath29 _ if it satisfies the following conditions : the rank of @xmath29 at @xmath26 is @xmath120 , the exterior derivative @xmath44 vanishes at @xmath26 , the hessian matrix of the function @xmath41 at @xmath26 is negative definite , the second derivative @xmath121 does not vanish at @xmath26 , where the prime means the derivative with respect to the null direction ( cf . and ) . [ def : b - fly ] a non - degenerate @xmath29-singular point @xmath119 is called a _ butterfly of @xmath29 _ if it satisfies @xmath122 where the prime means the derivative with respect to the null direction . [ fact : saji ] when a coherent tangent bundle is induced from a front , lips , beaks and butterfly correspond to cuspidal lips , cuspidal beaks and cuspidal butterfly , respectively ( fig . [ fig : front - peaks ] ) . see @xcite and @xcite . on the other hand , if a coherent tangent bundle is induced from a smooth map between @xmath0-manifolds , these singular points are corresponding to those on the map ( fig . [ fig : map - peaks ] ) . see @xcite . .,title="fig:",height=132 ] .,title="fig:",height=132 ] .,title="fig:",height=132 ] .,title="fig:",height=120 ] .,title="fig:",height=120 ] .,title="fig:",height=120 ] in this way , our intrinsic formulation can give a unified treatment of singular points on maps and on fronts at the same time . let @xmath76 be a coherent tangent bundle and fix a @xmath29-singular point @xmath123 which is an @xmath5-point . then there exists a neighborhood @xmath25 of @xmath26 such that @xmath124 consists of @xmath5-points . now we define the _ singular shape operator _ as follows : since the kernel of @xmath45 is transversal to @xmath48 at @xmath26 , @xmath125 is injective , where @xmath25 is a sufficiently small neighborhood of @xmath26 . then the metric @xmath126 is positive definite on @xmath52 . we take an orthonormal frame field @xmath127 , @xmath128 , , @xmath129 on @xmath124 with respect to @xmath126 . without loss of generality , we may assume that @xmath130 is smoothly extended on @xmath25 as an orthonormal @xmath6-frame field . then we can take a unique smooth section @xmath131 ( called the _ conormal vector field _ ) so that @xmath132 gives a positively oriented orthonormal frame field on @xmath17 . now , we set @xmath133 where the non - vanishing null vector field @xmath47 is chosen so that @xmath134 is compatible with respect to the orientation of @xmath2 . it holds that @xmath135 since @xmath29 is injective on each tangent space of @xmath48 and @xmath136 , the inverse element @xmath137 is uniquely determined . thus we get a bundle endomorphism @xmath138 which is called the _ singular shape operator _ on @xmath48 . [ prop : sym ] the definition of the singular shape operator @xmath139 is independent of the choice of an orthonormal frame field @xmath140 , the choice of an orientation of @xmath2 , and the choice of a co - orientation of @xmath17 . moreover , it holds that @xmath141 in fact , if one reverses the orientation of the basis @xmath142 , then @xmath143 and @xmath47 change sign at the same time , and thus @xmath144 is unchanged . similarly , if we reverse the orientation of @xmath2 ( resp . @xmath17 ) , then @xmath41 and @xmath47 ( resp . @xmath41 and @xmath143 ) change sign at the same time , and @xmath139 unchanged . we can get the final assertion from the following identity @xmath145\bigr)}\right\rangle}\\ & = -{\left\langle{{\boldsymbol{n}}},{d_y{\varphi}(x)}\right\rangle } = { \left\langle{d_y{\boldsymbol{n}}},{{\varphi}(x)}\right\rangle } = ds^2_{\varphi}\bigl({\varphi}^{-1}(d_y{\boldsymbol{n}}),x\bigr ) , \end{aligned}\ ] ] where @xmath72 and @xmath73 are both vector fields on @xmath48 . hence @xmath139 is symmetric with respect to @xmath126 . [ def : singular - curvature ] let @xmath146 be an @xmath5-point of @xmath29 . then @xmath147 is called the _ @xmath29-singular normal curvature _ at @xmath26 with respect to the direction @xmath72 . the eigenvalues of @xmath139 are called the _ @xmath29-singular principal curvatures _ , which give the critical values of the singular normal curvature on @xmath148 . when @xmath9 , the @xmath29-singular principal curvature is called ( simply ) the _ @xmath29-singular curvature _ , which is also denoted by @xmath149 . this definition of the singular curvature is the same as in ( * ? ? ? * ( 1.7 ) ) and ( * ? ? ? * ( 1.6 ) ) . more precisely , @xmath149 is computed as follows : let @xmath146 be an @xmath5-point of @xmath29 . then the @xmath29-singular set @xmath48 is parametrized by a regular curve @xmath150 ( @xmath151 ) on @xmath11 on a neighborhood of @xmath26 , and @xmath150 is an @xmath5-point of @xmath29 for each @xmath152 . since @xmath153 ( @xmath154 ) is not a null - direction , @xmath155 . take a section @xmath156 of @xmath17 along @xmath157 such that @xmath158 gives a positive orthonormal frame field on @xmath17 along @xmath157 , where @xmath159 . then we have @xmath160 where @xmath161 is a null - vector field along @xmath150 such that @xmath162 is compatible with the orientation of @xmath11 . by , it holds that @xmath163 we now prove the following : [ thm : infty ] let @xmath123 be a @xmath29-singular point of a coherent tangent bundle @xmath66 , and assume @xmath26 is non - degenerate but not an @xmath5-point of @xmath29 . take a regular curve @xmath164\ni t\mapsto \gamma(t)\in\sigma_{\varphi}$ ] such that such that @xmath165\bigr)$ ] consists only of @xmath5-points of @xmath29 and @xmath166 . then one of the @xmath29-singular principal curvatures along @xmath150 diverges to @xmath7 . we can take a local coordinate system @xmath167 on @xmath2 at @xmath26 satisfying the following properties : [ item : coord-1 ] @xmath168 is compatible with respect to the orientation of @xmath2 , [ item : coord-2 ] the @xmath29-singular submanifold is characterized by the zeros of the last coordinate function , that is , @xmath169 . [ item : coord-3 ] the vector field @xmath170 gives the null direction , where @xmath171 is a @xmath3-function satisfying @xmath172 . ( in fact , @xmath49 must tangent to the singular manifold , since @xmath26 is not an @xmath5-point . ) for the sake of simplicity , we set @xmath173 since @xmath150 ( @xmath174 $ ] ) is an @xmath5-point of @xmath29 , [ item : coord-2 ] and [ item : coord-3 ] imply that @xmath175 does not vanish for @xmath174 $ ] . then @xmath176 forms a positive frame field on @xmath28 along @xmath150 ( @xmath177 ) . by definition , the @xmath29-singular normal curvature along @xmath157 with respect to @xmath178 is given by @xmath179 since @xmath47 is a null vector , it holds that @xmath180 on @xmath48 . in particular , since @xmath181 at @xmath26 , we have that @xmath182 differentiating by @xmath183 , we have @xmath184 we can identify @xmath185 for each @xmath186 by using the inner product on @xmath17 , and then we set @xmath187 which gives a conormal vector field such that @xmath188 is a positive frame field on @xmath17 along @xmath150 ( @xmath177 ) . by , we have @xmath189 . then @xmath190 substituting and into , we have that @xmath191 set @xmath192 then substituting into the above equation and noticing that @xmath193 is proportional to @xmath194 , we have @xmath195 hence @xmath196 because of . since @xmath197 , we have that @xmath198 since @xmath26 is non - degenerate and @xmath199 , @xmath200 holds . thus , @xmath201 tends to @xmath7 as @xmath202 . hence , at least one of the singular principal curvatures tends to @xmath7 . at the end of this section , we give a definition of frontal bundles as an intrinsic characterization of wave fronts in space forms . let @xmath2 be an oriented @xmath1-manifold and @xmath76 a co - orientable coherent tangent bundle over @xmath2 . if there exists another bundle homomorphism @xmath203 such that @xmath204 is also a coherent tangent bundle and the pair @xmath205 of bundle homomorphisms satisfies a compatibility condition @xmath206 then @xmath207 is called a _ frontal bundle_. the bundle homomorphisms @xmath29 and @xmath208 are called the _ first homomorphism _ and the _ second homomorphism _ , respectively . we set @xmath209 for @xmath210 ( @xmath22 ) , and we call them _ the first , the second and the third fundamental forms _ , respectively . they are all symmetric covariant tensors on @xmath2 . [ def : front ] a frontal bundle @xmath211 is called a _ front bundle _ if @xmath212 holds for each @xmath22 . [ ex : front - bundle ] let @xmath213 be an @xmath90-dimensional space form , that is , a complete riemannian @xmath90-manifold of constant curvature @xmath214 , and denote by @xmath102 the levi - civita connection on @xmath215 . let @xmath216 be a co - orientable frontal . then there exists a globally defined unit normal vector field @xmath92 . since the coherent tangent bundle @xmath87 given in example [ ex : front ] is orthogonal to @xmath92 , we can define a bundle homomorphism @xmath217 then @xmath218 is a frontal bundle . moreover , this is a front bundle in the sense of definition [ def : front ] if and only if @xmath85 is a front , which is equivalent to @xmath219 being positive definite . as a fundamental theorem for hypersurfaces , the integrability condition for a given frontal bundle to be realized as a frontal in a space form is given in @xcite . we fix a front bundle @xmath207 over an @xmath1-dimensional manifold @xmath2 . [ def : ext ] when @xmath22 is not a singular point of @xmath29 , we define @xmath220 which is called the _ extrinsic curvature _ at @xmath26 with respect to the @xmath221-plane in @xmath222 . if a front bundle @xmath207 is induced from a front in @xmath215 , then it holds that @xmath223 where @xmath224 is the sectional curvature at each @xmath29-regular point @xmath26 of @xmath2 . as a generalization of ( * theorem 3.1 ) , relationships between singular principal curvatures and @xmath225 of wave fronts in space forms are investigated in @xcite . in this section , we give four gauss - bonnet formulas on a given front in @xmath15-dimensional space forms , and will point out several remarkable consequences of them . let @xmath2 be an oriented @xmath1-manifold and @xmath81 an oriented riemannian @xmath1-manifold . as in example [ ex : map ] , a @xmath3-map @xmath226 induces a coherent tangent bundle @xmath227 over @xmath2 . in this setting , an @xmath4-point ( cf . definitions [ def : a2-point ] and [ def : a3-point ] ) coincides with an @xmath228-morin singular point of @xmath85 ( see @xcite ) . now we restrict our attention to the case @xmath9 : an @xmath5-point ( resp . @xmath8-point ) of @xmath229 on @xmath11 is called a _ fold _ ( resp . a _ cusp _ ) ; namely , a fold ( resp . a cusp ) is right - left equivalent to the map @xmath230 at the origin . here , two map germs @xmath231 @xmath232 are right - left equivalent if there exist diffeomorphism germs @xmath233 and @xmath234 such that @xmath235 holds . the @xmath229-singular curvature @xmath236 of an @xmath5-point @xmath26 of @xmath229 ( i.e. a fold ) is called the _ singular curvature _ at a fold . the following assertion follows immediately . [ prop : geod ] let @xmath237 be a @xmath3-map and @xmath26 a fold singular point . suppose that @xmath150 is a regular curve which parametrizes the singular set so that @xmath238 lies on the left - hand side of @xmath239 . then the singular curvature @xmath236 at @xmath26 is equal to the geodesic curvature of @xmath240 . [ ex : parabola ] we set @xmath241 @xmath242 . if @xmath243 ( resp . @xmath244 ) , then all singular points consist of folds with positive ( resp . negative ) singular curvature , see fig . [ fig : pn - folds ] . and @xmath245 ( the folds are the left edges in both figures).,title="fig:",height=113 ] and @xmath245 ( the folds are the left edges in both figures).,title="fig:",height=113 ] now we recall the following two gauss - bonnet formulas on a given coherent tangent bundle : [ fact : suy ] let @xmath115 be a coherent tangent bundle over a compact oriented @xmath0-manifold @xmath11 , and suppose that the @xmath29-singular set @xmath48 consists of @xmath5-points and @xmath8-points . we denote by @xmath246 the gaussian curvature of the induced metric @xmath247 . then it holds that @xmath248 where @xmath249 is the length element on the @xmath29-singular set with respect to @xmath126 , and @xmath250 and @xmath251 are the numbers of positive and negative @xmath8-points of @xmath29 , respectively ( see ( * ? ? ? * figure 4 ) and also ( * ? ? ? * definition 2.9 ) ) . on the other hand , @xmath252 is the euler characteristic of the @xmath253-vector bundle @xmath17 . [ rem : gb ] let @xmath69 be an oriented vector bundle of rank @xmath0 with a metric over a compact oriented @xmath0-manifold @xmath11 , and @xmath68 a metric connection . suppose that there exists a bundle homomorphism @xmath254 . in the first section , @xmath5 and @xmath8-points are defined for such an arbitrary bundle homomorphism @xmath29 and _ holds without assuming the compatibility condition _ , as pointed out in @xcite : for an oriented orthonormal frame field @xmath255 of @xmath17 defined on @xmath256 , there is a unique @xmath120-form @xmath257 on @xmath25 such that @xmath258 then @xmath259 does not depend on the choice of @xmath260 , and there is a @xmath3-function @xmath261 on @xmath262 such that @xmath263 we call @xmath261 the _ gaussian curvature _ of @xmath68 with respect to @xmath29 . if @xmath17 is a coherent tangent bundle , the identity implies that @xmath261 coincides with the gaussian curvature of the metric @xmath264 . thus formulas and still hold for @xmath265 , without assuming . although depends on @xmath68 , is independent of the choice of metric connections . [ rem : peaks ] under the assumption that @xmath17 is a coherent tangent bundle , the identities and hold even if @xmath29 admits wider class of singular points called ` peaks ' . * definition 1.10 ) and also ( * ? ? ? * definition 2.1 ) ) . @xmath8-points , lips ( see definition [ def : lips ] ) , beaks ( see definition [ def : beaks ] ) , butterflies ( see definition [ def : b - fly ] ) are all examples of peaks . for a coherent tangent bundle over a compact oriented @xmath0-manifold whose singular points are at most peaks , the formulas and hold ( see ( * ? ? ? * theorem 2.3 ) and ( * ? ? ? * theorem b ) ) , here @xmath266 and @xmath267 in should be replaced by the numbers of positive ( resp . negative ) peaks ( cf . * definition 2.1 ) ) . the formula in our situation induces quine s formula . [ prop : quine ] let @xmath11 and @xmath268 both be compact oriented connected @xmath0-manifolds , and @xmath237 be a @xmath3-map whose singular set consists of folds and cusps . then the topological degree of @xmath85 satisfies @xmath269 where @xmath270 ( resp . @xmath271 ) is the set of regular points at which @xmath85 preserves ( resp . reverses ) the orientation , and @xmath272 ( resp . @xmath273 ) is the number of positive cusps ( resp . the number of negative cusps ) . on the other hand , the formula induces the following new formula . [ prop : id ] let @xmath274 be an oriented riemannian @xmath0-manifold , and @xmath11 a compact oriented @xmath0-manifold . let @xmath237 be a @xmath3-map whose singular set consists of folds and cusps . then the total singular curvature @xmath275 with respect to the length element @xmath276 ( with respect to @xmath86 ) on the singular set @xmath277 is bounded , and satisfies the following identity @xmath278 where @xmath279 is the gaussian curvature function on @xmath274 , and @xmath280 is the pull - back of the riemannian measure of @xmath274 . in particular , if @xmath274 is the euclidean plane @xmath281 , we have the following classical result : [ cor : id ] let @xmath11 be a compact oriented @xmath0-manifold , and @xmath282 a @xmath3-map whose singular set consists of folds and cusps . let @xmath283 be the disjoint union of the closed regular curves on @xmath11 such that the singular set of @xmath85 is @xmath284 . suppose that we give an orientation to each @xmath285 ( @xmath286 ) so that the image @xmath238 lies on the left - hand side of @xmath285 . then the rotation indices @xmath287 ( which takes values in the set of half - integers ) of @xmath285 ( @xmath286 ) as fronts satisfy @xmath288 we use the same notation as in the previous sections . let @xmath289 be a @xmath15-dimensional complete riemannian manifold with constant curvature @xmath214 . here , we do not assume that @xmath289 is simply connected , and we use the notation @xmath290 when it is in fact assumed to be simply connected , like in the previous section . let @xmath11 be an oriented closed @xmath0-manifold and @xmath291 a front . then the front bundle @xmath292 is defined as in example [ ex : front - bundle ] . we fix a positively oriented local coordinate system @xmath293 on @xmath11 . denote by @xmath294 ( resp . @xmath295 ) the @xmath29-jacobian function ( resp . @xmath208-jacobian function ) as in : @xmath296 where @xmath297 and @xmath23 is the volume form of @xmath87 giving co - orientation of @xmath87 . then the @xmath29-singular set ( resp . ) @xmath298 is expressed as @xmath299 [ rmk:2 ] if @xmath300 , @xmath301 is the singular set of the gauss map @xmath302 of @xmath85 . when @xmath303 , @xmath301 is the singular set of the map induced by the unit normal vector field @xmath304 , where @xmath305 is the unit @xmath15-sphere . this unit normal vector corresponds to the dual map of @xmath85 induced by a duality between @xmath305 and itself . when @xmath289 is the hyperbolic @xmath15-space @xmath306 , @xmath301 is the singular set of the map induced by the normal vector field @xmath307 into the de sitter @xmath15-space @xmath308 . we write the volume forms as in as @xmath309 and @xmath310 denote by @xmath246 ( resp . @xmath311 ) the gaussian curvature of the metric @xmath312 ( resp . @xmath313 ) , which is defined on @xmath314 ( resp . @xmath315 ) . the extrinsic curvature @xmath225 in definition [ def : ext ] can be considered as a @xmath95-function on @xmath314 , because we are working with the @xmath0-dimensional case . similarly , we denote by @xmath316 the extrinsic curvature with respect to the second homomorphism @xmath208 , namely , it is obtained by replacing @xmath317 with @xmath318 in . if we denote by @xmath319 , @xmath320 and @xmath321 the representation matrices of @xmath317 , @xmath322 and @xmath318 , respectively , with respect to @xmath323 : @xmath324 then the extrinsic curvatures are expressed as @xmath325 then is equivalent to @xmath326 [ lem : unbdd ] let @xmath291 be a front . then [ item : k:1 ] @xmath327 , [ item : k:1b ] @xmath328 if @xmath329 , [ item : k:2 ] @xmath330 and @xmath331 , [ item : k:2b ] @xmath332 and @xmath333 , [ item : k:3 ] @xmath334 . take a positive ( local ) orthonormal frame field @xmath255 of @xmath87 , and take a one - form @xmath257 as in , that is , @xmath257 is the connection form of @xmath68 with respect to the frame @xmath255 . by , @xmath335 is determined by just the connection @xmath68 . hence we have [ item : k:1 ] . take @xmath336-matrix - valued functions @xmath337 and @xmath338 as @xmath339 by definition , @xmath340 and @xmath341 in are expressed as @xmath342 in particular , we have that @xmath343 on the other hand , the matrices @xmath319 , @xmath320 and @xmath321 as in are written as @xmath344 where @xmath345 denotes transposition . thus , by , we have @xmath346 hence [ item : k:2 ] , [ item : k:2b ] and [ item : k:3 ] hold . finally , if @xmath329 , @xmath318 is just the pull - back metric of canonical metric of the unit sphere by the gauss map , so [ item : k:1b ] follows . now , we assume both of the singular sets @xmath277 and @xmath301 consist of @xmath5- points and @xmath8-points . then applying two abstract formulas and for @xmath29 and @xmath208 respectively , we have the following four gauss - bonnet formulas : @xmath347 where @xmath348 ( resp . @xmath349 ) is the singular curvature function along @xmath5-points in @xmath277 ( resp . @xmath301 ) as in , @xmath276 ( resp . @xmath350 ) is the length element on the singular curve with respect to @xmath317 ( resp . @xmath318 ) , @xmath351 ( resp . @xmath352 ) is the number of positive ( resp . negative ) @xmath8-points of @xmath29 , and @xmath353 ( resp . @xmath354 ) is the number of positive ( resp . negative ) @xmath8-points of @xmath208 . [ rem : peak2 ] as seen in remark [ rem : peaks ] , formulas hold for fronts such that the singular sets of @xmath29 and @xmath208 consist of at most peak . here , @xmath351 ( resp . @xmath352 ) in should be considered as a number of positive ( resp . negative ) peaks of @xmath29 , and @xmath355 ( resp . @xmath356 ) in should be considered as a number of positive ( resp . negative ) peaks of @xmath208 , in this case . let @xmath291 be an immersion of a compact orientable @xmath0-manifold @xmath11 . then the second homomorphism @xmath357 is the shape operator of @xmath85 , and the set of singular points @xmath301 of it coincides with the _ inflection points _ of @xmath85 ( see @xcite ) . then @xmath5-points and @xmath8-points in @xmath301 are called _ @xmath5-inflection points _ and _ @xmath8-inflection points _ , respectively . [ thm : a ] let @xmath11 be a compact oriented @xmath0-manifold and @xmath291 an immersion . suppose that the set of inflection points of @xmath85 consists of @xmath5-points and @xmath8-points . then we have @xmath358 moreover , @xmath359 coincides with the set @xmath360 . since @xmath5-inflection points and @xmath8-inflection points are non - degenerate singular points of @xmath208 , the set of inflection points @xmath301 is a regular submanifold of a compact manifold @xmath11 . hence @xmath301 is the disjoint union of a finite number of closed curves in @xmath11 , and thus the euler number @xmath361 vanishes . then we have @xmath362 since @xmath85 is an immersion , @xmath363 holds , and then @xmath364 thus , we have the equality . here , since @xmath363 , [ item : k:2 ] and [ item : k:2b ] of lemma [ lem : unbdd ] implies that @xmath365 . then @xmath366 hence we have the conclusion . [ rmk : bw ] when @xmath300 , is the classical formula given in @xcite . in this case , @xmath367 is equal to the total number of cusps appears in the gauss map @xmath92 of @xmath85 . romero - fuster @xcite discusses on this number for embedded surfaces . when @xmath303 or @xmath306 , the formula has been also proved in the authors previous work @xcite . however , this formula is a generalization of those results in @xcite and @xcite , since @xmath289 might not be simply connected in our setting . it should be also remarked that the formula can be generalized to any compact immersed surfaces in an arbitrary orientable riemannian 3-manifold : in fact , for @xmath357 is equivalent to the codazzi equation for the space form and does not hold for a general riemannian @xmath15-manifold . however , still holds without assuming , as mentioned in remark [ rem : gb ] , and the above proof works in this general setting ( see @xcite for details ) . here , we give two examples satisfying @xmath368 . it is well - known that there are embedded triply periodic minimal surfaces . although the gauss maps of minimal surfaces only have isolated singular points , if we perturb the surface , we get an immersion @xmath369 , where @xmath11 is a compact @xmath0-manifold with positive genus . since minimal surfaces have negative gaussian curvature , the perturbation @xmath85 satisfies @xmath368 , and the gauss map of @xmath85 must have cusps . [ ex : trinoid ] similarly , considering the jorge - meeks symmetric trinoid with the three ends rounding off to become closed discs , one can get a closed surface in @xmath12 as in fig . [ fig : trioid ] . then the resulting surface satisfies @xmath368 , and its gauss map must have cusps . in example [ ex : trinoid].,width=124 ] interchanging the roles of @xmath29 and @xmath208 , we get the following dual assertion in @xmath12 . let @xmath370 be a front with unit normal vector @xmath92 . then for each real number @xmath371 , @xmath372 is a front with unit normal @xmath92 , which is called the _ parallel front _ of signed distance @xmath371 of the front @xmath85 . [ thm : b ] let @xmath373 be a strictly convex surface and @xmath374 a parallel front of @xmath85 , assume that the set of singular points of @xmath375 consists of cuspidal edges and swallowtails . then @xmath376 holds , where @xmath377 ( resp . @xmath378 ) is the number of positive ( resp . negative ) swallowtails of @xmath375 , and @xmath379 is the subset of @xmath11 where the gaussian curvature @xmath380 of @xmath375 is negative . since @xmath85 is strictly convex , @xmath381 is a diffeomorphism , and then @xmath301 vanishes . then , exchanging the roles of @xmath29 and @xmath208 , the proof of theorem [ thm : a ] implies the conclusion . [ ex : ellipsoid ] let @xmath382 be a parametrization of the ellipsoid defined by @xmath383 and let @xmath92 be its unit normal vector field . then we can observe that @xmath384 for @xmath385 has four negative swallowtails and the euler number of the set @xmath386 is @xmath387 , where @xmath388 is the gaussian curvature of @xmath389 , see fig . [ fig : ellipsoid ] . in example [ ex : ellipsoid].,height=151 ] on the other hand , as a conclusion of , we have : [ thm : c ] let @xmath390 be an immersion and @xmath302 the gauss map of @xmath85 . assume that the singular set @xmath301 of @xmath92 consists of folds ( i.e. @xmath5-points ) and cusps ( i.e. @xmath8-points ) . then @xmath391 holds , where @xmath392 . in particular , the total dual singular curvature ( with respect to @xmath92 ) @xmath393 is non - positive . since @xmath85 is an immersion into @xmath12 ( i.e. @xmath329 ) , @xmath394 holds . then we have @xmath395 here , @xmath396 thus , by , we have the conclusion . [ ex : sin ] consider a rotation of the sine curve @xmath397 whose gauss map has @xmath5-singular points of positive singular curvature . this shows that the singular curvature of gauss maps can take positive values . [ cor : c1 ] under the same assumptions as in theorem [ thm : c ] , if there exists a point @xmath26 satisfying @xmath398 then there exists a fold point such that @xmath349 is negative . an embedded delaunay surface ( an unduloid ) as a rotationally symmetric periodic surface in @xmath12 can be considered as an immersion into a flat space @xmath399 which has an annular domain having negative gaussian curvature . as a consequence , the image of its gauss map of @xmath85 has folds of negative singular curvature . [ cor : c2 ] let @xmath390 be an immersion of a compact 2-manifold @xmath11 . then the inequality @xmath400 is called the chern - lashof inequality , where @xmath401 @xmath402 is the @xmath403-th betti number of @xmath11 . the equality holds if and only if @xmath85 is tightly immersed in @xmath12 . on the other hand , by the gauss - bonnet formula , it holds that @xmath404 hence , we have that @xmath405 where the equality holds if and only if @xmath85 is a tight immersion . similar to theorem [ thm : b ] , the dual version of theorem [ thm : c ] holds : [ thm : d ] let @xmath373 be a strictly convex immersion and @xmath375 a parallel front of @xmath85 for @xmath406 . assume that the set of singular points @xmath407 of @xmath375 consists of cuspidal edges and swallowtails . then @xmath408 holds , where @xmath380 is the gaussian curvature of @xmath375 , @xmath409 is the set where @xmath410 , and @xmath411 is the singular curvature of cuspidal edges of @xmath375 . the equality @xmath412 follows from the proof of theorem [ thm : c ] by exchanging the roles of @xmath29 and @xmath208 . on the other hand , in the proof of theorem [ thm : a ] , we proved that @xmath413 coincides with the set @xmath360 because of @xmath229 has no singular points . in our situation , since @xmath381 is a diffeomorphism , @xmath357 has no singular points . thus , by exchanging the roles of @xmath29 and @xmath208 , we have that @xmath414 because of [ item : k:3 ] in lemma [ lem : unbdd ] . moreover , by , we have @xmath415 and get the assertion . [ cor : d1 ] under the same assumptions as in theorem [ thm : d ] , if there exists a point @xmath26 satisfying @xmath416 then there exists a point such that @xmath411 is negative . from now on , we assume that the gaussian curvature is bounded . as shown below , there are many such surfaces as wave fronts . the following lemma holds . [ lem : bdd ] let @xmath11 be an oriented @xmath0-manifold , @xmath291 a front , and @xmath26 an @xmath5-singular point of @xmath85 . let @xmath417 be a front bundle associated to @xmath85 ( cf . example [ ex : front - bundle ] ) . suppose that there exists a neighborhood @xmath25 of a @xmath418-@xmath419singular point @xmath26 such that @xmath420 is bounded on @xmath421 , where @xmath225 is the extrinsic gaussian curvature and @xmath277 is the set of singular points of @xmath85 . then the following holds on @xmath25 : [ item : k:4 ] @xmath422 . [ item : k:5 ] by setting @xmath423 , it holds that @xmath424 and @xmath425 , where @xmath426 is the restriction of the function @xmath225 on @xmath25 . we fix a positive orthonormal frame field @xmath255 of @xmath87 and take matrices @xmath337 and @xmath338 as in . since @xmath225 is bounded on @xmath25 , @xmath427 holds on @xmath428 because of . this implies @xmath429 . on the other hand , by [ item : k:3 ] of lemma [ lem : unbdd ] and the assumptions , @xmath430 is also bounded . then exchanging the roles of @xmath29 and @xmath208 , we have @xmath431 . hence [ item : k:4 ] holds . we shall prove [ item : k:5 ] . parametrize the singular set @xmath422 by a regular curve @xmath150 on @xmath25 . since @xmath26 is an @xmath5-point , @xmath432 does not vanish , and one can take a @xmath95-function @xmath433 such that @xmath434 where @xmath154 and @xmath435 ( @xmath436 ) . then the conormal vector @xmath156 along @xmath150 is expressed as @xmath437 let @xmath257 be the connection form of @xmath68 with respect to the frame @xmath255 as in . then we have @xmath438 substituting this into , we have @xmath439 where @xmath161 is a null vector field such that @xmath162 is positively oriented . then we have @xmath440 similarly , if we set @xmath441 we have @xmath442 where @xmath443 is a null vector field with respect to @xmath208 such that @xmath444 is positively oriented . both of the frame fields @xmath162 and @xmath444 are compatible to the orientation of @xmath11 . since @xmath445 on @xmath150 , the relations and yield @xmath446 since @xmath447 , we have @xmath448 finally , by ( * ? ? ? * theorem 3.1 ) , the second fundamental form @xmath322 vanishes on @xmath150 . thus by and , we have @xmath449 and thus @xmath450 is constant . hence @xmath451 summing up , by , , and , we have [ item : k:5 ] . [ thm : add ] let @xmath11 be a compact oriented @xmath0-manifold , and @xmath390 a front such that @xmath452 and @xmath453 is bounded on @xmath454 . then the singular curvature at the @xmath5-points of the gauss map of @xmath85 is always negative . [ ex : p ] we set @xmath455 which is a rotation of a cycloid . by a straightforward calculation , @xmath246 is bounded . in particular , the gauss map of @xmath85 has a fold of negative singular curvature . [ thm : e ] let @xmath11 be a compact oriented @xmath0-manifold , and @xmath291 a front such that @xmath456 is bounded on @xmath314 . suppose that the singular sets @xmath277 of @xmath29 and @xmath301 ( @xmath457 ) of @xmath208 consist of @xmath5-points and @xmath8-points . then @xmath458 holds , where @xmath351 ( resp . @xmath352 ) is the number of positive ( resp . negative ) swallowtails of @xmath85 , and @xmath353 ( resp . @xmath353 ) is the number of positive ( resp . negative ) @xmath8-point of the map @xmath92 induced from the unit normal vector field of @xmath85 as in remark [ rmk:2 ] . combining [ item : k:1 ] of lemma [ lem : unbdd ] , , and [ item : k:5 ] of lemma [ lem : bdd ] , we have the conclusion . [ ex : cmc ] let @xmath370 be an immersion of constant mean curvature @xmath459 and @xmath460 its gauss map . then the parallel surface @xmath461 is a wave front with constant gaussian curvature @xmath120 . since the unit normal vector field of @xmath462 is also @xmath92 , the singular set @xmath301 of @xmath92 coincides with the set @xmath463 , where @xmath464 is the gaussian curvature of @xmath85 . then by lemma [ lem : bdd ] , @xmath465 . it is well - known that there are many constant mean curvature immersed tori in @xmath12 . on the other hand , groe - brauckmann @xcite constructed triply periodic surfaces in @xmath12 with constant mean curvature @xmath459 . such surfaces are compact surfaces in a @xmath15-dimensional flat torus , and then theorem [ thm : e ] can be applied for @xmath466 of them . moreover , for fronts of negative extrinsic curvature , we have the following : [ thm : g ] let @xmath11 be a compact oriented @xmath0-manifold , and @xmath291 a front such that @xmath467 and @xmath456 is bounded on @xmath11 . then @xmath468 holds . by lemma [ lem : bdd ] , we have @xmath422 and @xmath469 . then by and , we have @xmath470 on the other hand , by [ item : k:5 ] in lemma [ lem : bdd ] and [ item : k:1 ] in lemma [ lem : unbdd ] , we have @xmath471 combining these , we have the conclusion . [ ex : cyc ] it is well - known that there is a front of constant gaussian curvature @xmath472 which is diffeomorphic to a torus , that is , euler number vanishes . moreover , we can construct an example of a compact front of non - constant negative gaussian curvature as follows : we set @xmath473 which is another rotation of a cycloid ( cf . example [ ex : p ] ) . one can easily check that @xmath246 is bounded . since @xmath85 is periodic , it can be considered as a torus in a flat space form @xmath474 . let @xmath475 be a flat front in the unit @xmath15-sphere , that is , @xmath85 has vanishing gaussian curvature on its regular points . then the unit normal vector field @xmath304 also gives a flat fronts . if @xmath11 is compact , @xmath85 is weakly complete in the sense of @xcite and @xmath11 is orientable ( see ( * ? ? ? * section 1 ) ) . on the other hand by theorem [ thm : g ] , the euler characteristic of @xmath11 vanishes because @xmath476 for flat surfaces in @xmath305 . thus @xmath11 is diffeomorphic to a torus . moreover the singular set of @xmath85 coincides with that of @xmath92 , and theorem [ thm : e ] yields that @xmath477 holds . this identity is not trivial since the set of swallow tails of @xmath85 is not equal to that of @xmath92 . ( in fact , @xmath26 is an @xmath8-point , the null direction should be tangential direction of the singular set @xmath277 . on the other hand , null directions of @xmath29 and @xmath208 are linearly independent because @xmath85 is a front . ) several other properties of flat tori in @xmath305 as fronts are discussed in @xcite . the authors thank wayne rossman for careful reading of the first draft for giving valuable comments . they also thank the referee for valuable comments . s. izumiya , k. saji and m. takahashi , _ horospherical flat surfaces in hyperbolic @xmath15-space _ , j. math . japan * 62 * ( 2010 ) , 789849 . y. kitagawa and m. umehara , _ extrinsic diameter of immersed flat tori in @xmath305 _ , preprint . h. levine , _ mappings of manifolds into the plane _ , amer . j. math . * 88 * ( 1966 ) , 357365 . h. liu , m. umehara and k. yamada , _ the duality of conformally flat manifolds _ , to appear in bulletin of the brazilian mathematical society , arxiv:1001.4569 . j. r. quine , _ a global theorem for singularities of maps between oriented @xmath0-manifolds _ , trans . * 236 * ( 1978 ) 307314 . m. c. romero - fuster , _ sphere stratifications and the gauss map _ , proc . edinburgh sect . a * 95 * ( 1983 ) 115136 . k. saji , _ criteria for singularities of smooth maps from the plane into the plane and their applications _ , hiroshima math . j. * 40 * ( 2010 ) , 229239 . k. saji , m. umehara and k. yamada , _ the geometry of fronts _ , ann . of math . * 169 * ( 2009 ) 491529 . k. saji , m. umehara and k. yamada , _ behavior of corank one singular points on wave fronts _ , kyushu j. math . * 62 * ( 2008 ) , 259280 .

we review chiral ( klein ) tunneling in single - layer and bilayer graphene and present its semiclassical theory , including the berry phase and the maslov index . peculiarities of the chiral tunneling are naturally explained in terms of classical phase space . in a one - dimensional geometry we reduced the original dirac equation , describing the dynamics of charge carriers in the single layer graphene , to an effective schrdinger equation with a complex potential . this allowed us to study tunneling in details and obtain analytic formulas . our predictions are compared with numerical results . we have also demonstrated that , for the case of asymmetric n - p - n junction in single layer graphene , there is total transmission for normal incidence only , side resonances are suppressed .

1106.3042

since this paper is prepared for the proceedings of the nobel symposium on graphene we do not start with general explanations what graphene is and why it is important , it will be very well described in other presentations . we just refer to reviews @xcite . our particular subject is chiral , or klein ( as it was called in @xcite ) tunneling . this is one of the key phenomena determining peculiar electronic properties of graphene . in light of possible applications , the klein tunneling protects high charge carrier mobility despite unavoidable inhomogeneities . at the same time , due to the klein tunneling graphene electronics can not copy the standard semiconductor one : if you make graphene transistor based on n - p - n junction just like for silicon , it will not be efficient since you will not be able to lock it . these two remarks illustrate the importance of the subject , the more detailed discussion is presented below . the paper consists of two pieces . the first one ( sections [ kleinparadox ] [ kleinbi ] ) preserves the historical line of thoughts and presents the motivation of the problem , from @xcite to @xcite . in the second part ( sections [ sec::dimvar ] [ sec : numerics ] ) we present a systematic semiclassical theory of the chiral tunneling , together with numerical results . soon after the discovery of the dirac equation , o. klein @xcite noticed one of its strange properties which was afterwards called the `` klein paradox '' . klein considered the original four by four dirac equation , which governs the dynamics of a spin one half particle moving in three - dimensional space . to make a direct connection to the case of graphene without changing the essence of the paradox , we will consider a two by two matrix equation for a particle propagating in two - dimensional space : @xmath0 where @xmath1 and the hamiltonian @xmath2 here @xmath3 is the mass of the particle , @xmath4 is the speed of light and @xmath5 is the potential energy . to demonstrate the essence of the paradox we consider normal incidence on a one - dimensional potential barrier , which means that @xmath6 and @xmath7 . then equation ( [ eq : statschrod ] ) takes the form @xmath8 to make the problem exactly solvable we use a step - wise potential @xmath9 where @xmath10 is a positive constant . we consider a general scattering problem with an incoming wave @xmath11 and a reflected wave @xmath12 for @xmath13 , @xmath14 and a transmitted wave @xmath15 for @xmath16 , @xmath17 the @xmath18-dependence of the solutions for @xmath13 can be written as @xmath19 , where the wave vector @xmath20 satisfies the relativistic dispersion relation @xmath21 as can be found by diagonalizing equation ( [ eq : twoddirac ] ) with @xmath22 . alternatively the wave vector can be written as @xmath23 one easily sees that there are three distinct regimes , two of which are classically allowed , namely @xmath24 corresponding to electron states and @xmath25 corresponding to hole or positron states . there is also a classically forbidden region @xmath26 where the wave vector @xmath20 is imaginary and we have evanescent waves . in what follows we will assume that we are in the electron regime . by calculating eigenvectors of equation ( [ eq : twoddirac ] ) one obtains for the wavefunctions to the left of the barrier @xmath27 and @xmath28 where @xmath29 to the right of the barrier we have a new wave vector @xmath30 , which satisfies the relativistic dispersion relation @xmath31 , or @xmath32 consider a jump @xmath33 since in this case the paradox arises . the wave vector @xmath30 is real and we have a propagating wave on the right side of the barrier . note however that this particle belongs to the hole continuum rather than to the electron one . for smaller values of @xmath10 , there are either propagating electrons on both the left and the right side of the barrier , when @xmath34 , or evanescent waves on the right side of the barrier , when @xmath35 . solving the dirac equation ( [ eq : twoddirac ] ) on the right side of the barrier , one obtains for the transmitted wave @xmath36 where @xmath37 from the continuity of the wavefunction at @xmath38 , @xmath39 we find @xmath40 for the considered case we have @xmath41 , so that @xmath42 and @xmath43 to treat reflection and transmission coefficients properly one has to look at the probability current density for the one - dimensional dirac equation @xmath44 which is a conserved quantity . when we look at the current density ( [ eq : currdensity ] ) we see that it takes values @xmath45 for the incoming wave and @xmath46 for the reflected wave . therefore @xmath47 is nothing but the reflection coefficient and we come to the conclusion that the amplitude of the reflected wave is larger than the amplitude of the incident one . this strange effect that occurs when condition ( [ eq : kleinhighpot ] ) is fulfilled was initially called the klein paradox . in our further discussion we will follow @xcite and @xcite . for a rather complete list of references see @xcite . first of all note that the current density ( [ eq : currdensity ] ) on the right hand side equals @xmath48 , indicating that there is something wrong with the definition of the transmitted wave . what exactly is wrong was pointed out by pauli , who noticed that the group velocity for the case of equation ( [ eq : kleinhighpot ] ) , @xmath49 is opposite to the direction of the wave vector @xmath30 . since the group velocity determines the direction of propagation , the transmitted wave ( [ eq : kleintransmitted ] ) corresponds ( for positive @xmath30 ) to a particle moving to the left instead of to the right . therefore we should define our outgoing wave as @xmath50 which gives the currenty density @xmath51 . when we once again calculate @xmath52 from equation ( [ eq : continuity ] ) , it is seen that @xmath53 which is always smaller than one . therefore the formal paradox disappears , see also @xcite . the paradox reappears when we consider the problem from a different angle . instead of an infinitely broad barrier we will consider a finite barrier , @xmath54 the problem with the choice of the transmitted wave on the right side of the barrier has now disappeared , since it is simply @xmath55 . within the barrier one now has to consider both modes @xmath56 , representing the most general solution . reflection and transmission coefficients are then obtained from the continuity of the wave function at @xmath57 and @xmath58 , which gives after straightforward calculations ( see e.g. @xcite and @xcite ) @xmath59 there is no paradox in these expressions , since @xmath60 , @xmath61 and @xmath62 as it should be . note that we have total transmission through the barrier when @xmath63 with integer @xmath64 . we can consider an infinitely broad barrier by letting @xmath65 go to infinity in the above expressions . as @xmath65 becomes very large while other parameters remain fixed the sine will oscillate very rapidly . we can then average over the fast oscillations and replace @xmath66 by its average value @xmath67 to obtain the expressions @xmath68 one may be surprised that the results ( [ eq : kleinreflected ] ) and ( [ eq : kleinreflected2 ] ) do not coincide . it is however well known from electromagnetic wave theory @xcite that the reflection coefficients for the two situations should differ . from the last result we see once again that the paradox has disappeared in its mathematical form , but has reappeared as physically counterintuitive behaviour . in non - relativistic quantum mechanics a particle can tunnel through a classically forbidden region @xmath69 , but the probability is exponentially small when the barrier is high and broad . in the semiclassical approximation the transmission through the barrier with turning points @xmath70 , which satisfy @xmath71 , is given by @xmath72 where @xmath3 is the mass of the particle . for a relativistic particle incident on a sufficiently high barrier ( [ eq : kleinhighpot ] ) the situation is dramatically different . in the limit @xmath73 the probability of penetration ( [ eq : kleintransmitted2 ] ) is in general not small at all . even for an infinitely high barrier ( @xmath74 ) one has @xmath75 and @xmath76 this is of the order of one when @xmath77 is of the order of @xmath78 , while it is approximately equal to one in the ultrarelativistic limit @xmath79 this is the contemporary formulation of the klein paradox @xcite ; quantum relativistic particles can tunnel with large enough probabilities through barriers of arbitrarily large height and width . the tunneling effect can be hand - wavingly explained with the help of the heisenberg uncertainty principle . since one can not know both momentum and position with an arbitrary accuracy at a given instant , one can not separate the total energy into a potential and a kinetic part . so the kinetic energy can be `` a bit '' negative . in the relativistic regime the restriction is much stronger @xcite : one can not even know the coordinate with an accuracy higher than @xmath80 . therefore relativistic quantum mechanics can not be mechanics , but can only be field theory @xcite . this theory will always contain particles and antiparticles and to measure the coordinate better than @xmath81 one needs to apply such a high energy that particle - antiparticle pairs will be created . the original particle whose coordinate one wanted to measure will then be lost among the newly - born particles . a full field theoretic treatment of the problem was given in ref . the most important point is that although the problem of a high enough barrier looks like a static problem , this is actually not the case . one needs to study carefully how the state is reached and this involves positron emission by the growing barrier . for a more detailed discussion of the role of electron - positron pairs in the klein paradox , see @xcite . the hamiltonian for charge carriers in graphene near conical points @xmath82 and @xmath83 is given by the massless dirac hamiltonian @xmath84 where @xmath85 is the fermi velocity @xmath86 . to consider normal incidence on the one - dimensional potential barrier ( [ eq : potentialbarrier ] ) in this case , we can simply put @xmath87 in our previous results . from equations ( [ eq : defalpha ] ) and ( [ eq : defbeta ] ) it is seen that @xmath88 . therefore we have @xmath89 and @xmath90 in equations ( [ eq : kleinreflected2 ] ) and ( [ eq : kleintransmitted2 ] ) , regardless of the height of the potential . this result is not related to the specific shape of the potential @xcite . this property has an analog in two and three dimensions with @xmath91 or @xmath92 , namely that backscattering is forbidden . this was found long ago for scattering of ultrarelativistic particles in three dimensions ( see @xcite ) . an important consequence of this property for carbon materials was noticed in @xcite . absence of backscattering explains the existence of conducting channels in metallic carbon nanotubes , while in a non - relativistic one - dimensional system an arbitrarily small disorder leads to localization @xcite . the consideration in @xcite is very instructive since it explicitly shows the role of the berry phase and time - reversal symmetry , but it is also quite cumbersome . here we present a somewhat simplified scheme of this proof . to this aim we consider the equation for the @xmath93-matrix ( see e.g. @xcite ) @xmath94 where @xmath95 is the operator corresponding to the scattering potential , @xmath96 is the green s function of the unperturbed hamiltonian @xmath97 and @xmath98 is the electron energy , which is assumed to be larger than zero . if @xmath97 is the dirac hamiltonian for massless dirac fermions ( [ eq : graphenehamiltonian ] ) , we have @xmath99\ ; , \ ] ] where @xmath100 with @xmath101 . the probability of backscattering can be found by iterating equation ( [ eq : deftmatrix ] ) and is proportional to @xmath102 where @xmath103 is the contribution proportional to @xmath104 . we can always choose axes such that @xmath105 . in this case @xmath106 and @xmath107 have spinor structures @xmath108 and @xmath109 respectively . therefore , if @xmath110 is the two by two matrix @xmath111 one has @xmath112 now keeping in mind that @xmath85 is proportional to the identity matrix one can prove term by term that all contributions to @xmath113 and @xmath114 vanish by symmetry . actually this is because @xmath115 ; from the vectors @xmath116 and @xmath117 one can not construct anything with nonzero @xmath118 or @xmath119 components . strictly speaking this argument is only enough for an isotropic potential ; for a generic case one has to do a term by term analysis based on expansion ( [ eq : tmatrixexpansion ] ) , see ref . @xcite . for two nonparallel vectors @xmath120 and @xmath121 one can construct a matrix with nonzero @xmath118 or @xmath119 components , since one of the vectors has a nonzero @xmath118 component , so that @xmath122 . when one thinks about electrons in quantum electrodynamics , it is not easy to create potential jumps larger than @xmath123 mev . similar phenomena take place in electric or gravitational fields ( @xcite ; see @xcite for a detailed list of references ) , but the context is always quite exotic , such as collisions of ultraheavy ions or even black hole evaporation . there were no experimental data available which would require the klein paradox for their explanation . however shortly after the discovery of graphene it was realized that klein tunneling is one of the crucial phenomena for graphene physics and electronics @xcite . soon after this theoretial prediction the effect was confirmed experimentally @xcite . considering possible applications , klein tunneling in graphene is rather bad news . if one copied the construction from a silicon transistor to graphene , it would be impossible to lock the transistor . one would need to open a gap in the spectrum to be able to lock it . at the same time it is good news as well : due to the klein paradox inhomogeneities in the electron density do not lead to localization and their effect on the electron mobility is not very essential @xcite . let us now consider a massless dirac fermion incident on the potential barrier ( [ eq : potentialbarrier ] ) with positive energy under an angle @xmath124 , as it was done first in @xcite . of course , the potential can not be sharp on the atomic scale , since this would induce umklapp scattering between different valleys . therefore by a step - wise potential we mean that the electron wavelength @xmath125 is much larger than the typical spatial scale of the potential @xmath126 , which is in turn much larger than the size of the unit cell . within this assumption the solution is each region is given by traveling waves proportional to @xmath127 , where @xmath128 and @xmath129 satisfy the dispersion relation @xmath130 as can be found from equation ( [ eq : graphenehamiltonian ] ) . similarly to the original dirac equation we can distinguish three distinct regimes from this equation . for @xmath131 we have electrons and for @xmath132 we have holes , while the region @xmath133 is classically forbidden . as was done for the case of the massive dirac equation , we will now require that the potential @xmath10 in equation ( [ eq : potentialbarrier ] ) satisfies @xmath134 so that we have hole states within the barrier . let us denote by @xmath20 the wave vector for @xmath135 and by @xmath30 the wave vector for @xmath136 . at the potential jump the momentum in the @xmath118 direction should be conserved , so that the new angle @xmath137 is related to the new wave vector @xmath30 by @xmath138 from equation ( [ eq : graphenehamiltonian ] ) we see that the second component of the wavefunction is related to the first by @xmath139 so the solutions in the three regions are given by @xmath140 where we have introduced @xmath141 , @xmath142 , @xmath143 and @xmath144 . note that the reflected particle moves under the angle @xmath145 , assuming that the angle changes from @xmath146 to @xmath147 , so that we have the phase @xmath148 for the reflected wave . we can now determine the reflection coefficient @xmath52 , the transmission coefficient @xmath149 and the coefficients @xmath150 and @xmath151 as before , from the requirement that the wavefunction is continuous at @xmath152 . finally the result is given by @xmath153 - 2 i \sin(2 q_x a ) } \ ; . \label{eq : graphenereflectionbarrier}\ ] ] for the case under consideration we have @xmath154 , since the signs of @xmath98 and @xmath155 are opposite . the transmission probability can now easily be calculated as @xmath156 from equation ( [ eq : graphenereflectionbarrier ] ) we immediately see that the reflection is zero for normal incidence , as we proved for a more general potential in the previous section . there are also additional angles , called `` magic angles '' , at which the reflection coefficient is zero and we have full transmission . they are given by the condition @xmath157 where @xmath64 is an integer . we can compare the behaviour of electrons in single layer graphene with the behaviour of normal electrons . when the potential barrier contains no electronic states , the transmission decays exponentially with increasing barrier width and height , @xcite , so that the barrier would reflect electrons completely . but since single layer graphene is gapless , it seems more appropriate to compare it to a gapless semiconductor with non - chiral charge carriers , a situation which can be realized in certain heterostructures @xcite . for this case we find @xmath158 where @xmath128 and @xmath159 are the @xmath18-components of the wave vector outside and inside the barrier , respectively . as in the case of single layer graphene there are resonance conditions at which the barrier is transparent , given by @xmath160 , where @xmath64 is an integer . for normal incidence we see that the transmission coefficient is an oscillating function of the tunneling parameters and can exhibit any value between zero and one . this is in contrast to single layer graphene , where the transmission is always perfect . bilayer graphene consists of two layers of graphene on top of each other , the second layer being rotated by 120 degrees with respect to the first one . in this configuration the sublattices @xmath150 lie exactly on top of each other and the hopping parameter @xmath161 between them is approximately 0.4 ev @xcite , while the in - plane hopping parameter @xmath162 is approximately an order of magnitude larger . when we consider only low energy excitations , @xmath163 , the effective hamiltonian is given by @xcite @xmath164 where the effective mass @xmath165 , @xmath166 being the free electron mass @xcite . there is also hopping between the @xmath151 sublattices of both layers , which is denoted by @xmath167 ev . when we include this parameter into the description an extra term is added to the hamiltonian , which corresponds to so - called trigonal warping . this effect is however only important for small wave vectors @xcite , we will exclude it assuming that @xmath168 . let us consider an electron incident on the potential step ( [ eq : potentialbarrier ] ) under an angle @xmath124 , as was done in @xcite . since the potential is constant in the @xmath118-direction we can write the solution as @xmath169 inserting this into equation ( [ eq : statschrod ] ) with the hamiltonian ( [ eq : bilayergraphenehamiltonian ] ) , we obtain @xmath170 the solutions are therefore given by propagating waves @xmath171 and exponentially growing and decaying modes @xmath172 , @xmath173 the presence of evanescent modes is markedly different from both the schrdinger case and the dirac case . once again there are three regimes . there are electron states for @xmath174 and hole states for @xmath175 , while the region in between is classically forbidden . in what follows we assume that @xmath10 in equation ( [ eq : potentialbarrier ] ) satisfies @xmath176 to find the spinors that are the solutions to equation ( [ eq : bilayercstreduced ] ) we note that the components are related by @xmath177 as can be seen from the hamiltonian ( [ eq : bilayergraphenehamiltonian ] ) . now let @xmath178 be the wave vector for the propagating modes in the region @xmath135 , while @xmath179 is the wave vector in the region @xmath136 . then the solution for @xmath180 is given by @xmath181 where @xmath182 , @xmath183 , @xmath184 , @xmath185 and finally @xmath186 . the amplitude @xmath187 is the amplitude for the incoming wave in this expression , while @xmath188 corresponds to the reflected wave . for @xmath189 we have the general solution @xmath190 where @xmath191 is the transmission coefficient . inside the barrier we need the most general solution with two propagating modes and two modes with real exponentials , @xmath192 where @xmath193 because the transverse momentum is conserved . furthermore @xmath144 , @xmath194 , @xmath195 and @xmath196 . now the coefficients @xmath197 , @xmath198 , @xmath199 and @xmath200 have to be found from the continuity of @xmath201 and the derivative @xmath202 at the points @xmath152 . when the problem is solved numerically , one sees that the transmission probability at normal incidence is exponentially small . similar to the case of single layer graphene , there are once again `` magic angles '' in the spectrum , at which there is total transmission . the existence of magic angles in bilayer graphene has the same consequences as in single layer graphene , meaning that we can not lock a conventional transistor made from bilayer graphene . for the case of normal incidence @xmath203 we can also solve the problem analytically . the transmission coefficient is given by @xmath204 which is indeed exponentially small . when we let @xmath65 go to infinity , the transmission probability @xmath205 becomes zero at normal incidence . furthermore for a single n - p junction with @xmath206 the following analytical solution can be found for any @xmath124 @xmath207 which also gives @xmath208 at normal incidence , in contrast to the case of single layer graphene , where normally incident electrons are always transmitted . it is also different from the case of normal electrons , where the transmission is given by equation ( [ eq : normalelectrontransmission ] ) . in sections [ sec::kleinsingle ] , [ kleinbi ] it was discussed that the wavefunctions @xmath209 of charge carriers in single layer and bilayer graphene in a one - dimensional geometry obey equations @xmath210\psi=0 , \label{eq::single - dim}\ ] ] and @xmath211\psi=0 , \label{eq::bi - dim}\ ] ] respectively . here @xmath126 is a characteristic scale of a potential change . in dimensionless variables ( [ eq::single - dim ] ) takes the form @xmath212\psi=0 , \label{eq::single - dimless}\ ] ] where @xmath213 , @xmath214 , @xmath215 , @xmath216 , @xmath217 and @xmath218 . we denote some characteristic value of @xmath219 as @xmath220 . analogously , ( [ eq::bi - dim ] ) can be rewritten as @xmath221\psi=0 , \label{eq::bi - dimless}\ ] ] with @xmath213 , @xmath214 , @xmath215 , @xmath216 , @xmath222 and @xmath223 . we denote some characteristic value of @xmath219 as @xmath224 . thus we can introduce dimensionless hamiltonians ( we omitted tildes ) : @xmath225 for a single layer and @xmath226 for a bilayer . in both cases there are two substantial parameters in the problem : @xmath227 and @xmath228 . charge carriers in single layer graphene are described by the hamiltonian ( [ eq::single - dimless2 ] ) . this hamiltonian describes simultaneously coupled electron and hole states . according to [ app::ad - ham ] , in adiabatic approximation ( [ eq::single - dimless2 ] ) can be diagonalized up to any order of @xmath229 . the obtained scalar hamiltonians describe electrons and holes separately . the diagonalization is based on a series of unitary transformations of the original hamiltonian and traces back to the ideas of the foldy - wouthuysen transformation @xcite and the peierls substitution in blount s treatment @xcite . we use its variant @xcite . effective electron and hole hamiltonians @xmath230 and @xmath231 can be written as series with respect to the small parameter @xmath227 : @xmath232 to be precise we will assume that any function of @xmath233 and @xmath18 is defined in such a way that @xmath233 acts the first . as soon as the ordering of operators has been introduced , one can work with functions of @xmath4-numbers @xmath234 and @xmath18 . these functions are called `` symbols '' @xcite . it is shown in [ app::ad - ham ] , that leading terms @xmath235 of the effective hamiltonians @xmath236 are eigenvalues of @xmath237 : @xmath238 where @xmath239 are two eigenvectors of the matrix @xmath237 . this gives @xmath240 we note that in the absence of a magnetic field @xmath241 does not depend on @xmath18 . the first correction @xmath242 reads @xmath243 standard semiclassical treatment ( see [ app::sem - appr ] ) can be applied to scalar schrdinger - like equations @xmath244 . we are looking for a solution in the form @xmath245 , @xmath246 this gives @xmath247\end{aligned}\ ] ] with @xmath248 to be found from the hamilton - jacobi equation @xmath249 , where @xmath250 ^ 2-p_y^2\right)^{1/2}}{|e - u(x)|}.\ ] ] differentiating the hamilton - jacobi equation with respect to @xmath18 we find @xmath251 whence @xmath252 this gives @xmath253^{1/4}}e^{\pm is^+(x)/h+i\phi^\pm_p(x)/2}. \label{eq::amp - single - pm}\end{aligned}\ ] ] though it is possible to define locally @xmath241 in ( [ eq::l0-single ] ) as @xmath254 to obtain @xmath255 , such a choice does not provide a single - valued function in the classical phase space . let us first consider a scattering problem for @xmath256 following [ app::sem - appr ] and [ app::bohr - somm ] . for an electron coming from the left of the classically forbidden region we have @xmath257^{1/4}}%\nonumber\\&\times & \left(e^{is^+(x)/h+i\phi^+_p(x)/2}+e^{-is^+(x)/h+i\phi^-_p(x)/2-i\pi/2}\right ) \label{eq::amp - single}\end{aligned}\ ] ] and @xmath258^{1/4}}%\nonumber\\&\times & \left(\begin{array}{c}e^{is^+(x)/h - i\phi^+_p(x)/2}+e^{-is^+(x)/h - i\phi^-_p(x)/2-i\pi/2 } \\ e^{is^+(x)/h+i\phi^+_p(x)/2 } + e^{-is^+(x)/h+i\phi^-_p(x)/2-i\pi/2 } \end{array}\right ) , \label{eq::psi}\end{aligned}\ ] ] where @xmath259 note that @xmath260 continuously depends on @xmath228 when it passes through zero and @xmath261 undergoes a jump of @xmath262 . the reflection coefficient @xmath52 can be computed from ( [ eq::amp - single ] ) or ( [ eq::psi ] ) . it is usually defined as the coefficient in front of the semiclassical solution corresponding to the outgoing wave . one can also assume that the potential tends to a constant at infinity and take the coefficient in front of the plane wave , which is a particular case of the definition given above . obviously , the reflection coefficient defined in such a way does not depend on @xmath18 . choosing ( [ eq::amp - single - pm ] ) as incoming and outgoing solutions , we can write wavefunctions on the left of the classically forbidden region as @xmath263^{1/4}}%\nonumber\\&\times & \left(e^{is^+(x)/h+i\phi^+_p(x)/2}+r(p_y)e^{-is^+(x)/h+i\phi^-_p(x)/2}\right ) , \label{eq::psi - refl}\\ \fl\psi(x)=\frac{|e - u(x)|^{1/2}}{\left[(e - u(x))^2-p_y^2\right]^{1/4}}\nonumber\\ \times \left[\left(\begin{array}{c}e^{is^+(x)/h - i\phi^+_p(x)/2 } \\ e^{is^+(x)/h+i\phi^+_p(x)/2 } \end{array}\right)+r(p_y ) \left(\begin{array}{c}e^{-is^+(x)/h - i\phi^-_p(x)/2 } \\ e^{-is^+(x)/h+i\phi^-_p(x)/2 } \end{array}\right)\right ] , \label{eq::psi - refl}\end{aligned}\ ] ] comparing ( [ eq::psi - refl ] ) , ( [ eq::psi - refl ] ) and ( [ eq::amp - single ] ) , ( [ eq::psi ] ) we conclude that @xmath264 a similar calculation for a hole coming from the right gives , see also figure [ fig::phase - space ] , @xmath265 we paid attention to the definition of the reflection coefficient , since it may lead to discrepancy for the dirac particle . the problem appears due to a jump of @xmath262 in @xmath261 at any fixed @xmath18 as a function of @xmath228 when it goes through zero . this jump is a consequence of the cut at @xmath266 . at any @xmath256 this cut corresponds to infinite negative @xmath18-component of the momentum , and does not imply any discontinuities in the region , where the potential is finite . this jump results in the jump of @xmath267 in the phase of the wavefunction corresponding to the outgoing wave . however , the phase difference @xmath268 tends to zero when @xmath18 tends to a turning point @xmath269 and can therefore be treated as one half of the angle around the origin in @xmath270-space accumulating during the motion of a classicle particle from the point @xmath18 to the turning point @xmath269 and back . the peculiar behaviour of the phase difference can mathematicaly be expressed as the noncommutativity of limits : @xmath271=0,\nonumber\\ \lim_{x\to x_0}\lim_{p_y\to \pm0}[\phi_p^-(x)-\phi_p^+(x)]=\pm\pi.\end{aligned}\ ] ] the jump in the sign of the outgoing wave must be compensated by a kink in the reflection coefficient , since the whole wavefunction should analytically depend on @xmath228 . to get rid of the jump one can redefine the outgoing wave and write @xcite @xmath272^{1/4}}%\nonumber\\&\times & \left(e^{is^+(x)/h+i\phi^+_p(x)/2}+r(p_y)e^{-is^+(x)/h - i\phi^+_p(x)/2}\right).\end{aligned}\ ] ] though preserving the analyticity of @xmath273 , such a definition introduces an artificial jump of the phase as a function of @xmath18 upon reflection at negative @xmath228 . therefore we do not use it below . the reflection coefficient , defined in accordance with ( [ eq::psi - refl ] ) , ( [ eq::psi - refl ] ) does not depend on the sign of @xmath228 . it is completely defined by the orientation of the phase space , which is clockwise for an electron region and counterclockwise for a hole region ( see figure [ fig::phase - space ] ) . finally , the reflection can be written as @xmath274 where ` - ' corresponds to electron and ` + ' to hole regions . it is important to note , that the phase @xmath146 and the module @xmath275 of the reflection coefficient ( [ eq::refl - total ] ) were obtained under the assumption that there is no multiplicity change ! it is not the case when @xmath276 and the trajectory in the phase space tends to a separatrix , see figure [ fig::phase - space - full ] . let us now turn to bilayer graphene . the hamiltonian describing the charge carrier dynamics reads @xmath277 eigenvalues and eigenvectors of @xmath237 are @xmath278 we obtain @xmath279-p_y^2}dx . \label{eq::psi - bilayer}\end{aligned}\ ] ] obviously , the result ( [ eq::refl - total ] ) is valid for the bilayer as well , since the orientation of the phase space is the same . between two classically forbidden regions effective hamiltonians superimpose the following quantization conditions ( see [ app::bohr - somm ] for details ) : @xmath280 where @xmath281 for single and bilayer respectively , @xmath282 is the maslov index and @xmath283 is the total phase gain along the closed classical trajectory . the term @xmath284 is the berry phase in graphene @xcite . it is clear that @xmath283 acquires a non - zero value only if the trajectory in @xmath270-space encloses the origin . therefore in the absence of magnetic field @xmath285 . quantization condition ( [ eq::quantization ] ) allows one to determine resonance angles . though the considered diagonalization is very powerful to deal with complicated matrix hamiltonians in a classically allowed region , it possesses a substantial disadvantage : it treats electrons and holes separately neglecting tunneling effects . in the classically forbidden region when @xmath286 , i.e. @xmath287 effective hamiltonians @xmath288 become degenerate . at this point electron to hole transition may occur and the diagonalization fails . this transition is the origin of the klein tunneling . in the case of normal incidence @xmath289 and at the point @xmath269 , where @xmath290 there is a multiplicity change , i.e. effective hamiltonians @xmath288 become degenerate ( see figure [ fig::phase - space - full ] ) . to study wavefunctions in this case one can not apply a standard semiclassical treatment , described in section [ sec::standard - semi ] , since there may be a `` jump '' between @xmath291 and @xmath292 . fortunately , for the normal incidence in graphene there is an exact pseudospin conservation , which allows one to study this case in detail . for @xmath289 equations ( [ eq::single - dimless ] ) , ( [ eq::bi - dimless ] ) read @xmath293\psi=0 , \label{eq::normal - single - bi}\ ] ] where @xmath281 for single and bilayer respectively . eigenvectors of @xmath294 do not depend on @xmath295 , therefore ( [ eq::normal - single - bi ] ) can easily be diagonalized , which leads to @xmath296\eta_{1,2}=0 , \label{eq::normal}\ ] ] where @xmath297 in this case the eigenvalue of @xmath298 ( `` pseudospin '' ) persists . pseudospin conservation leads to very different physical consequences for single and bilayer graphene . for single layer graphene pseudospin conservation means the conservation of the @xmath18-component of the velocity . equation ( [ eq::normal ] ) is the first order differential equation , which can be solved exacly . we obtain @xmath299dx'\right ) , \label{eq::exact - single}\ ] ] where @xmath300 are some constants . the absence of the reflected wave in ( [ eq::exact - single ] ) means that for any potential shape one has a perfect transmission . thus we conclude that at the point @xmath301 there is a _ total transition _ between electron and hole states since hamiltonians ( [ eq::l0-single ] ) depend on @xmath302 in contrast to ( [ eq::normal ] ) ! for bilayer graphene pseudospin conservation is equivalent to the conservation of particle type , as is seen from the comparison of ( [ eq::lobi ] ) and ( [ eq::normal ] ) . therefore , an incoming particle obeys the schrdinger equation ( [ eq::normal ] ) everywhere . for a `` klein - setup '' this leads to exponentially decaying transmission as a function of a potential width and height . total transmission for normally incident electrons in single layer graphene and its exponential damped behaviour in bilayer have natural explanations of classical phase space ( figure [ fig::phase - space - full ] ) . in both cases the most probable process corresponds to the smoothest trajectory , constructed from separatrix pieces . for the single layer such a trajectory goes through the barrier and gives total transmission , while for bilayer one has to choose the trajectory reflected from the barrier to avoid discontinuity in the second derivative . in section [ sec::standard - semi ] it was discussed that the standard adiabatic diagonalization fails to describe klein tunneling , since it treats electrons and holes separately . however the existence of the exact diagonalization ( section [ sec::normal ] ) for a normal incidence raises the issue of a possible generalization for angular scattering . we shall show that for a single layer graphene there exists an _ exact _ transformation , reducing the original dirac equation to a scalar schrdinger - like equation with a _ complex _ potential . it is clear that such a procedure can not be a unitary transformation of the original hermitian hamiltonian . let us turn back to the dirac equation for single layer and write it in the form @xmath303 where @xmath304 , @xmath305 . let us act on the last equation from the left by the operator @xmath306 . then we get @xmath307\bigr)\psi\nonumber\\ \qquad=\bigl(\hat p_x^2 + p_y^2 - v(x)^2 - ih\sigma_x v'(x)\bigr)\psi=0 . \label{eq::red1}\end{aligned}\ ] ] remarkably , ( [ eq::red1 ] ) contains only the single matrix @xmath298 . therefore it can easily be diagonalized . we write @xmath308 and obtain @xmath309 functions @xmath310 are not independent and the connection formula can be obtained from ( [ eq::sigma - single ] ) . function @xmath311 can be reconstructed from @xmath312 as @xmath313 in figure [ fig::reduction ] one sees the initial potential landscape and the real part of the effective potential in ( [ eq::reduction ] ) . though this equation takes the form of a schrdinger equation , there are two substantial distinctions as compared to a common schrdinger particle : _ i _ ) the effective potential is complex and _ ii _ ) it depends on energy . let us first consider an exacly solvable model for a linear potential @xmath314 , @xmath315 @xcite . introducing a new variable @xmath316 and new @xmath118-component of the momentum @xmath317 we exclude @xmath227 and @xmath318 from ( [ eq::reduction ] ) . then it takes the form ( we omit primes ) : @xmath319 introducing a new variable @xmath119 such that @xmath320 we have : @xmath321 choosing an appropriate value for @xmath322 we can reduce ( [ eq::weber0 ] ) to the weber s equation @xcite @xmath323 indeed , choosing @xmath324 and solving the weber s equation @xcite we obtain @xmath325 where @xmath326 and @xmath327 are the parabolic cylinder functions . for these functions the following identities hold : @xmath328 applying the first equality from ( [ eq::pcyl - id ] ) we find @xmath329 from the second equality in ( [ eq::pcyl - id ] ) we have @xmath330 substituting ( [ eq::par - cyl1 ] ) , ( [ eq::par - cyl2 ] ) into ( [ eq::connection ] ) we obtain : @xmath331 from ( [ eq::eta1 ] ) , ( [ eq::eta2 ] ) we have for @xmath332 : @xmath333 using the asymptotic expansions of the parabolic cylinder functions ( see [ app::parcas ] ) , we find when @xmath334 ( hole region ) : @xmath335,\end{aligned}\ ] ] and when @xmath336 ( electron region ) : @xmath337\nonumber\\ & \pm&c_2\frac{\sqrt{2}e^{-i\pi/4}}{p_y } ( \bar z_1)^{-\nu } e^{ix^2/2},\end{aligned}\ ] ] where @xmath338 , @xmath339 and a bar means complex conjugation . now we turn to the discussion of the scattering problem . while tunneling through the barrier the dirac particle turns from an electron to a hole or vice versa . the of the group velocity of the hole @xmath340 has an opposite sign with respect to its momentum @xmath234 . let us consider an electron , coming from @xmath341 with a positive velocity @xmath342 . it corresponds to the action @xmath343 , since @xmath344 and @xmath345 . thus , the reflected electron corresponds to @xmath346 . the transmitted hole with a positive velocity has a negative momentum @xmath234 . hence it corresponds to the action @xmath347 . from the absence of the incoming wave in the hole region we find @xmath348 . let us consider ( [ eq::psi - refl ] ) at infinity . then we have for the action : @xmath349\right\},\end{aligned}\ ] ] where we assumed that @xmath350 . for large @xmath18 we obtain @xmath351 thus for large negative @xmath18 ( [ eq::psi - refl ] ) reads @xmath352 this gives for the reflection @xmath353 where @xmath354 . using equalities @xmath355 we can write the reflection coefficient as @xmath356 where @xmath357 . from the asymptotic expansion of the @xmath358-function at large arguments @xcite , one concludes that @xmath359 tends to @xmath360 when @xmath228 tends to infinity . at small @xmath228 the reflection coefficient is proportional to @xmath361 . this nonanalytic behaviour is compensated by the jump of the phase of the reflected wave as we discussed in section [ sec::standard - semi ] . comparing the coefficient in front of incoming and transmitted waves we find for the transmission amplitude @xmath362 for the transmission probability we thus have @xmath363 this result was first obtained by cheianov and falko @xcite . considering the scattering from the right to the left we find that the transmission amplitude in this case is @xmath364 the transfer matrix connecting incoming and outgoing waves from the right to the left of the barrier for positive @xmath318 is @xmath365 for negative @xmath318 the transfer matrix reads @xmath366 let us consider an outgoing hole on the right of a generic potential monotonously growing from the left to the right . in semiclassical approximation it is described by the wavefunction @xmath367^{1/4}}e^{-is^+(x)/h } \left(\begin{array}{c}e^{i\phi^+_p(x)/2 } \\ e^{-i\phi^+_p(x)/2 } \end{array}\right ) . \label{eq::psi - out}\end{aligned}\ ] ] using equalities @xmath368^{1/2}}{|e - u(x)|}\end{aligned}\ ] ] we write it as @xmath369^{1/4}\left[v+\sqrt{v^2-p_y^2}\right]^{1/2}}e^{-is^+/h } \left(\begin{array}{c}v+\sqrt{v^2-p_y^2}+ip_y \\ v+\sqrt{v^2-p_y^2}-ip_y \end{array}\right ) . \label{eq::psi - out2}\end{aligned}\ ] ] with @xmath305 . according to ( [ eq::psi - out2 ] ) the components of @xmath370 can be represented _ exactly _ as a sum or a difference of functions @xmath312 , @xmath311 obeying schrdinger - like equations ( [ eq::reduction ] ) with a complex potential . despite the complexity of the potential , according to @xcite , to connect a transmitted wave on the right of the barrier with an incoming wave on the left of the barrier one can still use an analytic continuation in the classically forbidden region similar to @xcite . in contrast to a usual schrdinger equation the transmitted hole has a negative momentum , so @xmath370 in ( [ eq::psi - out ] ) is proportional to @xmath371 , but not to @xmath372 . therefore the passage should be done in the lower complex half - plane . since both functions @xmath312 and @xmath311 allow analytic continuations in the lower half - plane , @xmath370 can also be continued in the lower half - plane . performing the passage and connecting the outgoing wave with an incoming wave , we obtain for the transmission coefficient : @xmath373 where @xmath374 and @xmath375 are two turning points , i.e. solutions for the equation @xmath376 . the standard complex wkb technique @xcite does not allow one to compute the reflection coefficient with an exponential accuracy . the module of the reflection coefficient can be computed from the unitarity of the scattering matrix : @xmath377 . this gives @xmath378 the semiclassical phase of the reflection coefficient can be reconstructed from ( [ eq::refl - np ] ) , ( [ eq::refl - pn ] ) . thus in semiclassical approxiamtion we obtain : @xmath379 where ` - ' corresponds to the electron and ` + ' to the hole region . for small @xmath228 any potential can be linearized in the classically forbidden region . comparing ( [ eq::tranmission ] ) and ( [ eq::tranmission - semi ] ) we see that in the limit @xmath276 the semiclassical transmission becomes exact . therefore ( [ eq::tranmission - semi ] ) can be used as a uniform approximation for the transmission coefficient at any @xmath228 . then the uniform approximation for the reflection coefficient reads @xmath380 where @xmath381 tends to zero when @xmath228 tends to infinity . for the phase @xmath381 we used the expression @xcite @xmath382 which was obtained by the replacement @xmath383 by @xmath384 in @xmath385 . let us consider some generic potential barrier @xmath386 . in graphene it implies n - p - n junction ( see figure [ fig::reduction ] ) . in terms of ( [ eq::reduction ] ) n - p - n junction becomes a complex double - hump potential . the transfer matrix in this case is @xmath387 where we assume that @xmath388 . there is no extra phase coming from @xmath389 since each of these functions takes the same values at both turning points @xmath375 , @xmath390 , lying in the hole region . the transmission coefficient reads @xmath391 the obtained transmission amplitude can easily be treated in terms of a sum of probability amplitudes of multiscattering processes leading to transmission @xcite . one sees that for normal incidence @xmath392 and the module of the transmission coefficient becomes one . transmission resonances can be found from the condition : @xmath393 which coincides with the quantization condition ( [ eq::quantization ] ) for large @xmath228 . for a symmetric n - p - n junction the resonant transmission is always one , since @xmath394 . for an asymmetric junction resonant values of transmission decay as @xmath395 when @xmath396 and @xmath397 . from ( [ eq::t - assym - res ] ) one sees that resonant values of transmission exponentially decay as a function of @xmath398 . such a fast decay can be crucial if one wants to weaken the influence of side resonances . in figures [ fig::single - hump]-[fig::single - asym ] we compare our semiclassical predictions with numerical results , obtained from a multistep approximation of the initial potential . a check on the accuracy of the calculation for constant @xmath228 is provided by the current @xmath399 to simulate an n - p junction we used the potential @xmath400 \\ \label{eq : numericpotential}\ ] ] with a characteristic length scale @xmath401 . an n - p - n junction was simulated as an n - p junction with a characteristic length @xmath401 , a p - n junction with a characteristic length @xmath402 and a constant potential in between of the length @xmath403 . in figure [ fig::single - hump ] we show the comparison of our numerical result for an n - p junction with the semiclassical transmission ( [ eq::tranmission - semi ] ) and the transmission for a linear potential ( [ eq::tranmission ] ) . while the semiclassical prediction works uniformly over the entire range of angles , the prediction obtained from a linear potential works only for small angles . in figure [ fig::double - hump ] we show the comparison of our numerical results with the prediction ( [ eq : pnptransmission ] ) for a symmetric n - p - n junction . we also show the semiclassical result , which is obtained by setting @xmath404 . the agreement between the latter answer and numerics gets better as the angle increases , i.e. deep in the semiclassical regime . the result ( [ eq::tranmission - semi ] ) uniformly approximates the numerical data over the entire range of angles . in figure [ fig::single - asym ] the result for an asymmetric n - p - n junction the height of resonances is seen to decay . the _ suppression of side resonances for asymmetric junctions _ in single layer graphene can have essential consequences for attemps to confine dirac particles ! ) , with @xmath405 nm . the blue line shows the numerical result with 49 steps , the dashed line shows the semiclassical result ( [ eq::tranmission - semi ] ) and the red line shows the result for a linear potential ( [ eq::tranmission ] ) , where the parameter @xmath318 was taken as the derivative at the central point of the junction.,height=264 ] nm and n - p and p - n regions have characteristic lengths @xmath406 nm . the blue line shows the numerical result for 99 steps , the red line shows the uniform approximation ( [ eq : pnptransmission ] ) and the orange line shows the semiclassical answer ( @xmath404).,height=264 ] nm and the n - p and p - n regions have characteristic lengths @xmath407 nm and @xmath408 nm , respectively . the blue line shows the numerical results for 99 steps , while the red line shows the uniform approximation ( [ eq : pnptransmission]).,height=264 ] numerical computations for bilayer graphene using the above procedure are less accurate , due to the presence of real exponentials everywhere . therefore we were unable to check the quantization condition ( [ eq::quantization ] ) numerically with a high precision . to check the accuracy of the computation we used the current @xmath409 in figure [ fig::bilayer ] we show our numerical results for a symmetric and an asymmetric n - p - n junction with the same shape of the potential as before . in contrast to the case of single layer , resonances _ do not seem to decay in this case_. nm . the blue line shows the numerical result for a symmetric junction with @xmath410 nm , while the red line shows an asymmetric junction with @xmath411 nm and @xmath412 nm . all calculations were done with 99 steps per junction.,height=264 ] let us summarize our main results . the detailed analysis of the reflection - and - transmission problem for the dirac electrons demonstrates essential differences from the conventional schrdinger case , due to the role of the berry phase . the reflection coefficient turns out to be nonanalytic function of the transverse momentum @xmath228 vanishing as @xmath361 at @xmath413 . we have presented a complete treatment of the chiral tunneling for both single and bilayer graphene in terms of a classical phase space . this gives a natural explanation of complete transmission of normally incident wave for single layer and its exponentially damped transmission in bilayer . we have also demonstrated that , for the case of nonsymmetric n - p - n junction in single layer graphene , there is total transmission for the normal incidence only , and other maxima are suppressed . our numerical studies show that for the case of bilayer there are always magic angles with total transmission . we are thankful to sergey dobrokhotov , andrey shafarevich , anna esina and andrey shytov for helpful discussions . we acknowledge financial support from the stichting voor fundamenteel onderzoek der materie ( fom ) , which is financially supported by the nederlandse organisatie voor wetenschappelijk onderzoek ( nwo ) . in this part of the appendix we show how to reduce in adiabatic approximation an initial matrix hamiltonian to a set of effective scalar hamiltonians . our consideration follows @xcite . let us consider an eigenvalue problem for an hermitian matrix hamiltonian @xmath414 , @xmath415 where we assume that @xmath416 acts first and @xmath18 acts second . in what follows we always assume this operator ordering . let us introduce a vector operator @xmath417 and a scalar wave function @xmath418 by the requirement @xmath419 we want @xmath418 to satisfy an eigenvalue problem @xmath420 with an effective hermitian hamiltonian @xmath421 . substituting ( [ eq : appeffschrodx ] ) in ( [ eq::stat - sch ] ) we obtain : @xmath422 the last equality will be fulfilled for any @xmath423 if the following operator equality holds : @xmath424 we will solve it by passing to symbols of operators ( see @xcite ) : @xmath425 = a(p_x - ihd / dx , x ) b(p_x , x ) \ ; .\ ] ] applying this formula to the above case , we obtain @xmath426 let us expand this expression with respect to the parameter @xmath229 . in zeroth order we get @xmath427 where @xmath428 , @xmath429 . the first order term in the expansion gives @xmath430 where @xmath431 and @xmath432 are the first order terms in @xmath433 and @xmath434 with respect to @xmath227 . the above expression can be rewritten as @xmath435 let us multiply the last equation by @xmath436 from the left . since both @xmath437 and @xmath438 are hermitian matrices , we obtain @xmath439 where we used the equality @xmath440 . to solve equation ( [ eq : appeffschrodx ] ) we will use the semiclassical ansatz @xmath441 then ( [ eq : appeffschrodx ] ) can then be rewritten as @xmath442 this equation can be expanded order by order in @xmath18 which to zeroth order gives the hamilton - jacobi equation @xmath443 from this equation we can determine the action @xmath444 , as is well known from classical mechanics @xcite . to first order in @xmath227 we obtain the equation @xmath445 where all terms should be evaluated at @xmath446 . when multiplying by the amplitude @xmath447 , equation ( [ eq : appexpfirstcorrectionx ] ) can be rewritten as @xmath448 where the total derivative acts on both @xmath18 and @xmath446 . this equation can be solved exactly to determine the amplitude @xmath449 \\ & \equiv&\left|\frac{\partial l_0}{\partial p_x}\right|^{-1/2}\exp\left[\int dx \left(\frac{\partial l_0}{\partial p_x}\right)^{-1}m\right ] \ ; , \label{eq : appamplx}\end{aligned}\ ] ] where we have defined @xmath450 . using equation ( [ eq : appl1solution ] ) it can be written as @xmath451 we can also solve equation ( [ eq : appeffschrodx ] ) by passing to @xmath270-representation @xcite : @xmath452 in this equation @xmath234 acts first , while @xmath453 acts second , contrary to the case we considered before . to solve this equation , we use the semiclassical ansatz @xmath454 similarly to equation ( [ eq : applongmomentumx ] ) equation ( [ eq : appeffschrodp ] ) can be rewritten as @xmath455 when this equation is expanded order to order in @xmath227 , one obtains to zeroth order the hamilton - jacobi equation @xmath456 from which the action @xmath457 can be determined . the first order term becomes @xmath458 where all terms have to be evaluated at @xmath459 . after multiplication by @xmath460 one finds @xmath461 which can be solved exactly to give @xmath462\\ & \equiv&\left|\frac{\partial l_0}{\partial x}\right|^{-1/2}\exp\left[-\int dp_x \left(\frac{\partial l_0}{\partial x}\right)^{-1}m\right ] \ ; . \label{eq : appamplp}\end{aligned}\ ] ] using equation ( [ eq : appl1solution ] ) the quantity @xmath450 can then be written as @xmath463 since the solutions ( [ eq : appsolutionx ] ) and ( [ eq : appsolutionp ] ) come from the same equation , they should be related . to find out what this relation is , we look at the fourier representation of ( [ eq : appsolutionx ] ) , which is defined by @xmath464 since the parameter @xmath227 is assumed to be small , we can calculate this integral using the stationary phase method . the result is @xmath465 where the point @xmath466 at which the phase is stationary is to be found from the equality @xmath467 from the comparison of ( [ eq : appsolutionp ] ) and ( [ eq : appftsolx ] ) we find @xcite @xmath468 now let us consider the phase space which is shown in figure [ fig : wkbmatlas ] . , @xmath469 , @xmath470 , @xmath471 . regular maps @xmath472 and @xmath470 can be uniquely projected onto the @xmath18-axis , while singular maps @xmath469 and @xmath471 can be uniquely projected onto the @xmath234-axis . the maps are chosen to overlap . , width=377 ] the part @xmath472 of the phase trajectory can be projected onto the @xmath18-axis . therefore we can use a standard wkb ansatz in this region . on the contrary , in region @xmath469 we can use a standard wkb ansatz in @xmath270-representation . since regions @xmath472 and @xmath469 overlap , functions are related according to ( [ eq : apprelationpxrep ] ) . since @xmath473 , we have @xmath474 . it is easily seen that the above reasoning can also be applied to regions @xmath469 and @xmath470 . since @xmath475 we obtain @xmath476 , or @xmath477 so , passing the turning point lying in the region @xmath469 we have picked up an extra factor @xmath478 . since @xmath479 and @xmath480 we pick up another factor of @xmath478 when we go through region @xmath471 and pass the second turning point , @xmath481 in passing one full turn along the circle , one therefore sees that @xmath482 . the wavefunction should be single - valued , which means that the exponent should be a multiple of @xmath483 . from equations ( [ eq : appsolutionx ] ) and ( [ eq : appamplx ] ) we therefore find @xmath484 where @xmath485 is to be found from the hamilton - jacobi equation ( [ eq : apphamiltonjacobix ] ) . the quantity @xmath486 is the berry phase , defined by @xmath487 equation ( [ eq : appprequant ] ) can be rewritten as the bohr - sommerfeld quantization rule @xmath488 looking back at the above derivation one sees that the term @xmath489 can be written as @xmath490 , where @xmath491 is the number of turning points ( maslov index in this particular case ) @xcite . for completeness we placed in this section the asymptotic expansions of the parabolic cylinder functions in different stokes sectors at @xmath492 according to @xcite : @xmath493 ) , & -\pi/2<\textrm{arg}\,(z)\leq\pi/2,\\ e^{-z^2/4}z^\nu(1+o[z^{-2}])-\frac{e^{z^2/4-i\pi\nu}\sqrt{2\pi}z^{-\nu-1}}{\gamma(-\nu)}(1+o[z^{-2 } ] ) , & \textrm{arg}\,(z)\leq-\pi/2\\ e^{-z^2/4}z^\nu(1+o[z^{-2}])-\frac{e^{z^2/4+i\pi\nu}\sqrt{2\pi}z^{-\nu-1}}{\gamma(-\nu)}(1+o[z^{-2 } ] ) , & \textrm{arg}\,(z)>\pi/2 \end{array}\right.\ ] ] we assume that @xmath494 . r. teissier , j. j. finley , m. s. skolnick , j. w. cockburn , j .- l . pelouard , r. grey , g. hill , m .- a . pate , and r. planel . experimental determination of gamma - x intervalley transfer mechanisms in gaas / alas heterostructures . , 54:8329 , 1996 . a. b. kuzmenko , i. crassee , d. van der marel , p. blake , and k. s. novoselov . determination of the gate - tunable band gap and tight - binding parameters in bilayer graphene using infrared spectroscopy . , 80:165406 , 2009 . k. s. novoselov , e. mccann , s. v. morozov , v. i. falko , m. i. katsnelson , u. zeitler , d. jiang , f. schedin , and a. k. geim . unconventional quantum hall effect and berry s phase of @xmath262 in bilayer graphene . , 2:177 , 2006 . j. heading . . methuen s monographs on physical subjects , london and new york , 1962 . russian translation with an appendix by v. p. maslov concerning the wkb method in the multi - dimensional case : mir , moscow 1965 .

we investigate the kinematics of the central gas disk of the radio - loud elliptical galaxy ngc 4335 , derived from hst / stis long - slit spectroscopic observations of h + [ ] along 3 parallel slit positions . the observed mean velocities are consistent with a rotating thin disk . we model the gas disk in the customary way , taking into account the combined potential of the galaxy and a putative black hole with mass , as well as the influence on the observed kinematics of the point spread function and finite slit width . this sets a 3 upper limit of on . the velocity dispersion at is in excess of that predicted by the thin rotating disk model . this does not invalidate the model , if the excess dispersion is caused by localized turbulent motion in addition to bulk circular rotation . however , if instead the dispersion is caused by the bh potential then the thin disk model provides an underestimate of . a bh mass is inferred by modeling the central gas dispersion as due to an isotropic spherical distribution of collisionless gas cloudlets . the stellar kinematics for ngc 4335 are derived from a ground - based ( wht / isis ) long - slit observation along the galaxy major axis . a two - integral model of the stellar dynamics yields . however , there is reason to believe that this model overestimates . reported correlations between black hole mass and inner stellar velocity dispersion predict to be in ngc 4335 . if our standard thin disk modeling of the gas kinematics is valid , then ngc 4335 has an unusually low for its velocity dispersion . if , on the other hand , this approach is flawed , and provides an underestimate of , then black hole masses for other galaxies derived from hst gas kinematics with the same assumptions should be treated with caution . in general , a precise determination of the relation and its scatter will benefit from ( i ) joint measurements of from gas and stellar kinematics in the same galaxies and ( ii ) a better understanding of the physical origin of the excess velocity dispersion commonly observed in nuclear gas disks of elliptical galaxies .

astro-ph0208132

the hubble space telescope has made it possible to measure the masses of black holes ( bhs ) in the centers of many nearby active and quiescent galaxies using stellar and/or gaseous kinematics ( for reviews , see e.g. , richstone et al . 1998 ; van der marel 1999 ; ho 1999 ; de zeeuw 2001 ; kormendy & gebhardt 2001 ) . to date , bh masses have been measured in about 40 galaxies , both spirals and ellipticals , and this number continues to increase . the bh masses correlate loosely with host spheroid luminosity ( kormendy & richstone 1995 ) and more tightly with inner stellar velocity dispersion ( gebhardt et al . 2000 ; ferrarese & merritt 2000 ) . central emission - line gas is detected in virtually all nearby radio galaxies , defined here as galaxies which harbor kpc - scale radio - jets . the equivalent widths of the gas emission lines are generally much larger than those of the absorption lines in the integrated stellar light , so that measurement of the kinematics of nuclear emission - line gas is an efficient way to determine the central gravitational potential and bh mass ( e.g. , harms et al . 1994 ; ferrarese , ford & jaffe 1996 ; macchetto et al . 1997 ; van der marel & van den bosch 1998 ; ferrarese & ford 1999 ; verdoes kleijn et al . 2000 , hereafter vk00 ; sarzi et al . 2001 ; barth et al . . a drawback of this approach is that the gas kinematics might be affected by non - gravitational motions . nevertheless , using the gas kinematics to determine accurate bh masses is particularly interesting for radio galaxies because determining the bh mass and the properties of the gas disk can advance our understanding of radio - jet formation and evolution . in turn , the radio - jet offers an extra diagnostic of the bh accretion and immediate bh surroundings . current questions include : what is the lower - limit to bh masses that can form kpc - scale jets ? is this lower - limit higher or lower than typical bh masses in spirals which ( almost ) never show kpc - scale jets ? is there a correlation between bh mass and jet properties , such as total power or jet velocity ? for these and other reasons we are performing a systematic study of a complete sample of nearby radio galaxies with fanaroff & riley ( 1974 ) type i radio jets , the ` ugc fr - i sample ' ( verdoes kleijn et al . 1999 , hereafter vk99 ; xu et al . 2000 ) using observations at multiple wavelengths . in particular , we have performed a stis spectroscopic survey of the inner gas distributions to measure the kinematics and the physical state of the gas ( noel - storr et al . 2002 ) . here we concentrate on one galaxy from our sample , ngc 4335 , which is a relatively unknown isolated giant elliptical ( @xmath9 ; paturel et al . 1997 ) at 66 mpc . the gas appears embedded in a well - defined dust disk ( diameter @xmath10 ) and gas kinematics can be traced sufficiently far out along the three slits to allow detailed gas dynamical modeling . in addition we perform stellar dynamical modeling for ngc 4335 using the stellar kinematics derived from a wht / isis long - slit observation . the paper layout is as follows . sections [ s : wfpc2 ] and [ s : spec ] present hst / wfpc2 broad- and narrowband imaging , hst / stis gas emission - line spectroscopy and ground - based wht / isis stellar absorption - line spectroscopy , including the basic data reduction and derivation of the gaseous and stellar kinematics . section [ s : modelh ] describes the modeling of the gas disk flux distribution , the derivation of the stellar mass distribution and the fits to the observed gaseous kinematics to estimate the bh mass . section [ s : starkin ] describes two - integral modeling of the wht stellar kinematics to determine the stellar mass - to - light ratio and to constrain the bh mass independently . section [ s : discon ] discusses the implications of the bh mass measurements in ngc 4335 for our understanding of bh demography , and for the techniques used to measure black hole masses . we adopt @xmath11 throughout this paper . this does not directly influence the data - model comparison for any of our models , but does set the length , mass and luminosity scales of the models in physical units . specifically , distances , lengths and masses scale as @xmath12 , while mass - to - light ratios scale as @xmath13 . we imaged ngc 4335 in the context of hst program go-6673 . we used the wfpc2 instrument ( described in , e.g. , trauger et al . 1994 ; biretta et al . 1996 ) on march 15 , 1997 to obtain images of ngc 4335 in the f555w ( @xmath14 ) and f814w ( @xmath15 ) filters and linear ramp filter fr680n which includes h@xmath0 + [ ] . the observing log is presented in table [ t : wfpc2 ] . the ` linear ramp filters ' ( lrfs ) of the wfpc2 have a central wavelength that varies as a function of position on the detector . the lrf fr680n image ( wf ccd ) was used as ` on - band ' image , with the galaxy position chosen so as to center the filter transmission on the h@xmath0 + [ ] emission lines . a combination of the broad - band images f555w and f814w ( pc ccds ) was chosen as ` off - band ' image and covers primarily stellar continuum . this combination of filters ensures that the effect of dust extinction is the same in the on- and off - band image . the final h@xmath0 + [ ] emission image was obtained by subtracting the stellar continuum , as determined by the off - band image , from the on - band image . the quality of the h@xmath0 + [ ] image is set by the poisson noise of input images and the uncertainty in the alignment and scaling of the on - band image and off - band image . the stellar surface brightness is steeply rising at the center of ngc 4335 . the resulting emission image depends quite sensitively on precise sub - pixel alignment and scaling . this introduces a @xmath16 flux error which is estimated by varying the alignment and scaling within a reasonable range . the @xmath14-band and h@xmath0 + [ ] images are shown in figure [ f : images ] . extended h@xmath0 + [ ] emission is present with a roughly elliptical morphology , elongated in the same direction as the dust disk . as described in section [ s : flux_gas ] , the emission - line gas is consistent with being in a thin disk , but the uncertainties involved in constructing the image are too large to prove this conclusively . we can not rule out a more spheroidal distribution , especially towards the nucleus . these data were presented previously in vk99 , who also give a more detailed description of the data reduction . the central dust detected in ngc 4335 extends to more than 2 kpc from the nucleus . in the outer regions ` arms ' of dust gradually align with the well - defined central dust disk which appears as roughly half an ellipse on the west side of the nucleus ( see fig . [ f : images ] ) . this is commonly seen in galaxies with inclined nuclear dust disks ( e.g. , ngc 7052 , van der marel & van den bosch 1998 ; ngc 6251 , ferrarese & ford 1999 ; 3c 449 , martel et al . on the east side no dust obscuration is evident in either the @xmath14- or @xmath15-band image and its @xmath17 color is typical for dust - free giant elliptical galaxies . we conclude that the eastern half of the dust disk produces negligible obscuration in the optical . we fitted by eye an elliptical contour to the outline of the dust disk in the @xmath14-band image , and measured the position angle of the major axis of the dust disk to be pa=@xmath18 . this is well - aligned with the galaxy major axis just outside the dust disk ( cf.fig . [ f : images ] and vk99 ) . if we assume that the dust disk is thin and intrinsically circular , then the inclination is @xmath19 ( vk99 ) . the total mass of the kpc - scale dust distribution estimated using the observed color excess is @xmath20 if the dust resides in the mid - plane of the galaxy ( vk99 ) . following the prescription by goudfrooij & de jong ( 1994 ) , the iras @xmath21 and @xmath22 flux measurements for ngc 4335 from knapp et al . ( 1989 ) indicate a dust mass @xmath23 . this is not inconsistent with the aforementioned result , given that the iras measurements cover a much larger area than that occupied by the nuclear dust disk studied with hst . furthermore , dust mass estimates using color excess are not sensitive to a smoothly distributed dust component , if present ( cf . e.g. , tran et al . in the context of go - program 8236 , we used stis ( see kimble et al . 1998 ) on september 22 , 1999 to obtain long slit spectra of ngc 4335 during two orbits of the hst telescope . we obtained spectra along three adjacent slit positions using a @xmath24 wide slit ( 52x0.2 ) . the lay - out of the slits is shown in figure [ f : images ] . the g750 m grating was used in combination with two pixel spatial and spectral on - chip rebinning ( in order to obtain enough signal to noise while retaining sufficient spatial and spectral resolution ) yielding spectra covering the wavelength range from 6436 to 7100 over 511 pixels . wavelength calibration spectra of the internal arc lamp were obtained at the beginning of the first orbit and at the conclusion of the second orbit . the slits were aligned along pa=@xmath25 . this is almost parallel ( @xmath26pa=@xmath27 ) to the galaxy major axis ( pa=@xmath18 ) . we denote the central slit by c , and the adjacent slits on the east- and west - side by e and w , respectively . a log of the observations is provided in table [ t : stissetup ] . target acquisition uncertainties and other possible systematic effects could cause the slit positions on the galaxy to differ slightly from those commanded to the telescope . we determined the actual slit positions from the data themselves , using the stis continuum and emission - line fluxes and the wfpc2 images . we compared the continuum counts as function of position along the slit for the three stis slits with the counts in the wfpc2 @xmath14 and @xmath15 images at these positions . the stis continuum was determined blue - ward of h@xmath0 + [ ] and red - ward of [ ] respectively . this analysis indicates that the positioning errors for the stis slits with respect to the wfpc2 image are less than @xmath28 in both the direction along and perpendicular to the slit . another estimate of the positioning of the slits with respect to the galaxy nucleus was obtained by interpolation of the central few h@xmath0 + [ ] fluxes assuming the maximum flux occurs at the galaxy nucleus . this analysis implies that the central slit was properly centered on the nucleus to an accuracy of @xmath29 . the small slews performed by the telescope between observations to dither the slits and observe the adjacent slits have a nominal positioning error of @xmath30 . most of the necessary data reduction steps were performed by the hst / stis calibration pipeline ( calstis version 2.3 ) , including flat - fielding , hot pixel and cosmic - ray removal , absolute sensitivity calibration and wavelength calibration which takes into account the earth s motion . to facilitate hot pixel and cosmic - ray removal two exposures were taken at each slit position dithered by 2 pixels in the direction along the slit . the uncertainty in the relative wavelength accuracy is @xmath31 or @xmath32 ( diaz - miller 2001 ) . the spectra show several emission lines , of which the h@xmath0 + [ ] complex at 6548 , 6563 , 6583 has sufficiently high signal - to - noise ratio @xmath33 for a kinematical analysis . the [ ] doublet at 6716 and 6731 can be fitted only in the central @xmath34 in the central slit and at a lower signal - to - noise ratio . the results for these few points are consistent with the h@xmath0 + [ ] kinematics described below . to quantify the h@xmath0 + [ ] gas kinematics we fitted the spectra assuming that each emission line is a gaussian on top of a flat continuum . this yields for each line the total flux , the mean velocity @xmath14 and the velocity dispersion @xmath2 . we fitted the [ ] doublet under the assumption that the individual lines have the same @xmath14 and @xmath2 and that the ratio of their fluxes equals the ratio of their transition probabilities ( i.e. , 3 ; see references in mendoza 1983 ) . the h@xmath0 + [ ] complex is influenced by blending of the lines , and we made the additional assumptions that h@xmath0 and the [ ] doublet have the same kinematics . we did the gaussian fitting using software described in van der marel & van den bosch ( 1998 ) . prior to the emission - line fit the spectra were rebinned along the slit where necessary to obtain sufficient @xmath33 , and hence sufficiently small errors on the gas kinematics ( table [ t : gaskin ] ) , to discriminate between the black hole masses in dynamical models discussed in section [ s : dyn_mod ] and later . the same gaussian fitting was performed on modeled emission - line spectra which resulted from the dynamical models . differences between the velocity profiles ( vp ) and the single gaussian fits are discussed in section [ s : dyn_mod ] . the gaussian fits to the emission lines are shown in figure [ f : emisfit ] . the resulting kinematics are shown in figure [ f : gaskin ] and tabulated in table [ t : gaskin ] . the systemic velocity used in figure [ f : gaskin ] and table [ t : gaskin ] was estimated from the hst data themselves , by including it as a free parameter in the dynamical models described below ( see section [ s : dyn_mod ] ) . this yields @xmath35 . this result is consistent with @xmath36 by huchra et al . ( 1983 ) and @xmath37 from the rc3 catalog ( de vaucouleurs et al . 1991 ) . the mean - velocity profiles are smooth functions of position and are indicative of a rotating circular disk of emission gas . the rotating disk picture is further supported by the fact that the mean velocities on the minor axis are close to @xmath38 in all three slits . the velocity dispersion profiles in all three slits show irregularities . we verified that this result does not depend sensitively on chosen region to fit the continuum and masked deviant pixels during the fit . the velocity dispersion peaks at @xmath39 at the nucleus in the central slit c. in slit w the dispersion peaks near the nucleus . however , in slit e the dispersion actually dips at the position closest to the nucleus . it seems therefore that the kinematics of ngc 4335 are consistent with a rotating gas disk , but with additional turbulent motion . we also performed single gaussian fits to h@xmath0 and the [ ] doublet independently . the two resulting flux profiles have very similar shapes . the mean velocities for the two sets of emission lines are very similar as well . the median difference between the mean velocities of the two components is @xmath40 . this is slightly larger than the typical formal errors given in table [ t : gaskin ] . these formal errors are derived by propagating the spectral flux errors . we will assume @xmath41 as a more realistic error estimate for the h@xmath0 + [ ] mean velocities . the h@xmath0 and [ ] velocity dispersion profiles show very similar behavior , but h@xmath0 has systematically lower velocity dispersions by about @xmath42 . the differences in dispersion could be caused by the fact that ( i ) the h@xmath0 and [ ] have intrinsically different kinematics or ( ii ) the h@xmath0 kinematic measurements are affected by stellar h@xmath0 absorption . a similar effect and of the same magnitude is seen for the gas disk in the s0 galaxy ngc 3245 ( barth et al . 2001 ) . a long - slit stellar absorption spectrum at pa@xmath43 ( i.e. , within @xmath44 from the major axis of ngc 4335 ) was obtained in service mode with the isis ( blue arm ) spectrograph on the 4.2 m william herschel telescope on la palma on june 13 , 2001 . a slit - width of @xmath45 was used with the r300b grating and the eev12 ccd detector . the dispersion is 0.86 / pix and the spatial scale is @xmath46/pix . the observed wavelength range was @xmath4734306900 . the observations included two 1800s exposures of ngc 4335 , a 60s exposure of the b2ivp star bd+33 2642 to facilitate flux calibration , and standard calibration exposures . the instrumental velocity dispersion @xmath48 was determined from sky lines and the emission - line widths in the calibration lamp spectra . both the star and galaxy spectra were observed with a seeing of fwhm @xmath47@xmath49 . the data reduction was performed in iraf using the ccdproc ( version dec93 ) and twodspec packages . the basic data reduction steps include bias subtraction , flat - fielding with dome - flats and cosmic - ray removal . the contribution from dark current is negligible . the wavelength calibration was done with contemporary cuar lamp spectra taken directly before and after the observations of ngc 4335 . either 31 or 32 arc - lamp emission lines were used to map the wavelength as a function of ccd coordinate . the final residuals per fitted emission line are @xmath50 ( or @xmath51 at 5000 ) . we subsequently performed the wavelength calibration and spatial rectification to align the wavelength direction with the ccd columns in one step . the flux calibration took into account telescope throughput and atmospheric extinction . the galactic extinction towards ngc 4335 is @xmath52 ( schlegel et al . 1998 ) and hence negligible . as a cross - check for the flux calibration we verified that the broad - band flux observed through a @xmath53 central aperture in the wht / isis spectrum and our previously observed hst @xmath14-band image ( vk99 ) agreed to within @xmath54 , correcting for differences in filter passbands and point spread function ( psf ) this level of agreement is satisfactory given the uncertainties in wht / isis seeing conditions and extinction corrections . finally , after verifying that the final calibrated spectra of stars and galaxies did not have any remaining large - scale patterns , we logarithmically rebinned in the wavelength direction to perform the kinematical analysis . we used the wavelength range @xmath555700 in the ngc 4335 absorption spectrum to derive the stellar kinematics . this range includes the red - shifted mg @xmath56 , fe@xmath57 and fe@xmath58 absorption features . to obtain the stellar kinematics we used the pixel space fitting method of van der marel ( 1994 ) , which compares the ngc 4335 spectrum to a stellar ` template ' spectrum . a suitable template star ( spectral type k0v ) was obtained during a different observing run with wht / isis in identical setup . to check for template mismatch we also used stellar templates ( spectral types k1iii and k2 ) described in van der marel & franx ( 1993 ) . these data have an instrumental velocity dispersion @xmath59 , i.e. , very similar to the @xmath60 of the ngc 4335 spectrum . the inferred stellar kinematics agree within the errors . the observed stellar kinematics as a function of radius are listed in table [ t : starkin ] ( and also shown in figure [ f : starkin ] which is discussed in section [ s : starkin ] ) . stars and gas rotate in the same direction in the core of ngc 4335 . the galaxy spectrum integrated along the slit over the central @xmath61 yields an aperture velocity dispersion @xmath62 . the @xmath61 length corresponds to two effective radii as determined by fitting the surface brightness distribution shown in figure [ f : sbprof ] . the listed formal errors include photon noise , uncertainties in the continuum fitting and the variation in derived @xmath63 observed by using different stellar templates . we are not aware of previous determinations of the central velocity dispersion for ngc 4335 , but note that the central value @xmath64 inferred from the faber & jackson ( 1976 ) relation agrees well with the observations . to model the h@xmath0 + [ ] gas kinematics we need a description of the intrinsic ( i.e. , deconvolved and de - inclined ) emission - line flux distribution . we use fluxes from both the wfpc2 h@xmath0 + [ ] image and stis spectra to constrain the intrinsic distribution . figure [ f : fluxcompare ] compares the actual flux distribution observed through the stis slits with the expected flux distribution through the stis slits as inferred from the wfpc2 image . the agreement for slits c and w is good considering the uncertainties in the absolute flux calibrations ( especially of the lrf filter ) . for slit e there is a clear discrepancy , but this does not affect the inferred bh mass ( see section [ s : bestfit ] ) . we model the ( face - on ) intrinsic flux profile as a double exponential , @xmath65 and assume that the disk is infinitesimally thin and viewed at an inclination @xmath66 , ( cf . section [ s : wfpc2_ana ] ) . the total flux contributed by each of the two exponential components is @xmath67 ( @xmath68 ) , and the overall total flux is @xmath69 . the best - fitting parameters of the model flux distribution were determined by a simultaneous fit to the wfpc2 and stis h@xmath0 + [ ] fluxes . we use a @xmath70 square region of the h@xmath0 + [ ] image centered on the nucleus . the flux errors take into account both poissonian errors and a @xmath71 error due to uncertainties in the alignment and subtraction of the on- and off - band image ( see section [ s : wfpc2_red ] ) . from the stis observations we use the h@xmath0 + [ ] flux of the gaussian emission - line fits ( section [ s : spec_ana ] ) . the fluxes from gaussian fits agree within the errors with fluxes estimated by directly measuring the spectral lines . the stis flux errors take into account readnoise and poissonian errors . typically , the stis fluxes have much smaller relative errors than the wfpc2 fluxes . we perform an iterative fit of the double exponential to the h@xmath0 + [ ] flux data , taking into account flux errors and the necessary convolutions with the appropriate psf , pixel size and aperture size for each setup . the diamonds in figure [ f : fluxfit ] show the predictions of the model that best fits all available data simultaneously . this model has parameters @xmath72 , @xmath73 and @xmath74 . the absolute calibration gives @xmath75 for h@xmath0 + [ ] . the lowest panel in figure [ f : fluxfit ] shows the intrinsic ( i.e. , deconvolved ) flux distribution as a function of radius . the thin disk flux model provides a decent fit to the observed stis fluxes , apart from a slight under - prediction of the central fluxes in slit w. the wfpc2 measurements are also reasonably well - fitted , although the observed flux profile on the major axis suggests a narrower central flux peak . a second fit was made as well in which we included only the wfpc2 data . this resulted in a considerably narrower intrinsic profile ( figure [ f : fluxfit ] ) . however , the two model emission - flux distributions result in bh masses which agree to within @xmath54 as discussed in section [ s : bestfit ] . for the purpose of dynamical modeling we parameterize the three - dimensional stellar luminosity density @xmath76 . we assume that @xmath76 is oblate axisymmetric , that the isoluminosity spheroids have constant flattening @xmath77 as a function of radius , and that @xmath76 can be parameterized as @xmath78^{\beta } , \quad m^2 \equiv r^2 + z^2 q^{-2 } . \label{e : lumdendef}\ ] ] here @xmath79 are the usual cylindrical coordinates , and @xmath0 , @xmath80 , @xmath81 and @xmath82 are free parameters . when viewed at inclination angle @xmath83 , the projected intensity contours are aligned concentric ellipses with axial ratio @xmath84 , with @xmath85 . the projected intensity for the luminosity density @xmath76 is evaluated numerically . we assume @xmath86 as discussed in section [ s : wfpc2_ana ] . we take @xmath87 within a radius of @xmath88 based on the isophotal analysis of vk99 ( their figure 2 ) . they determined that the isophotes are very close to elliptical and that @xmath84 varies from @xmath89 to @xmath90 between major - axis radius @xmath91 and @xmath92 . the isophotal major - axis pa=@xmath18 is nearly constant , varying by less than @xmath93 over the same radius range . a model with constant @xmath77 and pa is expected to be adequate in view of the main uncertainties for the bh mass from our dynamical modeling as discussed in section [ s : discon ] . for example , we verified that the variation in ellipticity changes the derived circular velocities by @xmath94 , which is small compared to our velocity errors . we determine the mean flux profile over three rows ( corresponding to a width @xmath95 ) along the unobscured eastern galaxy semi - minor axis ( cf . section [ s : wfpc2_ana ] ) . we use the @xmath15-band fluxes as opposed to @xmath14-band fluxes because the flux contribution of emission - lines is negligible in this wavelength region . the observed surface brightness profile is shown in figure [ f : sbprof ] . the projected intensity profile of the best - fit model using the fluxes on the semi - minor axis between @xmath96 and @xmath97 is shown by the solid curve . this model has @xmath98 , @xmath99 , @xmath100 and @xmath101 . the surface mass density of the dust disk , which is proportional to the optical extinction ( cf . vk99 ; and references therein ) , does not show a strong increase towards the center but varies by just about a factor of two over the disk ( cf . figure 2 in vk00 ) . taking a galactic value of the gas - to - dust ratio of 100 ( e.g. , bohlin , savage & drake 1978 ) , the total mass of the disk , which extends over a region @xmath102 is @xmath103 . the ratio of enclosed disk mass to stellar mass is at most @xmath104 in the range @xmath105 $ ] ( i.e. , @xmath106 $ ] . hence we can neglect the gravitational potential exerted by the gas and dust disk . there is one complication in estimating the stellar mass profile . the central @xmath107 of the galaxy has a color @xmath108 on both the eastern and western side of the galaxy nucleus . this color is typical of dust - free giant ellipticals . it implies that the dust either does not extend all the way to the center ( i.e. , has a ring structure ) or becomes optically thick towards the center . in the latter case it would hide approximately half of the stellar mass in this region as the dust is very close to the center . we discuss this scenario in section [ s : bestfit ] and find that it requires an unusually low stellar mass - to - light ratio to explain the observed gas kinematics . our thin - disk models for the gas kinematics are similar to those employed in van der marel & van den bosch ( 1998 ) and vk00 . the galaxy model is axisymmetric , with the stellar luminosity density @xmath109 chosen as in section [ s : mass_stars ] to fit the available surface photometry . the stellar mass density @xmath110 follows from the luminosity density upon the assumption of a constant mass - to - light ratio @xmath111 . @xmath111 is included as a free parameter in the dynamical model . we assume that the gas is in circular motion in an infinitesimally thin disk in the equatorial plane of the galaxy , and has the circularly symmetric flux distribution @xmath112 given in section [ s : flux_gas ] . thus the position angle of the gas disk is set by the position angle of the galaxy which is known to an accuracy of @xmath113 . the inclination @xmath83 of the gas disk is assumed to be identical to the dust disk inclination . below , we will verify pa and @xmath83 also from the two - dimensional gas kinematics independently . the circular velocity @xmath114 is calculated from the combined gravitational potential of the stars and a central bh of mass @xmath1 . the line - of - sight velocity profile of the gas at position @xmath115 on the sky is a gaussian with mean @xmath116 and dispersion @xmath117 , where @xmath118 is the radius in the disk . the velocity dispersion of the gas is assumed to be isotropic , with contributions from thermal and non - thermal motions : @xmath119 . for practical reasons , we refer to the non - thermal contribution as ` turbulent ' , although its origin is not certain ( see section [ s : dispersion ] ) . we parameterize @xmath120 through : @xmath121 . \label{e : turbdef}\ ] ] this functional form is meant merely to provide a fit to the observed dispersion and is not based on an underlying physical mechanism for the turbulence . the temperature of the gas is not an important parameter : the thermal dispersion for @xmath122 is @xmath123 , and is negligible with respect to @xmath124 for all plausible models . the predicted vp for any given observation is obtained through flux - weighted convolution of the intrinsic vps with the psf of the observation and the size of the aperture . the stis psf is represented as a sum of 5 gaussians ( see table [ t : psfstis ] ) . we ignore the effect of ` pixel - bleeding ' for the stis psf , since this changes the kinematics predicted by the model only by a few @xmath125 ( barth et al . the convolutions are described by the semi - analytical kernels given in appendix a of van der marel ( 1997 ) , and were performed numerically using gauss - legendre integration . a gaussian is fit to each predicted vp for comparison to the observed @xmath14 and @xmath2 . we define a @xmath126 quantity that measures the quality of the fit to the kinematical data , and the best - fitting model was found by minimizing @xmath126 using a ` downhill simplex ' minimization routine ( press et al . 1992 ) . as mentioned , the observed vp and the vp predicted by the dynamical modeling are fitted with a single gaussian . thus we have to verify that the deviations from a gaussian are the same in observations and model . figure 2 shows that in some spectra , especially near the galaxy nucleus , the emission lines tend to have smaller peaks and broader wings than the gaussian fit . the same trend is indeed seen in the modeled spectra . we verified that , apart from this difference , there are no systematic differences as a function of radius between the vp and the gaussian fit . furthermore , the modeled spectra allow us to determine the difference between the moments of the full vp and the gaussian fit qualitatively , also for blended h@xmath0 + [ ] emission lines , as is the case in many spectra . the representation by single gaussian fits turns out to be also quantitatively adequate in the sense that the differences between true and gaussian @xmath127 and @xmath2 of the vps are typically within the quoted errors for the observed gaussian moments . we first analyze the h@xmath0 + [ ] gas disk mean velocities and defer discussion of the velocity dispersions to section [ s : dispersion ] . we start out with models with no bh in which the gas disk has the orientation as determined from the photometry . thus the gas disk is assumed to have the same inclination @xmath66 and position angle pa=@xmath18 as the dust disk and galaxy ( see section [ s : wfpc2_ana ] ) . this corresponds to an intrinsic axial ratio of @xmath128 for our spheroid model . the stis long - slit observations were performed at a small angle @xmath129 to the major axis ( cf . section [ s : spec_red ] ) . first we vary the @xmath15-band mass - to - light ratio @xmath130 and find that a model with @xmath131 in solar units fits the observed gas kinematics best ( see figure [ f : mbhrange ] ) . we will refer to this model as the ` standard model ' . there are some differences between the model and observations . for slit e , the model predicts a steeper mean velocity gradient than observed . for slit w , the observed mean velocities around @xmath132 are systematically larger than predicted by the model . the total and reduced @xmath126 of this model with one free parameter are @xmath133 and @xmath134 , respectively . however , almost @xmath135 of the total @xmath126 is contributed by just the two points at @xmath136 in slit w and at @xmath137 in slit e. these two large deviations might be caused by local deviations from the model , for instance local turbulence . in fact , at the latter location the velocity dispersion shows an off - nuclear peak . discarding these two points , the reduced @xmath138 . thus given the idealization of the model , a rotating disk model with localized turbulence is not too unlikely . we can estimate @xmath130 independently from existing empirical correlations between @xmath111 and galaxy optical luminosity . for ngc 4335 , @xmath139 ( paturel et al . 1997 ) and we assume @xmath140 , @xmath141 and @xmath142 , which is typical for bright ellipticals and consistent with our wfpc2 observations ( cf . we derive @xmath143 using the relation between @xmath144 and absolute @xmath145 magnitude determined by van der marel ( 1991 ) . the value @xmath146 is found using the correlation between @xmath147 and absolute @xmath14 magnitude as determined by magorrian et al . ( 1998 ) . the two values are consistent because both correlations show scatter of a factor @xmath148 in the observed mass - to - light values at given host luminosity . the value @xmath149 derived from the gas kinematics falls nicely in the expected range @xmath150 . the inclination of the galaxy and its embedded gas disk are assumed to be similar to the larger scale dust disk . which @xmath83 is preferred by the gas disk kinematics alone ? to address this question , we derive the stellar mass model and intrinsic emission - line flux model assuming @xmath151 ( @xmath152 ) and @xmath153 ( @xmath154 ) . the profile shape of our stellar mass model ( section [ s : mass_stars ] ) does not depend on inclination but the central stellar mass density changes , i.e. , it increases for decreasing inclination . the changes in the intrinsic emission - line flux profile are negligible . for a given @xmath130 , the increase in central stellar density increases the predicted velocities , while the decrease in inclination decreases the line - of - sight component of the predicted velocities . thus these two effects compete . as figure [ f : upsilon ] shows , compared to the fit for @xmath66 the fits are poorer for @xmath151 and @xmath153 and the best fit @xmath155 is higher than expected from the empirical correlations with host luminosity . the pa of the gas disk is fixed in the model to be identical to that of the galaxy major axis , which is determined to within @xmath113 by the wfpc2 isophotal analysis . we have varied the pa@xmath156 of the gas disk major axis by @xmath157 , and find that the circular disk model fits best for pa@xmath158 , i.e. @xmath159 smaller than the pa of the galaxy major axis . the @xmath126 increases by @xmath160 for pa@xmath156 which are @xmath161 away from pa@xmath162 . thus the pa of the gas disk as preferred by the gas kinematics agrees well with the pa derived from the wfpc2 stellar isophotes . in summary , we see that the kinematics of the gas disk support the disk inclination and pa and the mass - to - light ratio @xmath130 implied by independent methods . the slope of the central velocity measurements for each slit are well fitted by this model with no bh . this a posteriori validates our inclusion of the central velocity gradient to determine @xmath130 because this gradient is evidently not affected significantly by the central bh mass . it also implies that the data provide an upper limit to the bh mass . we now determine this bh mass upper limit . the models indicate that only the gradient between the central three points closest in absolute radial distance from the nucleus ( i.e. , at @xmath163 in slit c ) are significantly affected by a bh with mass @xmath164 . we will therefore use these mean velocities to compute the @xmath126 . since the model parameters other than @xmath1 are optimized to fit the whole velocity profile , the @xmath126 from the central three mean velocities is expected to follow approximately a @xmath126 probability distribution with @xmath165 degrees of freedom ( 3 velocity measurements and 1 free parameter , @xmath1 ) and hence an expectation value @xmath166 . in agreement with this the standard model has @xmath167 . figure [ f : chisquare ] shows the increase in @xmath126 as a function of increasing bh mass . from the @xmath168 for the one - parameter model , a @xmath169 is formally ruled out at the 3 @xmath2 level . a more conservative upper - limit can be obtained by assuming the gas disk is maximally face - on . this will decrease the inferred intrinsic axis ratio of the galaxy ( see section [ s : mass_stars ] ) . since no giant ellipticals flatter than e6 are observed , we take @xmath152 as the minimum acceptable axis ratio for ngc 4335 . this implies a galaxy inclination @xmath151 . figure [ f : chisquare ] shows that a @xmath170 is ruled out at more than the @xmath171 level for @xmath151 . such a model can not fit the complete velocity curves as well as the standard model for any @xmath130 ( figure [ f : upsilon ] ) . moreover , the best fit is provided by @xmath172 for @xmath151 which is at least @xmath135 larger than expected , as discussed above . section [ s : flux_gas ] shows that the wfpc2 flux profile suggests a narrower flux profile than the stis data . we performed the gas dynamical modeling also with the double exponential fit to the wfpc2 data only . this results in a @xmath54 higher upper - limit to the inferred bh mass . the other notable discrepancy between the wfpc2 and stis fluxes in slit e does not influence the bh mass measurement directly . in section [ s : mass_stars ] we discussed the possibility that there might be a central dust distribution inside @xmath173 which is completely opaque . this will alter the assumed stellar mass density , because inside @xmath174 approximately half of the galaxy light will be obscured . we determine the stellar mass density assuming the intrinsic stellar flux inside @xmath174 is two times the observed flux and assume that the obscuration by dust is negligible outside @xmath175 . the fitted model has @xmath176 , @xmath177 , @xmath178 and @xmath179 . in this case a mass - to - light ratio @xmath180 is needed to obtain a good fit to the galaxy - dominated kinematics . this is an unusually low @xmath130 and lies well outside the observed scatter of observed absolute magnitude vs. @xmath130 correlations ( cf . section [ s : bestfit ] ) . moreover , the emission - line flux is strongly peaked towards the center which suggests we have a direct view of the nucleus without significant obscuration by dust . we conclude that ngc 4335 probably does not have an opaque central dust distribution . if however it is present , then our upper limit on the bh mass is an overestimate . in conclusion , there is good agreement between the disk inclination @xmath83 and pa as determined from gas kinematics and dust morphology and between the mass - to - light ratio @xmath130 from gas kinematics and from independent methods . these agreements form support for the thin disk model , which indicates @xmath181 . the quantities @xmath111 , @xmath83 and pa are mainly constrained by the gas kinematics at @xmath182 , which is outside the bh radius of influence @xmath183 ( with @xmath184 , and @xmath2 the central stellar velocity dispersion ) . the upper limit to the bh mass is determined from the velocity gradient at @xmath185 . the inferred bh mass is thus based on the assumption that the thin disk model holds for @xmath186 . the next section considers if this assumption is supported by the gas velocity dispersions . the velocity dispersion is only roughly fitted by a model of the form ( [ e : turbdef ] ) with @xmath187 , @xmath188 and @xmath189 ( figure [ f : turb ] ) . the observed velocity dispersion varies between @xmath190 and @xmath191 ( figure [ f : turb ] ) . the value of @xmath192 corresponds to the instrumental and thermal line broadening . in the absence of a bh , the induced line broadening due to differential rotation over the apertures increases the predicted dispersion to at most @xmath193 . the presence of a bh with mass @xmath3 would only affect the central aperture in each slit , increasing the predicted dispersions to @xmath194 . thus for many apertures , the gas shows a larger velocity dispersion than predicted by the dynamical model of a thin rotating disk . also a double gaussian fit to each emission - line , to fit a broad and a narrow component of the emission - line , yields a narrow - line component with a velocity dispersion in excess of that expected from the dynamical model ( e.g. , @xmath195 at the nucleus ) . thus , the ` excess ' velocity dispersion can not be ascribed easily to the presence of a second kinematic component in addition to the thin rotating disk . an excess velocity dispersion increasing towards the nucleus , i.e. , similar to seen here , is often observed in other gas disks ( e.g. , van der marel & van den bosch 1998 ; vk00 ; barth et al . 2001 ) , with m 87 being a notable exception ( macchetto et al . significant excess dispersion is only absent at @xmath196 . these points are well - fitted by our model which is strong evidence that ( i ) the model of a thin disk in rotation applies at these radii and ( ii ) the model parameters that depend sensitively on these points , i.e. , @xmath130 , pa , and @xmath83 , are trustworthy regardless of the effects of excess velocity dispersion observed closer to the nucleus . what causes the excess velocity dispersion ? is it gravitational or non- gravitational in origin ? a gravitational origin would imply that the gas rotates at speeds lower than the circular velocity . only if the excess in velocity dispersion is modest , @xmath197 , do approximate formulae for this effect of ` asymmetric drift ' ( e.g. , binney & tremaine 1987 ) allow the circular velocity to be derived from the observed mean velocity ( e.g. , barth et al . 2001 ) . this regime does not apply to the observed kinematics in the central regions if interpreted as asymmetric drift . unfortunately , the bh mass depends exclusively on these nuclear apertures . it is therefore useful to determine the bh mass in the most extreme case that all dispersion in the central spectrum is gravitational . one option is that the gas forms a spheroidal distribution of collisionless cloudlets . to first approximation we can model the gas assuming a spherical isotropic model and solve the jeans equations in similar fashion as done in vk00 . a black hole mass @xmath198 is then required to produce the observed @xmath199 . the model is simplistic and deviations from axisymmetry and isotropy can change the result by factors of a few ( de bruijne , van der marel , & de zeeuw 1996 ) . moreover the model assumes a constant number density to flux density ratio for the emission - line flux . different ionization mechanisms ( e.g. , central point source or shocks ) quite likely produce quite different relations between number density and emission - line emissivity as a function of radius . a second option is to assume that asymmetric drift is present _ except _ at the very center . the central velocity dispersion is then due entirely to differential rotation over the aperture . also in this case , a @xmath200 is required to produce the central @xmath199 for @xmath66 and @xmath131 . thus if the thin disk model breaks down within the central pixel we might have significantly under - predicted the bh mass . the thin disk model could be a good approximation if the excess velocity dispersion is caused by localized turbulence , which leaves the gas on average in circular rotation . the highly irregular behavior of the velocity dispersion points indeed to a non - gravitational origin . the velocity dispersion of the gas peaks towards the nucleus in slit c and w , but dips towards the nucleus for slit e. interestingly , the location of irregular peaks in the velocity dispersion profile seems to coincide with deviations of the observed mean velocities compared to the model , especially in slit e. the relative flux of [ ] 6583 and h@xmath0 can shed light on the excitation mechanism . figure [ f : lineratio ] shows that the flux ratio [ ] 6583 / h@xmath0 roughly peaks towards the nucleus in all slits . this is qualitatively in agreement with the excess dispersion being due to turbulent shocks as they tend to enhance the emission - line ratio ( dopita et al . interestingly , the maximum line ratio ( [ ] 6583 / h@xmath0 @xmath201 ) in fact occurs at the location of the dip in the velocity dispersion in slit e ( and a possible dip in slit w ) . no special feature is detected in the wfpc2 image ( figure [ f : images ] ) at these locations . however , the spine of the straight radio - jet observed at arcsecond - scale resolution with the vla ( pa=@xmath202 ; wrobel , machalski & condon 2002 ; xu et al . 2000 ) passes right through the most conspicuous dip in slit e ( between @xmath203 and @xmath204 ) if it continues straight down to the sub - arcsecond scales of interest here . unfortunately , no radio jet emission is detected at vlba scales ( xu et al . 2000 ) . surprisingly , the velocity gradient in slit e is regular and only slightly shallower than predicted by our standard model . thus it seems that the irregular excess in dispersion does not significantly alter the mean rotational velocity . the irregularity of the velocity dispersion profile and its correspondence to changes in the [ ] 6583 / h@xmath0 flux ratio favor a non - gravitational origin for the excess velocity dispersion . the irregularities do not correspond to irregularities in the gas mean velocities . instead , the mean velocities suggest normal rotation . these observations are consistent with a gas disk for which the bulk of the gas orbits at the circular velocity but locally posesses additional kinetic energy in the form of turbulent motions . testing the viability of such a hydrodynamical model is very interesting but beyond the scope of this paper . this leaves room for an alternative scenario , in which the excess velocity dispersion is predominantly gravitational ( despite its irregular behavior ) implying @xmath205 . the wfpc2 imaging and kinematics can not rule out such a spherical gas distribution at the very center ( section [ s : wfpc2_red ] ) . the wht / isis stellar kinematics can place constraints on the mass - to - light ratio @xmath130 and the bh mass independently from the gas kinematics . we in turn discuss the constraint on @xmath130 and bh mass . a fully general stellar dynamical modeling of an axisymmetric galaxy requires ` three - integral ' models in which the phase - space distribution function @xmath206 depends on the orbital energy @xmath207 , the angular momentum component parallel to the symmetry axis , @xmath208 , and a third , non - classical integral ( e.g. , binney & tremaine 1987 ) . however , the inferred mass - to - light ratio from ground - based observations is generally insensitive to assumptions about the structure and dynamics of the galaxy . models with different inclination ( van der marel 1991 ) or anisotropy ( van der marel 1994 ; 1999 ) yield the same value of @xmath111 to within @xmath209% . thus , to derive @xmath130 , we constructed axisymmetric two - integral models for ngc 4335 in which the phase - space distribution function @xmath210 depends only on the two classical integrals of motion . we use the same galaxy mass model as for the gas kinematics ( i.e. , @xmath211 with @xmath109 as derived in section [ s : mass_stars ] ) and we also include a bh mass . the modeling procedure is the same as used ( and described in more detail ) by vk00 and similar to applications of models to large samples by , e.g. , van der marel ( 1991 ) and magorrian et al . ( 1998 ) . figure [ f : starkin ] shows the observed and predicted kinematics as a function of major axis distance . a value of @xmath212 in solar units provides a good fit to the stellar mean velocity and velocity dispersion profiles at @xmath213 . this is @xmath214 higher than inferred from the gas . given the @xmath215 uncertainty for @xmath111 , the stellar kinematics out to a radius of @xmath216 are consistent with the mass - to - light ratio inferred by the gas velocity profile in the inner @xmath217 . the two - integral models can only reproduce the observed peak in the stellar velocity dispersion in the inner @xmath218 by invoking a bh of mass @xmath219 . if this bh mass is correct , the hst gas kinematics are resolving the radius of influence @xmath220 . however , these same gas kinematics firmly rule out such a massive bh . instead the gas models predict a bh mass which is at least between 5 and 30 times smaller . this is very surprising , because the two gas modeling variants discussed in section [ s : dispersion ] ascribe all nuclear gas kinetic energy to gravitation . the gas is assumed to be in either a disk or a spherical cloudlet distribution . these two models are expected to bracket the true nuclear gas distribution . however , there are reasons to mistrust the high bh mass from stellar dynamical modeling . if radial anisotropy in the stellar motions is present , the bh mass is overestimated . as an example , magorrian et al . ( 1998 ) determined the bh masses in a sample of 36 galaxies with similar modeling as performed here ( i.e. , two - integral modeling with ground - based kinematics and hst - based photometry ) . they found that for the 15 most spherical galaxies , a spherical model with radial anisotropy remove the need for a bh in all cases ( see also van der marel 1999 ) . it turns out that this degeneracy can be lifted if stellar kinematics at hst resolution is available ( i.e. , resolving the sphere of influence of the bh ) and the complete parameter - space for axisymmetric models is considered with three - integral modeling . early - type galaxies do indeed harbor bhs but the masses inferred from two - integral and three - integral axisymmetric modeling differ systematically . we compare nine e / s0 galaxies in magorrian et al ( 1998 ) which also have three - integral modeling ( as listed in kormendy & gebhardt 2001 ) . the average ratio of two - integral bh mass over three - integral bh mass is 3.7 with a minimum of 0.41 and a maximum of 10 . thus it is not too unexpected that , if we were to have stellar kinematics at hst resolution as well , three - integral modeling would indicate a bh mass in better agreement with the value @xmath221 as derived from the gravitational modeling of the gas velocity dispersion . by contrast , the factor @xmath222 difference in bh mass between the resolved thin disk modeling and the stellar dynamical modeling seems a far stretch . unfortunately , we can not expect to obtain a stronger constraint on bh mass with more general axisymmetric three integral modeling with the current data because we ( i ) lack stellar kinematics at hst resolution and hence do not resolve the @xmath223 of a @xmath224 and ( ii ) have ground - based stellar kinematics only along the major axis and with reliable measurements of only first and second order moments . before analyzing the implications of the inferred black hole mass we first discuss the two main differences between our gas disk dynamical modeling and the modeling used for ngc 3245 presented by barth et al . ( 2001 ) which constitutes the current state - of - the - art . first , instead of the observed emission - line flux distribution , we have used a double exponential fit to the flux distribution . for ngc 3245 the use of an exponential fit changes the inferred @xmath1 by @xmath225 . moreover , it accounted less well for the wiggles in the velocity profile . in our case the deviations of the exponential model from the observed photometric and spectroscopic emission - line fluxes are estimated to have a @xmath54 effect on bh mass ( see section [ s : bestfit ] ) . second , we do not incorporate the velocity shifts due to asymmetric illumination of the slit by the gas disk in the dispersion direction . in other galaxies this could be important to derive the bh mass as they produce surface brightness caustics ( maciejewski & binney 2001 ) . these caustics are not observed in ngc 4335 . stis observations of stars by one of us show that the velocity shift amounts to at most @xmath226 for a @xmath24 wide slit in the extreme case of a star at the edge of the slit . only the ( central ) apertures in the adjacent slits have an asymmetric flux gradient in the dispersion direction . the kinematics from these apertures are not taken into account in deriving the upper limit to @xmath1 . hence we conclude that the velocity shift has negligible effect on the derived @xmath1 . black hole mass @xmath1 correlates in general rather tightly with observed velocity dispersion @xmath2 in the inner region of galaxies ( gebhardt et al . 2000 ; ferrarese & merritt 2000 ) . recently , tremaine et al . ( 2002 ) completed a detailed analysis of the correlation and we will use their results in what follows . they find a best fit correlation of the form @xmath227 with @xmath228 and @xmath229 for @xmath230 . the observed spread in the correlation indicates that the intrinsic dispersion in @xmath231 is 0.3 dex ( perhaps smaller if observational errors are underestimated ) . measuring the central @xmath2 in similar fashion as in tremaine et al . ( 2002 ) gives @xmath232 for ngc 4335 ( cf . section [ s : stellarkin ] ) which corresponds to a predicted @xmath233 . the @xmath171 upper - limit @xmath234 from the thin disk modeling of the gas mean velocities ( section [ s : bestfit ] ) falls well below the relation , even when including the reported intrinsic dispersion in @xmath1 . the residuals are @xmath235 dex and @xmath236 dex for the upper - limit of @xmath237 ( best - fit model ) and @xmath238 ( for a maximally face - on gas disk ) , respectively ( see figure [ f : bhsigma ] ) . the residuals are even larger if we use the best - fit relation as determined by ferrarese ( 2002 ) who infers a larger @xmath239 . the @xmath219 derived from stellar dynamics corresponds to an equally large , but positive residual of @xmath240 dex . finally , the gravitational modeling of the gas velocity dispersions yields @xmath198 corresponding to a residual of @xmath241 dex , well within the reported intrinsic dispersion of the correlation . which bh mass are we to trust ? the analysis strongly supports a thin rotating gas disk model at @xmath196 . the model provides a reasonable fit to the gas mean velocities and dispersions , the dust disk morphology and gas disk kinematics indicate the same inclination and pa , and the @xmath111 from stellar and gas kinematics is consistent . however , there are doubts for the validity of the model at @xmath4 . the wfpc2 and stis fluxes are consistent with a thin disk surface brightness profile but the signal - to - noise of the wfpc2 image is too low to rule out a more spherical distribution . an excess velocity dispersion with an irregular profile is observed at @xmath242 . only an ad hoc explanation of localized random motion exists for this excess . it is not clear if such quasi - stationary turbulence in a thin disk is physically viable ( e.g. , wada , meurer & norman 2002 ; and references therein ) . if the excess velocity dispersion is ( partly ) due to gravitational motion around the bh , then the thin disk model underestimates the true bh mass . ascribing all gas kinetic energy , including the excess velocity dispersion , as counterbalancing the gravitational potential in a simple manner of isotropically moving collisionless cloudlets yields a bh mass which agrees well with the @xmath8 correlation ( tremaine et al . 2002 ) . ascribing the nuclear velocity dispersion to rotational motion from a spatially _ unresolved _ disk yields an equally good agreement ( section [ s : dispersion ] ) . as discussed in more detail in section [ s : starkin ] , it is not too unreasonable to assume that the stellar dynamical modeling overestimates the bh mass by a factor of @xmath201 . this would be due to radial anisotropy ( often observed in bright ellitpicals such as ngc 4335 ) which the two - integral modeling does not take into account . in conclusion , the gas spheroidal model infers a bh mass in accordance with the empirical @xmath8 relation , but remains very simplistic . if doubts about the validity of the thin disk modeling for @xmath4 were proven true , its inferred bh mass is expected to be an underestimate , driving the expected true bh mass to higher values in better agreement with the @xmath8 relation . similarly , the expected corrections for the stellar dynamical model bring its predicted bh mass in better agreement with the relation . what are the implications if in reality ngc 4335 indeed harbors a bh with @xmath243 ? our results then suggest that gas dynamical modeling assuming thin rotating disks can not be used to derive accurate black hole masses , _ even when all of the following are true _ : ( i ) the gas mean velocities very clearly suggest rotation ; ( ii ) the surrounding dust disk appears regular ; ( iii ) the inclination and position angle of the inner gas disk are well - constrained due to the use of three adjacent slits and these angles are both indicated by the gas disk kinematics and independent methods ; and ( iv ) the derived @xmath130 from gas and stars agree , ruling out asymmetric drift at @xmath196 . this then would cast doubt on other @xmath1 values determined from gas kinematics using similar models . figure [ f : bhsigma ] shows that these measurements have a large influence on the upper - end of the @xmath1 detections and hence on the slope of the correlation . moreover , to date all measurements of @xmath1 in nearby radio galaxies are based on gas kinematics . however , the fact that the inferred bh masses using this method follow the best - fit relation more closely than the stellar dynamical measurements argues against this worry of the validity of the gas dynamical modeling in general . the @xmath126 per degree of freedom as used in deriving the best - fit relation ( cf . equation 3 in tremaine et al . 2002 ) is 0.27 for the gas dynamical measurements while for the stellar dynamical measurements it is 1.30 ( 1.17 if one excludes the milky way bh mass measurement which is obtained from a stellar dynamical model which is completely different from those used for external galaxies ) . moreover , the average residual of the gas dynamical bh mass measurements is positive , counter to what is expected if the gas mean velocities systematically underestimate the circular velocity and hence the inferred bh mass . independent determinations of @xmath1 in galaxies from gaseous and stellar kinematics observed at high spatial resolution are crucial to address the worries about the gas dynamical modeling . this has been performed for ic 1459 ( verdoes kleijn et al . 2000 ; cappellari et al . verdoes kleijn et al . derive from the gaseous kinematics at 6 fos pointings a black hole mass ranging from @xmath244 ( thin disk model ) to @xmath245 ( isotropic spheroidal model ) . however , cappellari et al . ( 2002 ) infer @xmath246 based on three - integral modeling of combined ground - based and hst / stis stellar kinematics . unfortunately , the hst stellar kinematics can not be measured accurately inside the sphere of influence for this bh mass . cappellari et al . also present a more complete view of the gas kinematics from a recent stis long - slit observation . this indicates that while the data are consistent with the earlier fos measurements , and the inferred bh mass is similar ( modeled with independently developed software ) , the gas mean velocities are rather perturbed in this particular case . moreover , ic1459 has an irregular dust distribution . thus , the gas and dust properties in ic 1459 are quite different from the photometrically and kinematically well - behaved central disk in ngc 4335 . an independent @xmath1 measurement based on the stellar kinematics for ngc 4335 at hst resolution , and for other galaxies with similar gas disk kinematics , would test the validity of the gas dynamical modeling in well - behaved gas disks as opposed to irregular gas disks . it is similarly crucial to understand the origin of the excess velocity dispersion commonly seen in nuclear gas disks . presently , it is not clear under which circumstances the thin gas disks become locally turbulent ( for example , gravitational or mhd instabilities ) and if they remain globally stable ( e.g. , wada , meurer & norman 2002 ; and references therein ) . a better idea for the origin of the excess is needed to improve on the highly idealized collisionless spherical model discussed in this paper . in this respect , it is also useful to determine the ionization mechanism of the emission - line gas in nuclear disks . if it is shocks we should be wary about the modeling with unperturbed , infinitely thin rotating disks . furthermore , the disks of gas and dust might perhaps interact with ambient hot x - ray gas which is often present in the centers of bright ellipticals ( e.g. , gunn 1979 ) . the ultimate goal for gas dynamical modeling is thus to explain self - consistently the density , ionization state and dynamical state of the central gas distributions . this requires high s / n two - dimensional photometry , kinematics and line ratios to determine the dynamics , ionization state , density and temperature as a function of disk radius . this will provide not only accurate bh masses but also a vast improvement in our understanding of the fueling of bhs . kiloparsec - scale radio jets are only seen in early - type galaxies but never in spirals ( with one possible exception known to the authors ; ledlow et al . for instance , the ugc fr - i galaxies all have hubble types e - s0 ( cf . prior to this study , only bh masses @xmath247 have been reported in nearby radio galaxies ( i.e. , ngc 4261 , ngc 4374 , m87 , ngc 5128 , ngc 6251 and ngc 7052 , see tremaine et al . 2002 for references ) . interestingly , none of the seven galaxies with a hubble type later than s0 in the the sample compiled by tremaine et al . ( 2002 ) have @xmath248 . we are aware of only one galaxy with a hubble type later than s0 and a @xmath249 , which is the sombrero galaxy ( m104 , hubble type sa ) with @xmath250 , but this measurement is based on models less general than three - integral stellar dynamical models . by contrast , all bh mass upper limits and detections in a sample of 16 mostly early - type disk galaxies ( hubble type s0-sb ; sarzi et al . 2001 ) are below @xmath251 . these results suggest that the differences in bh mass might be the underlying factor for this host preference of radio - jets . however , if our determination of a @xmath171 upper limit of @xmath181 on the bh mass of ngc 4335 is correct , then ngc 4335 would illustrate that bhs with @xmath181 are also capable of producing fr - i radio jets . this would argue against bh mass being the ( only ) parameter underlying the host morphology preference of radio galaxies . it is a pleasure to thank marcella carollo and chris odea for stimulating advice , and to thank chris benn and andy longmore for carrying out our wht service observations and providing helpful suggestions during the analysis . we would like the referee for his / her comments which helped to improve the paper significantly . support for proposal # 8236 was provided by nasa through a grant from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc . , under nasa contract nas 5 - 26555 . this paper made use of the leda database : http://leda.univ-lyon1.fr . g.v.k . is grateful to the leids kerkhoven - bosscha foundation for financial support for a visit to stsci in february 2001 . e & 2 & 0.347 & -0.254 & 101.00 & 9.68 & 115.71 & 9.44 + & 1 & 0.196 & -0.232 & 57.01 & 9.58 & 84.86 & 9.34 + & 1 & 0.096 & -0.217 & 38.39 & 5.65 & 76.64 & 5.62 + & 1 & -0.005 & -0.202 & -2.00 & 5.48 & 80.75 & 5.47 + & 1 & -0.105 & -0.187 & -65.27 & 5.88 & 96.18 & 6.03 + & 1 & -0.205 & -0.171 & -105.18 & 6.50 & 112.63 & 6.26 + & 2 & -0.356 & -0.149 & -126.46 & 8.49 & 141.02 & 7.98 + & 2 & -0.556 & -0.119 & -292.72 & 4.77 & 53.30 & 4.62 + c & 3 & 1.128 & -0.169 & 320.72 & 6.12 & 77.20 & 6.19 + & 3 & 0.828 & -0.124 & 292.35 & 4.25 & 57.35 & 4.17 + & 2 & 0.577 & -0.086 & 284.91 & 5.49 & 55.16 & 5.44 + & 2 & 0.376 & -0.056 & 197.42 & 12.06 & 165.72 & 11.49 + & 1 & 0.226 & -0.034 & 142.31 & 8.88 & 145.00 & 8.72 + & 1 & 0.125 & -0.019 & 64.90 & 4.78 & 197.58 & 4.39 + & 1 & 0.025 & -0.004 & 15.66 & 3.55 & 240.79 & 2.98 + & 1 & -0.075 & 0.011 & -49.96 & 4.06 & 231.24 & 3.32 + & 1 & -0.175 & 0.026 & -89.63 & 5.59 & 142.48 & 5.62 + & 2 & -0.326 & 0.049 & -155.09 & 8.40 & 146.89 & 8.24 + w & 3 & 1.459 & -0.016 & 364.75 & 3.23 & 49.70 & 3.14 + & 3 & 1.158 & 0.029 & 354.71 & 6.85 & 82.90 & 6.86 + & 3 & 0.857 & 0.074 & 310.68 & 5.85 & 92.86 & 5.61 + & 2 & 0.606 & 0.111 & 315.11 & 5.76 & 87.42 & 5.92 + & 2 & 0.406 & 0.141 & 312.75 & 5.11 & 92.93 & 5.06 + & 1 & 0.255 & 0.164 & 209.58 & 11.82 & 147.67 & 11.28 + & 1 & 0.155 & 0.179 & 105.15 & 8.92 & 179.87 & 8.15 + & 1 & 0.055 & 0.194 & 28.02 & 5.92 & 163.13 & 5.79 + & 1 & -0.046 & 0.209 & -12.79 & 6.30 & 159.55 & 6.17 + & 1 & -0.146 & 0.224 & -68.63 & 8.27 & 128.64 & 8.32 + & 2 & -0.296 & 0.247 & -120.31 & 5.39 & 87.64 & 5.36 + & 2 & -0.497 & 0.277 & -155.89 & 9.89 & 113.07 & 9.61 + 36 & 7.39 & 1.04 & 51 & 10 & 210 & 11 + 12 & 3.18 & 0.45 & 95 & 8 & 245 & 9 + 5 & 1.65 & 0.23 & 91 & 8 & 297 & 9 + 3 & 0.92 & 0.13 & 49 & 7 & 296 & 7 + 2 & 0.47 & 0.07 & 34 & 7 & 295 & 7 + 2 & 0.09 & 0.01 & 17 & 8 & 314 & 6 + 2 & -0.28 & -0.04 & -22 & 7 & 291 & 6 + 3 & -0.73 & -0.10 & -46 & 8 & 287 & 7 + 6 & -1.54 & -0.22 & -101 & 7 & 258 & 7 + 16 & -3.52 & -0.50 & -119 & 8 & 232 & 8 +

a weak order on the set of maximal chains of the non - crossing partition lattice is introduced and studied . a-hecke algebra action is used to compute the radius of the graph on these chains in which two chains are adjacent if they differ in exactly one element .

1201.4669

consider the graph @xmath1 with vertex set consisting of all maximal chains in @xmath2 , the non - crossing partition lattice of type @xmath3 , where two chains are adjacent if they differ in exactly one element . this graph may be identified with the graph of all reduced words of a given coxeter element ( long cycle ) in the symmetric group @xmath4 , where the alphabet consists of all reflections ( transpositions ) and two words are adjacent if they agree in all but two adjacent letters , which are @xmath5 in one word and either @xmath6 or @xmath7 in the other ; here @xmath8 . this graph is known to be connected see , e.g. , ( * ? ? ? 1.6.1 ) ; we are interested in calculating its radius . this is motivated by @xcite and @xcite , where the analogous question for simple reflections was studied , and by @xcite which evaluated the radius of a related graph on labeled trees . recall that a classical result of hurwitz @xcite implies ( see @xcite ) that the number of maximal chains in the non - crossing partition lattice of type @xmath3 is equal to the number of labeled trees on @xmath9 vertices . our approach is to consider a @xmath0-hecke algebra action on the set of maximal chains . this allows us to define a well - behaved natural weak order on this set . each maximal interval is isomorphic to the weak order on the symmetric group . the number of maximal elements , refined by a generalized inversion number , is the carlitz - riordan @xmath10-catalan number . the resulting undirected hasse diagram spans the graph @xmath1 of maximal chains , implying an evaluation of the radius and an approximation of the diameter up to a factor of @xmath11 . the non - crossing partition lattice @xmath2 , first introduced by kreweras @xcite , may be defined as follows . let @xmath12 be the set of all reflections ( transpositions ) in the symmetric group @xmath4 , and let @xmath13 be the corresponding length function : @xmath14 is the minimal number of factors in an expression of @xmath15 as a product of reflections . let @xmath16 be a coxeter element in this group ( i.e. , a cycle of length @xmath9 ) ; for concreteness , take @xmath17 . then @xmath2 is the set @xmath18 ordered by @xmath19 let @xmath20 be the set of all maximal chains in the non - crossing partition lattice @xmath2 . clearly , an element of @xmath20 corresponds to a factorization of @xmath16 into a minimal number ( @xmath21 ) of transpositions . it can thus be written in the form @xmath22 , where @xmath23 @xmath24 and @xmath25 . [ d.hurwitz_graph ] the _ hurwitz graph _ @xmath1 is the ( undirected ) graph with vertex set @xmath20 , where two chains are adjacent if they differ in exactly one element . the hurwitz graph @xmath26 is drawn in figure [ figure1-tik ] ; note that @xmath27 , and the midpoint of the figure is not a vertex . ( -4,1.66 ) circle ( 0.1 ) node[left]12,23,34 ; ( -1.66,4 ) circle ( 0.1 ) node[above]12,24,23 ; ( 1.66,4 ) circle ( 0.1 ) node[above]14,12,23 ; ( 4,1.66 ) circle ( 0.1 ) node[right]14,13,12 ; ( 4,-1.66 ) circle ( 0.1 ) node[right]34,14,12 ; ( 1.66,-4 ) circle ( 0.1 ) node[below]34,24,14 ; ( -1.66,-4 ) circle ( 0.1 ) node[below]23,34,14 ; ( -4,-1.66 ) circle ( 0.1 ) node[left]23,13,34 ; ( -2,0 ) circle ( 0.1 ) node[left]13,12,34 ; ( -1.4,1.4 ) circle ( 0.1 ) node[below]12,34,24 ; ( 0,2 ) circle ( 0.1 ) node[below]24,14,23 ; ( 1.4,1.4 ) circle ( 0.1 ) node[below]14,23,13 ; ( 2,0 ) circle ( 0.1 ) node[right]13,34,12 ; ( 1.4,-1.4 ) circle ( 0.1 ) node[above]34,12,24 ; ( 0,-2 ) circle ( 0.1 ) node[above]24,23,14 ; ( -1.4,-1.4 ) circle ( 0.1 ) node[above]23,14,13 ; ( -4,1.66)(-1.66,4)(1.66,4)(4,1.66)(4,-1.66)(1.66,-4)(-1.66,-4)(-4,-1.66)(-4 , 1.66 ) ; ( -2,0)(2,0 ) ; ( 0,-2)(0,2 ) ; ( -1.4,-1.4)(1.4,1.4 ) ; ( -1.4,1.4)(1.4,-1.4 ) ; ( -4,1.66)(-1.4,1.4 ) ; ( -1.66,4)(-1.4,1.4 ) ; ( -1.66,4)(0,2 ) ; ( 1.66,4)(0,2 ) ; ( 4,1.66)(1.4,1.4 ) ; ( 1.66,4)(1.4,1.4 ) ; ( 4,1.66)(2,0 ) ; ( 4,-1.66)(2,0 ) ; ( -4,1.66)(-2,0 ) ; ( -4,-1.66)(-2,0 ) ; ( -1.66,-4)(0,-2 ) ; ( 1.66,-4)(0,-2 ) ; ( 4,-1.66)(1.4,-1.4 ) ; ( 1.66,-4)(1.4,-1.4 ) ; ( -4,-1.66)(-1.4,-1.4 ) ; ( -1.66,-4)(-1.4,-1.4 ) ; for each @xmath28 and @xmath29 let @xmath30 and @xmath31 where @xmath32 . @xmath33 and @xmath34 @xmath35 are operators on @xmath20 , where @xmath36 slides the @xmath37-th factor to the right while @xmath38 slides the @xmath39-st factor to the left . clearly @xmath40 and , for every @xmath41 , either @xmath42 ( if @xmath43 and @xmath44 commute ) or @xmath45 ( otherwise ) . furthermore , the following is true . [ t.arnold]@xcite for every @xmath28 , @xmath46 this clearly defines a braid group action on @xmath20 , a distinguished example of the hurwitz action on ordered tuples of group elements . for @xmath47 , @xmath48 is an edge of @xmath1 if and only if there exists an @xmath28 such that either @xmath49 or @xmath50 . the classical tits graph of a coxeter group is the graph whose vertices are all the reduced words for the longest element of the group , in terms of the set of _ simple _ reflections @xmath51 , and where two words are adjacent if one is obtained from the other by a braid relation . in other words , vertices correspond to maximal chains in the weak order on the group , and two chains are adjacent if they differ in a `` minimal cycle '' . replacing the set @xmath51 of simple reflections by the set @xmath12 of all reflections , one obtains the hurwitz graph . the diameters of the tits graphs of the coxeter groups of types @xmath52 and @xmath53 were evaluated in @xcite . bessis proved that the hurwitz action on @xmath20 is transitive ( * ? ? ? * prop . 1.6.1 ) , thus that the hurwitz graph @xmath1 is connected . its radius and bounds on its diameter will be given in section [ section : diameter ] . [ d.order ] for @xmath47 , denote @xmath54 if @xmath55 belongs to a geodesic in @xmath1 from @xmath56 to @xmath57 . this will be called _ the weak order on @xmath20 _ , and the resulting poset will be denoted by @xmath58 . the hasse diagram of @xmath59 is drawn in figure [ figure2-tik ] . this poset is neither a meet semilattice nor a join semilattice , as shown by the two pairs of elements @xmath60 and @xmath61 . ( 0,0 ) circle ( 0.1 ) node[below]e=12,23,34 ; ( -3,2.5 ) circle ( 0.1 ) node[left]23,13,34 ; ( -1,2.5 ) circle ( 0.1 ) node[left]13,12,34 ; ( 1,2.5 ) circle ( 0.1 ) node[right]12,34,24 ; ( 3,2.5 ) circle ( 0.1 ) node[right]12,24,23 ; ( -5,5 ) circle ( 0.1 ) node[left]23,14,13 ; ( -3,5 ) circle ( 0.1 ) node[left]23,34,14 ; ( -1,5 ) circle ( 0.1 ) node[left]13,34,12 ; ( 1,5 ) circle ( 0.1 ) node[right]34,12,24 ; ( 3,5 ) circle ( 0.1 ) node[right]14,12,23 ; ( 5,5 ) circle ( 0.1 ) node[right]24,14,23 ; ( -4,7.5 ) circle ( 0.1 ) node[above]14,23,13 ; ( -2,7.5 ) circle ( 0.1 ) node[above]14,13,12 ; ( 0,7.5 ) circle ( 0.1 ) node[above]34,14,12 ; ( 2,7.5 ) circle ( 0.1 ) node[above]34,24,14 ; ( 4,7.5 ) circle ( 0.1 ) node[above]24,23,14 ; ( 0,0)(-3,2.5)(-5,5)(-4,7.5)(3,5 ) ; ( 0,0)(3,2.5)(5,5)(4,7.5)(-3,5 ) ; ( 0,0)(-1,2.5)(-1,5)(0,7.5)(1,5)(1,2.5)(0,0 ) ; ( -3,2.5)(-3,5 ) ; ( 3,2.5)(3,5 ) ; ( -1,5)(-2,7.5)(3,5 ) ; ( 1,5)(2,7.5)(-3,5 ) ; [ weak - remark ] to motivate the name `` weak order '' consider a coxeter group @xmath62 , with standard generating set @xmath51 , acting transitively on a set @xmath63 . choose an element @xmath64 in @xmath63 . for every @xmath65 , let @xmath66 be the distance from @xmath64 to @xmath67 in the associated schreier graph @xmath68 ; and for every @xmath69 , define @xmath70 to _ cover _ @xmath67 if there exists some @xmath71 such that @xmath72 and @xmath73 . this defines a natural `` weak order '' on @xmath63 , with @xmath64 as a minimal element ; analogous concepts of strong and absolute orders on @xmath62-sets were studied in @xcite . in our case , by proposition [ t.arnold ] @xmath20 carries a transitive action of the braid group of type @xmath52 which covers the coxeter group @xmath74 , and can thus be equipped with a similar order . furthermore , it will be shown later that maximal intervals in this order are isomorphic to the weak order on @xmath74 . the poset @xmath58 has several nice properties , such as a @xmath0-hecke algebra action , the above mentioned structure of maximal intervals , and a catalan number of maximal elements . it will be used to evaluate the radius and bound the diameter of the graph @xmath1 . let @xmath20 be as in section [ section : intro ] . define a map @xmath75 as follows . [ d.sigma ] for @xmath76 define the partial products @xmath77 and the empty product @xmath78 by definition , @xmath79 for every @xmath80 . thus @xmath81 for exactly two values of @xmath82 , and @xmath83 for exactly one such value . ignoring @xmath84 , the set @xmath85 thus satisfies @xmath86 on the other hand , @xmath87 while @xmath88 , so that @xmath89 for every @xmath90 . it follows that @xmath91 combining ( [ e.1 ] ) with ( [ e.2 ] ) one concludes that @xmath92 and @xmath93 it follows that the map @xmath94 defined by @xmath95 is a permutation in @xmath74 . define @xmath96 by @xmath97 [ obs.phi ] for every @xmath98 and @xmath80 , if @xmath99 with @xmath100 then @xmath101 let @xmath102 . then @xmath103 , @xmath104 , @xmath105 , @xmath106 and @xmath107 , where permutations are written in cycle notation . thus @xmath108 , @xmath109 , @xmath110 and @xmath111 . for a sequence @xmath112 of @xmath113 transpositions in @xmath4 , let @xmath114 be the geometric graph whose vertices are @xmath115 drawn on a circle and whose edges are @xmath116 . the following characterization of the words in @xmath20 , due to goulden and yong , is very useful . [ gy22 ] ( * ? ? ? * theorem 2.2 ) + for @xmath117 , the word @xmath118 belongs to @xmath20 if and only if the following three conditions hold : * @xmath114 is a tree . * @xmath114 is non - crossing ; namely , two edges may intersect only in a common vertex . * cyclically decreasing neighbors : for every @xmath119 and @xmath120 , if @xmath121 and @xmath122 then @xmath123 . here @xmath124 is the linear order @xmath125 . proposition [ gy22](iii ) may be stated in the following equivalent form . [ t.avoid ] for @xmath126 and @xmath127 , if @xmath43 and @xmath128 do not commute then the pair @xmath129 is either @xmath130 , @xmath131 or @xmath132 , for some @xmath133 . [ phi.local-range ] for @xmath126 and @xmath134 , if @xmath122 with @xmath135 then @xmath136 satisfies : @xmath137 . the discussion leading to the definition of @xmath138 and @xmath139 above shows that , for any @xmath140 , the sequence @xmath141 is weakly decreasing except for a single step @xmath142 for which , if @xmath143 , it follows that @xmath122 and @xmath144 . in particular , it follows that @xmath145 and @xmath146 . [ phi.local-range-2 ] for @xmath126 and @xmath147 , if @xmath148 and @xmath149 with @xmath150 then @xmath136 satisfies : * either @xmath137 or @xmath151 . * deleting the edge @xmath152 from the tree @xmath114 leaves two connected components , @xmath153 ( containing @xmath154 ) and @xmath155 ( containing @xmath156 ) ; only one of them contains the edge @xmath157 . if this is @xmath155 then @xmath137 , otherwise @xmath151 . by lemma [ phi.local-range ] , @xmath158 . by the proof of that lemma , the directed path in @xmath114 leading from @xmath37 to @xmath159 consists of a single increasing step ( edge ) @xmath160 , together with some decreasing edges . assuming @xmath161 , this path can not contain the step @xmath162 , and also not @xmath163 ( since @xmath147 implies @xmath164 ) . however , since @xmath114 is non - crossing and @xmath165 , this path must pass through @xmath166 and @xmath16 . we thus get two distinct paths in @xmath114 from @xmath166 to @xmath16 , contradicting proposition [ gy22](i ) . thus either @xmath137 or @xmath151 . the rest of the claim now follows from the observation that @xmath37 and @xmath159 belong to distinct connected components of @xmath114 minus the edge @xmath167 . if @xmath155 contains the edge @xmath157 then , by the non - crossing property , it contains all the vertices from @xmath166 up to @xmath156 , and therefore @xmath137 . the complementary case leads , similarly , to the complementary conclusion . [ t.inversions ] ( characterization of inversions in @xmath168 ) + for @xmath126 and @xmath169 , @xmath170 if and only if one of the following holds : * @xmath171 such that either : * * @xmath172 and @xmath173 ; or : * * @xmath174 and @xmath175 . * @xmath176 such that @xmath177 and @xmath173 . * @xmath176 such that either : * * @xmath148 , @xmath149 and , in the notation of lemma [ phi.local-range-2 ] , the edge @xmath157 belongs to @xmath153 ; or * * @xmath174 , @xmath178 and , in the notation of lemma [ phi.local-range-2 ] , the edge @xmath157 belongs to @xmath155 . assume that @xmath169 and @xmath170 . if @xmath128 and @xmath179 do not commute then , by corollary [ t.avoid ] , there are 3 options for the pair @xmath180 . two of them appear in @xmath181 above while the third , @xmath182 , is ruled out since then , by lemma [ phi.local-range ] , @xmath183 . if @xmath128 and @xmath179 commute then there are 4 options for the pair @xmath180 : @xmath184 , @xmath185 , @xmath186 and @xmath187 ( for @xmath188 ) . the first option is ruled out by lemma [ phi.local-range ] , the second appears in @xmath189 above , and the third and fourth appear in @xmath190 above in the restricted versions allowed by lemmas [ phi.local-range ] and [ phi.local-range-2 ] . conversely , if @xmath169 and one of the options @xmath181 , @xmath189 and @xmath190 holds , then lemmas [ phi.local-range ] and [ phi.local-range-2 ] show that indeed @xmath170 . recall the operators @xmath191 and @xmath192 from section [ section : basic_concepts ] . by definition [ d.sigma ] , for all @xmath193 and @xmath194 , @xmath195 and @xmath196 [ t.main-lemma ] for every @xmath98 and @xmath197 , @xmath198 and @xmath199 we prove the formula for @xmath191 ; the proof of the other formula is similar . since @xmath200 for @xmath201 and @xmath202 for @xmath203 it follows , by observation [ obs.phi ] , that @xmath204 for all @xmath201 . thus either @xmath205 or @xmath206 . by observation [ obs.phi ] ( for @xmath207 ) , @xmath205 if and only if @xmath208 and @xmath209 for some @xmath154 , @xmath166 and @xmath16 such that @xmath100 and @xmath210 . these @xmath128 and @xmath211 do not commute , and by corollary [ t.avoid ] necessarily @xmath212 as well . recall the graded poset @xmath58 from section [ section : basic_concepts ] . note that , by definition , [ t.minimum ] the word @xmath213 is the unique minimal element in @xmath58 . [ t.unique_id ] @xmath214 clearly @xmath215 , since for @xmath216 one has @xmath217 , @xmath218 and thus @xmath219 by observation [ obs.phi ] . conversely , assume that @xmath220 has @xmath221 . then @xmath222 implies that @xmath223 , so that @xmath224 and @xmath225 . it follows that @xmath226 , so that necessarily @xmath227 and @xmath228 belongs to @xmath229 , with @xmath230 . induction on @xmath9 completes the proof . for any @xmath231 , let @xmath232 be its rank in the poset @xmath58 . for @xmath233 , let @xmath234 be its inversion number . [ t.rank_eq_inv ] for any @xmath98 , @xmath235 by induction on @xmath236 . if @xmath237 then @xmath221 . by lemma [ t.unique_id ] @xmath238 , so that indeed @xmath239 . for the induction step let @xmath240 , and assume that the claim holds for @xmath241 whenever @xmath242 . let @xmath243 . since @xmath244 there exists @xmath245 such that @xmath246 . by lemma [ t.main-lemma ] , @xmath247 unless @xmath248 such that @xmath209 and @xmath208 ( in @xmath249 ) ; whereas @xmath250 unless @xmath248 such that @xmath251 and @xmath252 . these two exceptions can not hold simultaneously , and thus @xmath253 for either @xmath254 or @xmath255 . for this @xmath55 , @xmath256 , so by the induction hypothesis @xmath257 . since @xmath55 and @xmath249 are connected by an arc in the graph @xmath1 , it follows that @xmath258 . on the other hand , in a geodesic from @xmath64 to @xmath249 in @xmath1 the value of @xmath259 changes by @xmath0 or @xmath260 on each arc ( again by lemma [ t.main-lemma ] ) , so that @xmath261 . thus @xmath262 . [ t.hasse ] the ( undirected ) hasse diagram of @xmath58 is obtained from the graph @xmath1 by deleting all edges connecting a vertex @xmath263 with a vertex @xmath264 for some @xmath265 . by lemma [ t.main-lemma ] and lemma [ t.rank_eq_inv ] . [ t.forbidden-wedge ] for every @xmath231 and @xmath193 , if @xmath266 and @xmath267 in @xmath58 then @xmath268 . we can assume that @xmath128 and @xmath211 do not commute , since otherwise @xmath268 . then , by corollary [ t.avoid ] , @xmath269 for some @xmath133 . the same holds for the @xmath142-th and @xmath270-st factors of @xmath271 , as well as for those of @xmath272 . by lemma [ t.rank_eq_inv ] , @xmath266 implies that @xmath273 , which is equivalent , by lemma [ t.main-lemma ] , to @xmath274 . by lemma [ t.inversions ] this rules out @xmath275 and @xmath276 , and leaves only @xmath277 as an option for the @xmath142-th and @xmath270-st factors of @xmath271 . a similar conclusion holds for @xmath272 , and this implies that @xmath278 and simultaneously @xmath279 , a contradiction . [ t.rl-cover ] lemma [ t.forbidden-wedge ] may be stated in the following two equivalent forms . * for every @xmath231 and @xmath193 , @xmath280 if and only if @xmath281 . * each orbit in @xmath20 of each @xmath191 ( or @xmath192 ) consists of either a pair @xmath282 with @xmath283 or a triple @xmath284 with @xmath283 and @xmath285 . the _ @xmath0-hecke algebra _ @xmath286 is an algebra over @xmath287 generated by @xmath288 with defining relations @xmath289 @xmath290 and @xmath291 @xmath286 can be viewed as the specialization @xmath292 of the iwahori - hecke algebra @xmath293 , deforming the group algebra of @xmath4 , or as the semigroup algebra of a suitable monoid . note that our notation is slightly non - standard , since equation ( [ e.0_hecke_relation ] ) is usually stated as @xmath294 , which amounts to replacing @xmath295 by @xmath296 ( for all @xmath37 ) . a faithful action of the @xmath0-hecke algebra @xmath297 on @xmath20 is introduced in this section . this action will be applied to show that every maximal interval in @xmath58 is isomorphic to the weak order on the symmetric group @xmath74 . recall the descent set of a permutation @xmath298 , @xmath299 [ def - down ] for @xmath300 define the _ @xmath37-th down operator _ @xmath301 by @xmath302 the down operators are well defined . by lemma [ t.main-lemma ] , for any @xmath231 and @xmath28 , either @xmath303 or @xmath304 ( or both ) . if @xmath305 then @xmath306 . thus , by lemma [ t.rank_eq_inv ] , either @xmath307 or @xmath308 , namely : either @xmath309 or @xmath310 in @xmath58 . if both hold then , by lemma [ t.forbidden-wedge ] , @xmath311 . [ obs.down ] for every @xmath312 and @xmath98 , @xmath313 [ t.down-relations ] the down operators @xmath314 satisfy the defining relations of the @xmath0-hecke algebra @xmath297 . if @xmath315 then , by definition [ def - down ] , @xmath316 . if @xmath305 then , by observation [ obs.down ] , @xmath317 and therefore , by definition [ def - down ] , @xmath318 . thus @xmath319 for every @xmath28 . if @xmath320 then , by definition [ def - down ] , @xmath321 . we have to verify that @xmath322 for every @xmath323 . if @xmath324 then , by definition [ def - down ] , @xmath325 . if @xmath305 but @xmath326 then , by definition [ def - down ] , @xmath327 and therefore @xmath328 . by observation [ obs.down ] , the permutation @xmath329 is the same as @xmath330 except that @xmath331 , which is not necessarily the case for @xmath330 . the assumption @xmath305 but @xmath326 implies that @xmath332 . denoting @xmath333 we get @xmath334 and therefore @xmath335 . hence @xmath336 and , by definition [ def - down ] , @xmath337 as well . for similar reasons , if @xmath315 and @xmath338 then @xmath339 and @xmath340 , so that @xmath341 . it remains to verify the braid relation @xmath322 when @xmath342 . by assumption @xmath343 , so that @xmath249 , @xmath344 , @xmath345 and @xmath346 are all distinct : @xmath347 in @xmath58 . similarly with @xmath348 and @xmath349 interchanged . for a word @xmath350 let @xmath351 . obviously , if @xmath352 then @xmath353 . it thus suffices to verify that @xmath354 for @xmath355 . for a subset @xmath356 let @xmath357 . by proposition [ gy22](i ) , @xmath358 . we shall consider these @xmath359 cases separately . * @xmath360 . + in this case @xmath361 and @xmath362 commute , so that @xmath363 etc . thus , by proposition [ t.arnold ] , @xmath364 * case 2 . * @xmath365 . + in this case one of the transpositions @xmath361 or @xmath362 commutes with other two . if @xmath43 commutes with @xmath44 and @xmath362 then , since @xmath338 , if @xmath366 with @xmath265 then , by lemma [ t.inversions](1 ) , either @xmath367 or @xmath368 and in both cases , after application of @xmath349 , this becomes @xmath277 . thus @xmath369 and also @xmath370 similarly , if @xmath362 commutes with @xmath43 and @xmath44 then , if @xmath371 with @xmath265 , @xmath372 if @xmath44 commutes with @xmath43 and @xmath362 then , if @xmath373 with @xmath265 , @xmath374 * case 3 . * @xmath375 . + let @xmath376 with @xmath377 . by proposition [ gy22 ] and lemma [ t.inversions](1 ) there are exactly @xmath378 possible options for @xmath379 , which correspond to the @xmath378 maximal elements of @xmath380 , as in figure [ figure2-tik ] . inspecting this figure , it is easy to verify that the braid relation hold in all cases . for a permutation @xmath233 let @xmath381 be a reduced word for @xmath15 in the alphabet of simple reflections . define the operator @xmath382 by @xmath383 the operator @xmath384 is well - defined . since any two reduced words for the same permutation are connected by a sequence of braid relations , the result follows from lemma [ t.down-relations ] . [ interval1 ] for every element @xmath231 , the lower interval @xmath385 $ ] is exactly @xmath386 . by definition , @xmath387 for any @xmath28 and therefore @xmath388 for any @xmath233 . in the other direction , if @xmath389 then @xmath57 belongs to a geodesic from @xmath249 to @xmath64 in @xmath1 , part of which is a saturated chain @xmath390 in @xmath58 . clearly each step is @xmath391 for some @xmath392 ( @xmath393 ) . if the word @xmath394 is not reduced then it can be brought , by relations as in lemma [ t.down-relations ] , to a reduced word for some @xmath15 , so that @xmath395 . [ interval2 ] let @xmath396 be any maximal element in @xmath58 . then , for any @xmath298 , @xmath397 where @xmath398 $ ] is the maximal element in the weak order on @xmath74 . by induction on @xmath234 . if @xmath399 then necessarily @xmath400 , since otherwise there exists @xmath193 such that @xmath401 ; and then , by lemma [ t.main-lemma ] and lemma [ t.rank_eq_inv ] , either @xmath402 or @xmath403 has @xmath404 , contradicting the maximality of @xmath396 . if @xmath405 then there exists @xmath193 such that @xmath406 has @xmath407 . then @xmath408 has @xmath409 with @xmath410 . by the induction hypothesis @xmath411 , thus @xmath412 . combining this with observation [ obs.down ] we obtain @xmath413 we conclude [ t.main2 ] each maximal interval in @xmath58 is isomorphic , as a poset , to the weak order on the symmetric group @xmath74 . by lemmas [ interval1 ] and [ interval2 ] , @xmath139 is onto @xmath74 even when restricted to a maximal interval @xmath414 $ ] . to prove that @xmath139 is also one - to - one when restricted to @xmath414 $ ] , it suffices to show that the cardinality of the interval @xmath414 $ ] does not exceed the cardinality of @xmath74 ; and this follows again from lemma [ interval1 ] . finally , to prove that @xmath139 is a poset isomorphism notice that , for @xmath415 $ ] , if @xmath416 then either @xmath417 or @xmath418 , and therefore either @xmath419 or @xmath420 . the converse implications are clear . thus , for @xmath415 $ ] , @xmath421 we can now summarize . * the @xmath0-hecke algebra @xmath297 acts faithfully on @xmath58 . * each lower interval is invariant under this action . by lemma [ t.down-relations ] , the down operators @xmath422 satisfy the defining relations of @xmath297 . furthermore , by lemma [ interval1 ] , lower intervals are invariant under these operators . by theorem [ t.main2 ] , the size of each maximal interval is equal to the dimension of the @xmath0-hecke algebra , so that the action is faithful . in this section we prove [ t.main3 ] the number of maximal elements in @xmath58 is the catalan number @xmath423 for a @xmath10-analogue of this result see corollary [ t.main3-q ] below . recall the map @xmath75 defined in section [ section : map ] . for @xmath298 , let @xmath424 for a subset @xmath51 of @xmath425 let @xmath426 be the subsequence of @xmath427 consisting only of the values in @xmath51 , and let @xmath428 be the pattern ( permutation ) obtained by mapping @xmath51 monotonically to @xmath429 , namely : the sequence obtained from @xmath426 by writing @xmath260 instead of the smallest number in @xmath51 , @xmath430 instead of the second smallest number , etc . for example , if @xmath431 and @xmath432 then @xmath433 and @xmath434 . also , for @xmath435 , let @xmath436 ( of size @xmath437 ) and @xmath438 ( of size @xmath439 ) . thus @xmath440 , a disjoint union . [ t.cover-number ] let @xmath298 such that @xmath441 @xmath442 . then @xmath443 for @xmath444 , @xmath445 . thus , summing over @xmath446 : @xmath447 assume that @xmath98 satisfies @xmath448 . then the first transposition in @xmath249 has the form @xmath449 for some @xmath450 , and the corresponding partial product @xmath451 is @xmath452 . in general , for any @xmath453 , @xmath454 is a product of @xmath455 disjoint cycles , each of which forming a ( not necessarily consecutive ) cyclic subsequence of @xmath456 , and the transition from @xmath457 to @xmath454 amounts to merging two of the cycles of @xmath457 into one cycle in which the original cycles form complementary cyclic intervals . it follows that @xmath458 is the product of those @xmath179 with @xmath459 which permute two elements of the set @xmath460 , and thus have @xmath461 in this set but of course @xmath462 ; and , similarly , @xmath463 is the product of those @xmath179 with @xmath459 which permute two elements of the set @xmath464 , and thus have @xmath461 in this set but of course @xmath465 . the word @xmath249 is a shuffle of these two sequences of transpositions , with the addition of @xmath449 . the @xmath139 values at these two sequences compose the ranges @xmath466 and @xmath467 , respectively . mapping these ranges monotonically to the ranges @xmath468 and @xmath469 we get @xmath470 and @xmath471 such that , for @xmath472 , @xmath473 and @xmath474 . conversely , for any @xmath475 , @xmath470 and @xmath471 we can construct @xmath231 by reversing the above process . this completes the proof of the formula . _ proof of theorem [ t.main3 ] . _ let @xmath476 be the longest element in @xmath74 , i.e. , @xmath477 @xmath24 ; denote @xmath478 . if @xmath98 has @xmath479 then @xmath480 and therefore , for each @xmath481 , @xmath482 and @xmath483 . thus @xmath484 and @xmath485 are decreasing sequences , and the corresponding permutations @xmath486 and @xmath487 are the longest elements in @xmath488 and @xmath489 , respectively . it follows , by lemma [ t.cover-number ] , that @xmath490 and clearly @xmath491 . we recognize the recursion formula for the catalan numbers @xmath492 , and therefore @xmath493 as claimed . the current section presents a bijection between the maximal elements in @xmath58 and alternating non - crossing trees , following a comment by an anonymous referee . recall that a tree with integer - labeled vertices is _ alternating _ if the vertices along any path in it form an alternating sequence : @xmath494 ; equivalently , if the neighbors of each vertex @xmath37 are all larger , or all smaller , than @xmath37 . [ t.max_to_alt ] if @xmath57 is a maximal element in @xmath58 then the corresponding non - crossing tree @xmath495 ( as in proposition [ gy22 ] ) is alternating . if @xmath57 is maximal in @xmath58 then , by lemma [ interval2 ] for @xmath496 , @xmath497 $ ] and thus all pairs of letters in @xmath498 are inversions . for @xmath133 , the pair @xmath277 does _ not _ correspond to an inversion in @xmath498 , by the characterization in lemma [ t.inversions ] . this implies that @xmath495 is alternating . the number of all alternating non - crossing trees on the set of vertices @xmath499 = \{1 , \ldots , n\}$ ] was shown ( * ? ? ? * theorem 6.4 ) to be the catalan number @xmath492 . this motivates the following result . [ t.max_eq_alt ] the mapping @xmath500 is a bijection between the set of all maximal elements in @xmath58 and the set of all alternating non - crossing trees on the set of vertices @xmath499 $ ] . by observation [ t.max_to_alt ] , this mapping , to be denoted @xmath501 , is well defined . let us construct a map @xmath502 in the other direction . for an alternating non - crossing tree @xmath503 on @xmath499 $ ] , defining @xmath504 amounts to imposing a linear order on the @xmath505 edges of @xmath503 . this can be done recursively as follows : if @xmath506 , there are no edges and the order is empty . for @xmath507 note that @xmath508 must be an edge of @xmath503 ( since the path from @xmath260 to @xmath9 in @xmath503 is monotone increasing , because of the non - crossing property , and thus is also alternating if and only if it consists of a single edge ) . the graph obtained from @xmath503 by deleting the edge @xmath508 has two connected components , @xmath509 ( containing the vertex @xmath260 ) and @xmath510 ( containing the vertex @xmath9 ) . they are alternating non - crossing trees on complementary subsets @xmath511 $ ] and @xmath512 $ ] of @xmath499 $ ] , for a suitable @xmath453 . they define , by recursion , corresponding words @xmath513 and @xmath514 ( using the same letter @xmath502 for the two corresponding maps ) . define the word @xmath504 as the word @xmath515 , followed by the letter @xmath516 , followed by the word @xmath517 . we need to show that @xmath57 is a maximal element of @xmath58 . first , @xmath518 since @xmath519 satisfies the goulden - yong conditions of proposition [ gy22 ] , the cyclic order on neighbors following recursively from the construction . maximality of @xmath57 in @xmath58 follows from the fact that all pairs in the permutation @xmath498 are inversions , by a careful application of lemma [ t.inversions ] . having defined the map @xmath502 , it is clear that the composition @xmath520 is the identity map on the set of all alternating non - crossing trees . thus @xmath501 is surjective , and since both sets have the same ( catalan ) size it follows that @xmath501 is a bijection . [ 1n ] every maximal element in @xmath58 contains @xmath516 as a factor . this follows from theorem [ t.max_eq_alt ] together with the observation ( mentioned in this theorem s proof ) that every alternating non - crossing tree of order @xmath521 contains the edge @xmath508 . recall that the weak order on @xmath4 is characterized by inversion sets , namely : @xmath522 in the ( right ) weak order on @xmath4 if and only if @xmath523 , where @xmath524 ; see , e.g. , ( * ? ? ? * cor . 1.5.2 and prop . 3.1.3 ) . the goal of this section is to provide a similar characterization for @xmath58 . [ defn - lr - inversions ] the set of _ inversions _ of @xmath525 is @xmath526 . a pair @xmath527 is * a _ right inversion _ if there exist @xmath265 such that @xmath528 and @xmath529 , or there exist @xmath530 such that @xmath531 and @xmath532 ; * a _ left inversion _ if there exist @xmath265 such that @xmath533 and @xmath534 , or there exist @xmath530 such that @xmath533 and @xmath535 ; * a _ neutral inversion _ if there exist @xmath530 such that @xmath536 and @xmath529 . [ left - right - explanation ] if @xmath529 , @xmath100 , write @xmath537 for the interval @xmath538\subseteq [ 1,n]$ ] . then @xmath527 is * a right inversion if and only if @xmath539 ; * a left inversion if and only if @xmath540 ; * a neutral inversion if and only if @xmath541 . denote the set of right , left and neutral inversions of @xmath98 by @xmath542 , @xmath543 and @xmath544 , respectively . lemma [ t.inversions ] implies [ t.ln ] for every @xmath98 , @xmath545 a disjoint union . for @xmath546 , @xmath547 $ ] . then @xmath548 is a disjoint union of @xmath549 , @xmath550 and @xmath551 . define the involution @xmath552 by @xmath553 where @xmath32 and @xmath554 \in s_n$ ] . in figure [ figure2-tik ] , for example , @xmath555 corresponds to a reflection of the hasse diagram of @xmath59 in a vertical line through @xmath64 . [ iota ] let @xmath398 \in s_{n-1}$ ] . * for every @xmath98 : @xmath556 and also @xmath557 , @xmath558 and @xmath559 . * @xmath555 is a rank preserving automorphism of the poset @xmath58 . [ determine ] + each of the pairs @xmath560 and @xmath561 determines the word @xmath562 uniquely ; namely , for @xmath47 , @xmath563 if @xmath564 then , of course , all the other equalities hold . let us prove that if @xmath565 and @xmath566 then @xmath564 . we shall use the shorthand notation @xmath567 for @xmath568 . of course , the inversion set @xmath569 determines @xmath570 uniquely since it determines , for any @xmath571 , whether or not @xmath572 . we shall show , by induction on @xmath9 , that @xmath573 and @xmath574 determine @xmath575 for all @xmath576 . this trivially holds for @xmath577 . for the induction step assume that @xmath578 with @xmath450 . by observation [ obs.phi ] , @xmath579 . we can recover @xmath142 from @xmath574 once we show that there are exactly @xmath580 left inversions of @xmath570 involving @xmath581 . indeed , the inversions involving @xmath581 are exactly the ordered pairs @xmath582 such that @xmath583 ( automatically satisfied ) and @xmath584 . by remark [ left - right - explanation ] , such a pair is a _ left _ inversion if and only if the @xmath455-th edge of @xmath585 has both endpoints in the interval @xmath586 $ ] , and is not the edge @xmath587 itself . by proposition [ gy22 ] , there are exactly @xmath580 such edges . thus we have recovered both @xmath37 and @xmath142 . now note that @xmath588 is a product of two disjoint cycles , @xmath589 and @xmath590 , of corresponding lengths @xmath591 and @xmath592 . denote @xmath593 and @xmath594 , and let @xmath595 and @xmath596 be monotone increasing bijections . the word @xmath597 is a shuffle of two subwords , one supported in @xmath598 and the other in @xmath599 . for @xmath600 we do nt yet know @xmath179 , but we do know whether it is supported in @xmath598 or in @xmath599 depending on whether @xmath601 belongs to @xmath598 or to @xmath599 . let @xmath602 and @xmath603 be the sets of indices @xmath455 for which @xmath179 is supported in @xmath598 and in @xmath599 , respectively . denoting @xmath604 , the word @xmath605 is in @xmath606 , and a similar definition gives @xmath607 . by induction , it suffices to show that @xmath573 and @xmath574 determine @xmath608 and @xmath609 ( since @xmath610 can then be retrieved from @xmath611 ) , and similarly for @xmath612 . actually , since @xmath613 is monotone increasing , @xmath614 for any @xmath615 . since @xmath616 and @xmath617 , it follows that @xmath618 and therefore @xmath619 . thus there is a bijection between the inversions of @xmath620 and the relevant inversions of @xmath55 , which may be written ( with a slight abuse of notation ) as @xmath621 the situation for @xmath612 is only slightly different , since indeed @xmath622 but an application of @xmath623 , exchanging @xmath37 with @xmath142 , does not change its relative position in the set . therefore @xmath624 . remark [ left - right - explanation ] now shows that similar relations hold for @xmath625 , completing the induction step . thus @xmath573 and @xmath574 uniquely determine @xmath55 . finally , by observation [ iota].1 , @xmath573 and @xmath626 determine @xmath627 and @xmath628 , thus @xmath629 and @xmath55 . @xmath574 and @xmath626 alone do not suffice to uniquely determine @xmath562 . here is an example from @xmath380 ( see figure [ figure2-tik ] ) : @xmath630 and @xmath631 are distinct , but have @xmath632 as well as @xmath633 ( while @xmath634 ) . [ criterion ] for @xmath635 , if @xmath636 in @xmath58 then @xmath637 , @xmath638 and @xmath639 . it suffices to prove the claim when @xmath57 covers @xmath55 in @xmath58 . thus @xmath640 in @xmath74 , for some @xmath28 , and consequently @xmath641 for the pair @xmath642 . there are three possible cases : @xmath643 , @xmath644 , and @xmath645 . we want to show that @xmath638 and @xmath639 . consider an arbitrary inversion @xmath646 of @xmath55 . it is also an inversion of @xmath57 . if @xmath647 then @xmath648 , where @xmath649 and similarly for @xmath57 . by remark [ left - right - explanation ] , the type of @xmath650 ( left , right or neutral ) is then the same in @xmath55 as in @xmath57 . if @xmath643 , namely if @xmath651 and @xmath652 commute , then @xmath653 and @xmath654 , and therefore _ all _ inversions keep their type upon transforming @xmath55 to @xmath57 : @xmath655 and @xmath656 . assume now that @xmath644 . then @xmath651 and @xmath652 do not commute , and by lemma [ t.inversions ] there exist @xmath133 such that @xmath657 and @xmath658 . since @xmath654 , the inversions involving @xmath659 do not change their type upon transforming @xmath55 to @xmath57 . it remains to consider the inversions involving @xmath660 . consider the graphs @xmath495 and @xmath585 , which are identical except that the former has an edge @xmath661 and the latter an edge @xmath662 ( and both have an edge @xmath157 ) . for any other edge of these graphs , the non - crossing property implies that the corresponding interval @xmath663 belongs to one of the following four classes : contained in @xmath664 $ ] ; contained in @xmath538 $ ] ; containing @xmath665 $ ] ; and disjoint from @xmath665 $ ] . we consider only edges corresponding to inversions with @xmath660 . using remark [ left - right - explanation ] again we see that , for the last three classes , transforming @xmath55 to @xmath57 ( namely replacing @xmath662 by @xmath661 ) does not change the type of inversion with @xmath660 . for the first class , the type changes from neutral to either right or left . thus we conclude that @xmath638 and @xmath639 . the treatment of the case @xmath645 , for which @xmath666 and @xmath658 , is similar . various natural stronger versions of proposition [ criterion ] can be shown to be false by examples from @xmath380 . for example , @xmath667 , @xmath668 and @xmath669 are not even weakly monotone increasing functions of @xmath55 ; and inclusions of any two of the sets @xmath670 , @xmath671 and @xmath625 does not imply @xmath636 in @xmath58 . it would be interesting to know whether the converse of proposition [ criterion ] holds . a @xmath10-analogue of theorem [ t.main3 ] is given in this section . carlitz and riordan @xcite defined a @xmath10-catalan number @xmath672 using the recursion @xmath673 with @xmath674 . for combinatorial interpretations of this number and further information see , e.g. , @xcite and sequence a138158 in @xcite . in particular , @xmath675 counts dyck paths of length @xmath9 with respect to the area above the path ; see , e.g. , @xcite . [ def : qt - catalan ] define _ @xmath676-catalan numbers _ by @xmath677 with @xmath678 [ obs.qt2 ] for every @xmath679 , @xmath680 and @xmath681 for @xmath98 , denote by @xmath682 the cardinality of @xmath542 and by @xmath683 the cardinality of @xmath543 . let @xmath684 denote the set of all maximal elements in @xmath20 . [ qt - prop ] for every positive integer @xmath9 , @xmath685 by induction on @xmath9 . for @xmath686 the claim obviously holds . for the induction step recall that , by corollary [ 1n ] , any maximal element in @xmath249 in @xmath687 contains the factor @xmath688 . denote by @xmath689 the set of maximal elements in @xmath687 whose @xmath690-st factor is @xmath688 . by the proof of theorem [ t.main3 ] , this set is characterized as follows : @xmath691 if and only if @xmath692 is a reduced word for the cycle @xmath693 , where @xmath694 $ ] , and @xmath695 is a reduced word for the cycle @xmath696 , where @xmath697 $ ] . for every @xmath698 the following holds : for every @xmath699 there exist @xmath700 such that @xmath701 ; for every @xmath702 there exist @xmath703 such that @xmath99 . thus , by definition [ defn - lr - inversions ] , all pairs @xmath704 , with @xmath705 and @xmath706 , are neutral inversions ; all pairs @xmath707 , @xmath699 , are left inversions and all pairs @xmath708 , @xmath709 , are right inversions . this implies that , for every @xmath710 , @xmath711 the last equality follows from the induction hypothesis . thus @xmath712 the last equality follows from the defining recursion of @xmath713 together with ( [ qt - eq2 ] ) . [ t.main3-q ] for every positive integer @xmath9 , @xmath714 or equivalently @xmath715 proposition [ qt - prop ] together with ( [ qt - eq1 ] ) . denote the set of alternating non - crossing trees of order @xmath9 by @xmath716 . [ edge_pairs ] a pair of distinct edges @xmath717 in a non - crossing alternating tree @xmath12 is * _ neutral _ if @xmath718 and @xmath719 for some @xmath188 ; * _ right _ if @xmath720 and @xmath721 for some @xmath722 and the edge @xmath723 is in the component of @xmath154 in @xmath724 ( for example , whenever @xmath725 ) ; and * _ left _ if @xmath726 and @xmath727 for some @xmath728 and the edge @xmath723 is in the component of @xmath156 in @xmath724 ( for example , whenever @xmath729 ) . denote the cardinality of right and left and pairs in an alternating non - crossing tree @xmath730 by @xmath731 and @xmath732 , respectively . [ qt - prop - trees ] for every positive integer @xmath9 , @xmath733 each pair of distinct edges is either neutral , right or left . moreover , the edge @xmath64 precedes @xmath501 in the canonical linear order on @xmath12 defined by the map @xmath502 as in the proof of theorem [ t.max_eq_alt ] . hence , a pair of edges in an alternating non - crossing tree is a right ( left ) pair if and only if the corresponding letters in the word @xmath504 form a right ( left ) inversion . it follows that @xmath734 proposition [ qt - prop ] completes the proof . consider the set @xmath735 of dyck paths of length @xmath736 . each @xmath737 is a sequence @xmath738 , with @xmath739 such that @xmath740 and @xmath741 geometrically , @xmath650 corresponds to a lattice path from @xmath742 to @xmath743 , with admissible steps @xmath744 and @xmath745 , which never goes below the horizontal line ( _ baseline _ , in the sequel ) connecting its endpoints . with each @xmath737 associate the following two statistics : @xmath746 geometrically , @xmath747 is the area between the path @xmath650 and the `` horizontal '' path @xmath748 , where the square with vertices @xmath742 , @xmath744 , @xmath749 and @xmath745 is considered to have area @xmath260 ; while @xmath750 is the sum , over all internal local minima of the path @xmath650 , of half the length of the maximal horizontal segment starting at the minimum point and stretching to the right without crossing ( but possibly touching ) @xmath650 . note that @xmath747 is ( a linear function of ) the inversion number of the sequence @xmath650 . [ dyck ] for every positive integer @xmath9 @xmath751 note that for @xmath752 this is a classical result @xcite . it suffices to show that the lhs , to be denoted @xmath753 , satisfies the recursion in definition [ def : qt - catalan ] ; it clearly satisfies @xmath754 . let @xmath755 , and define @xmath756 , so that @xmath757 is the first time that @xmath650 returns to the baseline . clearly @xmath710 . we can write @xmath650 as the concatenation of sequences @xmath758 where @xmath759 and @xmath760 . now clearly @xmath761 and @xmath762 since the path @xmath650 has an internal local minimum at @xmath763 , in addition to the internal local minima of @xmath764 and @xmath765 . thus @xmath766 and this completes the proof . @xmath767 and @xmath768 are equidistributed : @xmath769 proposition [ dyck ] together with both parts of observation [ obs.qt2 ] . @xmath770 proposition [ qt - prop ] together with proposition [ dyck ] . let @xmath771 be a finite connected undirected graph , and let @xmath772 be the corresponding metric on its vertex set @xmath773 . the _ radius _ of @xmath503 is @xmath774 and its _ diameter _ is @xmath775 recall the hurwitz graph @xmath1 ( definition [ d.hurwitz_graph ] ) , and let @xmath776 be the ( right ) cayley graph of the group @xmath74 with respect to its standard coxeter generators . [ t.phi_contracts ] the map @xmath777 is a contraction : @xmath778 it suffices to show that if @xmath779 then @xmath780 . indeed , if @xmath779 in @xmath1 then either @xmath419 or @xmath420 for some @xmath142 . thus , by lemma [ t.main-lemma ] , @xmath781 , so that indeed @xmath780 . the radius of the hurwitz graph @xmath782 by definition [ d.order ] , only edges connecting vertices of the same rank are deleted during the passage from @xmath1 to the hasse diagram of @xmath58 . it follows that the distance @xmath783 from the unique minimal element @xmath64 of @xmath58 to an arbitrary element @xmath231 is the same , whether measured in @xmath1 or in the hasse diagram . thus , by theorem [ t.main2 ] and a well - known property of @xmath74 , @xmath784 it follows that @xmath785 consider now an arbitrary @xmath518 . by the proof of theorem [ t.main2 ] , @xmath786 is surjective . thus there exists @xmath231 with @xmath787 where @xmath398 \in s_{n-1}$ ] . for this @xmath249 it follows from lemma [ t.phi_contracts ] that @xmath788 and therefore @xmath789 clearly , for any graph @xmath503 , @xmath790 the diameter of @xmath1 satisfies @xmath791 the upper bound can be improved . [ t.diameter ] the diameter of @xmath1 satisfies @xmath792 we shall use the bubble sort " algorithm to transform any two given words @xmath793 to a common word by applying various @xmath191 and @xmath192 operators . for the initial step note that , since every tree has at least one leaf ( vertex of degree @xmath260 ) , by proposition [ gy22](i ) applied to @xmath57 there is an @xmath794 which appears in exactly one of the @xmath795 factors ( transpositions ) in @xmath57 . this unique @xmath37-factor " is within distance @xmath796 from either the first or the last factor ( end factor " ) in @xmath57 . apply a sequence of @xmath191 and @xmath192 operators to push " this @xmath37-factor towards the closest end factor without creating any additional @xmath37-factors . this takes at most @xmath796 steps , resulting in a word whose only @xmath37-factor is an end factor . on the other hand , consider the word @xmath249 : it may contain more than one @xmath37-factor , and we want to push all of them toward the end factor ( at the same end as in @xmath57 ) . this can be done by pushing the @xmath37-factor which is furthest from that end factor . at each step the furthermost @xmath37-factor gets closer to the end factor , and the total number of @xmath37-factors is unchanged or is reduced by @xmath260 . after at most @xmath797 steps we get from @xmath249 a word whose only @xmath37-factor is the end factor . now observe that if the only occurence of @xmath37 in a decomposition of @xmath798 ( @xmath799 $ ] for this initial step ) is in the first ( leftmost ) factor then this factor is @xmath800 ; and if the only occurence is in the last factor then this factor is @xmath801 . this implies that the end @xmath37-factors in the two words obtained from @xmath57 and @xmath249 are equal . for the rest of the algorithm we keep this end factor fixed , and work only on the subwords containing the other factors . we repeat the procedure for the two words of length @xmath802 thus obtained from @xmath57 and @xmath249 . obviously , after @xmath802 such steps we get two words of length @xmath260 , which are necessarily equal . the total number of operators used is at most @xmath803 and this is an upper bound on the diameter of @xmath1 . an improvement of proposition [ t.diameter ] , bounding the diameter of the hurwitz graph , is desired . [ d.hurwitz_graph-coxeter ] let @xmath62 be a coxeter group , @xmath12 its set of reflections , and @xmath16 a coxeter element in @xmath62 . the _ hurwitz graph _ @xmath807 is the graph with vertices corresponding to all reduced words expressing @xmath16 in the alphabet @xmath12 of all reflections , where adjacency is determined by a local right or left shift : two words are adjacent in the graph if they agree in all but two adjacent letters , which are @xmath5 in one word and either @xmath6 or @xmath7 in the other . this conjecture is supported by computer results for @xmath810 . the computations also hint that the number of antipodes of @xmath811 ( i.e. , the number of elements of maximal rank in the associated weak order ) is given by sequence a054441 in @xcite which involves central binomial coefficients , namely the catalan numbers of type @xmath53 . * acknowledgements . * thanks to c. athanasiadis , o. bernardi , a. kalka , c. keller , h. y. khachatryan , s. margolis , p. mcnamara , a. postnikov and v. reiner for useful discussions , comments and references . thanks also to the anonymous referees for their stimulating comments . j. dnes , _ the representation of a permutation as the product of a minimal number of transpositions and its connection with the theory of graphs _ , publ . math . * 4 * ( 1959 ) , 6371 . i. m. gelfand , m. i. graev and a. postnikov , _ combinatorics of hypergeometric functions associated with positive roots _ , in : arnold - gelfand mathematical seminars : geometry and singularity theory , birkhuser , 1997 , 205-221 . n. j. a. sloane , the on - line encyclopedia of integer sequences . v. strehl , _ minimal transitive products of transpositions the reconstruction of a proof of a. hurwitz _ , sm . * 37 * ( 1996 ) , art . s37c , 12 pp . ( electronic ) .

a microscopic theory for the generation and propagation of coherent la phonons in pseudomorphically strained wurtzite ( 0001 ) multi - quantum well ( mqw ) diodes is presented . the generation of coherent la phonons is driven by photoexcitation of electron - hole pairs by an ultrafast gaussian pump laser and is treated theoretically using the density matrix formalism . we use realistic wurtzite bandstructures taking valence - band mixing and strain - induced piezoelectric fields into account . in addition , the many - body coulomb interaction is treated in the screened time - dependent hartree - fock approximation . we find that under typical experimental conditions , our microscopic theory can be simplified and mapped onto a loaded string problem which can be easily solved .

cond-mat0101459

in recent years , experiments have shown that optical excitation of electron - hole pairs in semiconductors by ultrafast lasers can coherently excite longitudinal optical phonon modes in semiconductors . @xcite in uniform bulk semiconductors , since the laser wavelength is much larger than the lattice spacing , the photo - generated carriers are typically excited by the optical pump over spatial areas that are much larger than the lattice unit cell . as a result , the excited carrier populations are generated in a macroscopic state and the carrier density matrix has only a @xmath2 fourier component . coupling of the photoexcited carriers to the phonons leads only to coherent optical phonon modes with @xmath2 . since the frequency of the @xmath2 acoustic phonon is zero , coherent acoustic phonons are not excited in bulk semiconductors . in semiconductor superlattices , even though the laser pump has a wavelength large compared to the lattice spacing , the pump can preferentially generate electron - hole pairs in the wells . the result is to create photoexcited carrier distributions that have the periodicity of the superlattice . since the density matrix of the photo - excited carrier populations now has a @xmath3 fourier component , the photo - excited carriers can not only couple to the optical phonon modes , but they can also generate coherent acoustic phonon modes with a nonzero frequency and wavevector @xmath4 where @xmath5 is the superlattice period . in superlattices , the coherent phonon oscillation of zone folded acoustic phonons has been observed in @xmath6 superlattices . however , the reflection modulation , observed to be on the order of @xmath7 , is very small . @xcite recently , c .- k . @xcite reported studies of coherent acoustic phonon oscillations in wurtzite ( 0001 ) @xmath0 multi - quantum well samples with strain induced piezoelectric fields . owing to the strong piezoelectric fields at the interfaces , huge coherent acoustic phonon oscillations were observed . the oscillations were strong enough to be seen in the transmision ( rather than the usual reflectivity ) with @xmath8 . the oscillation frequency , in the thz range , corresponding to the la phonon frequency with @xmath4 , varied between samples in accordance with their different superlattice periods , @xmath5 . in this paper , we formulate a microscopic model for the generation of coherent acoustic phonons in strained wurtzite superlattices via ultrafast laser photo - excitation of real carriers . whereas in bulk systems the microscopic theory of coherent lo phonons can be mapped onto a forced oscillator model @xcite , we show that coherent la phonon generation in superlattices , under appropriate conditions , can be mapped onto a _ loaded string model _ which is readily solved for the lattice displacement . since acoustic phonons are almost the same in the well as in the barrier , to lowest order we can treat the string as being uniform @xcite . the forcing term on the string , however , is not uniform since photoexcitation of carriers occurs only in the wells . our paper thus provides justification for using a simple , uniform string model with a nonuniform forcing term , rather than a more complicated microscopic theory . in addition , we provide a microscopic expression for the forcing term to use in the simplified string model . the string model provides additional insight into the physics of the coherent la phonons . in this section , we derive the microscopic theory for coherent acoustic phonon generation in superlattices and multiple quantum wells , including the effects of ( i ) bandstructure , ( ii ) strain , ( iii ) piezoelectric fields , ( iv ) coulomb interactions , and ( v ) laser optical excitation . in section iii , we will show how this reduces to a simplified driven uniform string model with a nonuniform forcing term and a microscopic expression for the forcing term . we model photogeneration of electrons and holes and the subsequant excitation of coherent acoustic phonons in a multi - quantum well ( mqw ) @xmath1 diode shown schematically in fig . [ diode ] . the intrinsic active region consists of a left @xmath9 buffer region , several pseudomorphically strained ( 0001 ) @xmath10 quantum wells sandwiched between @xmath9 barriers , and a right @xmath9 buffer region as indicated in the figure . the @xmath11 and @xmath12 regions are assumed to be abruptly terminated @xmath13- and @xmath14-doped @xmath9 bulk layers separated by a distance , @xmath5 , across which a voltage drop , @xmath15 , is maintained . photoexcitation of carriers is achieved by means of an ultrafast laser pulse incident normally along the ( 0001 ) growth direction , taken to coincide with the @xmath16-axis . in bulk systems , the conduction and valence bands in wurzite crystals including the effects of strain are treated using effective mass theory . near the band edge , the effective mass hamiltonian for electrons is described by a @xmath17 matrix which depends explicitly on electron wavevector , * k * , and the strain tensor , * @xmath18*. the electron bloch basis states are taken to be [ cbasis ] @xmath19 @xmath20 the conduction band hamiltonian is diagonal and we have ( relative to the bottom of the conduction band ) ( refs . and ) @xmath21 where @xmath22 is the identity matrix . the electron effective masses along z ( taken parallel to the c - axis ) and in the @xmath23 plane are @xmath24 and @xmath25 , respectively , @xmath26 , and @xmath27 , @xmath28 and @xmath29 are strain tensor components , and @xmath30 and @xmath31 are the deformation potentials . the hamiltonian for the valence bands is a @xmath32 matrix . following ref . , the hamiltonian ( relative to the top of the valence band ) can be block diagonalized into two degnerate @xmath33 submatrices if we adopt the bloch basis states [ vbasis ] @xmath34 @xmath35 @xmath36 @xmath37 @xmath38 @xmath39 the phase factors , @xmath40 and @xmath41 , are functions of the angle @xmath42 and are given by [ phases ] @xmath43 @xmath44 the block diagonalized hamiltonian can be written as @xmath45 where the upper and lower blocks of the hamiltonian are [ hulmatrix ] @xmath46 and @xmath47 the elements appearing in the @xmath33 hamiltonian matrices are [ valencematrixelements ] @xmath48 @xmath49 @xmath50 @xmath51 @xmath52 @xmath53 @xmath54 in eq . ( [ valencematrixelements ] ) , the @xmath55 are effective mass parameters , the @xmath56 are the bir - pikus deformation potentials , and the @xmath57 are related to the crystal field splitting , @xmath58 , and spin - orbit splitting , @xmath59 , by @xmath60 and @xmath61 . @xmath62 is the free electron mass . in quantum confined systems such as the @xmath1 diode shown in fig . [ diode ] , we must modify the bulk hamiltonian . the finite mqw structure breaks translational symmetry along the @xmath16 direction but not in the @xmath23 plane . thus , quantum confinement of carriers in the mqw active region gives rise to a set of two - dimensional subbands . the wavefunctions in the envelope function approximation are @xmath63 where @xmath64 refers to conduction or valence subbands , @xmath14 is the subband index , @xmath65 is the two - dimensional wavevector , and @xmath66 labels the spinor component . for conduction subbands , ( @xmath40=c ) @xmath67 while for valence subbands ( @xmath40=v ) @xmath68 . the slowly varying envelope functions @xmath69 are real and depend only on @xmath70 , while the rapidly - varying bloch basis states @xmath71 defined in eqs . ( [ cbasis ] ) and ( [ vbasis ] ) depend on @xmath72 in the case of valence subbands as given in eq . ( [ phases ] ) . the area of the mqw sample in the @xmath23 plane is @xmath73 , and @xmath74 is the projection of * r * in the plane . the envelope functions satisfy a set of effective - mass schrdinger equations @xmath75 \right\ } f^{\alpha}_{n , k , j ' } ( z ) = 0,\ ] ] subject to the boundary conditions @xmath76 where @xmath5 is again the length of the mqw diode structure ( c.f . [ diode ] ) , @xmath77 are the quantum confinement potentials for conduction and valence electrons . @xmath78 are the energy eigenvalues for the @xmath14th conduction or valence subband . note that in the envelope function approximation , the subband energy depends only on the magnitude @xmath79 of the transverse wavevector and not on the angle @xmath72 . for the quantum confined case , the matrix operators @xmath80 depend explicitly on @xmath16 and are obtained by making the replacement @xmath81 and letting all material parameters be @xmath16-dependent operators in the matrices @xmath82 given in eqs . ( [ hcmatrix ] ) and ( [ hvmatrix ] ) . to ensure the hermitian property of the hamiltonian , we make the operator replacements @xcite @xmath83 and @xmath84 .\ ] ] the quantum confinement potentials @xmath85 and @xmath86 arise from ( i ) bandgap discontinuities between well and barrier regions , ( ii ) the strain - induced piezoelectric field , and ( iii ) the time - dependent electric field due to photoexcited electrons and holes . thus , @xmath87 and the bandstructure is explicitly time - dependent . material parameters for @xmath88 and @xmath9 used in this work can be found in table [ parametertable ] . for @xmath10 alloys , we interpolate between the @xmath9 and @xmath88 values listed in the table . we obtain electron effective masses by linearly interpolating the reciprocals of the masses as a function of the indium concentration @xmath89 , i.e. , the concentration dependent effective masses are taken to be @xmath90 and @xmath91 for the alloy band gap , @xmath92 , we use an expression incorporating a bowing paremeter : @xmath93 where the bowing parameter , @xmath94 . @xcite for all other material parameters , we use linear interpolation in @xmath89 to obtain values for the alloy . since we ca nt find deformation potentials for @xmath88 , we use @xmath9 values by default . in the absence of values of @xmath95 for either @xmath9 or @xmath88 , we use linear interpolation in @xmath89 to obtain @xmath96 and simply take @xmath97 . if the z - dependent band gap in the mqw is @xmath98 as determined from eq . ( [ bandgap ] ) and the indium concentration profile , and @xmath99 $ ] is the minimum band gap in the structure , then the confinement potentials for conduction and valence electrons are defined as @xmath100 @xmath101 where the conduction band offset is taken as @xmath102 . @xcite with these definitions for the @xmath103 , the zero of the gap confinement potential is placed at the top of the valence band profile . the confinement potentials due to the strain - induced piezoelectric field are given by @xmath104 where @xmath105 is the electric charge and @xmath106 is the strain - induced piezoelectric field . in a pseudomorphically strained mqw diode , the bulk source and drain ( assumed to have identical composition ) are unstrained while the in - plane mqw lattice constants adjust to the source and drain values . for a mqw grown along [ 0001 ] ( the z - direction ) the z - dependent strain is @xcite @xmath107 here @xmath108 is the lattice constant in the source and drain and @xmath109 is the z - dependent lattice constant in the mqw . minimizing the overall strain energy , we find @xcite @xmath110 where @xmath111 and @xmath112 are z - dependent elastic constants . there are several issues concerning strain that one can worry about . one is the critical well thickness beyond which the strain relaxes . studies have shown @xcite that with 10 % indium , and 6 nm well width , a 350 kv / cm field is measured , implying that the well is not fully relaxed and justifies using the pseudomorphic strain approximation as we do in this paper for wells having 6 % indium and thickness near 4 nm . for thick wells , this pseudormorphic strain approximation clearly will begin to break down and a more detailed model will be needed . interface roughness can also play a role . the roughness will not significantly affect the acoustic phonon modes ( see introduction ) but may affect the photogeneration of carriers @xcite . for simplicity , we do not consider interface roughness . the strain - induced polarization directed along the z - direction is given by @xmath113 where @xmath114 and @xmath115 are z - dependent piezoelectric constants . the unscreened piezoelectric field in the diode is obtained from the requirement that the electric displacement vanishes @xcite . thus , @xmath116 where @xmath117 is a constant polarization induced by externally applied voltages and @xmath118 is the position - dependent static dielectric constant . the value of @xmath117 is obtained from the voltage drop across the diode ( of length @xmath5 ) in the unscreened limit , i.e. with no photoexcited carriers . in this limit , the voltage drop between source and drain due to the induced piezoelectric field is just @xmath119 from which @xmath117 can be determined . when photoexcited electrons and holes are generated by the laser , then there is an additional time - dependent confinement potential , @xmath120 this potential is obtained by solving the poisson equation in the diode for @xmath121 subject to the boundary condition @xmath122 here , @xmath123 is the total electric field and is just the sum of the strain induced electric field and the field due to photogenerated electrons and holes , i.e. , @xmath124 finally , we can write an effective - mass schrdinger equation for the conduction electron envelope functions in terms of an effedctive electron potential , @xmath125 : @xmath126 where the effective electron potential is @xmath127 similar expressions can be derived for valence electrons . we arrive at a set of coupled ordinary differential equations ( ode s ) subject to the two - point boundary value condition of eq . ( [ boundarycondition ] ) . these are solved for the envelope functions and subband energies . in practice , we introduce a uniform grid , @xmath128 , along the z - direction and finite - difference the effective mass schrdinger equations to obtain a matrix eigenvalue problem which can be solved using standard matrix eigenvalue routines . the resulting eigenvalues are the subband energies , @xmath78 , and the corresponding eigenvectors are the envelope functions , @xmath129 , defined on the finite difference mesh . we next describe the second quantized hamiltonians for electrons moving freely in the mqw interacting via a screened coulomb potential . we denote creation and destruction operators for electrons in conduction and valence subbands by @xmath130 and @xmath131 , respectively . the second quantized hamiltonian for free electrons and holes is simply @xmath132 the coulomb interaction hamiltonian is given by @xmath133 eq . ( [ coulombhamiltonian ] ) describes two - body interactions where electrons in states @xmath134 and @xmath135 scatter to subband states @xmath136 and @xmath137 , respectively . note that to simplify things , we have neglected terms corresponding to coulomb induced interband transitions ( the `` diagonal approximation '' ) since these are very unfavorable energetically . @xcite the electrons thus stay in their original subbands ( though they may scatter off of other electrons in different subbands ) and exchange crystal momentum @xmath138 . the matrix elements describing the strength of these transitions are given by @xmath139 from eq . ( [ exactvmatrix ] ) , it is apparent that the symmetry relation @xmath140 must hold . the fourier transform in the @xmath23 plane of the screened coulomb potential depends only on @xmath141 and @xmath142 and is given by @xmath143 to describe screening , we adopt an effective pseudodynamic dielectric function of the form @xmath144 in the pseudodynamic screening model , we completely neglect screening by the massive holes and treat screening by the lighter conduction electrons in the static screening limit . the screening wavevector , @xmath145 , is computed in the @xmath146 limit . thus @xcite @xmath147 where @xmath148 , the two - dimensional conduction electron density , is related to an effective chemical potential , @xmath149 , by @xmath150 \ . \label{n2d}\ ] ] in eq . ( [ n2d ] ) , @xmath151 is the conduction subband energy evaluated at @xmath152 . in our simulation , the value of @xmath149 is obtained by requiring that @xmath148 , evaluated using eq . ( [ n2d ] ) , be equal to the value @xmath153 obtained from the time dependent conduction electron distribution functions , @xmath154 . electron - hole pairs are created by the pump laser and we treat the electric field of the laser in the semiclassical dipole approximation . in this approximation , the electron - laser interaction hamiltonian is @xmath155 , \ ] ] where h.c . denotes the hermitian conjugate of the first term . the laser field is @xmath156 and the dipole matrix elements are @xmath157 the vector operator , @xmath158 , is a @xmath159 matrix with @xmath89 , @xmath160 , and @xmath16 components . thus , @xmath161 where @xmath162 , @xmath163 , and @xmath164 are unit vectors and [ dcv ] @xmath165 \ .\ ] ] @xmath166 , \ ] ] @xmath167 .\ ] ] the @xmath168 vector operator , @xmath169 , is related to the @xmath159 operator , @xmath158 , by @xmath170 in eq . ( [ dcv ] ) , @xmath40 and @xmath41 are the @xmath72-dependent phase factors defined in eq . ( [ phases ] ) and the kane parameters , @xmath171 and @xmath172 , for wurtzite materials are related to the effective masses and energy gaps by @xcite @xmath173 @xmath174 for the semiclassical laser field , we write the real electric field as @xmath175 \,\ ] ] where @xmath176 is the photon frequency , @xmath177 is a complex unit polarization vector , and @xmath178 is the pulse shape envelope function . we assume a gaussian pulse shape @xmath179 \label{gaussianpulseshape}\ ] ] centered at @xmath180 with an intensity full width at half maximum ( fwhm ) of @xmath181 . the maximum electric field strength , @xmath182 , is related to the pump fluence , @xmath183 , by @xmath184 where @xmath185 is the index of refraction at the photon frequency . for linearly polarized light incident normally on the mqw , the polarization vectors are real and given by either @xmath162 or @xmath163 . for circularly polarized light , the polarization vectors are complex and given by @xcite @xmath186 in eq . ( [ polarization ] ) , the upper sign refers to left circularly polarized light ( positive helicity ) and the lower sign refers to right circularly polarized light ( negative helicity ) . we treat the acoustic phonons in the mqw as bulk - like plane - wave states with wavevector @xmath187 . since the system exhibits cylindrical symmetry , only @xmath188 longitudinal acoustic phonons are coupled by the electron - phonon interaction . the free la phonon hamiltonian can be written as @xmath189 where @xmath190 and @xmath191 are creation and destruction operators for la phonons with wavevector @xmath188 . the wave vector component , @xmath192 , of la phonons in the mqw is thus defined in an extended zone scheme where @xmath193 . the phonon dispersion relation is given by a linear relation @xmath194 where @xmath195 is the mass density and @xmath196 is just the la phonon sound speed for propagation parallel to @xmath164 . @xcite in computing the la sound speed in the linear phonon dispersion relation of eq . ( [ phonondispersion ] ) , we neglect the @xmath16-dependence of the material parmeters and use bulk @xmath9 values for @xmath197 and @xmath195 . the la phonons in wurtzite mqw s interact with the electrons through deformation potential and screened piezoelectric scattering . the electron - la phonon interaction in an mqw is governed by the hamiltonian @xmath198 this hamiltonian describes the scattering of an electron from subband state @xmath199 to subband state @xmath134 with either the emission or absorption of an la phonon . we note that the electron wave vector , @xmath200 , in the @xmath23 plane is conserved in this process since , as noted earlier , the phonon wave vector in the @xmath23 plane is zero . the interaction matrix elements describing deformation and screened piezoelectric scattering are @xmath201 \ , \label{phononmatrix}\end{aligned}\ ] ] where @xmath202 is the crystal volume . the first term in eq . ( [ phononmatrix ] ) desribes deformation potential scattering while the second term describes screened piezoelectric scattering . the relative strengths of the various transitions are determined by form factors for deformation potential and piezoelectric scattering . the form factor for screened piezoelectric scattering is given by @xmath203 while the form factor for deformation potential scattering is defined to be @xmath204 the form factor for deformation potential scattering is similar to the form factor for piezoelectric scattering except that in summing over spinor components , @xmath66 , the terms are weighted by @xmath66-dependent deformation potentials @xmath205 which can be representd by the row vectors @xmath206 @xmath207 for conduction and valence electrons , respectively . we define statistical operators in terms of the electron and phonon eigenstates . the electron density matrix is @xmath208 where @xmath209 denotes the statistial average of the non - equlilbrium state of the system . the interband components of the density matrix , @xmath210 and @xmath211 , describe the coherence between conduction and valence electrons in subbands @xmath14 and @xmath212 and are related to the optical polarization . the intraband components of the density matrix , @xmath213 describe correlations between different subbands of the same carrier type if @xmath214 . if @xmath215 , @xmath216 is just the carrier distribution function for electrons in the subband state , @xmath217 , defined in eq . ( [ wavefunction ] ) . the coherent phonon amplitude of the @xmath192-th phonon mode , @xmath218 , is defined to be @xcite @xmath219 the coherent phonon amplitude is related to the macroscopic lattice displacement , @xmath220 , and velocity , @xmath221 , through the relations @xmath222 @xmath223 the coherent phonon amplitude , @xmath224 , will vanish if there are a definite number of phonons in the mode , i.e. , if the phonon oscillator is in one of its energy eigenstates , @xmath218 . in this case , there is no macroscopic displacement of the lattice . the coherent phonon distribution is @xcite @xmath225 and the total phonon distribution , @xmath226 , can be separated into coherent and incoherent contributions as follows ; @xmath227 in general , a mode can have a number of both coherent and incoherent phonons , but only the coherent phonons contribute to the macroscopic lattice displacement . we note that at the beginning of the experiment , there are no coherent phonons present , i.e. @xmath228 , and the incoherent phonon population is described by a thermal distribution , @xmath229 . in this section we develop equations of motion for the electron density matrices and coherent phonon amplitudes . the electron density matrices obey the general equations of motion @xmath230 \right > \ , \label{blochequation}\ ] ] where @xmath231 $ ] denotes the commutator and @xmath209 denotes the average over an initial ensemble . the density matrices are defined in the electron picture and initially the valence bands are filled while the conduction bands are empty . we have @xmath232 and @xmath233 which implies @xmath234 the total hamiltonian , @xmath235 , is the sum of the hamiltonians described in the previous sections , i.e. @xmath236 in deriving equations of motion for the density matrices , we make the ansatz that the density matrices depend only on @xmath237 . we use the rotating wave approximation ( rwa ) to factor out the rapid @xmath238 behavior of the interband density matrix elements , @xmath239 . in the rwa , we have @xmath240 where @xmath241 is a slowly varying envelope function . in addition , we treat the coulomb interaction in the time - dependent hartree - fock approximation by factoring four - operator averages arising from @xmath242 into appropriate products of two - operator averages as described in ref . . the resulting equations of motion for the density matrices are [ blochequations ] @xmath243 @xmath244 @xmath245 the equations of motion for @xmath246 are redundant since @xmath247 . the first terms on the right - hand side of eq . ( [ blochequations ] ) describe the free oscillation of the density matrices in the renormalized singel particle energy bands . the time - dependent single particle energies are @xmath248 where @xmath249 are the single particle subband energies in the absence of conduction electrons and holes and @xmath250 describes the time - dependent renormalization of the single particle subbands . the renormalization energies , @xmath250 , are the diagonal elements of a generalized renormalization energy matrix ( in the subband indices ) @xmath251 the first term in the renormalization energy matrix ( [ renormalization ] ) is the generalized exchange self - energy matrix arising from the coulomb interaction and is given by @xmath252 where @xmath253 are angular averaged coulomb interaction matrix elements . the second term in eq . ( [ renormalization ] ) accounts for renormalization due to coupling of carriers to coherent acoustic phonons . we have @xmath254 where @xmath224 is the coherent phonon amplitude and the electron - phonon matrix elements , @xmath255 , are defined in eq . ( [ phononmatrix ] ) . the self energy corrections in eq . ( [ selfenergies ] ) are small , though the can be important in some circumstances . in computing the angular averaged coulomb matrix elements in eq . ( [ selfenergies ] ) , we assume small momentum transfer , @xmath138 , and use the fact that the envelope functions , @xmath256 , depend weakly on @xmath79 to obtain an effective interaction , @xmath257 where , by definition , @xmath258 the effective coulomb interaction , @xmath259 , defined in eq . ( [ vapprox ] ) is an even function of @xmath260 and is symmetric in @xmath40 and @xmath79 , thus preserving the symmetry relation ( [ symmetry ] ) . preserving this symmetry is essential in order to maintain conservation of carriers in the scattering process . the second terms in eq . ( [ blochequations ] ) describe photoexcitation of electron - hole pairs by the pump laser . the system reacts to an effective field which is the sum of the applied field and the dipole field of the electron - hole excitations . this gives rise to a matrix of generalized rabi frequencies in the subband indices @xmath261 which can be shown to satisfy the symmetry relations @xmath262 the gaussian pump envelope function , @xmath178 , is defined in eq . ( [ gaussianpulseshape ] ) and the optical dipole matrix elements @xmath263 are angular averages in the @xmath23 plane of the vector dipole matrices dotted into the polarization vector . from eq . ( [ dcv ] ) , the @xmath72-dependence of @xmath264 only appears in @xmath158 and we can get the angular averages by setting @xmath265 and @xmath266 in the @xmath159 matrices @xmath267 , @xmath268 and @xmath269 defined in eq . ( [ dcv ] ) . the last terms in eq . ( [ blochequations ] ) are similar in structure to the renormalization corrections in the hartree - fock energies but are more complicated due to mixing among subbands and involve the off - diagonal components of @xmath270 . the prime on the summation sign indicates that terms containing factors of @xmath271 are excluded from the sum since these terms have already been incorporated in the renormalized hartree - fock energies in eq . ( [ hartreeenergies ] ) . the coherent phonon amplitudes , @xmath224 , satisfy the driven harmonic oscillator equations @xmath272 subject to the initial conditions @xmath273 the closed set of coupled partial differential equations , ( [ blochequations ] ) and ( [ dqequation ] ) , for the carrier density matrices and coherent phonon amplitudes are converted into a set of coupled ordinary differential equations ( ode s ) by discretizing @xmath79 and @xmath192 and solving for @xmath274 and @xmath275 for each of the mesh points @xmath276 and @xmath277 . the resulting initial value ode problem is then solved using a standard adaptive step - size runge kutta routine @xcite . the phonon distributions do not appear in the coupled set of equations ( [ blochequations ] ) and ( [ dqequation ] ) . if necessary , they can be determined from @xmath278 and the pair of equations @xmath279 and @xmath280 in eq . ( [ bqequation ] ) , @xmath281 satisfies the initial condition @xmath282 . for the incoherent phonon distribution , @xmath283 so no incoherent phonons are generated and the incoherent phonon population maintains its initial thermal equilibrium distribution . the microscopic equations are rather daunting and detailed . in this section , we show how they can be simplified ( under certain conditions ) to a more tractable model , manely that of a driven uniform string , provided one uses the appropriate driving function , @xmath284 , which is nonuniform . the microscopics , including details of the superlattice band structure and photogeneration process are included within the driving function . in our detailed numerical simulations , we use the full microscopic formalism discussed in the previous sections . however , we gain a lot of insight if we can deal with the lattice displacement , @xmath220 , directly . if we assume that the acoustic phonon dispersion relation is linear as in eq . ( [ phonondispersion ] ) , then we find that @xmath220 satisfies the loaded string equation @xmath285 subject to the initial conditions @xmath286 the la sound speed , @xmath196 , is defined in eq . ( [ phonondispersion ] ) , and the driving function , @xmath284 , is given by @xmath287 one may question whether a linear phonon dispersion relation is valid in a superlattice . for small wavevector , @xmath192 , for which elasticity theory holds , the dispersion relation for la phonons in a superlattice is linear with a dispersion , @xmath288 , where @xmath289 is the `` average '' sound speed of la phonons in the well and barriers @xcite . this , in fact , has been experimentally verified in ingan / gan superlattice samples studied by c .- k . al . @xcite . note that coherent acoustic phonon generation in a superlattice is qualitatively different than coherent optical phonon generation in a bulk system where only the @xmath2 optic mode can be excited . as a result , both the amplitude , @xmath220 , and the fourier transform of the amplitude , @xmath224 , for an optic mode in bulk satisfy a forced oscillator equation . for the nonuniform , multiple quantum well case , one can excite acoustic modes with @xmath3 . the fourier transform of the amplitude @xmath224 of a coherent acoustic phonon obeys a forced oscillator equation , but owing to the linear dependence of @xmath290 on @xmath192 , the amplitude itself , @xmath220 , obeys a 1-d wave equation with a forcing term , @xmath284 . another important point is that eq . ( [ waveequation ] ) can be taken to be a _ uniform _ string with a _ non - uniform _ forcing function . this is because the speed of sound is approximately the same in both the @xmath9 and @xmath291 layers ( a more detailed theory would take into account differences in the sound velocities in each layer ) . for propagation of acoustic modes one can neglect , to lowest order , the differences between the different layers ( this is not true for the optic modes ) . the non - uniformity of the forcing function @xmath284 results from differences in the absorption ( not sound velocity ) in the well and barrier layers and is therefore @xmath16 dependent . we thus see from eq . ( [ waveequation ] ) that understanding coherent acoustic phonons in multiple quantum wells is equivalent to understanding a _ uniform _ string with an _ inhomogeneous _ forcing term , @xmath284 , containing the microscopics . to simplify eq . ( [ szt ] ) , we neglect valence band mixing and assume the effective masses , sound speeds , and coupling constants are uniform over regions where @xmath292 , i.e. , in regions where carriers are being photogenerated . we also assume the pump pulses are weak enough so that screening of the piezoelectric interaction can be neglected . finally , if the pump duration is long enough so that transient effects associated with photogeneration of virtual carriers can be ignored , then the off - diagonal elements of the carrier density matrices in eq . ( [ szt ] ) can be dropped . in this case , the driving function takes on the simple form @xmath293 where the summation index , @xmath294 , runs over carrier species , i.e. , conduction electrons , heavy holes , light holes , and crystal field split holes . ( [ sztsimple ] ) says that each carrier species makes a separate contribution to the driving function . the partial driving functions , @xmath295 , are @xmath296 where the plus sign is used for conduction electrons and the minus sign is used for holes . here @xmath297 is the photognerated electron or hole number density , which is real and positive , and @xmath195 is the mass density . we note that the loaded string equation for the propagation of coherent phonons together with the simplified driving function in eqs . ( [ sztsimple ] ) and ( [ snu ] ) have also been independently derived by other authors in the limit @xmath298 . @xcite in eq . ( [ snu ] ) , the partial driving function for a given species is obtained by applying a simple operator to the photogenerated carrier density . this operator is a sum of two terms , the first due to deformation potential scattering and the second to piezoelectric scattering . the piezoelectric coupling constant , @xmath299 , is the same for all carrier species , while the deformation potential , @xmath300 , depends on the species . for conduction electrons , @xmath301 , for heavy or light holes , @xmath302 , and for crystal field split holes , @xmath303 . it is interesting to note that planck s constant does not appear in either the loaded string equation , ( [ waveequation ] ) , or in its associated driving function defined in eqs . ( [ sztsimple ] ) and ( [ snu ] ) . thus , we find that coherent la phonon oscillations in mqw s can be viewed as an essentially classical phenomenon , an observation that was made in the context of coherent lo phonon oscillations in bulk semiconductors by kuznetsov and stanton in ref . . the driving function , @xmath284 , satisfies the sum rule @xmath304 this is most easily seen from eqs . ( [ sztsimple ] ) and @xmath305 , but it also holds for the general expression in eq . ( [ szt ] ) . the significance of the sum rule is readily appreciated . after the pump dies away , the carrier density in eq . ( [ snu ] ) , neglecting tunneling between wells , is essentially constant and thus @xmath284 is time - independent . in the loaded string analogy , the integral of the driving function over position is proportional to the average force per unit length on the string . if this integral were non - zero , then the center of mass of the string would undergo a constant acceleration resulting in the buildup of an infinite amount of kinetic energy . such an alarming result in the context of coherent la phonons is precluded by the sum rule in eq . ( [ sumrule ] ) . for a given driving function , the wave equation ( [ waveequation ] ) , together with the initial conditions ( [ uboundarycondition ] ) , can be solved for the coherent phonon lattice displacement using the green s function method.@xcite thus , @xmath306 in our mqw diode model , the substrate is assumed to be infinite and the green s function in this case is just @xmath307 where @xmath308 is the heaviside step function . we note that the loaded string model described above is not restricted to the special case of an infinite substrate and can be extended to study the generation and propagation of coherent la phonons in more complicated heterostructures . if the driving function , @xmath284 , due to photoexcited carriers is localized , then the assumptions leading to eqs . ( [ sztsimple ] ) and ( [ snu ] ) need only hold in those regions where @xmath284 in nonvanishing . the wave equation applies in regions where the la sound speed , @xmath196 , is constant . heterostructure in which the la sound speed is piecewise constant have abrupt acoustic impedance mismatches which can be handled by introducing more complicated green s functions or by using other standard techniques.@xcite an example of such a problem would be a mqw structure embedded in a free standing substrate in which coherent la phonons generated in the mqw could bounce back and forth between two parallel substrate - air interfaces . in this section , we discuss simulations based on our microscopic theory of coherent la phonon generation in a @xmath1 diode structure with four periods of @xmath0 mqw s photoexcited by a gaussian pump normally incident along the ( 0001 ) @xmath16-direction . the parameters for our numerical example are listed in table [ simulationtable ] . the mqw dimensions and gaussian pump parameters were chosen to match those typically encountered in room temperature pump - probe differential transmission measurements of coherent la phonon oscillations carried out by c .- k . al . @xcite on @xmath309 mqw structures having fourteen periods . bulk wurtzite @xmath9 and @xmath88 are direct gap materials with band gaps of @xmath310 and 1.95 ev , respectively . the bulk bandstructure of unstrained wurtzite @xmath9 is shown in fig . [ bulkbands ] . as can be seen from equations ( [ hcmatrix ] ) and ( [ hulmatrix ] ) , the bandstructure is anisotropic and depends on @xmath311 , the wavevector along the ( 0001 ) @xmath16-axis , and @xmath312 , the wavevector within the @xmath23 plane perpendicular to the @xmath16-axis . the effective mass conduction band is two - fold degenerate and has a parabolic dispersion with anisotropic effective masses @xmath313 along the @xmath16-direction and @xmath314 in the @xmath23 plane . the two - fold degenerate valence bands are mixtures of heavy hole ( hh ) , light hole ( lh ) , and crystal - field splitoff hole ( ch ) character . at the zone center , the off - diagonal components of the @xmath33 upper and lower hamiltonians in eq . ( [ hulmatrix ] ) vanish and the valence bands can be labeled according to their pure state wavefunctions at @xmath315 . for the zone - center hh state , the degenerate wavefunctions are the basis states @xmath316 and @xmath317 defined in eq . ( [ vbasis ] ) . for the zone - center lh state , the wavefunctions are @xmath318 and @xmath319 , and for the ch band the zone - center wavefunctions are @xmath320 and @xmath321 . the heavy - hole effective masses along @xmath16 and @xmath23 are @xmath322 and @xmath323 for heavy holes , @xmath324 and @xmath325 for light holes , and @xmath326 and @xmath327 for crystal - field splitoff holes . bulk @xmath9 and @xmath88 have different lattice constants so when an ( 0001 ) @xmath291 mqw structure is grown , a significant lattice mismatch occurs between the @xmath328 wells and @xmath9 barriers . for the @xmath291 mqw diode specified in table [ simulationtable ] , we assume pseudomorphic strain conditions . in a pseudomorphically strained device , the lattice constant throughout the mqw adjusts to the value of the lattice constant in the bulk @xmath12 and @xmath11 substrates in order to minimize the overall strain energy . in our simulated diode , the substrates are @xmath14- and @xmath13- doped @xmath9 , so the lattice constant throughout the device takes on the @xmath9 value , i.e. @xmath329 . the non - vanishing position - dependent strain tensor components , @xmath27 , @xmath28 , and @xmath29 , for the mqw diode , as computed from eqs . ( [ epsilonxx ] ) and ( [ epsilonzz ] ) , are shown in fig . [ strainfigure ] as a function of @xmath16 . clearly , the @xmath9 barriers are unstrained since the @xmath12 and @xmath11 substrates are composed of @xmath9 and all the strain from the lattice mismatch is accomodated in the @xmath330 wells . the presence of strain in the mqw s results in the creation of a strain - induced polarization , @xmath331 , directed along @xmath16 as described by eq . ( [ pzeq ] ) . the strain - induced polarization , in turn , results in a strong bult - in piezoelectric field which can be computed from eqs . ( [ ezeq ] ) and ( [ p0eq ] ) , given the strain field and the dc bias , @xmath332 , applied across the diode . the computed strain - induced piezoelectric field , @xmath106 , and the piezoelectric confinement potential , @xmath333 , which result from the strain field in fig . [ strainfigure ] are shown in fig . [ piezofield ] . prior to the application of the pump pulse , we assume the applied dc bias , @xmath332 , has been adjusted so flat - band biasing in the diode is achieved , i.e. @xmath332 is such that the band edges seen in fig . [ piezofield ] are periodic functions of position . given the piezoelectric field and confinement potentials , position dependent band edges for the mqw can be computed . the conduction and valence band edges for our pseudomorphically strained mqw diode are shown as functions of position in fig . [ bandedgefigure ] . these are just the confinement potentials , @xmath334 , in the diode prior to photoexcitation . it is clear from fig . [ bandedgefigure ] that the confinement of elecrons and holes in the mqw is mostly due to strong built - in piezoelectric fields which result in the triangular confinement potentials seen in each well . in our numerical example , we simulate photoexcitation of electrons and holes and the generation and subsequent propagation of coherent la phonons in the hypothetical mqw diode when a gaussian pump laser pulse is normally incicent along the @xmath16-axis . as seen in table [ simulationtable ] , the gaussian pump pulse is assumed to be left circularly polarized with a photon energy of @xmath335 . the pump fluence is taken to be @xmath336 and the gaussian fwhm is taken to be @xmath337 . the expermient is assumed to take place at room temperature . in fig . [ subbandfigure ] , the computed conduction and valence subband energies are shown as functions of @xmath79 for the @xmath328 diode . at the chosen pump energy of @xmath338 , electrons from the first two valence subbands are excited into the lowest lying conduction subband . the computed densities of photoexcited electrons and holes , neglecting and including coulomb interaction effects , are shown as functions of position and time in figs . [ 3ddensity1 ] and [ 3ddensity2 ] , respectively and the total photoexcited electron density per unit area as a function of time is shown in fig . [ totaldensity ] . in fig . [ totaldensity ] the pulse shape is shown for comparison . we find that including coulomb effects decreases the total photogenerated carrier density . the electrons and holes screen the built - in piezoelectric field widening the effective band gap . this quantum confined stark effect acts to suppress the photogeneration of carriers . the driving function , @xmath284 , for the driven string equation ( [ waveequation ] ) is shown in fig . [ drivingfunction ] as a function of position and time . the driving function has units of acceleration and in fig . [ drivingfunction ] , we compute @xmath284 using the full microscopic formalism of eq . ( [ szt ] ) . for comparison , we also computed the driving function in the simplified loaded string model of eqs . ( [ sztsimple ] ) and ( [ snu ] ) using the carrier densities shown in fig . [ 3ddensity2 ] to facilitate the comparison . since the photoexcited holes are predominantly a mixture of heavy- and light- holes , we use @xmath302 in computing hole deformation potential contributions in eq . ( [ snu ] ) . the sum over species , @xmath294 , then yields the total driving function @xmath339 where @xmath340 and @xmath341 are the total conduction electron and valence hole densities plotted in fig . [ 3ddensity2 ] . the resulting @xmath284 is shown in fig . [ drivingfunction2 ] . by comparing figs . [ drivingfunction ] and [ drivingfunction2 ] , we see that for the diode structure and gaussian pump used in our simulation the simplified loaded string model produces essentially the same results as those obtained using the full microscopic formalism . acoustic la phonon generation due to the piezoelectric effect depends on the piezoelectric constant , @xmath299 , the number of photogenerated electrons and holes , as well as the spatial separation of electron and hole densities brought about by the strong built - in piezoelectric field in the mqw s . from eq . ( [ snu2 ] ) , the piezoelectric contribution to the driving function is given by @xmath342 in the absence of a built - in piezoelectric field ( such as found in a square well with infinite barriers ) we would have @xmath343 and hence @xmath344 even for relatively large values of @xmath299 . the built - in piezoelectric field serves to spatially separate the electrons and holes so that @xmath345 and hence @xmath346 . however , if the built - in piezoelectric field is too strong and the spatial separation of electrons and holes too large , then @xmath347 . this is because the overlap between the conduction and valence envelope functions enters into the optical dipole matrix elements in eq . ( [ dcv ] ) . if there s negligable overlap between electron and hole envelope functions due to strong piezoelectric fields then @xmath348 , no electron - hole pairs are photogenerated and once again @xmath344 . the deformation potential contribution to the driving function is given by @xmath349 from table [ parametertable ] , the conduction electron deformation potential , @xmath30 , is roughly twice the valence hole deformation potential , @xmath350 . thus , the two terms in eq . ( [ sdef ] ) are of comparable magnitude . the first term , due to conduction electrons , gives rise to a contribution to @xmath351 which is localized on the right side of each mqw while the second , due to valence holes , gives rise to contribution which is localized on the left hand side of each mqw . in our simulation , we find that piezoelectric and deformation potential contriburtions to the driving function are comparable . this is seen in fig . [ sztstring ] where @xmath352 and @xmath353 , along with their sum , are plotted at @xmath354 . in this example , we find that @xmath352 makes the dominant contribution to @xmath284 as can be seen in fig . [ sztstring ] . the macroscopic lattice displacement , @xmath220 , and velocity field , @xmath221 , can be obtained from the coherent phonon amplitudes , @xmath224 , using eqs . ( [ uzt ] ) and ( [ vzt ] ) . alternatively , they can also be obtained from the driving function , @xmath284 , through the green s function solution of the driven string equation . in fig . [ uvfigure ] , we plot the macroscopic lattice displacement , @xmath220 , and velocity field , @xmath221 , for coherent la phonon modes generated by the driving function shown in fig . [ drivingfunction ] . from the lattice displacement , @xmath220 , we can obtain the total energy density per unit volume associated with coherent la phonons , [ energydensityequations ] @xmath355 as the sum of a kinetic energy density term , @xmath356 and a potential energy density term , @xmath357 the coherent la phonon energy per unit area , @xmath358 , is obtained by integrating @xmath359 over position , @xmath16 : @xmath360 the total energy density , @xmath359 , for coherent la phonons as a function of position and time is shown in fig . [ energydensity ] and in fig . [ energydensitymovie ] the total energy density is plotted as a function of position for equally spaced values of the time ranging from @xmath361 to @xmath362 in increments of @xmath363 . the curves in fig . [ energydensitymovie ] for different times have been offset to avoid overlapping . at short times , the evolution of the total phonon energy density is complicated , but the long times behavior , @xmath364 , can be easily understood . as @xmath365 , a localized energy density appears in the mqw region due almost entirely to the potential energy term in eq . ( [ potentialenergydensity ] ) . this is due to near steady - state loading by the driving function at long times . assuming the driving function , @xmath284 , is approximately constant at long times , the loaded string equation , ( [ waveequation ] ) , can be integrated once in the steady state limit . we find the steady state solution @xmath366 from which the long time behavior of the la phonon energy density per unit volume in the mqw s , @xmath367 can be obtained . the fact that the energy density in eq . ( [ staticenergydensity ] ) is localized in the mqw s follows directly from the sum rule ( [ sumrule ] ) and is clearly seen in figs . [ energydensity ] and [ energydensitymovie ] . in addition to the localized energy density , which remains behind in the mqw s , two propagating wave trains consisting of four pulses each are seen to exit the mqw region and travel off to infinity at the acoustic phonon sound speed , @xmath196 . the distance between the pulses is just the inter - well separation distance . in these radiating wave trains , the kinetic and potential energy densities , @xmath368 and @xmath369 , are found to be equal as one would expect . the power spectrum of the coherent la phonon energy density in @xmath192-space can be written in terms of the coherent phonon amplitudes @xmath224 . the power spectrum for the total coherent la phonon energy density , [ densityequationsq ] @xmath370 is again the sum of a kinetic energy term , @xmath371 and a potential energy term , @xmath372 the phonon energy density per unit area is obtained by summing the power spectrum over positive phonon wavevectors , @xmath192 . thus , @xmath373 the total energy density power spectrum for coherent la phonons as a function of phonon wavevector , @xmath192 , and time is shown in fig . [ energypower ] . the peak near @xmath374 is associated with buildup of the steady state energy density localized in the mqw region . secondary peaks are seen near @xmath375 and twice this wavevector , i.e. @xmath376 . the wavevector @xmath377 corresponds to the the wavevector of the mqw period , @xmath378 where @xmath379 and @xmath380 are the well and barrier widths . the total coherent la phonon energy per unit area can be obtained from either eqs . ( [ elaz ] ) or ( [ elaq ] ) . in fig . [ totalenergy ] , we show the total phonon energy per unit area , @xmath381 , as a function of time for the coherent la phonons generated by the driving function shown in fig . [ drivingfunction ] . the total energy per unit area is the sum of kinetic- and potential- energy terms . for comparison , the pulse shape is shown as a dotted line . it is clear from the figure that the buildup of energy in coherent la phonons takes place on a time scale that is much longer than the pump duration . in addition , we see that the total energy buildup in the phonons saturates at around @xmath382 and that some strong but rapidly decreasing oscillations are superimposed on top of an increasing trend . the saturation phenomenon results from the fact that we have a finite number of quantum wells and not an infinite superlattice . the results can best be explained in terms of the driven string equation . in general , the rate at which energy is fed into the phonon field per unit area is described by the energy equation , @xcite @xmath383 in which @xmath284 and @xmath221 are the phonon driving function and velocity fields defined in eqs . ( [ szt ] ) and ( [ vzt ] ) , and @xmath195 is the @xmath9 mass density used for computing the sound speed , @xmath196 , in eq . ( [ phonondispersion ] ) . the energy equation simply says that the rate at which energy is added to a driven string is proportional to the local force times velocity integrated over the length of the string . the integral in eq . ( [ energyequation ] ) vanishes when the transient velocity field , @xmath221 , exits the mqw region in which the driving function , @xmath284 , is localized . thus , the time , @xmath384 , required for @xmath385 to saturate is just the time it takes for an la sound wave to cross the mqw , i.e. @xmath386 , where @xmath387 is the width of the mqw region over which the driving function is localized . in our example the la sound speed is @xmath388 in @xmath9 and the mqw width ( four well and three barrier layers ) is @xmath389 , from which we obtain @xmath390 . the oscillations of @xmath391 observed in fig . [ totalenergy ] reflect the number and periodicity of the diode mqw s . the pump laser generates spatially periodic electron and hole distributions , as seen in fig . [ 3ddensity1 ] , due to the fact that the pump photoexcites carriers in the wells but not the barriers . from each of the wells in the mqw , two sound pulses emerge traveling in opposite directions thus giving rise to an outwardly propagating velocity field pattern , @xmath221 , with four peaks traveling outward in each direction as seen in fig . [ uvfigure ] . the driving function , @xmath284 , on the other hand , is localized in the mqw s and is relatively constant in time after the pump pulse dies out . the driving functions localized in each well do work on four traveling velocity disturbances , the one generated in the well itself as well as the ones generated in the three neighboring wells that subsequently pass by . this gives rise to the three peaks plus saturation plateau seen in fig . [ totalenergy ] . the time interval between coresponding peaks in adjacent wells is just the time it takes la sound waves to travel between wells , i.e. @xmath392 , where @xmath379 and @xmath380 are the well and barrier thicknesses . for the mqw structure , @xmath393 and @xmath394 and we have @xmath395 which agrees with the peak - to - peak time between the first and second peaks seen in fig . [ totalenergy ] . we have developed a microscopic theory for the generation and propagation of coherent la phonons in pseudomorphically strained wurzite ( 0001 ) @xmath0 multi - quantum well ( mqw ) @xmath1 diodes . both @xmath9 and @xmath88 have different lattice constants so that a significant mismatch occurs between the wells and barriers . the presence of strain in the mqw s results in the creation of strain - induced built - in piezoelectric fields on the order of sevaral mv / cm which significantly alter the electronic and optical properties of the diode structure . in particular , the effective band gap can be lower than the band gap in unstrained @xmath291 wells due to the presence of triangular piezoelectric potentials . to a first approximation , the generation of coherent la phonons is driven by optical photoexcitation of electron - hole pairs by an ultrafast gaussian pump laser . under typical experimental conditions , the propagation of coherent la phonons is described by a _ uniform _ loaded string equation for the lattice displacement where the time- and position - dependent driving force on the string is a function of the photoexcited carrier density . this differs from coherent lo phonon oscillations in bulk systems where the coherent lo phonons obey a forced oscillator equation . both deformation potential and piezoelectric coupling mechanisms contribute to the driving force . we find that deformation potential coupling contributes a driving force proportional to the derivative of the carrier density while piezoelectric coupling contributes a driving force proportional to the photoexcited carrier density . we found that the driving term in the loaded string equation is suddenly turned on by rapid generation of electron - hole pairs by the pump and remains approximately constant theafter . this sudden displaceive loading of the string results in a new static equilibrium lattice displacement . this new static equilibrium displacement corresponds to a population of coherent la phonons with @xmath2 . as the lattice adjusts to the new equilibrium , coherent la phonons are transmitted in the positive and negative @xmath16-directions at the la sound speed . these traveling coherent la phonons are characterized by @xmath396 where @xmath5 is the superlattice period . the formalism described here can be applied to the analysis of more complicated device geometries as well as more complicated laser pulse sequences . this gives a simpler method for calculating the coherent la phonon generation in more complicated geometries and gives additional insight into the acoustic coherent response . the work of gds and cjs was supported by the national science foundation through grant no . dmr-9817828 and int-9414072 . the work of csk was supported by chonnam national university through a grant in the year 1999 . o. ambacher , j. smart , j. r. shealy , n. g. weimann , k. chu , m. murphy , w. j. schaff , l. f. eastman , r. dimitrov , l. wittmer , m. stutzmann , w. rieger , and j. hilsenbeck , j. appl . phys . * 85 * , 3222 ( 1999 ) . chichibu , a.c . abare , m.p . mack , m.s . minsky , t. deguchi , d. cohen , p. kozodoy , s.b . fleischer , s. keller , j.s . speck , j.e . bowers , e. hu , u.k . mishra , l.a . coldren , s.p . denbaars , k. wada , t. sota and s. nakamura , materials science and engineering , b59 , 298 ( 1999 ) . & & + @xmath108 ( ) & 3.540 & 3.189 + @xmath397 ( ) & 3.708 & 5.185 + @xmath398 & 0.377 & 0.376 + + & & + @xmath399 & 1.95 & 3.40 + + & & + @xmath400 & 0.10 & 0.18 + @xmath401 & 0.11 & 0.19 + + & & + @xmath402 & -9.28 & -7.24 + @xmath403 & -0.60 & -0.51 + @xmath404 & 8.68 & 6.73 + @xmath405 & -4.34 & -3.36 + @xmath406 & -4.32 & -3.35 + @xmath407 & -6.08 & -4.72 + + & & + @xmath60 & 17.0 & 22.0 + @xmath408 & 1.0 & 3.67 + @xmath409 & 1.0 & 3.67 + + & & + @xmath31 & & -4.08 + @xmath410 & & -4.08 + + & & + @xmath411 & & 0.7 + @xmath412 & & 2.1 + @xmath413 & & 1.4 + @xmath414 & & -0.7 + + & & + @xmath415 & -0.57 & -0.49 + @xmath299 & 0.97 & 0.73 + + & & + @xmath416 & 190 & 374 + @xmath417 & 104 & 106 + @xmath418 & 121 & 70 + @xmath197 & 182 & 379 + @xmath419 & 10 & 101 + + & & + @xmath420 & 15.3 & 8.9 & + left @xmath9 buffer width ( ) & 43.0 + + number of wells & 4 + well width ( ) & 63.0 + indium fraction in well & 0.06 + @xmath9 barrier width ( ) & 43.0 + + right @xmath9 buffer width ( ) & 43.0 + + & + @xmath332 ( @xmath202 ) & -0.261 + + & + @xmath421 ( @xmath422 ) & 300.0 + + & + photon energy ( ev ) & 3.21 + fluence ( @xmath423/@xmath424 ) & 160.0 + gaussian fwhm ( @xmath425 ) & 180.0 + polarization & left circular

we analyze the effect of screening provided by the additional graphene layer in double layer graphene heterostructures ( dlgs ) on transport characteristics of dlg devices in the metallic regime . the effect of gate - tunable charge density in the additional layer is two - fold : it provides screening of the long - range potential of charged defects in the system , and screens out coulomb interactions between charge carriers . we find that the efficiency of defect charge screening is strongly dependent on the concentration and location of defects within the dlg . in particular , only a moderate suppression of electron - hole puddles around the dirac point induced by the high concentration of remote impurities in the silicon oxide substrate could be achieved . a stronger effect is found on the elastic relaxation rate due to charged defects resulting in mobility strongly dependent on the electron denisty in the additional layer of dlg . we find that the quantum interference correction to the resistivity of graphene is also strongly affected by screening in dlg . in particular , the dephasing rate is strongly suppressed by the additional screening that suppresses the amplitude of electron - electron interaction and reduces the diffusion time that electrons spend in proximity of each other . the latter effect combined with screening of elastic relaxation rates results in a peculiar gate tunable weak - localization magnetoresistance and quantum correction to resistivity . we propose suitable experiments to test our theory and discuss the possible relevance of our results to existing data .

1208.3470

gate tunable anderson localization of dirac electrons is inaccessible in standard field effect heterostructures due to the strong disorder - induced inhomogeneity of the local doping level ( electron - hole puddles ) @xcite ubiquitous in the materials that show dirac - like low energy spectrum , such as graphene @xcite or topological insulator surfaces @xcite . basically , the low - density localization behavior of graphene is inaccessible experimentally because the dirac point itself is inaccessible due to the formation of electron - hole puddles around the charge neutrality point . however , very recently , a novel double layer graphene ( dlg ) heterostructure containing two graphene layers separated by an insulator allowed using additional screening effect of the second graphene layer to access a metal - insulator transition regime . anomalously large resistance , @xmath0 , in combination with insulating temperature dependence was observed in these experiments suggestive of a metal insulator transition ( mit ) @xcite . the physical nature of this behavior is currently a topic of a debate @xcite . in particular , whether the transport data of ref . represent an effective low - density high - temperature semi - classical resistivity @xcite or a low temperature strong anderson localization crossover behavior @xcite is unclear at this stage . on the one hand suppressing inhomogeneity allows access to the low - density regime @xmath1 in which anderson physics is expected to dominate ( here @xmath2 is the fermi wavevector and @xmath3 is the mean free path ) . in this regime , quantum interference of the two flavors ( due to two valleys ) of chiral dirac charge carriers in graphene may result in the insulating behavior in the case of sufficient mixing of the flavors by atomic - scale disorder @xcite . on the other hand , the experiments of ref . were restricted to a fairly high temperature regime ( 10 - 100k ) where quantum interference effects may be suppressed due to short phase breaking length . the observed mit is also in contrast with more recent transport measurements on ultra high quality suspended devices @xcite in which an extremely low density inhomogeneity is achieved , @xmath4 , nevertheless no mit is detected ( see also more recent experiment ref . ) . moreover , an anderson insulator would be characterized by the resistivity growing exponentially with decreasing temperature whereas the data @xcite demonstrates only a roughly power law growth . insulating power law temperature dependence ( resistivity growing with decreasing temperature ) is not unusual near charge neutrality in graphene ( see fig . 2 in ref . , and measurements on other low @xcite and high @xcite mobility samples ) . this behavior is explained by the combined effect of temperature dependent screening @xcite thermal excitation of electrons from the valence band and thermally activated hopping of electrons over the potential barriers between electron - hole puddles @xcite . nevertheless , very high values of resistance reported in ref . , @xmath0 suggest a novel behavior in this system possibly associated with mit . however , an alternative explanation for the observed behavior was recently suggested @xcite , in which case the anomalous resistance is explained by a boltzmann transport effect combined with strong suppression of inhomogeneity in dlg @xcite . the latter work is phenomenological and relies on the assumptions of the quasiclassical transport formalism and the phase breaking length being shorter than the elastic scattering length . therefore a more detailed analysis of quantum effects is required to justify the applicability of the latter approach . one way to make progress in the understanding of the observed behavior of the resistivity in these new devices is to analyze the metallic regime where perturbation expansion in disorder strength may be applied . such theoretical analysis of the quantum interference effect in the higher - density weak localization regime of the dlg system , which must be a precursor to any low - density strong localization crossover phenomenon , is one of the main goals of our work . in this paper we consider the effect of gate tunable screening provided by the additional layer in dlgs on transport characteristics of these devices . screening affects both the potential of charged defects and coulomb interactions of charge carriers . taking into account both effects we analyze the classical and quantum parts of the resistivity of graphene in dlg geometry in the metallic regime @xmath5 . we consider only low temperature regime such that elastic scattering by the disorder limits the transport characteristics and all inelastic scattering effects may be included perturbatively . below we outline the results of the following analysis : ( i ) we develop a framework describing the screening effect of the two graphene layers in dlg on the coulomb potential of charged defects in dlg heterostructure . ( ii ) we generalize the self - consistent theory of disorder induced electron density fluctuations at neutrality point of graphene @xcite ( electron - hole puddles ) and use it to estimate the screening effect of dlg on the amplitude of the electron density fluctuations . ( iii ) we analyze the effect of screening in dlg on the elastic scattering rate due to charged defects that determines the extent of gate tunability of mobility in dlg . ( iv ) we analyze the effect of the screening in dlg on the dephasing rate due to inelastic electron - electron collisions . ( v ) we analyze the combined effect of screening on the weak localization correction to the resistivity ( vi ) we predict a peculiar gate tunable magneto - resistance in the high quality dlg structures with mixed elastic scattering mechanisms . the present work is limited to the perturbative metallic regime and can not make any definitive conclusions about the regime of the low density studied in the experiments of ref . . however , the theoretical framework developed below provides a basis for analysis of a more detailed gate tunable quantum transport measurements which would allow characterization of various microscopic scattering mechanisms that determine the behavior of the system in the low density regime . in particular , we believe that it is imperative that we understand the high - density metallic regime of the dlg transport in some depth before trying to understand the low - density crossover mit regime since a systematic perturbative theory is available in the higher density regime whereas the low - density regime is inaccessible to analytical theory . we hope that our work would motivate experimental work focusing on the metallic regime of dlg in order to investigate conceptually important question of anderson localization and metal - insulator transition in graphene . we consider a dlg consisting of two layers of graphene , `` studied '' and `` control '' , see fig . [ fig0:sketch ] , which are used for resistivity measurements and to provide the additional screening effect , respectively . the two layers are separated by a thin layer , @xmath6 , of insulating hexagonal boron nitride ( hbn ) that is thick enough to suppress direct tunneling between the layers . the whole structure is separated by a thick layer , @xmath7 , of hbn from the standard sio@xmath8/si substrate . the combination of the top and bottom gates in this heterostructure allows independent `` control '' of the electron densities @xmath9 and @xmath10 in the `` control '' and `` studied '' layers respectively . this setup is motivated by the design of heterostructures in ref . . we assume that the mobility of the `` studied '' layer is limited by elastic scattering of charged impurities in accordance with the data @xcite . the additional screening effect of the `` control '' layer is expected to suppress the electron - hole puddles in the `` studied '' layer . we find that the effectiveness of the `` control '' layer screening strongly depends on the position of impurities in the dlg . it is efficient for the relatively small concentration of charged defects on top of the `` control '' layer , whereas it leads to only a moderate suppression of the inhomogeneity induced by charges in close proximity of the `` studied '' layer and by the remote charged defects in sio@xmath8 substrate . the latter inefficiency stems from the very high densities of charged defects in the sio@xmath8 substrate , @xmath11 , that would require screening charge density in the `` control '' layer several times @xmath12 which is beyond the experimentally accessible range . moreover , the nonlinear screening effects in the electron - hole puddle regime in the `` studied '' layer work against the screening effect of the `` control '' layer making the screening less efficient . the situation is slightly different in the case of elastic relaxation rate due to charged defects . the screening by `` control '' layer is more efficient in this case and leads to the suppression of the classical part of the resistivity near the dirac point of the `` studied '' layer . the resulting behavior of the resistivity as a function of the density in the `` control '' layer is a result of a competition of the suppression of electron - hole puddles that increases the resistivity and suppression of the elastic relaxation rate that decreases the resistivity . below we also demonstrate that the ratio of momentum relaxation rate to the transport rate varies substantially with the distance between graphene and charged defects . therefore measurement of the two elastic relaxation rates can be used to determine the location of charged defects limiting the mobility in dlg . transport scattering rate can be extracted from boltzmann resistivity . the momentum relaxation rate characterizes the broadening of single particle states and can be extracted from the decay of shubnikov de haas oscillations . a similar technique @xcite has been successfully applied to the analysis of the standard graphene based field effect transistors @xcite . since our work establishes that the inhomogeneous puddles induced by the distant charged impurities in sio@xmath8 can be suppressed only partially by the control layer screening , it is important to know where the dominant charged impurities reside and the measurement of the single - particle momentum relaxation ( sometimes also called the quantum scattering ) time could help locate the charged impurities in the dlg system . and @xmath13 considered in the text . , scaledwidth=25.0% ] very high values of the mobility of the `` studied '' layer encapsulated in hbn that is tunable in the range @xmath14 with the screening density in `` control '' layer @xmath9 indicate the dominant role of the charged defects in the elastic scattering in this system . furthermore , the very high value of the observed @xcite maximum of the gate tunable mobility suggests a weak effect of neutral ( immune to screening ) elastic scattering mechanisms that break locally the lattice symmetry of graphene . therefore these dlgs may demonstrate quantum interference behavior distinct from that observed in typical graphene flakes @xcite where strong lattice symmetry breaking disorder determines predominantly weak - localization behavior this is especially so at relatively low densities @xmath10 at which the effect of trigonal warping of the fermi line @xcite can be neglected . two identical carbon atoms in the unit cell of graphene give rise to two degenerate flavors ( valleys ) of chiral dirac charge carriers in its electronic spectrum confined to the vicinities of the two inequivalent corners of the hexagonal brillouin zone . quantum interference of two independent flavors of dirac quasiparticles is associated with weak anti - localization behavior @xcite . the latter is however protected only by the symmetry between the two sublattices in graphene , which can be easily broken by impurities . therefore quantum interference in graphene is very sensitive to the presence of elastic scattering mechanisms that break lattice symmetry . generic time inversion symmetric disorder in graphene can be categorized into three types according to their effect on quantum interference : ( i ) potential of the charged defects that does not break lattice symmetry ; ( ii ) smooth random vector potential disorder that breaks the symmetry between the two carbon atoms in the unit cell of graphene ( intravalley disorder ) ; ( iii ) atomic scale disorder that breaks lattice symmetry and causes mixing of the two valleys ( intervalley disorder ) . random vector potential disorder leads to suppression of the weak - antilocalization effect whereas the intervalley scattering restores the weak localization behavior typical for two dimensional electron gas . in the case of high quality dlg heterostructures with very weak lattice symmetry breaking scattering weak antilocalization effect of dirac electrons plays an important role . additional screening in dlg affects only the scattering rate due to charged defects in graphene and therefore changes the relative strength of intra and inter valley defects with respect to the strength of potential scatterers . this results in the dependence of quantum correction on the electron density in the `` control '' layer @xmath9 . in particular , an interesting possibility arises that the intervalley elastic scattering , while being unimportant in determining the semiclassical boltzmann resistivity , becomes important in determining the tuning of the quantum interference correction from anti - localization to localization . in a wide temperature range quantum interference is limited by decoherence rate due to inelastic electron - electron collisions . the latter is strongly affected by the screening effect in dlg for two reasons : due to suppression of the coulomb interaction between charge carriers , and due to enhancement of the diffusion coefficient that reduces the time electrons spend in proximity of each other where interaction is the strongest . the combination of these two effects results in strong variation of the decoherence rate with the screening density . this gives rise to a peculiar gate tunable quantum correction to the resistivity of graphene in dlg devices which may change sign depending on the screening density in the `` control '' layer especially at higher temperatures . it should be emphasized that this gate - tunable quantum correction to the boltzmann resistivity is manifest only when the phase breaking length is much larger than the elastic mean free path , i.e. at lower temperatures , so that the quantum interference is operational . since our theory treats the quantum correction perturbatively , it applies only in the metallic regime at densities well above the crossover to strong anderson localization . the most striking consequence of the gate tunable quantum correction can be observed in the intermediate density regime in which the transport in the `` studied '' layer is metallic , @xmath15 , whence the fermi energy is low enough so that the trigonal warping effect is negligible . the quantum correction in this case is strongly dependent on the electron density in the `` control '' layer . at sufficiently strong dephasing rate ( sufficiently high temperature ) this results in a gate tunable crossover from weak - antilocalizaiton type magnetoresistance to weak - localization type . a similar crossover that was temperature driven was demonstrated in graphene previously @xcite , our work demonstrates a similar tunability with the gate voltage in dlg systems . quantum part of the resistivity also includes altshuler - aronov correction to the conductivity due to the electron - electron interactions @xcite which typically has the sign enhancing the localizaiton effect and is expected to be suppressed by the additional screening effect which typically reduces the interaction strength . this does not have any effect on low field magnetoresistance which allows to study the quantum interference ( weak localization ) correction separately . the paper is organized as follows . in section ii we analyze the effect of `` control '' layer screening on coulomb potentials of the charged defects . sections iii and iv discuss classical and quantum parts of the resistivity respectively . we provide a discussion of the relevance of our work to experiments , particularly the data of ref . where the dlg system was studied experimentally in section v. we conclude in section vi . we start with a linear screening model for dlg heterostructure . for simplicity we assume that a density @xmath16 of charged defects is located in a plane a distance @xmath17 from the `` studied '' layer ( at @xmath18 ) , which can be both above @xmath19 and below @xmath20 it . the `` control '' layer is located at a distance @xmath21 above the `` studied '' layer . we include both intra and inter layer coulomb interactions between electrons in dlg described by a symmetric matrix @xmath22 with diagonal , @xmath23 , and off diagonal elements @xmath24 with @xmath25 being the 2d fourier transform of the coulomb potential . here @xmath26 correspond to the `` studied '' and `` control '' layers respectively . we calculate the interaction energy of an impurity charge @xmath27 and a conduction electron charge @xmath28 in the `` studied '' layer in the presence of the screening effect of electron gases in both `` studied '' and `` control '' layers . this can be found using linked cluster expansion of the thermodynamic potential @xcite of the dlg . in the resulting infinite series we keep only the term proportional to the product of the two charges that represents the linear screening of the interaction potential , @xmath29 where @xmath30 are 2d wave vector and position vector , respectively , and @xmath31 is the fourier transform of the bare coulomb interaction between the charges @xmath27 and @xmath28 . for electrons we can put @xmath32 . in eq . ( [ thermpot ] ) we introduced a dielectric function , @xmath33 where the matrix @xmath34 takes into account the dependence of coulomb interaction energy on the spatial separation between the charges in @xmath17 direction , @xmath35 . \end{aligned}\ ] ] here @xmath36 is the density - density correlator which is a matrix in the layer index , @xmath37 , including both intra and inter layer terms . the the density - density correlator is renormalized by the coulomb interaction matrix which is given by an infinite perturbative series , where we took the dielectric constant to be @xmath38 . ] . within the random phase approximation the result of the resummation of the perturbation series satisfies a matrix dyson equation @xcite , @xmath39 where @xmath40 is the diagonal matrix of free particle polarization operators in the two graphene layers . solving the dyson equation ( [ eq : dysmat ] ) we arrive at the dielectric function in the form , @xmath41 where @xmath42 . the dielectric function ( [ dielfunc ] ) demonstrates a strong dependence on both the impurity location @xmath17 and the interlayer distance @xmath21 , and in the limit @xmath43 approaches the monolayer form @xmath44 . static screening is described by the zero frequency limit of the free particle polarization operator @xcite , @xmath45 where @xmath46 stands for heaviside step function , @xmath47 stands for spin and valley degeneracy combined , @xmath48 and @xmath49 stand for the fermi wave vector and the density of states per spin per valley at the fermi level in the @xmath50-th layer , respectively . [ sec : e - h ] we estimate the effect of screening by the `` control '' layer on the density inhomogeneity in the `` studied '' layer using the simple self - consistent approach that had shown good agreement with more detailed thomas - fermi - dirac numerical calculations in the case of monolayer graphene @xcite . this estimate is based on the approximate relation @xmath51 , where the amplitude of the fluctuation of the local fermi energy is equated to the local fluctuation in the potential energy of electrons , which reflects the local electrochemical equilibrium condition . we further assume that the local fermi level is given by the free particle form @xmath52 , where @xmath53 and @xmath10 stands for the fermi velocity of dirac electrons and local density respectively . we also assume that the screening of impurity potential can be taken into account within linear approximation ( [ dielfunc ] ) . we then end up with a self - consistent equation for the typical density fluctuation @xmath54 . the solution of this self - consistent problem is plotted in fig . [ fig1:rmsns ] as a function of the `` control '' layer density @xmath9 . the discrepancy between the solid black line in fig . [ fig1:rmsns](a ) as @xmath55 and the root mean square density in monolayer without the screening layer under same conditions , dashed line , originates from the screening by interband transitions in the `` control '' layer at neutrality . the above analysis neglects non - linear screening effects in the `` control '' layer and therefore the saturation in the realistic dlg device may be at a slightly different value . nevertheless fig . [ fig1:rmsns ] and the comparison to the monolayer case gives a reasonable estimate of the suppression of the density inhomogeneity by screening . for a dlg with @xmath56 , see section [ sec : e - h ] for details . ( a ) solid line corresponds to @xmath57 , @xmath58 . the horizontal dashed line corresponds to the result of the self - consistent density calculation for isolated monolayer graphene encapsulated in hbn with the same parameters as the solid line . ( b ) solid and dashed lines correspond to @xmath59 and @xmath60 respectively with the impurity density @xmath61 . , title="fig:",scaledwidth=23.5% ] for a dlg with @xmath56 , see section [ sec : e - h ] for details . ( a ) solid line corresponds to @xmath57 , @xmath58 . the horizontal dashed line corresponds to the result of the self - consistent density calculation for isolated monolayer graphene encapsulated in hbn with the same parameters as the solid line . ( b ) solid and dashed lines correspond to @xmath59 and @xmath60 respectively with the impurity density @xmath61 . , title="fig:",scaledwidth=23.5% ] effectiveness of screening by the `` control '' layer is strongly dependent on the location of impurities ( @xmath17 ) which can accumulate in three possible areas : relatively small densities of charged defects , @xmath62 , can be located at hbn - graphene interfaces @xcite and on top of the `` control '' layer which can be exposed to air in realistic dlgs . a much larger concentration of charged impurities , @xmath63 , is expected to accumulate at the surface of sio@xmath8 substrate that is separated from the `` studied '' layer by @xmath64 nm of hbn , see fig . [ fig0:sketch ] . these defects do not play an important role in transport in the high density regime @xmath65 due to the exponential factors in eqs . ( [ thermpot ] ) and ( [ dielfunc ] ) . however , they dominate transport and density inhomogeneity at very low densities where @xmath66 and hence the long - range nature of the coulomb disorder becomes important at low densities near the dirac point . the additional screening effect in dlg is the strongest for impurities located on top of the `` control '' layer for which the inhomogeneity varies by an order of magnitude , dashed line in fig . [ fig1:rmsns](b ) . however , the screening is substantially less efficient for the impurities located in close vicinity of the `` studied '' layer and for the high concentration of impurities in sio@xmath8 substrate , see the solid line in fig . [ fig1:rmsns](b ) . note that our estimate of the density inhomogeneity relies on the assumption that locally electron density resembles quasiparticles in clean graphene . this is unjustified in the case of strong disorder for which @xmath67 , see sections [ sec : transrate ] and [ sec : momentumrate ] . additional screening in dlg leads to a substantial suppression of the transport scattering rate which in born approximation reads , @xmath68 where @xmath69 and @xmath70 are the scattering angle and momentum of electrons on the fermi surface in the `` studied '' layer , and @xmath71 is the matrix element of the coulomb potential of a defect in graphene . [ fig3:taup ] shows the variation of the transport mean free path approximated by eq . ( [ eq : tau ] ) within experimentally accessible densities in the `` control '' layer @xmath72 . , with @xmath57 , @xmath73 . dashed line corresponds to isolated monolayer graphene with the same parameters . for comparison the effect of a relatively small concentration of charged defects , @xmath61 , near `` studied '' layer , @xmath74 is shown as dash - dotted line . the dotted line corresponds to @xmath75 where quantum and classical parts of the conductivity are of the same order . , scaledwidth=40.0% ] different spatial configurations of impurities within dlg may be distinguished experimentally by analyzing the ratio of the momentum relaxation rate to the transport scattering rate . these ratios are substantially different in the cases of scattering on long range potential of remote and nearby impurities . the transport rate , eq . ( [ eq : tau ] ) , is mostly determined by backscattering with large momentum transfer , and therefore decays quickly with the distance @xmath17 between the charged defects and the `` studied '' layer . the momentum relaxation rate , given by , @xmath76 includes forward scattering with small momentum transfer which is only weakly dependent on @xmath17 , see fig . [ fig3:rateratio](a ) . the ratio of the momentum relaxation rate to the transport rate is shown in fig . [ fig3:rateratio](b ) for the three different location of impurities considered in sec . [ sec : locimp ] . note that born approximation used in eq . ( [ eq : taup ] ) is insufficient when @xmath77 , @xmath78 shown as dotted line in fig . [ eq : taup](a ) and further corrections to the elastic relaxation rate have to be included . in the case of remote impurities in sio@xmath8 there is a peculiar regime where @xmath77 and the born approximation is insufficient yet @xmath79 which suggests diffusive transport . note however that the perturbative analysis presented here is quantitatively correct only when @xmath80 . . solid lines from bottom to top ( red , blue , green and magenta ) correspond to @xmath81 respectively . dotted line corresponds to @xmath82 where born approximation becomes insufficient . ( b ) ratio of the transport and momentum relaxation times . solid lines from bottom to top ( blue , green and red ) correspond to ( @xmath83 ) , ( @xmath84 ) and ( @xmath85 ) respectively with @xmath86 . dashed lines correspond to the same parameters with @xmath87 . , title="fig:",scaledwidth=25.0% ] . solid lines from bottom to top ( red , blue , green and magenta ) correspond to @xmath81 respectively . dotted line corresponds to @xmath82 where born approximation becomes insufficient . ( b ) ratio of the transport and momentum relaxation times . solid lines from bottom to top ( blue , green and red ) correspond to ( @xmath83 ) , ( @xmath84 ) and ( @xmath85 ) respectively with @xmath86 . dashed lines correspond to the same parameters with @xmath87 . , title="fig:",scaledwidth=25.0% ] suppression of the relaxation rate is reflected in the boltzmann resistivity @xmath88 shown in fig . [ fig4:resistivity](a ) . it is the strongest near the neutrality point in the `` studied '' layer in which case both , @xmath89 and @xmath90 so that the high concentration of defects in sio@xmath8 substrate dominates the transport properties . in the formal limit @xmath91 , ( @xmath92 is the thomas - fermi screening wavevector in the `` control '' layer ) the charged impurity potential becomes effectively short ranged which leads to the resistivity independent of the density @xmath10 , @xmath93 where @xmath94 . this strong - screening limiting resistivity , however , is unlikely to be reached , since within the metallic regime and accessible densities @xmath95 is of the order of unity . moreover , the value of the resistivity at low densities @xmath10 is affected by the doping inhomogeneity . the resistivity maximum in the presence of density inhomogeneity ( electron - hole puddles ) can be estimated @xcite using the effective medium theory @xcite , which allows averaging of the resistivity of a system split into regions with random values of resistance characterized by a distribution @xmath96 $ ] . the averaged resistance , @xmath97 , in this case is given by the solution of a self - consistent effective medium equation , @xmath98\frac{\rho(n)-\rho_{emt}}{\rho(n)+\rho_{emt}}=0 . \end{gathered}\ ] ] we approximate the distribution of the density fluctuations by a gaussian with zero average and variance given by @xmath99 , which compares well with full thomas - fermi - dirac approach @xcite in the case of disordered monolayer graphene . the results are shown in fig . [ fig4:resistivity](a ) as dashed lines . note however that due to the relatively weak doping inhomogeneity induced by the remote impurities in sio@xmath8 the value of the metallicity parameters @xmath100 and @xmath77 indicate that quantum corrections could be important in this low - density regime . however , the long - range character of the disorder potential makes quantitative analysis of quantum interference difficult @xcite in this regime . at higher density @xmath10 the random inhomogeneity can be neglected and transport is metallic , therefore the linear screening model and the boltzmann resistivity provide an accurate description of the system . at even higher densities @xmath101 the exponential factors in eqs . ( [ dielfunc ] , [ eq : tau ] ) lead to the strong suppression of the effect of remote impurities in sio@xmath8 . as a result charged defects in the vicinity of graphene layer and neutral scatterers determine the resistivity . finally , at the very high density @xmath102 the typical monolayer behavior of the resistivity is restored , see fig . [ fig4:resistivity](a ) . . ( a ) solid lines from top to bottom ( red , green , blue and magenta ) correspond to @xmath103 respectively , with @xmath104 , @xmath57 , @xmath56 . dashed lines correspond to inhomogeneity induced maximum resistivity with the same parameters from top to bottom . ( b ) resistivity at intermediate density @xmath105 . top ( red ) line corresponds to @xmath104 , @xmath57 , bottom ( black ) and middle ( blue ) lines correspond to @xmath106 and @xmath60 , @xmath107 respectively.,title="fig:",scaledwidth=22.5% ] . ( a ) solid lines from top to bottom ( red , green , blue and magenta ) correspond to @xmath103 respectively , with @xmath104 , @xmath57 , @xmath56 . dashed lines correspond to inhomogeneity induced maximum resistivity with the same parameters from top to bottom . ( b ) resistivity at intermediate density @xmath105 . top ( red ) line corresponds to @xmath104 , @xmath57 , bottom ( black ) and middle ( blue ) lines correspond to @xmath106 and @xmath60 , @xmath107 respectively.,title="fig:",scaledwidth=22.5% ] there exists a regime of densities @xmath10 where eqs . ( 4,5 ) provide a quantitatively accurate description of the dependence of the elastic relaxation rate on both densities @xmath10 and @xmath9 . formally , the constraints defining this regime are : ( i ) the effect of the density inhomogeneity on the resistivity is negligible , @xmath108 , where @xmath109 stands for spatial average of @xmath110 ; ( ii ) transport in the `` studied '' layer is metallic @xmath111 . in addition we assume ( iii ) that the density in the `` studied '' layer is sufficiently small , @xmath112 , so that the resistivity changes substantially with changing @xmath9 . here @xmath9 is limited by the experimentally accessible density range @xmath113 ; and ( iv ) the trigonal warping of the fermi line has a negligible effect on weak localization corrections @xcite , @xmath114 , with @xmath115 and @xmath116is the distance between the nearest - neighbors in the hexagonal lattice of graphene . the latter constraint simplifies the analysis of characteristics of various scattering mechanisms that determine the quantum transport in the system . ( v ) scattering due to coulomb defects remains dominant @xmath117 . throughout the text we refer to the range of densities @xmath10 where all constraints ( i)-(v ) are satisfied as the `` intermediate density regime '' . the intermediate density regime can be used for a quantitative analysis of both boltzmann and weak localization contributions to the resistivity . the effect of screening electron density @xmath9 in the `` control '' layer on the resistivity is illustrated in fig . [ fig4:resistivity](b ) . the effect of @xmath9 on the weak localization correction in the dlg in the intermediate density regime is discussed below . we generalize the perturbative theory of weak localization in graphene @xcite to the dlg heterostructure in the intermediate density regime defined in sec . [ sec : intermeddensity ] above . we briefly review the theory of weak localization correction to the resistivity of graphene . weak localization originates from the interference contribution to the probability to scatter backwards . the latter is given by the modulus square of the sum of the quantum mechanical amplitudes @xmath118 associated with each possible backscattered trajectory , @xmath119 . this contains the classical part , @xmath120 , and the quantum interference contribution , @xmath121 . the latter vanishes for generic trajectories . however , in case @xmath122 is a time reversal image of @xmath118 , which is possible for loop - shaped self - intersecting trajectories , the phase factors cancel exactly , @xmath123 . therefore the average contribution of this special type of trajectories to the backscattering probability is non - zero . @xmath124 for a given self - intersecting trajectory is proportional to the ratio of the width of the trajectory @xmath125 to the 2d volume @xmath126 that can be spanned by diffusion during time @xmath127 . the resulting correction to the conductivity @xmath128 is proportional to the log of the ratio of the travel times along the longest @xmath129 and the shortest @xmath130 trajectories and is typically of the order of the conductance quantum . the longest trajectories @xmath129 that limit the extent of quantum interference are given by either the system size or by phase breaking effects . two degenerate flavors ( two valleys ) of chiral dirac fermions in graphene are characterized by an isospin degree of freedom that has a fixed projection on the direction of momentum . therefore backscattering of chiral charge carriers is accompanied by the reversal of isospin which gives rise to an additional phase difference @xmath131 between the two images of a looped backscattered trajectory , where @xmath132 is the winding number of the trajectory . the interference contribution to conductivity in this case acquires an additional minus sign which results in anti - localization @xcite effect in contrast to the usual weak localization in 2d @xcite . in realistic devices the vector potential ( intravalley ) disorder , characterized by a phenomenological rate @xmath133 , breaks the symmetry between the two sublattices of graphene and therefore suppresses the anti - localization effect . atomic scale disorder , characterized by @xmath134 , scatters electrons between different valleys and therefore restores the localization sign of the interference correction . quantitatively this effect is described by a perturbation expansion in @xmath5 which results in the conductivity correction @xcite , @xmath135.\label{eq : wl}\ ] ] here the first term on the left hand side describes intervalley interference sensitive only to the ratio @xmath136 , the second term describes the intravalley interference which is also sensitive to smooth intravalley scattering rate @xmath137 due to the vector potential disorder and the total transport scattering rate @xmath138 . the left hand side in the latter equation , given by eq . ( [ eq : tau ] ) , is obtained modeling charged defects by a gaussian disorder with a finite range in space and zero average of the potential . in fig . [ fig7:wlns](a ) we show the lines in the parameter space that separate the negative @xmath139 , localization - like correction from the positive one @xmath140 . in the presence of strong vector potential disorder , @xmath141 the intravalley interference term is strongly suppressed . this results in the weak localization sign of the correction even in the case of relatively weak intervalley scattering , see the blue line in fig [ fig7:wlns](a ) . by contrast , in the presence of relatively weak vector potential disorder @xmath142 , a substantially stronger intervalley scattering is required to compensate for the intravalley term in ( [ eq : wl ] ) and for the correction to have the localization sign . this effect is expected to be small in devices used in ref . since surface roughness of hbn substrate is an order of magnitude smaller than of sio@xmath8 . ] . we emphasize that these analytical considerations are entirely restricted to the perturbative regime , @xmath143 of the quantum interference correction to the conductivity and do not apply to the strong localization regime of metal - insulator transitions . microscopic origin of the intervalley scattering at intermediate densities therefore deserves a more careful consideration . electrostatic potential can give rise to a strong intervalley scattering only if its amplitude is larger than the bandwidth in which case the lowest order born approximation is insufficient , see appendix and ref . for details . such strong potentials can be induced by lattice defects and adsorbate molecules chemically bound to carbon atoms of the graphene lattice . scattering characteristics of such defects depend strongly on the detailed nature of the electronic structure , type of the chemical bonding and the position of the defect in the unit cell of graphene @xcite . such details are beyond the scope of the present analysis . instead we employ a more generic phenomenological model of two types of scatterers namely resonant scatterers and short range disorder . the former describes the effect of monovalent chemical groups and vacancies in the graphene lattice @xcite . the scattering rate in this case decreases with density @xmath144 which in monolayer graphene is practically indistinguishable from that due to charged defects @xmath145 , see ref . . short range scatterer model describes atomic defects that do not give rise to resonant scattering . the corresponding scattering amplitude grows with density @xmath146 and the resulting contribution to the conductivity is density independent . in the case of both sources of intervalley scattering a relatively weak @xmath147 is sufficient for the quantum correction to have the weak localization sign , fig [ fig7:wlns](a ) . there exists substantial experimental evidence for the presence of intervalley scattering in graphene . in particular , the observation of negative weak localization magnetoresistance suggests a rather high intervalley scattering rate corresponding to @xmath148 at high densities @xmath149 , @xcite . alternative measurements of a weak d - peak in the raman spectrum demonstrated the presence of intervalley scattering at high energies @xmath150ev @xcite . more detailed quantitative characteristics of the intervalley scattering such as density dependence are not currently available experimentally . the size of the self - intersecting trajectories that give rise to the interference correction is limited by either the system size or the phase breaking effect . the latter can be caused by scattering on magnetic defects @xcite or inelastic electron - phonon or electron - electron collisions @xcite . both latter effects result in temperature dependent dephasing rate @xmath151 , which is linear @xmath152 in the case of electron - phonon scattering and linear or quadratic in the case of electron - electron interactions in the diffusive and ballistic regime respectively . experiments report typical values for the dephasing length @xmath153 at the lowest temperatures @xmath154 @xcite and @xmath155 at higher temperatures @xmath156 . more recent measurements on large area epitaxial graphene devices accessed very high temperature regime @xmath157 corresponding to a short dephasing length @xmath158 @xcite . the available experimental information about the density , disorder , and temperature dependent phase breaking length in graphene is , however , not extensive , and more work is needed before definite quantitative statements can be made in details . respectively . red line corresponds to monovalent adsorbate disorder . the dashed lines show example trajectories that a dlg initially at point a takes with changing @xmath9 in case of fixed @xmath129 , straight line , and coherence limited by the system size , the curved trajectory @xmath159 . dash dotted line corresponds to the trajectory with coherence limited by electron - electron interactions in the diffusive regime eq . ( [ dephasing ] ) . ( b ) weak localization correction for different values @xmath160 from bottom to top with @xmath161 . we assume @xmath58 at @xmath57 and dephasing is limited by the electron - electron interactions in the diffusive regime @xmath162 , eq ( [ dephasing ] ) . we took @xmath163 at @xmath164.,title="fig:",scaledwidth=23.0% ] respectively . red line corresponds to monovalent adsorbate disorder . the dashed lines show example trajectories that a dlg initially at point a takes with changing @xmath9 in case of fixed @xmath129 , straight line , and coherence limited by the system size , the curved trajectory @xmath159 . dash dotted line corresponds to the trajectory with coherence limited by electron - electron interactions in the diffusive regime eq . ( [ dephasing ] ) . ( b ) weak localization correction for different values @xmath160 from bottom to top with @xmath161 . we assume @xmath58 at @xmath57 and dephasing is limited by the electron - electron interactions in the diffusive regime @xmath162 , eq ( [ dephasing ] ) . we took @xmath163 at @xmath164.,title="fig:",scaledwidth=23.0% ] in this section we estimate the effect of additional screening on the dephasing rate caused by inelastic electron - electron collisions with low energy transfer @xmath165 . we estimate the phase breaking rate up to a numerical coefficient using the self - consistent approach @xcite which is sufficient to analyze the dependence of the phase breaking rate on the densities in the two layers of dlg . there are two distinct dephasing temperature regimes , diffusive , @xmath166 , and ballistic , @xmath167 which correspond to distinct temperature dependences of @xmath151 . in the diffusive regime , @xmath168 , electrons undergo diffusive motion which increases the time they spend in vicinity of each other ( where interaction is the strongest ) and therefore enhances the inelastic scattering rate . electron - electron collision rate is given by fermi golden rule , @xmath169 where @xmath170 and @xmath171 . the interaction kernel in ( [ tauin ] ) is given by , @xmath172 the matrix element of the screened interaction is written in terms of exact eigenstates @xmath173 of non - interacting dirac hamiltonian in the presence of a given disorder potential . the screened electron - electron interaction potential in dlg is given by @xmath174 which includes the dynamical dielectric function of eq . ( [ dielfunc ] ) where we need to put @xmath18 for interaction within the `` studied '' layer . the expression for the collision rate ( [ tauin ] ) has to be averaged over disorder realizations . the result of this averaging can be then expressed in terms of the diffusion propagator in graphene @xmath175 which describes the transport of the charge density through the disordered system . the diffusion propagator is found by solving the diffusion equation @xmath176 in the `` studied '' layer @xcite . here @xmath177 is the diffusion coefficient in the `` studied '' ( @xmath178 ) or `` control '' ( @xmath179 ) layer respectively . as a result the eq . ( [ tauin ] ) at @xmath180 and @xmath181 takes the form , @xmath182 ^ 2\label{diffphi}\end{gathered}\ ] ] where we keep only the logarithmically divergent terms . also we introduce a low energy cutoff @xmath183 which reflects the fact that only collisions with large enough energy transfer @xmath184 produce a substantial dephasing of the electrons with diffusion time limited by @xmath129 @xcite . solving ( [ diffphi ] ) self - consistently gives an estimate for the dephasing rate . at high densities such that @xmath185 , @xmath186 , the screening is ballistic for which case we can approximate @xmath187 and eq . ( [ diffphi ] ) gives , @xmath188 with @xmath189 where we introduce the screening parameter @xmath190 for @xmath50-th layer and @xmath191 . at lower densities the polarization operator acquires a diffusion pole , @xmath192 the presence of the diffusion pole affects only the coefficient in eq . ( [ dephasing ] ) . in the case of short screening length @xmath193 we obtain for the coefficient @xmath194 , @xmath195 in the limit of @xmath196 the results ( [ diffusivephi ] ) and ( [ cball ] ) both approach , @xmath197 the dephasing rate ( [ dephasing ] ) is strongly dependent on the screening density @xmath9 in the `` control '' layer through the diffusion coefficient @xmath198 and the factor @xmath199 or @xmath200 . in the high temperature @xmath201 ballistic regime the polarization operator is approximated by the free particle form @xcite , @xmath202 at @xmath203 . the dephasing rate in this case is given by , @xmath204 where parameters @xmath190 and @xmath191 have the same meaning as above and we assume @xmath205 . in the presence of mixed scattering sources the suppression of the transport scattering rate with growing @xmath9 results in the increased relative strength of neutral scattering mechanisms . this is a simple consequence of the coulomb scattering being effectively screened out by the `` control '' layer . as a result the amplitude of the quantum interference correction varies as a function of @xmath9 . to illustrate this we assume a large concentration of charged defects @xmath12 located in the sio@xmath8 substrate and a small concentration of resonant scatterers @xmath206 . then , we have @xmath207 where we use @xmath208 valid for a small concentration of single site defects . in the present analysis we ignore the effect of screening by the `` control '' layer on the shift of the chemical potential in graphene due to charged defects . depending on the mechanism limiting the phase coherence a variety of behaviors of the conductivity correction as a function of the screening density @xmath9 is possible , see fig . [ fig7:wlns](a ) . for screening independent @xmath151 due to dephasing by magnetic impurities the system initially at point `` a '' in fig . [ fig7:wlns](a ) takes a trajectory represented by the straight dashed line . in smaller devices coherence is more likely to be limited by the system size , in which case the effective rate @xmath209 , depends on the diffusion coefficient and hence is sensitive to the screening effect of the `` control '' layer . this regime corresponds to the curved dashed trajectory in fig . [ fig7:wlns](a ) . the effect of screening by the `` control '' layer is the strongest in the case of dephasing limited by electron - electron interactions . in this case the ratio @xmath210 depends on @xmath9 through the coefficient @xmath199 in eq . ( [ dephasing ] ) , and the system takes the trajectory shown as the dash - dotted line in fig . [ fig7:wlns](a ) . this is the situation most likely to be realized in experiments on large exfoliated flakes of graphene . dependence of the weak localization correction on @xmath9 for this case is given by eq . ( [ eq : wl ] ) combined with ( [ eq : tau ] ) and ( [ diffusivephi ] ) , which is shown in fig . [ fig7:wlns](b ) . the effect of additional screening by `` control '' layer therefore expands the parameter range in which weak - localization sign of the quantum correction could be observed , and thus the possibility of a transition from the weak anti - localization to weak localization as a function of the `` control '' layer density ( i.e. gate - tunable ) in the nominally metallic studied layer is very high in this case . this predicted transition has not yet been seen in experiments , but may actually exist in the samples of ref . in the metallic regime of the studied layer . weak field magnetoresistance allows an unambiguous measurement of the weak - localization correction . at small fields @xmath211 the magnetoresistance can be approximated by , @xmath212.\nonumber \end{gathered}\ ] ] in this regime we expect a gate tunable crossover from weak anti - localization to weak localization magnetoresistance to be observable especially at higher temperatures . in this section we discuss the relation of the theory presented above to the recent transport measurements in dlg @xcite . the clever dlg design of ref . provides an opportunity to use the additional gate tunability of the resistivity in the `` studied '' layer , @xmath213 to distinguish the contributions of charged @xmath214 and neutral @xmath215 scattering mechanisms in the intermediate density regime , see sec . [ sec : intermeddensity ] . the latter can be deduced from the dependence of the mobility of the `` studied '' layer , @xmath216 , on the `` control '' layer density reported in the supplementary material of ref . . on the one hand these data are consistent with the presence of a high concentration of charged defects @xmath217 at the surface of sio@xmath8 substrate located at @xmath57 away from the `` studied '' layer @xcite . on the other hand a similar behavior could result from a smaller concentration of charged defects @xmath218 in close vicinity of the `` studied '' layer or on top of the `` control '' layer , see fig . [ fig4:resistivity](b ) for comparison . the two situations may be distinguished experimentally by comparing the transport and momentum relaxation rates , see fig . [ fig3:rateratio](b ) . at low carrier densities however the precise location of impurities does not play an important role and therefore the substantially larger concentration of defects likely to be present in sio@xmath8 dominates the transport properties . the situation is different in the case of density inhomogeneity . we find that dlg is relatively inefficient in screening out electron - hole puddles induced by the high concentration of defects in sio@xmath8 as well as defects in close proximity to the `` studied '' graphene layer itself . both allow some `` control '' in the experiments , the former can be reduced by changing the substrate and the latter may be reduced by improving the sample preparation . moreover , both improvements were recently implemented in suspended devices @xcite . in the absence of more systematic data we can estimate the residual mobility associated with neutral scattering mechanisms from the maximum of the mobility curve , @xmath216 at the maximum density in `` control '' layer reported in the supplementary material of ref . . the result is @xmath219 , which corresponds to a relatively small density of resonant scatterers , such that @xmath220 suggesting that the dlgs analyzed in ref . may demonstrate gate tunable weak anti - localization to localization crossover at intermediate densities . finally , we discuss the effect of coupling between graphene and hbn substrate which gives rise to spectacular new features in the electron spectrum at high densities @xcite . however , at low carrier densities this effect is not expected to be dramatic . transport characteristics of graphene could be affected in three possible ways : ( i ) triangular symmetry of hbn breaks the sublattice symmetry of graphene and therefore gives rise to an additional trigonal warping of the fermi surface @xcite ; our estimate shows negligible effect on the weak localization correction at low densities @xmath221 ; ( ii ) hbn induced moire pattern may dominate the density fluctuation at the neutrality point . this effect however has not been clearly identified in the local probe data @xcite to this point and therefore we expect it to be smaller than current levels of inhomogeneity induced by charged defects . ( iii ) additional intervalley scattering may be caused by strong coupling of graphene to hbn . however , existing experimental data and _ ab initio _ calculations @xcite suggest a rather weak coupling which is not expected to give rise to strong intervalley scattering . nevertheless , a more detailed analysis is required to completely rule out all of these possibilities . our goal in this work has been a careful theoretical study of the screening by the `` control '' layer in determining transport properties of the studied layer in double - layer graphene systems experimentally introduced in ref . . the problem of interest to us is both subtle and complex , involving many disparate aspects of electronic transport phenomena , and our study is at best an approximate one because of the highly complex nature of the problem . in particular , we are interested in studying both semiclassical boltzmann and quantum localization contributions to the dlg resistivity , including the `` control '' layer screening effect , on an equal footing as much as possible . since potential , intravalley and intervalley elastic scattering affect quantum localization properties qualitatively differently , our work must include all mechanisms in the presence of `` control '' layer screening with equivalent considerations for all scattering processes contributing to the semiclassical and the quantum part of graphene resistivity . since the phase breaking length arising from inelastic scattering processes is an important ingredient in determining localization effects arising from quantum interference , we consider electron - electron interaction induced inelastic phase decoherence effects in our theory , neglecting electron - phonon interaction which is known to be weak in graphene @xcite . we also neglect all direct contributions to transport from inelastic scattering processes since our interest is relatively low - temperature transport where quantum interference may play a role rather than high - temperature transport where inelastic phonon scattering plays a role . we investigate the role of the `` control '' layer screening on the density inhomogeneity ( i.e. electron - hole puddles ) near the dirac point finding that screening by the `` control '' layer is ineffective in suppressing the electron - hole puddles arising from charged impurities in the sio@xmath8 substrate far from the `` studied '' graphene layer ( in fact , this is an important qualitative new result of our work ) . finally , our work treats both long - range coulomb disorder arising from random charged impurities in the environment and the short - range disorder arising from neutral atomic defects on an equal footing . since the theory involves ( at least ) six different independent characteristic length scales ( elastic mean free path due to long - range and short - range disorder and intra- valley and inter - valley scattering length , the inelastic phase breaking length , the fermi wavelength , the density fluctuation correlation length , the thermal length , etc . ) , it is necessarily a complex problem necessitating various approximations focusing on different aspects of the transport phenomena . we concentrate on low - temperature transport at relatively high `` metallic '' densities where the quantum interference contribution to the resistivity can be treated as a perturbative weak localization correction to the semiclassical boltzmann resistivity . this precludes us from commenting directly on the very low density nonperturbative insulating behavior ( as observed in ref . ) from the perspective of strong anderson localization phenomenon , but our theory establishes the clear possibility of a gate - tunable transition from weak anti - localization to weak localization behavior in graphene at metallic densities , which would be a necessary precursor to a possible strong localization crossover behavior at low densities . whether such a weak localization transition , which must precede any strong localization induced low density metal - insulator transition , is operational in ref . was unfortunately not studied there , and the prediction of such a gate - tunable transition in the weak localization behavior is an important new prediction of our work . we emphasize that such transition must be studied at relatively low temperatures , and it may be missed at the higher temperature range ( 10 - 100k ) of study used in ref . . in addition , the strong localization induced insulating behavior at low carrier density should be exponential in temperature , and not a power law as observed in ref . . given our concrete theoretical predictions , we hope that future experiments on dlg systems will resolve the important question of a possible low - density quantum localization transition in clean graphene . our work shows that it is possible for the intervalley scattering in graphene to be strong enough to induce localization , but at the same time be weak enough not to strongly suppress high - density mobility , particularly in the presence of screening by the `` control '' layer . in conclusion , we analyze the effect of gate tunable screening provided by the additional `` control '' layer in dlg on both classical and quantum parts of the resistivity of a high quality `` studied '' layer in which elastic scattering is limited by charged disorder . we find that ( i ) the additional screening in dlg is relatively inefficient at screening out electron - hole puddles induced by the high concentration of defects in sio@xmath8 as well as defects in close proximity to the `` studied '' graphene layer itself . in the latter cases , the screening provides roughly a factor of 2 suppression within the `` control '' layer density range of @xmath222 . therefore the suppression of electron - hole puddles itself is insufficient to explain the contrast between the strongly insulating behavior observed in dlg @xcite structures and the absence of such in stand - alone graphene samples including highly homogeneous suspended samples @xcite . ( ii ) the screening effect of `` control '' layer results in a relatively strong suppression of the elastic relaxation rate resulting in gate tunable mobility of the `` studied '' layer in the dlg at intermediate density , @xmath10 . ( iii ) we show that the location of the charged defects limiting the mobility of the dlg can be determined by measuring the ratio of the momentum relaxation and transport relaxation rates . ( iv ) additional screening effect of the `` control '' layer strongly suppresses the potential of coulomb interaction between charge carriers . ( v ) combined effect of ( ii ) and ( iv ) results in a strong suppression of the dephasing rate due to inelastic electron - electron collisions which determines the sign and the magnitude of the quantum interference correction to the resistivity . ( vi ) as a consequence of ( ii ) and ( v ) weak localization correction to the resistivity is gate tunable . ( vii ) the very low concentration of resonant or short range scatterers that has little effect on the boltzmann part of the resistivity in high quality `` studied '' layer in dlg nevertheless can provide sufficient intervalley scattering for the quantum correction to resistivity to have a negative ( weak localization ) sign . moreover , the additional screening in dlg improves the coherence in the system and therefore expands the range in which quantum correction of weak localization sign is observable . ( viii ) we find that high quality dlg structures where charged defects are the dominant source of elastic scattering manifest a peculiar gate - tunable crossover between weak anti - localization and weak localization magneto - resistance . the latter could be observed in the intermediate density regime in which electron density in the `` studied '' layer is sufficiently high for random inhomogeneity to be neglected and the transport to be metallic and at the same time is low enough for the effect of the `` control '' layer screening to be substantial , see also sec . [ sec : intermeddensity ] . ) to intervalley scattering time ( @xmath223 ) , @xmath224 , or equivalently , the ratio of the scattering rates , @xmath225 , where @xmath226 for the long range coulomb potential . ( b ) and ( c ) show @xmath227 and @xmath223 , respectively , as a function of density for an impurity density @xmath61 . note that @xmath228 and @xmath229 . ] in this paper we developed a theoretical framework that can be used for a systematic experimental characterization of the symmetry breaking mechanisms in dlg in the intermediate density regime which could allow further understanding of the observed strongly insulating behavior at low densities . this work could also be used to describe a more generic double layer heterostructure involving materials other than graphene that realize other universality classes of dirac electrons such as surfaces of strong and weak topological insulators . such heterostructures could be useful to study experimentally gate tunable quantum interference and anderson localization effects in these systems which are a topic of active current theoretical interest @xcite . this work is supported by us - onr - muri and nri - swan . intervalley scattering rate due to coulomb impurities can be separated from the intravalley part by taking low and high momentum parts of the sum over @xmath230 in eq . ( [ eq : tau ] ) . where @xmath231 is the reciprocal lattice vector . for intravalley scattering we include only @xmath232 which results in , @xmath233 ^ 2 \sqrt{1-(q/2k_f)^2}\ ] ] for intervalley scattering we take the smallest allowed reciprocal vector that connects the two corner of hexagonal brillouin zone , @xmath234 , where @xmath235is the c - c distance . then the intervalley scattering time is given by @xmath236 ^ 2 \sin \theta,\ ] ] where @xmath237 and @xmath238 is given by the following relation @xmath239 in fig . [ fig : ratio_c ] we show the calculated ratio of intravalley scattering time to intervalley scattering time , @xmath240 , as a function of density for the long range coulomb impurities . since the ratio is very small the intervalley scattering plays a little role in the transport when the coulomb disorder dominates . at the same time a phenomenological model of delta - range disorder gives @xmath241 at all densities . 92ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1073/pnas.0704772104 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.83.407 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1021/nl2041673 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/revmodphys.82.3045 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.83.1057 [ * * , ( ) ] link:\doibase 10.1103/physrevb.85.235413 [ * * , ( ) ] link:\doibase 10.1038/nphys2114 [ * * , ( ) ] link:\doibase 10.1103/physrevb.85.195451 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.97.236801 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.97.236802 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.98.256801 [ * * , ( ) ] http://dx.doi.org/10.1140/epjst/e2007-00226-4 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.98.076602 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.99.106801 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.99.146806 [ * * , ( ) ] link:\doibase 10.1103/physrevb.76.214204 [ * * , ( ) ] link:\doibase 10.1103/physrevb.77.081410 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.100.036803 [ * * , ( ) ] link:\doibase 10.1103/physrevb.84.235420 [ * * , ( ) ] @noop ( ) , @noop ( ) , http://dx.doi.org/10.1140/epjst/e2007-00221-9 [ * * , ( ) ] link:\doibase 10.1103/physrevb.84.035421 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.101.096802 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevb.84.115442 [ * * , ( ) ] link:\doibase 10.1103/physrevb.77.195412 [ * * , ( ) ] link:\doibase 10.1103/physrevb.80.241415 [ * * , ( ) ] http://stacks.iop.org/0953-8984/22/i=20/a=205301 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.100.056802 [ * * , ( ) ] link:\doibase 10.1103/physrevb.78.125409 [ * * , ( ) ] http://stacks.iop.org/0957-4484/21/i=27/a=274014 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.103.226801 [ * * , ( ) ] link:\doibase 10.1103/physrevb.83.195417 [ * * , ( ) ] http://stacks.iop.org/1367-2630/13/i=11/a=113005 [ * * , ( ) ] link:\doibase 10.1063/1.3624611 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.98.136801 [ * * , ( ) ] link:\doibase 10.1016/j.ssc.2010.08.020 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.97.146805 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.89.266603 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.97.036802 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.97.196804 [ * * , ( ) ] @noop _ _ ( , , ) @noop _ _ ( , , ) @noop _ _ , ed . ( , , ) link:\doibase 10.1103/physrevb.23.805 [ * * , ( ) ] link:\doibase 10.1103/physrevb.34.979 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevb.75.205418 [ * * , ( ) ] link:\doibase 10.1103/physrevb.76.245405 [ * * , ( ) ] link:\doibase 10.1016/j.ssc.2009.02.041 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.101.166803 [ * * , ( ) ] link:\doibase 10.1103/physrevb.79.245423 [ * * , ( ) ] link:\doibase 10.1103/physrevb.83.121405 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.99.176801 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop ( ) , link:\doibase 10.1103/physrevb.74.235443 [ * * , ( ) ] link:\doibase 10.1140/epjst / e2007 - 00224 - 6 [ * * , ( ) ] link:\doibase 10.1016/0370 - 1573(84)90103 - 0 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.102.106401 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.105.056802 [ * * , ( ) ] link:\doibase 10.1103/physrevb.78.115432 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.101.196803 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.104.076802 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.105.266803 [ * * , ( ) ] @noop ( ) , link:\doibase 10.1021/nl101399r [ * * , ( ) ] link:\doibase 10.1103/physrevlett.107.166602 [ * * , ( ) ] link:\doibase 10.1103/physrevb.82.075424 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.108.106601 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevb.65.180202 [ * * , ( ) ] link:\doibase 10.1103/physrevb.83.155441 [ * * , ( ) ] link:\doibase 10.1103/physrevb.77.193403 [ * * , ( ) ] http://stacks.iop.org/0022-3719/15/i=36/a=018 [ * * , ( ) ] @noop * * , ( ) @noop ( ) , link:\doibase 10.1103/physrevb.84.195414 [ * * , ( ) ] link:\doibase 10.1103/physrevb.77.115449 [ * * , ( ) ] link:\doibase 10.1103/physrevb.85.235115 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.108.076804 [ * * , ( ) ] link:\doibase 10.1103/physrevb.86.045102 [ * * , ( ) ]

a _-box _ is the cartesian product of intervals of and a _ -box representation _ of a graph is a representation of as the intersection graph of a set of-boxes in . it was proved by thomassen in 1986 that every planar graph has a 3-box representation . in this paper we prove that every graph embedded in a fixed orientable surface , without short non - contractible cycles , has a 5-box representation . this directly implies that there is a function , such that in every graph of genus , a set of at most vertices can be removed so that the resulting graph has a 5-box representation . we show that such a function can be made linear in . finally , we prove that for any proper minor - closed class , there is a constant such that every graph of without cycles of length less than has a 3-box representation , which is best possible .

1512.02381

for @xmath9 , a _ @xmath0-box _ is the cartesian product @xmath10 of @xmath0 intervals of @xmath1 . @xmath0-box representation _ of a graph @xmath11 is a collection @xmath12 of @xmath0-boxes in @xmath3 such that any two boxes @xmath13 and @xmath14 intersect if and only if the corresponding vertices @xmath15 and @xmath16 are adjacent in @xmath2 . in other words , @xmath2 is the intersection graph of the boxes @xmath17 . the _ boxicity _ of a graph @xmath2 , denoted by @xmath18 and introduced by roberts in 1969 @xcite , is the smallest integer @xmath0 such that @xmath2 has a @xmath0-box representation . it was proved by thomassen in 1986 that planar graphs have boxicity at most 3 @xcite , which is best possible ( as shown by the planar graph obtained from a complete graph on 6 vertices by removing a perfect matching @xcite ) . it is natural to investigate how this result on planar graphs extends to graphs embeddable on surfaces of higher genus . let @xmath19 be the supremum of the boxicity of all graphs embeddable in a surface of euler genus @xmath5 . the result of thomassen on planar graphs was extended in @xcite by showing that for any @xmath20 , @xmath21 ( prior to this result , it was not known whether @xmath19 was finite ) . in @xcite , we proved the existence of two constants @xmath22 , such that @xmath23 . the proof of the upper bound relies on a connection between boxicity and acyclic coloring established in @xcite . it was noted there that using this connection and a result of kawarabayashi and mohar @xcite , it could be proved that there is a function @xmath4 such that any graph embedded in a surface of euler genus @xmath5 , such that all non - contractible cycles have length at least @xmath6 , has boxicity at most 42 . the main result of this paper is to reduce this bound to 5 for orientable surfaces ( theorem [ thm : locpla ] ) . similar ideas are then used to show that toroidal graphs have boxicity at most 6 , and toroidal graphs without non - contractible triangles have boxicity at most 5 ( theorem [ thm : tor ] ) . this improves on @xcite , where it was proved that toroidal graphs have boxicity at most 7 , while there are toroidal graphs of boxicity 4 . an immediate consequence of theorem [ thm : locpla ] is that if @xmath2 has genus @xmath5 , then a set of at most @xmath24 vertices can be removed in @xmath2 so that the resulting graph has boxicity at most 5 . however , the bound we obtain for @xmath6 in theorem [ thm : locpla ] is exponential in @xmath5 . we show the following improvement : if @xmath2 is embedded in a surface of euler genus @xmath25 , then a set of at most @xmath26 vertices can be removed in @xmath2 so that the resulting graph has boxicity at most 5 ( theorem [ thm : linex1 ] ) . note that this result is proved for any surface , orientable or not . in @xcite , it was proved that there is a function @xmath27 such that if @xmath2 is embedded in a surface of euler genus @xmath5 , and has no cycle of length less than @xmath28 , then @xmath2 has boxicity at most 4 . here , we show there is a function @xmath27 such that if @xmath2 has no @xmath29-minor and no cycle of length less than @xmath30 , then @xmath2 has boxicity at most 3 ( corollary [ cor : minor ] ) . this is best possible already for @xmath31 . this result follows from a more general theorem on path - degenerate graphs ( theorem [ thm:6path ] ) . this general result also implies , together with earlier results from @xcite , that there is a constant @xmath27 such that any graph embeddable in a surface of euler genus @xmath5 , with no cycle of length less than @xmath32 , has boxicity at most 3 ( corollary [ cor : girth ] ) . this is best possible up to the choice of the constant @xmath27 . some of the results we will use originate from the proof of thomassen @xcite that planar graphs have boxicity at most 3 , without being explicitly stated there . in section [ sec : pla ] , we explain how these results can be derived from @xcite , which might be of independent interest . we will then prove the main results of this paper in sections [ sec : locpla ] , [ sec : linext ] , and [ sec : girth ] . in the remainder of this section , we review the necessary background on boxicity and graphs on surfaces . we then give a simple proof that embedded graphs of large edge - width have boxicity at most 7 ( this will be improved to 5 in section [ sec : locpla ] ) . let @xmath11 be a graph , and let @xmath33 be a @xmath0-box representation of @xmath2 . for a @xmath0-box @xmath34 , and an integer @xmath35 , we refer to @xmath36 as _ the @xmath37-th interval of @xmath16_. for @xmath35 , let @xmath38 be the interval representation consisting of all @xmath37-th intervals of the vertices of @xmath2 in @xmath39 . each interval representation @xmath38 , @xmath35 , corresponds to an interval graph @xmath40 with vertex - set @xmath41 . observe that each graph @xmath40 is a supergraph of @xmath2 , and any two @xmath0-boxes @xmath13 and @xmath14 intersect if and only if for all @xmath42 the intervals corresponding to @xmath15 and @xmath16 intersect in @xmath38 , or equivalently , if the vertices @xmath15 and @xmath16 are adjacent in the interval graph @xmath40 . for two graphs @xmath43 and @xmath44 on the same vertex - set @xmath41 , the _ intersection @xmath45 _ of @xmath46 and @xmath47 is defined as the graph @xmath48 . the discussion above implies that the boxicity of a graph @xmath2 can be equivalently defined as the least @xmath49 such that @xmath2 can be expressed as the intersection of @xmath49 interval graphs on the same vertex - set . in this paper , the two definitions will be used and we will often switch from one to the other , depending of the situation , i.e whenever we consider some @xmath0-box representation @xmath33 we will implicitly consider it as a representation @xmath50 as defined in the previous paragraph , and vice - versa . if @xmath39 is a representation of @xmath2 , we will often say that @xmath39 _ represents _ @xmath2 , or _ induces _ @xmath2 . we refer the reader to the book by mohar and thomassen @xcite for more details or any notion not defined here . all the graphs in this paper are simple ( i.e. , without loops and multiple edges ) . a _ surface _ is a non - null compact connected 2-manifold without boundary . a surface can be orientable or non - orientable . the _ orientable surface @xmath51 of genus @xmath52 _ is obtained by adding @xmath53 _ handles _ to the sphere ; while the _ non - orientable surface @xmath54 of genus @xmath49 _ is formed by adding @xmath55 _ cross - caps _ to the sphere . the _ euler genus _ of a surface @xmath56 is defined as twice its genus if @xmath56 is orientable , and as its non - orientable genus otherwise . we say that an embedding is _ cellular _ if every face is homeomorphic to an open disk of @xmath57 . using the fact that the boxicity of a graph @xmath2 is at most @xmath49 if and only if all the connected components of @xmath2 have boxicity at most @xmath49 , we will always be able to assume in this paper ( using ( * ? ? ? * propositions 3.4.1 and 3.4.2 ) ) that the considered embeddings are cellular ( and we will do so implicitly ) . let @xmath2 be a graph embedded in a surface of euler genus @xmath25 . the _ edge - width _ of @xmath2 is defined as the length of a smallest non - contractible cycle of @xmath2 , and the _ face - width _ of @xmath2 is defined as the minimum number of points of @xmath2 intersected by a non - contractible curve on the surface . given a 2-sided cycle @xmath58 in @xmath2 , and an orientation of @xmath58 , the set of neighbors of @xmath58 incident to an edge leaving the left side of @xmath58 is denoted by @xmath59 , and the set of neighbors of @xmath58 incident to an edge leaving the right side of @xmath58 is denoted by @xmath60 ( see figure [ fig : lrc ] , left , for an example in a triangulation ) . note that @xmath59 and @xmath60 might not be disjoint . given two non - contractible cycles @xmath61 and @xmath62 , @xmath63 denotes the minimum distance between a vertex of @xmath61 and a vertex of @xmath62 in @xmath2 . if @xmath58 is a non - contractible 2-sided cycle , we also define @xmath64 as the minimum length ( number of edges ) of a path starting with an edge incident to @xmath58 and @xmath59 , and ending with an edge incident to @xmath58 and @xmath60 ( note that @xmath64 does not depend of the chosen orientation of @xmath58 ) . and @xmath60 in a triangulation ( left ) and the embedded graph resulting from the removal of @xmath58 ( right).[fig : lrc ] ] let @xmath2 be a triangulation of @xmath65 . a collection of pairwise disjoint non - contractible cycles @xmath66 of @xmath2 is _ planarizing _ if the graph obtained from @xmath2 by removing the vertices of @xmath66 is planar . if for any @xmath67 , @xmath68 , then the planarizing collection @xmath66 is said to have _ minimum distance _ at least @xmath0 . the following was proved by thomassen @xcite : @xcite[thm : placc ] let @xmath69 be integers , and let @xmath2 be a triangulation of @xmath65 of edge - width at least @xmath70 . then @xmath2 contains a planarizing collection of induced cycles with mininum distance at least @xmath0 . to give a taste of the proofs of the main results is this paper , we now prove that _ if @xmath11 is a triangulation of @xmath65 , @xmath71 , with edge - width at least @xmath72 , then @xmath2 has boxicity at most 7 . _ let @xmath66 be a planarizing collection of induced cycles with mininum distance at least 4 ( whose existence follows from theorem [ thm : placc ] ) in @xmath2 . we denote by @xmath58 the set of vertices lying in one of the @xmath73 s . we also denote by @xmath74 the set of vertices not in @xmath58 with at least one neighbor in @xmath58 , and set @xmath75 . since the collection @xmath66 is planarizing , the set @xmath76 induces a planar graph and therefore has a 3-box representation @xmath77 by the result of thomassen @xcite . we extend this representation to @xmath58 by mapping all the vertices of @xmath58 to a big 3-box containing all the other 3-boxes of the representation . since the distance in @xmath2 between any two cycles @xmath78 is at least 4 , there is no edge between a neighbor of @xmath73 and a neighbor of @xmath79 . as a consequence , the set @xmath80 induces a disjoint union of @xmath5 planar graphs ( and is therefore planar ) . let @xmath81 be a 3-box representation of this graph . we extend this representation to @xmath82 by mapping all the vertices of @xmath82 to a big 3-box containing all the other 3-boxes of the representation @xmath81 . finally , let @xmath83 be the interval graph mapping all the vertices of @xmath58 to @xmath84 , all the vertices of @xmath74 to @xmath85 $ ] , and all the vertices of @xmath82 to @xmath86 . we now prove that @xmath87 induces @xmath2 . each interval graph @xmath38 is clearly a supergraph of @xmath2 , so we only need to show that every pair of non - adjacent vertices @xmath88 in @xmath2 is non - adjacent in at least one of the @xmath38 s . this follows from the definition of @xmath77 if @xmath89 , and from the definition of @xmath81 if @xmath90 . finally if @xmath91 and @xmath92 then they are non - adjacent in @xmath83 , as desired.@xmath93 in section [ sec : locpla ] we will show that this argument can be applied to any embedded graph ( not only triangulations ) and that the bound on the boxicity can be improved to 5 . the idea will be to group @xmath94 , @xmath95 , and @xmath83 in a single interval graph . but for this we first need to see how much flexibility we have in choosing 3-box representations for planar graphs , which is the topic of the next section . recall that thomassen @xcite proved that planar graphs have boxicity at most 3 . he actually proved significantly stronger results , which we will need here . a @xmath0-box is _ non - degenerate _ if it is the cartesian product of @xmath0 intervals of positive length . a representation of a graph @xmath2 as the intersection of @xmath0-boxes is _ strict _ if ( 1 ) the boxes are non - degenerate , ( 2 ) the interiors of the boxes are pairwise disjoint , and ( 3 ) the intersection of any two boxes is a non - degenerate @xmath96-box . @xcite[thm : plasep ] a graph @xmath2 has a strict 2-box representation if and only if @xmath2 is a proper subgraph of a 4-connected planar triangulation . let @xmath2 be a graph embedded in some surface . a triangle of @xmath2 is said to be _ facial _ if it bounds a face of @xmath2 . we will use the following immediate corollary of theorem [ thm : plasep ] : [ cor : plasep ] let @xmath2 be a plane graph such that all the triangles of @xmath2 are facial . if @xmath2 is not a triangulation , then @xmath2 has a strict 2-box representation . let @xmath97 be the graph obtained from @xmath2 by doing the following , for every non - triangular face @xmath4 of @xmath2 : add a cycle @xmath98 of length @xmath99 ( the number of edges in a boundary walk of @xmath4 ) inside @xmath4 , join one vertex of @xmath98 to each of the other vertices of @xmath98 ( with the newly added edges staying inside the region bounded by @xmath98 ) , and then join each vertex of @xmath98 to two consecutive vertices in a boundary walk of @xmath4 . this can be done in such a way that @xmath97 is a ( simple ) triangulation . since all the triangles of @xmath2 are facial , @xmath97 is a triangulation without separating triangle . observe that @xmath97 contains at least 5 vertices , so @xmath97 is a 4-connected triangulation . the result now follows directly from theorem [ thm : plasep ] . consider a strict 3-box representation @xmath100 of a planar graph @xmath11 , and let @xmath101 be a triangle of @xmath2 . then @xmath101 has an _ empty inner corner _ in the representation if there is a small 3-box @xmath58 , whose interior is disjoint from the interior of all the boxes @xmath100 , such that the following holds : there is a corner @xmath27 of @xmath58 , and three faces @xmath102 of @xmath58 containing @xmath27 , such that @xmath103 , @xmath104 , and @xmath105 ( see figure [ fig : corner ] ) . in particular @xmath106 . note that if a triangle has an empty inner corner , then the corresponding 3-box @xmath58 defined above can be made arbitrarily small . ] we say that three intervals @xmath107 are _ strictly overlapping _ if @xmath108 is an interval of positive length , and none of the three intervals is contained in another one of the three . the following simple lemma about empty inner corners will be particularly useful . [ lem : corner ] let @xmath109 be a strict 3-box representation of a planar graph @xmath2 such that @xmath110 is a strict 2-box representation of @xmath2 . then any triangle @xmath101 of @xmath2 such that the intervals of @xmath111 are strictly overlapping in @xmath94 has an empty inner corner in @xmath109 . in a strict 2-box representation ( left ) , and an empty inner corner ( depicted by a white dot ) in the corresponding strict 3-box representation if the intervals of the third dimension are strictly overlapping ( right ) . for the sake of readability , we write @xmath112 instead of @xmath113 . [ fig : over ] ] let @xmath101 be a triangle of @xmath2 . without loss of generality , in @xmath114 , @xmath111 are mapped to rectangles @xmath115 such that the point @xmath116 is a corner of @xmath117 and @xmath118 , but lies on a side of @xmath119 ( see figure [ fig : over ] , left ) . moreover , no rectangle of @xmath114 other than @xmath115 contains @xmath120 . let @xmath121 be the intervals corresponding to @xmath111 in @xmath94 . since @xmath121 are strictly overlapping , @xmath122 is an interval of positive length , and none of the three intervals is contained in another one . let @xmath123=i_v \cap i_w$ ] . observe that the interior of @xmath124 intersects at least one of @xmath125 ( say @xmath126 by symmetry ) , otherwise @xmath124 would not intersect @xmath127 or @xmath128 ( in an interval of positive length ) , or would be contained in @xmath127 or @xmath128 . it is then easy to check that the triangle @xmath101 has an empty inner corner in @xmath129 , touching the point with coordinate @xmath130 ( see figure [ fig : over ] , right ) . the following result was proved by thomassen ( see theorem 4.5 in @xcite ) , using a strong variant of theorem [ thm : plasep ] . @xcite[thm : tho45 ] every planar triangulation @xmath2 has a strict 3-box representation such that every facial triangle of @xmath2 , except one prescribed triangle , has an empty inner corner . the proof of thomassen @xcite indeed shows a slightly stronger result , which will be extremely useful in this paper . @xcite[thm : tho45v2 ] let @xmath2 be a planar triangulation @xmath2 , with outerface @xmath101 , and let @xmath131 be any strict 3-box representation of @xmath111 such that @xmath101 has an empty inner corner @xmath58 . then @xmath131 extends to a strict 3-box representation @xmath39 of @xmath2 such that every facial triangle of @xmath2 , except possibly @xmath101 , has an empty inner corner in @xmath39 . moreover , all 3-boxes except that of @xmath111 are included in @xmath58 . let @xmath2 be a graph embedded in the plane , or in @xmath65 with @xmath25 , and such that all the triangles are contractible . then each triangle bounds a region homeomorphic to an open disk . this region is called the _ interior _ of the triangle . let @xmath97 be the graph obtained from @xmath2 by removing ( the vertices in ) the interior of each triangle ( distinct from the outerface of @xmath2 , in case @xmath2 is embedded in the plane ) . the graph @xmath97 is called the _ frame _ of @xmath2 . note that @xmath97 is embedded in the plane or @xmath65 with @xmath25 , with the property that all its triangles are facial . moreover , if @xmath2 is embedded in the plane , the outerfaces of @xmath2 and @xmath97 coincide . a triangle in a graph @xmath2 embedded in the plane is said to be _ internal _ if it is distinct from the outerface of @xmath2 . the following direct consequence of theorem [ thm : tho45v2 ] will be repeatedly used in the next section ( in conjunction with corollary [ cor : plasep ] and lemma [ lem : corner ] ) . [ cor : plaframe ] let @xmath2 be a planar graph and let @xmath97 be its frame . then any strict 3-box representation of @xmath97 in which each internal triangle has a fixed empty inner corner can be extended to a strict 3-box representation of @xmath2 . moreover , if @xmath101 is an internal triangle of @xmath97 and @xmath132 is a vertex of @xmath2 inside the region bounded by @xmath101 in @xmath2 , then @xmath132 is mapped to a 3-box lying in the empty inner corner of @xmath101 . we are now ready to prove the main result of this paper . [ thm : locpla ] let @xmath2 be a graph embedded in @xmath65 , @xmath71 , with edge - width at least @xmath72 . then @xmath2 has boxicity at most 5 . it is well - known that if @xmath2 has edge - width at least @xmath49 , then @xmath2 is an induced subgraph of a triangulation of @xmath65 of edge - width at least @xmath49 ( see for instance lemma 3.4 in @xcite ) . since having boxicity at most @xmath49 is a property that is closed under taking induced subgraphs , in the following we can assume that @xmath2 is a triangulation of @xmath65 . since @xmath2 has edge - width at least @xmath133 , all the triangles of @xmath2 are contractible . let @xmath97 be the frame of @xmath2 . observe that @xmath97 is a triangulation of @xmath65 of edge - width at least @xmath72 . by theorem [ thm : placc ] , @xmath97 has a collection of planarizing cycles @xmath66 with minimum distance at least 4 . consider the following 5 sets of vertices of @xmath2 , forming a partition of @xmath134 : 1 . : the union of all cycles @xmath66 , 2 . : the union of all neighbors in @xmath97 of a vertex of @xmath58 , 3 . : the vertices of @xmath97 not in @xmath80 , 4 . : the vertices of @xmath2 lying inside a non - facial triangle that does not intersect @xmath58 , 5 . : the vertices of @xmath2 lying inside a non - facial triangle intersecting @xmath58 . observe that since all the triangles of @xmath2 are contractible , all the triangles of @xmath97 are facial ( and moreover , the triangles of @xmath97 are precisely the faces of @xmath97 , since @xmath97 is a triangulation ) . let @xmath135 be the planar graph obtained from @xmath97 by deleting the vertices of @xmath58 ( i.e. @xmath135 is the subgraph of @xmath2 induced by @xmath136 ) . all the triangles of @xmath135 are facial , and we claim that @xmath135 is not a triangulation . to see this , observe that the deletion of each @xmath73 produces two faces with vertex - set @xmath137 and @xmath138 , respectively ( see figure [ fig : lrc ] , right ) . the two cycles bounding these faces in @xmath135 correspond to two cycles of @xmath97 that are ( freely ) homotopic to @xmath73 , and therefore non - contractible in @xmath65 . since @xmath97 has edge - width at least 4 , the boundary walks of the two faces of @xmath135 resulting from the deletion of @xmath73 have length at least 4 , and @xmath135 is not a triangulation . by corollary [ cor : plasep ] , @xmath135 has a strict 2-box representation @xmath110 . we extend this 2-dimensional representation of @xmath135 to @xmath139 by mapping all the vertices of @xmath140 to a large 2-box containing all the 2-boxes of @xmath141 . the images of the vertices of @xmath142 in @xmath141 will be defined later in the proof . let @xmath143 be the subgraph of @xmath97 induced by @xmath80 . since the collection @xmath66 has minimum distance at least 4 , @xmath143 is the disjoint union of @xmath5 graphs , each being a subgraph of @xmath97 induced by @xmath73 together with its neighborhood @xmath144 , for @xmath145 . as each of these graphs is embedded in a cylinder , they are all planar and so is @xmath143 . as before , we can argue that all the triangles of @xmath143 are facial , and @xmath143 is not a triangulation . it follows from corollary [ cor : plasep ] that @xmath143 has a strict 2-box representation @xmath146 . we extend this 2-dimensional representation of @xmath143 to @xmath147 by mapping all the vertices of @xmath147 to a large 2-box containing all the 2-boxes of @xmath146 . again , the images of the vertices of @xmath148 in @xmath149 will be defined later . for each vertex @xmath16 of @xmath97 , we choose a real @xmath150 , in such way that all the chosen @xmath151 are distinct . let @xmath152 be the following interval graph : each vertex @xmath16 of @xmath58 is mapped to @xmath153 $ ] , each vertex @xmath16 of @xmath74 is mapped to @xmath154 $ ] , and each vertex @xmath16 of @xmath82 is mapped to @xmath155 $ ] . observe that in the corresponding interval graph , @xmath58 , @xmath74 and @xmath82 are cliques , @xmath74 is complete to @xmath58 and @xmath82 , while there are no edges between @xmath58 and @xmath82 . in order to complete the description of our 5-box representation @xmath156 of @xmath2 , we only need to prescribe the images of the vertices of @xmath142 in @xmath157 and the images of the vertices of @xmath148 in @xmath158 . all these images will be defined using corollary [ cor : plaframe ] , as we now explain . consider first the 3-box representation @xmath159 , restricted to the vertices of @xmath135 ( i.e. the vertices of @xmath136 ) . since @xmath114 is a strict 2-box representation of @xmath135 and any three intervals of @xmath152 corresponding to some vertices of @xmath136 are strictly overlapping , by lemma [ lem : corner ] every ( facial ) triangle of @xmath135 has an empty inner corner in @xmath159 . by corollary [ cor : plaframe ] , this representation of @xmath135 can be extended to @xmath142 , so that each vertex @xmath15 of @xmath142 , lying inside some facial triangle @xmath160 of @xmath97 , is mapped to a 3-box @xmath13 such that all the points of @xmath13 are at @xmath161-distance at most @xmath162 from the empty inner corner of @xmath160 . this shows how to extend @xmath159 to @xmath142 ( in the sense that the restriction of @xmath159 to @xmath163 is a representation of the subgraph of @xmath2 induced by @xmath163 ) . similarly , consider the 3-box representation @xmath164 , restricted to @xmath143 ( i.e. to the vertices of @xmath165 ) . as before , we can prove using lemma [ lem : corner ] that all facial triangles of @xmath143 have an empty inner corner in this representation , and it follows from corollary [ cor : plaframe ] that this representation of @xmath143 can be extended to @xmath148 , so that each vertex @xmath15 of @xmath148 , lying inside some facial triangle @xmath160 of @xmath97 , is mapped to a 3-box @xmath13 such that all the points of @xmath13 are at @xmath161-distance at most @xmath162 from the empty inner corner of @xmath160 . recall that by definition of @xmath148 , such a triangle @xmath160 intersects @xmath58 , and therefore @xmath160 is also a facial triangle of @xmath143 . this shows how to extend @xmath164 to @xmath148 ( in the sense that the restriction of @xmath164 to @xmath166 is a representation of the subgraph of @xmath2 induced by @xmath166 ) and completes the description of our 5-box representation @xmath167 of @xmath2 . we now prove that @xmath39 is a representation of @xmath2 , i.e. @xmath2 is precisely the intersection of the interval graphs @xmath38 , for @xmath168 . by the definition of @xmath39 , the restriction of @xmath39 to @xmath169 represents the subgraph of @xmath2 induced by @xmath169 and the restriction of @xmath39 to @xmath170 represents the subgraph of @xmath2 induced by @xmath170 . so we only need to consider pairs of vertices @xmath171 and @xmath172 ( by definition , two such vertices @xmath88 are non - adjacent in @xmath2 ) , and show that the 5-boxes of @xmath15 and @xmath16 in @xmath39 are disjoint . to see this , consider only @xmath152 and observe that all vertices of @xmath148 are mapped to intervals that are at distance at most @xmath162 of an interval of @xmath58 , so all the intervals of @xmath173 end before @xmath174 . similarly , all the intervals of @xmath175 start after @xmath176 . it follows that the intervals of @xmath15 and @xmath16 are disjoint in @xmath152 and consequently , the boxes of @xmath15 and @xmath16 are disjoint in @xmath39 , as desired . it was proved in @xcite that toroidal graphs have boxicity at most 7 , while there are toroidal graphs of boxicity 4 . the proof was mainly based on the the following result of schrijver @xcite : @xcite[thm : sch ] every graph embedded in the torus with face - width @xmath49 contains @xmath177 vertex - disjoint ( and homotopic ) non - contractible cycles . using some ideas of the proof of theorem [ thm : locpla ] , we now improve the bound on the boxicity of toroidal graphs from 7 to 6 . theorem [ thm : locpla ] implies that graphs embedded on the torus with edge - width at least 40 have boxicity at most 5 . we also decrease this bound on the edge - width from 40 to 4 . [ thm : tor ] if @xmath2 is a graph embedded on the torus , then @xmath2 has boxicity at most 6 . if , moreover , @xmath2 has edge - width at least 4 , then @xmath2 has boxicity at most 5 . let @xmath2 be a graph embedded on the torus . as before , we can assume without loss of generality that @xmath2 is a triangulation , and thus the face - width and the edge - width of @xmath2 are equal . if @xmath2 has edge - width at least 6 , then by theorem [ thm : sch ] , @xmath2 has 4 vertex - disjoint ( and homotopic ) non - contractible cycles . let @xmath58 be one of them . we can assume that @xmath58 is chordless ( see @xcite ) . note that because of the 4 vertex - disjoint cycles , the subgraph of @xmath2 induced by @xmath58 and its neighborhood can be embedded in a cylinder , and is therefore planar . the exact same proof as that of theorem [ thm : locpla ] ( with @xmath178 ) then shows that @xmath2 has boxicity at most 5 . assume now that @xmath2 has edge - width at most 5 and at least 4 , and let @xmath58 be a shortest non - contractible cycle ( in particular , @xmath58 is chordless ) . observe that @xmath179 is planar ; in the remainder of the proof , we fix a planar embedding of @xmath179 . since @xmath2 has no non - contractible triangles , all the triangles of @xmath2 ( and @xmath179 ) are contractible . let @xmath97 be the frame of ( the fixed embedding of ) @xmath179 . let @xmath82 be the set of vertices of @xmath97 , and let @xmath180 be the set of vertices of @xmath2 lying in a non - facial triangle of ( the fixed planar embedding of ) @xmath179 . if one of the two faces of @xmath179 with vertex - set @xmath59 or @xmath60 is contained in the interior of some triangle @xmath101 of @xmath179 , then @xmath101 is homotopic to @xmath58 in @xmath2 , and thus non - contractible ( which contradicts the fact that @xmath2 has edge - width at least 4 ) . it follows that @xmath59 and @xmath60 are included in @xmath82 and consequently , no vertex of @xmath180 is adjacent to a vertex of @xmath58 . let @xmath110 be a strict 2-box representation of @xmath181 $ ] as in the proof of theorem [ thm : locpla ] ( since @xmath2 has edge - width at least 4 , a proof similar to that of theorem [ thm : locpla ] shows that such a representation exists ) . we first extend this 2-dimensional representation to @xmath58 by mapping all the vertices of @xmath58 to a large 2-box containing all the 2-boxes of @xmath110 . recall that @xmath58 induces a cycle of length 4 or 5 . we now partition the vertices of @xmath58 into three independent sets @xmath182 , each containing one or two vertices . for any vertex @xmath16 of @xmath2 , let @xmath183 denote the neighborhood of @xmath16 in @xmath184 . for each @xmath185 , we denote by @xmath38 the interval graph with vertex - set @xmath184 depicted in figure [ fig : interval ] ( left ) if @xmath186 or figure [ fig : interval ] ( right ) if @xmath187 . observe that in @xmath164 , @xmath58 induces a cycle , @xmath82 induces a clique , and the adjacency between @xmath58 and @xmath82 is the same as in @xmath2 . with @xmath186 ( left ) , and the description of @xmath38 with @xmath187 ( right ) . we use the following notation to avoid overloading the figure : @xmath188 ( left ) and @xmath189 ( right).[fig : interval ] ] since @xmath58 contains at most 5 vertices , one of the sets @xmath190 ( say @xmath191 ) contains only one vertex , call it @xmath132 . as before , we choose a small real @xmath192 for each vertex @xmath16 of @xmath97 , so that all the chosen @xmath151 are distinct , and we change each interval @xmath193 $ ] of @xmath16 in @xmath94 to @xmath194 $ ] . if each @xmath195 is small enough , this does not change the graph induced by @xmath94 ( and thus the graph induced by @xmath164 ) . let @xmath101 be a triangle of @xmath97 . since @xmath101 is disjoint from @xmath132 ( the unique vertex of @xmath191 ) , it follows from the definition of @xmath94 ( modified with the @xmath195 ) that the intervals corresponding to @xmath111 in @xmath94 are strictly overlapping . by lemma [ lem : corner ] , every triangle of @xmath97 has an empty inner corner in the strict 3-box representation @xmath129 . by corollary [ cor : plaframe ] , we can extend the strict 3-box representation @xmath129 of @xmath97 to @xmath180 ( so that in @xmath94 , all the newly added intervals are disjoint from the interval of @xmath132 ) . in @xmath196 , it remains to map all the vertices of @xmath180 to a 2-box of @xmath196 intersecting all the 2-boxes except that of the vertices of @xmath58 . this can be done for instance by mapping in @xmath38 each vertex of @xmath180 to the interval labelled @xmath197 in figure [ fig : interval ] ( left ) , or the interval labelled @xmath198 in figure [ fig : interval ] ( right ) . this ensures that in @xmath164 , the vertices of @xmath180 are adjacent to all the other vertices of @xmath179 , and non - adjacent to all the vertices of @xmath58 , as desired . the representation @xmath199 of @xmath2 then shows that @xmath2 has boxicity at most 5 . assume now that @xmath2 has edge - width at most 3 . then a set @xmath200 of at most 3 vertices can be removed from @xmath2 so that @xmath201 is planar , and thus has boxicity at most 3 . take a 3-box representation of @xmath179 , and extend it to @xmath58 by mapping all the vertices of @xmath58 to a large 3-box containing all the 3-boxes of the representation . now add 3 intervals graphs , one for each element of @xmath200 , defined as in figure [ fig : interval ] ( left ) , where we now define @xmath202 , with @xmath203 , as the neighborhood of @xmath132 in @xmath2 , and @xmath197 as @xmath204 . the obtained 6-box representation induces @xmath2 , so @xmath2 has boxicity at most 6 . theorem [ thm : locpla ] easily implies that there exists a function @xmath4 , such that for any @xmath20 and any graph @xmath2 embedded on @xmath65 , a set of at most @xmath6 vertices can be removed from @xmath2 so that the resulting graph has boxicity at most 5 . however , the function @xmath4 derived from theorem [ thm : locpla ] is exponential in @xmath5 . in this section , we show how to make @xmath4 linear in @xmath5 . note that the proof works for graphs of euler genus @xmath5 ( while theorem [ thm : locpla ] is only concerned with graphs embeddable on @xmath65 ) . the previously best known result of this type was that @xmath205 vertices can be removed in any graph of euler genus @xmath5 , so that the resulting graph has boxicity at most 42 @xcite . we will use a technique of kawarabayashi and thomassen @xcite , who used it to prove several results of this type . kawarabayashi and thomassen ( * ? ? ? * theorem 1 ) proved that any graph @xmath2 embedded on some surface of euler genus @xmath5 , with face - width more than @xmath206 ( for some constant @xmath126 ) has a partition of its vertex - set into three parts @xmath207 , such that @xmath208 has size at most @xmath209 , @xmath210 consists of the disjoint union of paths that are local geodesics ( in the sense that each subpath with at most @xmath126 vertices of a path of @xmath210 is a shortest path in @xmath2 and any two vertices at distance at least @xmath126 in some path of @xmath210 are at distance at least @xmath126 in @xmath2 ) and are pairwise at distance at least @xmath126 in @xmath2 , and @xmath211 induces a planar graph having a plane embedding @xmath97 such that the only vertices of @xmath211 having a neighbor in @xmath210 lie on the outerface of @xmath97 . we will also use the following technical lemma . [ lem : tech ] let @xmath2 be a graph whose vertex - set is partitioned into two sets @xmath212 and @xmath210 , such that @xmath212 induces a complete graph , @xmath210 induces a path , and for every vertex @xmath15 of @xmath212 , the neighbors of @xmath15 in @xmath210 lie in a subpath of @xmath210 of at most 3 vertices ( equivalently , any two neighbors of @xmath15 in @xmath210 are at distance at most two in @xmath210 ) . then for any real number @xmath126 , @xmath2 has a 3-box representation @xmath129 such that all the intervals of @xmath94 corresponding to some vertex of @xmath212 end at @xmath126 , while all the intervals of @xmath94 corresponding to some vertex of @xmath210 end strictly before @xmath126 . and the bottom corners @xmath213 of the vertices @xmath214 are depicted with white dots . for the sake of readability , the 3-boxes of @xmath215 and @xmath216 are not displayed ( only two of their corners are depicted ) . [ fig : kp ] ] we first construct a 3-box representation of @xmath2 , and then show how to slightly modify it so that it satisfies the additional constraint on @xmath94 . let @xmath217 . for every @xmath218 , @xmath219 is mapped to the 3-box @xmath220\times [ -1,i]\times [ 2i,2i+1]$ ] and @xmath221 is mapped to the 3-box @xmath222\times [ i , i+1]\times [ 2i+1,2i+2]$ ] ( see figure [ fig : kp ] , where both a 3-dimensional view and a 2-dimensional view from above are depicted for the sake of clarity ) . let @xmath15 be a vertex of @xmath212 . then @xmath15 is mapped to the 3-box with corners @xmath223 and @xmath224 , where @xmath224 is defined as follows . if @xmath15 has no neighbor in @xmath210 , then @xmath225 . if @xmath15 has a single neighbor @xmath226 in @xmath210 , then either @xmath227 and we define @xmath228 , or @xmath229 and we define @xmath230 ( see for example the 3-box of @xmath231 in figure [ fig : kp ] ) . if the neighbors of @xmath15 are two consecutive vertices of @xmath210 , say @xmath219 and @xmath221 ( the case @xmath232 can be handled analogously by switching the roles of the @xmath132- and @xmath233-axis ) , then we set @xmath234 ( see for example the bottom corner @xmath235 of @xmath215 in figure [ fig : kp ] ) . if the neighbors of @xmath15 in @xmath210 are @xmath219 and @xmath236 , then @xmath228 ( see for example the 3-box of @xmath237 in figure [ fig : kp ] ) . the case where the neighbors of @xmath15 in @xmath210 are @xmath221 and @xmath238 is handled analogously by switching the roles of the @xmath132- and @xmath233-axis . finally , if the neighbors of @xmath15 in @xmath210 are @xmath239 , for some @xmath37 , then we set @xmath240 . again , the case where the neighbors of @xmath15 in @xmath210 are @xmath241 is handled analogously by switching the roles of the @xmath132- and @xmath233-axis ( see for example the bottom corner @xmath242 of @xmath216 in figure [ fig : kp ] ) . all the 3-boxes of the vertices of @xmath212 contain the point @xmath223 , so @xmath212 induces a complete graph in the representation defined above . moreover , it readily follows from the definition of the 3-boxes of the vertices @xmath243 and the corners @xmath224 that for each vertex @xmath15 of @xmath212 , the neighbors of @xmath15 in @xmath210 are precisely the same in the graph @xmath2 and in the 3-box representation defined above . consequently , the constructed representation induces @xmath2 , as desired . let @xmath244 be the three interval graphs corresponding respectively to the @xmath132- , @xmath233- , and @xmath245-axis in the representation above . it follows from the construction of @xmath94 that all the vertices @xmath246 are mapped in @xmath94 to an interval of the form @xmath247 $ ] , while all the vertices of @xmath210 are mapped in @xmath94 to intervals ending strictly before @xmath248 . it is then easy to translate the whole representation along the @xmath245-axis so that it satisfies the additional property . finally , we will need the following direct consequence of theorem [ thm : tho45v2 ] . [ cor : plafixed ] let @xmath2 be a planar graph , let @xmath249 be a fixed vertex of @xmath2 , and let @xmath250 . then @xmath2 has a strict 3-box representation @xmath129 such that the interval @xmath128 of @xmath249 in @xmath94 ends at @xmath126 , while all the other intervals of @xmath94 are contained in @xmath251 . we can assume without loss of generality that @xmath2 is a triangulation ( since it is an induced subgraph of some triangulation ) , and that @xmath249 belongs to the outerface of @xmath2 . we first map the three vertices of the outerface to 3-boxes as in figure [ fig : corner ] , and then apply theorem [ thm : tho45v2 ] to extend this representation to a strict 3-box representation of @xmath2 , such that the boxes of all the internal vertices are inside the inner corner @xmath58 . note that some hyperplane separates the box @xmath252 of @xmath249 from the boxes of all the other vertices of @xmath2 , and the representation can therefore be translated in @xmath253 in order to satisfy the desired property . we are now able to prove the main result of this section . [ thm : linex1 ] let @xmath2 be a graph of euler genus @xmath25 . then @xmath2 contains a set @xmath208 of at most @xmath254 vertices such that @xmath255 has boxicity at most 5 . we prove the theorem by induction on @xmath25 . if @xmath2 has face - width at most @xmath256 , then @xmath2 contains a set @xmath208 of at most @xmath256 vertices such that @xmath255 has euler genus at most @xmath257 , or @xmath255 is the disjoint union of two graphs of euler genus @xmath258 and @xmath259 with @xmath260 ( see proposition 4.2.1 and lemma 4.2.4 in @xcite ) . in the first case , either @xmath255 is planar ( in which case the result clearly holds , since @xmath255 has boxicity at most 3 and @xmath261 ) , or by the induction , a set @xmath262 of at most @xmath263 can be removed from @xmath2 in order to obtain a graph with boxicity at most 5 . in the second case , by the induction , a set @xmath262 of at most @xmath264 can be removed from @xmath2 in order to obtain a graph with boxicity at most 5 . as a consequence , we can assume that @xmath2 has face - width at least @xmath256 , and apply the result of kawarabayashi and thomassen mentioned above , with @xmath265 . let @xmath207 be the corresponding partition of the vertex - set of @xmath2 ( and let @xmath97 be the planarly embedded subgraph of @xmath2 induced by @xmath211 , such that only the outerface @xmath266 of @xmath97 has neighbors in @xmath210 ) . note that @xmath208 contains at most @xmath267 vertices , and we will prove that @xmath268 $ ] , the subgraph of @xmath2 induced by @xmath211 and @xmath210 , has boxicity at most 5 . let @xmath269 be the planar graph obtained from @xmath97 by adding a new vertex @xmath270 adjacent to all the vertices of @xmath266 . by corollary [ cor : plafixed ] ( with @xmath271 ) , @xmath269 has a strict 3-box representation @xmath272 such that for some real number @xmath273 , all the intervals of @xmath274 corresponding to some vertex of @xmath266 start at @xmath273 , while all the intervals of @xmath274 corresponding to some vertex of @xmath275 start ( strictly ) after @xmath273 . let @xmath16 be a vertex of @xmath97 . since the paths of @xmath210 are local geodesics , and any two paths are at distance at least 3 apart , @xmath16 has at most 3 neighbors in @xmath210 and these neighbors lie on a subpath of at most 3 vertices of a path of @xmath210 ( i.e. they are either consecutive or at distance two on some path of @xmath210 ) . let @xmath276 be the paths of @xmath210 , and for each @xmath277 , consider the two endpoints of @xmath278 and decide arbitrarily which one is the left endpoint and which one is the right endpoint . let @xmath279 be the graph obtained from @xmath280 $ ] by adding , for each @xmath281 , a vertex @xmath243 adjacent ( only ) to the right endpoint of @xmath278 and the left endpoint of @xmath282 , and by adding an edge between any two ( non - adjacent ) vertices of @xmath266 . by lemma [ lem : tech ] , @xmath279 has a 3-box representation @xmath283 such that the intervals of @xmath284 either end at @xmath273 ( if they correspond to a vertex of @xmath266 ) , or end strictly before @xmath273 ( if they correspond to a vertex of @xmath210 ) . the restriction of @xmath283 to @xmath285 induces a 3-box representation of the graph obtained from @xmath280 $ ] by adding an edge between any two ( non - adjacent ) vertices of @xmath266 , with @xmath284 satisfying the same additional property as above . let @xmath94 be the interval representation obtained from @xmath274 and @xmath284 as follows . every vertex of @xmath275 is mapped to its image in @xmath274 , every vertex of @xmath210 is mapped to its image in @xmath284 , and every vertex of @xmath266 is mapped to the concatenation of its images in @xmath284 and @xmath274 ( note that the former ends at @xmath273 and the latter starts at @xmath273 ) . note that the adjacency between @xmath266 and @xmath210 is the same in @xmath94 and @xmath284 , and the adjacency between @xmath266 and @xmath286 is the same in @xmath94 and @xmath274 . moreover , the intervals of @xmath275 are disjoint from the intervals of @xmath210 in @xmath94 . it remains to map every vertex of @xmath210 in @xmath114 to a large 2-box containing all the other 2-boxes of @xmath114 , and to map every vertex of @xmath275 in @xmath196 to a large 2-box containing all the other 2-boxes of @xmath196 . let @xmath167 . note that the restrictions to @xmath211 of @xmath39 , @xmath129 , and @xmath272 , all induce the same graph ( namely , @xmath287 $ ] ) . similarly , the restrictions to @xmath288 of @xmath39 , @xmath164 , and @xmath289 , all induce the same graph ( namely , @xmath290 $ ] ) . since the intervals of @xmath275 are disjoint from the intervals of @xmath210 in @xmath94 , there are no edges between @xmath210 and @xmath275 in @xmath39 . it follows that @xmath82 represents @xmath291=g - x$ ] , as desired . the previous sections were devoted to graphs embedded in fixed surfaces , without short non - contractible cycles . here we consider graphs without short cycles at all . using the results of @xcite relating the boxicity of a graph and its second largest eigenvalue ( in absolute value ) , together with the existence of ramanujan graphs of arbitrarily large degree and girth @xcite , it directly follows that there is a constant @xmath292 such that for any integers @xmath0 and @xmath5 , there is a @xmath49-regular graph ( @xmath293 ) of girth at least @xmath5 and boxicity at least @xmath294 . as a consequence , there are ( regular ) graphs with arbitrarily large girth and boxicity . therefore , in order to bound the boxicity of graphs without short cycles , it is necessary to restrict ourselves to specific classes of graphs . let @xmath295 be an integer . a graph @xmath2 is said to be _ @xmath296-path - degenerate _ ( see @xcite ) if any subgraph @xmath97 of @xmath2 contains a vertex of degree at most 1 , or a path with @xmath296 internal vertices , each having degree two in @xmath97 . a 3-box representation of a graph @xmath2 is called a _ 3-segment representation _ if ( 1 ) each vertex is mapped to a _ segment _ , ( the cartesian product of two points and an interval of positive length ) , ( 2 ) the interiors of any two segments are disjoint ( in other words , the interior of a segment is only intersected by endpoints of other segments ) , and ( 3 ) no two segments lie on the same line . we will prove the following result : [ thm:6path ] any 5-path - degenerate graph @xmath2 has a 3-segment representation . and @xmath297 when @xmath298 and @xmath299 are parallel to different axes ( left ) and when @xmath298 and @xmath299 are parallel to the same axis ( right ) . [ fig:5path ] ] we will prove the result by induction on the number of vertices of @xmath2 . assume first that @xmath2 contains a vertex @xmath16 of degree at most 1 , and let @xmath300 . note that @xmath97 is 5-path - degenerate , so by the induction it has a 3-segment representation @xmath301 . if @xmath16 has degree 0 , then @xmath2 is the disjoint union of @xmath97 and @xmath302 , and clearly has a 3-segment representation . thus , we can assume that @xmath16 has a unique neighbor @xmath15 in @xmath2 . since @xmath303 contains a finite number of segments , @xmath304 contains a point @xmath296 such that some small ball @xmath305 centered in @xmath296 only intersects @xmath304 . we then represent @xmath306 as a segment orthogonal to @xmath304 ( there are two possible choices of dimension ) , with @xmath296 as one endpoint , and such that @xmath306 lies inside @xmath305 . in the remainder of the proof , we assume that @xmath2 contains a path @xmath307 , such that for any @xmath168 , the only neighbors of @xmath243 in @xmath2 are @xmath308 and @xmath309 . let @xmath97 be the graph obtained from @xmath2 by removing all the vertices @xmath243 with @xmath168 , and let @xmath301 be a 3-segment representation of @xmath97 . we now extend @xmath303 to the vertices @xmath310 . for @xmath311 , fix a point @xmath312 such that some small ball @xmath313 centered in @xmath314 only intersects @xmath315 . assume without loss of generality that either @xmath298 is parallel to the @xmath245-axis and @xmath299 is parallel to the @xmath233-axis ( this includes the case where @xmath316 and @xmath297 are adjacent ) , or both @xmath298 and @xmath299 are parallel to the @xmath233-axis ( this includes the case where @xmath316 and @xmath297 are the same vertex ) . assume that @xmath317 and @xmath318 have radius at least @xmath151 , for some @xmath319 . we map @xmath320 to the segment @xmath321 that has @xmath322 as an endpoint , is parallel to the @xmath132-axis , has length @xmath151 , and goes in the direction of @xmath323 ( parallel to the @xmath132-axis ) . similarly , we map @xmath324 to the segment @xmath325 that has @xmath323 as an endpoint , is parallel to the @xmath245-axis , has length @xmath151 , and goes in the direction of @xmath322 ( parallel to the @xmath245-axis ) . let @xmath326 be the endpoint of @xmath321 distinct from @xmath322 , and let @xmath327 be the endpoint of @xmath325 distinct from @xmath323 . let @xmath58 be the 3-box with corners @xmath326 and @xmath327 . we map @xmath328 to the edge @xmath329 of @xmath58 which contains @xmath326 and is orthogonal to @xmath298 and @xmath321 , and similarly we map @xmath330 to the edge @xmath331 of @xmath58 which contains @xmath327 and is orthogonal to @xmath325 and @xmath299 . finally , we map @xmath332 to the edge @xmath333 of @xmath58 connecting the endpoint of @xmath329 distinct from @xmath326 and the endpoint of @xmath331 distinct from @xmath327 ( see figure [ fig:5path ] ) . note that the description of @xmath315 , @xmath168 , above only depends of the choice of @xmath322 , @xmath326 , and @xmath151 . we now move each of @xmath322 , @xmath326 and the value of @xmath151 along a tiny interval . then the locus described by each @xmath315 , @xmath168 , is a non - degenerate 3-box . since @xmath303 is the union of a finite number of segments , it follows that we can choose @xmath322 , @xmath326 , and @xmath319 so that the @xmath315 ( @xmath168 ) are disjoint from @xmath303 . we can moreover choose @xmath322 , @xmath326 , and @xmath319 so that @xmath58 is a non - degenerate 3-box ( and so the @xmath315 are ( non - degenerate ) segments ) and no segment @xmath315 lie on the same line as a segment of @xmath303 . consequently , the obtained representation is a 3-segment representation and each vertex @xmath243 with @xmath168 is mapped to a segment that only intersects the segments of @xmath308 and @xmath309 , as desired . it was proved by galluccio , goddyn and hell @xcite that for any proper minor - closed class @xmath7 and for any @xmath49 , there is an integer @xmath334 such that any graph of @xmath7 with girth at least @xmath5 is @xmath49-path - degenerate . since 3-segment representations are 3-box representations , we have the following immediate consequence . [ cor : minor ] for any proper minor - closed class @xmath7 there is an integer @xmath335 such that any graph of @xmath7 of girth at least @xmath5 has boxicity at most 3 . note that the result of galluccio , goddyn and hell was recently extended by neetil and ossona de mendez @xcite to classes of subexponential expansion , i.e. expansion bounded by @xmath336 , for some @xmath319 ( see @xcite for definitions and further details ) . this shows that corollary [ cor : minor ] can be extended to this fairly broad setting as well . interestingly , the bound on the boxicity in corollary [ cor : minor ] is best possible already for the class of @xmath337-minor free graphs . the following example was given by stphan thomass . take a copy of @xmath338 , the complete graph on 5 vertices , and replace each edge by an arbitrarily large path . the resulting graph has arbitrarily large girth , no @xmath337-minor , and any 2-box representation of it would give a planar embedding ( without crossings ) of @xmath338 , a contradiction . for graphs of euler genus @xmath5 , theorem 3.2 in @xcite ( combined with theorem [ thm:6path ] ) implies the following interesting counterpart of theorem [ thm : locpla ] ( see the difference between the exponential bound there and the logarithmic bound here ) . [ cor : girth ] there is a constant @xmath27 such that any graph of euler genus @xmath5 and girth at least @xmath32 has boxicity at most 3 . observe that this is best possible up to the choice of the constant @xmath27 : for any integer @xmath49 , there is a constant @xmath339 and an infinite family of graphs of ( increasing ) euler genus @xmath5 , girth at least @xmath340 , and boxicity at least @xmath49 . this follows from the results of @xcite mentioned in the introduction of this section , and the fact that the ramanujan graphs described in @xcite have girth logarithmic in their number of vertices ( and euler genus linear in their number of vertices , at least for @xmath0-regular graphs with @xmath341 ) . it is worth noting that theorem [ thm:6path ] can also be applied to classes that do not fit well in the framework of neetil and ossona de mendez @xcite ( because their density is too high ) . examples of such classes include _ segment _ or _ strings _ graphs ( intersection graphs of segments , or strings in the plane ) , or _ circle _ graphs ( intersection graphs of chords of a circle ) . for example , it can be proved using theorem [ thm:6path ] and the results of @xcite that every circle graph of girth at least 9 has boxicity at most 3 . this is in contrast with the existence of a circle graph ( indeed , a permutation graph ) on @xmath342 vertices with boxicity @xmath343 , for every @xmath344 ( see @xcite ) . a natural problem is to find a counterpart of theorem [ thm : locpla ] for non - orientable surfaces . non - orientable versions of theorem [ thm : placc ] exist @xcite , so the only problem when applying the same arguments as in the proof of theorem [ thm : locpla ] to a graph embedded in a non - orientable surface is that some of the cycles in the planarizing collection might be one - sided , in which case the cycle @xmath58 together with its neighborhood @xmath74 does not necessarily embed in a cylinder , but instead on a mbius strip . as a consequence , these graphs are not necessarily planar . however , using lemma [ lem : tech ] , we can prove that the graph obtained from the subgraph induced by @xmath80 by adding an edge between any two vertices of @xmath74 , has boxicity at most 4 ( just remove a vertex of @xmath58 and apply lemma [ lem : tech ] ) . consequently , a proof along the lines of the proof of theorem [ thm : locpla ] easily shows that locally planar graphs embedded on non - orientable surfaces have boxicity at most @xmath345 . it would be interesting to improve this bound , as well as that of theorem [ thm : locpla ] . it was conjectured in @xcite that locally planar graphs have boxicity at most 3 ( which would be best possible since there are planar graphs of boxicity 3 ) . we also conjecture the following variant : note that the linear bound ( in @xmath5 ) would be best possible , since there are toroidal graphs with boxicity 4 ( for example @xmath347 minus a perfect matching , see @xcite ) , and the disjoint union of @xmath348 such graphs can be embedded in a surface of euler genus @xmath5 .

supersymmetry and extra dimension need not be mutually exclusive options of physics for the tev scale and beyond . intrinsically higher dimensional top - down scenarios , e.g. string models , often contain supersymmetry at the weak scale . in this paper , we envisage a more phenomenological scenario by embedding the 4d constrained minimal supersymmetric standard model in a flat 5d orbifold , with the inverse radius of compactification at the tev scale . the gauge and higgs supermultiplets are placed in the bulk . we assume that only the third generation matter multiplet accesses the bulk , while the first two generations are confined to a brane on an orbifold fixed point . from a 4d perspective , the bulk has supersymmetry which entails a special non - renormalization theorem giving rise to a significant numerical impact on the renormalization group running of various parameters . the brane supersymmetry corresponds to which we assume to be broken in an unspecified but phenomenologically acceptable way . given this set - up , we study how the gauge and yukawa couplings and the brane supersymmetry breaking soft masses run through the energy scale exciting the kaluza - klein states at regular interval . in the process , we ensure that electroweak symmetry does break radiatively . we confront our low energy parameters with the experimental measurements or limits of different observables , e.g. lep lower limits on the lightest higgs boson and the chargino , the of muon , the branching ratio of , and the wmap probe of relative dark matter abundance . we present our results by showing the allowed / disallowed zone in the plane of the common scalar mass ( ) and common gaugino mass ( ) for both positive and negative parameter . our plots are the first 5d versions of the often displayed 4d plots , and we provide reasons behind the differences between the 4d and 5d plots . ` pacs nos : 11.30.pb , 14.80.rt ` + ` key words : supersymmetry , extra dimension ` sinp / tnp/2010/03 * a phenomenological study of 5d supersymmetry * + gautam bhattacharyya and tirtha sankar ray +

1003.1276

the yet elusive higgs boson is the primary target of the cern large hadron collider ( lhc ) . but , lhc is also expected to reveal a new ruler of the tera - electron - volt ( tev ) territories . the standard model ( sm ) has so far been remarkably successful in explaining physics up to a few hundred gev energy scale . but theoretical inconsistencies of the sm ( like , gauge hierarchy problem ) and experimental requirements ( like , a candidate to account for the dark matter of the universe ) suggest that there are good reasons to believe that new physics beyond the sm is just around the corner crying out for verification . among the different possibilities , supersymmetry and extra dimension stand out as the two leading candidates for dictating terms in the tev regime . these two apparently distinct classes of scenarios cover a wide variety of more specific models . the usual practice from a bottom - up approach is to attach an ` either / or ' tag on supersymmetry and extra dimension , as if the presence of one excludes the other . a more careful thought would reveal that the relationship between these two is _ not necessarily _ mutually exclusive . in fact , the presence of higher dimensions is a common feature of any fundamental theory valid at high scale . we will get back to this issue a little later . for the moment , to put things into perspective , we recapitulate the chronological evolution of the extra dimensional scenarios without invoking supersymmetry _ a priori_. we restrict our discussion to the flat space scenarios , as we are not pursuing the warped path in this paper besides mentioning it as an aside . flat extra dimensions were first studied @xcite in a scenario where gravity propagates in a millimeter ( mm ) size compact space dimension , with the sm particles confined to a 4d brane . the motive was to bring down the fundamental planck scale to about a tev . subsequently , it was conceived that the brane where the sm particles live may actually have a very small size , like @xmath10 cm @xmath11 , leading to the concept of a ` fat brane ' @xcite . in the context of the present paper , we stick to the fat brane scenario . _ what are the experimental bounds on the fatness of such a brane , more precisely , on the radius of compactification ( @xmath12 ) _ ? for universal extra dimension ( ued ) models @xcite , in which _ all _ the sm particles access the extra dimensional bulk , a safe estimate is @xmath13 gev . more specifically , the @xmath14 of the muon @xcite , flavor changing neutral currents @xcite , @xmath15 decay @xcite , the @xmath16 parameter @xcite , and hadron collider studies @xcite reveal that @xmath17 gev . consideration of @xmath4 , however , implies a somewhat tighter bound ( @xmath18 gev @xcite ) . methods to decipher its signals from the lhc data have recently been discussed too @xcite . on the other hand , in the non - universal scenario where both the sm gauge bosons and the higgs boson propagate in the bulk but the fermions are confined to a 4d brane @xcite , @xmath19 can not be below @xmath20 tev @xcite . the reason behind the difference in constraints is the following . the kk parity , defined by @xmath21 for the @xmath22th kk label , is conserved in ued , while it is not a good symmetry in the non - universal scenario . as a result , while in the non - universal models kk states can mediate many processes at tree level yielding strong constraints , in the ued model , thanks to the kk parity , kk states appear only inside a loop leading to milder constraints . _ in any case , in the presence of supersymmetry , all those analyses need to be modified with more parameters , which would expectedly lead to a set of more relaxed bounds on @xmath19_. _ now , what is the motivation of studying a tev scale ( or , a fat brane ) extra dimension scenario _ ? the implications of such models have been investigated from the perspective of string theory , phenomenology , cosmology / astrophysics and high energy experiments . such models provide a cosmologically stable dark matter candidate @xcite , trigger electroweak symmetry breaking successfully through a composite higgs @xcite , address the fermion mass hierarchy problem from a different point of view @xcite tev due to large flavor - changing neutral currents and cp violation @xcite . ] , and stimulate power law renormalization group ( rg ) running yielding a lower ( few tens of a tev ) gauge coupling ( near-)unification scale @xcite . besides , the running of neutrino mixing angles generated from effective majorana mass operator in a 5d set - up has been studied both in non - supersymmetric @xcite and supersymmetric @xcite contexts . _ but , what is the advantage of supersymmetrizing it ? _ we argue that supersymmetry and extra dimension need not always be seen as _ two _ new physics considered simultaneously . in fact , they may nicely complement each other in _ some situations _ through mutual requirements . this can be seen as follows : 1 . from a top - down approach , string theory provides a rationale behind linking supersymmetry and extra dimension . the string models are intrinsically extra dimensional , and more often than not contain supersymmetry as an integral part . that said , we must also admit that establishing a rigorous connection between a _ realistic _ low energy supersymmetric model with string theory is still a long shot , though a lot of efforts have already been put in that direction @xcite . the higher dimensional field theories are non - renormalizable , and computation of quantum corrections in such scenarios require the existence of a uv completion . superstring theories , which contain supersymmetry , provide the main hope in this context . in fact , a joint rle of ` partial supersymmetry ' ( broken at very high scale ) and warped extra dimension / compositeness has been envisaged from ads / cft point of view to account for both the ` little ' and the ` big ' hierarchies @xcite . 3 . even after embedding the sm in an extra dimensional set - up , the scalar potential remains unstable under quantum correction . supersymmetrization stabilizes it and ameliorates the hierarchy problem . it is interesting to note that by admitting chiral fermions and their scalar partners in the same multiplet tacitly provides a rationale behind treating the higgs boson as an elementary object . an elementary higgs can be perfectly accommodated in a flat extra dimensional set - up . as a corollary , the upper limit on the lightest supersymmetric neutral higgs is relaxed beyond the 4d upper limit of 135 gev due to the presence of the kk towers of top / stop chiral multiplets , and the _ hitherto _ disfavored low @xmath23 region can be resurrected @xcite . each 4d supersymmetric scenario has its own supersymmetry breaking mechanism . the origin of this mechanism may be linked to the existence of extra dimension . in fact , one of the earliest motivations of a tev scale fat brane was to relate the scale of 4d supersymmetry breaking with @xmath19 @xcite . keeping these in mind , we outline the formalism of a 5d supersymmetric model in an @xmath24 orbifold which contains the 4d supersymmetric states as zero modes . in section [ 5dmssm ] , we state our assumptions leading to the construction of the 5d model and comment on supersymmetry breaking . furthermore , we explicitly write down the particle content and their 5d lagrangian and illustrate the kk decompositions of the different 5d fields . in section [ beta ] , we derive the beta functions of the gauge and yukawa couplings as well as those of the different soft supersymmetry breaking parameters _ diagram by diagram _ , pointing out how they are all modified from their 4d values due to the presence of kk states . in section [ numerical ] , we discuss the numerical effects of rg running and highlight the reason behind the differences between the 4d and 5d scenarios . we also point out under what conditions we can ensure electroweak symmetry breaking . in section [ m0mhalf ] , we exhibit the numerical impact of rg running through plots showing constraints in the @xmath5@xmath6 plane . we standardize our numerical codes by reproducing the known 4d plots before encoding the necessary alterations for producing the new plots pertaining to 5d supersymmetry . this also enables us to compare and contrast the 4d and 5d allowed / disallowed zones . finally , in section [ concl ] , we showcase the essential features we have learnt from this analysis . we highlight the salient features of supersymmetry in higher dimension and outline below the various assumptions that lead to a calculable phenomenological framework . 1 . we consider a 5d flat space time metric . the 5th dimension is compactified on a @xmath25 orbifold . orbifolding is necessary to reproduce chiral zero mode fermions as a 5d theory is vector - like . we embed the minimal supersymmetric standard model ( mssm ) in this higher dimensional set - up ( several consequences of such embedding , mainly the effects on gauge and yukawa couplings evolution , have been studied in @xcite ) . from a 4d point of view this leads to a tower of kk states . the massless sector corresponds to the 4d mssm states . 3 . since in 5d bulk the fermion representation is vectorial , the two - component spinor @xmath26 that generates 4d supersymmetry will in 5d be accompanied by its chiral conjugate mirror @xmath27 . thus a @xmath2 supersymmetry in 5d corresponds to two different @xmath2 supersymmetry , or equivalently , a @xmath1 supersymmetry from a 4d perspective . in fact , all members of a given kk mode must fall into a valid representation of @xmath1 supersymmetry . in fact , each 4d supermultiplet is augmented by new chiral conjugate states and together they form a hypermultiplet . we are talking about a massive representation of supersymmetry , where the supersymmetry preserving dirac mass plays the rle of central charge for @xmath1 supersymmetry . this charge is not renormalized , as a consequence of which the bulk hypermultiplets do not receive any wave - function renormalization supersymmetry it turns out that @xmath28 , which is analogous to @xmath29 for @xmath30 supersymmetry . here @xmath31 is the dirac mass ( central charge ) and @xmath32 is gauge coupling , while @xmath12 and @xmath33 are labels for renormalized and bare quantities . since the dirac mass of @xmath1 hypermultiplets appears on the right - hand side of the anti - commutation relation of the conserved supersymmetry charges , this mass can not be renormalized . this is intertwined with the observation that only those terms are renormalized which can be written as integrals over all superspace volume . the kinetic term of @xmath1 hypermultiplets can not be written as any such integral ( see discussions and related earlier references in @xcite . ] @xcite . we observe that this @xmath1 non - renormalization has serious numerical consequences in rg evolution of parameters . the most notable effect is the blowing up of the yukawa couplings into the non - perturbative regime around 18 tev , which we will take to be the cutoff of our theory . this is below the scale of perturbative gauge coupling unification , which is around 30 tev . recall that in 5d we encounter power law running which results in early ( compared to 4d ) unification . we allow the gauge and the higgs multiplets access the 5d bulk . thus far what we said is nothing but a supersymmetrization of ued . only the matter multiplets make the difference . in the ued framework , _ all _ sm particles access the bulk , and thus even though there are two fixed points , there is no brane . one could as well have kept some or all of the fermion generations in a brane at a fixed point ; the difference would be that the scenario would cease to be universal . in the present supersymmetric context too we have the freedom of keeping some or all of the matter multiplets at an orbifold fixed point . we note that unless we confine at least two generations of matter multiplets on a brane , the requirement of _ perturbative _ gauge coupling unification leads to a constraint @xmath34 gev @xcite , spoiling its relevance for lhc . on the other hand , unless we keep the third family of matter multiplet in the bulk we can not ensure electroweak breaking . in view of the above , we let the third generation matter multiplet access the bulk , but fix the first two generations at the @xmath35 brane . @xmath1 supersymmetry forbids yukawa interaction in the 5d bulk as this interaction involves odd ( three ) number of _ chiral _ multiplets . therefore , we localize yukawa interaction at the orbifold fixed point where the supersymmetry corresponds to @xmath2 . 7 . now we come to the important question as how we break the residual @xmath2 supersymmetry . different ideas have been advanced for its realization . one way is to break it by the scherk - schwarz mechanism @xcite in which fermions and bosons satisfy different periodic conditions over the compactified dimension . explicit realizations towards this using a tev - scale orbifold can be found in @xcite . another interesting approach was to break the residual supersymmetry by a second compactification on an orbifold with two reflection symmetries , viz . @xmath36 @xcite . this can be viewed as a discrete version of the scherk - schwarz mechanism . both these scenarios yield soft masses which are uv insensitive due to the non - local nature of supersymmetry breaking . from a completely different viewpoint , supersymmetry breaking may be infused from the brane - bulk interface @xcite , or transmitted from a distant brane @xcite , or arisen from a gaugino mediation set - up @xcite ( possibly with a much lower cutoff than @xmath37 gev ) , or triggered by some completely unknown brane dynamics , for example , by a spurion @xmath38-term vacuum expectation value ( vev ) . + in the context of the present analysis , we keep the exact mechanism of the @xmath2 brane supersymmetry breaking _ unspecified_. we assume that the supersymmetry breaking scale is of the order of the inverse radius of compactification , for example @xmath39 , where @xmath40 is an @xmath41 dimensionless parameter . our main goal is the following _ : just like in the conventional but constrained version of 4d supersymmetry one starts with a common scalar and a common gaugino mass at high scale ( e.g. the gut scale ) and then run them down using the mssm beta functions to find the weak scale spectrum , we do exactly the same here by assuming a common scalar mass ( @xmath5 ) and a common gaugino mass ( @xmath6 ) at low cutoff scale ( 18 tev ) and follow the running using the kk beta functions through successive kk thresholds to obtain the weak scale parameters . by adopting a phenomenological approach , we scan @xmath5 and @xmath6 over a set of values @xmath39 , with @xmath40 varying in the range @xmath42 to @xmath43 and @xmath19 fixed at @xmath43 tev . as mentioned in the introduction , from a 4d perspective the kk towers of matter and gauge fields rearrange in the form of @xmath1 hypermultiplets . a judicious choice of @xmath44 parity of the 5d fields allows us to break the @xmath1 supersymmetry to @xmath2 supersymmetry . we briefly review below the multiplet structures of the fields following the prescription suggested in @xcite . the 5d super yang - mills theory contains a 5-vector gauge field , a 4-component dirac gaugino and a real scalar . when dimensionally reduced to 4d , the gauge field splits into a 4-vector and a scalar , the gaugino splits into 2 majorana gauginos , and we still have the real scalar previously mentioned . all these fit into a vector multiplet and a chiral multiplet in @xmath2 language . if we represent the @xmath1 vector supermultiplet by @xmath45 , the 4-vector gauge field by @xmath46 , the gauginos by @xmath47 and @xmath48 , and define a complex scalar field @xmath49 , where @xmath50 is the 5d real scalar and @xmath51 is the 5th component of the 5-vector field , then one can schematically represent the 5d vector supermultiplet as @xmath52 from a 4d perspective ( where the compactified 5th coordinate @xmath53 is just a label ) , and in the @xmath2 language , one can visualize the vector hypermultiplet by a vector multiplet @xmath54 ( first column ) and a chiral multiplet in the adjoint representation by @xmath55 ( second column ) : @xmath56 the @xmath44 parity of @xmath45 is so chosen that the @xmath54 contains a zero mode , but @xmath55 does not have any zero mode . the gauge invariant action may be written as ( @xmath57 ) @xmath58,\ ] ] where the @xmath59 is the field strength superfield corresponding to @xmath60 . from the @xmath2 perspective , the @xmath1 hypermultiplet splits into two chiral multiplets . thus have a @xmath61 hypermultiplet and a @xmath62 hypermultiplet . we can represent them as ( the tilde symbol represents higgsino ) @xmath63 if we denote the two chiral multiplets inside the hypermultiplet @xmath64 as @xmath65 in left column and @xmath66 in right column , then one can expand the chiral superfield as @xmath67 we assign even @xmath44 parity to @xmath68 so that it has a zero mode , and odd @xmath44 parity to @xmath69 which does not have zero mode . the free action of the hypermultiplets can be written as @xmath70 \ , . \label{higgact}\ ] ] matters have similar hypermultiplet structures similar to higgs : @xmath71 where , @xmath72 ( @xmath44 even ) and @xmath73 ( @xmath44 odd ) represent the two @xmath2 chiral multiplets . the free matter hypermultiplet action will be similar to eq . ( [ higgact ] ) . there are five matter representations , two su(2 ) doublets @xmath26 and @xmath74 and three singlets @xmath75 , where the symbols have their standard meaning . when the hypermultiplets are charged under gauge symmetry , their free action can be promoted to take care of the interaction in the following way : @xmath76,\end{aligned}\ ] ] where , @xmath77 and @xmath78 are lie - algebra - valued gauge and matter superfields . we denote the yukawa part of the superpotential by @xmath79 , which contains the usual chiral superfield combinations @xmath80 , @xmath81 and @xmath82 . since an yukawa interaction involves three ( i.e. odd number ) chiral superfields , it is not possible to write a bulk yukawa interaction maintaining @xmath1 supersymmetry . for this reason , we confine yukawa interaction at the branes even though the associated superfields are in the bulk . a generic action ( e.g. involving the third generation bulk matter ) can be written as @xmath83 \ , .\ ] ] as the @xmath44 odd fields vanish at the fixed points , they do not contribute to yukawa interactions . we note that the first two matter generations can be kept either at either of the two fixed points . also , the above yukawa action could have been confined at either of the two branes instead of symmetrically on both . in order to obtain the action in terms of 4d component fields , we need to write down the kk decomposition of the 5d fields in terms of zero modes and higher kk modes @xcite . each 5d field is either @xmath44 even or @xmath44 odd . only the even fields have zero modes . the decomposition of the higgs fields will be exactly like the matter fields . @xmath84 the technical meaning of rg evolution in a higher dimensional context has been amply clarified in @xcite , and we merely reiterate it in the present context . the multiplicity of kk states renders any such higher dimensional scenario non - renormalizable . so ` running ' of couplings or parameters with the energy scale does not make much of a sense . rather , one can estimate the finite quantum corrections that these couplings / parameters receive whose size depends on some explicit cutoff @xmath85 . the contribution comes from @xmath86 number of kk states which lie between the scale of the first kk state , which is @xmath87 , and the cutoff @xmath85 . with this interpretation of rg running , we compute the one loop beta functions of the gauge and yukawa couplings and various soft supersymmetry breaking masses . we make the following observations : 1 . the contribution to the beta function from a given kk mode does not depend on its kk label . when we consider different kk thresholds we neglect their zero mode masses , i.e. we assume that the @xmath22th level kk state is kicked into life when we cross the energy scale @xmath88 . 3 . as we cross different kk thresholds , the beta functions also change . the beta function of the quantity @xmath89 at an energy scale @xmath26 , where @xmath90 , can be written as ( @xmath91 , where @xmath92 is a reference scale , e.g. the electroweak scale ) @xmath93 here @xmath94 is the contribution induced by the zero mode ( i.e. ordinary 4d mssm ) states ( which may be found , for example , in the review @xcite ) and @xmath95 arises from a single kk mode . ( [ master ] ) is our master equation using which we perform a diagram by diagram calculation for the estimation of @xmath95 for various couplings and parameters . the running of the gauge couplings ( @xmath96 ) and gaugino masses ( @xmath97 ) are controlled by @xmath98 \ , , ~~ \beta_{m_i } = \frac{g_i^2 m_i}{16\pi^2 } \left[b_i^0 + n \tilde{b}_i \right ] \ , .\end{aligned}\ ] ] for the gauge groups u(1 ) ( which cooresponds to @xmath99 , which unifies ) , su(2 ) and su(3 ) , @xmath100 @xcite , and @xmath101 @xcite , respectively . we recall that @xmath2 non - renormalization relates the beta functions of the yukawa couplings ( @xmath102 ) to the anomalous dimension matrices ( @xmath103 ) of the superfields . this theorem implies that logarithmically divergent contributions can always be written in terms of wave - function renormalizations . generically , @xmath102 may be written as @xmath104 the feynman diagrams showing the kk contributions to the wave - function renormalizations of the scalars and fermions are displayed in fig . [ selfenergies ] . the contribution from the gauge sector cancels exactly as a consequence of the @xmath1 non - renormalization theorem mentioned in section [ 5dmssm ] . diagrammatically , the origin of this cancellation may be traced to a relative sign between the @xmath105- and @xmath106-propagators - see eq . ( [ 5dv ] ) . only the brane localized yukawa interactions contribute to the yukawa evolution . we also keep track of the fact that the @xmath44 odd fields have vanishing wave - functions at the two branes , leaving the even fields alone to contribute to the diagrams in fig . [ selfenergies ] . here we have made a tacit assumption that although the yukawa interaction is brane localized , only one kk level ( @xmath22 ) states float inside the loop at a time . this is a technical assumption to ensure calculability by avoiding kk divergence which would have arisen while summing more than one kk index in a loop calculation . to appreciate the numerical impact of the bulk @xmath1 non - renormalization , we first write down the conventional 4d mssm beta functions ( i.e. those coming from zero mode states in the 5d context ) which contribute to the evolution of the third generation yukawa couplings @xcite : @xmath107,\nonumber \\ \beta_{b}^0 & = & { y_b \over 16 \pi^2 } \bigl [ 6 y_b^ * y_b + y_t^ * y_t + y_\tau^ * y_\tau - { 16\over 3 } g_3 ^ 2 - 3 g_2 ^ 2 - { 7\over 15 } g_1 ^ 2 \bigr ] \ , , \\ \beta_{\tau}^0 & = & { y_\tau \over 16 \pi^2 } \bigl [ 4 y_\tau^ * y_\tau + 3 y_b^ * y_b - 3 g_2 ^ 2 - { 9\over 5 } g_1 ^ 2 \bigr ] \ , . \nonumber \end{aligned}\ ] ] the corresponding kk contributions are given by @xmath108 where the vanishing gauge contributions is a direct consequence of the bulk @xmath1 non - renormalization . the effects of the above non - renormalization can also be felt in the evolution of the trilinear scalar couplings . the relevant feynman diagrams are displayed in fig . [ trilinear ] . again , for illustration , we first write down the contributions to the beta functions from the zero mode ( i.e. 4d mssm ) states @xcite : @xmath109 \ , , \nonumber \\ { \beta}_{a_b}^0 \!&=&\!\frac{1}{16\pi^2}\bigl [ a_b \left ( 18 y_b^ * y_b + y_t^ * y_t + y_\tau^ * y_\tau - { 16\over 3 } g_3 ^ 2 - 3 g_2 ^ 2 - { 7\over 15 } g_1 ^ 2 \right ) + 2 a_t y_t^ * y_b + 2 a_\tau y_\tau^ * y_b \bigr . \phantom{xxxx } \nonumber \\ & & \!+ y_b \bigl . \left ( { 32\over 3 } g_3 ^ 2 m_3 + 6 g_2 ^ 2 m_2 + { 14 \over 15 } g_1 ^ 2 m_1 \right ) \bigr ] \ , , \qquad { } \\ \beta_{a_{\tau}}^0 \!&=&\ ! \frac{1}{16\pi^2}\bigl [ a_\tau \left ( 12 y_\tau^ * y_\tau + 3 y_b^ * y_b - 3 g_2 ^ 2 - { 9\over 5 } g_1 ^ 2 \right ) + 6 a_b y_b^ * y_\tau + y_\tau \left ( 6 g_2 ^ 2 m_2 + { 18\over 5 } g_1 ^ 2 m_1 \right ) \bigr ] \ , . \nonumber\end{aligned}\ ] ] as expected , the beta functions of the _ soft supersymmetry breaking parameters _ are proportional _ not only _ to those parameters but to others as well , since any non - renormalization theorem ceases to work when supersymmetry is broken . for the computation of @xmath110 , we need to keep in mind the essence of eq . ( [ betaf ] ) , i.e. the _ absence of gauge contributions _ in @xmath111 , while solving the coupled differential equations . however , that part of the gauge contributions ( proportional to @xmath112 ) to the trilinear scalar couplings which multiply the gaugino masses ( @xmath97 ) in eq . ( [ atrge ] ) would still remain while computing the kk contribution . all in all , @xmath113 we make three observations regarding the kk contributions to the evolution of scalar masses ( see fig . [ 3higg ] ) : 1 . the two diagrams in the lower row of fig . [ 3higg ] depend on the yukawa couplings . hence , they are important only for the third generation matter fields . 2 . recall that in the evolution of the yukawa couplings the kk contributions from the gauge field @xmath105 and the complex scalar @xmath106 exactly cancelled thanks to the bulk @xmath1 non - renormalization . however , their fermionic superpartners contribute to the scalar mass evolution and those contributions add up instead of canceling out . this happens because these contributions yield gaugino masses which are @xmath2 supersymmetry breaking parameters and hence the non - renormalization theorem ceases to be applicable . 3 . each kk state in the two diagrams in the top row of fig . [ 3higg ] contributes twice that of the sm because of the doubling of the fermions ( this factor of 2 is highlighted in bold - face in eqs . ( [ scalarbeta ] ) and ( [ higgsbeta ] ) below , and also marked on the two diagrams in top row in fig . [ 3higg ] ) ) . however , each kk state at the lower row diagrams contributes the same as in the sm because the odd fermion modes vanish at the brane where yukawa interaction is confined . the zero mode contributions to the beta functions can be found in @xcite . below we display the kk contributions to the beta functions only for the third generation scalars in external lines ( @xmath114 $ ] , where the index @xmath115 runs over the third generation matter states which are in the bulk ) : @xmath116 , \nonumber \\ \tilde{\beta}_{d_3 } & = & \frac{1}{16\pi^2 } \left [ 2\left ( 2y_b^2 \left ( m_{hd}^2 + m_{\tilde{q}_3}^2 + m_{\tilde{d}_3}^2 \right ) + 2a_b^2 \right ) \right . - \left . { \bf 2 } \left ( \frac{32}{3}g_3 ^ 2|m_3|^2 + \frac{8}{15}g_1 ^ 2|m_1|^2 - \frac{2}{5}g_1 ^ 2 s \right ) \right ] , \nonumber \\ \tilde{\beta}_{q_3 } & = & \frac{1}{16\pi^2 } \left [ \left(2 y_t^2 \left ( m_{hu}^2 + m_{\tilde{q}_3}^2 + m_{\tilde{u}_3}^2 \right ) + 2a_t^2 \right ) \right . + \left . \left ( 2y_b^2 \left ( m_{hd}^2 + m_{\tilde{q}_3}^2 + m_{\tilde{d}_3}^2 \right ) + 2a_b^2 \right)\right . \nonumber\\ & & \left . - { \bf 2 } \left ( \frac{32}{3}g_3 ^ 2|m_3|^2 + 6g_2 ^ 2|m_2|^2 + \frac{2}{15}g_1 ^ 2|m_1|^2 - \frac{1}{5}g_1 ^ 2 s \right ) \right ] , \\ \tilde{\beta}_{l_3 } & = & \frac{1}{16\pi^2 } \left [ \left ( 2y_{\tau}^2 \left ( m_{hd}^2 + m_{\tilde{l}_3}^2 + m_{\tilde{e}_3}^2 \right ) + 2a_{\tau}^2 \right ) - { \bf 2 } \left ( \frac{6}{5}g_1 ^ 2|m_1|^2 + \frac{3}{5}g_1 ^ 2 s \right ) \right],\nonumber \\ \tilde{\beta}_{e_3 } & = & \frac{1}{16\pi^2 } \left [ 2\left ( 2y_{\tau}^2 \left ( m_{hd}^2 + m_{\tilde{l}_3}^2 + m_{\tilde{e}_3}^2 \right ) + 2a_{\tau}^2 \right ) - { \bf 2 } \left ( \frac{24}{5}g_1 ^ 2|m_1|^2 - \frac{6}{5}g_1 ^ 2 s \right ) \right ] \ , . \nonumber\end{aligned}\ ] ] the beta functions for the higgs scalars are given by @xmath117 \ , , \nonumber \\ \tilde{\beta}_{h_d } & = & \frac{1}{16\pi^2 } \left [ 3 \left ( 2y_b^2 \left ( m_{hd}^2 + m_{\tilde{q}_3}^2 + m_{\tilde{d}_3}^2 \right ) + 2a_b^2 \right ) \right.+ \left ( 2y_{\tau}^2 \left ( m_{hd}^2 + m_{\tilde{l}_3}^2 + m_{\tilde{e}_3}^2 \right ) + 2a_{\tau}^2 \right ) \\ & & - \left . { \bf 2 } \left ( 6g_2 ^ 2|m_2|^2 + \frac{6}{5}g_1 ^ 2|m_1|^2 + \frac{3}{5}g_1 ^ 2 s \right ) \right ] \nonumber \ , . \label{higgsrg}\end{aligned}\ ] ] note that for the first and second generation soft scalar masses only the second diagram in the top row of fig . [ 3higg ] would contribute . in this section , we highlight the special features of rg evolution in the 5d scenario . we also compare and contrast them with the 4d features . for all our numerical estimates we have fixed @xmath118 tev . the power law running of the gauge and yukawa couplings has been discussed in @xcite for the non - supersymmetric scenario and in @xcite for the supersymmetric case . as far as the higgs multiplets are concerned , there is a crucial difference between our model and that considered in @xcite . in our scenario there are separate up- and down - type higgs hypermultiplets - see eq . ( [ huhd - hyper ] ) . inside each hypermultiplet only the left column with label ( @xmath74 ) is @xmath44 even and its scalar zero mode receives a vev , whereas the right column with label ( @xmath12 ) is projected out . in other words , the hypermultiplet @xmath61 contains the vev @xmath119 and , similarly , @xmath62 contains @xmath120 . on the other hand , @xcite contains a single hypermultiplet , each column of which has a zero mode , one to be identified with the up - type chiral multiplet which contains the vev @xmath119 , and the other to be identified with the down - type containing @xmath120 . while our approach constitutes a straightforward generalization of @xmath61 and @xmath62 from chiral multiplets to hypermultiplets , the choice made in @xcite requires non - trivial boundary conditions . these two different assumptions lead to significant numerical differences . in our approach , the gauge couplings converge to one another but actually do not meet at a single point , while in @xcite the gauge couplings do meet at a point . the difference in the number of kk scalar excitations makes the difference between the two approaches . indeed , both gauge and yukawa couplings exhibit power law running due to summation over the kk states as one crosses the energy thresholds . as we have mentioned in section [ 5dmssm ] , keeping the first two matter generations confined at the brane ensures that the couplings remain perturbative even with @xmath19 as low as 1 tev . starting from their lep - measured values at the weak scale , as we extrapolate the gauge couplings using the kk beta functions we observe that the three couplings approach very close to one another near 32 tev , but they do not actually meet at any point , as mentioned in the previous paragraph . a crucial point of immense numerical significance is that on account of the special @xmath1 bulk non - renormalization , the third generation matter hypermultiplet kept in the bulk does not receive any wave - function renormalization from the gauge hypermultiplet , which we have illustrated below eq . ( [ nonrenorm ] ) . as an important consequence of this the yukawa coupling blows up to large ( non - perturbative ) values around @xmath121 18 tev , which we therefore take to be the effective cutoff of our theory . we assume that at the highest scale @xmath122 tev of our theory , i.e. just before the yukawa couplings blow up , all scalar masses unify to @xmath5 and all gaugino masses to @xmath6 . our high scale parameters are then @xmath123 and @xmath124 . the gaugino mass running is governed by the evolution of gauge couplings . since gauge couplings _ nearly _ meet around 32 tev , the gaugino masses tend to converge also at that scale . but in the present context , as mentioned before , we forced the gaugino masses to unify at 18 tev . recall that in 5d the running is short but fast ( power law ) , but in 4d it is long and slow ( logarithmic ) . this leads to a general expectation that starting from a given high scale value , the low scale predictions would be similar in 4d and 5d . but since we forcibly unified the gaugino masses in our set - up , earlier than otherwise expected , we obtain a somewhat different set of low scale values . the gaugino mass scaling in 5d is shown in fig . [ gauge ] , while in the inset is displayed the 4d running . a rough comparison of the weak scale ratios of the three gaugino masses is the following : @xmath125 if @xmath12-parity remains conserved , the lightest neutralino remains the lightest supersymmetric particle ( lsp ) , only that its mass is heavier than what is expected in the standard 4d scenario - see eq . ( [ gauginom ] ) . fig . [ rewsb ] shows the running of the soft scalar masses . the large top quark yukawa coupling continues to play a crucial rle as in 4d . a rough comparison of the weak scale predictions in 4d and 5d is : @xmath126 even for the brane localized scalars , the 5d model predicts slightly higher weak scale masses compared to 4d supersymmetry breaking is induced by brane - localized spurion vev , the structure of the third generation soft masses may , in principle , be different from what we assumed due to kk mixing . for calculational simplicity we have ignored kk mixing which , we believe , would not affect our main qualitative features . ] . the relatively _ compressed _ spectrum in 5d , as expressed through eqs . ( [ gauginom ] ) and ( [ scalarm ] ) , is indicative of a lesser fine - tuning compared to the conventional mssm . during power law running we ensure that radiative breaking of electroweak symmetry does happen at the desired scale . just like in 4d , only @xmath127 turns negative while all other scalars remain positive . again , the large top quark yukawa coupling drives this phase transition . a point to note is that _ unless _ we keep the third generation matter in the bulk , electroweak symmetry would never break radiatively in our class of models . furthermore , we had to take in 5d a factor 1 to 3 larger ( than 4d ) value of @xmath7 at the cutoff scale . for our numerical estimates we go through the following steps : 1 . we scan @xmath5 and @xmath6 through the range @xmath129/r$ ] . we fix @xmath130 and take both positive and negative values of @xmath7 . our cutoff is 18 tev , which is the scale where the yukawa couplings blow up to non - perturbative values . we use one loop rg equations as displayed in section [ beta ] . to obtain the weak scale parameters we follow the logarithmic evolution like in the 4d desert , but now using different beta functions for different energy intervals separated by successive kk thresholds - see eq . ( [ master ] ) . so , operationally , power law running is treated by taking _ many _ logs . 2 . for each input combination , we perform a consistency check to ensure correct electroweak symmetry breaking , and accept only those inputs which admit this phenomenon . we then feed the weak scale spectrum into the code micromegas @xcite , and using this software package calculate the dark matter density ( @xmath131 ) , @xmath132 , @xmath133 , and @xmath134 . since we consider @xmath118 tev , even the lightest kk states are somewhat heavier than the lighter section of the zero mode spectra . as a result , without any significant numerical compromise we neglect the direct loop contributions of the virtual kk particles , @xmath135 , @xmath136 , @xmath137@xmath138 are all finite _ upto one loop _ in ued or its supersymmetric version thanks to the kk number conservation at the _ tree _ level lagrangian , in scenarios containing brane - bound fermions one encounters a uv sensitivity even at one loop @xcite . a sensible uv completion is needed to counter this . see also @xcite and footnote 2 of this paper . ] . in other words , the kk effects feed into the calculation of low energy spectra _ via _ power law running , but after that we rely on the standard 4d computations encoded in micromegas . this approximation is good enough for our purpose . we compare the predictions of the above observables with their experimental values / constraints , and translate the information into the inclusion / exclusion plots ( figs . [ momh1 ] and [ momh2 ] ) in the @xmath139 plane . the 4d plots have been reproduced to serve as a guide to the eyes for capturing the 5d subtleties . we note that our 4d plots are in agreement with the ones in the existing literature , e.g. with @xcite . we highlight only the major differences between the 4d and 5d models that appear on the @xmath5@xmath6 plane . 1 . we assume that @xmath12-parity is conserved . in the 4d scenario the lightest neutralino is the most likely candidate for an lsp . in the 5d model the situation is somewhat tricky . indeed , the 4d lsp is still an lsp here which is the zero mode neutralino . besides , _ if _ the kk parity remains conserved , then the @xmath140 mode of photon tower , namely @xmath141 , and its supersymmetric partner @xmath142 are also stable dark matter candidates . however , the kk parity is unlikely to respected by the brane - bulk interaction . in our numerical analysis , we have treated the zero mode lsp as the dark matter candidate . we have taken a @xmath143 range of the five year average of wmap dark matter density ( @xmath144 ) @xcite . we raise a caution here that _ if _ kk parity remains conserved and we have two _ more _ dark matter candidates , as mentioned above , then the edge of the allowed band arising from the lower limit of @xmath131 would be further stretched . note further that in the 5d case there is a slight broadening of the wmap allowed strip compared to 4d . this happens because of a combined effect of eqs . ( [ gauginom ] ) and ( [ scalarm ] ) leading to a reduced sensitivity to @xmath6 variation . the region where the lightest neutralino satisfies the dark matter constraints extends to a _ higher _ value of @xmath6 in 5d compared to 4d . what matters most in this context is the mass difference between the chargino and the lightest neutralino , i.e. @xmath145 . it is clear from eq . ( [ gauginom ] ) that starting from a given @xmath6 this difference is smaller in 5d than in 4d . this explains the shift of the band to the right side . we have taken @xmath146 @xcite , and @xmath147 @xcite . there is nothing much to distinguish between 4d and 5d from these two observables . we have _ not _ included the _ direct _ loop effects of the virtual kk states for any of the weak scale observables . for @xmath148 tev or more , for processes like muon anomalous magnetic moment or @xmath149 , such effects are numerically negligible , but _ only _ for the higgs mass it makes a difference . in figs . [ momh1 ] and [ momh2 ] , the entire region to the left of the line marked with @xmath150 gev which looks otherwise disfavored will be resurrected if we include the kk loop correction to the higgs mass @xcite . we reiterate once again that supersymmetry and extra dimension are interestingly intertwined both from theoretical and phenomenological perspectives . the presence of extra dimensions is an essential part of any high scale fundamental theory , and supersymmetry is quite often an integral component of such theories . supersymmetry , on the other hand , provides a sensible uv completion of the extra dimensional models . the connection between supersymmetry and extra dimension is further strengthened by the consideration that the origin of supersymmetry breaking may reside in extra dimension . in a general class of scenarios in which tev - size soft masses are generated at a relatively low cutoff scale ( like 20 - 30 tev ) with intermediate resonances , one can find a common ground where our analysis will be applicable . the logarithmic running in 4d from 100 gev to @xmath37 gev is replaced in 5d by fast power law running on a shorter interval from 100 gev to about 30 tev thanks to the kk states . this is a feature of extra dimension . what is technically / operationally special about 5d supersymmetry is that it provides a special @xmath1 non - renormalization that forces us to consider an early cutoff ( @xmath151 18 tev ) . the constraints in the @xmath5@xmath6 plane have been placed for the _ first time _ in this paper . the ratio @xmath152 is higher in 5d compared to 4d . for this reason the _ allowed _ region in the 5d plot extends to _ larger _ values of @xmath6 compared to the 4d plot . it should be noted that @xmath19 could in principle be much smaller , e.g. a few hundred gev , than our reference value of 1 tev . in fact , the first kk excitations of sm particles could very well be lighter than zero mode superpartners . it is neither our intention , nor it is possible within the scope of the present paper , to distinguish the two different options . two issues require further studies beyond the scope of the present paper : ( i ) besides the lightest neutralino ( the usual 4d lsp ) , _ if _ kk parity remains conserved , there may be two other candidates of dark matter in this model . one is @xmath141 , the @xmath140 level photon , and the other is its superpartner @xmath142 . although brane - bulk interface is likely to frustrate kk parity conservation , yet for the sake of completeness one should study the _ combined effects _ of all three dark matter candidates . ( ii ) it will also be interesting to revisit the lower limit on @xmath19 in a _ scenario , which we suspect would be relaxed . * acknowledgments : * we thank e. dudas , s. pokorski and a. raychaudhuri for making valuable comments related to our work . gb acknowledges hospitality of the physics department , university of warsaw , during the final phase of the work . gb s work has been supported in part by the project no . 2007/37/9/brns of brns ( dae ) , india . tsr thanks hri , allahabad , for hospitality during the early phase of this work . tsr also acknowledges the s.p . mukherjee fellowship of csir , india . 99 n. arkani - hamed , s. dimopoulos and g. r. dvali , `` the hierarchy problem and new dimensions at a millimeter , '' phys . b * 429 * ( 1998 ) 263 [ arxiv : hep - ph/9803315 ] ; + i. antoniadis , n. arkani - hamed , s. dimopoulos and g. r. dvali , `` new dimensions at a millimeter to a fermi and superstrings at a tev , '' phys . b * 436 * ( 1998 ) 257 [ arxiv : hep - ph/9804398 ] ; + n. arkani - hamed , s. dimopoulos and g. r. dvali , `` phenomenology , astrophysics and cosmology of theories with sub - millimeter dimensions and tev scale quantum gravity , '' phys . d * 59 * ( 1999 ) 086004 [ arxiv : hep - ph/9807344 ] . a. j. buras , m. spranger and a. weiler , `` the impact of universal extra dimensions on the unitarity triangle and rare k and b decays . ( ( u ) ) , '' nucl . b * 660 * ( 2003 ) 225 [ arxiv : hep - ph/0212143 ] ; + a. j. buras , a. poschenrieder , m. spranger and a. weiler , `` the impact of universal extra dimensions on b @xmath153 x / s gamma , b @xmath153 x / s gluon , b @xmath153 x / s mu+ mu- , k(l ) @xmath153 pi0 e+ e- , and epsilon/epsilon , '' nucl . b * 678 * ( 2004 ) 455 [ arxiv : hep - ph/0306158 ] . j. f. oliver , j. papavassiliou and a. santamaria , `` universal extra dimensions and z @xmath153 b anti - b , '' phys . d * 67 * ( 2003 ) 056002 [ arxiv : hep - ph/0212391 ] . t. appelquist and h. u. yee , `` universal extra dimensions and the higgs boson mass , '' phys . rev . d * 67 * ( 2003 ) 055002 [ arxiv : hep - ph/0211023 ] . i. antoniadis , k. benakli and m. quiros , `` production of kaluza - klein states at future colliders , '' phys . b * 331 * ( 1994 ) 313 [ arxiv : hep - ph/9403290 ] ; + t. g. rizzo , `` probes of universal extra dimensions at colliders , '' phys . rev . d * 64 * ( 2001 ) 095010 [ arxiv : hep - ph/0106336 ] ; + c. macesanu , c. d. mcmullen and s. nandi , `` collider implications of universal extra dimensions , '' phys . d * 66 * ( 2002 ) 015009 [ arxiv : hep - ph/0201300 ] ; + c. macesanu , c. d. mcmullen and s. nandi , `` new signal for universal extra dimensions , '' phys . b * 546 * ( 2002 ) 253 [ arxiv : hep - ph/0207269 ] . h. c. cheng , `` universal extra dimensions at the @xmath154 colliders , '' int . j. mod . a * 18 * ( 2003 ) 2779 [ arxiv : hep - ph/0206035 ] ; + a. muck , a. pilaftsis and r. ruckl , `` probing minimal 5d extensions of the standard model : from lep to an e+ e- linear collider , '' nucl . phys . b * 687 * ( 2004 ) 55 [ arxiv : hep - ph/0312186 ] . g. bhattacharyya , a. datta , s. k. majee and a. raychaudhuri , `` exploring the universal extra dimension at the lhc , '' nucl . b * 821 * ( 2009 ) 48 [ arxiv:0904.0937 [ hep - ph ] ] ; + d. choudhury , a. datta and k. ghosh , `` deciphering universal extra dimension from the top quark signals at the cern lhc , '' arxiv:0911.4064 [ hep - ph ] . e. accomando , i. antoniadis and k. benakli , `` looking for tev - scale strings and extra - dimensions , '' nucl . phys . b * 579 * ( 2000 ) 3 [ arxiv : hep - ph/9912287 ] . p. nath , y. yamada and m. yamaguchi , `` probing the nature of compactification with kaluza - klein excitations at the large hadron collider , '' phys . b * 466 * ( 1999 ) 100 [ arxiv : hep - ph/9905415 ] ; + m. masip and a. pomarol , `` effects of sm kaluza - klein excitations on electroweak observables , '' phys . d * 60 * ( 1999 ) 096005 [ arxiv : hep - ph/9902467 ] ; + t. g. rizzo and j. d. wells , `` electroweak precision measurements and collider probes of the standard model with large extra dimensions , '' phys . d * 61 * ( 2000 ) 016007 [ arxiv : hep - ph/9906234 ] ; + a. strumia , phys . b * 466 * ( 1999 ) 107 [ arxiv : hep - ph/9906266 ] ; c. d. carone , `` electroweak constraints on extended models with extra dimensions , '' phys . d * 61 * ( 2000 ) 015008 [ arxiv : hep - ph/9907362 ] . g. servant and t. m. p. tait , `` is the lightest kaluza - klein particle a viable dark matter candidate ? , '' nucl . b * 650 * ( 2003 ) 391 [ arxiv : hep - ph/0206071 ] ; + f. burnell and g. d. kribs , `` the abundance of kaluza - klein dark matter with coannihilation , '' phys . d * 73 * ( 2006 ) 015001 [ arxiv : hep - ph/0509118 ] ; + d. hooper and g. d. kribs , `` kaluza - klein dark matter and the positron excess , '' phys . d * 70 * ( 2004 ) 115004 [ arxiv : hep - ph/0406026 ] ; + f. fucito , a. lionetto and m. prisco , `` extra - dimensions and dark matter , '' jcap * 0606 * ( 2006 ) 002 [ arxiv : hep - ph/0603042 ] . n. arkani - hamed , h. c. cheng , b. a. dobrescu and l. j. hall , `` self - breaking of the standard model gauge symmetry , '' phys . d * 62 * ( 2000 ) 096006 [ arxiv : hep - ph/0006238 ] . k. r. dienes , e. dudas and t. gherghetta , `` extra spacetime dimensions and unification , '' phys . lett . b * 436 * ( 1998 ) 55 [ arxiv : hep - ph/9803466 ] . g. bhattacharyya , a. datta , s. k. majee and a. raychaudhuri , `` power law blitzkrieg in universal extra dimension scenario , '' nucl . b * 760 * ( 2007 ) 117 [ arxiv : hep - ph/0608208 ] . a. hebecker and a. westphal , `` power - like threshold corrections to gauge unification in extra dimensions , '' annals phys . * 305 * ( 2003 ) 119 [ arxiv : hep - ph/0212175 ] . g. bhattacharyya , s. goswami and a. raychaudhuri , `` power law enhancement of neutrino mixing angles in extra dimensions , '' phys . d * 66 * ( 2002 ) 033008 [ arxiv : hep - ph/0202147 ] . a. deandrea , j. welzel , p. hosteins and m. oertel , `` quantum corrections to the effective neutrino mass operator in 5d mssm , '' phys . d * 75 * ( 2007 ) 113005 [ arxiv : hep - ph/0611172 ] . b. s. acharya , k. bobkov , g. l. kane , p. kumar and j. shao , `` explaining the electroweak scale and stabilizing moduli in m theory , '' phys . d * 76 * ( 2007 ) 126010 [ arxiv : hep - th/0701034 ] ; + b. s. acharya , k. bobkov , g. kane , p. kumar and d. vaman , `` an m theory solution to the hierarchy problem , '' phys . * 97 * ( 2006 ) 191601 [ arxiv : hep - th/0606262 ] ; + j. j. heckman , c. vafa , h. verlinde and m. wijnholt , `` cascading to the mssm , '' jhep * 0806 * ( 2008 ) 016 [ arxiv:0711.0387 [ hep - ph ] ] ; + i. antoniadis and s. dimopoulos , `` splitting supersymmetry in string theory , '' nucl . b * 715 * ( 2005 ) 120 [ arxiv : hep - th/0411032 ] ; + r. blumenhagen , j. p. conlon , s. krippendorf , s. moster and f. quevedo , `` susy breaking in local string / f - theory models , '' jhep * 0909 * ( 2009 ) 007 [ arxiv:0906.3297 [ hep - th ] ] ; + t. banks and m. dine , `` couplings and scales in strongly coupled heterotic string theory , '' nucl . b * 479 * ( 1996 ) 173 [ arxiv : hep - th/9605136 ] . r. sundrum , `` susy splits , but then returns , '' arxiv:0909.5430 [ hep - th ] ; + t. gherghetta and a. pomarol , `` the standard model partly supersymmetric , '' phys . d * 67 * ( 2003 ) 085018 [ arxiv : hep - ph/0302001 ] . g. bhattacharyya , s. k. majee and a. raychaudhuri , `` extra - dimensional relaxation of the upper limit of the lightest supersymmetric neutral higgs mass , '' nucl . b * 793 * ( 2008 ) 114 [ arxiv:0705.3103 [ hep - ph ] ] . i. antoniadis , c. munoz and m. quiros , `` dynamical supersymmetry breaking with a large internal dimension , '' nucl . b * 397 * ( 1993 ) 515 [ arxiv : hep - ph/9211309 ] ; + k. benakli , `` perturbative supersymmetry breaking in orbifolds with wilson line backgrounds , '' phys . b * 386 * ( 1996 ) 106 [ arxiv : hep - th/9509115 ] ; + i. antoniadis , s. dimopoulos and g. r. dvali , `` millimeter range forces in superstring theories with weak - scale compactification , '' nucl . b * 516 * ( 1998 ) 70 [ arxiv : hep - ph/9710204 ] . j. scherk and j. h. schwarz , `` how to get masses from extra dimensions , '' nucl . b * 153 * ( 1979 ) 61 ; + j. scherk and j. h. schwarz , `` spontaneous breaking of supersymmetry through dimensional reduction , '' phys . b * 82 * ( 1979 ) 60 . a. pomarol and m. quiros , `` the standard model from extra dimensions , '' phys . b * 438 * ( 1998 ) 255 [ arxiv : hep - ph/9806263 ] ; + a. delgado , a. pomarol and m. quiros , `` supersymmetry and electroweak breaking from extra dimensions at the tev - scale , '' phys . d * 60 * ( 1999 ) 095008 [ arxiv : hep - ph/9812489 ] ; + i. antoniadis , s. dimopoulos , a. pomarol and m. quiros , `` soft masses in theories with supersymmetry breaking by tev - compactification , '' nucl . b * 544 * ( 1999 ) 503 [ arxiv : hep - ph/9810410 ] ; + d. diego , g. von gersdorff and m. quiros , `` the mssm from scherk - schwarz supersymmetry breaking , '' phys . d * 74 * ( 2006 ) 055004 [ arxiv : hep - ph/0605024 ] ; + g. von gersdorff , `` the mssm on the interval , '' mod . a * 22 * ( 2007 ) 385 [ arxiv : hep - ph/0701256 ] . r. barbieri , l. j. hall and y. nomura , `` a constrained standard model from a compact extra dimension , '' phys . d * 63 * ( 2001 ) 105007 [ arxiv : hep - ph/0011311 ] . e. a. mirabelli and m. e. peskin , `` transmission of supersymmetry breaking from a 4-dimensional boundary , '' phys . d * 58 * ( 1998 ) 065002 [ arxiv : hep - th/9712214 ] . d. e. kaplan , g. d. kribs and m. schmaltz , `` supersymmetry breaking through transparent extra dimensions , '' phys . d * 62 * ( 2000 ) 035010 [ arxiv : hep - ph/9911293 ] . z. chacko , m. a. luty , a. e. nelson and e. ponton , `` gaugino mediated supersymmetry breaking , '' jhep * 0001 * ( 2000 ) 003 [ arxiv : hep - ph/9911323 ] . n. arkani - hamed , t. gregoire and j. g. wacker , `` higher dimensional supersymmetry in 4d superspace , '' jhep * 0203 * ( 2002 ) 055 [ arxiv : hep - th/0101233 ] . n. arkani - hamed , l. j. hall , y. nomura , d. tucker - smith and n. weiner , `` finite radiative electroweak symmetry breaking from the bulk , '' nucl . b * 605 * ( 2001 ) 81 [ arxiv : hep - ph/0102090 ] . t. kobayashi , j. kubo , m. mondragon and g. zoupanos , `` running of soft parameters in extra space - time dimensions , '' nucl . b * 550 * ( 1999 ) 99 [ arxiv : hep - ph/9812221 ] . g. belanger , f. boudjema , a. pukhov and a. semenov , `` micromegas : a program for calculating the relic density in the mssm , '' comput . commun . * 149 * ( 2002 ) 103 [ arxiv : hep - ph/0112278 ] ; + g. belanger , f. boudjema , a. pukhov and a. semenov , `` micromegas2.0 : a program to calculate the relic density of dark matter in a generic model , '' comput . commun . * 176 * ( 2007 ) 367 [ arxiv : hep - ph/0607059 ] . p. dey and g. bhattacharyya , `` a comparison of ultraviolet sensitivities in universal , nonuniversal , and split extra dimensional models , '' phys . d * 70 * ( 2004 ) 116012 [ arxiv : hep - ph/0407314 ] ; + p. dey and g. bhattacharyya , `` ultraviolet sensitivity of rare decays in nonuniversal extra dimensional models , '' phys . d * 69 * ( 2004 ) 076009 [ arxiv : hep - ph/0309110 ] . e. komatsu _ et al . _ [ wmap collaboration ] , `` five - year wilkinson microwave anisotropy probe observations : cosmological interpretation , '' astrophys . j. suppl . * 180 * ( 2009 ) 330 [ arxiv:0803.0547 [ astro - ph ] ] . e. barberio _ et al . _ [ heavy flavor averaging group ] , `` averages of @xmath155hadron and @xmath156hadron properties at the end of 2007 , '' arxiv:0808.1297 [ hep - ex ] . r. m. carey _ et al . _ , `` the new ( g-2 ) experiment : a proposal to measure the muon anomalous magnetic moment to + -0.14 ppm precision , '' , fermilab - proposal-0989 .

we present an analysis of the dust and gas in andromeda , using _ herschel _ images sampling the entire far - infrared peak . we fit a modified - blackbody model to quasi - independent pixels with spatial resolution ofpc and find that a variable dust - emissivity index ( ) is required to fit the data . we find no significant long - wavelength excess above this model suggesting there is no cold dust component . we show that the gas - to - dust ratio varies radially , increasing from in the center to in the star - forming ring at 10kpc , consistent with the metallicity gradient . in the 10kpc ring the average is.9 , in good agreement with values determined for the milky way ( mw ) . however , in contrast to the mw , we find significant radial variations in , which increases from 1.9 at 10kpc to.5 at a radius of 3.1kpc and then decreases to 1.7 in the center . the dust temperature is fairly constant in the 10kpc ring ( ranging from 1720k ) , but increases strongly in the bulge tok . within 3.1kpc we find the dust temperature is highly correlated with the 3.6 flux , suggesting the general stellar population in the bulge is the dominant source of dust heating there . at larger radii , there is a weak correlation between the star formation rate and dust temperature . we find no evidence for ` dark gas ' in m31 in contrast to recent results for the mw . finally , we obtained an estimate of the co x - factor by minimising the dispersion in the gas - to - dust ratio , obtaining a value of^{-1}$ ] .

1204.0785

astronomy at long infrared wavelengths ( 201000 ) is a relatively young field due to the need for space missions to avoid the absorption of the atmosphere in this waveband . this waveband , however , is vital for astronomical studies as this is where dust in the interstellar medium ( ism ) radiates . this is important for studies of galaxy evolution as star formation regions tend to be dusty , and therefore the use of uv and optical measurements to trace the star formation rate can lead to it being underestimated ( see * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? calibrating the relationship between infrared emission and star formation rate has been difficult due to uncertainties from the contribution of the general stellar population to heating the dust , the fact that not all optical / uv emission is absorbed , and uncertainties in the properties of the dust . the dust emission is affected by the composition of the dust , and the proportion of non - equilibrium to equilibrium heating . studies with previous space missions _ iras _ , _ iso _ , _ spitzer _ and _ akari _ tried to address these questions ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) . however , as they were limited to wavelengths less than 160 , they were not sensitive to the cold dust ( @xmath415k ) and missed up to 50% of the dust mass in galaxies . the continuum emission from the dust has been + + @xmath5__h _ _ esa e p i nasa . + proposed as a potential method of measuring the total mass of the ism @xcite ; traditionally the amount of gas has been measured using hi and co surveys , but due to sensitivity and resolution issues this method is limited to low redshift and small numbers of galaxies . @xcite found that for early - type galaxies ( e / s0 ) the ism was detected for 50% of objects through its dust emission but only 22% through its co emission . in addition , the conversion of the co tracer to molecular gas mass is highly uncertain and is believed to vary with metallicity ( see * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) . this is a topical area as recent studies using the planck all - sky survey and _ iras _ maps have made an estimate of the amount of ` dark gas ' @xcite in the milky way . the ` dark gas ' is thought to be molecular gas which is traced by dust , but not detected with the standard co method . previous works have also suggested the presence of ` dark gas ' by using @xmath6-ray emission from cosmic - ray interactions with clouds of gas @xcite and by the kinematics of recycled dwarf galaxies @xcite . _ herschel _ is one of the european space agency s flagship observatories and observes in the far - infrared ( fir ) in the range 55671 ( see * ? ? ? due to the large space mirror and cryogenic instruments , it has a high sensitivity and unprecedented angular resolution at these wavelengths . _ herschel _ has the ability to target both large numbers of galaxies and map large areas of sky . it has two photometric instruments : pacs @xcite which can observe in 3 broad bands around 70 , 100 and 160 ( 70 and 100 can not be used simultaneously ) and spire @xcite which observes simultaneously in filter bands centered at 250 , 350 and 500 . spire provides flux measurements on the longer wavelength side of the fir dust peak ( @xmath0160 ) , allowing us to obtain a full census of dust in nearby galaxies ( e.g. , * ? ? ? andromeda ( m31 ) and the milky way ( mw ) are the only two large spirals in the local group . studies of andromeda therefore provide the best comparison to observations of our own galaxy with the advantage that we get a ` global ' picture , whereas investigations of the milky way are limited by our location within the galaxy . the total size of m31 and the scalelength of its disk are both approximately twice those of the mw ( see * ? ? ? * and references therein ) . however , the star - formation rate of the mw is @xmath036@xmath7 @xcite compared to only @xmath00.31.0 in m31 ( * ? ? ? * ; * ? ? ? * g. p. ford et al . , in preparation ) , despite similar amounts of gas present @xcite . for this reason , m31 is often labelled as ` quiescent ' . the dust emission from m31 is dominated by a dusty star - forming ring at a radius of 10kpc , and was first observed in the infrared by _ iras _ @xcite . many projects to map the ism in m31 , have been undertaken using observations in the mid - infrared ( mir ) with _ spitzer _ @xcite , in the fir with _ spitzer _ @xcite , in the hi atomic line @xcite and in the co(@xmath8=1 - 0 ) line @xcite . studies of the gas kinematics and dust emission show m31 has a complicated structure ( e.g. , * ? ? ? * ; * ? ? ? * ) ; @xcite attribute many of the features observed to density waves from a possible head - on collision with m32 . @xcite investigated the relation between gas , dust and star formation using fir _ spitzer _ data . @xcite used _ spitzer _ and gas observations to investigate the conversion between co(@xmath8=1 - 0 ) line flux and molecular hydrogen column - density in a sample of nearby galaxies including m31 . in their analysis they found m31 has the lowest dust temperatures in their sample and therefore would benefit most by including _ herschel _ data . the _ herschel _ exploitation of local galaxy andromeda ( helga ) is a survey covering a @xmath9 area centered on m31 . we use pacs - spire parallel mode , observing at 100 , 160 , 250 , 350 and 500 simultaneously . further details of the observations can be found in section [ sec : firobs ] and in @xcite . helga observations have been used in other works to investigate structures in dust and hi at large radii @xcite , the relationship between gas and star formation ( g. p. ford et al . , in preparation ) , and the structure of m31 and the cloud - mass function ( j. kirk et al . , in preparation ) . in this paper we use the _ herschel _ data combined with the wealth of ancillary data to investigate the distribution , emission properties and the processes heating the dust in m31 on spatial scales of @xmath0140kpc . there have still been relatively few attempts to map the dust within a galaxy ; recent studies with _ herschel _ include @xcite , @xcite , @xcite , @xcite , and @xcite , but they are often limited to small numbers of independent pixels . the advantage of m31 over these other studies is the close proximity which allows us to investigate the dust at higher spatial scale and with many more independent pixels . we also apply the _ planck _ method for finding ` dark gas ' @xcite to m31 and use this method to determine a value of the x - factor the relationship between the molecular hydrogen column density and the observed co tracer . in section [ sec : data ] we present the data used for this analysis and section [ sec : firsed ] describes our method for fitting the spectral energy distribution of dust . section [ sec : res ] presents our maps of the dust properties , including the distribution of dust surface density , temperature and spectral index . this section also describes a comparison of the distributions of gas and dust and of the search for an excess emission at long wavelengths . in section [ sec : disc ] we discuss the dust properties including the composition of the dust and the processes heating the dust . in this section we also search for ` dark gas ' and determine the value of the x - factor . the conclusions are presented in section [ sec : conc ] . _ herschel _ observations of m31 were taken using both pacs and spire in parallel mode covering an area of @xmath10 centered on m31 . to observe an area this large , the observations were split into two halves with a cross - scan on each half , which produced data at 100 , 160 , 250 , 350 and 500 simultaneously ( observation ids : 1342211294 , 1342211309 , 1342211319 , 1342213207 ) . full details of the observing strategy and data reduction can be found in @xcite . the pacs data reduction was performed in two stages . the initial processing up to level-1 ( i.e. , to the level where the pointed photometer timelines have been calibrated ) is performed in hipe v8.0 @xcite using the standard pipeline . to remove low - frequency noise ( or drifts ) in the arrays , residual glitches and create the final maps we use scanamorphos ( v15 , * ? ? ? the final maps have a pixel scale of 2 and 3 and a spatial resolution of 12.5 and 13.3 full - width half maximum ( fwhm ) at 100 and 160 respectively . the spire data were processed up to level-1 with a custom pipeline script adapted from the official pipeline . we apply the latest flux correction factors from the spire instrument control centre to update the maps to the latest calibration product @xcite . for the baseline subtraction we use a custom method called brigade ( m. w. l. smith et al . , in preparation ) which uses an alternative technique for correcting temperature drifts . the final maps were created using the naive mapper with pixel sizes of 6 , 8and 12 with spatial resolution of 18.2 , 24.5 , 36.0 fwhm for the 250 , 350 and 500 maps , respectively . all _ herschel _ images are shown in figure [ fig : her - maps ] . in addition to the _ herschel _ data , we make use of the 70 _ spitzer _ mips @xcite map published in @xcite . this observation covers a region of m31 approximately @xmath11 in size and has an exposure time of @xmath12 s pixel@xmath13 . the data was processed using the mips data analysis tool , version 2.90 @xcite and full details of the reduction can be found in @xcite . to investigate the atomic hydrogen in andromeda we use the hi moment - zero map presented in @xcite . the observations were made with the westerbork synthesis radio telescope ( wrst ) covering a region of @xmath14 with a resolution of 18@xmath15 15 . in this work we present our results using a map which has not been corrected for opacity effects since this correction is uncertain . for the results which make use of the hi map we also test how our results are affected by using a hi map corrected for self - opacity using the prescription outlined in @xcite . the uncorrected hi map has a sensitivity of @xmath16 . for the molecular hydrogen gas content we use co(@xmath8=1 - 0 ) observations presented in @xcite made with the iram 30 m telescope . this covers an area of @xmath17 with a sensitivity of @xmath00.35kkms@xmath13 . all maps other than the _ herschel _ images used for this analysis are shown in figure [ fig : oth - maps ] . to investigate how the dust properties vary across m31 we undertake a pixel - by - pixel dust analysis , using the _ spitzer _ 70 , _ herschel _ pacs and spire images . we first convolve the data to the same resolution as the 500map ( our largest fwhm ) by using a kernel which matches the point - spread function ( psf ) in a particular band to the 500 band . this procedure is described in detail in @xcite . the images are then re - binned into 36 pixels , chosen to be about the same size as the 500 beam so that each pixel is approximately independent from each other . for each band we subtract a background value for the whole map , estimated from regions around the galaxy . the uncertainties on the flux in each pixel are found by measuring the standard deviation of the pixels in these background regions and adding this in quadrature with the calibration uncertainty . the flux errors in the majority of pixels are dominated by the calibration uncertainty which we take as 7% for _ spitzer _ 70 @xcite , 10% for pacs ( see * ? ? ? * ) and 7% for spire ( see section [ sect : sedfit ] for more details ) . for each pixel we fit the far - infrared sub - millimetre sed with a one - temperature modified black - body model described by : @xmath18 where @xmath19 is the dust mass with dust temperature @xmath20 , @xmath21 is the planck function and @xmath22 is the distance to the galaxy . @xmath23 is the dust absorption coefficient , described by a power law with dust emissivity index @xmath1 such that @xmath24 . we assume a typical value for the ism of @xmath25 @xcite . while the absolute value of @xmath23 is uncertain , its value will not affect trends with other parameters as it is a fixed scaling constant . the distance to andromeda was taken in this work to be @xmath26 @xcite . we initially used a fixed value of @xmath1 across the whole galaxy , but we found that with a fixed value it was impossible to adequately fit the seds ; we therefore let @xmath1 be a free parameter . to ensure the simplex fitting routine did not get stuck in a local minimum , we ran the sed - fitter in two ways : first with all three parameters ( m@xmath27 , t@xmath27 , @xmath1 ) free to vary ; second by fixing the value of @xmath1 while allowing m@xmath27 and t@xmath27 to vary and repeating the process for all values of @xmath1 in the range 0.20 to 5.90 in 0.01 intervals , selecting the result with the lowest @xmath28 . both methods gave consistent results but we created the final maps of the dust properties by choosing the result with the lowest @xmath28 for each pixel . the spire calibration has an overall systematic uncertainty of 5% due to the uncertainty in the prime calibrator neptune , and a statistical uncertainty of 2% determined from instrumental reproducibility ; the spire observer s manual recommends linearly adding these to give an overall uncertainty of 7% . to implement this in practice in our fitting algorithm we increased the uncorrelated uncertainty to give the same overall uncertainty when the errors are added in quadrature . this is implemented in the sed fitting by using the full covariance matrix in the @xmath28 calculation . we apply our own color correction to the _ herschel _ maps by removing the _ herschel _ pipeline ` k4 ' parameter and then convolving the sed with the filter transmission in the fitting process ( for spire the filters appropriate for extended sources are used ) . this method takes full account of all the wavelength - dependant effects associated with pacs and spire . in previous work @xcite we found that there is a significant contribution from a warmer component of dust at wavelengths @xmath470and so the _ spitzer _ flux at 70 is treated as an upper limit in the fitting ( i.e , if the flux value is higher than the model it does not contribute to @xmath28 ) . the omission of the warm component in the fitting process has a negligible effect on the derived dust mass as the cold component dominates the total dust mass . @xcite suggest that a warmer component could influence the dust emission up to wavelengths of 160 ; to test this we repeated the sed - fitting by treating the _ spitzer _ and pacs fluxes as upper limits and found it made a negligible difference to our results . to estimate the uncertainties on the results of our fits we use a monte - carlo technique . for each pixel we create a set of 1000 artificial seds , created by taking the original flux densities and adding a random value selected from a normal distribution with a mean of zero and a standard deviation equal to the uncertainty in the measured flux ( the correlations between the calibration uncertainties for spire are also included ) . we estimate the error in the derived parameters for each pixel from the 1000 fits . we find that for each pixel there is an uncertainty of 20% in our estimate of the surface density of the dust , of @xmath291.4k in our estimate of the dust temperature , and @xmath290.31 in our estimate of @xmath1 . values from pixels fitted with the modified - blackbody model . the red line represents the 10% significance value for one degree of freedom.,scaledwidth=49.0% ] in producing the final dust mass , temperature and @xmath1 images in this work , we only used pixels where the fluxes in all six bands ( five _ herschel _ & mips 70 ) have a signal - to - noise greater than 5@xmath30 . while this potentially causes us to miss the very coldest dust due to weak emission in the shortest wavelength bands , we choose it as a conservative approach to ensure we have accurate estimates of temperature . in practice it is the 100 map , which has the lowest sensitivity , which limits the number of pixels in our selection . despite this very conservative selection , there are still @xmath31 pixels in our resultant maps . to see if our model is a statistically reasonable fit to the data , we created a histogram of the @xmath28 values for all pixels ( figure [ fig : chisq ] ) . as the 70 flux is used as an upper limit ( see section [ sect : sedfit ] ) and is usually higher than the best - fit model it does not usually contribute to @xmath28 and therefore we only have 1 degree of freedom ( @xmath32 ) . the 10% significance level for @xmath28 occurs at 2.71 which is shown by the red line in figure [ fig : chisq ] . we find 9.8% of our fits have @xmath28 above this level showing our model is an adequate representation of the data . we have also checked to see if our radial gradients in temperature , @xmath1 , and dust surface density are affected by lowering the criteria to include fluxes greater than 3@xmath30 and find no significant changes . recent results from the key insight on nearby galaxies : a far - infrared survey with _ herschel _ ( kingfish ) presented in @xcite have suggested that the one - temperature modified black - body model underestimates the dust mass compared to the @xcite prescription . they attribute this difference to the contribution of warm dust emitting at shorter wavelengths . for our analysis this does not appear to be the case . first when we set the pacs fluxes ( @xmath4160 ) as upper limits there is no significant change in our results . second , if multiple temperature components were present this would bias our @xmath1 values to lower values ( since each temperature component peaks at a different wavelength , * ? ? ? * ) , but we mostly find higher @xmath1 values than expected ( see section [ sec : res ] ) . third the @xmath28 analysis suggests the model is consistent with the data . in addition , @xcite , using a similar analysis , have found in m51 that the dust mass distribution is similar when using the @xcite or the single modified blackbody prescription . we applied the same fitting technique as outlined in section [ sect : sedfit ] to the global flux densities , measured in an elliptical aperture with a semi - major axis of 2.0@xmath33 and semi - minor axis of 0.73@xmath33 . this produces a total dust mass of @xmath34 with a dust temperature of @xmath35k and @xmath36 . the total dust mass from summing all the pixels in our pixel - by - pixel analysis gives a value @xmath37 , a factor of @xmath02.5 lower . this is expected as the fraction of 500 flux in the pixels which satisfy our signal to noise criterion is approximately half the global value . combining our global dust mass with gas measurements , we find a global gas - to - dust mass ratio of 72 . the global temperature is consistent with these for other spiral galaxies obtained using similar methods ( e.g. , * ? ? ? the pixel - by - pixel analysis shows a large range of temperatures and @xmath1 values as discussed in detail in section [ sec : res ] . examples of fits for individual pixels are shown in figure [ fig : exampleseds ] . to help understand the significance of our results and any potential biases or degeneracies in the parameters , we ran a monte - carlo simulation . assuming the dust emits as a single - component modified blackbody , we generated synthetic flux values for a range of temperatures and @xmath1 values , with the same wavelengths as our real data . noise was then added to the simulated fluxes with a value equal to the errors in the real fluxes ( excluding the calibration errors ) for 2000 repetitions per t , @xmath1 combination . the calibration error was not included as it would systematically shift the fluxes for all pixels . we chose an input mass surface density of 0.5@xmath38 as this roughly corresponds to the values found in the 10kpc ring . the quantity that is most important for our work is the dust mass , which from equation [ equ : sed ] is just a multiplicative term . figure [ fig : sim - mass ] shows the mean and median mass returned for the 2000 repetitions as the input temperature is varied between 12 and 30k for a @xmath1 of 2 ; the error bars show the error on the mean . between 15 and 30k , the mean and median dust mass returned matches within the errors the input dust surface density of 0.50 @xmath39 . at dust temperatures of 15k and below , there are large errors on the returned mass , which is due to the fluxes at the pacs wavelengths not reaching the required sensitivity to be included in the fit . for the actual data , we only included pixels in which there is at least a 5@xmath30 detection in all bands , which will avoid the wildly incorrect estimates of the dust mass seen in the simulation . to fully estimate the dust mass and temperature of very cold dust ( t @xmath40 15k ) , we need flux measurements at @xmath41500 . nevertheless , the lack of an excess at 500 ( section [ sec : excess ] ) is circumstantial evidence that andromeda does not contain very cold dust . we investigated if the results in figure [ fig : sim - mass ] were different for a different choice of @xmath1 but found no systematic differences . by plotting the difference in the resultant temperature in each pixel compared to the input model temperature , we find that between 15k and 25k , the temperature uncertainty is @xmath00.6k . above 25k the uncertainty increases , although in the simulation we did not include a 70 point which would likely provide a constraint to our fits if the dust temperature was @xmath4125k . the simulation suggests that for input @xmath1 between 1.5 and 2.4 , the uncertainty in the returned value of @xmath1 for each pixel is @xmath00.1 . as expected these uncertainties are lower than returned by the monte - carlo technique in section [ sect : sedfit ] as we have not included a calibration uncertainty . with @xmath42 modified - blackbody . the mean of the returned masses is shown by the blue points and the median is shown by the black points . the input mass surface density of 0.50 @xmath38 is shown by the red dashed line . to avoid overlapping data points the blue have been shifted by -0.1k and the black by + 0.1k.,scaledwidth=49.0% ] a degeneracy is known to exist between temperature and @xmath1 ; figure [ fig : sim - delt - delt ] shows that if there is an error in one parameter this is anti - correlated with the error in the other parameter . as the distribution is centered on the origin , there is no systematic offset in the returned value of t or @xmath1 . as there is no systematic offset , the error in the mean values over many pixels will therefore be much smaller than the error for a single pixel . this simulation is based on the dust emission arising from a one - component modified blackbody . fitting a one - component modified blackbody to pixels for which there are multiple temperature components would produce a bias towards smaller values of @xmath1 @xcite , although we would hope to detect this by finding high @xmath28 values . in section [ sec : beta - t ] we show that different regions of m31 have different @xmath43 relations and discuss why this is unlikely to be due to multiple temperature components . to summarise , while there is a @xmath1t degeneracy from the fitting algorithm this does not create any systematic offsets in the value returned . if the dust temperature falls below 15k we are unable to constrain the sed due to the lack of data points beyond 500 . and temperature . the uncertainties are taken from the simulated data with @xmath44k and @xmath45 . a clear anti - correlation is observed with the distribution centered on the correct values . to fully populate this graph we increased the simulation to 20,000 repetitions for this @xmath1 , t combination.,scaledwidth=49.0% ] by fitting seds to every pixel , we have created maps of dust surface density , temperature , @xmath1 and gas - to - dust ratio which are shown in figure [ fig : sed - maps ] ( for details on how the gas surface density is calculated see section [ sec : radgd ] ) . the dust surface density distribution , un - surprisingly , is more similar to the maps of gas and star - formation rate than to the 3.6 image ( see figure [ fig : her - maps ] & [ fig : oth - maps ] ) , which traces the stellar mass distribution . the @xmath1 and temperature maps show significant radial variations . figure [ fig : sed - rad ] shows how the dust column density , temperature and @xmath1 vary with radius ( the physical radius is calculated assuming an inclination of 77@xmath46 and pa of 38@xmath46 , @xcite ) . in the star - forming ring at 10kpc , @xmath1 has an average value of @xmath01.8 but increases with decreasing radius reaching a maximum of @xmath02.5 at a radius of @xmath03kpc . this is higher than found in global studies of galaxies @xcite . however similarly high values have been reported recently in @xcite and @xcite for dust within galaxies . the value for the ring agrees well with early results from _ @xcite for dust in the galactic disk and the solar neighbourhood . the 10kpc ring has an average dust surface density of @xmath00.6@xmath47 with a dust temperature of 18k . towards the very center of the galaxy the dust column density decreases to @xmath00.04@xmath47 , @xmath1 values fall to @xmath01.9 and the dust temperature increases to @xmath030k . ) , the dashed green line represents the transition radius ( 3.1kpc ) found when fitting the @xmath1 results.,scaledwidth=49.0% ] . the solid red line represents the best - fit exponential profile to the gas - to - dust ratio.,scaledwidth=49.0% ] in figure [ fig : sed - rad ] we see a clear break in the radial variation for both temperature and @xmath1 at a radius @xmath03kpc . we fit a model with two straight lines and a transition radius ( using a simplex routine ) to the @xmath1 results . the same method is used with the temperature values but we set the transition radius to the value obtained from the fit to the @xmath1 values , which occurs at 3.1kpc ( shown by the dashed green line in figure [ fig : sed - rad ] or black ellipse in figure [ fig : sed - maps ] ) . at radii smaller than the transition radius , the temperature decreases with radius from @xmath030k in the center to @xmath017k , with an associated increase in @xmath1 from @xmath01.8 to @xmath02.5 . at radii larger than the transition radius the dust temperature slowly increases with radius while @xmath1 decreases with radius to @xmath01.7 at 13kpc . the best - fit relationships between @xmath20 , @xmath1 and @xmath48 are shown in figure [ fig : sed - rad ] and listed below : @xmath49 @xmath50 we discuss the cause of the temperature and @xmath1 variations in section [ sec : disc ] . for sections [ sec : heating ] & [ sec : beta - t ] we consider the inner ( @xmath51@xmath403.1@xmath52 ) and outer regions separately . in figure [ fig : gas - rad ] the radial variations of the atomic gas , the molecular gas and the gas - to - dust ratio are shown . we assume an x - factor of @xmath53^{-1}$ ] @xcite to convert a co line flux to a h@xmath54 column density , although this value is notoriously uncertain and has been found to vary with metallicity @xcite . for our analysis of m31 , the choice of x - factor makes very little difference as the total gas is dominated by the atomic phase . out of the 3974 pixels plotted in the gas - to - dust figure , only 101 have a molecular fraction of @xmath4150% and globally the molecular gas only constitutes @xmath07% of the atmoic gas @xcite . to estimate the total gas surface density we include the contribution of the atomic or molecular gas in each pixel if the value is greater than 3@xmath30 in their respective maps . only 86 of our 5@xmath30 dust pixels are not covered by the co(@xmath8=1 - 0 ) observations ; these pixels are in the outskirts of the galaxy and we assume the contribution of molecular gas is negligible . we find a tight relation between the gas - to - dust ratio and radius ( spearman rank coefficient of 0.80 ) as shown in figure [ fig : gas - rad ] , which is described by an exponential profile , shown by the red line . the gas - to - dust ratio increases exponentially from low values of @xmath020 in the center of the galaxy to @xmath0110 at 15kpc typical of the milky way in the local environment ( see * ? ? ? * and references therein ) . note that the values are not corrected for helium in the ism . metallicity gradients have been estimated from oxygen line ratios [ oiii]/h@xmath1 , [ oii]/[oiii ] and r@xmath55 by @xcite . they infer a radially decreasing metallicity gradient of @xmath56 from the r@xmath55 parameter . @xcite calculate oxygen abundance gradients based on a set of 11 hii regions from @xcite and find gradients in the range of -0.027 to -0.013@xmath57 depending on the calibration used . if a constant fraction of the metals in the ism is incorporated into dust grains @xcite , one would expect the gas - to - dust gradient to be @xmath58 metallicity gradient . we find a gas - to - dust gradient of @xmath59 consistent within the uncertainties to the gradient measured by @xcite ( if the hi map corrected for self - opacity is used the gradient slightly increases to @xmath60 ) . this supports the claim that gas can be well traced by dust at a constant metallicity . to see if the uncertainty in the choice of x - factor could affect this result , we carried out the same procedure but limited the analysis to pixels where the molecular hydrogen contribution is less than 10% of the total gas mass . this produced only a small change in the gas - to - dust gradient to @xmath61 which is still consistent with the metallicity gradient . searches for a long - wavelength sub - mm excess ( i.e. @xmath41500 ) has been made in both the milky way @xcite and for nearby galaxies . a sub - mm excess is important as it could suggest very cold dust is present ( @xmath4015k ) which would dominate the dust mass in galaxies . other possible explanations of an excess include variations in the dust emissivity index with wavelength @xcite , different dust populations or contamination from a synchrotron radio source . sub - mm excesses have been reported from observations of low - metallicity dwarf galaxies ( see * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) and spiral galaxies ( see * ? ? ? * ; * ? ? ? most detections have been made by combining fir data with ground - based data at 850 or 1 mm data rather than from data only at @xmath62500 . we searched for a sub - millimetre excess in m31 by comparing the 500 flux to our best - fit models . a 500 excess is defined to be any observed 500 flux that is greater than the expected flux at 500 from our modified blackbody fit to the data . figure [ fig : excess ] is a histogram of the ratio of the excess at 500 and the noise on the 500 map . the distribution of the excess is consistent with a gaussian function with a mean of 0.54@xmath30 and standard deviation of 0.31@xmath30 . the fact the distribution of the histogram is a gaussian suggests that what we are seeing is random noise with a fixed offset ( not centered on zero ) . two non - astronomical sources could explain a fixed offset , either an incorrect background subtraction or a calibration error . the background correction applied is quite small ( 0.2@xmath30 ) , and thus an error in this is unlikely to be the entire explanation of the offset . the distribution of the excess is consistent with the 2% random spire calibration uncertainty , so both factors together could explain the small offset . if the excess is generated from dust with a 10k temperature and @xmath1 of 2 , the dust mass is @xmath010@xmath63@xmath64 in our @xmath415@xmath30 pixels , which corresponds to only 13% of the mass of the warmer dust . if we used a model containing dust at more then one temperature we would not get a useful constraint on the colder dust component without additional data at longer wavelengths . in particular observations at @xmath0850 ( e.g. with scuba2 ) would be useful to determine if any cold dust present . recent studies by @xcite and @xcite have used _ herschel _ colors to confirm previous works ( e.g. , * ? ? ? * ; * ? ? ? * ) that emission from dust in nearby galaxies at wavelengths longer than 160 is mostly from dust heated by the general stellar population . these authors conclude that at wavelengths shorter than 160 , the emission tends to be dominated from warmer dust heated by newly formed stars . @xcite in a study of m31 using _ spitzer_mir / fir , uv and optical data , also concluded that the dust is mostly heated by an old stellar population ( a few gyr old ) . to investigate the relation between our derived dust properties and the sfr and the general stellar population , we have used the 3.6 map presented in @xcite to trace the general stellar population ( rather than the luminous newly formed stars that dominate the uv ) and a map of star - formation rate ( sfr ) . the sfr map is created from far - uv and 24 images which have been corrected for the contribution of the general stellar population to the emission at these wavelengths ( for details see g. p. ford et al . in preparation ) . these maps were all convolved to the same resolution and binned to the same pixel size as all the other maps in our analysis . we have plotted the fluxes from these maps against the results of our sed - fits ( see figure [ fig : heating ] ) . in section [ sec : spatial - dist ] we showed there is a clear break in the dust properties at a radius of 3.1kpc . therefore in figure [ fig : heating ] the pixels at a radius less than 3.1kpc are shown in blue and those at a radius above 3.1kpc in green . for each graph ( and both sets of pixels ) the spearman rank coefficient is computed and an estimate made of its significance ( see table[table : spearman ] ) . with such a large number of pixels , all but one of our correlations are formally significant . in the discussion below we have concentrated on the strongest correlations as measured by the spearman correlation coefficient . there are strong correlations in figure [ fig : heating ] between @xmath1 and 3.6 emission . we believe these are most likely to be caused by radial variations in @xmath1 seen in figure [ fig : sed - rad ] and the decrease in 3.6 brightness with radius . we discuss the possible cause of the radial variation in @xmath1 in section [ sec : beta - t ] . we find a strong correlation between dust surface - density and the sfr in the outer regions , but not with the surface density of total stars traced by the 3.6 emission . this correlation is expected as stars are formed in clouds of gas and dust . in the inner region there is an anti - correlation between the dust surface density and the 3.6 flux , which seem most likely to be explained by both quantities varying with radius : the 3.6 emission from the bulge increasing towards the center and the dust surface density decreasing towards the center . ccccc & & inner & -0.73 & @xmath65 + & & outer & -0.16 & @xmath66 + & & inner & 0.31 & @xmath67 + & & outer & 0.74 & 0.00 + ( rl)1 - 5 & & inner & 0.90 & @xmath68 + & & outer & -0.09 & @xmath69 + & & inner & 0.04 & @xmath70 + & & outer & 0.14 & @xmath71 + ( rl)1 - 5 & & inner & -0.52 & @xmath72 + & & outer & 0.56 & @xmath73 + & & inner & 0.16 & @xmath74 + & & outer & -0.24 & @xmath75 [ table : spearman ] the strongest correlation is seen between the dust temperature and the 3.6 flux in the inner region . figure [ fig : ftcorr ] shows a log - log graph of the two quantities and a linear fit to the points . the gradient represents the power _ n _ where @xmath76 where @xmath77 . while the correlation suggests the dust in the bulge is heated by the general stellar population , for a modified blackbody with @xmath42 we would expect a gradient of 6 . the difference is probably explained by the simplicity of our assumptions : that there is only a single stellar population in the bulge and that the bulge has a constant depth in the line of sight . if these assumptions are incorrect , the 3.6 surface brightness of m31 will only be an imperfect tracer of the intensity of the interstellar radiation field ( isrf ) . looking at the temperature beyond 3.1kpc , we find a weak , but still highly significant , correlation with sfr , suggesting that the isrf has a significant contribution from star - forming regions . as most of the star formation in m31 occurs in the 10kpc ring this is to be expected . this correlation can be seen in figure [ fig : sfrcont ] where most but not all of the temperature peaks in the 10kpc ring appear aligned with the peaks in the sfr map . in the same region there is a slight anti - correlation of temperature with the 3.6 flux which could be explained by the radial decrease in the 3.6 flux whilst the dust temperature increases slightly with radius . the fact that dust temperature increases slightly with radius , while the number - density of stars , traced by the 3.6 , is falling with radius suggests that outside the bulge the dust is mainly heated by young stars . nevertheless , the lack of a strong correlation between dust temperature and either sfr or 3.6 flux suggest the optical / uv light absorbed by a dust grain is from photons from a large range of distances ( e.g. , photons from the bulge heating dust in the disk ) . @xcite have studied fir colour ratios in m81 , m83 and ngc4203 . they find the 250/350 color ratio has the strongest positive correlation with 1.6 emission . an increase in the 250/350 ratio would indicate either an increase in dust temperature or a decrease of @xmath1 . @xcite conclude that the most likely explaination is the temperature effect , with the dust being heated by the general stellar population traced by the 1.6 emission . we find a similar correlation to @xcite , only we trace the stellar radiation field with the 3.6 band . however , our sed - fitting results suggest that this is caused by a combination of changes in temperature and @xmath1 . note that since @xcite only use color ratios they are unable to discriminate between changes of temperature and @xmath1 . at radii greater than 3.1kpc our results suggest the variation in the 250/350 is mainly caused by a change in @xmath1 . @xcite found that the 70/160 color ratio has the greatest correlation with star - formation rate , which is evidence that there is dust at more than one temperature contributing to the 70500 emission . one possible explanation of our failure to find a correlation between the 3.6 emission and dust temperature outside 3.1kpc , instead of the negative correlation between 3.6 and @xmath1 we find , might be if the _ herschel _ emission at short wavelengths contains a contribution from a warmer dust component . this would mean our fits of a one - component modified blackbody would produce misleading results . however , as stated in section [ sect : sedfit ] when we attempted the same sed fitting process but using all flux densities @xmath4160 as upper limits , we see little difference in our results . the dust emissivity index ( @xmath1 ) is related to the physical properties of the dust grains , including the grain composition , grain size , the nature of the absorption process and the equilibrium temperature of the dust . we would also expect to see a change in @xmath1 due to environment from the processing of the grains via grain growth ( e.g. coagulation , mantle accretion ) or destruction through surface sputtering by ions / atoms or shattering by shocks . in m31 , we detect an apparent inverse correlation between @xmath78 and @xmath1 for the inner and outer regions of m31 , as shown in figure [ fig : b - tmodels ] . we find the form of the relation is different for the two regions . such an inverse relationship has been observed in the milky way with previous fir - submm experiments and surveys including archeops @xcite , which showed @xmath1 ranging from 4 to 1 with the dust temperature varying between 7 and 27k , and pronaos @xcite , which shows a variation of @xmath1 from 2.40.8 for dust temperatures between 11 and 80k . @xcite used ir - mm data of galactic high latitude clouds and found a similar trend , and more recently @xcite with _ herschel _ found a similar inverse relationship with @xmath79 for @xmath80 for galactic longitude @xmath81 ( at longitude @xmath82 @xmath83 for 1823k ) . recently @xcite used _ herschel_-atlas observations to investigate @xmath1 variations in low - density , high - latitude galactic cirrus , measuring values of @xmath1 ranging from 4.51.0 for @xmath84 . these @xmath85 relationships could be indicative of a problem with the temperature@xmath1 degeneracy arising from the sed fitting , the presence of dust with a range of temperatures along the line of sight @xcite , or real variations of the properties of the dust grains . on the assumption that the inverse correlations between @xmath1 and @xmath78 in figure [ fig : b - tmodels ] are not simply caused by the two variables being separately correlated with radius , we looked for other possible causes of the relationships : 1 . the fitting can lead to a spurious inverse correlation between @xmath1 and @xmath78 ( * ? ? ? * section [ sec : simulation ] ) . the most striking feature in figure [ fig : b - tmodels ] is the clear separation in points between the inner 3.1kpc and the outer regions . to test whether these different distributions might be produced by the fitting artifact , we used the monte - carlo simulations from section [ sec : simulation ] to simulate the effect of fitting a modified blackbody for various combinations of @xmath1 and @xmath86 . the grey lines in figure [ fig : b - tsim ] show the best - fit relationships for different input @xmath78 and @xmath1 combinations and clearly show that the two different relationships in the two regions can not be obtained from a single population of dust grains . the green and blue data points represent the range of output @xmath78 and @xmath1 for an input modified blackbody with @xmath87k ( green ) and @xmath88k ( blue ) , with @xmath89 . a comparison of figures [ fig : b - tmodels ] and [ fig : b - tsim ] shows that in both regions of m31 there is a larger range of temperature and dust emissivity for the real data than that found in the monte - carlo simulation , indicating there are genuine variations of @xmath78 and @xmath1 in both regions . moreover , the fitting artifact can not explain the observed relationships of @xmath1 and temperature with radius ( figure [ fig : sed - rad ] ) . 2 . artificial inverse @xmath90 relationships can also be produced if a one - component modified blackbody model is used to fit dust which contains a range of dust temperatures @xcite . since we are averaging through the disk of a galaxy along the line of sight ( los ) , it is obviously possible that the dust contains a range of dust temperatures . while we can not fully address this issue , our @xmath1 values are higher than expected , which is the opposite of what happens from a los averaging of temperatures . we also find no statistical evidence from our fits that there is more than one component of dust . also @xcite and @xcite show inverse @xmath85 relationships still exist in places where it is unlikely there is dust at more than one temperature . 3 . variation of @xmath1 with wavelength has been reported by some authors both from theoretical models and laboratory experiments and from observations ( e.g. , * ? ? ? * ; * ? ? ? * ) , with a transition around 500 . in section [ sec : excess ] we show that there is no evidence for excess 500 emission , suggesting this is not an explanation of our results . to describe the @xmath85 relationships we use an empirical model of the form @xmath91 , commonly used in the literature ( e.g. , * ? * ; * ? ? ? * ; * ? ? ? * ) to fit the @xmath85 anti - correlation . the observed anti - correlation between t and @xmath1 may arise due to a change in the physical properties of grains including the grain optical constants changing with temperature for amorphous grains or changes in the dust emissivity with wavelength ( see * ? ? ? * and references therein ) . other possibilities include grain growth or quantum mechanical effects ( though these latter grain properties only arise at lower dust temperatures than observed in m31 ) . the best - fit relationship which describes @xmath85 for r@xmath92 and for @xmath93 are ( shown in figure [ fig : b - tmodels ] ) : @xmath94 where the steeper @xmath95 relationship at r@xmath96 agrees well with the relationship found in the plane of the milky way at longitudes of @xmath81 @xcite and in low - density , high - latitude cirrus @xcite . there is some evidence that the @xmath95 relationship in m31 is slightly steeper , so that for the same temperature compared to the galactic plane , m31 has a lower @xmath1 ( but this is only at @xmath97 level ) . what could be a physical ( or chemical ) explanation of the different @xmath98 relationships in the two regions ? typical values of @xmath1 are in the range 1.52.0 for interstellar dust grains , and have been found in global extra - galactic studies ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) and _ average _ global values measured in the milky way ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? low values of @xmath1 for large grains would typically represent freshly - formed dust grains in circumstellar disks or stellar winds . alternatively , @xmath99 has been observed in regions where small grains dominate @xcite . high values of @xmath1 ( @xmath100 ) might occur due to grain coagulation or to the growth of icy mantles on the surface of the grains in denser regions @xcite . studies have also suggested high values of @xmath1 are associated with very cold dust ( @xmath101 e.g. , * ? ? ? * ) possibly caused by a change in the absorption properties due to quantum effects at low temperatures , increasing self - absorption in amorphous grains via tunnelling @xcite . the highest value of @xmath1 is seen at the 3.1kpc boundary between the two regions . this can not be caused by changes in the quantum mechanical absorption , since this is only thought to be important for cold dust at temperatures @xmath102 . the high @xmath1 values could be due to efficient grain coagulation or mantle growth in dense molecular clouds , although this too seems unlikely as little co(@xmath103 ) is observed in this region . while there is no obvious explanation for the high @xmath1 values at this radius , there are many indications that this 3kpc `` boundary '' is an interesting regime , we discuss this further in section [ sec : transition ] . in the inner region , we suggest that the decrease in @xmath1 with corresponding increase in @xmath86 might be caused by the increased intensity of the isrf . towards the center of m31 , we would expect increased sputtering or sublimation of mantles from the increased isrf , shown by the increased temperature of the dust and the increased x - ray emission observed in the center @xcite . the lack of gas in the central regions ( figure [ fig : gas - rad ] ) also suggests that dust is less likely to be shielded and thus more efficiently sputtered and leading to smaller grain sizes . as we mentioned above , a problem with this analysis is that it is difficult to determine which are the causal relationships . for example , we have argued that the radial variation in @xmath86 is due to the radial variation in the isrf . the radial variation in @xmath1 might then be due to a physical relationship between @xmath104 and @xmath1 or it might be the case that there is no causal relationship between these parameters but the radial variation in @xmath1 is caused by a different effect . for example , an interaction between m32 and m31 might have caused a wave of star formation which has moved out through the galaxy , which might have led ( by a number of processes ) to the radial variations in @xmath1 . therefore , we can rule out some hypotheses but we can not conclusively determine which is the true explanation using this data set . interpreting the transition in dust properties seen at 3.1kpc ( figure [ fig : sed - rad ] ) is difficult . one possible clue comes from previous gas kinematics studies . @xcite found that the hi rotation curve inside a 4kpc radius is warped with respect to the rest of disk . @xcite suggest the inner hi data is consistent with a bar extended to 3.2kpc , while a newer analysis by @xcite explains the hi distribution as the result of a triaxial rotating bulge . @xcite , using _ spitzer _ irac observations , identified a new inner dust ring with dimensions of 1.5@xmath151kpc . by using the stellar and gas distributions and from the presence of the 10kpc ring , they conclude that an almost head - on collision has occurred between m31 and m32 around 210 million years ago . this collision could explain the perturbation of the gas observed in the central 4kpc . these other observations all show that the inner 3kpc of m31 is an intriguing region , although it it not clear what are the causes of the difference in the dust properties . the perturbation of the gas may have lead to the processing of dust grains , or potentially material from m32 could have been deposited after the interaction . the total dust mass for the pixels in our selection within 3.1kpc is @xmath105 which is a plausible amount to be deposited as recently dust masses of @xmath0@xmath106 have been reported in virgo dwarf ellipticals @xcite . another possibility is the dust properties are affected by the differences between conditions in the bulge and disk . @xcite decomposed the luminosity profile of _ spitzer _ irac data into a bulge , disk and halo . from their figure 16 we can see that our transition radius of @xmath03kpc is approximately where the bulge emission begins to become a significant fraction of the optical disk emission . whether the transition in the dust parameters is due to the changing contribution to the isrf from the general stellar population and star formation or if there is another influence in the bulge is unknown . the detection of ` dark gas ' in the milky way was a surprising early result from _ planck _ @xcite , obtained by combining _ iras _ 100 data and the six - band _ planck _ data from 3503 mm . the _ planck _ team compared the dust optical - depth with the total column density of hydrogen ( @xmath107 ) , where the optical depth at each wavelength is given by @xmath108 where @xmath109 is the flux density in that band and @xmath110 is the blackbody function . they assumed that at low @xmath107 the atomic hydrogen dominates over the molecular component while at high column density the molecular hydrogen dominates the emission . for these two regimes they found a constant gas - to - dust ratio , but at intermediate column densities they found an excess of dust compared to the gas . this excess is attributed to gas traced by dust but not by the usual hi and co lines , and is found to be the equivalent 28% of the atomic gas or 118% of the molecular gas . the excess dust emission was typically found around molecular clouds , suggesting that the most likely cause is the presence of molecular gas not traced by the co line . we attempted the same analysis as the @xcite for m31 using our sed fitting results from section [ sec : res ] . instead of using equation [ equ : planck ] , we compare the column density of gas estimated from the hi and co with the column density of dust ( for convenience we call this @xmath111 ) . we use this parameter as it is calculated with data from all wavelengths , whereas if we used equation [ equ : planck ] , small errors in temperature would cause large uncertainties in @xmath112 for wavelengths close to the peak of emission . the _ planck _ team found no radial variation in the gas - to - dust ratio in the milky way ( see figure 10 * ? ? ? * ) . in andromeda we show that the gas - to - dust ratio does vary radially ( section [ sec : radgd ] and figure [ fig : gas - rad ] ) , as expected from the metallicity gradient . to determine if there is an excess at intermediate column density in dust compared to that expected from the gas we have to correct for the radial change in gas - to - dust ratio . to remove this dependence , we adjust dust column density by using the exponential fit ( shown by the red line in figure [ fig : gas - rad ] ) so the gas - to - dust ratio at all radii has the same value as the center of m31 . to avoid biasing this correction by assuming a value for the x - factor which is highly uncertain , we estimate this relationship from pixels where the atomic hydrogen column density is @xmath113% of @xmath107 ( 1569 pixels out of the 3600 ) . in figure [ fig : optdep - gas ] we show that the relationship between corrected dust column density ( @xmath111 ) and gas column density is well represented by assuming the two quantities are directly proportional with no excess in dust column - density that could be attributed to ` dark gas ' . although only a small proportion of the gas is molecular , we still need to use a value for the x - factor . we can estimate this quantity from the data itself by finding the values of the x - factor and the constant of proportionality between the gas column density and the corrected @xmath111 that gives the minimum @xmath114 value ( the fitted line is shown in figure [ fig : optdep - gas ] . ) we find a best value for the x - factor of @xmath115^{-1}$ ] ( or expressed as @xmath116^{-1}$ ] ) , where the random error is estimated using a monte - carlo technique ( similar to one used in section [ sect : sedfit ] ) . for the dust column densities in each pixel we use the uncertainties provided by the sed fitter as explained in section [ sect : sedfit ] , which is on average @xmath022% . @xcite quote a calibration error of 15% for the co observations ( which directly results in at least a 15% uncertainty in the x - factor ) which we combine with the noise in each pixel of our processed moment - zero co map . for the hi observations we use an uncertainty map provided by robert braun ( private communication ) which has an average uncertainty of 12% on the raw 10 map . we also include a 5% systematic uncertainty ( e.g. , calibration uncertainties ) . if the opacity corrected hi map is used our best fit x - factor is @xmath117^{-1}$ ] . @xcite find values of the x - factor between @xmath118 to @xmath119^{-1}$ ] when analysing the southern , northern and inner regions for m31 with _ spitzer our average value of the x - factor falls within their range of values . a full analysis of the spatial variations of the x - factor with our data will be undertaken in a future paper . there are two main problems with this method which could lead us to miss ` dark gas ' . first , unlike the milky way we have to average through the whole disk of m31 and , second , m31 has a significantly lower molecular gas fraction . the latter prevents us from fitting the model to pixels with just very high and low values of @xmath107 as the pixels with highest molecular gas fraction are not clustered to high @xmath120 values ( this is illustrated in figure [ fig : codist ] ) . this suggests andromeda may not be the best galaxy for this analysis as the molecular contribution to the overall column density is quite low . however , we can also try an alternative method of looking for ` dark gas ' because we have one important advantage over the _ planck _ team : we can see m31 from the outside . we can therefore make a map of the ratio of radially corrected dust column density ( @xmath111 ) to gas column density to look for regions of enhancements in this ratio ( figure [ fig : tauimage ] ) . the image clearly shows spatial variations which could either suggest regions of ` dark gas ' or local variations in the metallicity or emissivity of dust . to distinguish between these scenarios an independent measurement of the ` dark gas ' is required . on the outskirts of molecular clouds , co could be photodissociated and the carbon gas would therefore reside in c or c@xmath121 @xcite . planned observations of the c[ii ] 158 line in andromeda with _ herschel _ could then be a potential test for investigating whether ` dark gas ' exists in m31 . in this paper we present the results of an analysis of dust and gas in andromeda using new _ herschel _ observations from the helga survey . we have @xmath04000 independent pixels with observations in the range of 70500 . we find the following results : 1 . we find that a variable dust emissivity index , @xmath1 , is required to adequately fit all the pixels in andromeda . when a variable @xmath1 is used , the modified blackbody model with a single temperature is found to be a statistically reasonable fit to the data in the range 100500 . there is no significant evidence of an excess of dust emission at 500 above our model . 2 . there are two distinct regions with different dust properties , with a transition at @xmath122kpc . in the center of andromeda , the temperature peaks with a value of @xmath030k and a @xmath1 of @xmath01.9 . the temperature then declines radially to a value of @xmath017k at 3.1kpc with a corresponding increase in @xmath1 to @xmath02.5 . at radii larger than 3.1kpc @xmath1 declines but only with a small associated increase in temperature . the drop in @xmath1 towards the center of the galaxy may be caused by increased sputtering or sublimation of mantles from an increased isrf . the origin of the high @xmath1 values at 3.1kpc from the center is less clear but may be indicative of either grain coagulation or an increase in the growth of icy mantles . the dust surface density for our pixels in which flux is detected at @xmath123 in all herschel bands range is between @xmath00.12.0@xmath38 . we find the gas - to - dust ratio increases exponentially with radius . the gradient matches that predicted from the metallicity gradient assuming a constant fraction of metals are included into dust grains . the dust surface density is correlated with the star formation rate rather than with the stellar surface density . 5 . in the inner 3.1kpc the dust temperature is correlated with the 3.6 flux . this suggests the heating of the dust in the bulge is dominated by the general stellar population . beyond 3.1kpc there is a weak correlation between dust temperature and the star formation rate . we find no evidence for ` dark gas ' , using a similar technique as the _ planck _ team . however , we find this technique may not be as effective for m31 due to poor angular resolution and line of sight effects . we have used an alternative technique by constructing a gas - to - dust map after correcting for the radial gradient . we do find regions with enhancements ( i.e. , higher values of @xmath124 ) , which may show places where ` dark gas ' exists , or may be due to local variations in the gas - to - dust ratio . a detection of a potential component of co - free molecular gas will be possible with future observations to measure the c[ii ] line planned with _ herschel_. 7 . by minimising the scatter between our radially corrected dust column - density and the column - density of gas inferred from the hi and co line we find a value for the x - factor of @xmath115^{-1}$ ] ( or expressed as @xmath116^{-1}$ ] ) . our results of andromeda represent the largest resolved analysis of dust and gas in a single galaxy with _ the results of this analysis on m31 is strikingly different from those obtained by the planck team in the milky way , since we find no clear evidence for ` dark gas ' , a radial gradient in the gas - to - dust ratio and evidence for radial variation in the dust emissivity index ( @xmath1 ) . in future work it will be important to understand these differences between the two big spirals in the local group . we thank everyone involved with the _ herschel _ observatory . pacs has been developed by a consortium of institutes led by mpe ( germany ) and including uvie ( austria ) ; ku leuven , csl , imec ( belgium ) ; cea , lam ( france ) ; mpia ( germany ) ; inaf- ifsi / oaa / oap / oat , lens , sissa ( italy ) ; iac ( spain ) . this development has been supported by the funding agencies bmvit ( austria ) , esa - prodex ( belgium ) , cea / cnes ( france ) , dlr ( germany ) , asi / inaf ( italy ) , and cicyt / mcyt ( spain ) . cvn , pr , dh , gr , yn and ke acknowledge support from the belgian federal science policy office via the prodex programme of esa . spire has been developed by a consortium of institutes led by cardiff univ . ( uk ) and including : univ . lethbridge ( canada ) ; naoc ( china ) ; cea , lam ( france ) ; ifsi , univ . padua ( italy ) ; iac ( spain ) ; stockholm observatory ( sweden ) ; imperial college london , ral , ucl - mssl , ukatc , univ . sussex ( uk ) ; and caltech , jpl , nhsc , univ . colorado ( usa ) . this development has been supported by national funding agencies : csa ( canada ) ; naoc ( china ) ; cea , cnes , cnrs ( france ) ; asi ( italy ) ; mcinn ( spain ) ; snsb ( sweden ) ; stfc , uksa ( uk ) ; and nasa ( usa ) .

the dressed heavy quark mass with being the binding energy is introduced to characterize the heavy hadrons containing a single heavy quark . a heavy quark expansion in terms of the inverse of the dressed heavy quark mass is presented with a complete decomposition of the full field and integrating out the small components . the heavy quark - antiquark coupling effects are included in the finite mass corrections . it is shown that the expansion is more favorable in application . the extraction of from exclusive b decays is studied by using such a new expansion approach .

1004.5444

for a heavy hadron which contains a single heavy quark ( bottom or charm ) , the heavy quark mass @xmath4 is much larger than the qcd energy scale @xmath5 which characterizes the light degrees of freedom in the heavy hadron . the four momentum of the heavy quark can be expressed as @xmath6 , where @xmath7 is taken to be the velocity of hadron at the rest frame , and @xmath8 is the residual momentum of the order of binding energy , which is much smaller than @xmath4 . the heavy quark symmetry @xcite and its breaking effects are of particular importance in studying such hadrons . consequently , the heavy quark effective theory ( hqet ) has been developed , where the effective lagrangian is expanded in @xmath9 . in deriving the lagrangian of hqet , the quark and antiquark are assumed to be conserved separately . namely the heavy antiquarks are regarded as completely decoupled from the heavy quarks at the beginning . the transition matrix elements can also be represented in series of @xmath9 through the heavy quark expansion ( hqe ) and evaluated order by order . hqet and hqe have been discussed by many authors @xcite . in the past two decades they are widely used in studying heavy hadrons . when the momentum of the heavy quark is much lower than the quark - antiquark pair creation threshold , an alternative framework of effective field theory for heavy quarks can directly be derived from the full qcd @xcite . just like for other effective theories , the basic idea is that some degrees of freedom characterizing higher scale physics can be decomposed and integrated out when we consider physics at low energy scales . concretely speaking , for heavy quarks with @xmath10 , one may perform a complete decomposition of the qcd full field into quark field and antiquark field via positive and negative energy components ( see below ) of a full field , and integrate out the small components of quark field and antiquark field , which leads to the so - called @xmath9 corrections . when considering heavy quark ( or antiquark ) systems , one should further integrate in the contributions of heavy antiquark ( or quark ) components . as a consequence , additional @xmath9 corrections arise from the quark - antiquark coupling terms in the full qcd . it should be noted that such a framework is distinguishable from the usual hqet in which the particle and antiparticle were assumed to be conserved separately and treated in a different way . for convenience , we refer to such a framework as a heavy quark effective field theory ( hqeft ) . though the heavy quark - antiquark coupling effects vanish in the heavy quark limit , they are actually nonzero when one considers the finite mass contributions . it is then not surprising that the @xmath9 corrections evaluated in hqeft and hqet could be different though they are the same in the infinity mass limit . for instance , the transition matrix elements of @xmath9 corrections concern less independent wave functions in hqeft than in the usual hqet and the @xmath9 order corrections at zero recoil automatically vanish in hqeft , and there exist some relations between wave functions and heavy hadron masses in hqeft . hqeft has been applied to explore various processes of heavy hadrons . in particular , the cabibbo - kobayashi - maskawa ( ckm ) matrix elements @xmath3 and @xmath11 are extracted from both inclusive @xcite and exclusive @xcite b decays . in the treatment of inclusive b decays , the dressed heavy quark mass @xmath0 as a whole enters the formulation , which implies that a dressed heavy quark "- hadron duality is more reasonable than the naive heavy quark - hadron duality . as a consequence , when the inclusive decay rates of heavy hadrons are expressed in terms of the physical hadron masses , they receive no @xmath2 order corrections . this treatment not only suppresses the next - to - leading order contributions and makes the operator product expansion ( ope ) reliable , but also diminishes the large uncertainties arising from the heavy quark mass . for exclusive decays , hqeft has also been demonstrated to be reliable . whereas in our previous works the heavy quark expansion for the effective lagrangian and transition matrix elements is carried out in powers of @xmath9 . in this paper , we briefly review the description of hqeft and show that a @xmath2 expansion is also consistently applicable to the heavy quark effective lagrangian and transition matrix elements . in sec . [ sectionlagrangian ] , we first outline the derivation of a complete hqeft and then extend it to the formulation in terms of @xmath2 expansion . in sec . [ sectionmatixelement ] , we present new formulae for hqe of heavy - to - heavy transition matrix elements by applying the @xmath2 expansion . in sec . [ sectionvcb ] , we extract the ckm matrix element @xmath3 based on the new formulation and the most recent experimental data . our conclusions are given in sec . [ sectionsummary ] . the lagrangian in the full qcd is @xmath12 @xmath13 is the full field for heavy quark , and @xmath14 represents the section containing no heavy quarks . based on the principle of superposition , the field @xmath13 in quantum field theory is actually the superposition of two parts which correspond to the positive and negative energy components . namely the full field @xmath13 can always be decomposed formally into positive and negative energy parts @xmath15 where @xmath16 and @xmath17 may be expressed explicitly in the energy - momentum space as @xmath18 it is clear that @xmath16 and @xmath17 correspond to the positive and negative energy parts of the full field @xmath13 , which are the so - called quark field and antiquark field , respectively . in the case for free quark fields , they can be expanded in terms of plane waves as @xmath19 where @xmath20 is the spin index , @xmath21 and @xmath22 are the annihilation and creation operators respectively , and @xmath23 and @xmath24 are four - component spinors . in dirac representation , they can be explicitly written as @xmath25 with @xmath26 being the two component pauli spinor field that annihilates a heavy quark , and @xmath27 being the pauli spinor field that creates a heavy antiquark . later on we will see that the effective heavy quark and antiquark fields at @xmath28 exactly come from the large " components of @xmath16 and @xmath17 . the generating functional in the full theory can be represented as @xmath29&=&\int { \cal d}q { \cal d}q e^{i\int d^4x ( { \cal l}_{qcd}[q , q]+j \phi ) } .\end{aligned}\ ] ] in eq.([generatingfun1 ] ) and all the following relevant equations , we simply denote the source terms as @xmath30 for convenience . introducing a vector @xmath7 with @xmath31 , one can define the projecting operators @xmath32 which satisfies @xmath33 then @xmath34 can be written as @xmath35 with @xmath36 at @xmath28 , the field components @xmath37 and @xmath38 become small components " of quarks and antiquarks , while @xmath39 and @xmath40 are the large components " @xcite . to be more explicit , taking @xmath41 , one then has in the momentum space @xmath42 in this case one can decompose the full field @xmath13 as eqs.([qzf ] ) , ( [ qdec1 ] ) and ( [ qdec2 ] ) , and write the generating functional as @xmath43 = \int { \cal d}q { \cal d}\hat{q}^{+}_v { \cal d}\hat{q}^{-}_v { \cal d}r^{+}_v { \cal d}r^{-}_v e^{i \int d^4x ( { \cal l}_{qcd}[q,\hat{q}^{+}_v , \hat{q}^{-}_v , r^{+}_v , r^{-}_v ] + j \phi ) } .\end{aligned}\ ] ] then one may integrate out the small components @xmath37 and @xmath38 to get @xmath44= \int { \cal d}q { \cal d}\hat{q}^{+}_v { \cal d}\hat{q}^{-}_v e^{i\int d^4x ( { \cal l}_{light}+\hat{\cal l}_{q , v}[\hat{q}^{+}_v , \hat{q}^{-}_v ] + j \phi ) } .\end{aligned}\ ] ] @xmath45 is the resulting lagrangian for heavy section with the small components integrated out . it can also be derived equivalently by using the relevant dirac equation of motion @xmath46 where @xmath47 , @xmath48 , @xmath49 , @xmath50 and @xmath51 are defined as @xmath52 clearly one has @xmath53 = 0.\end{aligned}\ ] ] @xmath45 is found to be @xcite @xmath54 with @xmath55 { \hat{q}^{\pm}_v } , \nonumber \\ \hat{\cal l}^{(\pm \mp)}_{q , v } & = & \frac{1}{2m_q } { \bar{\hat{q}}_v{\vspace{-0.3cm}\hspace{-0.2cm}{^{\pm } } } } ( -i \overleftarrow { \hat { \cal \not \!\ ! d}}_{\ ! v } -m_q ) \big ( 1-\frac{i{d\hspace{-0.25cm}\slash_{\|}}+{m_{q}}}{2{m_{q } } } \big)^{-1 } ( i{d\hspace{-0.25cm}\slash_{\bot } } ) { \hat{q}^{\mp}_v}\nonumber \\ & = & \frac{1}{2m_q } { \bar{\hat{q}}_v{\vspace{-0.3cm}\hspace{-0.2cm}{^{\pm } } } } ( -i \overleftarrow{{d\hspace{-0.25cm}\slash}}_{\ ! \bot } ) \big ( 1-\frac{-i \overleftarrow { d\hspace{-0.25cm}\slash_{\|}}+{m_{q}}}{2{m_{q } } } \big)^{-1 } ( i \hat { \cal \not \!\ ! d}_{\ ! v}-m_q ) { \hat{q}^{\mp}_v } , \end{aligned}\ ] ] where @xmath56 is defined as @xmath57 and the operator @xmath58 can be obtained from @xmath59 by replacing @xmath60 with @xmath61 . to get a reliable expansion at low energies , the large momentum carried by the heavy quark should be removed . generally this can be achieved by introducing new field variables @xmath62 @xmath63 is a parameter with mass dimension . it can be chosen appropriately according to the physical picture for the process studied . in general a heavy quark within a hadron can not truly be on shell due to strong interaction among heavy and light quarks as well as soft gluons . thus one may write the total momentum of the heavy quark in a hadron as @xmath64 , where @xmath65 is the part which depends on heavy flavor and is suppressed by the heavy quark mass ; @xmath66 is defined as the sum of the heavy quark mass and the binding energy @xmath1 that reflects the nonperturbative effects of strong interaction and relates to the light constituents in the heavy hadron , @xmath0 . in such a consideration , the residual momentum @xmath67 as a whole characterizes the off - shellness of the heavy quark in the heavy hadron . namely , the total residual momentum @xmath68 of the heavy quark is assumed to comprise the main contributions of the light degrees of freedom in the heavy hadron containing a single heavy quark . in such a physical picture , the heavy quark may be regarded as a dressed heavy quark " , and the heavy hadron containing a single heavy quark is more reliable to be considered as a dualized particle of a dressed heavy quark " . different from the heavy quark "- hadron duality in the usual heavy quark effective theory , what we are considering is the physical picture of the dressed heavy quark "- hadron duality . as a consequence , the wave functions defined in the next section should have a weaker dependence on the light constituents of heavy hadrons . with the definition ( [ momentumshift ] ) , the lagrangian ( [ lhatqv ] ) can be written in terms of @xmath69 and @xmath70 , which carry only the small residual momentum @xmath71 . explicitly , one has@xcite @xmath72 with @xmath73 in eq.([lqv ] ) we use @xmath74 to denote the leading term in the @xmath2 expansion of the lagrangian @xmath75 , and @xmath76 contains all @xmath2 corrections to @xmath74 . from eqs.([lqv])-([lqv2 ] ) one has @xmath77 note that @xmath78 so the effective lagrangian @xmath75 is complete for the large component of heavy quark and antiquark . in deriving @xmath75 we have only integrated over the small component ( @xmath79 ) of heavy quark and antiquark fields . in the above discussions , the binding energy @xmath1 is introduced based on physical consideration of the heavy - light systems . formally it can be defined consistently via the normalization of hadron states as follows . the hadron state @xmath80 in full theory is normalized as @xmath81 where @xmath82 is the momentum of the heavy hadron @xmath83 . in the effective theory at low energies , one may introduce an effective heavy hadron state @xmath84 , which is heavy flavor independent , and normalized as @xmath85 it is then related to the heavy hadron state @xmath86 via @xmath87 where @xmath88 denotes dirac matrixes , @xmath89 and @xmath90 are the mass differences between heavy hadrons and heavy quarks , while @xmath91 is independent of the heavy flavor and reflects the contributions of light degrees of freedom in the hadron . @xmath92 in eq.([matrixhhv ] ) is derived from the current @xmath93 . it will be given explicitly in sec.[sectionmatixelement ] . the basic framework of hqeft has been derived and discussed in the previous papers@xcite . here we reexpress the effective lagrangian in terms of the expansion @xmath2 instead of @xmath9 . the binding energy arising from the longitudinal residual momentum " of the heavy quark is absorbed into the heavy quark mass to be given as the dressed heavy quark mass @xmath94 , so that the flavor independent nonperturbative contributions of the light degrees of freedom are effectively included in the dressed heavy quark . this treatment is consistent with the physical picture of a heavy kernel ( the dressed heavy quark ) surrounded with the clouds of light degrees of freedom which mainly reflect the small transverse residual momentum " of heavy quark . as @xmath95 , an expansion in @xmath2 is expected to be more convergent and reliable , especially for charm quark systems . actually , it has been shown in the inclusive decays@xcite that by using @xmath63 instead of adopting @xmath4 and @xmath1 separately in the calculations , the results get less uncertainties and the order @xmath2 corrections are automatically absent . in this note , we will show that based on the new formulation of lagrangian given in eqs.([lqv])-([lqv2 ] ) , one can also consistently perform @xmath2 expansion for the exclusive decays . similar to the derivation of the effective lagrangian @xmath75 , the heavy quark current @xmath96 in full qcd can also be transformed into the following form by integrating out the small component @xmath97 , @xmath98 where @xmath99 with @xmath100 which have minor differences to the formulae in ( [ eq:12 ] ) . using eq.([relation ] ) , and noting that @xmath101 where @xmath102 and @xmath103 is a c - number , we write @xmath92 as @xmath104 furthermore , both @xmath92 and @xmath75 can be expanded in terms of @xmath105 . explicitly one obtains @xmath106 \frac{1+\sigma { \not\!{v}}}{2}q_v(x)\end{aligned}\ ] ] and @xmath107 \frac{1 + \varepsilon { \not\!{v}}}{2 } q_v , \\ \label{lqvexp2 } { \cal l}^{ii(1/\hat{m}_q)}_{q , v } & = & \sum _ { \varepsilon = \pm 1 } e^{2i\varepsilon \hat{m}_{q}v\cdot x } \bar{q}_v \frac{1 + \varepsilon { \not\!{v}}}{2}\big [ \frac{1}{2\hat{m}_q}(-i \overleftarrow{d\hspace{-0.25cm}\slash_{\|}}+\bar{\lambda } ) i{d\hspace{-0.25cm}\slash_{\bot}}\nonumber \\ & & + \frac{1}{4\hat{m}_{q}^{2}}(-i \overleftarrow{d\hspace{-0.25cm}\slash_{\|}}+\bar{\lambda } ) ( i{d\hspace{-0.25cm}\slash_{\|}}+\bar{\lambda})i{d\hspace{-0.25cm}\slash_{\bot}}\nonumber \\ & & + \frac{1}{4\hat{m}_{q}^{2}}(-i\overleftarrow { d\hspace{-0.25cm}\slash_{\bot}})^{2}i{d\hspace{-0.25cm}\slash_{\bot}}+ o\big(\frac{1}{\hat{m}_q^3 } \big ) \big]\frac{1- \varepsilon { \not\!{v}}}{2 } q_v .\end{aligned}\ ] ] then the effective current in terms of @xmath2 expansion is obtained : @xmath108\frac{1+\sigma { \not\!{v}}}{2}q_{v } , \end{aligned}\ ] ] where the operators are defined as @xmath109 eq.([neweffectivecurrent ] ) can be derived by using @xmath110 as the propagator when @xmath111 and @xmath112 fields are contracted . the feasibility of this treatment is shown in the appendix . note that the effective current @xmath113 consists of @xmath114 to @xmath114 and @xmath115 to @xmath115 components as well as mixing ones of @xmath114 to @xmath115 and @xmath115 to @xmath114 . when the effective field @xmath111 in @xmath113 acts on a specific hadron state , the state will pick up the proper component . that is , the hadron containing a single heavy quark ( antiquark ) picks up @xmath114 ( @xmath115 ) and cancels @xmath115 ( @xmath114 ) . now the heavy quark expansion for any heavy - to - heavy transition matrix elements can be represented as @xmath116 when @xmath117 , one gets from eqs.([normalizationqcd])-([matrixhhv ] ) and ( [ matrixhqe ] ) @xmath118 it is seen from eq.([lambdah ] ) that the heavy flavor dependence of @xmath119 can be attributed to the heavy - to - heavy transition matrix elements , or relevant wave functions . the heavy hadron mass is then given by @xmath120 which is the fact that the hadron mass consists of the dressed heavy quark mass and the residual mass suppressed by @xmath121 . to be concrete , we study the weak transition matrix elements between ground state pseudoscalar and vector mesons . they can be described by 18 form factors : @xmath122 , \nonumber\\ & & \hspace{-0.7cm}\langle d^{\ast}(v^{\prime},\epsilon^{\prime})| \bar{c}\gamma^{\mu } b | b(v)\rangle = i \sqrt{m_{d^{\ast } } m_b } h_{v}(\omega ) \epsilon^{\mu \nu \alpha \beta } \epsilon^{\prime \ast}_{\nu } v^{\prime}_{\alpha } v_{\beta } , \nonumber\\ & & \hspace{-0.7cm}\langle d^{\ast}(v^{\prime},\epsilon^{\prime})| \bar{c}\gamma^{\mu } \gamma^{5}b | b(v)\rangle = \sqrt{m_{d^{\ast } } m_b } [ h_{a_1}(\omega)(1+\omega ) \epsilon^{\prime \ast \mu } -h_{a_2}(\omega)(\epsilon^{\prime \ast } \cdot v)v^{\mu } \nonumber \\ & & \hspace{1 cm } -h_{a_3}(\omega)(\epsilon^{\prime \ast } \cdot v)v^{\prime\mu } ] , \nonumber\\ & & \hspace{-0.7cm}\langle d^{\ast}(v^{\prime},\epsilon^{\prime})| \bar{c}\gamma^{\mu}b| b^{\ast}(v,\epsilon)\rangle = \sqrt{m_{d^{\ast } } m_{b^{\ast } } } \{-(\epsilon\cdot \epsilon^{\prime\ast})[h_{1}(\omega)(v+v^{\prime})^{\mu } + h_{2}(\omega)(v - v^{\prime})^{\mu } ] \nonumber\\ & & \hspace{1 cm } + h_{3}(\omega)(\epsilon^{\prime\ast}\cdot v)\epsilon^{\mu } + h_{4}(\omega)(\epsilon\cdot v^{\prime})\epsilon^{\prime\ast\mu } -(\epsilon\cdot v^{\prime})(\epsilon^{\prime\ast}\cdot v ) [ h_{5}(\omega)v^{\mu}+h_{6}(\omega)v^{\prime\mu } ] \ } , \nonumber \\ & & \hspace{-0.7cm}\langle d^{\ast}(v^{\prime},\epsilon^{\prime})| \bar{c}\gamma^{\mu } \gamma^{5}b | b^{\ast}(v,\epsilon)\rangle = i\sqrt{m_{d^{\ast } } m_{b^{\ast } } } \{\epsilon^{\mu\nu\alpha\beta}\{\epsilon_{\alpha}\epsilon^{\prime\ast}_{\beta } [ h_{7}(\omega)(v+v^{\prime})_{\nu}\nonumber\\ & & \hspace{1cm}+h_{8}(\omega)(v - v^{\prime})_{\nu } ] + v^{\prime}_{\alpha}v_{\beta}[h_{9}(\omega)(\epsilon^{\prime\ast}\cdot v ) \epsilon_{\nu}+h_{10}(\omega)(\epsilon\cdot v^{\prime } ) \epsilon^{\prime\ast}_{\nu } ] \ } \nonumber\\ & & \hspace{1cm}+\epsilon^{\alpha\beta\gamma\delta}\epsilon_{\alpha } \epsilon^{\prime\ast}_{\beta}v_{\gamma}v^{\prime}_{\delta } [ h_{11}(\omega)v^{\mu}+h_{12}(\omega)v^{\prime\mu } ] \ } , \end{aligned}\ ] ] where @xmath123 is the polarization vector of the vector meson , and @xmath124 is the product of the four - velocities of heavy mesons , @xmath125 . on the other hand , for such transitions between @xmath126 states one can rewrite eq.([matrixhqe ] ) as @xmath127 -\frac{1}{4\hat{m}_{q'}^2 } \big [ ( -iv^{\prime } \cdot { \overleftarrow d}-\bar{\lambda } ) \big ( { \overleftarrow d^2_{\ ! \ \bot } } + \frac{i}{2 } \sigma_{\alpha\beta } f^{\alpha\beta } \big ) + i f^{\alpha\beta } v'_\alpha { \overleftarrow d}_{\beta } \nonumber \\ & & -f^{\nu \alpha } { \overleftarrow d}^\mu v'_\alpha \sigma_{\mu \nu } \big ] \frac{p'_+}{\bar{\lambda}-iv^{\prime } \cdot { \overleftarrow d } } \gamma + \frac{1}{4\hat{m}^2_q } \gamma \frac{p_+}{\bar{\lambda}+iv\cdot d } \big ( d^2_{\ ! \bot}+ \frac{i}{2 } \sigma_{\alpha\beta } f^{\alpha\beta } \big ) \frac{p_+}{\bar{\lambda}+iv\cdot d } \nonumber\\ & & \big ( d^2_{\ ! \bot}+ \frac{i}{2 } \sigma_{\gamma\delta } f^{\gamma\delta } \big ) + \frac{1}{4\hat{m}^2_{q ' } } \big ( { \overleftarrow d^2_{\ ! \ \bot } } + \frac{i}{2 } \sigma_{\alpha\beta } f^{\alpha\beta } \big ) \frac{p'_+}{\bar{\lambda}-iv ' \cdot { \overleftarrow d } } \big ( { \overleftarrow d^2_{\ ! \ ! \bot } } + \frac{i}{2 } \sigma_{\gamma\delta } f^{\gamma\delta } \big ) \nonumber \\ & & \frac{p'_+}{\bar{\lambda}-iv ' \cdot { \overleftarrow d } } \gamma + \frac{1}{4\hat{m}_q \hat{m}_{q ' } } \big ( { \overleftarrow d^2_{\ ! \ ! \bot } } + \frac{i}{2 } \sigma_{\alpha\beta } f^{\alpha\beta } \big ) \frac{p'_+}{\bar{\lambda}-iv ' \cdot { \overleftarrow d } } \gamma \frac{p_+}{\bar{\lambda}+iv\cdot d } \nonumber \\ & & \big ( d^2_{\ ! \bot}+ \frac{i}{2 } \sigma_{\gamma\delta } f^{\gamma\delta } \big ) + o\big ( 1/\hat{m}^3_{q^ { ( ' ) } } \big ) \big\ } { q_v^{+}}| h_v \rangle \end{aligned}\ ] ] with @xmath128 $ ] , @xmath129 , and the gluon field strength tensor @xmath130 $ ] . then a set of heavy flavor and spin independent wave functions can be introduced as follows , @xmath131 , \nonumber \\ & & \langle m^{\prime}_{v^{\prime}}| { \bar{q}^{\prime + } _ { v^{\prime } } } \gamma \frac{-p_+}{\bar{\lambda}+iv\cdot d } d^2_{\ ! \bot } { q_v^{+}}| m_v \rangle = -\kappa_1(\omega ) \frac{1}{\bar{\lambda } } tr[\bar{\cal m}^{\prime}\gamma { \cal m } ] , \nonumber \\ & & \langle m^{\prime}_{v^{\prime}}| { \bar{q}^{\prime + } _ { v^{\prime } } } \gamma \frac{-1}{\bar{\lambda}+iv\cdot d } p_{+ } \frac{i}{2 } \sigma_{\alpha\beta}f^{\alpha\beta}{q_v^{+}}| m_v \rangle = \frac{1}{\bar{\lambda}}tr [ \kappa_{\alpha\beta}(v , v^{\prime } ) \bar{\cal m}^{\prime}\gamma p_{+}\frac{i}{2}\sigma^{\alpha\beta}{\cal m } ] , \nonumber\\ & & \langle m^{\prime}_{v^{\prime}}| { \bar{q}^{\prime + } _ { v^{\prime } } } \gamma \frac{p_+}{\bar{\lambda}+iv\cdot d } [ { d_{\bot}}^2(i v\cdot d-\bar{\lambda})-i v_\mu d_\nu f^{\mu\nu}]{q_v^{+}}| m_v\rangle = -\varrho_1(\omega)\frac{1}{\bar{\lambda}}tr[\bar{\cal m}^{\prime}\gamma { \cal m } ] , \nonumber \\ & & \langle m^{\prime}_{v^{\prime } } | { \bar{q}^{\prime + } _ { v^{\prime } } } \gamma \frac{p_+}{\bar{\lambda}+iv\cdot d } \big [ \frac{i}{2}\sigma_{\alpha\beta}f^{\alpha\beta}(i v\cdot d-\bar{\lambda})+ v_\alpha \sigma_{\mu\nu } d^\mu f^{\nu\alpha } \big ] { q_v^{+}}| m_v \rangle \nonumber\\ & & \hspace{2 cm } = \frac{1}{\bar{\lambda}}tr[\varrho_{\alpha\beta}(v , v^{\prime})\bar{\cal m}^{\prime}\gamma p_{+ } \frac{i}{2}\sigma^{\alpha\beta}{\cal m } ] , \nonumber \\ & & \langle m^{\prime}_{v^{\prime}}| { \bar{q}^{\prime + } _ { v^{\prime } } } \gamma \frac{p_+}{\bar{\lambda}+iv\cdot d } d^2_{\ ! \bot } \frac{p_+}{\bar{\lambda}+iv\cdot d } d^2_{\!\bot } { q_v^{+}}| m_v \rangle = -\chi_1(\omega)\frac{1}{\bar{\lambda}^2 } tr[\bar{\cal m}^{\prime}\gamma { \cal m } ] , \nonumber\\ & & \langle m^{\prime}_{v^{\prime}}| { \bar{q}^{\prime + } _ { v^{\prime } } } \gamma \frac{p_+}{\bar{\lambda}+iv\cdot d } \big [ d^2_{\ ! \bot } \frac{p_+}{\bar{\lambda}+iv\cdot d } \frac{i}{2 } \sigma_{\alpha\beta } f^{\alpha\beta } + \frac{i}{2}\sigma_{\alpha\beta } f^{\alpha\beta } \frac{p_+}{\bar{\lambda}+iv\cdot d } d^2_{\!\bot } \big ] { q_v^{+}}| m_v \rangle \nonumber\\ & & \hspace{2 cm } = \frac{1}{\bar{\lambda}^2 } tr[\chi_{\alpha\beta}(v , v^{\prime})\bar{\cal m}^{\prime}\gamma p_{+ } \frac{i}{2}\sigma^{\alpha\beta}{\cal m } ] , \nonumber\\ & & \langle m^{\prime}_{v^{\prime } } | { \bar{q}^{\prime + } _ { v^{\prime } } } \gamma \frac{p_+}{\bar{\lambda}+iv\cdot d } \frac{i}{2}\sigma_{\alpha\beta } f^{\alpha\beta } \frac{p_+}{\bar{\lambda}+iv\cdot d } \frac{i}{2}\sigma_{\gamma\delta}f^{\gamma\delta}{q_v^{+}}| m_v \rangle \nonumber\\ & & \hspace{2 cm } = - \frac{1}{\bar{\lambda}^2 } tr[\chi_{\alpha\beta\gamma\delta}(v , v^{\prime } ) \bar{\cal m}^{\prime}\gamma p_{+ } \frac{i}{2}\sigma^{\alpha\beta } p_{+ } \frac{i}{2}\sigma^{\gamma\delta } { \cal m } ] , \nonumber \\ & & \langle m^{\prime}_{v^{\prime } } | { \bar{q}^{\prime + } _ { v^{\prime } } } { \overleftarrow d^2_{\ ! \ ! \bot } } \frac{p^{\prime}_{+}}{\bar{\lambda}-iv ' \cdot { \overleftarrow d } } \gamma \frac{p_+}{\bar{\lambda}+iv\cdot d } d^2_{\!\bot } { q_v^{+}}| m_v \rangle = -\eta_{1}(\omega ) \frac{1}{\bar{\lambda}^2 } tr[\bar{\cal m}^{\prime}\gamma { \cal m } ] , \nonumber \\ & & \langle m^{\prime}_{v^{\prime } } | { \bar{q}^{\prime + } _ { v^{\prime } } } { \overleftarrow d^2_{\ ! \ ! \bot } } \frac{p^{\prime}_{+ } } { \bar{\lambda } -iv'\cdot { \overleftarrow d } } \gamma \frac{p_+}{\bar{\lambda}+iv\cdot d } \frac{i}{2 } \sigma_{\alpha\beta } f^{\alpha\beta } { q_v^{+}}| m_v \rangle \nonumber\\ & & \hspace{2 cm } = \frac{1}{\bar{\lambda}^2 } tr[\eta_{\alpha\beta}(v , v^{\prime } ) \bar{\cal m}^{\prime}\gamma p_{+}\frac{i}{2}\sigma^{\alpha\beta}{\cal m } ] , \nonumber \\ & & \langle m^{\prime}_{v^{\prime } } | { \bar{q}^{\prime + } _ { v^{\prime } } } \frac{i}{2 } \sigma_{\alpha\beta } f^{\alpha\beta } \frac{p'_{+}}{\bar{\lambda}-iv ' \cdot { \overleftarrow d } } \gamma \frac{p_+}{\bar{\lambda}+iv\cdot d } \frac{i}{2 } \sigma_{\gamma\delta } f^{\gamma\delta } { q_v^{+}}| m_v \rangle \nonumber \\ & & \hspace{2 cm } = -\frac{1}{\bar{\lambda}^2 } tr[\eta_{\alpha\beta\gamma\delta}(v , v ' ) \bar{\cal m}^{\prime}\frac{i}{2}\sigma^{\alpha\beta } p'_{+ } \gamma p_{+ } \frac{i}{2}\sigma^{\gamma\delta}{\cal m } ] , \end{aligned}\ ] ] where @xmath132 is the spin wave function @xmath133 and @xmath134 . the decomposition of the tensors @xmath135 , @xmath136 , @xmath137 , @xmath138 , @xmath139 , and @xmath140 are the same as that presented in the appendix b of ref.@xcite . for simplicity , when the variable @xmath124 is not written explicitly , we refer to the zero recoil values of relevant functions , i.e. @xmath141 , @xmath142 , etc . with the definition in eqs.([formfactordef ] ) and ( [ wavefunctiondef ] ) , one obtains from eqs.([matrixhhv ] ) and ( [ matrixhqeexp ] ) @xmath143 where the normalization of the isgur - wise function @xmath144 @xcite has been used . at the zero recoil point , eqs.([matrixhhv ] ) and ( [ formfactordef])-([lambdahadronvec ] ) yield @xmath145 , \nonumber\\ h_{a_{1}}&=&1+\frac{1}{8\bar{\lambda}^2 } \big [ \frac{1}{\hat{m}_b}(\kappa_1 + 3\kappa_2 ) -\frac{1}{\hat{m}_c}(\kappa_1 -\kappa_2 ) \big]^2 -\frac{1}{8\hat{m}^2_b \bar{\lambda}^2 } ( \eta_{1}+6\eta_{2}-3\eta_{4}-9\eta_{5}-6\eta_{6 } ) \nonumber\\ & & -\frac{1}{8\hat{m}^2_c \bar{\lambda}^2 } ( \eta_{1}-2\eta_{2 } -3\eta_{4 } -\eta_{5}+2\eta_{6 } ) + \frac{1}{4\hat{m}_b \hat{m}_c \bar{\lambda}^2 } ( \eta_{1}+2 \eta_{2}+\eta_{4}+3\eta_{5}+2\eta_{6 } ) , \nonumber\\ h_{1}&=&1+\frac{1}{8\bar{\lambda}^2 } \big(\frac{1}{\hat{m}_b}-\frac{1}{\hat{m}_c}\big)^2 \big[(\kappa_{1}-\kappa_{2})^2 -\eta_1 + 2\eta_2 + 3\eta_{4}+\eta_{5}-2\eta_{6 } \big ] , \nonumber \\ \big [ 1+\frac{1}{8\bar{\lambda}^2}\big(\frac{1}{\hat{m}_b}-\frac{1}{\hat{m}_c}\big)^2 ( \kappa_{1}-\kappa_{2})^2 -\frac{1}{8\bar{\lambda}^2}\big(\frac{1}{\hat{m}^2_b}+\frac{1}{\hat{m}^2_c}\big ) ( \eta_1 - 2\eta_2 - 3\eta_{4}-\eta_{5}+2\eta_{6 } ) \nonumber\\ & & + \frac{1}{4\hat{m}_b \hat{m}_c \bar{\lambda}^2}(\eta_1 - 2\eta_2+\eta_{4}-\eta_{5 } -2\eta_{6 } ) \big ] . \end{aligned}\ ] ] then it is clear that all matrix elements in ( [ formfactordef ] ) are protected from @xmath2 order corrections at zero recoil . furthermore , one has @xmath146 @xcite because in the new framework of hqeft the effective current @xmath113 contains only terms with even powers of @xmath48 . the @xmath147 differential decay rates are @xmath148 |v_{cb } |^2 { \cal f}^2(\omega ) , \\ \label{widthb2d } \frac{d\gamma(b\rightarrow dl\nu)}{d\omega}&=&\frac{g^2_f}{48\pi^3}(m_b+m_{d})^2 m^3_{d}(\omega^2 - 1)^{3/2 } \vert v_{cb } \vert^2 { \cal g}^2(\omega ) \end{aligned}\ ] ] with @xmath149 , \end{aligned}\ ] ] where the qcd radiative corrections to two loops give the short distance coefficients @xmath150 and @xmath151 @xcite . the form factors @xmath152 contain long distance effects and can be estimated by nonperturbative methods such as lattice simulations , qcd sum rules or quark models . here we do not perform such calculations but try to make model independent prediction on @xmath3 using the hqe discussed in the previous sections . suppose that the residual momenta of the heavy quarks approximately equal and the longitudinal residual momenta of dressed heavy quarks be much smaller than the binding energy , we then have in a good approximation @xmath153 which implies @xmath154 . consequently , we arrive at the following relations among the wave functions : @xmath155 which will be adopted in the following discussions . since the contribution of the chromomagnetic moment operator is generally much smaller than that of the kinetic energy operator , we neglect operators containing two field strength tensors of gluon but remain those containing only one . as a result , @xmath156 and @xmath157 for @xmath158 will be neglected . thus we get from eqs.([lambdahadronpse])-([formfactorinwavefunction ] ) @xmath159 and @xmath160 ^ 2 -\frac{1}{24 \hat{m}^2_b\bar{\lambda}^2 } ( 2\bar{\lambda } \varrho_1 + 6\bar{\lambda } \varrho_2 -f_1 - 3f_2 ) \nonumber \\ & & -\frac{1}{24\hat{m}_c^2 \bar{\lambda}^2 } ( 2\bar{\lambda } \varrho_1 -2 \bar{\lambda } \varrho_2 -f_1 + f_2 ) + \frac{1}{12 \hat{m}_b \hat{m}_c \bar{\lambda}^2 } ( 2\bar{\lambda } \varrho_1 + 2\bar{\lambda } \varrho_2 -f_1-f_2 ) , \\ \label{formfactorhz } h_{+}&= & 1+\frac{1}{8\bar{\lambda}^2}\big(\frac{1}{\hat{m}_b}-\frac{1}{\hat{m}_c}\big)^2 \big[(\kappa_1 + 3\kappa_2)^2-\frac{1}{3 } ( 2\bar{\lambda } \varrho_1 + 6\bar{\lambda } \varrho_2 -f_1 - 3f_2 ) \big ] , \\ \label{formfactorhf } h_{-}&= & 0 , \end{aligned}\ ] ] where @xmath161 and @xmath162 are defined as @xmath163 as already mentioned in the previous section , the form factors @xmath164 and @xmath165 are protected from @xmath2 order correction , and @xmath166 holds up to order @xmath167 in our expansion . these make both the semileptonic decays of @xmath168 and @xmath169 the appropriate channels for the @xmath3 extraction . from eqs.([massb ] ) and ( [ massbs ] ) one can estimate the zero recoil values of @xmath170 and @xmath171 from the bottom and charm meson masses ( @xmath172 , @xmath173 , @xmath174 and @xmath175 ) . @xmath170 and @xmath171 as functions of the variables @xmath176 and @xmath177 are shown in fig.[fig1 ] . it is found that @xmath178 and @xmath179 are independent of @xmath1 , and the change of @xmath1 value only affects @xmath161 and @xmath162 quite slightly . @xmath178 is sensitive to @xmath176 and also influenced by @xmath177 . @xmath179 changes slightly against @xmath177 but is almost independent of @xmath176 . both @xmath161 and @xmath162 heavily depend on @xmath177 , and @xmath161 is also sensitive to @xmath176 . when taking @xmath180 we obtain @xmath181 the data in eqs.([variable ] ) and ( [ kappafrommass ] ) are consistent with the results in ref.@xcite . in that reference @xmath182 , @xmath183 and @xmath184 , @xmath185 are given via different methods of analysis on the sum rule equations that include only one - loop perturbative contributions . when the two - loop perturbative contributions are considered in the sum rule for the decay constant , @xmath186 , @xmath187 are obtained . @xmath164 and @xmath165 in ( [ formfactorha1 ] ) and ( [ formfactorhz ] ) also depend on @xmath188 and @xmath189 . note that the definition of @xmath170 and @xmath190 in ( [ wavefunctiondef ] ) can be written as @xmath191 \nonumber\\ & & \hspace{4 cm } + \frac{1}{\bar{\lambda}}tr[\kappa_{\alpha\beta}(v , v^{\prime } ) \bar{\cal m}^{\prime}\gamma p_{+}\frac{i}{2}\sigma^{\alpha\beta}{\cal m } ] , \\ & & \langle m^{\prime}_{v^{\prime } } | { \bar{q}^{\prime + } _ { v^{\prime } } } \gamma \frac{1}{\bar{\lambda } + iv\cdot d } ( i{d\hspace{-0.25cm}\slash_{\bot } } ) ( i{d\hspace{-0.25cm}\slash_{\|}}+ \bar{\lambda})(i{d\hspace{-0.25cm}\slash_{\bot } } ) { q_v^{+}}| m_v \rangle = -\varrho_1(\omega)\frac{1}{\bar{\lambda}}tr[\bar{\cal m}^{\prime}\gamma { \cal m } ] \nonumber\\ & & \hspace{4 cm } + \frac{1}{\bar{\lambda}}tr[\varrho_{\alpha\beta}(v , v^{\prime})\bar{\cal m}^{\prime}\gamma p_{+ } \frac{i}{2}\sigma^{\alpha\beta}{\cal m } ] . \end{aligned}\ ] ] then the approximation ( [ approxeq ] ) implies @xmath192 . the resulting @xmath3 value is shown in figs.[fig2]-[fig5 ] . using ( [ variable ] ) and allowing @xmath190 change in the range @xmath193 we get @xmath194 consequently , the averages of measurements @xcite @xmath195 give @xmath196 so the @xmath3 values extracted from @xmath197 and @xmath198 decays are consistent within the errors of experimental data . it is noticed that the value extracted from @xmath199 suffers from relatively larger experimental uncertainty , which can be seen in eq.([exp2 ] ) . the result for @xmath3 in ( [ vcbdsb ] ) is marginally consistent with the value given in ref.@xcite ( @xmath200 ) but has a smaller center value . for more precise determination of @xmath3 , it would be helpful to evaluate @xmath201 order wave functions such as @xmath171 and @xmath190 through other methods like lattice or qcd sum rule calculation . we have briefly reviewed the derivation of a heavy quark effective field theory . this hqeft is complete in that the effective lagrangian contains the heavy quark - antiquark coupling terms , which appear as finite mass corrections . unlike the usual naive heavy quark - hadron duality , we do not simply treat the light components of hadrons as spectators . instead , the flavor independent nonperturbative effects of light degrees of freedom are attributed to the dressed heavy quark with the dressed mass @xmath202 and the total momentum @xmath203 , and it is such a dressed heavy quark that dualizes the heavy hadron . consequently , hqeft has been extended into the formulation in terms of @xmath2 expansion . such an expansion is consistent with the picture of dressed heavy quark and becomes more convergent . the hqe of heavy - to - heavy transition matrix elements has been consistently extended into a @xmath2 expansion form , in which the contribution of heavy antiquark ( or quark ) field is integrated into the effective current . @xmath3 extraction from @xmath204 decays has been discussed by using the hqe in @xmath2 . due to the appropriate definition of effective states , zero recoil values of the relevant form factors can be estimated from the hadron masses . using some approximate relations between wave functions @xmath3 is found to be @xmath205 from @xmath168 decay and @xmath206 from @xmath169 decay . for these two channels , experimental study of @xmath169 is more difficult and it has larger uncertainties . nevertheless , the current data of @xmath207 and the form factors extracted within the framework of hqeft give the @xmath3 value consistent with that from @xmath208 decay , which shows the reliability of the @xmath2 expansion in this application . this work was supported in part by the national science foundation of china ( nsfc ) under grant # no . 10821504 , 10805005 , 10975170 and the project of knowledge innovation program ( pkip ) of the chinese academy of science . this appendix is devoted to the derivation of the effective current @xmath113 in eqs.([neweffectivecurrent ] ) and ( [ matrixhqe ] ) . in particular , we would like to show why one can use ( [ propagator ] ) as the propagator of @xmath111 field . firstly , one may notice that the gluon couplings arising from @xmath47 can be trivialized by the wilson - line transformation @xcite . one can introduce new field variable @xmath209 by @xcite @xmath210 where @xmath211 denotes path ordering with @xmath212 . since @xmath213 one can write eqs.([jqvexp0])-([lqvexp2 ] ) as @xmath214 \frac{1+\sigma { \not\!{v}}}{2}q^0_v(x)\end{aligned}\ ] ] and @xmath215 \frac{1 + \varepsilon { \not\!{v}}}{2 } q^0_v , \\ \label{lq0vexp2 } { \cal l}^{ii(1/\hat{m}_q)}_{q , v } & = & \sum _ { \varepsilon = \pm 1 } e^{2i\hat{m}_{q}v\cdot x\varepsilon } \bar{q}^0_v \frac{1 + \varepsilon { \not\!{v}}}{2}\big [ \frac{1}{2\hat{m}_q}(-i{v\hspace{-0.2cm}\slash}v\cdot \overleftarrow\partial + \bar{\lambda } ) i{d\hspace{-0.25cm}\slash_{\bot}}\nonumber \\ & & + \frac{1}{4\hat{m}_{q}^{2}}(-i { v\hspace{-0.2cm}\slash}v\cdot \overleftarrow\partial + \bar{\lambda } ) ( i{v\hspace{-0.2cm}\slash}v\cdot \partial + \bar{\lambda})i{d\hspace{-0.25cm}\slash_{\bot}}\nonumber \\ & & + \frac{1}{4\hat{m}_{q}^{2}}(-i\overleftarrow { d\hspace{-0.25cm}\slash_{\bot}})^{2}i{d\hspace{-0.25cm}\slash_{\bot}}+ o\big(\frac{1}{\hat{m}_q^3 } \big ) \big]\frac{1- \varepsilon { \not\!{v}}}{2 } q^0_v .\end{aligned}\ ] ] to illustrate the details of deriving the effective current @xmath113 , we consider as an example the two - point correlation function @xmath218 where @xmath219 and @xmath220 can be any local operators that may contain @xmath221 ( @xmath47 ) and @xmath222 ( @xmath48 ) on the left ( right ) of the dirac matrixes @xmath88 . @xmath223 and @xmath224 are obtained from @xmath219 and @xmath220 by replacing @xmath221 and @xmath47 in the operators with @xmath225 and @xmath226 , respectively . using the propagator ( [ propagatorq0v ] ) for field contraction and applying the integration by parts , the two - point function ( [ 2pointfunction ] ) becomes @xmath227 \nonumber\\ & & \hat{o}^0_2(w^{-1}(y , v)\gamma_2 w(y , v''))_{(y ) } q^0_{v''}(y ) \nonumber\\ & = & \int d^4y \bar{q}^0_{v'}(x ) \hat{o}^0_1(w^{-1}(x , v')\gamma_1 w(x , v))_{(x ) } \int \frac{d^4k}{(2\pi)^4 } e^{-ik\cdot ( x - y ) } \frac{i}{\bar{\lambda } } \sum ^{\infty}_{n=0 } \big ( -\frac{i}{\bar{\lambda } } { v\hspace{-0.2cm}\slash}v\cdot \partial_{(y ) } \big)^n \nonumber\\ & & \hat{o}^0_2(w^{-1}(y , v)\gamma_2 w(y , v''))_{(y)}q^0_{v''}(y ) \nonumber\\ & = & \bar{q}^0_{v'}(x ) \hat{o}^0_1(w^{-1}(x , v')\gamma_1 w(x , v))_{(x ) } \frac{i}{\bar{\lambda } } \sum ^{\infty}_{n=0 } \big ( -\frac{i}{\bar{\lambda } } { v\hspace{-0.2cm}\slash}v\cdot \partial_{(x ) } \big)^n \nonumber\\ & & \hat{o}^0_2(w^{-1}(x , v)\gamma_2 w(x , v''))_{(x)}q^0_{v''}(x ) \nonumber\\ & = & \bar{q}_{v'}(x ) \hat{o}_1(\gamma_1)_{(x ) } \frac{i}{\bar{\lambda } } \sum ^{\infty}_{n=0 } \big ( -\frac{i}{\bar{\lambda } } { { d\hspace{-0.25cm}\slash_{\|}}}_{(x ) } \big)^n \hat{o}_2(\gamma_2 ) _ { ( x ) } q_{v''}(x ) \nonumber \\ & = & \bar{q}_{v'}(x ) \hat{o}_1(\gamma_1)_{(x ) } \frac{i}{i{{d\hspace{-0.25cm}\slash_{\|}}}_{(x ) } + \bar{\lambda } } \hat{o}_2(\gamma_2)_{(x)}q_{v''}(x ) , \end{aligned}\ ] ] the final expression of which can be obtained directly from ( [ 2pointfunction ] ) with using ( [ propagator ] ) as @xmath111 field propagator . e. v. shuryak , phys . lett . * b 93 * , 134 ( 1980 ) ; nucl . phys . * b 198 * , 83 ( 1982 ) . s. nussinov and w. wetzel , phys . rev . * d 36 * , 130 ( 1987 ) . n. isgur and m. wise , phys . * b 232 * , 113 ( 1989 ) ; * b 237 * , 527 ( 1990 ) ; * b 206 * , 681 ( 1988 ) .

we have recently proposed an ultra - strong " version of the equivalence principle ( ep ) that is not satisfied by standard semiclassical gravity . in the theory that we are conjecturing , the vacuum expectation value of the ( bare ) energy momentum tensor is exactly the same as in flat space : quartically divergent with the cut - off and with no spacetime dependent ( subleading ) terms . the presence of such terms seems in fact related to some known difficulties , such as the black hole information loss and the cosmological constant problem . since the terms that we want to get rid of are subleading in the high - momentum expansion , we attempt to explore the conjectured theory by ir - completing " gr . we consider a scalar field in a flat frw universe and isolate the first ir - correction to its fourier modes operators that kills the quadratic ( next to leading ) time dependent divergence of the stress energy tensor vev . analogously to other modifications of field operators that have been proposed in the literature ( typically in the uv ) , the present approach seems to suggest a breakdown ( here , in the ir , at large distances ) of the metric manifold description . we show that corrections to gr are in fact very tiny , become effective at distances comparable to the inverse curvature and do not contain any adjustable parameter . finally , we derive some cosmological implications . by studying the consistency of the canonical commutation relations , we infer a correction to the distance between two comoving observers , which grows as the scale factor only when small compared to the hubble length , but gets relevant corrections otherwise . the corrections to cosmological distance measures are also calculable and , for a spatially flat matter dominated universe , go in the direction of an effective positive acceleration .

0907.0765

modifications of general relativity ( gr ) on the largest scales have been advocated in order to give account for the present acceleration of the universe . attempts in this direction @xcite typically maintain the metric manifold structure of gr and change the dynamics of the gravitational field by effectively giving the graviton a small mass . it is thoroughly assumed that any infra - red ( ir ) modification of gravity must bring in at least one new mass parameter , corresponding to the scale at which such a modification becomes effective . however , that is not necessarily the case . in @xcite we put forward an infra - red modification of gr that does not involve any new adjustable parameter but that might require a substantial reassessment of the present understanding of gravity in the semi - classical regime . in this paper we review and refine the model sketched in @xcite and derive some of its cosmological implications . we start from usual semi - classical gravity , where matter fields are quantized on a curved background manifold , and we modify in the ir the operators of the matter field theory . in a cosmological setup , operators corresponding to fourier modes of physical momentum @xmath0 are corrected by order @xmath1 , @xmath2 being the hubble parameter . the coefficient of such a modification is fixed by imposing an ultra - strong " version of the equivalence principle , which we are going to recall in the following . we argue that such ir - modified operators fail to be supported by a metric manifold structure at large scales . hence , the proposed modification seems to indicate a breakdown , for systems of size comparable to the inverse curvature , of the geometrical description of space - time as a metric manifold . it is known that there is a strict connection between the geometric properties of a manifold and the spectrum of the differential operators @xcite or the algebra of functions @xcite therein defined ; such abstract characterizations have occasionally been used for generalizing common geometrical concepts and the description of spacetime itself @xcite . however , so far , attempts in this direction have always been applied to the uv and intended to modify spacetime at the smallest scales . although perhaps less intuitive , a modification effective in the ir and at length scales set by the curvature is appealing on both theoretical and observational grounds : * the standard paradigm gr + matter fields " presents some difficulties that are arguably uv - insensitive , such as the cosmological constant problem @xcite and the black hole information paradox @xcite . such difficulties appear already at the semi - classical / low - energy effective level and therefore it seems unlikely that any modification at the plank scale or uv - completion of the theory might possibly address them . questioning the ir - side of the picture seems thus a valuable alternative . * the apparent acceleration of the universe is described in gr as something kicking in at some late - time epoch in the cosmological evolution and related to an extremely tiny mass scale . such a scale is adjusted _ ad hoc _ in the models of dark energy " , which , therefore , generically suffer from fine - tuning and coincidence " problems . the alternative view suggested here is based on the simple remark that any high redshift object is placed from us at distances of order the inverse curvature ( i.e. the hubble radius @xmath3 ) . hence , modifying the usual framework at a length scale set by the curvature rather than fixed _ a priori _ will systematically affect any cosmological observation at high redshift , regardless of when such an observation takes place and without the need of any external mass parameter . before summarizing the cosmological implications of the proposed deformation let us review the main theoretical idea : the ultra strong equivalence principle @xcite . consider the vacuum expectation value of the local energy density of a ( massless ) field according to standard semi - classical gravity ; this can be expanded at high momenta as @xmath4 & = & \ \ { \rm local\ terms}\ \ + \ \ { \rm non \ local\ terms}. \nonumber\end{aligned}\ ] ] here spatial homogeneity has been assumed for simplicity and the @xmath5s are functions of time of appropriate dimensions ( see eq . below , or reference @xcite , for the explicit expression in the case of a massive scalar in a flat frw universe ) . the stress tensor renormalization is a well - defined prescription to make sense of the infinities appearing in : the local terms can be subtracted by local gravitational counterterms ( cosmological constant , newton constant and higher order local operators ) . the finite non - local pieces left represent the genuine particle / energy content of the choosen vacuum " state . it is interesting to note that some of the difficulties afflicting the standard low - energy framework for gravity seem to be related to the spacetime dependent subleading terms ( the @xmath5s in the case considered above ) in the energy momentum vev expansion . the black hole information loss paradox is related to the presence of the non - local pieces , as they are the ones responsible for particle creation and for the evaporation of the black hole . we argue that also the cosmological constant problem is related to the presence of those terms , at least if we restrict the analysis to the level of non - quantized semi - classical gravity . such a problem appears in the renormalization procedure that we described earlier as the huge amount of fine tuning that is required to match the first quartically divergent piece of with its observed value . however , the quartically divergent term is not particularly worrisome by itself , essentially because it is a constant . in flat space we deal with it , by normal ordering or just by subtracting a constant , albeit infinite , contribution . in other words , as brutal as it might appear , the vacuum does not gravitate " is a well defined prescription , as long as the contribution to be neglected is a spacetime - independent scalar . we are arguing that the main source of trouble is not the first term in , but the remaining spacetime dependent pieces , as they make the normal ordering prescription meaningless and the renormalization procedure necessary in the first place . based on the above reasoning and encouraged by few ( more philosophical ) arguments explained below in sec . [ sec2 ] , the conjecture at the basis of this paper is that _ there exists a theory that resembles standard semiclassical gravity at small scales but in which , as in flat space , all time - dependent pieces of ( the @xmath5s ) at any point - event , and for some vacuum " state , just do not exist_. since the terms that we want to get rid of are effectively ir with respect to the first quartically divergent part and can be calculated , we should be able to probe this theory by ir - completing standard semi - classical gravity . it is clearly not guaranteed that a consistent theory with the required properties exists in full generality beyond the example considered here ; if it does , its formulation will certainly require a major theoretical breakthrough . giving up the metric manifold structure and dealing only with quantum operators constitutes a significant technical difficulty . on the other hand , compared to some of the subtleties of standard qft on curved spacetime @xcite , the physical picture that we are insinuating is somewhat simplified , made trivial . we are conjecturing the possibility of describing , locally , the physical phenomena in exactly the same way as we do in flat space , and that the large scale - properties of spacetime have no effect whatsoever inside einstein s free - falling elevator . this is the spirit of the equivalence principle , that we are upgrading to the level of semi - classical gravity and making ultra - strong " @xcite . the local resemblance to flat space is pushed to the level of the quantized fields , whose energy momentum tensor vev at any point - event is conjectured to amount , for some unambiguously chosen vacuum state , to the constant value @xmath6 . in this paper we try to explore the conjectured theory through a sort of taylor expansion around gr , the zeroth order being a flat frw universe with a given scale factor @xmath7 and a spectator free scalar field in it . we isolate the first correction to the scalar field fourier operators that reabsorbs the quadratic divergence @xmath8 in . in a similar fashion , if we did not know gr , we might at least try to explore it by expanding around flat space at a given point in riemann normal coordinates . the proposed modification is first introduced , in sec . [ sec4.5 ] , through modified commutation relations between the global creators and annihilators of the field , @xmath9 = \delta^3(\vec{n } - \vec{n}')\left(1 - \frac{h^2 a^2}{2 n^2 } + { \rm higher\ orders}\ , \right).\ ] ] in the above , @xmath10 is the comoving momentum and the form of the correction term is dictated by the ultra - strong equivalence principle . at this order of approximation , however , there seems to be another more practical way to see the proposed modification , which we show in sec . [ sec5 ] . after a simple rescaling , one can deal with the usual manifold - operators " , labeled by the manifold - momenta " @xmath10 , that satisfy the usual commutation relations , @xmath11 = \delta_{{\vec{n}},{\vec{n}}'}$ ] . the proposed modification then boils down to introducing a slight mismatch between the global , manifold " label @xmath10 and the physical ( comoving ) momentum @xmath12 that locally defines the infinitesimal translations and the derivatives of the local fields : @xmath13 note that the comoving physical momentum is time dependent in this model , although only negligibly at sub - hubble scales . the modified dispersion relations come from requiring the cancellation of the quadratically divergent time - dependent piece @xmath14 in the high momentum expansion at some given point @xmath15 " . in sec . [ secglobal ] we try to make sense , globally , of such modified dispersion relations and explore this spacetime a finite distance away from @xmath16 by using the accordingly modified translation operator . as much as the physical comoving momentum is conserved only approximately , it turns out from the analysis of sec . [ secglobal ] that the physical distance @xmath17 between two comoving observers grows as the scale factor @xmath7 only if that distance is much smaller than hubble . at large scales , the expansion rate effectively detaches from the local expansion @xmath7 . this is the most striking signal of a breakdown of the manifold description . the comoving distance @xmath18 of two faraway comoving observers , instead of being constant , satisfies @xmath19 the effect is clearly negligible at sub - hubble scales but will affect the observed cosmological dynamics at high redshift . the cosmological implications / predictions of the present model , obtained by applying the proposed ir modification to a spatially flat frw universe , are derived and discussed in sec . [ seccosm ] and can be summarized as follows : * comoving momenta @xmath12 are conserved only when they are very high compared to hubble , i.e. when @xmath20 . more generally , what is conserved is the quantity + @xmath21 * the physical distance @xmath17 between two comoving observers grows in time proportionally to the scale factor @xmath7 only if the two are well within the hubble radius . more generally , their distances at two different times @xmath22 and @xmath23 are related by + @xmath24.\ ] ] * the angular - diameter distance is given , in terms of the redshift , by + @xmath25 + in a matter dominated universe , @xmath26 is the solution of @xmath27 with initial condition @xmath28 . * the luminosity distance @xmath29 , at least at this order of approximation , is related to @xmath30 in the usual way , @xmath31 . the cosmological distances derived with equation in a matter dominated einstein - de sitter universe are corrected in the direction of an apparent acceleration " and show some resemblance to those of a @xmath32cdm universe ( figs . [ fig1 ] and [ fig2 ] , sec . [ seccosm ] ) . a more quantitative analysis will follow @xcite . the paper is organized as follows . the general motivations and the main idea are given in secs . [ sec2 ] and [ sec3 ] which are basically taken from the shorter note @xcite . in sec . [ sec4 ] we review the standard calculation of for a free scalar in a flat frw universe . in sec . [ sec4.5 ] we modify the fourier operators in order to get rid of the quadratic divergences there encountered . modified dispersion relations are introduced in sec . [ sec5 ] as an essentially equivalent , but more practical , way to look at the proposed modification . in sec . [ secglobal ] we attempt to make sense , globally , of the proposed modification and study its implications on large scales . in sec . [ seccosm ] we finally derive some cosmological consequences , such as the modified luminosity distance . potential problems and future perspective are finally summarized in sec . [ sec9 ] . the equivalence principle ( ep ) can be formulated simply as follows : _ inside a sufficiently small free - falling elevator you do not see the ( classical ) effects of gravity_. amusingly , such a cornerstone of modern physics is actually stating what ( where ) gravity is not , rather than what ( where ) gravity is ! among the many celebrated implications of general relativistic physics , the view that we aim to stress here is that ep forces us to consider and describe gravity as an ir phenomenon , whose effects are visible only outside the free - falling elevator . how ep turned into a consistent theory is well known : gravity is beautifully encoded in gr as the geometry of the physical space - time and therefore its effects are automatically suppressed within those systems that are much smaller than the inverse curvature . by changing ( curving ) the large - scale structure of spacetime , gr leaves the smallest systems free of classical gravitational effects . notably , the ir scale where non - gravitational physics breaks down is not a parameter of the theory , but is set by the local curvature @xmath33 . schematically , in three dimensions , the area of a two - sphere of radius @xmath34 and volume @xmath35 receives corrections from the flat - space expectation of the type @xmath36 the effects of gravity , originally banned from the free - falling elevator , reappeared , after the developments of quantum theory , through what one might call the back door . the fields quantized on a curved manifold are sensitive to the global properties of spacetime because their modes are defined on the whole of it . as a result , inside the elevator , you will generically experience , and possibly detect with your local instruments , the first excited level is typically at energy @xmath37 above the ground state and can not be excited by photons of much lower energies . however , the structure of the higher energy meta - stable states is often very rich , and energy differences among such states can be arbitrarily small . an atom , for example , has extremely fine energy transitions that , in principle , allow to detect " photons of wavelengths much larger than its size . ] , particle creation because of non - local gravitational effects . clearly , the process of quantum particle creation does not contradict ep , which was formulated within the framework of classical physics . nevertheless , it is tempting to try a different path than the one historically followed : rather than imposing field quantization on top of a curved manifold , here we attempt to upgrade the equivalence principle and extend it to the quantum phenomena . thus , we will consider a stronger version of ep , in which all the effects of gravity are definitely forbidden inside the elevator , including the quantum effects that in the standard semi - classical treatment lead to particle creation . effectively , what we want is that _ for each matter field or sector sufficiently decoupled from all other matter fields , there exists a state , the vacuum " , that is experienced as empty of particles by each free - falling observer . _ following the discussion given in the introduction , a possible concrete strategy would be trying to reabsorb in the ir the non - local terms inside after the stress energy tensor has been renormalized by usual means . however , we will follow a different , more radical , plan . we will circumvent the usual procedure of stress - tensor renormalization and just try to cancel the time - dependent terms of the _ bare _ energy momentum tensor ( all terms inside except the first ) . in fact , the time - dependent pieces can arguably be reabsorbed with some ir modification , because they are effectively ir with respect to the quartically divergent term . moreover , bypassing the usual procedure of stress tensor renormalization might , at the same time , shed some new light on the cosmological constant problem , at least at the level of non - quantized semi - classical gravity . if , in the ir - completed theory , all the time dependent terms just do not exist , then we do not need to renormalize the stress tensor anymore . of course , we are still left with the leading spacetime - constant quartic divergence , but we can live with it and treat it , as we do in flat space , by normal ordering . finally , it is interesting to reabsorb the quadratic divergence in the ir , rather than with a local counterterm , because the required modification , as we are going to show , is of the right order of magnitude to give interesting cosmological implications . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * ultra strong equivalence principle : * for each matter field or sector sufficiently decoupled from all other matter fields , there exists a state , the vacuum " , for which the expectation value of the ( bare ) energy momentum tensor reads the same as in flat space , regardless of the configuration of the gravitational field . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ as a rather obvious warning to the reader we mention that , while ep is extremely well tested ( see e.g. @xcite ) , the proposed ultra strong version is not . experimental hints might actually be pointing against it : the mechanism of quantum particle creation is successfully exploited by inflationary models in order to produce the primordial spectrum of cosmological fluctuations . the present knowledge of the earliest universe is , however , too model dependent ( and largely based on the assumption that such a mechanism is at work ) for that to be considered as an experimental evidence . before describing the model , it is worth examining closely what happens in the standard formulation of semi - classical gravity , where fields are quantized on a curved background manifold . consider a spacetime with a global time foliation labeled by a time parameter @xmath23 . the universe as a whole at time @xmath23 is a three - dimensional manifold @xmath38 in the gr description . the matter quantum fields are instead described by a quantum state living in a hilbert space @xmath39 . thus , if now we consider a region of space ( at time @xmath23 ) of finite volume @xmath35 ( this room , now " ) , that has two complementary descriptions @xcite : it is a _ submanifold _ according to gr and a _ quantum subsystem _ for the quantized fields . the correspondence sub - manifold / sub - system is explicitly realized by the set of local operators @xmath40 of the field theory . those act on the quantum system @xmath39 but have labels @xmath41 living on the three - dimensional manifold @xmath38 . as a consequence , we can take integrals of scalar local operators over some region of volume @xmath35 , @xmath42 which are still operators acting on the total hilbert space @xmath39 . more precisely , @xmath43 acts non - trivially only on the quantum subsystem @xmath44 corresponding to the region of space that has been integrated over , and as the identity on the rest of the system @xmath45 . in this way , the algebra of operators @xmath46 determines @xcite the partition @xmath47 of the quantum system / universe and therefore can be taken as a definition of the region of space @xmath35 . defining regions of space as quantum subsystems @xcite looks like a useless complication in the standard formalism because the correspondence sub - system / sub - manifold is always implicitly at work : such a correspondence is set once and for all by , given the set of local operators @xmath48 . when we say that the metric - manifold description might break - down , we mean , more precisely , that regions of space , still perfectly defined as quantum subsystems , may not have the nice property that the corresponding operators integrate as in . we argue therefore that is valid only in the limit of zero curvature and that operators corresponding to extended regions of space can be written as only up to order @xmath49 , where @xmath33 is some curvature scalar . here is the mnemonic thumb - rule that will guide us in our further developments : @xmath50 .\ ] ] we are deliberately mimicking the general type of corrections that non - extensive geometrical quantities undergo in the transition from flat space to curved space . the idea is to extend this type of behavior also to extensive quantities , that in the standard description are just proportional to the volume . the implications of are quite striking . consider , for instance , a perfectly homogeneous universe . by definition , each comoving observer measures in its surrounding the same energy density @xmath51 . . implies that in that universe , if one starts considering regions of space of hubble size , the total energy inside that region will drastically differ from the three - dimensional integral of the local densities measured by the observers living therein . in a metric manifold , that would be a dramatically non - local effect . in this section we briefly review the quantization of a free scalar field in a frw spatially flat metric manifold . in the spirit of this paper , what we are considering here is therefore the zeroth order approximation of the theory that we are trying to explore . from the action @xmath52 and the metric element @xmath53 we derive the equation of motion for the field , @xmath54 and its energy momentum tensor , @xmath55 , from which we get the expression of the energy density of @xmath56 at time @xmath23 and comoving spatial coordinate coordinate @xmath57 : @xmath58.\ ] ] a dot means derivation with respect to the proper time @xmath23 while a prime denotes derivation with respect to conformal time @xmath59 ; @xmath60 is the hubble parameter . upon quantization in the heisenberg picture , the field can be expanded in creators and annihilators : @xmath61.\ ] ] for reasons that will be clear in the following , we have called @xmath10 ( and not @xmath12 ) the comoving momentum label in this frw space ; @xmath10 is a conserved quantity , related to the proper physical momentum @xmath62 by @xmath63 . the suffix @xmath64 stands for zeroth order " since the creation operators just introduced are intended to be the zeroth order ( metric manifold ) approximation of the operators of the ir - completed theory . from the action one derives the conjugate momentum @xmath65 and from the canonical commutation relations @xmath66 = i\delta^3(\vec{x}-\vec{x}\ , ' ) , \ ] ] one derives the canonical commutators for the zeroth order global operators @xmath67 = \delta^3(\vec{n } - \vec{n}\ , ' ) \ , .\ ] ] it is customary to choose @xmath68 as the operator that always annihilate the vacuum . the chosen vacuum state is therefore implicitely characterized by the choice of the mode functions @xmath69 , that , by and , satisfy @xmath70 where @xmath71 there is an extensive literature about the correct vacuum choice , i.e. , the appropriate solutions of . the classical argument in favor of the _ adiabatic vacuum _ @xcite is its resemblance to the flat space behavior for those modes that are deep inside the hubble scale . within the standard framework , the adiabatic vacuum is as close as we can get to implement the general ideas described in the previous sections and , therefore , will be our starting point . the adiabatic vacuum corresponds to the choice @xmath72 where the function @xmath73 can be approximated at arbitrary adiabatic order with a formal wkb expansion . for the next to leading order in the momentum @xmath74 expansion we need no more than the next to leading adiabatic order @xcite : @xmath75 where the function @xmath76 can be derived from a wkb expansion and amounts to @xmath77 = -\frac{a^2}{n^2}\left(\frac{\ddot a}{a } + h^2\right)\left[1 + { \cal o}\left(\frac{m^2}{n^2}\right)\right].\ ] ] the vacuum expectation value of the energy density at a given point is calculated straightforwardly : @xmath78[a_n^{(0 ) } , a_{- n'}^{(0)\dagger}]\\ & = & \frac{1}{4 \pi^2 } \int_0^\infty n^2 d n \ , ( |{\dot \psi}_n|^2 + \omega_n^2 |\psi_n|^2),\end{aligned}\ ] ] where commutation relations have been used in the second line . from and then we get @xmath79\\ & = & \frac{1}{a^3 \omega_n}\left[\omega_n^2 + \frac{1}{8 } \left(3 h + \frac{{\dot \omega}_n}{\omega_n}\right)^2 + { \cal o } ( n^{-2})\right ] . \nonumber\end{aligned}\ ] ] note that @xmath76 drops from the expression of the energy momentum tensor vev if we do not consider beyond the quadratic divergence . in other words , all we need of @xmath80 is its expression at leading adiabatic order , @xmath81 . we finally get @xmath82,\ ] ] to be opposed to the flat space result @xmath83 in order to enforce the ultra strong equivalence principle we want to get rid of the terms in the local energy momentum vev that we normally would not find in flat space . in this paper we deal only with the time dependent quadratic divergence , the second term inside the square brackets in . first of all , we focus our attention on a given comoving observer / trajectory at any proper time @xmath23 and , say , at point @xmath16 . for brevity , when we write @xmath84 , we refer to a region of space around that trajectory small enough that the manifold description holds and we can define spatial gradients and local commutation relations between neighboring points . we want to strictly preserve the local dynamics of the scalar field . our starting points will be , therefore , the field equations in the heisenberg picture for the field operator in @xmath16 , @xmath85 and the hamiltonian density @xmath86 at point 0 , which , again , is an explicitely time dependent operator : @xmath87 .\ ] ] while and are both elegantly derivable from the same action , for the moment we are forced to consider and as two basic and in principle independent ingredients of this novel framework . note moreover that , in a sort of pre - geometric fashion , we can take the above equations as the implicit definition of the local expansion @xmath7 . by homogeneity , we will require in sec . [ secglobal ] that the translated fields at some other point in space still obey with the same scale factor @xmath7 . commutation relations between local quantities are strictly preserved : @xmath88 = i\delta^3(\vec{x}-\vec{x}'),\ ] ] ( the momentum conjugate to @xmath56 can still be defined completely locally in terms of the lagrangian density in @xmath16 : @xmath89 ) . as before , we formally expand the field in fourier modes : @xmath90,\ ] ] where the suffix @xmath91 means that we now want to consider the modes corrected at first order in the ir expansion . so far the field is only defined at @xmath92 ; the @xmath57 labels at the exponent in the equation above are there just to remind that , if we take space derivatives of the such a local field , a factor @xmath93 drops . the crucial observation now is that the commutator between @xmath94 and @xmath95 is proportional to the total volume and therefore will receive the postulated corrections . by looking at and , it is immediate to note that the quadratic divergence is reabsorbed if we impose the commutation relation @xmath9 = \delta^3(\vec{n } - \vec{n}')\left(1 - \frac{h^2 a^2}{2 n^2}\right).\ ] ] the correction is in fact of the expected form . this is perhaps more evident in the approach followed in @xcite where a compact toy - universe of total linear size @xmath96 much smaller than the hubble length was choosen to begin with . in that framework , fourier labels assume discrete values and the correction @xmath97 is clearly of order @xmath98 or smaller . although we previously referred to modifications at scales of the inverse curvature , we find that , more precisely , it is the extrinsic curvature which appears to regulate the correct modification . equivalently , we can just say that the annihilators are modified with respect to gr by @xmath99 the corrected fourier modes are effectively time dependent . at least at this order of approximation , however , the time dependence is just a multiplicative factor that leaves the vacuum untouched . note also that , by applying the postulated local equation , the time dependence of @xmath100 introduces additional spurious terms inside . those , however , are of higher order in the @xmath74 expansion and therefore do not affect the present result . finally , inverse fourier transforms ( such as inverting by integrating over @xmath57 ) are not allowed . following the reasoning of section [ sec3 ] , we argue that whenever we take integrals @xmath101 of operators over a volume @xmath35 we are making an error of order @xmath102 . so far we concentrated only on local operators at some given point @xmath15 and made connection with the global fourier modes . a crucial ingredient to understand the large - scales properties of this deformed spacetime is the momentum operator constructed with the modified fourier modes , @xmath103 in fact , we are not provided with a metric element that we can integrate over , but we can exploit the assumed homogeneity of this spacetime and define a translation operator by simply exponentiating . by applying the translation operator , in section [ secglobal ] we are able to define the local operators a finite distance away from @xmath15 . the form of the momentum operator suggests another way to look at the proposed deformation . we can write in terms of the usual gr operators @xmath104 , satisfying the usual commutation relations , and modify the momentum labels instead : @xmath105 where , at this order of approximation , @xmath106 we are trading modified fourier operators for modified dispersion relations . in other words , in this equivalent , but more practical , formulation , we are postulating a slight mismatch between the metric - manifold " labels @xmath10 and the physical momenta @xmath12 that locally define the infinitesimal translations and the derivatives of the local fields . since @xmath10 is the usual comoving manifold " label , we assume that it is time - independent . moreover , since we will use only the usual manifold " creators operators @xmath104 from now on , we drop the the suffix @xmath64 to simplify the notation . the recipe can finally be summarized as follows : * there are two different ( comoving ) fourier coordinates : the manifold "- coordinate @xmath10 and the locally defined physical momentum @xmath12 . the integration measure in fourier space is flat in the @xmath74-coordinates . moreover , during time evolution , @xmath107 is conserved , @xmath12 is not . the relation between the two is @xmath108 * the local field in @xmath109 is expanded in fourier modes as follows : @xmath110 \equiv \frac{1}{(2 \pi)^{3/2}}\int d n^3 \phi_{\vec k}(t ) e^{i \vec{k } \cdot \vec{x}}\ ] ] and the @xmath111 satisfy the usual commutation relations : @xmath112 = \delta^3(\vec{n } -\vec{n}').\ ] ] + again , the @xmath57 coordinate in extends only as far as is needed to take the space derivatives of the field in @xmath15 . when a derivative is taken , a factor of @xmath12 , instead of @xmath10 , drops . * the local equation for @xmath113 , , fixes the equations for the functions @xmath114 : @xmath115 where @xmath116 and @xmath12 is given in . the solution of the above equations are choosen to correspond to the adiabatic vacuum ( in the physical momentum @xmath0 ) , i.e. @xmath117 the function @xmath118 can be approximated at arbitrary adiabatic order as before . the only difference is that @xmath119 should substitute @xmath120 in and . * the energy density in @xmath121 is expressed in terms of the local field @xmath113 in the usual way ( equation ) , @xmath122 .\ ] ] the above equations can be used to calculate the vacuum expectation value of @xmath123 . by direct substitution of into we get @xmath124[a_n , a_{- n'}^{\dagger}]\\ & = & \frac{1}{4 \pi^2 } \int_0^\infty n^2 d n \ , ( |{\dot \psi}_k|^2 + \omega_k^2 |\psi_k|^2),\end{aligned}\ ] ] by substitution of we then get @xmath125 , \\ & = & \frac{1}{4 \pi^2 a^3 } \int_0^\infty n^2 d n \ , \left[\omega_n + { \cal o}(n^{-3})\right]\end{aligned}\ ] ] note that the quadratic divergences are here reabsorbed just be re - expressing @xmath126 in terms of the appropriate flat measure " time - independent fourier coordinates @xmath74 . so far , we have defined local operators only in @xmath109 and made connection with the global ones . however , we want to be able to move in this space arbitrarily far in any direction and define local operators also elsewhere . we are provided with a momentum operator , @xmath127 where , again , @xmath128 since we are assuming homogeneity , we obtain a global translation operator by exponentiation , @xmath129 thus , we can define local operators at a finite distance from @xmath109 by the action of @xmath130 : @xmath131 where @xmath132 stands for at distance @xmath133 if one keeps going in the @xmath134 direction " and @xmath133 is made up by all the infinitesimal steps that have been taken in order to go from one point to another and is therefore the ( comoving ) proper distance . the momentum operator is given in - only approximately . therefore , also should be trusted only as far as @xmath135 is smaller than the hubble radius . in the present case , because of homogeneity , translations in different directions commute with each other , so we can label the local operators with three parameters and describe , approximately , a finite patch of universe around @xmath109 . explicitely , @xmath136.\ ] ] although the above equation has a familiar form , the labels @xmath137 should really be thought here as the parameters of the translation group . we will soon show that , if we use them as integration variables at @xmath138 const . , as we would do on a standard frw flat space , we make an error of order @xmath139 . it is interesting to calculate , at some given time @xmath23 , the canonical commutator between @xmath140 and the field itself at a distance . by using the standard commutation relations we obtain @xmath141 = - \frac{a^3}{(2\pi)^3 } \int d^3 n \ , \left[\dot{\psi}_k^ * \psi_k e^{i \vec{k } \cdot \vec{\lambda } } - \dot{\psi}_k \psi_k^ * e^{- i \vec{k } \cdot \vec{\lambda}}\right ] .\ ] ] by using the wronskian condition @xmath142 , immediately derivable from , and we get , up to higher orders in @xmath143 , @xmath141 = -\frac{i}{(2\pi)^3 } \int d^3 n\ , e^{- i\ , \vec{\lambda}\cdot \vec{n } \left(1- \frac{h^2 a^2}{2 n^2}\right ) } \simeq -\frac{i}{(2\pi)^3 } \int d^3 n\ , e^{- i\ , \vec{\lambda}\cdot \vec{n } } \left(1 + i \vec{\lambda}\cdot \vec{n } \frac{h^2 a^2}{2 n^2}\right),\ ] ] where we have expanded the exponential in the last expression . the first term gives the usual delta function . the last term gives @xmath144 in this way we obtain the first correction to the canonical commutator , @xmath145 = -i\left(\delta^3(\vec{\lambda } ) + \frac{1}{8 \pi } \frac{h^2 a^2}{\lambda } \right).\ ] ] the above relation finally gives a more precise sense to . at face value , is a signal of a non - locality that becomes more and more important at large scales . but again , that is the interpretation that we would have of on a metric manifold . in section [ sec3 ] we argued that regions of space at a given time can still be associated with a well defined algebra of averaged " operators of the matter theory , but that those operators can only approximately be expressed as integrals over a metric manifold . the expression in is in fact a three - dimensional density and the correction term is just signaling the mistake that we make if we naively take the integral @xmath146 \ = \ -i \left[1 + \frac{1}{4 } h^2 a^2 \lambda^2 + { \cal o}(h^2 v^{2/3})^2 \right]\ ] ] and use @xmath133 as a flat - measure integration variable . we nicely recover the expected pattern . it is equally interesting to calculate @xmath147 $ ] . first , we note that there is a potential ambiguity in defining the time derivative of a displaced operator . by deriving we get @xmath148 \\ & + & i \frac{a^3}{(2 \pi)^{3/2 } } \int d^3n\ , ( \vec{k } \cdot \vec{\lambda})\dot{}\ \left[\psi_k \ e^{i \vec{k } \cdot \vec{\lambda } } a_{\vec{n } } - \psi^*_k \ i \vec{k } \cdot \vec{\lambda } } a^\dagger_{\vec{n } } \right ] . \label{ambig}\end{aligned}\ ] ] the second line in the above equation is there because @xmath0 is time dependent . however , if we just applied the translation to @xmath149 , instead of deriving @xmath150 , those terms would not be there . therefore , for consistency , we need to make them ineffective at the required order of approximation . in other words , we have to impose that @xmath151 = - [ \pi(0),\phi(\vec{\lambda})]$ ] . because of the second line of , the commutator between @xmath152 and @xmath153 gives @xmath154 = - [ \pi(0),\phi(\vec{\lambda } ) ] - 2i \frac{a^3}{(2\pi)^3 } \int d^3 n \ , e^{- i \vec{k } \cdot \vec{\lambda } } \left| \psi_n \right|^2 ( \vec{k } \cdot \vec{\lambda})\dot { } \ , \ ] ] the last term being the spurious contribution . in order to get rid of it , we have to make the comoving physical distance @xmath155 also time dependent . this effectively means that , after an infinitesimal time step @xmath156 , we have to reconsider the field translated , from @xmath84 , not by the same comoving distance @xmath133 , but by a slightly different amount . at high momenta / small distances , since @xmath157 , the integral in the second term of reads @xmath158\ , .\ ] ] we make the ansatz @xmath159 , @xmath160 being a number to be determined . we get @xmath161 the last integral can be regularized by setting @xmath162 and taking the @xmath163 limit only after deriving with respect to @xmath164 . the result is null for @xmath165 , which fixes the time dependence of @xmath133 : @xmath166 the above formula is one of the main results of this paper and can be rephrased as follows . in this homogeneous universe , in the vicinity of each point , there is a local expansion given by the same scale factor @xmath7 . in essence , @xmath7 rules the evolution equations for the local operators and determines the rate at which the distance between two comoving observers grows , as far as such a distance is extremely small compared to hubble . in that limit , the comoving distance @xmath133 is conserved . however , if you pick up a pair of comoving observers further apart , their distance grows slightly differently , which means that the expansion rate on large scales is effectively dependent on the distance . the modified expansion rate is given , at first order , by . this effect , which is the most striking signal of the breakdown of the manifold description , is clearly negligible well within the hubble scale . cosmology is where the predicted departures from gr are more relevant , since every high red - shift object is located from us at distances comparable to the inverse curvature . . can be integrate straightforwardly : @xmath167.\ ] ] equivalently , the proper distance @xmath135 between two comoving observers evolves from time @xmath22 to @xmath23 as @xmath168^{-1/2}.\ ] ] according to and the distance between two comoving observers grows slightly less than in a frw universe with scale factor @xmath7 . we now want to calculate the corrected " trajectory @xmath169 for a light ray , where @xmath170 is its comoving distance and @xmath59 the conformal time . a light ray passing through @xmath109 satisfies @xmath171 . however , on top of the usual contribution , the trajectory of a light ray receives a correction from the modified global expansion : @xmath172 now we specialize to a matter dominated universe for simplicity . therefore , we set @xmath173 and we get @xmath174 normally , at the rhs we would simply have @xmath175 and , again , the second term should be considered just as the first order correction . we could nt find an analytic solution to the above equation , and therefore we will solve it numerically . with respect to an einstein - de sitter universe , a light ray propagating forward in time will go a shorter way . a light ray that we are detecting now , on the other hand , is actually coming from farer . for a light ray propagating back in the past the above equation becomes , in terms of the redshift , @xmath176 the above should be solved with initial condition @xmath28 and will give some @xmath26 to be used in what follows . note that , as in the normal framework , the hubble parameter @xmath177 is just a total normalization factor for cosmological distances . light rays going backward in time in any direction describe , at some redshift @xmath178 , the surface of a two - sphere of area @xmath179 . therefore , the angular diameter distance is @xmath26 multiplied by @xmath180 : @xmath181 the redshift of the frequences , on the other hand , is basically unmodified in this framework , @xmath182 . it is true that the comoving momentum @xmath12 is not strictly conserved , but the effect is negligible for momenta much bigger than @xmath2 . + obtaining the formula for the luminosity distance looks , in the present framework , slightly less straightforward . we discuss its derivation in the appendix where we argue that , at least of this order of approximation , angular diameter and luminosity distances are related by the usual formula @xmath183 we plot the obtained luminosity distance as a function of the redshift in fig . [ fig2 ] for different values of the hubble parameter . + our correction makes cosmological distances larger than in a usual einstein - de sitter universe . this goes in the right direction of a @xmath32cdm universe , at least qualitatively ( figs . [ fig1 ] and [ fig2 ] , left panel ) . however , this effect kicks in only at a relatively high redshift . by expanding @xmath26 as @xmath184 it is easy to see that the proposed effect only modifies the curve at forth order in the small - z expansion , while dark energy models usually give already contributions at second order . qualitatively , in order to raise the overall angular diameter distance to the plateau reached by @xmath32cdm , we have to lower the hubble constant @xmath177 , so that @xmath185 starts out with a slightly higher derivative . this makes our curves systematically higher than @xmath32cdm at small redshift . note , on the other hand , that lowering @xmath177 in the absence of this effect produces curves of a completely different shape ( figs . [ fig1 ] and [ fig2 ] , right panel ) . finally , the runaway behaviour of @xmath30 after about @xmath186 ( fig . [ fig1 ] , left panel ) is probably due to the fact that the distance covered by the light ray becomes dangerously comparable to hubble at that redshift ( indeed , @xmath187 ) and therefore higher order corrections to the present model are needed . while the low - redshift discrepancies with the successful concordance model suggest that this mechanism might not be able , by itself , to give full account for the apparent acceleration of the universe ( for a more quantitative analysis see @xcite ) , it is worth noting that the case for an einstein - de sitter universe with a lower hubble constant has occasionally been made in the literature . interestingly , the authors of @xcite claim that a model with @xmath188 and @xmath189 km / s / mpc fits cosmological observations , with the exception of supernovae , better than the concordance model . within the established low - energy effective framework gr + matter fields " @xmath190 there are models that can fit our cosmological data with impressive accuracy . however , those explanations do not come without a cost : we need to assume two epochs of accelerating expansion ( inflation and dark energy " ) , characterized by widely separated mass scales , that require appropriate negative pressure components and , expecially for dark energy , a tremendous amount of fine tuning . with the possible exception of inflation , these difficulties look oblivious to the possible uv - completions of . the candidate theories of quantum gravity , as long as they reproduce at low - energy , are not expected to essentially simplify the picture , although they proved able to suggest new mechanisms , give insights on the cosmological singularity , motivate the introduction of new degrees of freedom / sectors needed for the acceleration etc in this paper we questioned the ir - side of the paradigm gr + matter fields " and attempted to look at as a small distance approximation . the proposed modification is radical because it challenges the usual geometrical description of spacetime on the largest scale , but compelling because it does not contain any adjustable parameter : in a cosmological set - up deviations from gr are calculable and become relevant , at any epoch , at the hubble scale . abandoning the well understood and reliable framework is adventurous to say the least ; but we tried to do that by following a physical principle that seems to address , at least in part , some theoretical difficulties such as the cosmological constant problem and the black hole information paradox and , at the same time , to give observable cosmological predictions at high redshift . the most relevant cosmological effect is a departure at large scales from the local expansion rate ( e.g. eq . ) . the general picture that this model suggests is that the present acceleration of the universe may not be due to any additional dark energy " component but to a systematic breakdown of the geometrical description of spacetime at length scales comparable to hubble . the proposed effect does not kick in at some given time by coincidence " during the cosmological evolution , but is permanently present at any cosmological epoch whenever a comoving observer looks at very far away objects . exploring new directions often opens up potential problems and questions which have been answered here only in part and surely need future work and further double checks . we isolated the ir corrections that apply to a free scalar field in a spatially flat frw universe . it is not clear if those corrections have some sort of universality i.e. if they automatically apply to , or are inherited by , other types of fields . for instance , in a cosmological set - up , applying the ultra - strong equivalence principle to the electromagnetic field is just trivial , because of conformal invariance . for photons , the time dependent terms that we got rid of in the high momentum expansion are absent from the very beginning . if the proposed model is really a modification of the underlying geometry , such a modification should be universal , and this , at present , is not guaranteed . it would be interesting to consider the implications of this model in the very early universe and close to the singularity , where ir corrections of order hubble might become important . we note also that the usual cosmological kinematic of the modes ( exiting " during acceleration , entering " during deceleration ) is modified in the present framework , as the physical comoving momentum is not conserved on large scales ( eq . [ conserved ] ) . it would be interesting to see what are the implications of this effect in the early universe , although higher orders in the ir expansion might be needed in order to have a more complete picture . whether or not the proposed modification applies to reality , our findings are of some interest and can be rephrased as follows : how much should we modify gr in order to renormalize " the stress tensor from the ir side with an ir modification instead than with local gravitational counterterms ? how different from gr would the conjectured theory be ? we have shown that the required modification is indeed extremely tiny . for a free scalar in a cosmological set - up , fourier modes of physical momentum @xmath191 get modified by an amount of order @xmath192 . on cosmological scales , such a modification goes in the direction of an effective acceleration but is still too weak to give full account of supernovae observation @xcite . it will also be interesting to modify , along these same lines , a black hole or a stellar solution , also to check the compatibility with constraints coming from the solar system and strongly self gravitating objects . at the level of order of magnitude estimate , however , we do not expect relevant corrections . the proposed modification affects those phenomena whose relevant length scales are of order the inverse curvature . atomic spectra and and nuclear physics should be untouched , even in the presence of strong gravitational fields . the reason is that the inverse ( ricci ) curvature is roughly of order @xmath193 , @xmath51 being the energy density , and this typically gives a huge number in atomic size units . for what concerns the solar system , its entire dynamics is well within the average curvature radius . if we try to compare the typical solar system distances to , say , the weyl curvature , those are always extremely small . an order of magnitude estimate gives @xmath194 , @xmath170 being the distance from the sun . this means that , within the solar system , and as opposed to high - redshift cosmology , objects are always at distances much smaller than the inverse curvature . we finally note that some recent observational inputs in cosmology may look as an encouragement for ir modifications of gravity . beside the acceleration of the universe , which is already puzzling by itself , some emerging tensions between standard @xmath32cdm cosmology and large - scale observations might be indicating that there is something going wrong with the standard gr description of the largest scales . interestingly , other ir modifications of gr have already been tested @xcite against those discrepancies , which include , among other , the lack of large - scale cmb correlations @xcite , the anomalously large integrated sachs - wolfe cross - correlation @xcite and the abnormally large bulk flows on very large scales @xcite . * acknowledgments * i thank niayesh afshordi , sergio cacciatori , fabio costa , paolo creminelli , olaf dreyer , justin khoury , jennifer lin , christian marinoni , savvas nesseris , burt ovrut , filippo passerini , costantinos skordis , brian schmidt , lee smolin , shinji tsujikawa , andrew tolley and mark wyman . research at the perimeter institute is supported in part by the government of canada through nserc and by the province of ontario through the ministry of research & innovation . the luminosity distance accounts for the flux of energy per unit time arriving from the source . in standard cosmology it is known that there are two in principle - independent dampening cosmological effects for the flux . the usual analysis nicely relies on the wave - particle duality : first , each photon has its energy red - shifted by a factor @xmath195 . second , the number of them per unit time decreases , from emission to absorption , by the same factor . as already discussed , the present framework does not change the first effect : comoving momenta much higher than hubble are conserved and therefore physical momenta redshift in the usual way . here we argue that also the second dampening effect is proportional to @xmath195 , at least at the present order of approximation . in principle , this is not so obvious . we might in fact reason as follows . consider a bunch of photons leaving the source at redshift @xmath178 and propagating towards us . say that , at a given redshift @xmath178 , the comoving distance between the first photon and the last is @xmath196 , where both @xmath197 and @xmath198 are solutions of . in the limit in which those two solutions are very close to each other it is easy to estimate @xmath199 . by writing equation for @xmath200 we get @xmath201 we can now use the equation for @xmath202 and drop higher orders in @xmath203 . this gives an estimate for a small spread @xmath199 around a given solution @xmath169 , @xmath204 by reconverting in redshift this gives @xmath205 numerical integration shows that the above is indeed a very accurate estimate . this way of reasoning would suggest that , on top of the usual spreading effect due to the expansion , there is a sort of refocussing " effect , due to the fact that two infinitely close solutions of tend to get closer and closer to each other as they evolve forward in time . for faraway objects that is a very relevant brightening effect . what is not convincing about the above derivation is that it heavily relies on the estimate of a small spread @xmath203 even when that is calculated very far away from us . we know that the translation operator that we are using is very accurate only in shifting objects relatively close to the center @xmath84 . the error in estimating a small shift @xmath199 at high redshift can be comparatively very large . a more convincing way to proceed is simply to note that two parallel light rays , in the limit when they are very close to each other , should feel only the local expansion @xmath7 . at each time step , by homogeneity , we can put one of the two at @xmath84 and calculate the position of the other . this amounts to re - set , at each time step , @xmath206 in eq . and gives , for the comoving spread @xmath203 , an equation identical to . the effect is negligible each time the spread is much smaller than hubble . we therefore obtain , at least at this order of approximation , the known relation @xmath183 see , among many others , c. deffayet , g. r. dvali and g. gabadadze , `` accelerated universe from gravity leaking to extra dimensions , '' phys . d * 65 * , 044023 ( 2002 ) [ arxiv : astro - ph/0105068 ] ; n. arkani - hamed , h. c. cheng , m. a. luty and s. mukohyama , `` ghost condensation and a consistent infrared modification of gravity , '' jhep * 0405 * , 074 ( 2004 ) [ arxiv : hep - th/0312099 ] ; g. dvali , s. hofmann and j. khoury , `` degravitation of the cosmological constant and graviton width , '' phys . d * 76 * , 084006 ( 2007 ) [ arxiv : hep - th/0703027 ] . f. piazza , `` modifying gravity in the infra - red by imposing an ultra - strong equivalence principle , '' arxiv:0904.4299 [ hep - th ] . d. v. vassilevich , `` heat kernel expansion : user s manual , '' phys . * 388 * , 279 ( 2003 ) [ arxiv : hep - th/0306138 ] . a. connes , _ noncommutative geometry _ , academic press , san diego , ca , 1994 . y. b. zeldovich and a. a. starobinsky , `` particle production and vacuum polarization in an anisotropic gravitational field , '' sov . jetp * 34 * , 1159 ( 1972 ) [ zh . fiz . * 61 * , 2161 ( 1971 ) ] ; s. a. fulling and l. parker , `` renormalization in the theory of a quantized scalar field interacting with a robertson - walker spacetime , '' annals phys . * 87 * , 176 ( 1974 ) . n. d. birrell and p. c. w. davies , `` quantum fields in curved space , '' _ cambridge , uk : univ . pr . ( 1982 ) 340p _ s. schlamminger , k. y. choi , t. a. wagner , j. h. gundlach and e. g. adelberger , `` test of the equivalence principle using a rotating torsion balance , '' phys . lett . * 100 * , 041101 ( 2008 ) [ arxiv:0712.0607 [ gr - qc ] ] . f. piazza , `` glimmers of a pre - geometric perspective , '' arxiv : hep - th/0506124 ; f. piazza , `` quantum degrees of freedom of a region of spacetime , '' aip conf . proc . * 841 * , 566 ( 2006 ) [ arxiv : hep - th/0511285 ] . s. cacciatori , f. costa and f. piazza , `` renormalized thermal entropy in field theory , '' phys . d * 79 * , 025006 ( 2009 ) [ arxiv:0803.4087 [ hep - th ] ] . p. zanardi , d. a. lidar and s. lloyd `` quantum tensor product structures are observable - induced '' , phys . lett . * 92 * , 060402 ( 2004 ) [ arxiv : quant - ph/0308043 ] ; a. blanchard , m. douspis , m. rowan - robinson and s. sarkar , `` an alternative to the cosmological concordance model , '' astron . astrophys . * 412 * , 35 ( 2003 ) [ arxiv : astro - ph/0304237 ] . n. afshordi , g. geshnizjani and j. khoury , `` observational evidence for cosmological - scale extra dimensions , '' arxiv:0812.2244 [ astro - ph ] . j. khoury and m. wyman , `` n - body simulations of dgp and degravitation theories , '' arxiv:0903.1292 [ astro-ph.co ] . g. hinshaw _ et al . _ , `` 2-point correlations in the cobe dmr 4-year anisotropy maps , '' arxiv : astro - ph/9601061 ; d. n. spergel _ et al . _ [ wmap collaboration ] , `` first year wilkinson microwave anisotropy probe ( wmap ) observations : determination of cosmological parameters , '' astrophys . j. suppl . * 148 * , 175 ( 2003 ) ; c. copi , d. huterer , d. schwarz and g. starkman , `` the uncorrelated universe : statistical anisotropy and the vanishing angular correlation function in wmap years 1 - 3 , '' phys . d * 75 * , 023507 ( 2007 ) . a. kashlinsky , f. atrio - barandela , d. kocevski and h. ebeling , `` a measurement of large - scale peculiar velocities of clusters of galaxies : technical details , '' arxiv:0809.3733 [ astro - ph ] ; arxiv:0809.3734 [ astro - ph ] ; r. watkins , h. a. feldman and m. j. hudson , `` consistently large cosmic flows on scales of 100 mpc / h : a challenge for the standard lcdm cosmology , '' arxiv:0809.4041 [ astro - ph ] .

we present the results of a study of a large sample of luminous ( ) lyman break galaxies ( lbgs ) in the redshift interval , selected from a contiguous 0.63 square degree area covered by the ukidss ultra deep survey ( uds ) and the subaru xmm - newton survey ( sxds ) . utilising the large area coverage and the excellent available optical+nearir data , we use a photometric redshift analysis to derive a new , robust , measurement of the bright end ( ) of the uv - selected luminosity function at high redshift . when combined with literature studies of the fainter lbg population , our new sample provides improved constraints on the luminosity function of redshift lbgs over the luminosity range 0.1l ll . a maximum likelihood analysis returns best - fitting schechter function parameters of and for the luminosity function at , and and at . in addition , an analysis of the angular clustering properties of our lbg sample demonstrates that luminous lbgs are strongly clustered (mpc ) , and are consistent with the occupation of dark matter halos with masses of . moreover , by stacking the available multi - wavelength imaging data for the high - redshift lbgs it is possible to place useful constraints on their typical stellar mass . the results of this analysis suggest that luminous lbgs at have an average stellar mass of , consistent with the results of the clustering analysis assuming plausible values for the ratio of stellar to dark matter . finally , by combining our luminosity function results with those of the stacking analysis we derive estimates of mpc and mpc for the stellar mass density at and respectively . # 1to 0pt#1 # 1 # 1 to size # 1 to size = cmr10 scaled2 = cmbx10 scaled2 = cmti10 scaled2 ps . 2 = 6truein galaxies : high - redshift - galaxies : evolution - galaxies : formation

0805.1335

accurately determining the global properties of the high - redshift galaxy population is a powerful method of constraining current models of galaxy evolution , and identifying the sources responsible for reionisation . large , statistical samples of high - redshift galaxies can be identified efficiently using two complementary photometric techniques . firstly , lyman - alpha emitters ( laes ) can be selected via deep imaging with narrow - band filters centred on the redshifted lyman - alpha emission line ( hu , mcmahon & cowie 1999 ) . alternatively , so - called lyman - break galaxies ( lbgs ) can be selected from deep broad - band photometry using the lyman - break , or `` dropout '' , technique pioneered by guhathakurta , tyson & majewski ( 1990 ) . when combined with deep multi - wavelength photometry and spectroscopic follow - up , narrow - band selection of laes is an efficient technique for producing large samples of high - redshift galaxies within narrow redshift intervals , free from significant levels of contamination . by exploiting the wide - field imaging capabilities of suprime - cam on subaru , several recent studies have investigated the number densities , luminosity functions and clustering properties of laes at @xmath21 and @xmath22 ( e.g. ouchi et al . 2008 ; shimasaku et al . 2006 ; taniguchi et al . moreover , at present , the highest - redshift galaxy with spectroscopic confirmation ( @xmath23 ; iye et al . 2006 ) was identified using the narrow - band technique ( see stark et al . 2007a for candidate laes at @xmath24 ) . however , although lae studies have many advantages , they are constrained by the fact that any individual study can only survey a comparatively small cosmological volume and that only a minority of lbgs at high redshift appear to be strong laes ( e.g. shapley et al . therefore , although the selection function of the lyman - break technique is more difficult to quantify , because it includes both laes and non - laes , it should provide a more complete census of the high - redshift galaxy population . consequently , many studies in the recent literature have exploited both ground - based and hubble space telescope ( hst ) data to study the properties of high - redshift galaxies selected via the lyman - break technique . in particular , due to its unparallelled point - source sensitivity , deep hst imaging has allowed studies of the high - redshift luminosity function to reach @xmath25 , and has therefore been crucial to constraining both the normalization and faint - end slope of the luminosity function at @xmath26 ( e.g. dickinson et al . 2004 ; malhotra et al . 2005 ; oesch et al . 2007 ; bouwens et al . 2006 , 2007 , 2008 ; reddy et al . 2007 ) . in terms of best - fitting schechter function parameters the results of these studies have described a fairly consistent picture , with the majority finding a very steep faint - end slope , typically in the range @xmath27 ( although see sawicki & thompson 2006 for a different result at @xmath28 ) , and little evidence for significant evolution in either the faint - end slope or normalization at @xmath29 . although studies based on deep hst imaging have been successful at exploring the faint end of the high - redshift luminosity function , their small areal coverage means that they are not ideally suited to accurately constraining the bright end of the luminosity function . to address this issue studies based on shallower , but wider area , ground - based imaging have been important ( e.g. ouchi et al . 2004a ; shimasaku et al . 2005 ; yoshida et al . 2006 ; iwata et al . 2007 ) . to date , although the existing wide - area , ground - based optical imaging surveys have been very useful for constraining the bright end of the lbg luminosity function in the redshift range @xmath30 , it has been difficult to successfully extend ground - based studies to higher redshift . the reason for this is that without the benefits of hst spatial resolution to exclude low - redshift contaminants ( i.e red galaxies at @xmath31 and ultra - cool galactic stars ) , selecting clean samples of @xmath32 lbg candidates requires near - infrared photometryfilters to quantify the uv slope of a small sample of 12 redshift @xmath33 candidates in the subaru deep field by shimasaku et al . ( 2005 ) is one notable exception . ] to confirm a relatively blue spectral index long - ward of lyman alpha . unfortunately , due to the lack of wide area near - infrared detectors , until very recently it was not possible to obtain suitably deep near - infrared imaging over areas commensurate with the existing wide - area optical surveys . however , with the advent of deep near - infrared imaging over a 0.8 square degree field provided by the ukidss ultra deep survey ( uds ) this fundamental issue is now being addressed . in a previous paper ( mclure et al . 2006 ) we combined the early data release of the uds with optical imaging from the subaru / xmm - newton survey ( sxds ) to explore the properties of the most luminous ( @xmath34 ) galaxies at @xmath35 . in this paper we extend this previous study by using new , deeper , @xmath36 imaging of the uds field to study the properties of a much larger sample of @xmath37 lbgs in the redshift range @xmath1 . the specific aim of this study was to use the large , contiguous area of the uds to provide an improved measurement of the bright end of the uv - selected luminosity function at @xmath9 and @xmath12 , and to measure the clustering properties , and thereby halo masses , of the luminous lbg population at @xmath35 . the structure of the paper is as follows . in section 2 we briefly describe the properties of the near - infrared and optical data - sets used in this study . in section 3 we describe our initial selection criteria and our photometric redshift analysis . in section 4 we describe our adopted technique for estimating the lbg luminosity function . in section 5 we present our luminosity function results , and the results of combining our ground - based data with existing constraints derived from deep hst imaging data . in section 6 we report the results of our study of the clustering properties of the luminous lbg population in the redshift range @xmath3 and link the lbg population to that of the underlying dark matter halos . in section 7 we provide an estimate of the typical stellar mass for luminous @xmath3 lbgs based on a stacking analysis of the available multi - wavelength imaging data . in section 8 we use this information to derive an estimate of the stellar mass function at @xmath9 and @xmath12 and compare our results with the predictions of recent semi - analytic models of galaxy formation . in section 9 we summarise our main conclusions . throughout the paper we adopt the following cosmology : @xmath38 kms@xmath39mpc@xmath39 , @xmath40 , @xmath41 , @xmath42 , @xmath43 . all magnitudes are quoted in the ab system ( oke & gunn 1983 ) . the ultra deep survey ( uds ) is the deepest of five near - infrared surveys currently underway at the uk infrared telescope ( ukirt ) with the new wfcam imager ( casali et al . 2007 ) which together comprise the ukirt infrared deep sky survey ( ukidss ; lawrence et al . 2007 ) . the uds covers an area of 0.8 square degrees centred on ra=02:17:48 , dec=@xmath4405:05:57 ( j2000 ) and is already the deepest , large area , near - infrared survey ever undertaken . the data utilized in this paper were taken from the first ukidss data release ( dr1 ; warren et al . 2007 ) , which included @xmath36 imaging of the entire uds field to @xmath45 depths of @xmath46 ( 1.6diameter apertures ) . the ukidss dr1 became publicly available to the world - wide astronomical community in january 2008 and can be downloaded from the wfcam science archive . the uds field is covered by a wide variety of deep , multi - wavelength observations ranging from the x - ray through to the radio ( see cirasuolo et al . 2008 for a recent summary ) . however , for the present study the most important multi - wavelength observations are the deep optical imaging of the field taken with suprime - cam ( miyazaki et al . 2002 ) on subaru as part of the subaru / xmm - newton deep survey ( sekiguchi et al . the optical imaging consists of 5 over - lapping suprime - cam pointings , and covers an area of @xmath47 square degrees . the whole field has been imaged in the @xmath48 filters , to typical @xmath45 depths of @xmath49 , @xmath50 , @xmath51 , @xmath52 and @xmath53 ( 1.6diameter apertures ) . the reduced optical imaging of the sxds is now publicly available and full details of the observations , data reduction and calibration procedures are provided in furusawa et al . the high - redshift galaxies investigated in this study were selected from a contiguous area of 0.63 square degrees ( excluding areas contaminated by bright stars and ccd blooming ) covered by both the uds near - infrared and sxds optical imaging . one of the principle motivations for this study was to exploit the unique combination of areal coverage and optical+nearir data available in the uds / sxds to study the high - redshift galaxy population without recourse to strict dropout / lbg selection criteria . the reasoning behind this is to allow the luminosity function to be studied over a relatively wide range of redshifts from a single sample , and to reduce as much as possible the strong bias towards young , blue star - forming galaxies inherent to traditional lyman - break selection . although single - colour selection techniques ( e.g. @xmath54-drop , @xmath55drop ) have proven to be highly successful at isolating high - redshift galaxies , because of the need to adopt fairly strict colour - cut criteria ( e.g. @xmath56 for @xmath57 ) , it is at least possible that a significant population of old / redder , perhaps more massive , galaxies which marginally fail to satisfy these strict criteria could be excluded ( e.g. dunlop et al . 2007 ; rodighiero et al . 2007 ) . consequently , from the outset the decision was taken to make the initial selection criteria as simple and inclusive as possible , and to then use a photometric redshift analysis to both define the final high - redshift galaxy sample and to exclude likely low - redshift interlopers . within this context , the original sample for this study was a @xmath58band selected catalog covering the @xmath59 square - degree area uniformly covered by the subaru optical and uds near - infrared data , produced with version 2.5.2 of the sextractor software ( bertin & arnouts 1996 ) . the original catalog was then reduced to @xmath60 objects by adopting a @xmath58band magnitude limit of @xmath61 , which corresponds to the @xmath62 detection threshold ( @xmath63 diameter apertures ) and the @xmath64% completeness limit . in addition , the only other selection criterion applied was the exclusion of all objects which were detected in the @xmath65band at more than @xmath66 significance . this criterion was adopted in order to exclude the vast majority of galaxies which are simply too bright in the @xmath65band to be robust high - redshift candidates ; given that the lyman - limit at @xmath67 is redshifted out of the suprime - cam @xmath65band filter at a redshift of @xmath68 . following the application of the initial selection criteria the parent sample of potential high - redshift galaxies consisted of @xmath69 candidates in the apparent @xmath58band magnitude range @xmath70 . due to the decision not to apply strict drop - out criteria in our initial selection , in principle , this sample should contain _ all _ galaxies with @xmath37 lying between @xmath68 ( when the lyman limit [ @xmath67 ] is redshifted out of the @xmath65band ) and @xmath71 ( when lyman alpha [ @xmath72 ] is redshifted out of the @xmath73-band ) . however , given that the primary selection is performed in the rest - frame uv , it is clear that the sample is still biased against objects with significant levels of intrinsic reddening . the next stage in the analysis was to process each of the potential high - redshift candidates with our own photometric redshift code . the photometric redshift code is an extended version of the publicly available hyperz package ( bolzonella , miralles & pell 2000 ) and is described in more detail by cirasuolo et al . however , briefly , the code fits a wide range of different galaxy sed templates to the available multi - wavelength ( @xmath74 ) photometry of each candidate , returning a best - fitting value of redshift , sed type , age , mass and reddening . it is worth noting at this point that for calculating the photometric redshifts , and the final luminosity function estimates , the @xmath63 diameter magnitudes were corrected to total using point - source corrections in the range @xmath75@xmath76 magnitudes ( depending on the seeing of the @xmath74 images ) . the primary set of stellar population models we adopted were those of bruzual & charlot ( 2003 ) , with solar metallicity and assuming a salpeter initial mass function ( imf ) with a lower and upper mass cutoff of 0.1 and 100 @xmath77 respectively . both instantaneous burst and exponentially declining star formation models with e - folding times in the range @xmath78 were included , with the restriction that the sed models did not exceed the age of the universe at the best - fitting redshift . the code accounts for dust reddening by following the calzetti et al . ( 2000 ) obscuration law within the range @xmath79 , and accounts for lyman series absorption due to the hi clouds in the inter galactic medium according to the madau ( 1995 ) prescription . based on the results of the photometric redshift analysis , 8% of the sample was excluded because it was not possible to find an acceptable @xmath80 fit ( at the @xmath81 level ) with a galaxy sed template at any redshift . the vast majority of these objects were spurious artifacts on the @xmath73-band images which had contaminated the original catalog . however @xmath82% of the excluded contaminants with @xmath83 had @xmath84 colours consistent with ultra - cool galactic stars . following the photometric redshift analysis our sample consisted of 1621 high - redshift galaxy candidates with primary redshift solutions at @xmath85 , and 4350 objects with primary redshift solutions at @xmath86 . however , due to the problem of multiple redshift minima which is inherent to photometric redshift techniques , when computing the luminosity function is it clearly not optimal to simply exclude the majority of candidates with primary redshift solutions at @xmath86 . indeed , even for candidates which have well defined primary redshift solutions at @xmath85 , it is often the case that a plausible secondary redshift solution exists at lower redshift ; and vice - versa . the simple reason is that , within the constraints of the available photometry , it is often statistically impossible to cleanly differentiate between a high - redshift lyman - break solution and a competing , low - redshift , @xmath87 - break solution at @xmath31 . therefore , during the derivation of the luminosity functions presented in section 5 , each candidate ( including those with primary redshift solutions at @xmath86 ) was represented by its normalised probability density function @xmath88 , which satisfied the condition : @xmath89 in addition to making better use of the available information , this methodology immediately deals with the problem of multiple redshift solutions in a natural and transparent fashion . armed with a normalised probability density function for each high - redshift candidate , it is possible to construct their luminosity function using the classic @xmath90 maximum likelihood estimator of schmidt ( 1968 ) . therefore , within a given redshift interval ( @xmath91 ) , the luminosity function estimate within a given absolute magnitude bin was calculated as follows : @xmath92 where the summation runs over the full sample , and @xmath93 is the redshift range within which a candidate s absolute magnitude lies within the magnitude bin in question . the quantity @xmath94 is a correction factor which accounts for incompleteness , object blending and contamination from objects photometrically scattered into the sample from faint - ward of the @xmath58band magnitude limit . the corrections for sample incompleteness and object blending were calculated by adding thousands of randomly placed point - sources into the @xmath58band images and recovering them with sextractor using the same extraction parameters adopted for the original object catalogues . the results of this procedure revealed that the @xmath58band images are 100% complete at @xmath34 and fall to 80% complete at the adopted @xmath61 magnitude limit . in addition , this process also demonstrated that it is necessary to correct for a constant @xmath95% of objects which are missed from the object catalogues due to blending with nearby objects on the crowded suprime - cam images . in this study we adopt @xmath61 as our limiting magnitude , which corresponds to the @xmath96 detection limit . this fact , combined with the steepness of the luminosity function in the region we are exploring ( see fig . 1 ) means that the prospect of contamination of the faint end of our sample from objects photometrically up - scattered from below the flux limit has to be considered carefully . in order to quantify this effect it is necessary to rely on simulation . in outline , the adopted procedure was to produce a realistic simulated population of lbgs in the redshift interval @xmath97 , based on a known model for the evolving luminosity function . the simulated photometry for this population was then passed through the same initial selection criteria and photometric redshift analysis as employed on the real data . the average effect on the faint end of our luminosity function determination could then be calculated by running many monte - carlo realisations of the simulated lbg population . in order to make the simulated lbg population as realistic as possible we based our model for the evolving luminosity function on the schechter function fits to the latest determination of the uv - selected luminosity functions centred on @xmath98 , @xmath99 and @xmath100 by bouwens et al . ( 2007 ) . over the full redshift range the adopted model had a fixed normalisation ( @xmath101 mpc@xmath17 ) , a fixed faint - end slope ( @xmath102 ) and an evolving characteristic magnitude parameterised as : @xmath103 where @xmath104 is calculated at a rest - frame wavelength of 1500 ( which corresponds to the middle of the @xmath58filter for an object at @xmath9 ) . although simple , this parameterisation nonetheless closely reproduces the bouwens et al . ( 2007 ) luminosity function fits over the redshift range of interest . more importantly , this model should provide an accurate prediction of the relative numbers of objects just above and just below our adopted magnitude limit , the region in which our sample is most vulnerable to potential contamination from photometric scatter . this parameterisation of the evolving luminosity function was used to populate the redshift@xmath44magnitude plane using bin sizes of @xmath105 and @xmath106 . within each redshift bin the luminosity function was integrated down to an absolute magnitude corresponding to @xmath107 , one magnitude fainter than our adopted magnitude limit . in addition to its absolute magnitude and redshift , each simulated lbg was allocated @xmath108 photometry based on an appropriate model sed drawn randomly from a distribution of parameter values ( i.e. age , reddening , metallicity ) representative of those displayed by the real lbg sample . appropriate photometric errors were then calculated for the synthetic photometry in each filter , according to the average depths of the real data . the final stage of the process involved the use of monte - carlo simulation whereby , in each realisation , the photometry of each candidate was randomly perturbed according to its photometric errors . the perturbed photometry of the synthetic lbg sample was then passed through our initial optical selection criteria to establish the number of objects which would have been included in our sample . the final result of this process was a finely sampled grid of correction factors which accounted for the fraction of the objects scattered into , and out of , the sample as a function of redshift and apparent @xmath58band magnitude . in terms of the estimated luminosity functions presented in section 5 , this simulation process demonstrated that photometric up - scatter from below the magnitude limit has a relatively minor effect . as would be anticipated , the effect is most pronounced on the two faintest absolute magnitude bins in the @xmath9 and @xmath12 luminosity function estimates . however , even here the effect is only to increase the volume density by @xmath109% and @xmath110% respectively . the luminosity function estimates shown in fig . 1 are corrected for this effect . following the procedures outlined in the previous two sections , the uv - selected luminosity functions were estimated in two redshift intervals centred on @xmath9 and @xmath12 . for the @xmath9 luminosity function we considered the redshift range @xmath111 and for the @xmath12 luminosity function we considered the redshift range @xmath112 . these redshift ranges were chosen to be wide enough to include sufficient numbers of objects to provide a robust measurement of the bright end of the luminosity function , and narrow enough to limit the evolution of the luminosity function across each bin . the resulting estimates of the luminosity functions at @xmath9 and @xmath12 are shown in fig . 1 . the advantages provided by the large area and deep optical+nearir data available in the uds / sxds are immediately obvious from fig . 1 . as a result of the current data - set covering an area of 0.63 square degrees , it has been possible to robustly estimate the bright end of the luminosity function to absolute magnitudes as bright as @xmath113 and @xmath114 at @xmath9 and @xmath12 respectively . a second obvious feature of fig . 1 is the significant evolution in the bright end of the luminosity function within the @xmath3 redshift interval ; as previously demonstrated by bouwens et al . ( 2007 ) . from the data presented in fig . 1 it is tempting to assume that this evolution is due entirely to a change in @xmath104 between @xmath9 and @xmath12 ( a shift of @xmath115 magnitudes results in excellent agreement between the two luminosity functions ) . in fact , based on our data - set alone , it is not possible to determine whether the apparent evolution is due entirely to a change in @xmath104 , or whether significant evolution in the normalisation ( @xmath116 ) is also occurring . however , the differential change in galaxy number density as a function of absolute magnitude does provide some information on the form of the luminosity function evolution . for example , while the number density of @xmath117 galaxies only decreases by a factor of @xmath118 between @xmath9 and @xmath12 , the number density of @xmath114 galaxies decreases by a factor of @xmath119 . in terms of schechter function parameters , it is not possible to reproduce this change in luminosity function shape through evolution of @xmath116 alone , and immediately confirms that some evolution of @xmath104 must be taking place . an attempt to quantify the relative contribution of evolution in @xmath104 and @xmath116 is pursued in section 5.2 . before proceeding to explore the evolution of the luminosity function from @xmath9 to @xmath12 further , it is worthwhile comparing our results with those of recent studies in the literature . the @xmath19 luminosity function has been investigated by several recent studies using both ground - based and hst imaging ( e.g. ouchi et al . 2004 ; yoshida et al . 2006 ; iwata et al . 2007 ; oesch et al . 2007 ; bouwens et al . 2007 ) . as an illustration of the range of results which have recently been reported we show in fig . 2a a comparison of our @xmath9 luminosity function estimate with the ground - based results of yoshida et al . ( 2006 ) and iwata et al . ( 2007 ) , and the hst - based results of bouwens et al . the yoshida et al . ( 2006 ) study is based on deep subaru suprime - cam optical imaging data covering an area of @xmath120 square degrees in the subaru deep field . yoshida et al . selected a large lbg sample centred on @xmath121 , based on cuts in the @xmath122 vs @xmath123 colour - colour diagram , the reliability of which were confirmed via limited follow - up spectroscopy . the schechter function fit derived by yoshida et al . from their @xmath121 lbg sample is shown as the dark grey curve in fig . it can be seen that our luminosity function estimate is in excellent agreement with the yoshida et al . determination at @xmath124 . at @xmath125 our number densities are somewhat lower than determined by yoshida et al . , although only by @xmath126% . this difference is not very significant , and can largely be explained by the fact that the schechter function fit derived by yoshida et al . does not have a well constrained faint - end slope due to the lack of data fainter than @xmath127 . we note here that the yoshida et al . results at @xmath19 are very similar to those of ouchi et al . ( 2004a ) , who analysed an combined sdf+sxds data - set covering a larger area ( 0.35 square degrees ) but with a magnitude limit @xmath128 magnitudes brighter . the solid black curve plotted in fig . 2a is the schechter function fit derived by bouwens et al . ( 2007 ) from a sample of 1416 @xmath129drop galaxies selected from a compilation of several deep hst imaging surveys ( goods n+s , hudf & hudf - p ) . at @xmath130 our data are in excellent agreement with the bouwens et al . luminosity function , which is remarkable given the very different techniques , data - sets and sky areas utilised by the two studies ( the largest area in the bouwens et al . study is the combined goods n+s fields ; @xmath131 square degrees ) . at the brightest magnitudes fig . 2a suggests that our sample contains a higher number density of galaxies than predicted by the bouwens et al . luminosity function . however , this is not unreasonable given that the large survey area of this study provides improved statistics on the bright end of the luminosity function , where the bouwens et al . fit to the luminosity function is less well constrained . finally , in fig . 2a we plot the recent luminosity function fit of iwata et al . ( 2007 ) which is based on deep subaru suprime - cam imaging covering an area of @xmath132 square - degrees ( hdf - n and the j0053 + 1234 region ) . based on cuts in the @xmath133 vs @xmath134 colour - colour diagram , iwata et al . ( 2007 ) select 853 @xmath19 lbg candidates and derive the luminosity function fit shown as the light grey curve in fig . it can be seen that our luminosity function estimate is in poor agreement with the iwata et al . ( 2007 ) results , with the latter study finding a number density of objects at the brightest absolute magnitudes a factor of @xmath118 higher than found here . it is difficult to understand how the large numbers of bright lbgs detected in the iwata et al . study could be missing from our sample . objects this bright would have very high signal - to - noise photometry in our data - set , and should return the most robust photometric redshifts . the iwata et al . ( 2007 ) results are in good agreement with their previous estimate of the @xmath19 luminosity function ( iwata et al . 2003 ) and it has been suggested by several authors ( e.g. ouchi et al . 2004a ; yoshida et al . 2006 ; bouwens et al . 2007 ) that the selection criteria used in iwata et al . ( 2003 , 2007 ) may allow contamination from low - redshift interlopers . we note that to produce number densities at @xmath135 as high as reported by iwata et al . ( 2007 ) , we would have to reinstate into our sample objects which have been excluded as suspected ultra - cool galactic stars due to their photometry being a poor match to any galaxy sed template previous constraints on the @xmath20 luminosity function have been based , virtually exclusively , on the deep hst imaging available in the two goods regions , the hubble ultra - deep field ( hudf ) and the hudf parallel fields ( e.g. bunker et al . 2004 ; dickinson et al . 2004 ; yan & windhorst 2004 ; malhotra et al . 2005 ; bouwens et al . 2006 , 2007 ) . in fig . 2b we show the schechter function fit to the @xmath20 luminosity function derived by bouwens et al . ( 2007 ) from a sample of 627 @xmath55dropouts selected from a compilation of the deepest available hst imaging data luminosity function based on hst imaging data ] . 2b shows that our estimate of the @xmath12 luminosity function is in reasonable agreement with the luminosity function fit of bouwens et al . ( 2007 ) , although our number densities are lower by @xmath136% in the region around @xmath117 , where the bouwens et al . luminosity function fit is still well constrained by data . however , this discrepancy is only apparent in comparison to the schechter function fit derived by bouwens et al . , and the actual data from the two studies are in excellent agreement ( see fig . 3b ) . in fig . 2b we also show the number density of objects at @xmath137 derived by shimasaku et al . ( 2005 ) , based on a sample of 12 objects selected using imaging of the subaru deep field with two medium - band @xmath138filters . the shimasaku et al . estimate is a factor of @xmath139 higher than our estimated number density at @xmath137 . the source of this discrepancy is not clear . given the rigorous nature of their selection criteria , it seems unlikely that the shimasaku et al . sample is contaminated by interlopers at the @xmath140% level . however , the survey area covered by this work is a factor of @xmath118 larger than that of the shimasaku study , so it is likely that the apparent off - set is simply due to a combination of statistical errors and cosmic variance . although the results of the present study provide much improved constraints on the bright end of the luminosity function at @xmath12 and @xmath9 , they do not reach faint enough to constrain the normalisation or faint - end slope of the @xmath35 luminosity function . consequently , in this section we explore the possibility of providing an improved measurement of the evolution of the luminosity function between @xmath9 and @xmath12 by combining our results with those of the hst - based study of bouwens et al . ( 2007 ) . in fig . 3 we show a comparison of our luminosity function estimates at @xmath9 and @xmath12 with those of bouwens et al . ( 2007 ) . it can be seen that in the absolute magnitude range common to both studies the two , independent , estimates of the luminosity functions are in excellent agreement . motivated by this agreement , it was decided to combine our results with those of bouwens et al . ( 2007 ) to derive schechter function fits to the luminosity function at @xmath9 and @xmath12 based on data spanning a factor of @xmath141 in luminosity . the adopted procedure was to use a maximum likelihood technique to fit a schechter function to a finely sampled grid on the apparent @xmath58band magnitude @xmath44 redshift plane @xmath142 . to fit the luminosity function at @xmath9 the grid was populated within the redshift interval @xmath143 and the apparent @xmath58band magnitude range @xmath144 . in the magnitude range @xmath145 the grid was populated using the data from this study , while in the magnitude range @xmath146 the grid was populated according to the schechter function fit to the @xmath9 luminosity function derived by bouwens et al . this process was repeated to fit the @xmath12 luminosity function , with the exception that the redshift range considered was @xmath147 . the fitting procedure was to maximise the following likelihood function : @xmath148 where @xmath149 is the number of galaxies in cell @xmath150 and @xmath151 is the probability of finding a galaxy within cell @xmath150 given the choice of model parameters . in this context , the probability @xmath151 is naturally defined as @xmath152 , where @xmath153 is the predicted number of galaxies within cell @xmath150 , for a given set of model parameters , and @xmath154 is the corresponding integrated luminosity function . the best - fitting schechter functions to the combined ground - based+hst data - sets at @xmath9 and @xmath12 are shown as the solid curves in fig . the best - fitting schechter function parameters and their associated uncertainties are reported in table 1 , and the @xmath155 and @xmath81 confidence regions around the best - fitting values of the faint - end slope and characteristic magnitude are plotted in fig . 5 . it can be seen from fig . 4 that a schechter function provides a good description of luminosity function at @xmath9 and @xmath12 over the full luminosity range ( @xmath156 magnitudes ) sampled by the combined ground - based+hst data - set . a comparison of the best - fitting schechter function parameters derived here ( table 1 ) with those reported by bouwens et al . ( 2007 ) shows that , at both redshifts , our best - fitting parameters are within the @xmath157 uncertainties reported by the earlier study . with respect to the faint - end slope and normalisation this is unsurprising , given that the constraints on these two parameters are largely provided by the deeper hst imaging data . however , the principle advantage provided by fitting the combined data - set is the improved constraints on the value of @xmath158 ; particularly at @xmath12 . our results suggest that @xmath158 dims by @xmath159 magnitudes between @xmath9 and @xmath12 . this is more dramatic than , although still completely consistent with , the dimming of @xmath160 suggested by the schechter function fits derived by bouwens et al . with respect to the normalisation of the luminosity function , our results suggest that @xmath116 increases by a factor of @xmath139 between @xmath9 and @xmath12 , although the comparatively large uncertainty on the normalisation of the schechter function at @xmath12 means that this evolution is not statistically significant ( @xmath161 . as expected , our fit to the combined ground - based+hst data - set confirms the bouwens et al . ( 2007 ) result that the faint - end slope of the luminosity function at @xmath35 is steep ( @xmath162 ) and shows no sign of evolution in the redshift interval @xmath3 . .the best - fitting schechter function parameters from our maximum likelihood fitting of the combined ( ground - based+hst ) luminosity function data - sets at @xmath9 and @xmath12 . the errors on the faint - end slope and characteristic magnitude have been derived from the likelihood ratio contours shown in fig . the values of @xmath116 and their associated errors have been calculated by fitting the number counts in each redshift interval . [ cols="^,^,^,^",options="header " , ] in this section we evaluate the 2-point angular correlation function for those lbgs in our sample with primary photometric redshift solutions in the interval @xmath163 . to measure the angular correlation function @xmath164 and estimate the related poissonian errors we used the landy & szalay ( 1993 ) estimators . the correlation function derived from our @xmath163 lbg sample is shown in fig . the best fit for the angular correlation is assumed to be a power - law as in groth & peebles ( 1977 ) : @xmath165 with the amplitude at 1 degree @xmath166 , the slope @xmath167 , and the integral constraint due to the limited area of the survey @xmath168 ( determined over the unmasked area ) . it should be noted that because the contamination of our @xmath3 lbg sample from low - redshift interlopers is securely @xmath169 we have made no correction for dilution of the clustering signal . in fig . 6 we also indicate the best - fitting curve determined using only the largest scale measurements ( @xmath170 ) and a fixed value for the slope of @xmath171 . based on a sample of @xmath19 lbgs selected from the goods fields , lee et al . ( 2006 ) find that their angular clustering measurments show a strong transition on the smallest scales ( @xmath172 ) , indicative of a intra / inter dark matter halo transition , while on larger scales the angular correlation is well approximated by a power - law with a slope of @xmath171 . however , as can be seen from fig . 6 , our results are consistent with a value of @xmath171 on large scales , and only present a non - significant deviation from this slope on the smallest scales ( @xmath173 corresponding to @xmath174kpc ) . in order to compare the clustering of galaxy populations at different redshifts it is necessary to derive spatial correlation measurements . using the redshift distribution of our sample we can derive the correlation length using the relativistic limber equation ( magliocchetti & maddox 1999 ) and , in addition , derive a linear bias estimation ( magliocchetti et al . 2000 ) assuming that the dark matter behaves as predicted by linear theory . to perform this calculation we have adopted the redshift distribution for our sample as derived from integrating over the redshift probability density functions for those objects with primary photometric redshift solutions in the interval @xmath163 . the correlation length can then be evaluated by fixing the slope of the correlation function to @xmath175 . this calculation produces a measurement of @xmath176mpc for the correlation length and a value of @xmath177 for the linear bias . although we consider this to be our best estimate of the clustering properties of the sample , we note that if we instead adopt the redshift distribution based on the best - fitting photometric redshifts alone ( @xmath178 ; see fig a2 ) we derive slightly higher , although consistent , values of @xmath179mpc and @xmath180 for the correlation length and linear bias respectively . it is of interest to compare our clustering results for @xmath181 lbgs in the redshift interval @xmath163 with recent literature results for lbg samples with similar limiting magnitudes and mean redshifts . our results are in good agreement with ouchi et al . ( 2004b ) who derived a correlation length of @xmath182mpc for a sample of @xmath183 lbgs with a mean redshift of @xmath184 in the subaru deep field . in addition , our results are in agreement with those of overzier et al . ( 2006 ) who derived a correlation length of @xmath185mpc for a sample of @xmath186 refers to @xmath138band magnitudes obtained through the hst f850lp filter , rather than the subaru @xmath58filter . ] lbgs with a mean redshift of @xmath187 in the goods fields . finally , our results are also in good agreement with those of lee et al . ( 2006 ) , who derived a value of @xmath188mpc ( fixed slope of @xmath171 ) from their sample of @xmath189 lbgs in the goods fields with a mean redshift of @xmath190 . based on the predictions of dark matter halo models ( sheth et al . 2001 ; mo & white 2002 ) , and assuming a one - to - one correspondance between galaxies and halos , we infer that our @xmath163 lbgs are likely to be hosted by dark matter halos with masses of @xmath191 , comparable to what is observed for @xmath192 lbgs at @xmath193 . in this section we explore the possibility of estimating the typical stellar mass of our @xmath3 lbgs from a stacking analysis of the available optical+nearir imaging data . the aim is to test whether the typical stellar mass is compatible with the dark matter halo mass suggested by the clustering analysis . in fig . 7 we show 20@xmath194 20 postage stamps produced by taking a median stack of the available imaging data , centred on the positions of the 754 lbg candidates with primary photometric redshift solutions in the interval @xmath163 . there are several features of fig . 7 which are worthy of comment . firstly , in the stack of the @xmath65band photometry the lbg candidates are robustly non - detected ( @xmath195 ; @xmath157 ) , as expected from our initial selection criteria . secondly , it can be seen that there is a faint detection in the stacked @xmath129band data . however , the @xmath129band detection is very faint ( @xmath196 ) and is expected given that the lyman - limit does not pass out of the subaru suprime - cam @xmath129filter until @xmath197 . finally , the two panels on the lower right of fig . 7 show the results of stacking the available spitzer irac data at @xmath198 m and @xmath199 m from the swire survey ( lonsdale et al . although relatively shallow , the swire data is deep enough to produce detections at @xmath198 m and @xmath199 m in the median stack . together with the robust photometry from the stacked @xmath200 imaging data , it is the extra wavelength coverage provided by the detections at @xmath198 m and @xmath199 m which will allow us to place useful constraints on the typical stellar mass of the @xmath3 lbgs . using the same procedure as described in section 3.1 , the best - fitting galaxy template to the stacked photometry is shown in fig . it can be seen from the lower panel of fig . 8 that the fit to the stacked photometry provides a robust redshift solution at @xmath201 and that any alternative low - redshift solution is securely ruled out . the best - fitting sed template has an age of 400 myrs , consistent with a typical formation redshift of @xmath202 , and no intrinsic reddening ( @xmath203 ) . by considering the range of sed templates which provide an acceptable fit to the stacked photometry , we calculate that the stellar mass of the lbg stack is @xmath15 ( salpeter imf ) . although the uncertainties in the determination of the typical stellar and dark halo mass for the @xmath3 lbg sample are obviously large , combined with the results of the clustering analysis reported in the previous section , the sed fit to the stacked photometry suggests that the typical dark matter to stellar mass ratio for the @xmath3 lbgs is @xmath204 . given that our lbg sample are exclusively drawn from the bright end of the @xmath3 luminosity function it would seem reasonable to assume that they will evolve into objects occupying the high - mass end of the low - redshift galaxy stellar mass function ( i.e. @xmath205 ) . in which case , it is noteworthy that @xmath206 is in good agreement with the measured value for luminous red galaxies ( lrgs ) in the sdss from galaxy - galaxy lensing ( mandelbaum et al . moreover , @xmath206 is also in good agreement with the predicted mean value for low - redshift lrgs from recent semi - analytic galaxy formation models ( e.g. almeida et al . although it remains a possibility that a substantial population of massive , comparatively red , objects exist at high redshift , it seems likely that the luminous ( @xmath2 ) lbgs selected in this study are amongst the most massive galaxies in existence at @xmath3 . as such , their number densities and stellar masses have the potential to provide constraints on the latest generation of galaxy evolution models . in fig . 9 we show an estimate of the galaxy stellar mass function at @xmath3 based on combining the luminosity function fits presented in section 5 with the results of the stacking analysis presented in section 7 . the adopted procedure for estimating the stellar mass function is extremely straightforward . based on the schechter function fits to the luminosity function at @xmath9 and @xmath12 , we have estimated the corresponding stellar mass function by simply multiplying the luminosity function fits by the mass - to - light ratio derived from our sed fit to the stacked lbg imaging data . in fig . 9 the upper envelope of the grey shaded area is our estimate of the stellar mass function at @xmath9 , while the lower envelope is our corresponding estimate at @xmath12 ( both include the @xmath157 uncertainties on the best - fitting schechter luminosity function parameters ) . for comparison we also plot the stellar mass function estimates at @xmath207 based on the de lucia & blaizot ( 2007 ) and bower et al . ( 2006 ) semi - analytic galaxy formation models ( thick and thin curves respectively ) . are adopted because they are the closest available to the mean photometric redshift of our @xmath3 lbg sample ( @xmath178 ) . ] in order to perform a fair comparison , we have shifted our stellar mass function estimates to lower masses by a factor of 1.8 , to account for the fact that the de lucia & blaizot ( 2007 ) and bower et al . ( 2006 ) models are based on imfs which typically return stellar masses a factor of @xmath208 lower than our salpeter - based estimates ( de lucia & blaizot 2007 ) . finally , for reference , the dashed curve in fig . 9 is the stellar mass function at @xmath209 from cole et al . ( 2001 ) . the mass function predictions from the bower et al . ( 2006 ) and de lucia & blaizot ( 2007 ) models are in good agreement , at least qualitatively , with our observational estimate . the general agreement between our estimate , based on the uv - selected luminosity function , and the model predictions suggests that lbgs constitute the majority of the stellar mass density at @xmath35 and , if the model predictions are correct , a putative population of reddened / older galaxies at @xmath3 does not dominate the stellar mass density . compared to the @xmath209 mass function , our estimate of the high - redshift stellar mass function suggests that the number density of @xmath210 and @xmath211 galaxies in place by @xmath19 is @xmath119% and @xmath212% of its local value respectively . given that we have estimated the stellar mass function at @xmath9 and @xmath12 by simply applying a constant mass - to - light ratio derived from our stacking analysis , it is obviously somewhat speculative to proceed to estimate the co - moving stellar mass density . however , for completeness , integrating our stellar mass function estimates down to a mass limit of @xmath213 and @xmath12 luminosity functions to a limit of m@xmath214 , which involves an extrapolation @xmath215 magnitude fainter than the @xmath58band magnitude limit of the uds / sxds data set . ] provides stellar mass density estimates of @xmath216 mpc@xmath17 and @xmath217 mpc@xmath17 at @xmath9 and @xmath12 respectively ( salpeter imf ) . it was decided to integrate our stellar mass function down to a limit of @xmath213 for two reasons . firstly , a mass limit of @xmath213 corresponds well to the mass limits of stark et al . ( 2007 ) and yan et al . ( 2006 ) , who previously estimated the stellar mass density at @xmath9 and @xmath12 respectively . secondly , given that our data - set provides no information on the mass - to - light ratios of objects with masses smaller than @xmath218 , it would clearly be ill advised to extrapolate to smaller stellar masses . our stellar mass density estimate at @xmath9 is in reasonable agreement with the estimate of @xmath219 mpc@xmath17 derived by stark et al . ( 2007 ) from a combined sample of spectroscopic and photometrically selected galaxies in the southern goods field . moreover , our estimate of the stellar mass density at @xmath12 is in agreement with the lower limit of @xmath220 mpc@xmath17 derived by yan et al . ( 2006 ) , based on a sample of @xmath55drop galaxies in the north and south goods fields . the stellar mass densities quoted by both stark et al . ( 2007 ) and yan et al . ( 2006 ) are also based on a salpeter imf . in this paper we have reported the results of a study of a large sample of luminous ( @xmath2 ) lbgs in the redshift interval @xmath1 . by employing a photometric redshift analysis of the available optical+nearir data we have derived improved estimates of the bright end of the uv - selected luminosity function at @xmath9 and @xmath12 . moreover , by combining our new results with those based on deeper , but small area , hst data we have derived improved constraints on the best - fitting schechter function parameters at @xmath9 and @xmath12 . in addition , by studying the angular clustering properties of our sample we have determined that luminous lbgs at @xmath3 typically lie in dark matter halos with masses of @xmath221 . finally , based on the results of a stacking analysis , we have estimated the galaxy stellar mass functions and integrated stellar mass densities at @xmath9 and @xmath12 . our main conclusions can be summarised as follows : 1 . our new determination of the bright end of the high - redshift luminosity function confirms that significant evolution occurs over the redshift interval @xmath3 . based on our results it is clear that the luminosity function evolution can not be described by evolution in normalisation ( @xmath116 ) alone , and that some level of evolution in @xmath158 is also required . a comparison of our new results with those in the literature demonstrates that , within the magnitude range where the two studies overlap , our estimates of the luminosity function at @xmath9 and @xmath12 are in excellent agreement with those derived from ultra - deep hst imaging data by bouwens et al . 3 . by combining our estimate of the bright end of the luminosity function with the corresponding estimates of the faint end by bouwens et al . ( 2007 ) , it is possible to fit the luminosity function at @xmath9 and @xmath12 over a luminosity range spanning a factor of @xmath141 . based on this combined ground - based+hst data - set we find the following best - fitting schechter function parameters : @xmath7 and @xmath8 for the luminosity function at @xmath9 , and @xmath222 and @xmath11 at @xmath12 . 4 . these results are consistent with the corresponding schechter function parameters derived by bouwens et al . ( 2007 ) although , due to the improved statistics at the bright end provided by our wide survey area , the fits to the combined ground - based+hst data - set provide improved constraints on the evolution of @xmath158 in particular . 5 . an analysis of their angular clustering properties demonstrates that luminous @xmath2 lbgs at @xmath3 are strongly clustered , with a correlation length of @xmath176mpc . comparison of these clustering results with theoretical models suggests that these lbgs typically reside in dark matter halos with masses of @xmath223 . an sed fit to a stack of the available optical+nearir data for our sample suggests that luminous lbgs at @xmath3 have typical stellar masses of @xmath210 . combined with the results of the clustering analysis this suggests that the typical ratio of dark matter to stellar mass for luminous lbgs at @xmath3 is @xmath224 . assuming that the mass - to - light ratio derived from the sed fit to the stacked lbg imaging data is representative , we use our best - fitting schechter function parameters for the @xmath9 and @xmath12 luminosity functions to estimate the corresponding stellar mass functions . although clearly subject to large uncertainties , our stellar mass function estimates are consistent with the latest predictions of the semi - analytic galaxy formation models of de lucia & blaizot ( 2007 ) and bower et al . 8 . based on our stellar mass function estimates we calculate that the stellar mass in place at @xmath9 and @xmath12 is @xmath225 mpc@xmath17 and @xmath226 mpc@xmath17 respectively . although uncertain , these independent estimates of the integrated stellar mass density are consistent with the results of studies utilising the ultra - deep spitzer irac data in the goods fields ( stark et al . 2007 ; yan et al . 2006 ) . the authors would like to thank the anonymous referee whose comments and suggestions significantly improved the final version of this manuscript . rjm and oa would like to acknowledge the funding of the royal society . mc and sf would like to acknowledge funding from stfc . we are grateful to the staff at ukirt and subaru for making these observations possible . we also acknowledge the cambridge astronomical survey unit and the wide field astronomy unit in edinburgh for processing the ukidss data . almeida c. , baugh c.m . , wake d.a . , lacey c.g . , benson a.j . , bower r.g . , pimbblet k. , 2008 , mnras , in press , arxiv:0710.3557 [ astro - ph ] bertin e. , arnouts s. , 1996 , a&as , 117 , 393 bolzonella m. , miralles j.m . , pell , 2000 , a&a , 363 , 476 bouwens r.j . , illingworth g.d . , blakeslee j.p . , franx m. , 2006 , apj , 653 , 53 bouwens r.j . , illingworth g.d . , franx m. , ford h. , 2007 , apj , 670 , 928 bouwens r.j . , illingworth g.d . , franx m. , ford h. , 2008 , apj , submitted , arxiv:0803.0548 [ astro - ph ] bower r.g . , benson a.j . , malbon r. , helly j.c . , frenk c.s . , baugh c.m . , cole s. , lacey c.g . , 2006 , mnras , 370 , 645 bruzual g. , charlot s. , 2003 , mnras , 344 , 1000 bunker a.j . , stanway e.r . , ellis r.s . , mcmahon r.g . , 2004 , mnras , 355 , 374 calzetti d. , armus l. , bohlin r. c. , kinney a. l. , koornneef j. , storchi - bergmann t. , 2000 , apj , 533 , 682 casali m. , et al . , 2007 , a&a , 467 , 777 cirasuolo m. , et al . , 2007 , mnras , 380 , 585 cirasuolo m. , mclure r.j . , dunlop j.s . , almaini o. , foucaud s. , simpson c. , 2008 , mnras , submitted , arxiv:0804.3471 [ astro - ph ] cole s. , et al . , 2001 , mnras , 326 , 255 de lucia g. , blaizot j. , 2007 , mnras , 375 , 2 dickinson m. , et al . , 2004 , apj , 600 , l99 dunlop j.s . , cirasuolo m. , mclure r.j . , 2007 , mnras , 376 , 1054 eyles l.p . , bunker a.j . , stanway e.r . , lacy m. , ellis r.s . , doherty m. , 2005 , mnras , 364 , 443 furusawa h. , et al . , 2008 , apjs , 176 , 1 groth e.j . , peebles p.j.e . , 1977 , apj , 217 , 385 guhathakurta p. , tyson j.a . , majewski s.r . , 1990 , apj , 357 , l9 iye m. , et al . , 2006 , nature , 443 , 186 iwata i. , ohta k. , tamura n. , ando m. , wada s. , watanabe c. , akiyama m. , aoki k. , 2003 , pasj , 55 , 415 iwata i. , ohta k. , tamura n. , akiyama m. , aoki k. , ando m. , kiuchi g. , sawicki m. , 2007 , mnras , 376 , 1557 hu e.m . , mcmahon r.g . , cowie l.l . , 1999 , apj , 521 , l9 landy s.d . , szalay a.s . , 1993 , apj , 412 , 64 lawrence a. , et al . , 2007 , mnras , 379 , 1599 lee k.s . , giavalisco m. , gnedin o.y . , somerville r.s . , ferguson h.c . , dickinson m. , ouchi m. , 2006 , apj , 642 , 63 lonsdale c.j . , et al . , 2003 , pasp , 115 , 897 madau p. , 1995 , apj , 441 , 18 magliocchetti m. , maddoz s.j . , 1999 , mnras , 306 , 988 magliocchetti m. , bagla j.s . , maddox s.j . , lahav o. , 2000 , mnras , 314 , 546 malhotra s. , et al . , 2005 , apj , 626 , 666 mandelbaum r. , seljak u. , kauffmann g. , hirata c.m . , brinkmann j. , 2006 , mnras , 368 , 715 mclure r.j . , et al . , 2006 , mnras , 372 , 357 mo h.j . , white s.d.m . , 2002 , mnras , 336 , 112 moscardini l. , coles p. , lucchin f. , materrese s. , 1998 , mnras , 299 , 95 miyazaki s. , et al . , 2002 , pasj , 54 , 833 oesch p.a . , et al . , 2007 , apj , 671 , 1212 oke j.b . , gunn j.e . , 1983 , apj , 266 , 713 ouchi m. , et al . , 2004a , apj , 611 , 660 ouchi m. , et al . , 2004b , apj , 611 , 685 ouchi m. , et al . , 2005 , apj , 635 , l117 ouchi m. , et al . , 2008 , apjs , in press , arxiv:0707.3161 [ astro - ph ] overzier r.a . , bouwens r.j . , illingworth g.d . , franx m. , 2006 , apj , 648 , l5 reddy n.a . , steidel c.c . , pettini m. , adelberger k.l . , shapley a.e . , erb d.k . , dickinson m. , 2008 , apjs , 175 , 48 rodighiero g. , cimatti a. , franceschini a. , brusa m. , fritz j. , bolzonella m. , 2007 , a&a , 470 , 21 sawicki m. , thompson d. , 2006 , apj , 642 , 653 schmidt m. , 1968 , apj , 151 , 393 shapley a.e . , steidel c.c . , pettini m. , adelberger k.l . , 2003 , apj , 588 , 65 shimasaku k. , ouchi m. , furusawa h. , yoshida m. , kashikawa n. , okamura s. , 2005 , pasj , 57 , 447 sekiguchi k. , et al . , 2005 , in renzini a. & bender r. ed . , multiwavelength mapping of galaxy formation and evolution . springer - verlag , berlin , p82 sheth r.k . , mo h.j . , tormen g. , 2001 , mnras , 323 , 1 stark d.p . , bunker a.j . , ellis r.s . , eyles l.p . , lacy m. , 2007b , apj , 659 , 84 stark d.p . , ellis r.s . , richard j. , kneib j.p . , smith g.p . , santos m.r . , 2007a , apj , 663 , 10 steidel c.c . , adelberger k.l , giavalisco m. , dickinson m. , pettini m. , 1999 , apj , 519 , 1 taniguchi y. , et al . , 2005 , pasj , 57 , 165 warren , s. j. , et al . , 2007 , mnras , 375 , 213 yan h. , windhorst r.a . , 2004 , apj , 612 , l93 yan h. , dickinson m. , giavalisco m. , stern d. , eisenhardt p.r.m . , ferguson h.c . , 2006 , apj , 651 , 24 yoshida m. , et al . , 2006 , apj , 653 , 988 when deriving our estimates of the lbg luminosity functions at @xmath9 and @xmath12 in section 4 we employed several techniques which differ somewhat from those which have been commonly adopted in previous literature studies . consequently , it is perhaps instructive to investigate what influence these techniques have on the derived luminosity function estimates . in this section we first investigate the influence of our completeness corrections , before proceeding to investigate the influence of our decision to make use of the full photometric redshift probability density function for each lbg candidate . as previously discussed in section 4 , to estimate the lbg luminosity function at @xmath9 and @xmath12 we used the @xmath90 maximum likelihood estimator of schmidt ( 1968 ) , adapted to incorporate the photometric redshift probability density function of each lbg candidate . consequently , within a given redshift interval ( @xmath91 ) , our estimate of the luminosity function within a given absolute magnitude bin was : @xmath92 where the summation runs over the full sample , and @xmath93 is the redshift range within which a candidate s absolute magnitude lies within the magnitude bin in question . as described in detail in section 4 , the quantity @xmath94 is a correction factor derived from realistic monte - carlo simultions of the evolving lbg population , and accounts for sample incompleteness , object blending and contamination from objects photometrically scattered into the sample from faint - ward of the @xmath58band magnitude limit . to illustrate the influence of the correction factor we show in fig a1 our estimates of the @xmath9 and @xmath12 lbg luminosity functions with and without the inclusion of @xmath94 . in fig a1 the open and filled black symbols are the estimates of the @xmath9 and @xmath12 luminosity functions including the correction factor ( identical to the estimates displayed in fig 1 ) , while the grey symbols are the equivalent estimates setting @xmath227 . it can be seen from fig a1 that the correction factor does not have a significant impact on the shape of the luminosity function estimates . at the bright end , where our sample is most complete , the off - set between the black and grey data - points is simply due to the @xmath95% correction for objects lost due to blending in the crowded subaru images . in the faintest absolute magnitude bins the inclusion of the correction factor boosts the number densities by a roughly constant factor of @xmath228% ( 0.1 dex ) . this effect is due to a balance between the increasing positive correction that has to be made for incompleteness as the data approach the @xmath61 magnitude limit , and the increasingly negative correction that has to be made for objects photometrically scattered into the sample from faint - ward of the magnitude limit . when deriving our estimates of the lbg luminosity functions at @xmath9 and @xmath12 in section 4 we used the full photometric redshift probability density functions provided by our sed template fitting code in order to make full use of the available information . in this section we investigate the influence of this technique on the derived luminosity function estimates . of particular interest is to investigate how the derived luminosity function estimates change if we simply treat the best - fitting photometric redshift for each lbg as unique , and ignore the possibility of secondary redshift solutions . the first straightforward comparison of the two techniques is to compare the estimated redshift distributions . in fig a2 we show the lbg redshift distributions in the three redshift intervals utilised in this paper : @xmath143 , @xmath163 and @xmath147 . the redshift distributions shown in the left - hand column are derived by treating the best - fitting photometric redshift for each lbg as unique and robust . the redshift distributions shown in the right - hand column are derived by integrating over the full redshift probability density functions as done in section 4 . the number of objects included in each redshift distribution is listed in the top right - hand corner of each panel . although the redshift distributions derived via both techniques are broadly similar , it can be seen from fig a2 that they do differ in detail . perhaps the most noticeable difference is that adopting the best - fitting photometric redshifts alone predicts @xmath140% more objects in the highest redshift interval ( @xmath147 ) . armed with this information we now proceed to investigate how the differences in redshift distributions affect the estimated luminosity functions . in fig a3 we show estimates for the bright end of the @xmath9 and @xmath12 lbg luminosity functions . the left - hand panels show the estimates derived in section 4 by adopting the photometric redshift probability density function for each lbg . the right - hand panels show the luminosity function estimates derived from simply adopting the best - fitting photometric redshift for each source . in each panel we also plot the maximum likelihood schechter function fits to the lbg luminosity function at either @xmath9 or @xmath12 as derived in section 5.2 . it can be seen from fig a3 that the estimate of the @xmath9 luminosity function is virtually identical using both techniques . this is as expected given the similarity of the @xmath143 redshift distributions shown in the top panel of fig a2 , and demonstrates that the photometric redshifts at @xmath19 are sufficiently robust that both techniques lead to indistinguishable conclusions . in contrast , although the two luminosity function estimates at @xmath12 are clearly consistent , there are differences in detail . specifically , it can be seen from fig a3 that the luminosity function estimate based on the best - fitting photometric redshifts predicts more objects in the three faintest absolute magnitude bins . this effect is entirely consistent with the differences between the @xmath147 redshift distributions shown in fig a2 , which demonstrate that using the best - fitting photometric redshifts alone predicts @xmath140% more objects at @xmath20 . in conclusion , it is clear from the information displayed in figs a2 & a3 that the only substantive difference between the two techniques is for the faintest lbgs at @xmath20 . this is perhaps to be expected , given that these are the very objects for which the best - fitting photometric redshifts are least robust , and was the primary motivation for exploiting the extra information contained within the full redshift probability density function for each object . however , the good agreement between the luminosity function estimate shown in the bottom right panel of fig a3 and the best - fitting schechter function derived in section 5.2 confirms that adopting either technique would lead to the same conclusions regarding the form , and evolution , of the lbg luminosity function between @xmath9 and @xmath12 .

we present a new numerical code which is designed to derive a spectral energy distribution ( sed ) for an arbitrary spatial distribution of stellar and gaseous components in a dusty starburst galaxy . we apply a ray tracing method to numerical simulations and thereby estimate extinction and reemission of stellar light by dusty gas in an explicitly self - consistent manner . by using this code , we can investigate simultaneously dynamical and photometric evolution of a dusty galaxy based on stellar and gaseous dynamical simulations . as an example , we demonstrate when and how a major galaxy merger with dusty starburst becomes an ultra - luminous infrared galaxy owing to strong internal dust extinction . we furthermore discuss advantages and disadvantages of the present new code in clarifying the nature and the origin of low and high redshift dusty starburst galaxies .

astro-ph0004391

recent observational studies have discovered possible dusty starburst candidates in various classes of galactic objects such as low z ultra - luminous infrared galaxies ( ulirgs ) ( e.g. , sanders et al . 1988 ; sanders & mirabel 1996 ) , intermediate z ones ( e.g. , tran et al . 1999 ) , faint submillimeter sources detected by the sub - millimeter common - user bolometer array ( scuba ) ( holland et al . 1999 ) on the james clerk maxwell telescope ( smail , ivison , & blain 1997 ; hughes et al . 1998 ; barger et al . 1998 ; smail et al . 1998 ; ivison et al . 1999 ; lilly et al . 1999 ) , extremely red objects ( eros ) ( elston , rieke , & rieke 1988 ; graham & dey 1996 ; cimatti et al . 1998 ; dey et al . 1999 ; smail et al . 1999 ) , optically faint radio sources detected by the very large array ( vla ) ( e.g. , richards et al . 1999 ) , lyman - break galaxies ( steidel et al . 1996 ; lowenthal et al . 1997 ) , and near - infrared emission line galaxies ( e.g. , mannucci et al . these dusty starburst candidates are generally considered to provide valuable information on the formation and the evolution of galaxies , accordingly the origin and the nature of these galaxies have been extensively discussed in variously different contexts . these discussions include , for example , the importance of dissipative dynamics in the formation of elliptical galaxies at low and high redshift ( kormendy & sanders 1992 ; barnes & hernquist 1992 ) , an evolutionary link between starburst and active galactic nuclei ( norman & scoville 1988 ) , fueling of dusty gas to the central @xmath0 100 pc for nuclear starburst in major galaxy mergers ( e.g. , mihos & hernquist 1996 ) , physical correlations between morphological and photometric properties in ulirgs ( bekki , shioya , & tanaka 1999 ; bekki & shioya 2000a ) , merging and clustering process of high @xmath1 dusty mergers within a hierarchical clustering scenario ( e.g. , somerville & primack 1998 ) , and cosmic star formation history ( e.g. , pei & fall 1995 ; madau et al . 1996 ; meurer et al . 1997 ; madau , pozzetti , & dickinson 1998 ; pascarelle , lanzetta , & fernandez - soto 1998 ; blain et al . 1999 ; steidel et al . 1999 ) . one of important problems in extra - galactic astronomy is to settle the above discussions on high redshift dusty starburst galaxies , though observational analysis still has some difficulties in determining precisely redshifts ( sanders 1999 ) , identifying optical counterparts ( richards 1999 ) , and estimating the degree of dust extinction ( e.g. , meurer , heckman , & calzetti 1999 ) for high @xmath1 dusty starburst galaxies . spectral energy distributions ( seds ) are generally considered to be one of essentially important factors that can determine the nature of dusty starburst galaxies . accordingly several theoretical attempts have been made to derive the seds of galaxies with dusty starburst regions . rowan - robinson & crawford ( 1989 ) discussed how the far - infrared spectra of a sample of galaxies observed by @xmath2 @xmath3 @xmath4 ( iras ) can be reproduced by a theoretical model with a given temperature of stars embedded by dust , the optical depth , and the density distribution ( or geometry ) of dust . witt et al . ( 1991 ) investigated transfer processes of stellar light for models with variously different spherical geometries of dust with special emphasis on the effects of scattered light on photometric properties of galaxies and seds . calzetti , kinney , & storchi - bergmann ( 1994 ) analyzed ultraviolet ( uv ) and optical spectra of 39 starburst and blue compact galaxies and thereby provided an analytical formulation of the effects of dust extinction in galaxies and an effective extinction law for correcting the observed uv and optical spectral continua . witt & gordon ( 1996 ) investigated the radiative transfer processes of a central stellar source surrounded by a spherical , statistically homogeneous but clumpy two - phase scattering medium and found that the structure of dusty medium can greatly affect the conversion of uv , optical , and near - infrared radiation into thermal far - ir dust radiation in a dusty system . by comparing the observed seds of 30 starburst galaxies with theoretical radiative transfer models of dusty systems , gordon , calzetti , and witt ( 1997 ) discussed the importance of geometry of stellar and gaseous components in determining the sed of a dusty starburst galaxy . takagi , arimoto , & vansevi@xmath5ius ( 1999 ) investigated radiative transfer models with variously different ages of secondary starburst components and optical depths for dusty starburst galaxies and proposed a new method for estimating precisely the effect of ages of young starburst populations and that of dust attenuation on the shape of sed in uv and near@xmath6infrared bands . efstathiou , rowan - robinson , & siebenmorgen ( 2000 ) treated starburst galaxies as an ensemble of optically thick giant molecular clouds ( gmcs ) centrally illuminated by recently formed stars and thereby constructed a new radiative transfer model for calculating seds from uv to millimeter band of dusty starburst galaxies . by using this model , they discussed how the age and the star formation history of a dusty starburst galaxy can control the sed particularly for the prototypical starburst galaxy m82 and ngc 6090 . although the above previous studies have succeeded in clarifying important dependences of seds on physical parameters such as spatial distribution of dust , geometries of stellar and gaseous components , and dust properties in dusty starburst galaxies , many of them did not discuss @xmath7 @xmath8 @xmath9 @xmath10 @xmath11 . main reasons for @xmath12 previous theoretical studies not discussing the time evolution of seds are the following three . firstly previous models did not follow the time evolution of stellar and gaseous distribution and accordingly could not derive the time evolution of seds in dusty starburst galaxies . secondly , hydro - dynamical evolution of interstellar gas ( e.g. , time evolution of gaseous density ) was not included in previous studies , and accordingly the time evolution of optical depth could not be derived . thirdly , some previous studies did not consider chemical evolution of dusty interstellar medium and that of stellar components , they could not follow time evolution of metallicity and that of dust properties . considering that low z infrared luminous galaxies with dusty starburst and high @xmath1 faint scuba sources with possible dusty starburst show very peculiar morphology ( sanders et al . 1988 ; sanders & mirabel 1996 ; smail et al . 1998 ) , spherical symmetric approximation or axisymmetric one adopted in previous theoretical and numerical studies for the stellar and gaseous distribution of a dusty galaxy should be also relaxed in order that the nature and the origin of low and high z dusty galaxies can be extensively investigated by theoretical studies . the purpose of the present paper and our future papers is to investigate in detail the formation and the evolution of dusty starburst galaxies by using a new numerical code for deriving the time evolution of seds of these galaxies . we first perform numerical simulations that can follow dynamical evolution of stellar and gaseous components , star formation history , and chemical evolution for dusty starburst galaxies in an explicitly self - consistent manner . we then derive stellar and gaseous distribution and age and metallicity distribution of stellar populations and thereby calculate galactic seds . furthermore we describe how the present numerical code is useful and helpful for investigating variously different physical properties of low and high z dusty galaxies . as an example , we here present the results of a major merger with dusty interstellar gas . we particularly demonstrate how dynamical evolution of stellar and gaseous components controls the time evolution of seds in dusty galaxy mergers and thus their photometric evolution . the layout of this paper is as follows . in 2 , we summarize numerical models used in the present study and describe in detail the methods for deriving the seds corrected by internal dust extinction . in this section , we also point out the limitations of the present numerical code . in 3 , we present numerical results on the time evolution of morphology , sed , and photometric properties in a gas - rich major merger . in 4 , we discuss the origin of high @xmath1 dusty starburst galaxies such as faint scuba sources and eros . the conclusions of the preset study are given in 5 . the most remarkable difference in deriving seds of dusty galaxies between the present model and previous models ( e.g. , mazzei , xu , & de zotti 1992 ; franceschini et al . 1994 ; witt & gordon 1996 ; gordon st al . 1997 ; guiderdoni et al . 1998 ) is that we derive seds of galaxies based on the result of numerical simulations that can follow both dynamical and chemical evolution of galaxies . the derivation of a sed for a galaxy with dusty starburst consists of the following three steps . in the first step we simultaneously derive age and metallicity distribution of stellar component in the galaxy , based on numerical simulations . secondly , we use a stellar population synthesis code ( e.g. bruzal & charlot 1993 ) and thereby calculate the sed of the galaxy , based on the derived age and metallicity distribution of stellar population in the galaxy . in this second step , the dust effects are not included . thirdly , we consider extinction and reemission of stellar light by dusty interstellar gas and modify the sed of the galaxy , by using our new code . the details of our new method for modifying galactic seds are given in 2.3 . we construct models of galaxy mergers between gas - rich disk galaxies with equal mass by using fall - efstathiou model ( 1980 ) . the total mass and the size of a progenitor disk are @xmath13 and @xmath14 , respectively . from now on , all the mass and length are measured in units of @xmath13 and @xmath14 , respectively , unless specified . velocity and time are measured in units of @xmath15 = @xmath16 and @xmath17 = @xmath18 , respectively , where @xmath19 is the gravitational constant and assumed to be 1.0 in the present study . if we adopt @xmath13 = 6.0 @xmath20 @xmath21 @xmath22 and @xmath14 = 17.5 kpc as a fiducial value , then @xmath15 = 1.21 @xmath20 @xmath23 km / s and @xmath17 = 1.41 @xmath20 @xmath24 yr , respectively . in the present model , the rotation curve becomes nearly flat at 0.35 radius with the maximum rotational velocity @xmath25 = 1.8 in our units . the corresponding total mass @xmath26 and halo mass @xmath27 are 5.0 and 4.0 in our units , respectively , which means that the baryonic mass fraction of the initial disk galaxy is 0.2 . the radial ( @xmath28 ) and vertical ( @xmath29 ) density profile of a disk are assumed to be proportional to @xmath30 with scale length @xmath31 = 0.2 and to @xmath32 with scale length @xmath33 = 0.04 in our units , respectively . the velocity dispersion of halo component at a given point is set to be isotropic and given according to the virial theorem . in addition to the rotational velocity made by the gravitational field of disk and halo component , the initial radial and azimuthal velocity dispersion are given to disk component according to the epicyclic theory with toomre s parameter ( @xcite ) @xmath34 = 1.2 . the vertical velocity dispersion at given radius is set to be 0.5 times as large as the radial velocity dispersion at that point , as is consistent with the observed trend of the milky way ( e.g. , wielen 1977 ) . as is described above , the present initial disk model does not include any remarkable bulge components , and accordingly corresponds to ` purely ' late - type spiral without galactic bulge . although it is highly possible that galactic bulges greatly affect the chemical evolution of galaxy mergers , we however investigate this issue in our future papers . the collisional and dissipative nature of the interstellar medium is modeled by the sticky particle method ( @xcite ) . it should be emphasized here that this discrete cloud model can at best represent the @xmath35 interstellar medium of galaxies in a schematic way . as is modeled by mckee & ostriker ( 1977 ) , the interstellar medium can be considered to be composed mainly of ` hot ' , ` warm ' , and ` cool ' gas , each of which mutually interacts hydrodynamically in a rather complicated way . actually , these considerably complicated nature of interstellar medium in disk galaxies would not be so simply modeled by the ` sticky particle ' method in which gaseous dissipation is modeled by ad hoc cloud - cloud collision : any existing numerical method probably could not model the @xmath35 interstellar medium in an admittedly proper way . in the present study , as a compromise , we only try to address some important aspects of hydrodynamical interaction between interstellar medium in disk galaxies and in dissipative mergers . more elaborated numerical modeling for real interstellar medium would be necessary for our further understanding of dynamical evolution in dissipative galaxy mergers . we assume that the fraction of gas mass ( @xmath36 ) in a disk is set to be 0.5 initially . actually , the gas mass fraction in precursor disks of a merger is different between galaxy mergers and depends on the epoch of the merging . for example , recent observational results on the total mass of molecular gas for faint scuba sources ( e.g. , frayer et al . 1999 ) revealed that smm j14011 + 0252 with some indications of major merging at z = 2.565 has @xmath0 5.0 @xmath20 @xmath21 @xmath37 . if the progenitor of this gas - rich high redshift merger has a mass of 6.0 @xmath20 @xmath21 @xmath22 ( i.e. , the same as that of the galaxy ) , the gas mass fraction is @xmath0 0.42 for this galaxy . the gas mass fraction is observationally suggested to be larger for higher redshift mergers ( e.g. , evans , surace , & mazzarella 1999 ) . guided by these observational results , we adopt the above value ( 0.5 ) as the initial gas mass fraction in order to discuss higher redshift galaxy mergers . although the difference of gas mass fraction probably could yield a great variety of chemical and dynamical structures in mergers , we do not intend to consider this important difference for simplicity in the present paper and will address in our future paper . the radial and tangential restitution coefficient for cloud - cloud collisions are set to be 1.0 and 0.0 , respectively . the number of particles for an above isolated galaxy is 10000 for dark halo , 10000 for stellar disk components , and 10000 for gaseous ones . in all of the simulations of mergers , the orbit of the two disks is set to be initially in the @xmath38 plane and the distance between the center of mass of the two disks , represented by @xmath39 , is assumed to be the free parameter which controls the epoch of galaxy merging . the pericenter distance , represented by @xmath40 , is also assumed to be the free parameter which controls the initial total orbital angular momentum of galaxy mergers . the eccentricity is set to be 1.0 for all models of mergers , meaning that the encounter of galaxy merging is parabolic . the spin of each galaxy in a pair merger is specified by two angle @xmath41 and @xmath42 , where suffix @xmath43 is used to identify each galaxy . @xmath41 is the angle between the @xmath1 axis and the vector of the angular momentum of a disk . @xmath42 is the azimuthal angle measured from @xmath44 axis to the projection of the angular momentum vector of a disk onto @xmath44-@xmath45 plane . in the present study , we show the results of only one model with @xmath46 = 0.0 , @xmath47 = @xmath48 , @xmath49 = 0.0 , @xmath50 = 0.0 , @xmath40 = 17.5 kpc , and @xmath39 = 140 kpc : this model describes a nearly prograde - retrograde merger . the results of the models with variously different @xmath41 , @xmath42 , @xmath40 and @xmath39 will be described in our future papers . the time when the progenitor disks merge completely and reach the dynamical equilibrium is less than 15.0 in our units for most of models and does not depend so strongly on the history of star formation in the present calculations . star formation is modeled by converting the collisional gas particles into collisionless new stellar particles according to the algorithm of star formation described below . we adopt the schmidt law ( schmidt 1959 ) with exponent @xmath51 = 2.0 ( 1.0 @xmath52 @xmath51 @xmath52 2.0 , @xcite ) as the controlling parameter of the rate of star formation . the amount of gas consumed by star formation for each gas particle in each time step , @xmath53 , is given as : @xmath54 where @xmath55 and @xmath56 are the gas density around each gas particle and the mean gas density at 0.48 radius of an initial disk , respectively . this density - dependent star formation model is similar to that of mihos , richstone , & bothun ( 1992 ) and the probability approach of the gas consumption rate is similar to that described by katz ( 1992 ) and oneil et al . ( 1998 ) . in order to avoid a large number of new stellar particles with different mass , we convert one gas particle into one stellar one according to the following procedure . first we give each gas particle the probability , @xmath57 , that the gas particle is converted into stellar one , by setting the @xmath57 to be proportional to the @xmath53 in equation ( 1 ) estimated for the gas particle . then we draw the random number to determine whether or not the gas particle is totally converted into one new star . this method of star formation enables us to control the rapidity of star formation without increase of particle number in each simulation thus to maintain the numerical accuracy in each simulation . the @xmath58 in the equation ( 1 ) is the parameter that controls the rapidity of gas consumption by star formation . we determine the initial value of the @xmath58 so that mean star formation rate of an isolated late - type disk galaxy model with the typical gas mass fraction of @xmath0 0.2 can become an order of 1 @xmath59 @xmath60 for the first 1 gyr evolution . the positions and velocity of the new stellar particles are set to be the same as those of original gas particles . all the calculations related to the above dynamical evolution including the dissipative dynamics , star formation , and gravitational interaction between collisionless and collisional component have been carried out on the grape board ( @xcite ) at astronomical institute of tohoku university . the parameter of gravitational softening is set to be fixed at 0.03 in all the simulations . the time integration of the equation of motion is performed by using 2-order leap - flog method . energy and angular momentum are conserved within 1 percent accuracy in a test collisionless merger simulation . most of the calculations are set to be stopped at t = 20.0 in our units unless specified . chemical enrichment through star formation during galaxy merging is assumed to proceed both locally and instantaneously in the present study . the model for analyzing metal enrichment of each gas and stellar particle is as follows . first , as soon as a gas particle is converted into a new stellar one by star formation , we search neighbor gas particles locating within @xmath61 from the position of the new stellar particle and then count the number of the neighbor gas particles , @xmath62 . this @xmath61 is referred to as chemical mixing length in the present paper , and represents the region within which the neighbor gas particles are polluted by metals ejected from the new stellar particle . the value of @xmath61 relative to the typical size of a galaxy is set to be 0.01 . next we assign the metallicity of original gas particle to the new stellar particle and increase the metals of the each neighbor gas particle according to the following equation about the chemical enrichment : @xmath63 where the @xmath64 represents the increase of metal for each gas particle . @xmath65 , @xmath66 , @xmath67 , and @xmath68 in the above equation represent the metallicity of the new stellar particle ( or that of original gas particle ) , the fraction of gas returned to interstellar medium , the mass of the new star , and the chemical yield , respectively . the value of the @xmath66 and that of @xmath68 are set to be 0.2 and 0.03 , respectively . furthermore , the time , @xmath69 , when the new stellar particle is created , is assigned to the new stellar particle in order to calculate the photometric evolution of merger remnants , as is described later . to verify the accuracy of the above treatment ( including numerical code ) for chemical enrichment process , we checked whether or not the following conservation law of chemical enrichment is satisfied for each time step in each test simulation : @xmath70 where @xmath71 , @xmath72 , and @xmath65 are the mass of each gas particle , that of each stellar one , and the metallicity of each particle , respectively , and the summation ( @xmath73 ) is done for all the gas particles or stellar ones . strictly speaking , the above equation holds when the value of @xmath66 is 0.0 . thus , in testing the validity of the present code of chemical enrichment , we set the value of @xmath66 to be 0.0 and then perform a simulation for the test . we confirmed that the above equation is nearly exactly satisfied in our test simulations and furthermore that even if the @xmath66 is not 0.0 , the difference in the value of total metallicity between the left and right side in the above equation is negligibly small . our numerical results do not depend so strongly on @xmath61 , @xmath66 , @xmath68 within plausible and realistic ranges of these parameters . it is assumed in the present study that the spectral energy distribution ( sed ) of a model galaxy is a sum of the sed of stellar particles . the sed of each stellar particle is assumed to be a simple stellar population ( ssp ) that is a coeval and chemically homogeneous assembly of stars . thus the monochromatic flux of a galaxy with age @xmath74 , @xmath75 , is described as @xmath76 where @xmath77 and @xmath67 are a monochromatic flux of ssp of age @xmath78 and metallicity @xmath79 , where suffix @xmath43 identifies each stellar particle , and mass of each stellar particle , respectively . the age of ssp , @xmath78 , is defined as @xmath80 , where @xmath69 is the time when a gas particle is converted into a stellar one . the metallicity of ssp is exactly the same as that of the stellar particle , @xmath79 , and the summation ( @xmath73 ) in equation ( 4 ) is done for all stellar particles in a model galaxy . a stellar particle is assumed to be composed of stars whose age and metallicity are exactly the same as those of the stellar particle and the total mass of the stars is set to be the same as that of the stellar particle . thus the monochromatic flux of ssp at a given wavelength is defined as @xmath81 where @xmath82 is mass of a star , @xmath83 is a monochromatic flux of a star with mass @xmath82 , metallicity @xmath79 and age @xmath78 . @xmath84 is a initial mass function ( imf ) of stars and @xmath85 , @xmath86 are upper and lower mass limit of imf , respectively . we here adopt the salpter imf with @xmath85 = 120@xmath37 and @xmath86 = 0.1 @xmath37 . in this paper , we use the @xmath87 of gissel96 , which is the latest version of bruzual & charlot ( 1993 ) . using the derived each stellar particle s sed not corrected by dust ( @xmath88 shown in equation ( 4 ) ) and stellar and gaseous distribution , we can obtain the sed of a galaxy corrected by dust . from now on the sed of a stellar particle , @xmath89 , is referred to as @xmath90 for convenience . we first calculate dust extinction of star light for @xmath91 stellar particle and dust temperature for @xmath91 gaseous particle , based on the three dimensional spatial distribution of stellar and gaseous particles . we then sum each stellar particle s sed corrected by dust extinction and the dust reemission of each gaseous particle . the method to derive the dust extinction and reemission for each particle consists of the following three steps . firstly we assume that each stellar particle with the sed @xmath90 emits rays with the sed for each of the rays equal with @xmath90/@xmath92 , where @xmath92 is the total number of the rays . we generate random numbers for each ray and thereby determine the direction of each ray from the stellar particle . figure 1 represents a schematic explanation of this first step . @xmath92 is set to be 10 for all stellar particles in the present study . secondly , we investigate whether each of rays emitted from a stellar particle can penetrate a gas particle for all gas particles and thereby estimate dust extinction of stellar light for each of the rays . this investigation and estimation is done for all stellar particles . in this second step , we regard a ray as penetrating a gas particle and thus being affected by internal dust extinction if the ray passes through the region within the gas cloud radius ( @xmath93 ) . this second step is also schematically shown in figure 1 . here gas clouds with the masses of @xmath94 are assumed to be spherical and the cloud radius @xmath93 is changed following the relation ( kwan & valdes 1987 ) , @xmath95 although physical processes associated with the growth and the disruption of gas clouds have been investigated ( e.g. , kwan 1979 ; kwan & valdes 1987 ; olson & kwan 1990a , b ) , it is not so clear which the most probable and physically reasonable value of @xmath96 is for evolving galaxies . a probable value of @xmath96 is @xmath0 1.0 for disk galaxies ( kwan & valdes 1987 ) whereas the typical cloud mass and size depend strongly on time in merging galaxies owing to coalescence and disruption of gas clouds ( olson & kwan 1990a , b ) . olson & kwan ( 1990b ) demonstrated that although most of the cloud - cloud collisions induced by galaxy merging is disruptive , several very large gas clouds with masses greater than @xmath97 @xmath37 can form after merging owing to successive coalescence of colliding clouds . furthermore the density of gas clouds in ulirgs considered to be ongoing mergers is suggested to be very high from that in disk galaxies ( aalto et al . 1991a , b ) , which implies that the size and the mass might be also different from those of isolated disk galaxies . accordingly it is reasonable for us to assume that @xmath96 is a free parameter for the present study . in the present paper , however , we show only the results for the model with @xmath96 = 4.68 . the @xmath96 dependence in major mergers will be described in our future papers . absorption of stellar light for each of rays from a stellar particle is modeled according to the following reddening formulation ( black 1987 ; mazzei et al . 1992 ) ; @xmath98 , where @xmath99 and @xmath100 are gaseous column density and gaseous metallicity , respectively . since dust absorption is estimated for each of gas particles penetrated by a ray emitted from a stellar particle , the above @xmath99 and @xmath100 correspond to column density of each gas cloud particle ( @xmath101 ) and metallicity of the particle ( @xmath102 ) , respectively . the @xmath102 is derived from our numerical simulations and the @xmath101 is derived from the relation between cloud mass and size shown in equation ( 6 ) . using the derived redding @xmath103 and extinction law by cardelli et al . ( 1989 ) , we adopt the so - called screen model for rays from a stellar particle and calculate the dust absorption of the stellar particle . thirdly , by assuming the modified black body radiation with the emissivity ( @xmath104 ) law @xmath105 , we determine the dust temperature of a gas particle such that total energy flux of dust absorption is equal to that of dust reemission . in the present study , we show the results of a model with @xmath106 = 2 . we here do not include the albedo for dust grains , which means that only stars and new stars heat dust . thus the sed of a galaxy consists of the stellar continuum modified by dust extinction and the reemission of dusty gas . the method adopted in the present numerical code calculating dust absorption and reemission is essentially the same as the so - called ray tracing method adopted in previous theoretical calculations on radiative transfer problems ( e.g. witt 1977 ; efstathiou & rowan - robinson 1990 ) . this paper is the first step toward a sophisticated combination of n - body simulations with the already existing ray tracing method . therefore there are still some problems of the present new code in calculating correctly the sed within feasible time scale for a simulation with the total particle number approximately equal to @xmath107 . the most significant problem is on the time which we spent in calculating the sed . the time necessary for the sed calculation for each time step in a simulation can be roughly scaled to @xmath108 , where @xmath109 , @xmath110 , and @xmath111 are total particle number of stars , that of gas , and that of rays emitted from each stellar particle . therefore it is very time - consuming to derive a sed in each time step of a simulation , in particular , for large n - body simulations with both @xmath109 and @xmath112 larger than @xmath107 : it is obviously our future work to develop a new method for calculating seds of such large n - body systems . furthermore , our sed calculation becomes also time - consuming , unless we carefully choose the parameter value of @xmath111 . although we can more correctly estimate seds of galaxy mergers with dusty starburst for the models with larger @xmath111 , we should adopt a feasible and reasonable value of @xmath111 , considering that the time for our sed calculation is roughly proportional to @xmath113 . in order to determine a plausible value of @xmath111 , we investigated the dependence of a sed on @xmath111 ( 1 @xmath114 @xmath111 @xmath114 50 ) in a dusty starburst merger model and found that if @xmath111 @xmath115 10 , the differences in seds between models with different @xmath111 become negligibly small . we thus adopt @xmath111 = 10 as a feasible and plausible value for our sed calculations in the present study . although our new numerical method allows us to derive a sed for an arbitrary distribution of stellar and gaseous component for a dusty galaxy in numerical simulations , it still has some problems such as those described above . both our method and previous ones ( i.e. , those by which numerical studies can precisely solve radiative transfer processes and estimate galactic seds within a certain spherical symmetric approximation or axisymmetric one ) have advantages and disadvantages , and therefore we stress that the present numerical study is complementary to previous ones . our numerical model can derive a sed for a galaxy with an arbitrary distribution of stellar and gaseous components and accordingly enable us to examine a sed of a galaxy with very peculiar mass distribution , such as galaxy mergers and forming galaxies . however , our model has the following four limitations in investigating the time evolution of seds in galaxies . firstly , we do not include physical processes related to dust production , grain destruction , and grain growth in interstellar medium in the present model and thus can not follow the time evolution of dust composition . dwek ( 1999 ) developed a new self - consistent model for the evolution of the compositions and abundances of elements and the dust in galaxies by including condensation , accretion , and destruction processes of grains in the model . by using this model , dwek ( 1999 ) investigated the time evolution of dust composition ( e.g. , whether interstellar gas is silicate rich or not ) and suggested that extinction law , which is an important factor for calculating a galactic sed , depends on the evolution . the method developed by dwek ( 1999 ) is mainly for one - zone models of galactic chemical evolution , accordingly it can not be so simply applied to n - body numerical simulations . it is thus our future study to combine the method provided by dwek ( 1999 ) with our numerical code and thereby to derive a galactic sed in a more self - consistent manner . furthermore chang , schiano , & wolfe ( 1987 ) investigated how the radiation from qsos can affect the hydrodynamical evolution of dusty interstellar gas , considering ion - field emission , ion sputtering , coupling between gas ions and dust grains , photoionization , and photoelectrical processes . they found that owing to the high sputtering on the grain surface in dusty interstellar gas , most of the grains inside 7 kpc of a qso host galaxy can be destroyed within @xmath0 3 @xmath20 @xmath97 yr . although their results are not directly related to the dust destruction processes in dusty starburst galaxies , we consider that the essence of their results can be applied to the case of starburst galaxies . the time scale for intense starburst in ulirgs formed by major mergers is suggested to be @xmath0 @xmath116 yr and the total luminosity of ulirgs are similar to qsos ( e.g. , sanders et al . 1988 ) , and accordingly the radiation from nuclear strong starburst can also destruct the dust grains in ulirgs . if the dust grains are efficiently destructed by radiation of starburst , the total flux of far - infrared reemission from dusty starburst can be greatly decreased . our present model does not include this effects of starburst radiation on dust grains , we stress that the total value of infrared and far - infrared luminosity can be overestimated in major mergers with dusty starburst . secondly , we do not consider the effects of small grains with the size between 1 and 10 nm , and consequently can not precisely estimate a sed around @xmath107 @xmath117 in a dusty system . devriendt , guiderdoni , & sadat ( 1999 ) included contributions from polycyclic aromatic hydrocarbons ( pahs ) , small grains , and big ones in their model and succeeded in reproducing reasonably well the seds of several ulirgs . to construct multi - component dust model is accordingly the next step of our studies toward the more precise estimation of seds of dusty galaxies . thirdly , the parameter @xmath96 is fixed at a possibly reasonable value during a simulation , accordingly our model can not follow the time evolution of cloud size and mass . witt & gordon ( 1996 ) demonstrated that the change of the structure of interstellar medium in a dusty system provides the change in effective optical depth and consequently controls the sed of the system . they furthermore suggested that a breakup of large interstellar cloud complexes into numerous smaller clouds and a intercloud medium of enhanced density in a galaxy merger can greatly affect the sed of the merger . fourthly , we do not include the effects of albedo at all in the present study , though several authors have already investigated in detail the effects of dust albedo ( e.g. , multiple scattering of stellar light ) on seds in dusty systems ( e.g. , witt & gordon 1996 ; gordon et al . 1997 ) . although the total number of parameters increase if we include the effects of dust albedo in our numerical simulations , it is our future work to investigate the importance of dust albedo in determining galactic seds . thus our numerical results should be carefully interpreted owing to the above four limitations of the code used in the present study . mainly important advantages of the present study are the following two . firstly we can investigate simultaneously morphological , structural , kinematical , and photometric properties at a given time step in a galaxy merger with dusty starburst based on the derived sed of the merger . secondly , by using the sed derived for a merger at a given z and considering the effects of k - correction , we can investigate how the morphological and photometric properties of the merger change with redshift . accordingly we mainly investigate the time evolution of morphology , star formation , global colors , luminosity and @xmath118 in a dusty starburst galaxy merger . furthermore we investigate two - dimensional distributions of colors , luminosity , and @xmath118 , which have not been investigated at all in previous theoretical studies of dusty galaxies . in order to discuss apparent morphology of intermediate and high @xmath1 dusty galaxies , we also investigate how dust - enshrouded starburst galaxy mergers at z = 0.4 , 1.0 , and 1.5 can be seen in the hubble space telescope ( @xmath119 ) . the method to construct the synthesized @xmath119 images of galactic morphology in the present study is basically the same as that described by mihos ( 1995 ) and bekki , shioya , & tanaka ( 1999 ) . we adopt madau s model ( madau 1995 ) for intervening absorption of neutral hydrogen and consider k - correction of galactic seds in order to derive the synthesized images . for comparison , we also show how dust - enshrouded starburst galaxy mergers at z = 0.4 , 1.0 , and 1.5 can be seen in the subaru which is a japanese large ( 8.2 m ) grand - based telescope . the results on the synthesized @xmath119 and subaru images are presented in 3.3 and for discussing the origin of high @xmath1 faint scuba sources and eros . in order to calculate the seds of the merger model at each redshift , we assume that mean ages of old stellar components initially in a merger progenitor disk at the redshift z=0.4 , 1.0 , and 1.5 are 7.14 , 3.80 , and 2.46 gyr , respectively . we here describe mainly the results of only one merger model with @xmath96 = 4.68 ( i.e. , @xmath93 is 0.011 in our units corresponding to 200 pc ) because the main purpose of the present study is not to give dependences of seds on physical parameters of galaxy mergers but to demonstrate usefulness of the present code in studying theoretically dusty starburst galaxies . from now on this mode is referred to as the standard model . we do not intend to change the values of important free parameters such as @xmath13 , @xmath26 , @xmath39 , @xmath40 , @xmath42 , @xmath41 , and @xmath36 in the present study . dependences of the evolution of dusty starburst mergers on parameters will be described in our future papers ( bekki & shioya 2000b ) . although each of the results described in the following sections have much implications on the nature of ulirgs and can be compared with up - to - date observational results such as near - infrared colors ( e.g. , scoville et al . 1999 ) , surface brightness distribution ( e.g. , sanders et al . 1999 ; scoville et al . 1999 ) , and high resolution optical / near - infrared images of ulirgs ( e.g. , surace , sanders , & evans 1999 ) , we do not intend to discuss so extensively the origin and the nature of ulirgs in the present study : this is simply because the most important purpose of the paper is to demonstrate the importance and the usefulness of the present code in investigating dusty starburst galaxies . we will discuss the origin of ulirgs in our future papers . in the followings , the cosmological parameters @xmath120 and @xmath121 are set to be 50 km s@xmath122 mpc@xmath122 and 0.5 respectively . figure 2 and 3 describe time evolution of morphology of the standard model for each of four components , dark halo , stars initially located within two disks , gas , and new stars formed during galaxy merging . from now on , for convenience , the time @xmath123 represents the time that has elapsed since the two disks begin to merge . as galaxy merging proceeds , the two disks are strongly disturbed to form a long tidal tail in the disk orbiting in a prograde sense at 0.6 @xmath124 @xmath123 @xmath124 1.1 gyr . a remarkable tidal tail is not developed in the disk orbiting in a retrograde sense . this one long tidal arm is characteristics of prograde - retrograde mergers . the two disks finally sink into the center of massive dark halos owing to dynamical friction during merging and consequently are completely destroyed by violent relaxation of galaxy merging ( 1.1 @xmath124 @xmath123 @xmath124 1.7 gyr ) . as a result of this , the two disks form an elliptical galaxies with the structure and kinematics similar to the observed ones ( @xmath123 = 1.7 gyr ) . during violent major merging , interstellar gas is very efficiently transferred to the central region of the two disks owing to gaseous dissipation of colliding gas clouds and gravitational torque . gas accumulated in the central region of the merger is then consumed by massive starburst and consequently converted into new stars . as is shown in figure 4 , the star formation rate becomes maximum ( @xmath125 @xmath37 @xmath126 ) at @xmath123 = 1.3 gyr when two disks of the merger become very close to suffering from violent relaxation . the maximum star formation rate is roughly two orders of magnitude larger than the mean star formation rate of an isolated disk in the present study and comparable to that required for explaining the strong infrared luminosity observed in ulirgs . after the strong starburst , star formation rate rapidly declines ( @xmath0 1 @xmath127 ) within less than 1 gyr after @xmath123 = 1.3 , essentially because most of the gas is consumed up by the starburst . these results are qualitatively consistent with those in mihos & hernquist ( 1996 ) . we here stress that the above one - time starburst in the late phase of major merging is applied only to the present prograde - retrograde merger with the adopted inclination of two disks and internal structure of the disks . mihos & bothun ( 1998 ) demonstrated that physical details such as internal structure of merger progenitor disks and their initial gas mass fraction are important for the star formation history and the luminosity evolution in ulirgs , by using imaging fabry - p@xmath128rot observations on four ulirgs . they furthermore found that the spatial distribution of h@xmath129 emission from starburst regions is very diverse in the four ulirgs ( i.e. , some ulirgs show the strong h@xmath129 emission only in the central region and h@xmath129 emission is quite extended in some ulirgs ) and accordingly suggested that several different factors play a role in triggering starburst in ulirgs . considering these important observational results , we suggest that the results on the star formation history and the resultant luminosity evolution described in the present study are only true for some ulirgs . figure 5 and 6 give mass distribution projected onto @xmath44-@xmath45 plane ( orbital plane ) and @xmath44-@xmath1 plane at @xmath123 = 1.1 ( pre - starburst phase ) and 1.7 gyr ( post - starburst phase ) for stars , gas , and new stars . new stars formed mainly by secondary massive starburst are more compactly distributed in the merger than old stars initially located within disks at the pre - starburst ( weak starburst ) epoch when the two cores in the merger have not yet merged with each other to form an elliptical galaxy . this is essentially because new stars experience much more gaseous dissipation when they were previously gaseous components . consequently new stars with younger ages are more heavily obscured by dusty gas than old stars during galaxy merging . this result that young stellar components can be preferentially obscured by dust during evolution of galaxy mergers is suggested to be very important for understanding the nature of post - starburst galaxies detected by smail et al . ( 1999 ) in scuba surveys of intermediate redshift clusters of galaxies ( shioya & bekki 2000 ) . three very small compact stellar clusters composed mainly of new stars ( and gas ) can be seen above the two cores at @xmath123 = 1.1 . this results suggests that efficient star formation during galaxy merging can occur not only in the central part of a merger but also in stellar clusters located in the outer part of the merger . as is shown in figure 6 , the morphology of the merger only @xmath0 0.4 gyr after the maximum starburst looks like an elliptical galaxy , which implies that the time scale within which a merger can be seen as an ulirg with the very peculiar morphology is very short ( less than 0.5 gyr ) . figure 7 describes the seds of the merger which we derive based on the mass distribution of stellar and gaseous component shown in figure 2 and 3 and the age and metallicity distribution of stellar populations of the merger at @xmath123 = 0.6 , 1.1 , 1.3 ( the epoch of maximum starburst ) , 1.7 , 2.3 , and 2.8 gyr . we can clearly observe how the dust extinction of interstellar gas can change the shape of the sed of the merger at each time by comparing the results of the model with dust extinction with those without dust extinction . the uv flux with the wavelength less than 3000 @xmath130 rapidly increases during 1.1 @xmath124 @xmath123 @xmath124 1.3 gyr owing to a large number of young massive stars formed by the strong starburst whereas it decreases during 1.3 @xmath124 @xmath123 @xmath124 2.8 gyr because of very small star formation rate after the starburst in the model without dust extinction . the infrared and submillimeter fluxes become larger during 1.1 @xmath124 @xmath123 @xmath124 1.3 gyr in the model with dust extinction . this is firstly because star formation rate , which is closely associated with the total mount of stellar light absorbed by interstellar dust , becomes considerable higher owing to the efficient gas transfer to the central region of the merger and secondly because the density of dusty gas becomes also very high in the later phase of the merging ( 1.1 @xmath124 @xmath123 @xmath124 1.3 gyr ) so that the gas can heavily obscure the strong starburst . after the maximum starburst at @xmath123 = 1.3 gyr , both infrared and submillimeter fluxes rapidly decline ( 1.3 @xmath124 @xmath123 @xmath124 1.7 gyr ) this is principally because most of interstellar gas indispensable for strong starburst and dust obscuration is rapidly consumed by star formation in the merger till @xmath123 @xmath0 1.7 gyr . thus the time evolution of sed of a merger depends strongly on that of the star formation rate , which is basically controlled by dynamical evolution of the merger . figure 8 and 9 describe the mass distribution and the sed at the epoch of maximum starburst in the standard model , respectively . figure 8 clearly shows that both new stars and gas are more centrally concentrated than old stellar components , which means that star light from new stars formed by the massive starburst are more heavily obscured by dusty interstellar medium than that of old stellar components . the sed at @xmath123 = 1.3 gyr in figure 9 is rather similar to the observed sed of typical ulirgs , which implies that the present dusty starburst merger model can be observed as an ulirg in the late of galaxy merging . these results in figure 8 and 9 clearly demonstrate that starburst population in the merger is so heavily obscured by dusty interstellar medium ( the mean @xmath131 mag ) that the dust reemission in far - infrared ranges ( @xmath132 ) becomes very strong at the maximum starburst ( @xmath133 ) . the reasons for this heavy dust extinction are firstly that column density of dusty interstellar medium becomes extremely high owing to the strong central accumulation of gas and secondly that the central gaseous metallicity , which is a measure of total amount of dust , also becomes rather large ( @xmath134 ) because of efficient and rapid chemical evolution in galaxy merging with secondary starburst . based on the derived mass distribution and seds at the maximum starburst ( at ulirg phase ) , we can examine two - dimensional distribution of @xmath135 band surface brightness ( figure 10 ) , rest - frame @xmath136 color ( 11 ) , and @xmath137 ( 12 ) , all of which have been observationally investigated in a very extensive manner for the better understanding of the nature of ulirgs . as is shown in figure 10 , near - infrared surface brightness is the highest ( the smallest in number ) in the central region within which most of young bright massive stars are located . this is essentially because most of young stars are formed in the central region at @xmath123 = 1.3 gyr . the @xmath136 color is also the largest in the central part of the merger ( figure 11 ) , because the central starburst components are most heavily obscured by dust owing to the very high gaseous density there . accordingly , this result implies that an ulirg formed by major merging can show negative color gradients . as is shown in figure 12 , @xmath137 is larger in the central region of the merger than in the outer part , which indicates that dust absorption and its reemission is larger in the central region . this result is consistent with those described in figure 11 . figure 13 describes the time evolution of absolute magnitude from optical to near - infrared wavelength ( @xmath138 , @xmath139 , and @xmath140 ) for the standard model with and without dust extinction . as galaxy merging proceeds , @xmath138 , @xmath139 , and @xmath140 gradually rise up owing to the increase of star formation rate ( 1.1 @xmath124 @xmath123 @xmath124 1.3 gyr ) . these @xmath138 , @xmath139 , and @xmath140 then become @xmath141 , @xmath142 , and @xmath143 mag , respectively , at the epoch of maximum star formation rate ( @xmath123 = 1.3 gyr ) in the model with dust extinction . the difference in magnitude between the models with and without dust extinction at the epoch of maximum starburst is the largest for @xmath144 band ( 2.5 for @xmath144 , 2.1 for @xmath28 and 1.3 for @xmath135 band ) , which reflects the fact that stellar light with shorter wavelength can be more greatly absorbed by dust . the mean value of @xmath140 in ulirgs ( @xmath145 mag ; sanders & mirabel 1996 ) is about 1 mag brighter ( smaller in number ) than the @xmath140 of the present merger at @xmath123 = 1.3 gyr , which probably means that initial total mass of the host disk of a typical ulirg is about 2 times larger than that of the galaxy adopted as a merger progenitor disk in the present study . furthermore , irrespectively of wavelength , the difference in absolute magnitude between the two models in the preset study is the largest at the epoch of maximum starburst . this is essentially because young and very luminous stellar components formed by starburst are the most heavily obscured by dust at the maximum starburst epoch when a larger amount of dusty gas is transfered to form very high density region around the compact starburst . as the star formation rate rapidly declines , @xmath138 , @xmath139 , and @xmath140 also rapidly decline ( 1.3 @xmath124 @xmath123 @xmath124 2.8 gyr ) because of the death of young and luminous stellar components ( ob stars formed in starburst ) and the aging of stellar populations . the present results thus suggest that time variation of absolute magnitude during galaxy merging becomes moderate owing to dust extinction . figure 14 shows the time evolution of @xmath146 , @xmath147 , and @xmath136 colors in the standard model with and without dust extinction . although the colors in the model without extinction become very blue at the epoch of maximum starburst , these colors are changed into redder ones for a starburst galaxy because of the heavy dust extinction in the merger . the color difference between the models with and without dust extinction is the most remarkable at the epoch of maximum starburst . figure 15 describes the time evolution of infrared flux at 60 and 100 @xmath148 m and submillimeter one at 450 and 850 @xmath148 m for the model with dust extinction . clearly these fluxes resulting mainly from dust reemission become maximum when the magnitude of starburst is the largest and the density of dusty gas in the central region of the merger is also considerably high . furthermore these fluxes become an order of magnitude smaller than the maximum values within @xmath0 1 gyr after the maximum starburst . this result implies that the time scale during which a dusty starburst merger can be identified as a submillimeter source detected recently by the scuba is very short ( an order of @xmath24 year ) . as is shown in figure 16 , the dust temperature of the model becomes maximum ( @xmath0 32 k ) when the star formation rate becomes maximum and total amount of radiation from young massive stars formed by starburst becomes the largest . we here note that the dust temperature is the mean value for all gas particle for each of which the dust temperature is calculated . figure 16 implies that the dust temperature in mergers with dusty starburst depends strongly on the strength of starburst . figure 17 , 18 , and 19 describe how the difference in the epoch of galaxy merging affects the global colors ( @xmath146 , @xmath147 , and @xmath136 , respectively ) in observed frame for the standard model . for clarity , we give the dependence of colors on the merging epoch z only for the three different phases of the merger , the pre - starburst epoch ( @xmath123 = 1.1 gyr ) , the maximum starburst one ( 1.3 ) , and the post - starburst one ( 1.7 ) . as is shown in figure 17 , optical color @xmath146 in observed frame is the reddest at the epoch of post - starburst ( @xmath123 = 1.7 gyr ) at each redshift ( i.e. , at the epoch of galaxy merging ) . furthermore , the @xmath146 color difference between different redshifts is about 0.7 mag for the epoch of maximum starburst when infrared flux becomes the largest . as has been shown in bekki et al ( 1999 ) , the submillimeter flux at 850 @xmath148 m at z @xmath149 1.0 for the maximum starburst phase ( @xmath123 = 1.3 gyr ) is a few mjy that is well above the current detection limit of the scuba ( @xmath0 2 mjy ) . these results thus imply that the observed color difference in @xmath146 for the faint scuba sources ( smail et al . 1998 ) is due partly to the difference in redshifts between dusty starburst galaxies detected by the scuba . irrespectively of redshifts , near - infrared colors @xmath147 and @xmath136 are the reddest for the post - starburst phase ( figure 18 and 19 ) . furthermore these colors are redder in higher z for the post - starburst phase ( @xmath123 = 1.7 gyr ) because of the k - correction of stellar populations in this phase . it should be noted here that the @xmath136 color for the post - starburst phase becomes larger than @xmath0 6.0 around z = 1.5 . these results imply that an ero with the @xmath136 color larger than 6.0 is more likely to be formed in the post - starburst phase of a gas - rich major merger with the strong dusty starburst and at higher redshifts ( z @xmath149 1 @xmath0 2 ) . these results on the redshift evolution of global colors are only for one merger model . therefore , we stress that although the above results are useful and helpful for understanding the nature of the scuba sources and eros , the results can be true only for some high z dusty mergers . it is our future work to investigate a much larger number of dusty merger models and thereby clarify the nature and the origin of high z dusty starburst galaxies such as the faint scuba sources and eros . one of advantages of the present new numerical code is that we can investigate both dynamical and photometric evolution of dusty starburst galaxies in an explicitly self - consistent manner . therefore we can address some important questions that classical one - zone models on galaxy evolution have not yet answered so clearly for the origin of dusty starburst galaxies . here , we particularly enumerate three problems which the newly developed code is rather helpful and useful for clarifying and thus will be discussed in our future papers . recent observational studies with the scuba have revealed possible candidates of heavily dust - enshrouded starburst galaxies at intermediate and high redshift , which could be counterparts of low redshift ulirgs ( smail , ivison , & blain 1997 ; barger et al . 1998 ; hughes et al . 1998 ; smail et al . 1998 ; ivison et al . 1999 ; lilly et al . although optical morphology of these submillimeter extra - galactic sources should be treated with caution owing to the absence of high resolution sub - mm imaging capability ( richards 1999 ) , more than 50 % of those are suggested to show the indication of galactic interaction and merging ( smail et al . furthermore observational studies of an extremely red object ero j 164502 - 4626.4 ( hr 10 ) with the redshift of 1.44 by the hubble space telescope and the scuba have found that this high redshift galaxy is also a dust - enshrouded starburst galaxy with clear indication of galaxy merging / interaction ( graham & dey 1996 ; cimatti et al . 1998 ; dey et al . although low redshift ulirgs are generally considered to be ongoing galaxy mergers with the triggered prominent nuclear activities ( starburst or agn ) heavily obscured by dust ( sanders et al . 1988 ; sanders & mirabel 1996 ) , the origin of the faint scuba sources is not so clearly understood . in particular , it is not clear whether the observed very large submillimeter luminosity , which could result either from active galactic nuclei ( agn ) obscured by dust or from dusty starburst , is due essentially to physical processes associated closely with major galaxy merging . concerning this problem , one of tests to assess the validity of merger scenario of the faint scuba source formation is to compare the observed morphological , structural , and kinematical properties of the sources with those of a certain theoretical merger model @xmath150 @xmath7 @xmath151 @xmath152 @xmath153 @xmath154 when submillimeter flux of the merger can exceed the detection limit of the scuba . therefore , theoretical studies should investigate simultaneously the time evolution of submillimeter flux at 850 @xmath148 m , morphology , structure , and kinematics of a dusty major galaxy merger in order to compare the results with the corresponding observational ones . although classical one - zone models can investigate photometric evolution of dusty starburst galaxies at each wavelength and thus have contributed greatly to the understanding of dusty galaxies ( e.g. , mazzei et al 1992 ) , they can not predict dynamical evolution of galaxies simultaneously . the present model , on the other hand , can predict not only photometric evolution from uv to submillimeter wavelength but also dynamical evolution in a dusty galaxy . for example , as is shown in the present study , we can predict optical and near - infrared morphology of a merger with the 850 @xmath148 m flux larger than 2 mjy ( the current detection limit of the scuba ) at a given redshift . thus we expect that future theoretical studies with our new code will provide valuable clues to the origin of the faint scuba sources . recent observational studies have discovered a significant number of high z eros with @xmath136 @xmath115 @xmath155 ( elston et al . 1988 ; cimatti et al . 1998 ; dey et al . 1999 ; bentez et al . 1998 ; cimatti et al . 1999 ; smail et al . 1999 ; soifer et al . 1999 ) , and accordingly the nature and the origin of eros have now been discussed very extensively both in observational studies and in theoretical ones . thompson et al . ( 1999 ) argued that eros are most likely to lie in the redshift range 1 @xmath124 @xmath1 @xmath124 2 and they represent an important population in high redshift universe . beckwith et al . ( 1998 ) argued that the surface density of eros with @xmath156 mag and @xmath157 mag is 0.14 @xmath158 for eros with @xmath156 and @xmath157 . furthermore thompson et al . ( 1999 ) derived a surface density of eros with @xmath159 mag and @xmath160 of 0.039 @xmath161 0.016 @xmath158 and estimated that the volume density of bright eros to be as high as that of nearby seyfert galaxies . there are mainly two possible interpretations for eros ( we here do not intend to discuss whether some eros are actually low - mass galactic stars such as main - sequence m stars brown dwarfs ) . one is that eros are passively evolving red elliptical galaxies at high @xmath1 ( e.g. , those recently discovered by the vlt ; bentez et al . 1998 ) , and the other is that eros are dusty starburst galaxies with starburst components obscured heavily by dust ( e.g. , hr10 ; dey et al . it still remains unclear which interpretation among the two is more plausible , essentially because spectroscopic studies of eros have not been so accumulated yet which can reveal unambiguously the redshift of eros and thus discriminate the effects of aging of stellar populations and those of dust extinction . recent observational results on eros have begun to ( or will soon begin to ) provide valuable information concerning this problem . for example , smail el al . ( 1999 ) discovered two eros with @xmath162 and 6.8 among their scuba sources samples , which implies that the submillimeter telescopes with the improved detectability ( the detection limit of the order of @xmath163 @xmath148jy ) can detect submillimeter flux indicative of the obscured dusty starburst in a significant number of eros . furthermore statistical spectroscopic studies of eros by the already existing large grand - based telescopes ( e.g. , subaru with ircs and fmos ) will reveal the strength of @xmath164 emission that is not affected so strongly by dust extinction and thereby clarify star formation rate and history of eros . it is also possible that the future improved @xmath119 acs and the 8m - class grand - based telescopes provide the detailed morphology of an ero . accordingly , one of important works for clarifying the origin of eros is to compare emission line strengths , submillimeter flux , and morphological properties observed ( or those that will be observed in near future ) in eros with those predicted by a certain theoretical dusty starburst model . in order to make this comparison possible , theoretical models of dusty starburst galaxies should investigate the time evolution of @xmath136 color , @xmath135 band magnitude , emission lines , submillimeter flux , and morphology for a dusty starburst galaxy @xmath165 @xmath166 @xmath167-@xmath168 @xmath169 . although the present code still can not predict emission and absorption line strengths of a dusty starburst galaxy , our new code enables us to investigate most of the above important properties simultaneously in numerical simulations . thus it is one of important issues for future numerical and theoretical studies with the present new code to clarify the origin of eros . we present a new code which is designed to derive a sed for an arbitrary spatial distribution of stellar and gaseous components for a dusty starburst galaxy . by using this code , we can calculate seds based on numerical simulations that can analyze simultaneously dynamical and chemical evolution , structural and kinematical properties , morphology , star formation history , and transfer of metals and dust in interstellar medium for a starburst galaxy . accordingly , we can investigate variously different properties of starburst galaxies , such as effects of dynamical evolution on galactic seds , physical correlations between morphology and seds , photometric evolution from uv to submillimeter wavelength , two dimensional distribution of @xmath118 , and dependence of seds on line - of - site of an observer . thanks to this code , we can furthermore try to clarify the origin of possible candidates of starburst galaxies , such as low @xmath1 ulirgs ( e.g. , sanders et al . 1988 ; sanders & mirabel 1996 ) , intermediate z ones ( e.g. , tran et al . 1999 ) , faint scuba sources ( smail , ivison , & blain 1997 ; barger et al . 1998 ; hughes et al . 1998 ; smail et al . 1998 ; ivison et al . 1999 ; lilly et al . 1999 ) , eros ( elston , rieke , & rieke 1988 ; dey et al . 1999 ; smail et al . 1999 ) , optically faint radii sources detected by vla ( e.g. , richards et al . 1999 ) , lyman - break galaxies ( steidel et al . 1996 ; lowenthal et al . 1997 ) , and emission line galaxies ( e.g. , mannucci et al . 1998 ) . by using a new code developed in the present study , we try to answer the following seven questions in our forthcoming papers : ( 1 ) when and how a gas - rich major galaxy merger becomes an ulirg during the dynamical evolution of the merger ? ( 2 ) how high z faint scuba sources with dusty starburst form and evolve ? ( 3 ) what are physical conditions for high z dusty galaxies to become eros ? ( 4 ) are there any evolutionary links between high z possible dusty starburst galaxies , such as faint scuba sources , eros , optically faint radio sources recently detected by vla , emission line galaxies , and lyman - break ones ? ( 5 ) when does a forming disk galaxy show the strongest submillimeter flux ? ( 6 ) what physical processes can determine the shapes of seds observed in lyman - break galaxies ? ( 7 ) how dusty interstellar gas affects apparent morphology of intermediate and high @xmath1 dusty galaxies ? although this code has some disadvantages in deriving very precise seds of dusty starburst galaxies , we believe that this code enables us to grasp some essential ingredients of physical processes related to galaxy formation with starburst at low and high z universe . we are grateful for the referee g. d. bothun for giving us valuable comments and suggestions that greatly improve our paper . y.s . thank to the japan society for promotion of science ( jsps ) research fellowships for young scientist .

we calculate theoretical isochrones , in a consistent way , for five filters in the atmospheric window between and , , , , f205w , and f222 m , using the padova stellar evolutionary models by girardi et al . even when displayed in the same vega magnitude system , the near - infrared colors of the same isochrone can differ by up to 0.18 mag at its bright end , depending on the filter . we present magnitude transformations between-band filters as a function of color from & band filters . isochrones with extinction at of up to 6 mag are also presented . we find that care is needed when comparing extinction values that are estimated using different filter sets in the-band , in particular when comparing those between atmospheric and space filter sets : extinction values for space filters can be in error by up to 0.3 mag . to reduce this error , we introduce an `` effective extinction slope '' for each filter set and isochrone model , which describes the extinction behaviour of isochrones in the color - magnitude diagram more correctly than the actual extinction law . our calculation also suggests that the extinction law implied by the observations of rieke , rieke , & paul for wavelengths between and bands is better described by a power - law function with an exponent of 1.61 , instead of 1.55 , which is commonly used with an assumption that the transmission functions of and filters are dirac delta functions .

astro-ph0505089

deep near - infrared stellar photometry is an important tool for studies of the structure and stellar population in the galactic bulge and the disk , where the extinction by dust is quite significant . in the near - infrared ( @xmath6 m ) , deep ground - based imaging is generally limited to the three shorter atmospheric windows , @xmath7 ( @xmath8 m ) , @xmath5 ( @xmath9 m ) , and @xmath2 ( @xmath10 m ) . longward of @xmath11 m , thermal emission from a warm telescope , instrument , and sky ( for ground - based observations ) makes photometry highly inefficient . the @xmath2-band is the most popular among these three bands , because it suffers from the least extinction and provides the widest wavelength baseline when combined with visible bands . contamination by the thermal emission in the @xmath2 filter ( johnson et al . 1966 ) is not negligible , so astronomers have introduced several filters with slightly shorter long wavelength cutoffs , i.e. @xmath3 ( wainscoat & cowie 1992 ) and @xmath4 ( @xmath2-short ; developed by m. skrutskie see the appendix of persson et al . 1998 ) , that effectively reduce the total background from a warm sky and telescope by a factor of two . on the other hand , the near - infrared camera on board _ hubble space telescope _ ( _ hst _ ) , nicmos , has a unique filter system , as it is not subject to the effects of atmospheric absorption . the nicmos f160w filter and the @xmath5 filter ( johnson et al . 1966 ) have similar central wavelengths , and the nicmos f205w and f222 m filters transmit wavelengths that fall within the passband of the johnson @xmath2 filter . as enumerated above , there are at least five different broad- and medium - band filters in common usage that cover wavelengths near @xmath12 : three atmospheric filters @xmath2 , @xmath3 , & @xmath4 , and two space filters f205w & f222 m . transmission functions for these five filters are quite different from each other ( see figure [ fig : filters ] ) , but there have been no studies that consistently compare the magnitudes and colors of stars as observed with these five filters as a function of interstellar extinction . there have been a few studies that compare theoretical isochrones or colors for _ some _ of the five filters . girardi et al . ( 2002 ) presented theoretical isochrones for @xmath2 , @xmath4 , and f205w filters among others , but not for @xmath3 and f222 m . origlia & leitherer ( 2000 ) provided color transformations between @xmath13 & f110w@xmath14f160w , and between @xmath15 & f110w@xmath14f222 m , and stephens et al . ( 2000 ) presented transformation between @xmath2 magnitude and f110w@xmath14f205w color , but none for @xmath3 , @xmath4 , or f205w . wainscoat & cowie ( 1992 ) investigated the transformation between the @xmath3 and @xmath2 filters . in the present paper , we present and analyze theoretical isochrones that are calculated in a consistent way for the five filters near the @xmath16 atmospheric window over a range of extinctions . the magnitude system we adopt here is explained in [ sec : magsys ] , and the stellar spectral library and evolutionary tracks we use are discussed in [ sec : spec_track ] . theoretical isochrones are presented in [ sec : isochrones ] , and our findings are then summarized in [ sec : summary ] . here , we briefly discuss our adopted magnitude system , following the discussion by girardi et al . ( 2002 ) . the flux of a star measured at the top of the earth s atmosphere , @xmath17 , is related to the flux at the stellar photosphere , @xmath18 , by @xmath19 where @xmath20 is the extinction in magnitudes at the wavelength @xmath21 , @xmath22 is the stellar radius , and @xmath23 is the distance to the star . the apparent magnitude , @xmath24 , as measured using a filter with transmission function @xmath25 is given by @xmath26 where @xmath27 is the reference spectrum corresponding to a known apparent magnitude @xmath28 . the integrands of equation ( [ mag ] ) correspond to the total photon energy , which suitably represents data obtained with traditional energy - collecting devices , such as bolometers . however , data from modern photon - collecting devices , such as ccds and infrared arrays , would be better represented by an integration over the collected photon flux , @xmath29 the difference between energy integration and photon integration is usually very small when extinction is negligible . in the present paper , we adopt the photon integration ( eq . [ mag_ph ] ) for defining the magnitudes . by combining the equations above , we derive the absolute magnitude , for which @xmath23 is defined to be 10 pc : @xmath30 + m_{s_\lambda}^0.\ ] ] stellar radius @xmath22 can be obtained if the luminosity @xmath31 and the effective surface temperature @xmath32 of the star are given . we adopt a vega - based photometric system ( vegamag system ) , which uses vega ( @xmath33 lyr ) as the calibrating star . following girardi et al . ( 2002 ) , we adopt the synthetic atlas9 model ( kurucz 1993 ) for vega with @xmath34 k , @xmath35 , @xmath36 } = -0.5 $ ] , microturbulent velocity @xmath37 , @xmath38 , and photometric zeropoint of 0.03 . the dilution factor @xmath39 is needed to calculate @xmath40 from the model spectrum @xmath41 . now , equation ( [ mag ] ) can be written as @xmath42 + m_{s_\lambda}^{\rm vega},\ ] ] where @xmath43 and @xmath44 . the nicmos data handbook ( dickinson 2002 ) defines the vega magnitude system for nicmos filters by using a filter - averaged flux density of a synthetic model for vega , @xmath45 , as a photometric zeropoint . therefore , for nicmos filters , the magnitude is defined to be @xmath46 + m_{s_\lambda}^{\rm vega}.\ ] ] we adopt @xmath45 values from the nicmos data handbook ver . 5.0 : 1040.7 jy for f160w , 703.6 jy for f205w , and 610.4 jy for f222 m . we adopt filter transmission functions of bessell & brett ( 1988 ) for @xmath5 and @xmath2 , and those of the quirk camera of the university of hawaii for @xmath3 and @xmath4 . in the case of nicmos filters , transmission functions include both throughputs and detector quantum efficiencies , which come from synphot , a synthetic photometry package distributed as a part of stsdas ( space telescope science data analysis system ) software . when calculating the magnitudes for ground - based filters , we do not consider the atmospheric transmission , telescope throughputs , and detector quantum efficiencies with an assumption that the wavelegnth dependence of these effects is negligible for the wavelength ranges covered by the filters considered here . for the spectra of synthetic stellar atmospheres , we adopt kurucz atlas9 no - overshoot models calculated by castelli et al . the metallicities of these models cover the values of [ m / h ] = @xmath47 to @xmath48 . a microturbulent velocity @xmath49 , and a mixing length parameter @xmath50 are adopted in the present study . for the temporal evolution of @xmath32 and @xmath31 as a function of stellar mass , i.e. stellar evolutionary tracks , we adopt the basic set " of the padova models ( girardi et al . we consider isochrones with a metallicity @xmath51 = 0.0001 , 0.001 , 0.019 , & 0.03 . the stellar spectral library and the evolutionary tracks we adopted assume a solar chemical mixture . atlas9 models cover a wide range of @xmath32 and @xmath52 : @xmath53 , & @xmath54 . however , as shown in figure [ fig : tg ] , some of the @xmath55 and @xmath52 values from the stellar tracks are located outside this range , in particular for small @xmath55 and @xmath52 . for this reason , empirical m giant spectra ( e.g. , bessell et al . 1989 , 1991 ; fluks et al . 1994 ) are often adopted . however , as noted by girardi et al . ( 2002 ) , the magnitudes from the atlas9 models , and those from the supplementary spectra , do not nicely match at their boundaries . we find that the isochrones in this overlap region are sensitive on how one applies the interpolation between the two libraries . the goal of the present paper is simply to present similarities / disimilarities of the isochrones for different filters , not the most complete isochrones , so we choose not to supplement the atals9 models . consequently , our isochrones do not include m giants cooler than @xmath56 k , relatively rare metal - rich stars that are usually in the thermally pulsing agb phase of evolution , and are found at the upper tip of the red giant branch . we similarly choose not to supplement the atlas9 models with synthetic spectra of m , l , & t dwarfs ( low @xmath32 and high @xmath57 ) , as was done for the padova models by girardi et al . ( 2002 ) using the spectra calculated by allard et al . we first calculate a table of magnitudes for all spectra in atlas9 models in the @xmath5 and @xmath2 band filters , covering a large range in @xmath32 , @xmath52 , and [ m / h ] , using equations ( [ mag2 ] ) and ( [ mag2nic ] ) . we use this table as a set of interpolates for the @xmath32 , @xmath52 , and @xmath51 values predicted by the stellar evolution models for a given age in order to estimate synthetic isochrones . isochrones for @xmath58 , calculated in this way , are shown in figures [ fig : iso1]@xmath14[fig : iso4 ] for four different metallicities , and four ages . as expected for color - magnitude diagrams ( cmd ) in the infrared , the isochrones have relatively narrow color spreads . the color differences between filters is more prominent for the highest metallicity isochrones . in most cases , isochrones for @xmath3 and @xmath4 are nearly indistinguishable , and those for f205w and f222 m are quite close to each other . the largest color difference , 0.18 mag , is seen between the bright end of the @xmath2 and f222 m isochrones for @xmath51 = 0.019 and age = @xmath59 yr . in general , intrinsic color differences of red giants between the atmospheric filters and the space filters are @xmath600.1 - 0.15 mag . such color differences at the bright end of red giant isochrones between different filters are due to a kink in the continuum near @xmath61 and the co absorption band longward of @xmath62 . figure [ fig : spec ] compares the spectrum of a k giant with @xmath63 & age = @xmath64 yr ( @xmath65 mag , @xmath66 , @xmath67 ) , whose f160w@xmath14f222 m color is 0.13 mag redder than its @xmath68 color , and that of an f dwarf ( @xmath69 mag , @xmath70 , @xmath71 ) , whose f160w@xmath14f222 m color is only 0.01 mag redder than its @xmath68 color . the spectrum of an f dwarf is very close to that of vega , and has no strong features in @xmath5 and @xmath2 bands , making all colors in our filter sets nearly zero . on the other hand , the spectrum of a k giant 1 ) significantly deviates from that of vega at the wavelengths shortward of @xmath72 due to the h@xmath73 opacity in cool stars ( j. valenti 2005 , private communication ) , which causes f160w magnitudes to be fainter than @xmath5 magnitudes , and 2 ) has the co absorption band longward of @xmath62 , which causes @xmath2 magnitudes to be fainter than f222 m magnitudes . these two effects make f160w@xmath14f222 m redder than @xmath68 , and the same analysis can be applied to other filter sets . as an independent check of our procedure , we compare our @xmath68 vs. @xmath2 isochrones to those calculated by girardi et al . ( 2002 ) in figure [ fig : padova ] . the isochrones match nicely , except at the extremes . the discrepancy in the bright end is caused by the empirical m giant spectra that girardi et al . ( 2002 ) added to their spectral library , and that in the faint end is by the late m dwarf spectra . the discrepancies are considerable only at the top 1 or 2 magnitudes of the isochrone , where only a small number fraction of giants reside . magnitude transformations between @xmath2-band filters can be obtained from our isochrones . figures [ fig : trans1a ] through [ fig : trans1e ] show magnitude differences between @xmath2-band filters as a function of color for all ages considered ( @xmath74 , @xmath75 , @xmath64 , & @xmath59 yr for @xmath76 & 0.019 models , and @xmath77 , @xmath75 , @xmath64 , & @xmath59 yr for @xmath78 & 0.03 models ) , for @xmath2-band magnitudes brighter than 4 mag . we find that the magnitude difference can be well fit by a quadratic function for @xmath79 mag , and by a separate straight line for @xmath80 mag ( we do not present figures for @xmath80 mag ) . the largest residuals from the fit are 0.019 mag for the former and 0.009 mag for the latter . the coefficients of the best - fit functions are presented in tables [ table : trans1 ] and [ table : trans2 ] , along with the residuals and fitting ranges . extinction in the @xmath2-band is much lower than that in visual bands , but there are still cases where the @xmath2-band extinction is significant , such as the central region of the milky way ( @xmath81 ) . here , we calculate isochrones with @xmath2-band extinctions up to 6 mag , some of which are shown in figures [ fig : red1]@xmath14[fig : red5 ] . for @xmath20 in equations ( [ mag2 ] ) & ( [ mag2nic ] ) , we adopt a power - law extinction law for wavelengths between @xmath5 and @xmath2 bands : @xmath82 where we choose @xmath83 m , and @xmath84 is the extinction at @xmath85 . when assuming that the transmission functions of @xmath5 and @xmath2 filters are dirac delta functions centered at 1.65 @xmath86 m and 2.2 @xmath86 m , respectively , the extinction law by rieke , rieke , & paul ( 1989 ) gives @xmath87 . however , as discussed later in this section , the apparent extinction behaviour of isochrones in the cmd can be different from the actual extinction law , due to non - zero width and asymmetry of the filter transmission functions . we find that @xmath88 makes the isochrone for the @xmath89 , age = @xmath64 yr model behave in the cmd as if it follows an extinction law with @xmath87 . we choose this particular isochrone for calibrating the extinction law , with an assumption that the stars used in rieke et al . ( 1989 ) to derive their extinction law , which are the stars in the central parsec of our galaxy , can be represented by such metallicity and age . for the sake of easier comparison , isochrones in figures [ fig : red1]@xmath14[fig : red5 ] are dereddened by an amount of @xmath90 , where the central wavelength of the filter @xmath91 is defined by @xmath92 and is given in table [ table : lambdac ] . as seen in the figures , isochrones with larger extinction values are more vertically straight , in particular for @xmath64 and @xmath59 yr models . this is probably because the extinction makes the longer wavelength part of the spectrum more important , which is closer to the rayleigh - jeans regime of the spectrum . since we have dereddened the isochrones with the known amount of extinction at @xmath91 , all the dereddened isochrones with different extinction values in figures [ fig : red1]@xmath14[fig : red5 ] should be coincident , if the filter transmission functions were dirac delta functions centered at @xmath91 . the figures show that the dereddened isochrones do not align , and the amount of misalignment is rather large for some filters . this implies that the inferred amount of extinction can be sensitively dependent on the shape of the filter transmission function . let us see how this will affect the analyses of actual observations . when one estimates the amount of extinction from an observed near - infrared cmd , where isochrones are nearly vertically straight , one converts an observed color excess to an extinction value following an assumed extinction law , which usually has the form of a power - law . when one has photometric data from a set of two filters , @xmath93 and @xmath94 , the amount of extinction can be estimated by @xmath95 where @xmath96 & @xmath97 and @xmath98 & @xmath99 are the magnitudes and the central wavelengths of the two filters , respectively , and subscript 0 denotes the intrinsic value . for estimating extinction from our isochrones , we first use @xmath87 . figure [ fig : adiff1 ] shows the difference between the inferred extinction values , using equation ( [ a_est ] ) and colors from our reddened isochrones , and the actual extinction values . here , the extinction of each isochrone has been calculated using the mean color ( for @xmath100 ) and magnitude ( for @xmath101 ) of the reddened isochrone data points having @xmath2-band magnitudes between @xmath146 and 0 mag . as the figure shows , the differences between estimated and actual extinction values are larger for the space filter sets in general . the largest relative difference is @xmath102 % , and the largest absolute difference is 0.29 mag . note that the extinction estimates for the @xmath89 and age = @xmath64 yr model inferred from @xmath5 & @xmath2 are very close to the actual extinction values , justifying our choice of @xmath88 for equation ( [ extinction ] ) . the problems seen in figure [ fig : adiff1 ] can be alleviated by finding an `` effective extinction slope '' for each filter set and isochrone model , which is defined such that it better describes the extinction behaviour in the cmd . figure [ fig : extlaw ] shows reddened @xmath2-band magnitudes and colors for the @xmath89 & age = @xmath64 yr isochrone ( the figure only shows an isochrone data point whose intrinsic @xmath2 magnitude is 0 , as an example ) . the reddened magnitudes and colors of each filter set form a nearly straight line ( except for filter set f160w & f222 m ) , thus we may assume that @xmath103 in equation ( [ a_est ] ) is not a function of extinction . then the effective slope of the extinction , @xmath104 , can be calculated by @xmath105 where @xmath106 is the slope of the straight line that fits the distribution of reddened magnitudes vs. reddened colors , as in figure [ fig : extlaw ] . we calculate @xmath106 for data points of each isochrone whose intrinsic @xmath2 magnitudes are between @xmath107 and 0 mag , and take an average for each isochrone model . when finding the best - fit straight line , we forced the line to include the @xmath108 point , i.e. , we fitted @xmath109 , instead of @xmath110 , where subscript 0 denotes the values for @xmath108 . table [ table : alpha_eff ] shows the averages and standard deviations of @xmath104 values for each isochrone model . note that the standard deviations of @xmath104 in an isochrone is generally much smaller than the differences of average @xmath104 values between different isochrones . the average @xmath104 values range from 1.489 to 1.574 , which are 7.5 % to 2.2 % smaller than the original @xmath33 value we adopted for extinction , 1.61 . as seen in figure [ fig : adiff3 ] , extinction values estimated by equation ( [ a_est ] ) with @xmath104 are now very close to the actual values , except for filter set f160w & f222 m , whose extinction behaviour in the cmd is poorly described by a straight line . the nonlinear extinction behavior of the filter set f160w & f222 m appears to be due to a significant width difference in the two filters : f160w covers @xmath111 while f222 m covers only @xmath112 . here we discuss the effect of the width difference in terms of an extinction - weighted central wavelength , @xmath113 as the amount of extinction @xmath84 in equation ( [ extinction ] ) becomes large , @xmath114 shifts to a longer wavelength because @xmath20 decreases with an increasing wavelength . however , the amount of this shift is limited by the filter width . when the widths of two filters differ significantly , the relative shifts in @xmath114 become considerably different for the two filters . since the slope of the reddening vector in the cmd is determined by a ratio between @xmath114 s of the two filters ( see eq . [ a_est ] ) , the slope of the reddening vector will become a function of @xmath84 when the ratio between @xmath114 s is sensitively dependent on @xmath84 , i.e. when the widths of two filters differ significantly . figure [ fig : nonlin ] shows the effect of the filter width by comparing the extinction behaviour of five imaginary filter sets . filter set @xmath115 represents f160w & f222 m , and its extinction behaviour in the cmd is not linear , as in the real f160w & f222 m set . but this nonlinearity becomes much less significant when the two filters are both wide ( filter set @xmath106 ) or both narrow ( filter set @xmath116 ) . filter sets @xmath23 and @xmath117 show that the nonlinearity becomes more prominent when the two filters with significantly different widths are closer to each other . this is because the dependence of @xmath114 ratio on @xmath84 is larger when the two filters are closer to each other , and this supports our interpretation of the nonlinear reddening vector of the filter set f160w & f222 m . now we discuss how one may transform an extinction value estimated for one filter in the @xmath2-band to the extinction value for another filter . when the empirical extinction law from observations is not avaialable for the two filters , transformation of extinction values between filters @xmath94 and @xmath118 is normally obtained by @xmath119 assuming that the extinction has a functional form of a power - law , and the filter transmission functions are dirac delta functions centered at their central wavelengths . we find that once @xmath101 is estimated with @xmath104 , equation ( [ a_trans ] ) with the original @xmath33 value , 1.61 , gives good estimates for @xmath120 . figure [ fig : adiff4 ] shows the difference between the extinction values that are estimated by equation ( [ a_est ] ) with our @xmath121 , @xmath122 , f160w@xmath14f205w , & f160w@xmath14f222 m colors and then converted to an extinction value at @xmath2 by equation ( [ a_trans ] ) with @xmath88 , and the actual extinction values for @xmath2 . except for filter set f160w@xmath14f222 m again , extinction values can be reliably transformed between different filter sets following the above procedure . it is interesting to note that the original @xmath33 value of 1.61 is not suitable for estimating the extinction , but can be used for transformation between appropriately estimated extinction values . we have calculated in a consistent way five near - infrared theoretical isochrones for filter sets @xmath68 , @xmath121 , @xmath122 , f160w@xmath14f205w , and f160w@xmath14f222 m . we presented isochrones for @xmath51 of 0.0001 to 0.03 , and age of @xmath74 to @xmath59 yr . even in the same vega magnitude system , near - infrared colors of the same isochrone can be different by up to 0.18 mag at the bright end of the isochrone for different filter sets . the difference of intrinsic colors for a red giant between atmospheric filters and the space filters is generally @xmath600.1 - 0.15 mag . we provided magnitude transformations between @xmath2-band filters as a function of color from @xmath5 & @xmath2 band filters . we also presented isochrones with @xmath123 of up to 6 mag . isochrones for larger extinction values are found to be more vertically straight . we found that care is needed when comparing extinction values that are estimated using different filter sets , in particular when comparing those between atmospheric and space filter sets : extinction values inferred using space filters can be in error by up to 0.3 mag . to alleviate this problem , we introduced an `` effective extinction slope '' for each filter set and isochrone model , which describes the extinction - dependent behaviour of isochrones in the observed cmd . thanks hwankyung sung for helpful discussion . we thank jeff valenti for identifying h@xmath73 as a source of opacity in cool stars . we thank the anonymous referee whose suggestions and comments significantly improved the content of the present paper . this work was supported by the research fund from kyung hee university . s.s.k . was supported by the astrophysical research center for the structure and evolution of the cosmos ( arcsec ) of korea science and engineering foundation through the science research center ( src ) program . m.g.l . was in part supported by the abrl ( r14 - 2002 - 058 - 01000 - 0 ) and the bk21 program . allard , f. , haushildt , p. h. , alexander , d. r. , tamanai , a. , & ferguson , j. w. 2000 , in proc . of from gian planets to cool stars , eds . c. a. griffith , & m. s. marley , asp conf . ser . , 212 , 127 bessell , m. s. , & brett , j. m. 1988 , pasp , 100 , 1134 bessell , m. s. , brett , j. m. , scholz , m. , wood , p. r. 1989 , a&as , 77 , 1 bessell , m. s. , brett , j. m. , scholz , m. , wood , p. r. 1991 , a&as , 89 , 335 castelli , f. , gratton , r. g. , & kurucz , r. l. 1997 , a&a , 318 , 841 dickinson , m. , ed . 2002 , nicmos data handbook ( ver . 5.0 ; baltimore : stsci ) fluks , m. a. , plez , b. , the , p. s. , de winter , d. , westerlund , b. e. , & steenman , h. c. 1994 , a&as , 105 , 311 girardi , l. , bertelli , g. , bressan , a. , chiosi , c. , groenewegen , m. a. t. , marigo , p. , salsasnich , b. , & weiss , a. 2002 , a&a , 391 , 195 johnson , h. l. , iriarte , b. , mitchell , r. i. , & wisniewskj , w. z. 1966 , comm . lunar plan . lab . , 4 , 99 kurucz , r. l. 1993 , in the stellar populations of galaxies , eds . b. barbuy , & a. renzini , ( dordrecht , kluwer ) , iau symp . , 149 , 225 origlia , l. , & leitherer . c. 2000 , aj , 119 , 2018 persson , s. e. , murphy , d. c. , krzeminski , w. , roth , m. , & rieke , m. j. 1998 , aj , 116 , 2475 rieke , g. h. , rieke , m. j. , & paul , a. e. 1989 , apj , 336 , 752 stephens , a. w. , grogel , j. a. , ortolani , s. , davies , r. , jablonka , p. , renzini , a. , & rich , r. m. 2000 , aj , 119 , 419 wainscoat , r. j. , & cowie , l. l. 1992 , aj , 103 , 332 cclrrrcc @xmath124 & @xmath3 @xmath14@xmath2 & 0.0001 & @xmath125 & @xmath126 & @xmath127 & @xmath128 & @xmath129 + @xmath124 & @xmath3 @xmath14@xmath2 & 0.001 & @xmath130 & @xmath131 & @xmath132 & @xmath133 & @xmath134 + @xmath124 & @xmath3 @xmath14@xmath2 & 0.019 & @xmath135 & @xmath136 & @xmath137 & @xmath138 & @xmath139 + @xmath124 & @xmath3 @xmath14@xmath2 & 0.03 & @xmath140 & @xmath141 & @xmath142 & @xmath143 & @xmath144 + @xmath124 & @xmath4@xmath14@xmath2 & 0.0001 & @xmath145 & @xmath146 & @xmath147 & @xmath128 & @xmath129 + @xmath124 & @xmath4@xmath14@xmath2 & 0.001 & @xmath130 & @xmath148 & @xmath149 & @xmath150 & @xmath134 + @xmath124 & @xmath4@xmath14@xmath2 & 0.019 & @xmath135 & @xmath151 & @xmath152 & @xmath153 & @xmath139 + @xmath124 & @xmath4@xmath14@xmath2 & 0.03 & @xmath154 & @xmath155 & @xmath156 & @xmath157 & @xmath144 + @xmath124 & f205w@xmath14@xmath2 & 0.0001 & @xmath158 & @xmath159 & @xmath160 & @xmath161 & @xmath129 + @xmath124 & f205w@xmath14@xmath2 & 0.001 & @xmath162 & @xmath163 & @xmath164 & @xmath165 & @xmath134 + @xmath124 & f205w@xmath14@xmath2 & 0.019 & @xmath166 & @xmath167 & @xmath168 & @xmath169 & @xmath139 + @xmath124 & f205w@xmath14@xmath2 & 0.03 & @xmath158 & @xmath170 & @xmath171 & @xmath172 & @xmath144 + @xmath124 & f222m@xmath14@xmath2 & 0.0001 & @xmath173 & @xmath174 & @xmath175 & @xmath176 & @xmath129 + @xmath124 & f222m@xmath14@xmath2 & 0.001 & @xmath177 & @xmath178 & @xmath179 & @xmath150 & @xmath134 + @xmath124 & f222m@xmath14@xmath2 & 0.019 & @xmath180 & @xmath181 & @xmath182 & @xmath183 & @xmath139 + @xmath124 & f222m@xmath14@xmath2 & 0.03 & @xmath184 & @xmath185 & @xmath186 & @xmath143 & @xmath144 + @xmath187 & @xmath2 @xmath14@xmath3 & 0.0001 & @xmath145 & @xmath188 & @xmath189 & @xmath128 & @xmath190 + @xmath187 & @xmath2 @xmath14@xmath3 & 0.001 & @xmath140 & @xmath191 & @xmath192 & @xmath165 & @xmath193 + @xmath187 & @xmath2 @xmath14@xmath3 & 0.019 & @xmath194 & @xmath195 & @xmath196 & @xmath197 & @xmath198 + @xmath187 & @xmath2 @xmath14@xmath3 & 0.03 & @xmath194 & @xmath199 & @xmath200 & @xmath201 & @xmath202 + @xmath187 & @xmath4@xmath14@xmath3 & 0.0001 & @xmath203 & @xmath204 & @xmath205 & @xmath128 & @xmath190 + @xmath187 & @xmath4@xmath14@xmath3 & 0.001 & @xmath125 & @xmath206 & @xmath207 & @xmath208 & @xmath193 + @xmath187 & @xmath4@xmath14@xmath3 & 0.019 & @xmath145 & @xmath180 & @xmath209 & @xmath128 & @xmath198 + @xmath187 & @xmath4@xmath14@xmath3 & 0.03 & @xmath145 & @xmath158 & @xmath210 & @xmath128 & @xmath202 + @xmath187 & f205w@xmath14@xmath3 & 0.0001 & @xmath158 & @xmath211 & @xmath212 & @xmath128 & @xmath190 + @xmath187 & f205w@xmath14@xmath3 & 0.001 & @xmath158 & @xmath213 & @xmath214 & @xmath128 & @xmath193 + @xmath187 & f205w@xmath14@xmath3 & 0.019 & @xmath177 & @xmath215 & @xmath216 & @xmath161 & @xmath198 + @xmath187 & f205w@xmath14@xmath3 & 0.03 & @xmath177 & @xmath217 & @xmath173 & @xmath161 & @xmath202 + @xmath187 & f222m@xmath14@xmath3 & 0.0001 & @xmath173 & @xmath218 & @xmath181 & @xmath176 & @xmath190 + @xmath187 & f222m@xmath14@xmath3 & 0.001 & @xmath219 & @xmath220 & @xmath221 & @xmath176 & @xmath193 + @xmath187 & f222m@xmath14@xmath3 & 0.019 & @xmath222 & @xmath223 & @xmath224 & @xmath176 & @xmath198 + @xmath187 & f222m@xmath14@xmath3 & 0.03 & @xmath219 & @xmath225 & @xmath226 & @xmath176 & @xmath202 + @xmath227 & @xmath2 @xmath14@xmath4 & 0.0001 & @xmath125 & @xmath228 & @xmath229 & @xmath128 & @xmath230 + @xmath227 & @xmath2 @xmath14@xmath4 & 0.001 & @xmath140 & @xmath125 & @xmath231 & @xmath176 & @xmath232 + @xmath227 & @xmath2 @xmath14@xmath4 & 0.019 & @xmath203 & @xmath233 & @xmath234 & @xmath169 & @xmath235 + @xmath227 & @xmath2 @xmath14@xmath4 & 0.03 & @xmath194 & @xmath236 & @xmath237 & @xmath133 & @xmath238 + @xmath227 & @xmath3 @xmath14@xmath4 & 0.0001 & @xmath135 & @xmath239 & @xmath240 & @xmath208 & @xmath230 + @xmath227 & @xmath3 @xmath14@xmath4 & 0.001 & @xmath145 & @xmath241 & @xmath242 & @xmath208 & @xmath232 + @xmath227 & @xmath3 @xmath14@xmath4 & 0.019 & @xmath125 & @xmath243 & @xmath244 & @xmath128 & @xmath235 + @xmath227 & @xmath3 @xmath14@xmath4 & 0.03 & @xmath125 & @xmath245 & @xmath246 & @xmath128 & @xmath238 + @xmath227 & f205w@xmath14@xmath4 & 0.0001 & @xmath158 & @xmath247 & @xmath248 & @xmath176 & @xmath230 + @xmath227 & f205w@xmath14@xmath4 & 0.001 & @xmath158 & @xmath249 & @xmath250 & @xmath128 & @xmath232 + @xmath227 & f205w@xmath14@xmath4 & 0.019 & @xmath166 & @xmath251 & @xmath252 & @xmath128 & @xmath235 + @xmath227 & f205w@xmath14@xmath4 & 0.03 & @xmath166 & @xmath253 & @xmath254 & @xmath128 & @xmath238 + @xmath227 & f222m@xmath14@xmath4 & 0.0001 & @xmath219 & @xmath255 & @xmath256 & @xmath165 & @xmath230 + @xmath227 & f222m@xmath14@xmath4 & 0.001 & @xmath219 & @xmath257 & @xmath258 & @xmath165 & @xmath232 + @xmath227 & f222m@xmath14@xmath4 & 0.019 & @xmath222 & @xmath136 & @xmath148 & @xmath259 & @xmath235 + @xmath227 & f222m@xmath14@xmath4 & 0.03 & @xmath222 & @xmath260 & @xmath261 & @xmath259 & @xmath238 + f160w@xmath14f205w & @xmath2 @xmath14f205w & 0.0001 & @xmath262 & @xmath263 & @xmath264 & @xmath259 & @xmath265 + f160w@xmath14f205w & @xmath2 @xmath14f205w & 0.001 & @xmath266 & @xmath267 & @xmath268 & @xmath161 & @xmath269 + f160w@xmath14f205w & @xmath2 @xmath14f205w & 0.019 & @xmath262 & @xmath255 & @xmath270 & @xmath157 & @xmath271 + f160w@xmath14f205w & @xmath2 @xmath14f205w & 0.03 & @xmath262 & @xmath272 & @xmath273 & @xmath133 & @xmath274 + f160w@xmath14f205w & @xmath3 @xmath14f205w & 0.0001 & @xmath245 & @xmath275 & @xmath276 & @xmath161 & @xmath265 + f160w@xmath14f205w & @xmath3 @xmath14f205w & 0.001 & @xmath245 & @xmath277 & @xmath278 & @xmath161 & @xmath269 + f160w@xmath14f205w & @xmath3 @xmath14f205w & 0.019 & @xmath245 & @xmath279 & @xmath131 & @xmath176 & @xmath271 + f160w@xmath14f205w & @xmath3 @xmath14f205w & 0.03 & @xmath266 & @xmath280 & @xmath281 & @xmath161 & @xmath274 + f160w@xmath14f205w & @xmath4@xmath14f205w & 0.0001 & @xmath262 & @xmath282 & @xmath283 & @xmath259 & @xmath265 + f160w@xmath14f205w & @xmath4@xmath14f205w & 0.001 & @xmath262 & @xmath280 & @xmath284 & @xmath176 & @xmath269 + f160w@xmath14f205w & @xmath4@xmath14f205w & 0.019 & @xmath245 & @xmath155 & @xmath285 & @xmath161 & @xmath271 + f160w@xmath14f205w & @xmath4@xmath14f205w & 0.03 & @xmath245 & @xmath155 & @xmath286 & @xmath176 & @xmath274 + f160w@xmath14f205w & f222m@xmath14f205w & 0.0001 & @xmath135 & @xmath287 & @xmath288 & @xmath176 & @xmath265 + f160w@xmath14f205w & f222m@xmath14f205w & 0.001 & @xmath135 & @xmath289 & @xmath211 & @xmath208 & @xmath269 + f160w@xmath14f205w & f222m@xmath14f205w & 0.019 & @xmath140 & @xmath290 & @xmath291 & @xmath176 & @xmath271 + f160w@xmath14f205w & f222m@xmath14f205w & 0.03 & @xmath154 & @xmath292 & @xmath293 & @xmath128 & @xmath274 + f160w@xmath14f222 m & @xmath2 @xmath14f222 m & 0.0001 & @xmath294 & @xmath295 & @xmath296 & @xmath176 & @xmath297 + f160w@xmath14f222 m & @xmath2 @xmath14f222 m & 0.001 & @xmath245 & @xmath298 & @xmath299 & @xmath176 & @xmath300 + f160w@xmath14f222 m & @xmath2 @xmath14f222 m & 0.019 & @xmath243 & @xmath301 & @xmath302 & @xmath201 & @xmath303 + f160w@xmath14f222 m & @xmath2 @xmath14f222 m & 0.03 & @xmath243 & @xmath304 & @xmath283 & @xmath172 & @xmath305 + f160w@xmath14f222 m & @xmath3 @xmath14f222 m & 0.0001 & @xmath262 & @xmath126 & @xmath306 & @xmath128 & @xmath297 + f160w@xmath14f222 m & @xmath3 @xmath14f222 m & 0.001 & @xmath294 & @xmath307 & @xmath308 & @xmath128 & @xmath300 + f160w@xmath14f222 m & @xmath3 @xmath14f222 m & 0.019 & @xmath294 & @xmath233 & @xmath309 & @xmath128 & @xmath303 + f160w@xmath14f222 m & @xmath3 @xmath14f222 m & 0.03 & @xmath294 & @xmath211 & @xmath160 & @xmath128 & @xmath305 + f160w@xmath14f222 m & @xmath4@xmath14f222 m & 0.0001 & @xmath294 & @xmath310 & @xmath311 & @xmath176 & @xmath297 + f160w@xmath14f222 m & @xmath4@xmath14f222 m & 0.001 & @xmath294 & @xmath312 & @xmath313 & @xmath176 & @xmath300 + f160w@xmath14f222 m & @xmath4@xmath14f222 m & 0.019 & @xmath243 & @xmath314 & @xmath219 & @xmath176 & @xmath303 + f160w@xmath14f222 m & @xmath4@xmath14f222 m & 0.03 & @xmath294 & @xmath315 & @xmath316 & @xmath176 & @xmath305 + f160w@xmath14f222 m & f205w@xmath14f222 m & 0.0001 & @xmath145 & @xmath317 & @xmath140 & @xmath161 & @xmath297 + f160w@xmath14f222 m & f205w@xmath14f222 m & 0.001 & @xmath203 & @xmath317 & @xmath318 & @xmath208 & @xmath300 + f160w@xmath14f222 m & f205w@xmath14f222 m & 0.019 & @xmath194 & @xmath285 & @xmath319 & @xmath176 & @xmath303 + f160w@xmath14f222 m & f205w@xmath14f222 m & 0.03 & @xmath194 & @xmath320 & @xmath321 & @xmath128 & @xmath305 + cclrrcc @xmath124 & @xmath3 @xmath14@xmath2 & 0.0001 & @xmath203 & @xmath322 & @xmath323 & @xmath324 + @xmath124 & @xmath3 @xmath14@xmath2 & 0.001 & @xmath135 & @xmath325 & @xmath128 & @xmath326 + @xmath124 & @xmath3 @xmath14@xmath2 & 0.019 & @xmath327 & @xmath328 & @xmath161 & @xmath329 + @xmath124 & @xmath3 @xmath14@xmath2 & 0.03 & @xmath330 & @xmath331 & @xmath176 & @xmath332 + @xmath124 & @xmath4@xmath14@xmath2 & 0.0001 & @xmath194 & @xmath216 & @xmath323 & @xmath324 + @xmath124 & @xmath4@xmath14@xmath2 & 0.001 & @xmath203 & @xmath245 & @xmath208 & @xmath326 + @xmath124 & @xmath4@xmath14@xmath2 & 0.019 & @xmath333 & @xmath167 & @xmath128 & @xmath329 + @xmath124 & @xmath4@xmath14@xmath2 & 0.03 & @xmath334 & @xmath167 & @xmath161 & @xmath332 + @xmath124 & f205w@xmath14@xmath2 & 0.0001 & @xmath177 & @xmath335 & @xmath323 & @xmath324 + @xmath124 & f205w@xmath14@xmath2 & 0.001 & @xmath336 & @xmath293 & @xmath128 & @xmath326 + @xmath124 & f205w@xmath14@xmath2 & 0.019 & @xmath228 & @xmath337 & @xmath161 & @xmath329 + @xmath124 & f205w@xmath14@xmath2 & 0.03 & @xmath174 & @xmath338 & @xmath161 & @xmath332 + @xmath124 & f222m@xmath14@xmath2 & 0.0001 & @xmath219 & @xmath339 & @xmath323 & @xmath324 + @xmath124 & f222m@xmath14@xmath2 & 0.001 & @xmath222 & @xmath340 & @xmath128 & @xmath326 + @xmath124 & f222m@xmath14@xmath2 & 0.019 & @xmath341 & @xmath342 & @xmath128 & @xmath329 + @xmath124 & f222m@xmath14@xmath2 & 0.03 & @xmath206 & @xmath343 & @xmath128 & @xmath332 + @xmath187 & @xmath2 @xmath14@xmath3 & 0.0001 & @xmath135 & @xmath282 & @xmath323 & @xmath344 + @xmath187 & @xmath2 @xmath14@xmath3 & 0.001 & @xmath203 & @xmath250 & @xmath128 & @xmath345 + @xmath187 & @xmath2 @xmath14@xmath3 & 0.019 & @xmath346 & @xmath347 & @xmath176 & @xmath348 + @xmath187 & @xmath2 @xmath14@xmath3 & 0.03 & @xmath295 & @xmath349 & @xmath161 & @xmath350 + @xmath187 & @xmath4@xmath14@xmath3 & 0.0001 & @xmath145 & @xmath340 & @xmath323 & @xmath344 + @xmath187 & @xmath4@xmath14@xmath3 & 0.001 & @xmath194 & @xmath147 & @xmath208 & @xmath345 + @xmath187 & @xmath4@xmath14@xmath3 & 0.019 & @xmath130 & @xmath351 & @xmath323 & @xmath348 + @xmath187 & @xmath4@xmath14@xmath3 & 0.03 & @xmath352 & @xmath257 & @xmath323 & @xmath350 + @xmath187 & f205w@xmath14@xmath3 & 0.0001 & @xmath166 & @xmath301 & @xmath323 & @xmath344 + @xmath187 & f205w@xmath14@xmath3 & 0.001 & @xmath158 & @xmath213 & @xmath323 & @xmath345 + @xmath187 & f205w@xmath14@xmath3 & 0.019 & @xmath166 & @xmath353 & @xmath323 & @xmath348 + @xmath187 & f205w@xmath14@xmath3 & 0.03 & @xmath177 & @xmath322 & @xmath323 & @xmath350 + @xmath187 & f222m@xmath14@xmath3 & 0.0001 & @xmath222 & @xmath354 & @xmath323 & @xmath344 + @xmath187 & f222m@xmath14@xmath3 & 0.001 & @xmath219 & @xmath355 & @xmath208 & @xmath345 + @xmath187 & f222m@xmath14@xmath3 & 0.019 & @xmath166 & @xmath356 & @xmath208 & @xmath348 + @xmath187 & f222m@xmath14@xmath3 & 0.03 & @xmath166 & @xmath357 & @xmath208 & @xmath350 + @xmath227 & @xmath2 @xmath14@xmath4 & 0.0001 & @xmath154 & @xmath358 & @xmath323 & @xmath359 + @xmath227 & @xmath2 @xmath14@xmath4 & 0.001 & @xmath135 & @xmath336 & @xmath208 & @xmath360 + @xmath227 & @xmath2 @xmath14@xmath4 & 0.019 & @xmath361 & @xmath362 & @xmath161 & @xmath363 + @xmath227 & @xmath2 @xmath14@xmath4 & 0.03 & @xmath364 & @xmath365 & @xmath161 & @xmath366 + @xmath227 & @xmath3 @xmath14@xmath4 & 0.0001 & @xmath125 & @xmath367 & @xmath323 & @xmath359 + @xmath227 & @xmath3 @xmath14@xmath4 & 0.001 & @xmath154 & @xmath314 & @xmath208 & @xmath360 + @xmath227 & @xmath3 @xmath14@xmath4 & 0.019 & @xmath368 & @xmath369 & @xmath323 & @xmath363 + @xmath227 & @xmath3 @xmath14@xmath4 & 0.03 & @xmath370 & @xmath371 & @xmath323 & @xmath366 + @xmath227 & f205w@xmath14@xmath4 & 0.0001 & @xmath166 & @xmath372 & @xmath323 & @xmath359 + @xmath227 & f205w@xmath14@xmath4 & 0.001 & @xmath336 & @xmath373 & @xmath208 & @xmath360 + @xmath227 & f205w@xmath14@xmath4 & 0.019 & @xmath219 & @xmath374 & @xmath323 & @xmath363 + @xmath227 & f205w@xmath14@xmath4 & 0.03 & @xmath173 & @xmath210 & @xmath323 & @xmath366 + @xmath227 & f222m@xmath14@xmath4 & 0.0001 & @xmath180 & @xmath375 & @xmath323 & @xmath359 + @xmath227 & f222m@xmath14@xmath4 & 0.001 & @xmath180 & @xmath376 & @xmath323 & @xmath360 + @xmath227 & f222m@xmath14@xmath4 & 0.019 & @xmath222 & @xmath240 & @xmath323 & @xmath363 + @xmath227 & f222m@xmath14@xmath4 & 0.03 & @xmath180 & @xmath377 & @xmath323 & @xmath366 + f160w@xmath14f205w & @xmath2 @xmath14f205w & 0.0001 & @xmath243 & @xmath378 & @xmath323 & @xmath379 + f160w@xmath14f205w & @xmath2 @xmath14f205w & 0.001 & @xmath380 & @xmath381 & @xmath176 & @xmath382 + f160w@xmath14f205w & @xmath2 @xmath14f205w & 0.019 & @xmath383 & @xmath384 & @xmath133 & @xmath385 + f160w@xmath14f205w & @xmath2 @xmath14f205w & 0.03 & @xmath386 & @xmath387 & @xmath133 & @xmath388 + f160w@xmath14f205w & @xmath3 @xmath14f205w & 0.0001 & @xmath262 & @xmath389 & @xmath323 & @xmath379 + f160w@xmath14f205w & @xmath3 @xmath14f205w & 0.001 & @xmath243 & @xmath136 & @xmath128 & @xmath382 + f160w@xmath14f205w & @xmath3 @xmath14f205w & 0.019 & @xmath306 & @xmath390 & @xmath208 & @xmath385 + f160w@xmath14f205w & @xmath3 @xmath14f205w & 0.03 & @xmath391 & @xmath392 & @xmath208 & @xmath388 + f160w@xmath14f205w & @xmath4@xmath14f205w & 0.0001 & @xmath243 & @xmath393 & @xmath323 & @xmath379 + f160w@xmath14f205w & @xmath4@xmath14f205w & 0.001 & @xmath380 & @xmath127 & @xmath161 & @xmath382 + f160w@xmath14f205w & @xmath4@xmath14f205w & 0.019 & @xmath241 & @xmath394 & @xmath161 & @xmath385 + f160w@xmath14f205w & @xmath4@xmath14f205w & 0.03 & @xmath241 & @xmath395 & @xmath161 & @xmath388 + f160w@xmath14f205w & f222m@xmath14f205w & 0.0001 & @xmath203 & @xmath384 & @xmath323 & @xmath379 + f160w@xmath14f205w & f222m@xmath14f205w & 0.001 & @xmath352 & @xmath396 & @xmath176 & @xmath382 + f160w@xmath14f205w & f222m@xmath14f205w & 0.019 & @xmath346 & @xmath244 & @xmath165 & @xmath385 + f160w@xmath14f205w & f222m@xmath14f205w & 0.03 & @xmath131 & @xmath397 & @xmath259 & @xmath388 + f160w@xmath14f222 m & @xmath2 @xmath14f222 m & 0.0001 & @xmath294 & @xmath398 & @xmath323 & @xmath399 + f160w@xmath14f222 m & @xmath2 @xmath14f222 m & 0.001 & @xmath294 & @xmath400 & @xmath208 & @xmath401 + f160w@xmath14f222 m & @xmath2 @xmath14f222 m & 0.019 & @xmath402 & @xmath391 & @xmath128 & @xmath403 + f160w@xmath14f222 m & @xmath2 @xmath14f222 m & 0.03 & @xmath404 & @xmath391 & @xmath208 & @xmath405 + f160w@xmath14f222 m & @xmath3 @xmath14f222 m & 0.0001 & @xmath245 & @xmath406 & @xmath323 & @xmath399 + f160w@xmath14f222 m & @xmath3 @xmath14f222 m & 0.001 & @xmath266 & @xmath407 & @xmath128 & @xmath401 + f160w@xmath14f222 m & @xmath3 @xmath14f222 m & 0.019 & @xmath295 & @xmath247 & @xmath161 & @xmath403 + f160w@xmath14f222 m & @xmath3 @xmath14f222 m & 0.03 & @xmath408 & @xmath284 & @xmath161 & @xmath405 + f160w@xmath14f222 m & @xmath4@xmath14f222 m & 0.0001 & @xmath294 & @xmath409 & @xmath323 & @xmath399 + f160w@xmath14f222 m & @xmath4@xmath14f222 m & 0.001 & @xmath294 & @xmath369 & @xmath323 & @xmath401 + f160w@xmath14f222 m & @xmath4@xmath14f222 m & 0.019 & @xmath266 & @xmath410 & @xmath208 & @xmath403 + f160w@xmath14f222 m & @xmath4@xmath14f222 m & 0.03 & @xmath266 & @xmath286 & @xmath208 & @xmath405 + f160w@xmath14f222 m & f205w@xmath14f222 m & 0.0001 & @xmath135 & @xmath411 & @xmath323 & @xmath399 + f160w@xmath14f222 m & f205w@xmath14f222 m & 0.001 & @xmath368 & @xmath412 & @xmath161 & @xmath401 + f160w@xmath14f222 m & f205w@xmath14f222 m & 0.019 & @xmath334 & @xmath413 & @xmath176 & @xmath403 + f160w@xmath14f222 m & f205w@xmath14f222 m & 0.03 & @xmath327 & @xmath414 & @xmath176 & @xmath405 + lcccccc 0.0001 & @xmath74 & 1.574@xmath4150.001 & 1.572@xmath4150.001 & 1.574@xmath4150.001 & 1.544@xmath4150.001 & 1.566@xmath4150.001 + 0.0001 & @xmath75 & 1.570@xmath4150.002 & 1.567@xmath4150.003 & 1.569@xmath4150.003 & 1.539@xmath4150.005 & 1.557@xmath4150.006 + 0.0001 & @xmath64 & 1.566@xmath4150.002 & 1.563@xmath4150.002 & 1.565@xmath4150.002 & 1.528@xmath4150.008 & 1.544@xmath4150.008 + 0.0001 & @xmath59 & 1.564@xmath4150.002 & 1.560@xmath4150.002 & 1.562@xmath4150.002 & 1.520@xmath4150.007 & 1.537@xmath4150.006 + 0.001 & @xmath416 & 1.570@xmath4150.003 & 1.567@xmath4150.003 & 1.570@xmath4150.003 & 1.539@xmath4150.005 & 1.557@xmath4150.007 + 0.001 & @xmath75 & 1.568@xmath4150.004 & 1.565@xmath4150.003 & 1.568@xmath4150.004 & 1.536@xmath4150.009 & 1.553@xmath4150.009 + 0.001 & @xmath64 & 1.562@xmath4150.005 & 1.560@xmath4150.003 & 1.563@xmath4150.003 & 1.520@xmath4150.009 & 1.538@xmath4150.006 + 0.001 & @xmath59 & 1.556@xmath4150.009 & 1.557@xmath4150.005 & 1.559@xmath4150.005 & 1.511@xmath4150.013 & 1.532@xmath4150.009 + 0.019 & @xmath74 & 1.574@xmath4150.001 & 1.571@xmath4150.001 & 1.574@xmath4150.001 & 1.545@xmath4150.001 & 1.566@xmath4150.001 + 0.019 & @xmath75 & 1.561@xmath4150.013 & 1.562@xmath4150.008 & 1.564@xmath4150.009 & 1.526@xmath4150.021 & 1.547@xmath4150.017 + 0.019 & @xmath64 & 1.549@xmath4150.009 & 1.555@xmath4150.006 & 1.555@xmath4150.006 & 1.505@xmath4150.015 & 1.529@xmath4150.010 + 0.019 & @xmath59 & 1.539@xmath4150.010 & 1.548@xmath4150.007 & 1.548@xmath4150.007 & 1.489@xmath4150.016 & 1.517@xmath4150.012 + 0.03 & @xmath416 & 1.568@xmath4150.008 & 1.567@xmath4150.005 & 1.569@xmath4150.006 & 1.538@xmath4150.012 & 1.557@xmath4150.011 + 0.03 & @xmath75 & 1.560@xmath4150.014 & 1.562@xmath4150.008 & 1.563@xmath4150.009 & 1.526@xmath4150.022 & 1.547@xmath4150.017 + 0.03 & @xmath64 & 1.547@xmath4150.010 & 1.554@xmath4150.006 & 1.554@xmath4150.006 & 1.503@xmath4150.014 & 1.528@xmath4150.010 + 0.03 & @xmath59 & 1.538@xmath4150.009 & 1.548@xmath4150.006 & 1.548@xmath4150.006 & 1.489@xmath4150.013 & 1.517@xmath4150.010 +

we aim to settle the debate regarding the fraction of the local group s peculiar velocity that is induced by structures beyond the great attractor by calculating the dipole anisotropy of the largest , all - sky , truly x - ray selected cluster sample compiled to date . the sample is the combination of the reflex catalog in the southern hemisphere , the ebcs sample in the north , and the ciza survey in the galactic plane . the composite reflex+ebcs+ciza sample overcomes many of the problems inherent to previous galaxy and cluster catalogs which limited their effectiveness in determining the origin of the local group s motion . from the dipole anisotropy present in the cluster distribution we determine that of the local group s peculiar velocity is due to infall into the great attractor region , while is in the form of a large - scale flow induced by more distant overdensities between and mpc away . in agreement with previous analyses , we find that the shapley supercluster is the single overdensity most responsible for the increase in the dipole amplitude beyond mpc , generating of the large - scale contribution . despite the dynamical significance of both the great attractor and shapley regions , we find that additional superclusters play an important role in shaping the local group s peculiar velocity . locally , the perseus - pisces region counteracts much of the great attractor s effect on the acceleration field beyond mpc . at larger distances we find that numerous groupings and loose associations of clusters at roughly the same distance as the shapley region induce a significant acceleration on the local group . these include the well known horologium - reticulum concentration , as well as newly noted associations centered on abell 3667 and abell 3391 and a string of ciza clusters near c1410 which may trace an extension of the shapley complex into the zone of avoidance . we also note the presence of a significant underdensity of clusters in the northern hemisphere roughly mpc away and suggest that the large - scale anisotropy observed in the cluster distribution near this distance may have as much to do with the presence of large overdensities in the south as it does with the lack of superclusters in the north . finally we discuss reasons for the discordant results obtained using cluster and galaxy samples in determining the origin of the local group s motion .

astro-ph0510106

it is generally accepted that the dipole signature in the cosmic microwave background ( cmb ) is due to a doppler effect arising from the motion of the local group ( lg ) through the cosmological reference frame . although the direction and amplitude of this motion is known to high accuracy , its source has yet to be conclusively determined . while linear perturbation theory predicts that the lg s peculiar velocity is induced by anisotropies in the surrounding matter distribution , there has been disagreement regarding the distance out to which inhomogeneities in the density field continue to affect the lg s dynamics . the debate has centered on whether the lg is principally accelerated by a massive , nearby attractor which has remained hidden behind the galactic plane , or whether a significant portion of its motion is in the form of a large - scale bulk flow induced by more distant structures , as has been suggested by an increasing number of studies . resolving the source of the lg s motion and the bulk flow in which it participates carries many interesting cosmographical and cosmological implications . for example , in order for a distant supercluster like the shapley concentration ( shapley 1930 ) to induce an infall at the lg as large as the one produced by the more nearby hydra - centaurus complex , its mass would need to be sixteen times greater than that of the largest overdensity observed in the local volume . furthermore , if distant structures contribute to the lg s peculiar velocity , then anisotropies in the large - scale matter distribution must exist to at least those structures , implying that the universe becomes isotropic only at larger distances . early attempts to determine the source of the lg s peculiar velocity largely suggested a local origin . the nearest large - scale overdensity , the virgo cluster , which generates a 240 km s@xmath8 infall velocity at the lg ( jerjen & tammann 1993 ) , accounts for 27% of the lg s 627 km s@xmath8 velocity toward the cmb dipole ( kogut 1993 ) . virgo s inability to explain all of the lg s motion led shaya ( 1984 ) to suggest an additional flow toward the hydra - centaurus supercluster . this flow was subsequently detected by lynden - bell et al . ( 1988 , hereafter lb88 ) as a systematic distortion in the peculiar velocities of 400 early - type galaxies within @xmath9 mpc and its amplitude at the lg was estimated to be 570 km s@xmath8 . assuming this motion was due to infall into a single `` great attractor '' ( ga ) , lb88 estimated that the source of the flow was located roughly @xmath10 mpc away , or about @xmath11 mpc behind centaurus . this distance , coupled with the large infall velocity , implied the rather high mass of @xmath12 m@xmath13 for the ga complex . the lb88 findings suggested that the ga was responsible for the remaining @xmath14 of the lg s peculiar velocity not induced by the virgo cluster . despite the lb88 findings , subsequent redshift surveys that have encompassed the ga region have failed to detect a mass overdensity as large as the one implied by the lb88 peculiar velocity data ( dressler 1988 ; strauss et al . 1992 ; hudson 1993 , 1994 ) , nor have they conclusively measured the backside infall into the ga one would expect if the region were best described as a single , stationary attractor ( mathewson et al . 1992 , courteau et al . 1993 ) . even with the more recent discoveries of rich clusters such as norma ( abell 3627 , kraan - korteweg et al . 1996 ) and ciza j1324.7 - 5736 ( ebeling , mullis & tully 2002 ) near the hydra - centaurus region , there remains a significant discrepancy between the mass concentration observed in the ga and the mass originally proposed by lb88 ( staveley - smith et al . 2000 , kocevski et al . 2006 ) . there is now a growing volume of work suggesting the dynamical significance of the ga was originally overestimated and that some component of the lg s peculiar velocity is in the form of a large - scale bulk flow which continues past the ga region and is induced by more distant structures . one of the first suggestions of a non - local component was put forth by plionis ( 1988 ) , who proposed that the anisotropy observed in the galaxy number - counts of the lick catalog ( shane & wirtanen 1967 ) implied a portion of the lg s motion originating from beyond @xmath15 mpc . more recently , hudson et al . ( 2003 ) show that results from the peculiar velocity studies of hudson et al . ( 1999 ) , dale et al . ( 1999 ) , willick ( 1999 ) , and colless et al . ( 2001 ) are all consistent with a 350 km s@xmath8 bulk flow continuing beyond @xmath16 mpc , ruling out nearby overdensities such as the ga as the source of the motion . likewise , zaroubi et al . ( 1999 ) show that if the local velocity field is decomposed into its divergent ( locally produced ) and tidal ( externally produced ) components , only @xmath17 of the flow toward the ga is due to infall into the hydra - centaurus region , while the remaining velocity is due to a continuing bulk flow generated by attractors beyond @xmath18 mpc . in addition , tonry et al . ( 2000 ) , using surface brightness fluctuation ( sbf ) distances , find a mass for the ga that is @xmath19 times less than the original lb88 estimate . their best - fit models suggest the ga is well centered on the centaurus cluster ( as opposed to @xmath11 mpc behind it ) and that part of the flow toward the ga is due to a @xmath20 km s@xmath8 residual bulk motion . tonry et al . essentially propose that the lg s motion , which was mistakenly identified as infall into a single , massive attractor slightly behind centaurus , is in fact the result of two flows , one into the hydra - centaurus region and a second flow toward a more distant source . many of the studies finding evidence for a continuing flow beyond the ga have suggested that some fraction of the motion may be due to infall into the shapley supercluster ( ssc ) , located @xmath21 mpc behind the hydra - centaurus complex . the ssc region is unique in the local universe , containing the richest concentration of clusters out of all the 220 identified superclusters out to @xmath22 ( einasto et al . in fact , the ssc contains more than 4 times the number of rich clusters present in the ga region . the ssc s possible dynamical significance was pointed out by scaramella et al . ( 1989 , 1991 ) and plionis & valdarnini ( 1991 ) because of the high concentration of abell clusters in the region and its directional alignment with the ga . since the two regions are only separated by @xmath23 on the sky , kocevski et al . ( 2004 ) suggest this alignment causes a bootstrap effect that sets in place the large - scale density anisotropy that is responsible for the lg s peculiar velocity . despite the significant overdensity present in the ssc , its dynamical impact on the lg has been a matter of debate . rowan - robinson et al . ( 2000 ) use the dipole anisotropy present in the iras point source catalog redshift sample ( pscz , saunders et al . 1995 ) to conclude that the ssc has a marginal influence on the lg , inducing only @xmath24 km s@xmath8 of its motion . likewise , erdogdu et al . ( 2005 ) , using the dipole anisotropy of the 2 micron all - sky redshift survey ( 2mrs , huchra et al . 2005 ) , find that structures beyond @xmath25 mpc induce only a negligible acceleration on the lg . on the other hand , lucey et al . ( 2005 ) combine the cluster peculiar velocity data from the smac ( hudson et al . 1999 ) , sci / scii ( giovanelli et al . 1999 ) , enearc ( bernardi et al . 2002 ) and sbf ( tonry et al . 2000 ) surveys and find that both the ga and the ssc generate an equal amount of the lg s peculiar velocity . this higher estimate concurs with the results from dipole analyses of various cluster samples ; branchini & plionis ( 1996 ) and plionis & kolokotronis ( 1998 ) use the dipole anisotropy in the distribution of optically selected abell / aco ( abell 1958 , abell et al . 1989 ) and x - ray brightest abell - type cluster ( xbac , ebeling et al . 1996 ) samples , respectively , to determine that @xmath26 of the lg s motion is due to the ssc . kocevski et al . ( 2004 ) recently added x - ray selected clusters from the ciza survey ( named for clusters in the zone of avoidance , ebeling , mullis , & tully 2002 ) to the xbac distribution to fill in the zone of avoidance ( zoa ) and found an even larger ssc contribution ( @xmath27 ) . in this study we aim to settle the debate regarding the fraction of the lg s peculiar velocity that is induced by structures beyond the ga by calculating the dipole anisotropy of the largest , all - sky , truly x - ray selected cluster sample compiled to date . the sample is the combination of the rosat - eso flux limited x - ray catalog ( reflex , bhringer et al . 2004 ) in the southern hemisphere , the extended brightest cluster sample ( ebcs , ebeling et al . 1998 , 2000 ) in the north , and the ciza survey in the galactic plane . the composite reflex+ebcs+ciza sample ( hereafter rbc ) overcomes many of the problems inherent to previous galaxy and cluster catalogs which limited their effectiveness in determining the origin of the lg s motion . first of all , due to its x - ray selected nature , the rbc sample maps the distribution of massive x - ray luminous clusters which tend to be the accelerators of large - scale flows . in this sense , the rbc sample traces the peaks of the density fluctuation field better than galaxy samples such as the pscz , which has been shown to undersample dense regions such as the ssc ( kaiser et al . x - ray bright clusters are also luminous enough for statistically complete samples to be constructed out to larger distances than galaxy samples , which face an increasing incompleteness beyond 60 @xmath4 mpc . we suspect that the estimate for a marginal ssc acceleration from the pscz data may be a result of these two factors . second , the x - ray selected rbc sample overcomes several limitations of early cluster catalogs such as abell / aco , which are optically selected . since optical selection methods rely on identifying individual clusters through overdensities in the projected galaxy distribution and assigning cluster masses based on the size , or richness , of those overdensities , fluctuations in the surface density of field galaxies as well as superpositions of poor clusters or filamentary structure along the line of sight can lead to false detections or overestimates of a system s richness ( e.g. van haarlem 1997 , see sutherland 1988 ) . these projection effects work to systematically amplify the measured dipole amplitude . in addition , optical cluster searches suffer from severe extinction and stellar obscuration in the direction of the milky way , leading to catalogs with poor coverage in a @xmath28 wide strip centered on the plane of the galaxy . this is particularly troubling since large - scale structures associated with the ga and the ssc are known to exist in , or extend into , the zoa ( tully et al . 1992 ) . a variety of techniques have been used to reconstruct the zoa , ranging from uniform filling ( strauss & davis 1988 ; lahav 1987 ) to a spherical - harmonics approach which extends structures above and below the plane into the zoa ( plionis & valdarnini 1991 , cf . brunozzi et al . 1995 ) , but the value of these reconstruction techniques is limited if the milky way does indeed obscure dynamically significant regions , as has been suggested . the x - ray selected nature of the rbc sample makes it preferable over its optically selected counterparts since ( 1 ) cluster x - ray emission originates from the @xmath29 k intracluster medium , which is more peaked at the gravitational center of the cluster than the projected galaxy distribution . this minimizes projection effects since clusters would need to be in almost perfect alignment to be mistaken for a single , more luminous object , ( 2 ) x - ray luminosity is closely correlated with cluster mass ( reiprich & bhringer 1999 ) , thus providing a better estimate to a system s dynamical impact than a clusters projected galaxy richness , and ( 3 ) x - ray emission does not suffer as severe an extinction in the plane of the galaxy ( ebeling , mullis & tully 2002 ) . finally , the rbc s relatively low flux limit makes it a better tracer of large - scale structures than previous x - ray confirmed cluster samples , such as the xbac catalog which has recently been used by plionis & kolokotronis ( 1998 ) and kocevski et al . ( 2004 ) to investigate the origin of the lg s motion . due to the xbac s fairly high x - ray flux limit , the sample is limited to only the most massive clusters , which , although tracing the deepest potential wells , only sparsely sample the underlying density field and can lead to an increased level of shot - noise . in addition , although x - ray confirmation effectively eliminates projection effects in xbac , the catalog remains optically selected , therefore clusters missed by the abell / aco sample will not be included in the xbac sample . this essentially means very nearby , very extended clusters are systematically missed as they often do not contrast strongly with the background galaxy population . the undersampling of clusters at low redshifts would lead to an overestimate of the contribution to the dipole from distances greater than 60 @xmath4 mpc , where the xbac incompleteness is minimal . the rbc s x - ray flux limit is nearly half of the one used in the xbac survey , which means the sample is no longer limited to extremely massive clusters . this increased depth more than triples the number of clusters present in the rbc sample , which in turn affords us a greater resolution in tracing the overdensities which give rise to the lg s motion . the rbc sample is currently the largest , most complete x - ray selected cluster sample for which to trace the large - scale structure of the local universe and determine the origin of the lg s peculiar velocity . in what follows we use the rbc sample to ( 1 ) determine the role of the cluster distribution in producing the lg s peculiar velocity , ( 2 ) determine the fraction of the lg s motion that is induced from distances greater than 60 @xmath4 mpc , and ( 3 ) estimate the relative contribution of various superclusters such as the ga and ssc regions , to the final dipole amplitude . we proceed in the following manner : in 2 we give an overview of the reflex , ebcs , and ciza samples , 3 describes the details of the dipole analysis , and our results are put forward in 4 . finally we summarize our primary conclusions in section 5 . throughout this paper we assume an einstein - de sitter universe with @xmath30 and @xmath31 @xmath32 km s@xmath8mpc@xmath8 unless otherwise stated , so that our results are directly comparable to those of previous studies . we aim to measure the dipole anisotropy present in the all - sky , x - ray selected cluster sample created by combining the reflex , ebcs and ciza catalogs . in this section we describe the attributes of the reflex , ebcs and ciza samples in further detail and the methods used to homogenize them into one all - sky catalog . the reflex catalog is the most comprehensive x - ray selected cluster sample compiled for the southern hemisphere , consisting of 447 clusters with x - ray fluxes greater than @xmath33 erg @xmath34 s@xmath8 in the 0.12.4 kev band . the survey is limited to declinations of @xmath35 , redshifts of @xmath36 and galactic latitudes away from the galactic plane ( @xmath37 ) . reflex hails from the rosat all - sky survey ( rass , voges 1992 ) and is therefore truly x - ray selected . as described in bhringer et al . ( 2001 ) , the rass bright source catalog ( voges et al . 1999 ) provided 54076 x - ray detections which were reexamined by means of a growth curve analysis ( bhringer et al . 2000 ) , resulting in 1417 targets with x - ray fluxes above @xmath33 erg @xmath34 s@xmath8 . cross - correlating these targets with galaxy overdensities in the cosmos optical database ( macgillivray & stobie 1984 ) produced 673 candidate clusters . subsequent screening and follow - up observations led to 447 confirmed galaxy clusters in the final reflex sample . estimated to be over 90@xmath38 complete , reflex is the largest cluster catalog produced from the rass to date . the ebcs catalog is the most complete x - ray flux - limited cluster sample constructed from the rass for the northern hemisphere , consisting of 290 clusters with x - ray fluxes greater than @xmath33 erg @xmath34 s@xmath8 in the 0.12.4 kev band . the sample is limited to declinations of @xmath39 and redshifts of @xmath36 and , like reflex , the survey avoids the galactic plane ( @xmath37 ) . the ebcs sample was compiled by cross - correlating the rass with existing cluster catalogs such as the abell / aco and zwicky ( zwicky et al . 1961 - 68 ) samples , while also using the voronoi tessellation and percolation ( vtp , ebeling 1993 ; ebeling & wiedenmann 1993 ) algorithm to allow for an improved determination of basic cluster characteristics , as well as to detect additional clusters based on their x - ray properties alone . the fact that the ebcs is not selected from the rass bright source catalog in the same manner as reflex leads to slight differences in the global properties of the two samples ; corrections for this and other systematic effects are discussed in 2.4 . the ebcs catalog is estimated to be 75% complete within the 2190 deg@xmath40 processed by the vtp algorithm . the ciza sample is the product of the first systematic search for x - ray luminous clusters behind the plane of the galaxy . as described in ebeling , mullis & tully ( 2002 , hereafter emt ) , ciza targets were selected from the rass bright source catalog if they met three criteria : ( 1 ) location in the zoa , @xmath41 , ( 2 ) an x - ray flux greater than @xmath42 erg @xmath34 s@xmath8 ( 0.12.4 kev ) and ( 3 ) a spectral hardness ratio exceeding a preset threshold value to discriminate against softer , non - cluster sources . the resulting target list of 1901 sources was cross - correlated with existing databases to identify known clusters and remove obvious non - clusters . the remaining cluster candidates were then subjected to a comprehensive imaging and spectroscopic follow - up campaign . the use of the bright source catalog as a target list means ciza is not correlated with any optically selected catalog and is therefore truly x - ray selected . a subsample of 73 ciza clusters with fluxes above @xmath43 erg @xmath34 s@xmath8 ( the b1 sample ) has recently been published ( emt ) . to allow the combination of ciza with reflex and ebcs , we use the total fluxes of the b1 sample listed by emt and also add clusters from a second , yet unpublished extended sample , which includes all clusters whose total fluxes within a metric 1.5 mpc aperture exceed @xmath33 erg @xmath34 s@xmath8 . the resulting sample , limited to @xmath44 and @xmath45 , contains 151 clusters . the combined effects of differing selection techniques , flux measuring algorithms and other systematic effects make merging the reflex , ebcs and ciza catalogs a non - trivial task . the method used to combine the individual catalogs into one homogeneous all - sky sample will be described in greater detail in a forthcoming paper and is therefore only briefly discussed here . first we need to ensure that the x - ray flux of each cluster is measured in a consistent manner . the need for uniformly measured fluxes is two - fold : it ensures that ( 1 ) all three samples are complete to the same depth and ( 2 ) the weight given to each cluster in the dipole analysis , which is derived from the cluster s x - ray luminosity , is determined in the same manner throughout the sky . the depth and weighting of the sample must be considered carefully when measuring the dipole anisotropy , since systematic differences in either can introduce spurious contributions to the dipole amplitude and pointing . as published , reflex fluxes are determined using a growth curve analysis which corrects for any x - ray flux missed outside the detection aperture by extrapolating out to the cluster s estimated virial radius . conversely , the ebcs flux correction extrapolates out to infinity , which leads to an @xmath46 difference in the fluxes measured by the two approaches ( bhringer et al . finally , ciza fluxes are measured by summing all of the x - ray emission within a metric 1.5 mpc radius ( at the cluster redshift ) about the centroid of the cluster s x - ray emission . in this study , we adopt this latter approach and recalculate the flux of all reflex , ebcs and ciza clusters using a metric 1.5 mpc aperture . using rass data , we redetermine the centroid of each cluster s x - ray emission , remove point sources within the detection aperture , and calculate a new x - ray count rate at the location of each cluster , taking into account the local rass exposure time . the x - ray background at each position is determined in a 1 mpc wide annulus ( 2 - 3 mpc from the cluster centroid ) about the measurement aperture and subtracted from the observed count rate . the measured count rates are then converted to unabsorbed fluxes in the 0.1 - 2.4 kev band by taking into account the line - of - sight interstellar hydrogen column density as given by dickey & lockman ( 1990 ) . next , clusters whose x - ray emission appears to be dominated by a point source are removed if more than 50% of their x - ray emission originates from within the central 10% of the detection aperture . finally a flux cut at @xmath33 erg @xmath34 s@xmath8 is applied , leaving 359 reflex , 248 ebcs , and 151 ciza clusters in the resulting sample . the distribution of the combined sample in galactic coordinates is shown in figure 1 and its redshift distribution is shown in figure 2 . the distribution of cluster luminosities versus redshift is shown in figure 3 . in addition to the different flux measuring procedures of the three catalogs , differences also exist in the selection technique employed . whereas reflex and ciza had the bright source catalog source detection algorithm run on rass data covering their entire survey area , the ebcs employed the vtp to search for clusters only over regions which were correlated with previously known clusters . while this led to the detection of both known and unknown clusters , the serendipitously detected clusters could only be found in the 2190 deg@xmath40 which were processed with vtp . given the frequency of serendipitously detected clusters in the surveyed area ( 17 in 2190 deg@xmath40 ) , 84 additional clusters should exist with fluxes above @xmath33 erg @xmath34 s@xmath8 in regions of the northern sky not processed by the ebcs . since reflex , by design , should detect all such serendipitous clusters in its survey area , a density variation is introduced between the northern and southern portions of the combined sample . left uncorrected such a systematic density difference will bias the dipole pointing toward the southern portion of the sample . to compensate for this effect in the dipole analysis , we weight each ebcs cluster by the difference in the comoving cluster density between the ebcs and reflex . this is essentially equivalent to weighting each ebcs cluster by a factor of @xmath47 , which is the multiplicative correction needed to make up for the 84 clusters not included in the ebcs catalog . this weighting scheme is similar to the one employed to correct the density variation between the abell and aco cluster samples ( brunozzi et al . 1995 , plionis & kolokotronis 1998 ) , although our weighting is distributed uniformly throughout the ebcs since we do not expect the missing clusters to be correlated with any particular part of the sky or with redshift . to check the effectiveness of this weighting scheme , we fit a schechter function to the x - ray luminosity functions of the reflex and ebcs portions of our sample and find that the best - fit parameters agree quite well between the two sections . the binned luminosity functions , which were constructed using our recalculated x - ray fluxes and the corrective weighting for the ebcs , are shown in figure 4 . finally , systematic effects in the ciza sample are corrected for in much the same manner as for the ebcs . the catalog s primary incompleteness comes from the fact that clusters are systematically missed in the very central regions of the zoa , which is due to the increased difficulty of obtaining spectroscopic confirmation of clusters through the severe extinction within @xmath48 of the galactic plane and toward the galactic center . to correct for this we again weight each ciza cluster by the difference in the comoving cluster density between the ciza and reflex samples , which amounts to a factor of @xmath49 . given the relatively sparse nature of cluster samples and the known clustering of clusters , we find that distributing the weight over the entire sample is preferable to adding random clusters to the galactic plane . to ensure that the weighting of ebcs and ciza clusters in this manner does not overly influence our findings , we calculate the dipole with our weights set to unity and find that our results are quite robust to variations in the adopted weighting scheme . linear theory of gravitational instability dictates that the peculiar velocity of a reference frame can be related to the gravitational acceleration induced by the mass distribution surrounding it via @xmath50 ( peebles 1976 ) , where @xmath51 and @xmath52 is the biasing parameter relating the mass tracers to the underlying mass distribution they represent , and @xmath53 is the average mass - tracer number density . in other words , equation 1 tells us that the dipole moment of a mass - tracer distribution can be directly related to the peculiar velocity that sample would induce on the lg . for a non - continuous , flux - limited tracer sample , such as a cluster catalog , a more useful version of equation ( 1 ) is @xmath54 @xmath55 where @xmath56 is the distance to each cluster , @xmath57 is the sample s selection function at @xmath56 , @xmath58 is weight assigned to the @xmath59th cluster , and @xmath60 are the unit vectors pointing to the position of each cluster . @xmath61 is the vector quantity we refer to as the dipole throughout the rest of this paper . from equation ( 2 ) it can be seen that the characteristics of the dipole are such that its amplitude will increase with distance until the largest inhomogeneity in the sample is encompassed and isotropy is reached , after which the dipole flattens out to its final value . this flattening signals that the convergence depth , @xmath62 , has been reached . assuming that @xmath61 is well aligned with the lg s peculiar velocity , @xmath63 , at this convergence depth , equation ( 2 ) provides the means to estimate the @xmath64 parameter . the rest of this section is devoted to describing the distance , selection function and cluster weights used in equation ( 2 ) . to convert our observed redshifts to distances , we use the formula of mattig ( 1958 ) : @xmath65,\ ] ] which reduces to @xmath66 for our assumed value of @xmath67 . to account for possible peculiar - velocity contamination in our redshifts , which may alter the perceived distance to a cluster , we follow the approach of kocevski et al . ( 2004 ) and perform our analysis in both the lg and cmb reference frames , since it has been shown that the cluster dipole calculated in these frames over- and under - estimate , respectively , the real - space dipole ( branchini & plionis 1996 ) . we transform our measured heliocentric redshifts , @xmath68 , into the lg and cmb rest frames using @xmath69 and @xmath70 \nonumber\end{aligned}\ ] ] where @xmath71 is the amplitude of the lg velocity as inferred from the cmb dipole anisotropy and @xmath72 is the direction of this motion in galactic coordinates . authors have traditionally removed the lg s virgocentric infall velocity from @xmath71 since previous samples such as xbac - ciza did not include virgo . for a meaningful comparison between our results and those of earlier studies we follow the same approach and set @xmath73 km s@xmath8 and @xmath72 = @xmath74 . the inverse of the sample s selection function , @xmath75 , is needed in equation ( 2 ) whenever a flux - limited catalog is employed in order to correct for the non - detection of intrinsically less luminous objects with increasing distance . the selection function is defined as the fraction of the cluster number density that is observed above the flux limit at a given distance : @xmath76 where @xmath77 is the average cluster density , @xmath78 is the x - ray cluster luminosity function and @xmath79 , where @xmath80 is the flux limit . we estimate @xmath77 by integrating the luminosity function over the entire luminosity range of the sample @xmath81 where the lower luminosity is @xmath82 @xmath83 ergs s@xmath8 . for the luminosity function , @xmath78 , we adopt a single schechter - like function of the form @xmath84 and fit it to our combined rbc sample . our best - fit values for @xmath85 and @xmath86 are @xmath87 @xmath88 mpc@xmath89 , @xmath90 @xmath91 ergs s@xmath8 @xmath34 , and @xmath92 , respectively . using these parameters we find @xmath93 @xmath94 mpc@xmath89 . figure 4 shows the binned x - ray luminosity functions for the reflex and ebcs portions of our sample , which were constructed using our recalculated x - ray fluxes and corrective weights for the ebcs . finally , in addition to the corrective weights described in 2.4 , the @xmath58 term in equation ( 2 ) includes a component which weights clusters based on their relative mass . we estimate the mass of each cluster in our sample through the empirical relationship @xmath95 ( allen et al . 2003 ) which links a cluster s x - ray luminosity to its mass contained within the radius @xmath96 , defined as the distance where the mean enclosed density is 200 times the critical density of the universe at the redshift of the cluster . we perform our dipole analysis with and without these mass weights and present both sets of results . the use of discrete objects such as clusters to trace the underlying density field introduces a level of shot noise to the dipole calculation . we estimate the amount of noise introduced by the sparseness of the rbc sample by the method of hudson ( 1993 ) : @xmath97 such that @xmath98 is the root - mean - square ( rms ) of the cumulative variance of the dipole vector sum . assuming that the shot - noise variance along each dipole component is equal , the mean one - dimensional error is @xmath99 . we find that shot - noise produces roughly @xmath100 of the rbc cluster dipole signal at 300 @xmath4 mpc ( number - weighted amplitude ) ; this is roughly a @xmath101 reduction of the shot - noise contribution in the sparser xbacs - ciza dipole . it should be noted that while we calculate and plot the shot - noise amplitude along with the cluster dipole , we do not subtract the shot - noise component from the dipole amplitudes shown in figures 4 , 5 and 7 . the measured amplitude of the dipole anisotropy present in the composite rbc sample , in both the number- and mass - weighting schemes , is shown in figures 5 and 6 , respectively . the dipole amplitude at any given distance is directly proportional to the peculiar velocity that is induced on the lg by the cluster distribution within that distance . the amplitude is shown in both the lg and cmb rest frames ; these should be taken as upper and lower estimates of the true dipole ( branchini & plionis 1996 ) . the direction of the dipole in both rest frames and weighting schemes is shown in figure 7 . in agreement with previous cluster dipole analyses , we find that the amplitude of the anisotropy in the rbc sample is dominated by cluster concentrations at @xmath102 and @xmath20 @xmath4 mpc . these distances largely match the locations of cluster overdensities in the ga and ssc regions . unlike the anisotropy in galaxy samples such as the pscz and 2mrs , which show only a negligible contribution to the lg s dynamics from distances beyond 60 @xmath4 mpc , we find that the rbc dipole amplitude continues to grow until roughly @xmath3 @xmath4 mpc , after which it temporarily increases and then returns to its final value . this would suggest that the rbc sample becomes isotropic with respect to the lg in the distance range of @xmath3 to @xmath103 @xmath4 mpc . the large increase in the amplitude at @xmath20 @xmath4 mpc implies that clusters beyond the ga induce a significant acceleration on the lg , which would seem to support studies that find evidence for a large amplitude bulk flow continuing beyond the ga . averaging the fraction of the number - weighted amplitude generated from beyond 60 @xmath4 mpc in both rest frames , we estimate @xmath104 of the lg s peculiar velocity originates from beyond the hydra - centaurus complex . this contribution rises to @xmath105 if the clusters are weighted by their mass . the increased large - scale contribution in the mass - weighted dipole is due to the fact that distant rbc clusters are intrinsically more luminous than their nearby counterparts ( due to the flux - limited nature of the sample ) and therefore have an increased weight in the mass - weighting scheme , which in turn increases the overall contribution from beyond 60 @xmath4 mpc . averaging the results in both rest frames and weighting schemes , we find that roughly @xmath1 of the lg s peculiar velocity is produced by clusters beyond @xmath6 @xmath4 mpc . we also find that the misalignment between the cmb and cluster dipoles varies little throughout our study volume , implying that an anisotropy in the direction of the cmb dipole is present in the cluster distribution as early as the ga region @xmath106 mpc away . this agrees with previous studies which found a large - scale anisotropy in both the optically - selected cluster distribution ( plionis & valdarnini 1991 ) and infrared - selected galaxy samples ( basilakos & plionis 1998 ) . after encompassing the ga and perseus - pegasus supercluster ( also known as perseus - pisces , hereafter pp ) , the cluster dipole is pointed within @xmath107 of the cmb dipole direction ( lg - frame , number - weighting ) . after 180 @xmath4 mpc the dipole in both rest - frames and weighting schemes makes its closest approach to the cmb dipole direction at roughly 220 @xmath4 mpc , with the smallest misalignment angle , @xmath108 , being achieved in the number weighted , lg - frame calculation . the misalignment angle then remains fairly constant until roughly 250 @xmath4 mpc , after which it steadily rises , presumably due to the increased shot - noise beyond that distance . our results are in good agreement with the findings of plionis & kolokotronis ( 1998 ) and kocevski et al . ( 2004 ) , who measured the dipole in the sparser xbac and xbac - ciza samples , respectively . the xbac - ciza dipole amplitude is shown with the results of this work in figure 8 . the qualitative similarity between the rbc and xbac - ciza dipoles is not a surprise given that the latter sample is a subset of the former and therefore traces , to a large extent , the same superclusters that influence the lg s peculiar velocity . the primary difference between our results and those of kocevski et al . ( 2004 ) is a slightly reduced large - scale contribution to the number - weighted dipole amplitude : we find 49% of the dipole is induced from beyond @xmath109 @xmath4 mpc using the rbc sample versus 62% using xbac - ciza . the reason for this difference stems from the xbac s known incompleteness at low redshifts , where very extended clusters are systematically missed due to the optical selection of the catalog . the xbac sample contains 32 clusters within 100 @xmath4 mpc , whereas the rbc has 82 within the same volume . the undersampling of xbac clusters at low redshifts leads to an increased contribution to the dipole from greater distances , where the xbac incompleteness is minimal . despite the difference in the dipole amplitude , our findings seem to confirm the general conclusions reached by kocevski et al . ( 2004 ) : a significant component of the lg s peculiar velocity is induced by overdensities beyond the ga . despite the agreement with previous cluster - based dipole analyses , our findings differ from the results obtained using galaxy catalogs such as the pscz and 2mrs . whereas we find a significant contribution to the dipole arising from 130 to 180 @xmath4 mpc , rowan - robinson et al . ( 2000 ) find only a marginal contribution to the pscz dipole over the same distance range ( see also saunders et al . 1999 and schmoldt et al . 1999a , b ) . likewise , erdogdu et al . ( 2005 ) report that the dipole of the 2mrs sample largely converges by @xmath6 @xmath4 mpc . we suspect the reason for this discrepancy is two - fold . first , while the pscz and 2mrs surveys sample the nearby galaxy population quite well , their redshift distributions reach a maximum near @xmath110 ( @xmath111 mpc ) and rapidly decline afterwards ( more so for the 2mrs than the pscz , which has a longer redshift tail ) . the rbc sample , on the other hand , has limited resolution nearby ( simply due to the volume under consideration ) but does not exhibit such a turnover until a much larger distance ( @xmath112 mpc ) . second , the iras catalog , from which the pscz sample hails , is known to undersample the elliptical galaxy population and therefore the dense regions that our x - ray selected cluster sample traces best . as we will discuss in the next section , most of our dipole signal near 150 @xmath4 mpc is produced by superclusters such as the ssc and the horologium - reticulum ( hr ) system . in other words , the densest regions in the rbc sample produce most of our signal at large distances and it is precisely these regions that are undersampled by the pscz . we propose that poor sampling by galaxy samples of both distant superclusters and dense regions such as the ssc most likely explains the discordant results from the pscz/2mrs and rbc dipole analyses . the increased size of the rbc catalog over the xbac - ciza sample affords us greater resolution in tracing the large - scale structures which give rise to the lg s peculiar velocity . since the catalog probes the cluster distribution to fainter fluxes and is therefore not limited to extremely massive clusters which only sparsely sample the underlying density field , the rbc sample allows us to better discern which superclusters have the most dynamical impact on the lg . to this end , we have examined the dipole profile on a cluster - by - cluster basis to investigate the relative contribution of various supercluster concentrations to the final dipole amplitude . a diagnostic dipole profile , which includes the location of individual cluster and superclusters , is shown in figure 9 . as previously mentioned , the rbc dipole amplitude is dominated by cluster concentrations at @xmath102 and @xmath20 @xmath4 mpc . the initial increase in the amplitude at @xmath102 @xmath4 mpc is produced almost exclusively by the ga region , which is the first supercluster encountered in the dipole calculation ( see figure 10 ) . within the rbc sample we find 13 clusters that are associated with the ga region ( i.e. in the vicinity of centaurus , norma and ciza j1324.7 - 5736 ) ; these make up @xmath113 of all the rbc clusters within @xmath114 @xmath4 mpc . at this distance the ga overdensity creates an anisotropy that is pointed within @xmath108 of the cmb dipole . we propose that this early alignment with the cmb dipole direction is the reason that many studies which have only probed the peculiar velocity field out to the ga distance manage to obtain a good alignment with the cmb anisotropy , despite overlooking more distant structures , such as the ssc . moving to larger distances we find the initial rise in the amplitude is counteracted between 50 and 60 @xmath4 mpc due to the effects of the pp supercluster which lies on the opposite side of the sky relative to the ga . between 70 to 130 @xmath4 mpc the dipole varies little in both amplitude and direction , with the largest fluctuations arising from the addition of the lepus , hercules and pisces - cetus ( pc ) superclusters . the former two overdensities are responsible for the rise and fall of the amplitude between 100 and 120 @xmath4 mpc , while the latter creates the downturn between 120 and 130 @xmath4 mpc . the next supercluster to be encompassed beyond 130 @xmath4 mpc is the ssc , which is by far the densest region in the rbc sample . although the increase near @xmath115 mpc is largely due to the ssc , we find that its origin is more complex than the signal at 40 @xmath4 mpc which was solely due to a single supercluster . to investigate the source of the increase we identify each cluster which produces a positive jump in the amplitude between @xmath2 and @xmath3 @xmath4 mpc and determine its association to known and unknown overdensities . we find that a total of 75 clusters contribute to the increase , of which 54 are reflex clusters , 12 are ciza members and 7 belong to the bcs sample . we can associate 17 of these clusters with the ssc region and 13 to the hr supercluster ( i.e. near abell 3128 ) . in addition to these well known regions , there exists numerous groupings and loose associations at roughly the same distance that have a significant effect on the amplitude . the most notable among these is a previously unknown association of clusters around a3667 . while a3667 is a well known merging cluster ( knopp et al . 1996 , markevitch et al . 1999 ) , we can find no reference of its participation in an overdensity such as a supercluster . this is not surprising since a3667 has only two abell cluster moderately nearby , a3651 and a3716 . kocevski et al . ( 2004 ) noticed a3667 since its sizable mass and favorable location provided a boost to the xbac - ciza dipole , but the cluster was not associated with a group since it has only one companion in the xbac catalog ( i.e. a3716 ) . with the reduced flux limit of the rbc sample , we can now resolve that a3667 is actually the core member of a loose association of 7 clusters . another notable group is a set of 6 clusters centered on a3391 , which lies between hr and the zoa . the group contains a3391 , a3395 , and a3380 and may be an extension of the larger hr region located 50 @xmath4 mpc away ( see figure 10 ) . also of interest are numerous ciza clusters which cross the zoa between @xmath2 and @xmath3 @xmath4 mpc . one notable structure is a string of 6 ciza clusters that traverse the zoa at the same distance as the ssc and the a3667 association . one these clusters , c1410 , is near enough to the ssc to be considered a member of the supercluster ; the smooth transition from ssc members to ciza clusters in this region suggests this string of clusters may trace an extension of the filament network in which the ssc is embedded into the zoa . this association also includes the well - known triangulum australis cluster ( c1638 ) and the ciza cluster c1652 , which was noted by kocevski et al . ( 2004 ) ; we will refer to this region as the c1410 filament hereafter . another set of interesting ciza members are 3 clusters near the outer portions of the a3391 group which may hint at an extension of this association behind the plane of the galaxy ; the groups central cluster is c0821 . the locations of these superclusters is displayed in supergalactic coordinates in figure 10 . we next calculate the contribution of each overdensity to the increase at @xmath115 mpc ; our results are listed in table 1 . we find the ssc is the greatest single contributor to the increase , with its clusters producing @xmath5 of the signal . the c1410 filament is second , contributing @xmath116 to the overall increase . the c1821 extension produces @xmath117 of the jump near @xmath115 mpc , while the a3391 association induces the next largest fraction at @xmath118 . the final two regions , the hr and a3667 associations , produce @xmath46 and @xmath119 , respectively . it should be noted that if one considers the a3391 group an extension of hr region , the combined system would be second only to the ssc in its effect of the dipole amplitude , contributing a total of @xmath120 . although the hr is further south than the ssc and on the opposite side of the zoa ( see figure 10 ) , the location of the supercluster is near enough to the dipole direction at 150 @xmath4 mpc to provide a sizable contribution to the increase . in total these cluster associations account for @xmath121 of the large - scale contribution to the cluster dipole . the remaining @xmath122 is produced by isolated clusters in the general direction of the resultant dipole pointing near @xmath115 mpc . lrr shapley & 17 & 30.4 c1410 filam . & 5 & 12.5 c1821 ext . & 3 & 9.3 a3391 assoc . & 6 & 8.6 hor - ret . & 13 & 8.3 a3667 assoc . & 7 & 5.8 all reflex & 54 & 65.4 all ciza & 12 & 29.6 all ebcs & 7 & 5.0 one final note regarding the origin of the increase near @xmath7 @xmath4 : while it is evident that the presence of numerous superclusters and overdensities near this distance and in the direction of the lg s peculiar velocity play a large role in creating the increase in the dipole amplitude , of equal importance is the absence of clusters in the opposite direction . examining the redshift distribution of the ebcs and reflex samples , it is readily apparent that compared to reflex , the ebcs is underdense between @xmath2 and @xmath3 @xmath4 mpc . we find that the region of sky within @xmath123 of the dipole pointing in this shell is 2.7 times as dense as the opposite part of the sky . there is no reason to believe this density variation is artificial , indicating that the anisotropy in the cluster distribution near @xmath7 @xmath4 mpc has as much to do with the presence of large overdensities such as the ssc and the hr concentrations in the south as it does with the lack of superclusters in the north . the fact that galaxy samples do not detect a significant dipole signal at this distance may be in part because they do not detect as large a contrast between the density field in the northern and southern hemispheres owing to their sparse sampling of the galaxy population at these distances . + + 0.6 mm fig . 8. schematic dipole profile ; see text for details . cluster associations are grouped by color to highlight their impact on the overall dipole amplitude . abell and ciza clusters begin with the letters a and c , respectively . acronyms are ga : great attractor , hor - ret : horologium - reticulum , per - peg : perseus - pegasus . + + + 0.6 mm fig . 9. the location of cluster associations highlighted in figure 9 as projected into the supergalactic coordinate system . colors and acronyms are the same as in figure 9 . the dashed line represents the boundary of the traditional zoa ( @xmath124 ) , while the dotted lines demark 50 mpc steps . with the exception of the hr and a3391 associations , the structures which are largely responsible for inducing the lg s motion lie on a relatively thin sgx - sgz plane . + by comparing the peculiar velocity of the lg as inferred from the cmb dipole , @xmath125 , to that predicted by the x - ray cluster distribution , @xmath126 we can use equation 2 to determine the biasing parameter , @xmath51 , which relates the rbc clusters to the underlying mass distribution which they trace . although the virgo cluster is included in the rbc sample , authors have traditionally removed the infall velocity of the lg toward virgo since previous samples such as xbac - ciza did not include the local supercluster . for a meaningful comparison between our results and those of plionis & kolokotronis ( 1998 ) and kocevski et al . ( 2004 ) we follow the same approach . to obtain @xmath126 , we calculate the median of the dipole amplitude between 240 and 300 @xmath4 mpc , where we assume isotropy has been reached and the amplitude has arrived at its final value . without virgo , the lg and cmb reference frame , number - weighted amplitudes are @xmath127 and @xmath128 km s@xmath8 , respectively , where the errors listed are from the variation of the amplitude over the given range . to remove the lg s virgocentric infall velocity from its peculiar motion we use @xmath129 where @xmath130 is 627 km s@xmath8 , @xmath131 is the angle between the cmb dipole and virgo directions and is roughly @xmath132 and we set @xmath133 , the infall velocity , to the literature average value of 170 km s@xmath8 , as used by plionis and kolokotronis ( 1998 ) and kocevski et al . ( 2004 ) . with these values we obtain a corrected peculiar velocity of @xmath134 km s@xmath8 . dividing this by the dipole amplitudes we obtain the following upper and lower estimates on the @xmath64 parameter : @xmath135 where the errors are again from the variation of the amplitude over the distance of 240 and 300 @xmath4 mpc . performing the same analysis on the mass - weighted amplitude we find @xmath136 and @xmath137 . we have combined the reflex catalog in the southern hemisphere , the ebcs sample in the north , and the ciza survey in the galactic plane to produce the largest , all - sky , truly x - ray selected cluster sample compiled to date and used it to investigate the origin of the lg s peculiar velocity . the x - ray selected nature of the rbc catalog largely does away with projection effects , the poor sampling of nearby clusters , and the incompleteness in the plane of the galaxy introduced as a result of the optical selection methods used to construct previous cluster catalogs . in addition , the rbc s relatively low x - ray flux limit allows us to better sample the underlying density field , which in turn affords us a greater resolution in tracing the structures which give rise to the lg s motion . from the dipole anisotropy present in the cluster distribution we determine that @xmath0 of the lg s peculiar velocity is due to infall into the ga region , while @xmath1 is induced by more distant overdensities between @xmath2 and @xmath3 @xmath4 mpc away . of the large - scale contribution , we find that the ssc has the largest dynamical impact , being responsible for @xmath5 of the increase in the dipole amplitude beyond @xmath2 @xmath4 mpc . despite the significance of the ga and ssc regions , our findings are not consistent with a simple two - attractor model for generating the lg s motion . there exists numerous groupings and loose associations of clusters at roughly the same distance as the ssc that have a significant effect on the dipole amplitude . these include the well - known hr region , as well as the newly noted a3667 and a3391 associations and the c1410 filament which may trace an extension of the ssc complex into the zoa . in addition , we find that the region of sky within @xmath123 of the dipole pointing between @xmath2 and @xmath3 @xmath4 mpc is 2.7 times as dense as the opposite part of the sky . we suggest that the anisotropy in the cluster distribution near @xmath7 @xmath4 mpc has as much to do with the presence of large overdensities such as the ssc and the hr concentrations in the south as it does with the lack of superclusters in the north . we thank brent tully and chris mullis for many valuable discussions and contributions , as well as manolis plionis for his useful suggestions . dk is supported by the nasa graduate student research program . allen , s.w . , schmidt , r.w . & fabian , a.c . 2003 , mnras , 342 , 287 basilakos , s. , & plionis , m. 1998 , mnras , 299 , 637 bhringer , h. , voges , w. , huchra , j.p . , mclean , b. , giacconi , r. , rosati , p. , burg , r. , mader , j. , schuecker , p. , simic , d. , komossa , s. , reiprich , t.h . , retzlaff , j. , & trmper , j. 2000 , apjs , 129 , 435 bhringer , h. , schuecker , p. , guzzo , l. , collins , c.a . , voges , w. , schindler , s. , nuemann , d.m . , cruddace , r.g , de grandi , s. , chincarini , g. , edge , a. , macgillivray , h.t . , & shaver , p. 2001 , a&a , 369 , 826 bhringer , h. , schuecker , p. , guzzo , l. , collins , c.a . , voges , w. , cruddace , r.g , ortiz - gil , a. , chincarini , g. , de grandi , s. , edge , a. , macgillivray , h.t . , nuemann , d.m . , schindler , s. , & shaver , p. 2004 , a&a , 425 , 367 branchini , e. , et al . 1999 , mnras , 308 , 1 branchini , e. , & plionis , m. 1996 , apj , 460 , 569 brunozzi , p. tini , borgani , s. , plionis , m. , & moscardini , l. 1995 , 277 , 1210 courteau , s. , faber , s.m . , dressler , a. , & willick , j.a . 1993 , apj , 412 , l51 colless m. , saglia r.p . , burstein , d. , davies , r.l . , mcmahan , r.k . , & wegner , g. 2001 , mnras , 321 , 277 dale , d.a . , giovanelli , r. , haynes , m.p . , campusano , l.e . , hardy , e. & borgani , s. 1999 , apj , 510 , l11 dickey , j.m . , & lockman , f.j . 1990 , ara&a , 28 , 215 dressler , a. 1988 , apj , 329 , 519 ebeling , h. 1993 , ph.d . thesis , ludwig - maximilians - univ . mnchen ebeling , h. , & wiedenmann , g. 1993 , phys . e , 47 , 704 ebeling , h. , edge , a.c . , bhringer , h. , allen , s.w . , crawford , c.s . , fabian , a.c . , voges , w. , huchra , j.p . 1998 , mnras , 301 , 881 ebeling , h. , edge , a.c . , allen , s.w . , crawford , c.s . , fabian , a.c . , huchra , j.p . 2000 , mnras , 318 , 333 ebeling , h. , mullis , c.r . , & tully r.b . 2002 , apj , 580 , 774 einasto , m. , tago e. , jaaniste , j. , einasto , j. , & andernach , h. 1997 , a&as , 123 , 119 erdogdu , p. , et al . 2005 , mnras , _ in press _ huchra , j.p . , et al . 2005 , _ in prep _ hudson , m.j . 1993 , mnras , 265 , 72 hudson , m.j . 1994 , mnras , 266 , 475 hudson , m.j . , smith , r.j . , lucey , j.r . , schlegel , d.j . , & davies , r.l . 1999 , apj , 512 , l79 jerjen , h. , & tammann , g.a . 1993 , a&a , 276 , 1 kaiser , n. , efstathiou , g. , saunders , w. , ellis , r. , frenk , c. , lawrence , a. , rowan - robinson , m. 1991 , mnras , 252 , 1 kocevski , d.d . , mullis , c.r . , & ebeling , h. 2004 , apj , 608 , 721 kocevski , d.d . , ebeling , h. , mullis , c.r . , tully , r.b . 2006 , _ in press _ ( astro - ph/0512321 ) kogut , a. , et al . 1993 , apj , 419 , 1 knopp , g.p , henry , j.p . , briel , u.g . 1996 , apj , 472,125 kraan - korteweg , r.c . , woudt , p.a . , cayatte , v. , fairall , a.p . , balkowski , c. , & henning , p.a . 1996 , nature , 379 , 519 lahav , o. 1987 , mnras , 225 , 213 lynden - bell , d. , et al . 1988 , apj , 326 , 19 macgillivray , h.t . & stobie , r.s . 1984 , vistas astron . , 27 , 433 markevitch , m. , sarazin , c.l . , vikhlinin , a. , apj , 521 , 526 mathewson , d.s . , ford , v.l . , & buchhorn , m. 1992 , apj , 389 , l5 mattig , w. 1958 , astron . nachr . , 284 , 109 peebles , p.j.e . 1976 , apj , 205 , 318 plionis , m. 1988 , mnras , 234 , 401 plionis , m. , & kolokotronis , v. 1998 , apj , 500 , 7 plionis , m. , & valdarnini , r. 1991 , mnras , 249 , 46 reiprich , t.h . , & bhringer , h. 1999 , astron . nachr . , 320 , 296 rowan - robinson , m. , et al . 2000 , mnras , 314,375 rowan - robinson , m. , et al . 1990 , mnras , 247,1 saunders , w. , et al . 1999 , in _ density and velocity fields from the pscz survey _ , ed courteau , s. & willick , j. ( asp conf . ser . , 201 ; san francisco : astron . soc . pac . ) , 228 saunders , w. , et al . 1995 , in _ wide field spectroscopy and the distant universe _ , ed maddox , s.j . ( singapore : world scientific ) , 88 scaramella , r. , baiesi - pillastrini , g. , chincarini , g. , vettolani , g. , & zamorani , g. 1989 , nature , 338 , 562 scaramella , r. , vettolani , g. , & zamorani , g. 1991 , apj , 376 , l1 shapley , h. 1930 , harvard obs . , 874 , 9 shane , c.d . , & wirtanen , c.a . 1967 , publs lick obs . , xxii , part 1 shaya , e.j . 1984 , apj , 280 , 470 schmoldt , i.m . , 1999 , aj , 118 , 1146 staveley - smith , l. , juraszek , s. , henning , p.a . , koribalski , b.s . , & kraan - korteweg , r.c . 2000 , in _ mapping the hidden universe : the universe behind the milky way - the universe in hi _ , ed kraan - korteweg , r.c . , henning p.a . , & andernach h. ( asp conf . ser . , 218 ; san francisco : astron . soc . pac . ) , 207 strauss , m.a . , davis , m. , yahil , a. , & huchra , j.p . 1992 , apj , 385 , 421 strauss , m. , & davis , m. 1988 , in large - scale motions in the universe , ed . rubin & g.v . coyne ( princeton : princeton univ . press ) , 256 sutherland , w. 1988 , mnras,234 , 159 tonry , j.l . , blakeslee , j.p . , & ajhar , e.a . , apj , 530 , 625 tully , r.b . , scaramella , r. , vettolani , g. , zamorani , g. 1992 , apj , 388 , 9 van haarlem , m.p . , frenck , c.s . , & white , s.d.m . 1997 , mnras , 287 , 637 voges , w. 1992 , proc . satellite symp . 3 , esa isy-3 , 9 voges , w. , et al . 1999 , a&as , 349 , 389 willick , j.a . , courteau , s. , faber , s.m . , burstein , d. , & dekel , a. 1995 , apj , 446 , 12 willick , j.a . , courteau , s. , faber , s.m . , burstein , d. , dekel , a. , & strauss , m. 1997 , apjs , 109 , 333 willick , j.a . 1999 , apj , 522 , 647 zaroubi , s. , hoffman , y. , & dekel , a. 1999 , apj , 520 , 413 zwicky , f. , herzog , e. , wild , p. , karpowicz , m. , & kowal , c.t . 1961 - 68 , catalogue of galaxies and cluster galaxies , vols 1 - 6

the particle number projection method is formulated for density dependent forces and in particular for the finite range gogny force . detailed formula for the projected energy and its gradient are provided . the problems arising from the neglection of any exchange term , which may lead to divergences , are thoroughly discussed and the possible inaccuracies estimated . numerical results for the projection after variation method are shown for the nucleuser and for the projection before variation approach for the nucleicr . we also confirm the coulomb antipairing effect found in mean field theories . , , , effective interactions , particle number projection , hartree - fock - bogoliubov . a=48 , 50 and 164 .

nucl-th0105003

the simplest approach to the many - body system is the mean field approximation ( mfa ) . the most powerful mfa is the hartree fock bogoliubov ( hfb ) theory , where particle - hole and particle - particle correlations are taken into account at the same foot . the success of this simple approach is based on the large variational hilbert space generated by symmetry breaking wave functions . for an atomic nucleus and in the most general hfb approach as many symmetries as possible , in particular the rotational invariance and the particle number symmetries are broken . in these approaches the quantum numbers are only conserved on the average , a semiclassical approach to the full quantum requirement to the wave function of being an eigenstate of the symmetry operators . it is well known @xcite that the plain hfb approach works well in the case of a strongly correlated regime and with a large number of nucleons participating in the symmetry breaking process . in the case of the rotational invariance , the ideal situation is represented by a strongly deformed heavy nucleus where a large number of nucleons contribute to the collective phenomenon of deformation . in the case of the particle number symmetry , in which we are interested in this paper , the situation is far from being satisfactory because only a few pairs of nucleons ( four to five ) in the vicinity of the fermi surface are thought to cause the phenomenon of nuclear superfluidity . thus , it is well known that the treatment of pairing correlations in the mean field approach produces an unrealistic transition from the superfluid to the normal phase along the yrast band , not expected in a finite system . in order to get an improved description of the nucleus it is clear that one has to go beyond the mean field approximation to include correlations . this task can be achieved in a simple and effective way within the particle number projection ( pnp ) @xcite formalism , another possibility is to build correlations by the random - phase - approximation ( rpa ) @xcite . the latter is , however , from the applications point of view , very limited , since the required calculation of the ground state correlations can be performed only for very simple models and/or separable interactions . concerning this point one has to keep in mind that not only the rpa but also other sophisticated theories have been developed only in simple models or with separable forces . thus the first calculation with exact particle number projection before the variation was done about twenty years ago @xcite with the pairing plus quadrupole hamiltonian and the baranger - kumar configuration space . in the last years in an attempt to achieve some progress in calculations with effective forces and large configuration spaces , the lipkin nogami method @xcite has experienced a revival . the great advantage of the lipkin nogami method is that corrects , in an approximate way , for the zero point energy associated with the breakdown of the particle number symmetry at the _ mean field level _ , i.e. the calculations do not get much involved as compared to the full hfb ones . some applications of this method with effective forces have been done with the skyrme force @xcite , the gogny force@xcite and the relativistic mean field approach @xcite . the lipkin nogami method on the other hand is just a recipe and can not be obtained from a variational principle . it is therefore desirable to develop an exact particle number projection which can be applied to effective forces . the purpose of this paper is to develop an exact particle number projection before ( and after ) the variation for density dependent forces and in particular for the gogny force . as we shall see , the exchange terms play an important role in pnp , in fact the neglection of any exchange term may lead to divergences . as it is well known , in general , a zero range force is unable to provide good pairing properties . as a matter of fact , normally the pairing terms of the force are neglected in order to get reasonable hartree fock results . to perform hfb calculations a pairing force has been usually added ad hoc . the finite range of the density dependent gogny force , on the other hand , provides good pairing properties without neglection of any exchange term . in fact , the force was designed with this intention@xcite and this property makes the gogny force one of the few effective density dependent forces for which particle number projection is feasible without problem due to divergences . all the skyrme parametrizations ( with the few exceptions of those which provide reasonable pairing correlations ) and all relativistic mean field approaches may give in general divergent contributions to the projected energy . in section [ sec : gpt ] , we introduce the particle number projection method for non - separable density dependent forces . in section [ sec : ddp ] we present the formalism for the density dependent part of the interaction and discuss some alternative prescriptions to this dependence . the gradient of the projected energy to be implemented in the calculations is developed in section [ sec : vap ] . in section [ sec : convl ] , we present results for the projection after variation ( pav ) method for the nucleus @xmath0er and discuss the problems associated with the divergent terms . results for the variation after projection ( vap ) method for the nuclei @xmath1cr are discussed in section [ sec : vapr ] . the particle number projected hartree fock bogoliubov theory is the simplest symmetry conserving mean field approach which takes into account particle - particle correlations . this theory has been derived in the past for non - density dependent interactions and a detailed formulation has only been given for separable forces like the pairing plus quadrupole model @xcite . in order to generalize this theory to non - separable and density dependent forces , like the gogny or the skyrme force , and to introduce the pertinent notation we shall also summarize the known formulation . let @xmath2 be a product wave function of the hfb type , i.e , the vacuum to the quasiparticle operators given by the bogoliubov transformation @xmath3 where @xmath4 are the particle annihilation and creation operators ( usually in the harmonic oscillator basis ) . this transformation does not conserve the particle number symmetry and the wave function @xmath2 is not an eigenstate of the particle number operator @xmath5 . one can generate , however , out of @xmath2 , an eigenstate , @xmath6 , of @xmath5 by the well known projection technique @xcite @xmath7 where the integration interval can be reduced to @xmath8 $ ] whenever the intrinsic wave function has a well defined `` number parity '' @xcite quantum number . furthermore , the integral can be discretized in a sum using the fomenko expansion @xcite @xmath9 where @xmath10 is the number of points used in the calculations , for @xmath11 one gets the exact solution , usually @xmath12 is used in the calculations . @xmath13 projects only on one system of particles , in the general case , however , one has to perform a double projection , for the number of neutrons and protons . in this case the projected matrix element of any operator @xmath14 , such that @xmath15=0 $ ] , is given by @xcite @xmath16 where we have introduced @xmath17 and the @xmath14-operator overlap @xmath18 with @xmath19 . another useful quantity is the norm overlap , which is given by @xmath20 clearly , @xmath21 and @xmath22 are related by @xmath23 , with @xmath24 . as it will be illustrated for the hamiltonian operator , any operator overlap can be calculated using the generalized wick theorem @xcite . this theorem allows to express matrix elements of the form @xmath25 , exactly in the same way as the ordinary wick theorem , as the sum of all possible contracted products , where @xmath26 and @xmath27 are quasiparticle vacua . thus , in particular @xmath28 \label{extwick}\end{aligned}\ ] ] looking at eq . ( [ pme ] ) we see that this theorem allows to express projected expectation values in terms of mean field matrix elements . in eq . ( [ extwick ] ) the first term is the hartree term , the second one the fock term and the last one the bogoliubov term . > from this expression we see that problems may appear when @xmath29 . this point will be discussed later on and its consequences for the effective force will be analyzed in appendix [ appen : poles ] . for the pnp method , the basic building blocks of wick s theorem are the @xmath30-dependent generalized density matrix and pairing tensors @xmath31 where we have used the notation @xmath32 , the indices @xmath33 belonging to the same isospin channel @xmath34 . the matrices @xmath35 and @xmath36 are related to @xmath37 and @xmath38 of eq . ( [ bogt ] ) by @xmath39 and the contraction @xmath40 is given by @xmath41 here again the indices @xmath42 belong to the same isospin channel and @xmath43 with @xmath44 and @xmath45 . the matrices @xmath46 and @xmath47 represent the number operator in the quasiparticle basis of eq.([bogt ] ) . the nuclear hamiltonian can be written as @xmath48 where the density independent part , @xmath49 , represents the nuclear interaction ( coulomb included ) with the exception of a possible density - dependent term ( the well known density dependent term of the skyrme and gogny forces ) which is represented by @xmath50 . in our case @xmath50 is given by the last term of eq . ( [ eq : vgog ] ) . with the definitions above we can formulate the particle number projected theory for non separable interactions , like the gogny force ( appendix [ appen : gogf ] ) , in a similar way as in the mean field approximation . using eq . ( [ pme ] ) and the generalized wick theorem , the density independent part of the projected energy is given by @xmath51 where the traces run over the configuration space , here assumed to be the same for protons and neutrons , and @xmath52 in @xmath53 the indices @xmath54 belong to the isospin @xmath34 and @xmath55 to @xmath56 , in @xmath57 all indices belong to the isospin @xmath34 does not mix isospin . ] . since @xmath58 , we can write the proton or neutron projected density as @xmath59 with @xmath60 belonging to the isospin @xmath34 , and the projected energy as @xmath61 where @xmath62 and @xmath63 . by @xmath64 in @xmath65 and in some notation to be introduced later on ( see also eq . [ eq : rhop ] ) we want to indicate that these quantities are @xmath66 independent , i.e. , the integration on @xmath66 has been performed . in these fields , the indices @xmath67 belong to the isospin @xmath34 . the expressions ( [ eq : rhop ] ) and ( [ eq : epdi ] ) can be rewritten taking into account that @xmath68 in this way we can simplify the sum in eq . ( [ eq : rhop ] ) for the projected density to ) from @xmath69 to @xmath10 can be written as a sum from @xmath70 to @xmath71 . ] @xmath72 } { \mathcal c}_{l_{\tau } } x_{l_{\tau } } { \rho}^{\tau}({\varphi}_{l_{\tau } } ) \right \ } \ , , \end{aligned}\ ] ] where we have taken into account that @xmath73 . by @xmath74 $ ] we represent the integer part of @xmath75 and @xmath76 in the same way , for the projected energy , eq . ( [ eq : epdi ] ) , we obtain @xmath77 } x_{l_{\tau } } \frac{\,{\mathcal c}_{l_{\tau}}\,}{2 } \bigl ( { \gamma}^{\tau \tau } ( { \varphi}_{l_{\tau } } ) \rho ( { \varphi}_{l_{\tau } } ) - { \delta}^{10,\tau}({\varphi}_{l_{\tau}}){\kappa}^{01}({\varphi}_{l_{\tau } } ) \bigr ) \biggr \}. \label{eq : ep2}\end{aligned}\ ] ] density dependent interactions have a term which simulates a g - matrix through an explicit dependence of the nuclear density . this term , @xmath78 , was introduced in the mean field approximation . in the mfa only _ diagonal _ matrix elements between product wave functions are needed to calculate the energy and , consequently , @xmath78 is constructed to depend on the mean field density . in theories beyond mean field , the density dependent part is given by @xmath79 | \psi_n \rangle } { \langle \psi_n | \psi_n \rangle } \nonumber \\ & = & \frac { { \displaystyle \int } d { \varphi}_{\pi } { \displaystyle \int } d { \varphi}_{\nu } \langle \phi | { \hat{v}}_{dd } \left [ \overline{\rho } ( \vec{r } ) \right ] e^{i{\varphi}_{\pi } { \hat{n}}_{\pi } } e^{i{\varphi}_{\nu } { \hat{n}}_{\nu } } | \phi \rangle } { { \displaystyle \int } d { \varphi}_{\pi } { \displaystyle \int } d { \varphi}_{\nu } \langle \phi | e^{i{\varphi}_{\pi } { \hat{n}}_{\pi } } e^{i{\varphi}_{\nu } { \hat{n}}_{\nu } } | \phi \rangle } \label{eq : epdd}\end{aligned}\ ] ] where @xmath80 $ ] indicates the explicit dependence of @xmath78 on a density @xmath81 to be specified . looking at these expressions it is not obvious which dependence should be used . there are two more or less straightforward prescriptions @xcite for @xmath81 : 1 . the first prescription is inspired by the following consideration : in the mfa , the energy is given by @xmath82 and @xmath78 is assumed to depend on the density @xmath83 . on the other hand , if the wave function which describes the nuclear system is the projected wave function @xmath84 , we have to calculate the matrix element @xmath85 ( see the middle term in eq . ( [ eq : epdd ] ) ) . it seems reasonable , therefore , to use in @xmath78 the density @xmath86 , i.e. the projected density . 2 . the other prescription has been guided by the choice usually done in the generator coordinate method with density dependent forces @xcite . the philosophy behind this prescription is the following : to evaluate eq . ( [ eq : epdd ] ) we have to calculate matrix elements between different product wave functions @xmath2 and @xmath87 ( @xmath88 ) ( see last term in eq . ( [ eq : epdd ] ) ) . then , to calculate matrix elements of the form @xmath89 we choose the mixed density @xmath90 to be used in @xmath91 . we shall call this approach the mixed density prescription . both prescriptions have been tested with the gogny force in the lipkin nogami approach @xcite and practically no difference was found in the numerical applications . one should notice that in the second prescription @xmath92 depends on the angle @xmath30 at variance with prescription 1 . we shall now proceed to evaluate the projected energy with both prescriptions . using the projected density prescription , the projected energy of the density dependent term ( @xmath93 ) is obtained in a simple way . one just has to use in @xmath91 the projected density . in the case of the gogny force ( see appendix [ appen : gogf ] ) which we are considering here , the value @xmath94 for the spin exchange parameter guarantees that there is no contribution from @xmath91 neither to the pairing field @xmath95 nor to the hartree - fock field of the type @xmath96 , i.e. , without isospin mixing . we find @xmath97 where @xmath98 \right ) } _ { k_1k_2k_3k_4 } { \rho}_{k_4 k_2}^{p\tau'}\ ; , \end{aligned}\ ] ] with @xmath99 and @xmath100 \right ) } _ { k_1k_2k_3k_4 } = \langle k_1k_2 | v_{dd } [ { \rho}^p ] |k_3k_4 \rangle - \langle k_1k_2 | v_{dd } [ { \rho}^p ] |k_4k_3 \rangle \ ; , \end{aligned}\ ] ] the symbol @xmath101 is used to specify that we are dealing with the dd part of the interaction . we assume in this section that @xmath78 depends on the mixed density @xmath102 the contribution of @xmath103 to the energy is given by eq . ( [ eq : epdd ] ) with @xmath80= { \rho}_{{\varphi}_{\pi}{\varphi}_{\nu } } ( \vec{r } ) $ ] . since the mixed density depends on the angles , we can not simplify the double integral to a single one as we did before . replacing the integrals by sums and using ( [ eq : descsum ] ) again , we obtain @xmath104 } y_{l_{\tau } } \sum_{l_{\tau'}=0}^{[l/2 ] } y_{l_{\tau ' } } { \mathcal c}_{l_\tau l_{\tau ' } } \cdot { \displaystyle tr } \bigl ( { \breve{\gamma}}^{\tau \tau ' } \bigl [ \rho({\varphi}_{l_{\tau } } ) , \rho({\varphi}_{l_{\tau ' } } ) \bigr ] \cdot \rho ( { \varphi}_{l_{\tau } } ) \bigr ) \right ) \nonumber \right . \\ & + & \hspace{-0.1 cm } \left . 2\cdot { \displaystyle re } \left ( \sum_{l_{\tau}=1}^{[l/2]-1 } y_{l_{\tau } } \sum_{l_{\tau'}=1}^{[l/2]-1 } y^*_{l_{\tau ' } } { \displaystyle tr } \bigl ( { \breve{\gamma}}^{\tau \tau ' } \bigl [ \rho({\varphi}_{l_{\tau } } ) , \rho^*({\varphi}_{l_{\tau ' } } ) \bigr ] \cdot \rho ( { \varphi}_{l_{\tau } } ) \bigr ) \right ) \right \ } \label{eq : epdd2}\end{aligned}\ ] ] where @xmath74 $ ] represents , as before , the integer part of @xmath75 and @xmath105 in this expression we have furthermore introduced the following notation @xmath106 & = & \sum_{k_4 k_2 } { \left ( \bar{v}_{dd } \left [ \rho({\varphi}_{l_{\tau}}),\rho({\varphi}_{l_{\tau ' } } ) \right ] \right ) } _ { k_1 k_2 k_3 k_4 } \rho_{k_4 k_2 } ( { \varphi}_{l_{\tau ' } } ) \nonumber \\ { \breve{\gamma}}^{\tau \tau'}_{k_1 k_3 } \bigl [ \rho({\varphi}_{l_{\tau } } ) , \rho^*({\varphi}_{l_{\tau ' } } ) \bigr ] & = & \sum_{k_4 k_2 } { \left ( \bar{v}_{dd } \left [ \rho({\varphi}_{l_{\tau}}),\rho^*({\varphi}_{l_{\tau ' } } ) \right ] \right ) } _ { k_1 k_2 k_3 k_4 } \rho^*_{k_4 k_2 } ( { \varphi}_{l_{\tau ' } } ) . \label{eq : gammatil } \end{aligned}\ ] ] here again the indices @xmath54 belong to the isospin @xmath34 and @xmath55 to @xmath56 @xmath107 and @xmath108 \right ) } _ { k_1 k_2 k_3 k_4 } $ ] is the density dependent matrix element calculated with the density @xmath109 in @xmath78 , similarly , @xmath110 \right ) } $ ] is calculated with @xmath111 in @xmath78 . if we compare the expression ( [ eq : epdd2 ] ) with the result obtained for @xmath112 with the projected prescription ( [ eq : epdd1 ] ) , it is easy to observe that ( [ eq : epdd2 ] ) implies a more elaborated calculation . with the projected prescription , we only have to calculate the projected density @xmath113 , the field @xmath114 $ ] and then one easily gets @xmath93 . but in order to evaluate expression ( [ eq : epdd2 ] ) it is necessary to calculate @xmath115 different fields . for example , for @xmath12 , we have to calculate @xmath116 fields . using equation ( [ eq : ep2 ] ) , we can perform a restoration of the particle number symmetry , projecting after the variational hartree fock bogoliubov equations have been solved . in this case the hfb wave function is not self consistently determined . in particular , if the hfb calculation collapses to the pairing uncorrelated hf one , the pnp method does not provide a better approximation . in this section we shall derive the projected variational equations in order to solve the self consistent problem , that means , the variation after the projection method . according to the ritz variational principle , the intrinsic wave function @xmath117 has to be determined by minimization of the projected energy , i.e. , @xmath118 the variational hilbert space generated by product wave functions can be parametrized by the thouless theorem @xcite , which allows to write any product wave function , @xmath2 , in the form @xmath119 where @xmath120 is a reference product wave function non orthogonal to @xmath2 and @xmath121 are the variational parameters . we shall first study the variation of the density independent part of the interaction , eq . ( [ eq : ep2 ] ) . an arbitrary variation of the energy is given by @xmath122 where we have introduced the projected gradient @xmath123 . here the indices @xmath124 run over the proton and neutron configuration space . the solution of the projected variational problem can be obtained by the conjugate gradient method @xcite . using the generalized wick theorem @xcite , @xmath123 can be written as @xmath125 with the index @xmath34 corresponding to the isospin of the quasiparticle operators @xmath126 . we have furthermore introduced @xmath127 and @xmath128 with @xmath129 again @xmath99 , and @xmath130 where @xmath131 taking into account eq . ( [ eq : descsum ] ) , we can simplify the gradient expression to the region @xmath132 $ ] in the following way : @xmath133 } c_{l_{\tau } } x_{l_{\tau } } \left\ { { \mathcal h}^{20}_{\mu \nu,\tau } ( { \varphi}_{l_{\tau } } ) + { \mathcal e}^{\tau}({\varphi}_{l_{\tau } } ) \left ( { \mathcal x}_{\mu \nu}({\varphi}_{l_{\tau } } ) - { \mathcal x}^{p\tau}_{\mu \nu } \right ) \right\}. \label{eq : h20psindd}\end{aligned}\ ] ] these expressions apply to the non - density dependent part of the interaction . to calculate the gradient of the density dependent part of the interaction we proceed as with the evaluation of the energy , accordingly we shall distinguish the two prescriptions adopted . in a similar way as in the case of the density independent part of the interaction , the variation of @xmath134 is defined by @xmath135 . here the expression of @xmath136 is somewhat more complicated due to the explicit dependence on the density of the interaction : @xmath137 using eq . ( [ rhodef ] ) one can calculate @xmath138 very easily : @xmath139 using the notation @xmath140 for the rearrangement term @xmath141 where the indices @xmath60 and @xmath142 in eqs . ( [ rhoptau],[rearr ] ) belong to the isospin @xmath34 , we can write the projected gradient of @xmath143 in the following way : @xmath144 } c_{l_{\tau } } x_{l_{\tau } } \biggl \ { \breve{{\mathcal h}}^{20}_{\mu \nu,\tau } ( { \varphi}_{l_{\tau } } ) + { \breve{\mathcal e}}^{\tau}({\varphi}_{l_{\tau } } ) \bigl ( { \mathcal x}_{\mu \nu}({\varphi}_{l_{\tau } } ) - { \mathcal x}^{p\tau}_{\mu \nu } \bigr ) \biggr \ } \label{eq : h20pcondd}\end{aligned}\ ] ] where @xmath145 and @xmath146 this is a similar expression to the projected gradient of the non density dependent part , ( [ eq : h20psindd ] ) , they differ only by the term @xmath147 , which does not contribute in this case because the specific value of the coefficient @xmath148 in the gogny force makes it zero . we shall now derive the gradient of the density dependent term with the second prescription . the projected gradient is given by an expression similar to eq . ( [ h20pdd ] ) but with @xmath78 depending on the mixed density , eq . ( [ mixdens ] ) , and the rearrangement term given by @xmath149 } { \partial { \rho}_{{\varphi}_{\pi}{\varphi}_{\nu } } ( \vec{r } ) } } f_{kk'}(\vec{r } ) e^{i{\varphi}_{\pi } { \hat{n}}_{\pi } } e^{i{\varphi}_{\nu } { \hat{n}}_{\nu } } | \phi \rangle } { \langle \phi | e^{i{\varphi}_{\pi } { \hat{n}}_{\pi } } e^{i{\varphi}_{\nu } { \hat{n}}_{\nu } } | \phi \rangle } .\end{aligned}\ ] ] as before , we obtain @xmath150 } \sum_{l_{\tau'}=0}^{[l/2 ] } { \mathcal c}_{l_{\tau}l_{\tau ' } } x_{l_{\tau } } x_{l_{\tau ' } } \left ( \breve{\mathcal h}^{20}_{\mu \nu , \tau } ( { \varphi}_{l_{\tau } } , { \varphi}_{l_{\tau ' } } ) + { \mathcal x}_{\mu \nu}({\varphi}_{l_{\tau } } ) e^p_{dd } ( { \varphi}_{l_{\tau}},{\varphi}_{l_{\tau ' } } ) \right ) + \nonumber \\ & + & 2 \ , \sum_{l_{\tau}=1}^{[l/2]-1 } \sum_{l_{\tau'}=1}^{[l/2]-1 } x_{l_{\tau } } x^*_{l_{\tau ' } } \left ( \breve{\mathcal h}^{20}_{\mu \nu , \tau } ( { \varphi}_{l_{\tau } } , { \varphi}_{l_{\tau'}}^ * ) + { \mathcal x}_{\mu \nu}({\varphi}_{l_{\tau } } ) e^p_{dd } ( { \varphi}_{l_{\tau}},{\varphi}_{l_{\tau'}}^ * ) \right ) \biggr \ } \nonumber \\ & - & e^p_{dd } { \mathcal x}^{p\tau}_{\mu \nu } \label{h20pmd}\end{aligned}\ ] ] where @xmath151 @xmath152 and @xmath153 in eq . ( [ h20pmd ] ) are given by the corresponding expressions in eq . ( [ hhdd ] ) but with the replacement @xmath154 . in this section we discuss some problems that may arise in pnp calculations as well as some theoretical results . as we have seen in eq . ( [ extwick ] ) , the extended wick theorem allows to factorize matrix elements of the two - body part of the interaction in terms of the generalized matrix density and pairing tensors . the matrix element of eq . ( [ extwick ] ) takes a simple form in the canonical basis . using eqs . ( [ eq : rhophi]-[eq : kap0phi ] ) , we obtain @xmath155 \nonumber \\ & = & \langle \phi | \tilde{\phi } \rangle \left [ \frac{v_{k_1}^2 e^{2{{\mathrm{i}}}\varphi}}{u_{k_1}^2 + v_{k_1}^2 e^{2 { { \mathrm{i}}}\varphi } } \cdot \frac{v_{k_2}^2 e^{2{{\mathrm{i}}}\varphi}}{u_{k_2}^2 + v_{k_2}^2 e^{2 { { \mathrm{i}}}\varphi } } \left ( \delta_{k_3k_1 } \delta_{k_2k_4 } - \delta_{k_4k_1 } \delta_{k_3k_2 } \right ) \right . \nonumber \\ & + & \left . \frac{u_{k_1 } v_{k_1}}{u_{k_1}^2 + v_{k_1}^2 e^{2 { { \mathrm{i}}}\varphi } } \cdot \frac{u_{k_3 } v_{k_3 } e^{2{{\mathrm{i}}}\varphi}}{u_{k_3}^2 + v_{k_3}^2 e^{2 { { \mathrm{i}}}\varphi } } \delta_{k_2 \bar{k}_1 } \delta_{k_4 \bar{k_3 } } \right ] \label{eq : overlap1}\end{aligned}\ ] ] the dangerous cases are for @xmath156 and @xmath157 when the denominators become zero . taking into account that @xmath158 , it is easy to see that as long as there is only one pole or two poles but @xmath159 ( or @xmath160 ) there are no problems because they will be cancel out by the multiplying overlap @xmath161 . the only case where problems arise is with the matrix elements of the form @xmath162 , because in this case no cancellation with the norm overlap @xmath161 is possible . it is easy to check , however , that in this case , though each term of eq.([eq : overlap1 ] ) by itself is divergent , their sum gives a finite contribution , namely @xmath163 , which behaves properly when @xmath164 and @xmath156 . that means , as long as all three terms of the wick factorization of eq . ( [ extwick ] ) , i.e. , the hartree , the fock and the bogoliubov term , are taken into account no divergences appear in the pnp formalism . notice that the arguments given here are quite general and independent of the kind of force ( coulomb force included ) used in the calculations -see appendix [ appen : poles ] for a detailed discussion . this point has already been noticed in ref . obviously , one body operators will never cause problems to the pnp . in the case of the density dependent forces , one must be aware that this dependence itself may cause problems . as we have seen before , two different prescriptions have mainly been used to treat the density dependence of the hamiltonian . as discussed in appendix [ appen : poles ] , the projected density prescription does not have divergences and the mixed density prescription has an integrable divergence . for the moment , we use the projected density prescription ( pav1 ) , at the end of this section we show results with the mixed density prescription ( pav2 ) . unfortunately most of the calculations performed so far with effective forces neglect exchange terms . some of them because different forces are used in the particle - hole and the particle - particle channel ( like most of the skyrme force parametrizations and the relativistic mean field calculations ) . the contributions to the p - p interaction from the force used in the p - h channel being neglected and vice - versa . others , though with the same force in both channels , like the gogny force , neglect some exchange terms ( coulomb among others ) just to save cpu time . only recently hfb calculations including _ all terms _ in a triaxial basis has been performed with the gogny force , see @xcite . obviously , the most interesting issue to address is to investigate how large the divergences are in the case that some exchange terms are neglected . though this point will ultimately depend on the particular force used in the calculation , we shall illustrate it here for the gogny force where we are able to perform the usual calculations which neglect some exchange terms and the exact ones . to be precise , we present two types of calculations , the first one is the usual one used in calculations with this force in a triaxial basis @xcite , namely , the pairing field is calculated _ only _ from the brink boeker term , i.e. all additional pairing exchange terms are neglected and the fock term of the coulomb potential is either neglected or calculated in an approximate way . we shall call this approximation @xmath165 , @xmath166 for standard calculation -see ref . @xcite for further details . the wave function solution of the selfconsistent hfb equations with this approximate hamiltonian will be denoted @xmath167 . in the second calculation , the exact hfb equations ( @xmath168 ) with the gogny force are solved , i.e. , _ all _ exchange terms are considered in the variational principle -see @xcite for further details . the wave function resulting from the selfconsistent solution of the @xmath168 equations will be denoted @xmath169 . in a later stage the intrinsic wave functions obtained in these ways are projected onto good particle number in order to calculate the projected energy . in the calculations , we have expanded the quasiparticle operators in a triaxial harmonic oscillator basis . the configuration space was chosen by the condition @xmath170 where the coefficients @xmath171 depend on the relations between the axes @xmath172 and @xmath173 in the form @xmath174 , @xmath175 and @xmath176 . this basis has been symmetrized with respect to the simplex operation and is an eigenstate of the operator @xmath177 , with @xmath178 , @xmath179 is the time reversal operator , and @xmath180 the parity operator @xcite . in the calculations we take @xmath181 . to take into account the rotational motion we add to the hamiltonian the cranking term @xmath182 , as in ref . @xcite . the most stringent test one can perform is to study the yrast band of a nucleus , for example , since the pairing correlations and the quasiparticle occupancies @xmath183 strongly depend on the angular momentum . to separate the different issues we shall first study only the convergence of the projected energy in terms of the parameter @xmath10 of the fomenko expansion . in order not to mix this issue with the divergence problem associated with the neglection of the exchange terms , we shall solve first the exact problem , i.e. the hfb@xmath184 case . to illustrate more clearly our results we shall perform particle number projection after the variation ( pav ) , i.e. , first the cranked hfb@xmath184 equations with the gogny force are solved , as in ref . this provides us with the wave function @xmath169 . then a projection on the particle number ( pav@xmath184 ) is performed onto the wave function @xmath169 ( for different @xmath10 values ) to calculate the projected energy . in fig . [ fig : convlex ] , we display the results of the transition gamma ray energy in the pav@xmath184 approach for different @xmath10-values . we find that the points for the @xmath10-dependent calculation are on top of each other , which indicates a clean convergence and that @xmath12 is already a good approximation to the exact pnp . this value is in agreement with earlier calculations @xcite . we find , by the way , a good agreement with the experiment @xcite up to @xmath185 and a considerable improvement to the hfb@xmath184 approach ( dashed line ) . for higher spin values , the agreement with the experiment is not as good because the proton pairing energy has already collapsed to zero . we now turn to the problem of the divergences in calculations where the exchange terms are neglected . as we have seen in eq . ( [ eq : overlap1 ] ) and in appendix [ appen : poles ] , the poles appear whenever @xmath186 and @xmath187 . the first condition , obviously , can not be eluded . in the non - selfconsistent pav method because the occupancies @xmath188 of the wave function are already determined by the solution of the hfb equations and in the self - consistent vap because it would be against the variational principle . to investigate the behavior of the occupancies @xmath188 , we have solved the selfconsistent cranked hfb@xmath189 equations along the yrast band for the nucleus . the wave functions @xmath167 determined in this way have been analyzed in the canonical basis . in fig . [ fig : uk2vk2 ] , we show the smallest values of the quantity @xmath190 , for protons and neutrons , found in @xmath167 , as a function of the angular momentum . at small spins and up to spin 8 @xmath191 , the quantity @xmath190 for neutrons is rather small ( around @xmath192 ) , at spin 10 @xmath191 it reaches the critical value @xmath193 and from this point on it steadily increases up to the maximal value at a spin value of 28 @xmath191 . this behavior is typical of the high spin regime . at small angular momentum , the system is well paired and we may find a pair of nucleons close to the fermi surface with @xmath194 . as the spin increases the pairs align ( states start being predominantly occupied or empty ) and at high spins , when the pairing collapse takes place , the states become either fully occupied or empty . for protons the situation is even worse , we find two critical points , one around @xmath195 and the other around @xmath196 . the second condition for having the pole , i.e. that @xmath187 , is in principle avoidable , since @xmath197 , taking l odd in the fomenko expansion should be enough . this recipe , as we shall see below , helps in some cases but it does not guarantees that one is error free . the reason is very easy , to find convergence in the fomenko expansion one must take l large enough , but then one may get too close to the critical @xmath198 . that means in this way one avoids to get nonsense but the results may not be reliable . as we have seen above and in appendix [ appen : poles ] , in the neighborhood of the poles we will get separately big contributions to the hartree - fock energy and to the pairing energy . if all exchange terms have been taken into account in the calculations , the sum of the divergent terms provides a finite value . this cancellation , however , is not guaranteed if some exchange terms have been neglected , as in the @xmath165 approach , which we now discuss . in table [ tab : convl ] , we show for the nucleus the differences in the pairing , eq . ( [ paienergy ] ) , hartree - fock , eq . ( [ hfenergy ] ) , and total , eq . ( [ eq : ep ] ) , projected energies calculated for @xmath12 and @xmath199 ( column labeled l - even ) and for @xmath200 and @xmath201 ( column labeled l - odd ) as a function of the angular momentum . in the even - l calculations in the pav@xmath189 approach , we find large @xmath202 values in the pairing and the hf energies around @xmath203 and @xmath204 . these deviations do not cancel each other and give rise to large @xmath205 , indicating that no convergence is reached . for odd - l values the situation is much better . the largest @xmath205 is around 20 kev which is more than one order of magnitude smaller than in the even - l case . lastly , in the last column the difference of the total energy between @xmath12 and @xmath200 is displayed . we clearly observe that again large values around @xmath203 and @xmath204 are obtained . interestingly , a comparison with fig . [ fig : uk2vk2 ] shows that these spin values are those where @xmath206 is very small , i.e. the poles . the neutron system is responsible for the increase at @xmath207 , see second column , and the proton one around @xmath208 and @xmath204 . .differences ( in kev ) between the projected energies calculated with @xmath200 ( @xmath209 ) and @xmath201 ( @xmath210 ) and @xmath12 ( @xmath211 ) and @xmath199 ( @xmath212 ) in the fomenko expansion for the nucleus , calculated in the pav@xmath189 approach . the last column displays the difference between the total energy calculated with @xmath12 and @xmath200 . [ cols="^,^,^,^,^,^,^,^ " , ] we shall now discuss the convergence of our results with respect to the density dependence of the hamiltonian used in the calculations . in order to isolate this issue , we do not neglect any exchange terms in the calculations , that means we solve the hfb@xmath184 equations . this provides us with the wave function @xmath169 which we use to calculate the projected energy , which can calculated according to two prescriptions . notice that @xmath169 is different from @xmath167 and that the levels occupancies of both wave functions are not necessarily the same . as it is shown in section [ sect : approden ] the prescription using the projected density , which we shall label dd1 , was divergence free , while the second prescription , the mixed density one ( section [ sect : apmixden ] ) , from now on called dd2 , was showing an integrable divergence . as in table [ tab : convl ] we shall calculate separately the pairing and the hartree - fock contributions to the total energy . since the density dependent term does not contribute to the pairing energy ( the parameter @xmath148 in the gogny force is equal to @xmath213 ) this energy is independent of the prescription used for the density dependent term . in table [ tab : convlex ] we display the differences @xmath214 of projected energies calculated with the pairing energy , eq . ( [ paienergy ] ) , the hf energy , eq . ( [ hfenergy ] ) and the sum of both , the total energy . at low and medium spins , the difference of pairing energies for @xmath12 and 20 is not equal to zero indicating the known divergences . we find the largest value for @xmath215 indicating that the wave function @xmath169 has a level occupancy at this spin value such that @xmath216 . inspection of the canonical basis indicates that the orbital is a neutron one , the proton system in this case does not have orbitals with @xmath188 close to 1/2 . at high spins , the wave function @xmath169 is not paired in the proton system ( see ref . @xcite ) and in the neutron system it does not have orbitals with @xmath188 close to 1/2 . to get convergence , either with the dd1 or the dd2 prescription , the difference of the total energy calculated with different l values , must be zero . this is the case in the dd1 approach where , even with even - l values , the hf energy differences are almost exactly the same as the pairing ones . as a result , both cancel out and the total energy is convergent , as expected . with the dd2 prescription and even - l values , however , the hf results are different from the pairing ones and as a result no convergence for the total energy is found . we know , however , that the dd2 divergence is integrable , that means if the integral is carefully performed the convergence must be found . this is what happens if we do the calculations with odd - l values ( see last column ) where the largest deviation found is 3 kev . obviously , more sophisticated integration methods will provide better convergence . an interesting point concerning the density prescriptions dd1 and dd2 is to know if there is a big difference in the results calculated with dd1 or dd2 . in fig . [ fig : dd1dd2 ] we show the @xmath217-ray energies along the yrast line of the nucleus @xmath0er in the hfb@xmath218 with the dd1 ( in the figure pav1 ) and with dd2 ( in the figure pav2 ) . as we can see the behavior is very similar and the physics is clearly the same . a similar conclusion was found in @xcite in the lipkin - nogami approach . in ref . @xcite and @xcite a good agreement between the hfb approximation , the exact shell model diagonalization and the experimental results was found for the rotational states of the nuclei @xmath219cr and @xmath220cr . only the @xmath217-ray energies along the yrast bands in the hfb approach were systematically around 0.5 mev smaller than experimentally observed . a simulation with the exact shell model diagonalization in @xcite showed that the shift was due to a deficient treatment of the pairing correlations of the hfb method in the weak pairing regime . it is well known that the variation after projection method is able to provide pairing correlations even at the limiting case when the hfb approach collapses to the hf one . in this section we shall investigate the mentioned nuclei in the vap method . the hfb calculations of refs . @xcite and @xcite were performed in the standard approximation , i.e. hfb@xmath189 . for nuclei like these with little pairing -low level density at the fermi surface- is very unlikely to find orbitals with @xmath221 , nevertheless we have checked during the minimization process of the vap@xmath189 that no orbitals were populated in that way . additionally , we have performed also the hfb@xmath184 and the vap@xmath184 calculations . in fig . [ fig : paircrom_s ] , we represent the hfb@xmath189 and vap@xmath189 ( see eq.([paienergy ] ) ) pairing energies for the nuclei @xmath219cr , left hand side , and @xmath220cr , right hand side . in the hfb@xmath189 approach the pairing energies are rather small , for @xmath219cr practically zero for all spin values , with the exception of the @xmath222 and @xmath223 values where they are very small . for @xmath220cr they are also very small and only for @xmath224 and @xmath225 differ from zero . in both nuclei we find the coriolis antipairing effect that leads to a sharp transition from a ( weakly ) superfluid system to a normal one . the vap@xmath189 solutions , on the other hand , are well paired and no sharp transition is seen , only a smooth decrease in the correlation energy is found . in fig . [ fig : gammacrom_s ] , the @xmath226-ray energies along the yrast band are displayed in the hfb@xmath189 and vap@xmath189 approaches together with the experimental results for @xmath219cr , @xcite and @xmath220cr , @xcite . as discussed in @xcite , the hfb@xmath189 @xmath226-ray energies for @xmath219cr show the same trend as the experimental ones but are about 0.5 mev smaller . the vap@xmath189 approach , on the other hand , due to the larger pairing energy leads to smaller moment of inertia improving the agreement with the experiment . the vap@xmath189 results come closer to the experiment and the backbending is slightly better described . for @xmath220cr the situation is similar , the vap@xmath189 approach improves considerably the hfb@xmath189 one specially at low and medium spins . one may ask , however , why we do not get enough pairing correlations in the vap@xmath189 method as to match the experimental results . there are several possible answers , first in our calculations we do not include p - n pairing which is known @xcite to play an important role in these nuclei . second , the fact that in these nuclei , in the hfb approximation , there are barely any pairing correlations indicates that the potential energy curve is rather flat against pairing fluctuations . this kind of fluctuations is not included in the pnp method . in figs . [ fig : paircrom_e ] and [ fig : gammacrom_e ] , we finally present the results of the exact hfb calculations , hfb@xmath184 , and the corresponding vap@xmath184 method . the pairing correlation energies are represented in fig . [ fig : paircrom_e ] . as expected @xcite , the neglected pairing terms in the hfb@xmath189 approach have an antipairing effect , mainly the coulomb term . the small pairing energy of the hfb@xmath189 approach is quenched in the hfb@xmath184 , in particular the proton pairing energies completely vanish . the pairing energies in the vap@xmath184 approach display this effect even more clearly . the total vap@xmath184 pairing energy is considerably reduced with respect to the vap@xmath189 approach . furthermore , while in the vap@xmath189 the proton pairing energy is always larger than the neutron one in the vap@xmath184 the proton pairing energy becomes always smaller than the neutron one . in fig . [ fig : gammacrom_e ] we finally present the @xmath217-ray energies in the hfb@xmath184 and the vap@xmath184 together with the experimental ones . since the pairing energies are negligible in the hfb@xmath184 and hfb@xmath189 for both nuclei , both sets of @xmath217-ray energies practically coincide . with respect to the vap@xmath184 values they are of the same quality as the vap@xmath189 though somewhat smaller than these one , as expected from the behavior of the pairing correlations . in this paper we have derived the expressions needed to perform a particle number projected hfb calculation with a finite range and density dependent force in the triaxial basis . we have studied and thoroughly discussed the divergences that appear for special values of orbital occupancies when exchange terms are neglected . we have performed exact and approximate calculations ( neglecting some exchange terms ) and found that in the latter case some problems associated with the convergence of the particle number projection arise leading in some cases to an inaccuracy of about 10 % in the values of the @xmath217-ray energies . interestingly , _ all terms _ of the force , even of the coulomb part , must be taken into account . this feature presents a challenge for interactions where a given force is used in the particle - hole channel and a different one in the particle - particle channel . we have furthermore performed variation after the projection for the @xmath1cr nuclei with the gogny force . we found strong pairing correlations , at variance with the weekly correlated hfb approach . these correlations smoothly decrease with increasing angular momentum indicating that no sharp phase transition takes place as predicted by the hfb approach . the projected calculations improve the hfb ones but some correlations ( probably t=0 pairing ) not present in our approach are still missing . this work was supported in part by dgicyt , spain under project pb97 - 0023 . one of us ( m.a ) acknowledges a grant from the spanish ministry of education under project pb94 - 0164 . in the calculations we use the gogny interaction @xcite as the effective force . the main ingredients of this force are the phenomenological density dependent term which was introduced to simulate the effect of a g matrix interaction and the finite range of the force which allows to obtain the pairing and hartree fock fields from the same interaction . we use the parametrization d1s , which was fixed by berger _ et al . _ the force is given by @xmath227 and the coulomb force @xmath228 the density operator , @xmath229 , is given by @xmath230 in the two - body interaction used in the calculations we also include the one body and two body center of mass corrections . @xmath231 in this appendix we show that the projected energy is always well defined if the direct ( hartree ) , exchange ( fock ) and pairing terms of each component of the force are taken into account , i.e. if no exchange terms are neglected . the projected energy can be written in the following way @xmath232 where @xmath233 these energies are given in terms of the @xmath234-dependent density matrix @xmath235 and the pairing tensors @xmath236 and @xmath237 . these quantities can be easily evaluated in the canonical basis . in this basis , denoted by @xmath238 , the usual density matrix @xmath239 is diagonal and the usual pairing tensor @xmath240 has the canonical form . in this basis the quantities we are interested in are given by @xmath241 these matrix elements can be easily evaluated @xmath242 here @xmath243 denotes the canonical conjugated state of @xmath244 . to calculate @xmath245 , we shall proceed in the same way . let us first assume @xmath246 , then @xmath247 that means , the @xmath234-dependent density matrix @xmath248 , like the normal density matrix @xmath249 , is diagonal in the canonical basis . for @xmath250 we obtain @xmath251 therefore , in the canonical basis , the matrix @xmath252 is given by @xmath253 we see in this expression that for @xmath254 and @xmath255 , the matrix element diverges and one has to be very careful in the calculations . it is easy to show that @xmath256 . let us now calculate the @xmath234-dependent pairing tensors in the canonical basis @xmath257 in the same way @xmath258 the non - zero matrix elements are given by @xmath259 obviously , @xmath260 and @xmath261 . again the dangerous terms are for @xmath254 and @xmath262 . > from eq . ( [ eq : ephf ] ) and using the expressions derived above , we obtain @xmath263 taking into account that @xmath264 , we see that the possible poles of @xmath265 and @xmath266 are canceled when @xmath267 , i.e. we may only have problems if @xmath268 . the contribution to the hartree- fock energy in this case is given by @xmath269_{pole } & = & \sum_{k > 0 } \sum_l { \bar{v}}_{k \bar{k } k \bar{k } } \ , \ , y_{l } \frac{v_k^4e^{4 { { \mathrm{i}}}\varphi_l } } { { ( u_k^2 + v_k^2 e^{2 { { \mathrm{i}}}\varphi_l})}^2 } \nonumber \\ & = & { \mathcal n } \sum_{k > 0 } \sum_l { \bar{v}}_{k \bar{k } k \bar{k } } \frac{v_k^4e^{4 { { \mathrm{i}}}\varphi_l } } { { ( u_k^2 + v_k^2 e^{2 { { \mathrm{i}}}\varphi_l } ) } } \cdot \prod_{m>0,m \ne k } ( u_m^2 + v_m^2 e^{2 { { \mathrm{i}}}\varphi_l } ) . \label{dirterm}\end{aligned}\ ] ] where we have made use of eq . ( [ yltau ] ) and @xmath270^{-1}$ ] . this contribution clearly diverges for @xmath157 and @xmath187 . the pairing energy term is given by eq . ( [ eq : eppa ] ) , in the canonical basis it takes the form @xmath271 as before , @xmath272 cancels one possible pole , and if @xmath268 the contribution to the energy is @xmath273_{pole } & = & \sum_{k > 0 } \sum_l { \bar{v } } _ { k \bar{k } k \bar{k } } \ , \ , y_{l } \frac{u_k^2 v_k^2 e^{2 { { \mathrm{i}}}\varphi_l } } { { ( u_k^2 + v_k^2 e^{2 { { \mathrm{i}}}\varphi_l})}^2 } \nonumber \\ & = & { \mathcal n}\sum_{k > 0 } \sum_l { \bar{v}}_{k \bar{k } k \bar{k } } \frac{u_k^2 v_k^2 e^{2 { { \mathrm{i}}}\varphi_l } } { { ( u_k^2 + v_k^2 e^{2 { { \mathrm{i}}}\varphi_l})}}\cdot \prod_{m>0,m \ne k } ( u_m^2 + v_m^2 e^{2 { { \mathrm{i}}}\varphi_l } ) \label{exchterm}\end{aligned}\ ] ] which also diverges for @xmath157 and @xmath187 . the total contribution to the energy is given by the sum of eq.([dirterm ] ) and eq.([exchterm ] ) , we obtain @xmath269_{pole}+ [ e_g^p]_{pole}&= & \sum_{k > 0 } \sum_l { \bar{v } } _ { k \bar{k } k \bar{k } } \ , \ , \frac{y_{l } v_k^2 e^{2 { { \mathrm{i}}}\varphi } } { { ( u_k^2 + v_k^2 e^{2 { { \mathrm{i}}}\varphi_l})}^2 } \nonumber \\ & = & { \mathcal n}\sum_{k > 0 } \sum_l { \bar{v}}_{k \bar{k } k \bar{k } } v_k^2 e^{2 { { \mathrm{i}}}\varphi_l } \cdot \prod_{m>0,m \ne k } ( u_m^2 + v_m^2 e^{2 { { \mathrm{i}}}\varphi_l } ) \label{totalterm}\end{aligned}\ ] ] in the previous demonstration we have not considered the case of density dependent interactions @xmath274 in the analysis of the convergence . we shall distinguish between the two choices for the hamiltonian density in order to study this problem . taking into account the expression ( [ eq : epdd1 ] ) , one sees that the density dependent energy depends only on the proton and neutron projected densities . we must therefore analyze the behavior of these densities in the presence of poles . the projected density is given by @xmath275 as we saw in the previous section , any pole of @xmath276 will be canceled by a zero of the same order from the factor @xmath272 . therefore , if we take the projected density for the density dependent term , the total energy including this term is well defined . in this case the energy of the density dependent term is given by eq . ( [ eq : epdd2 ] ) . as we can see in this expression and in eq . ( [ eq : gammatil ] ) , the densities @xmath277 and the norms @xmath278 appear pairwise indicating a cancellation of possible poles and zeros . however , problems may arise if the interaction @xmath279 itself has some poles . in this prescription @xmath279 depends on @xmath280 given by eq . ( [ mixdens ] ) . writing the density matrix elements in the canonical basis we find that where the functions @xmath282 are defined in eq . ( [ rhodef ] ) . clearly , if one of the @xmath283 , let s say @xmath284 , equals @xmath285 then the density is singular at @xmath286 and in the neighborhood of this point it behaves as @xmath287 which is clearly divergent . this singularity , however , does not pose any real drawback as the density dependence of the interaction is proportional to @xmath288 with @xmath289 and therefore the singularity is integrable . 99 p. ring and p. shuck , _ the nuclear many body problem _ ( springer - verlag , berlin , 1980 ) . k. dietrich , h. j. mang and j. h. pradal , phys . * 135 * ( 1964 ) b22 . egido , h.j . mang , p. ring , nucl . phys . a 339 ( 1980 ) 390 - 414 . j. l. egido and p. ring , nucl . phys . a 383 , 189 ( 1982 ) ; nucl a 388 , 19 ( 1982 ) . lipkin , , ann . ( ny ) 12 ( 1960 ) 425 ; y. nogami phys . b 134 ( 1964)313 b. gall , p. bonche , j. dobaczeski , h. flocard , p. h. heenen , z. phys . a 348(1994 ) 183 . p. bonche , h. flocard , p. h. heenen , nucl . phys . a 598 , 169 ( 1996 ) . a. valor , j.l . egido and l. m. robledo , phys . c 53 ( 1996 ) 172 . afanasjev et al . , nucl . a 676(2000 ) 196 . j. decharg and d. gogny , phy . c 21(1980 ) 1568 . e. caurier , j.l . egido , g. martinez - pinedo , a. poves , j. retamosa , l. m. robledo and a. zuker , _ phys . lett . _ * 75 * , 2466 ( 1995 ) g. martinez - pinedo , a. poves , l. m. robledo , e. caurier , f. nowacki , j. retamosa and a. zuker , _ phys . rev . _ * c54 * , r2150 ( 1996 ) j.a . cameron et al . , phys . rev c49 , 1347 ( 1994 ) ; j.a . cameron et al . , phys . b319 , 58 ( 1993 ) ; t.w . burrows , nucl . data sheets 68 , 1 ( 1993 ) . burrows , nuclear data sheets ( 1990 ) ; s.m . lenzi et al . , phys . rev c56 , 1313 ( 1997 ) . a. poves and g. martinez - pinedo , phys . b430 * ( 1998)203 . d. gogny , in nuclear selfconsistent fields " , eds . g. ripka and m. porneuf ( north holland , 1975 ) . j. f. berger , m. girod and d. gogny , comp . . comm . * 63 * , 365 ( 1991 ) .

we study the effects of subgroup distortion in the wreath products , where is finitely generated abelian . we show that every finitely generated subgroup of has distortion function equivalent to some polynomial . moreover , for infinite , and for any polynomial , there is a-generated subgroup of having distortion function equivalent to the given polynomial . also a formula for the length of elements in arbitrary wreath product easily shows that the group has distorted subgroups , while the lamplighter group has no distorted ( finitely generated ) subgroups .

1009.4364

the notion of subgroup distortion was first formulated by gromov in @xcite . for a group @xmath7 with finite generating set @xmath8 and a subgroup @xmath9 of @xmath7 finitely generated by @xmath10 , the distortion function of @xmath9 in @xmath7 is @xmath11 where @xmath12 represents the word length with respect to the given generating set @xmath10 , and similarly for @xmath13 . this function measures the difference in the word metrics on @xmath7 and on @xmath9 . as usual , we only study distortion up to a natural equivalence relation . for non - decreasing functions @xmath14 and @xmath15 on @xmath16 , we say that @xmath17 if there exists an integer @xmath18 such that @xmath19 for all @xmath20 . we say two functions are equivalent , written @xmath21 , if @xmath17 and @xmath22 . when considered up to this equivalence , the distortion function becomes independent of the choice of finite generating sets . if the subgroup @xmath9 is infinite then the growth of the distortion function is at least linear , and therefore one does not extend the equivalence classes using the equivalence defined by the inequality @xmath23 a subgroup @xmath9 of @xmath7 is said to be undistorted if @xmath24 . if a subgroup @xmath9 is not undistorted , then it is said to be distorted , and its distortion refers to the equivalence class of @xmath25 . [ poa ] suppose there exists a subsequence of @xmath16 given by @xmath26 where @xmath27 for @xmath28 . if there exists @xmath29 such that @xmath30 , for all @xmath28 , and @xmath31 , then @xmath32 . here we study the effects of distortion in various subgroups of the wreath products @xmath33 , for @xmath34 , and more generally , in @xmath0 where @xmath1 is finitely generated abelian . note that wreath products @xmath35 where @xmath1 is abelian play a very important role in group theory for many reasons . given any semidirect product @xmath36 with abelian normal subgroup @xmath37 then any two homomorphisms from @xmath38 and @xmath39 ( uniquely ) extend to a homomorphism from @xmath35 to @xmath40 also , if @xmath41 is presented as a factor - group @xmath42 of a @xmath43-generated free group @xmath44 then the maximal extension @xmath45 $ ] of @xmath41 with abelian kernel is canonically embedded in @xmath46 ( see @xcite . ) wreath products of abelian groups give an inexhaustible source of examples and counter - examples in group theory . for instance , the group @xmath47 is the simplest example of a finitely generated ( though not finitely presented ) group containing a free abelian group of infinite rank . in @xcite the group @xmath47 is studied in connection with diagram groups and in particular with thompson s group . in the same paper , it is shown that for @xmath48 , where the group @xmath49 appears @xmath50 times , there is a subgroup @xmath51 having distortion function @xmath52 . in contrast to the study of these iterated wreath products , here we obtain polynomial distortion of arbitrary degree in the group @xmath47 itself . in @xcite the distortion of @xmath47 in baumslag s metabelian group is shown to be at least exponential , and an undistorted embedding of @xmath47 in thompson s group is constructed . in this note , rather than embedding the group @xmath47 into larger groups , or studying multiple wreath products , we will study distorted and undistorted subgroups in the wreath products @xmath0 with @xmath1 finitely generated abelian . the main results are as follows . [ x ] let @xmath1 be a finitely generated abelian group . 1 . for any finitely generated infinite subgroup @xmath53 there exists @xmath54 such that the distortion of @xmath9 in @xmath0 is @xmath55 2 . if @xmath1 is finite , then @xmath56 ; that is , all subgroups are undistorted . if @xmath1 is infinite , then for every @xmath54 , there is a @xmath3-generated subnormal subgroup @xmath9 of @xmath0 having distortion function @xmath55 the following will be explained in subsection [ fs ] . [ ot ] for every @xmath54 , there is a @xmath3-generated subgroup @xmath9 of the free @xmath57-generated metabelian group @xmath58 having distortion function @xmath59 [ xx ] if we let the standard generating set for @xmath47 be @xmath60 , then the subgroup @xmath61,b ] , \cdots , b ] \rangle$ ] , where the commutator is @xmath62-fold , is @xmath63 subnormal , isomorphic to the whole group @xmath47 , with distortion @xmath64 . in particular the normal subgroup @xmath65 \rangle$ ] has quadratic distortion . corollary [ xx ] is proved at the end of this paper . because the subgroup @xmath66 , b \rangle$ ] of @xmath47 is normal , it follows by induction that the distorted subgroup @xmath67 is subnormal . [ mmm ] there are distorted embeddings from the group @xmath47 into itself as a normal subgroup . for example , the map defined on generators by @xmath68 $ ] extends to an embedding , and the image is a quadratically distorted subgroup by corollary [ xx ] . by lemma [ t2 ] , @xmath47 is the smallest example of a metabelian group embeddable to itself as a normal subgroup with distortion . [ sc ] there is a distorted embedding of @xmath47 into thompson s group @xmath69 . under the embedding of remark [ mmm ] , @xmath47 embeds into itself as a distorted subgroup . it is proved in @xcite that @xmath47 embeds to @xmath69 . therefore , corollary [ sc ] is true . it is interesting to contrast theorem [ x ] part @xmath70 with the following , which will be discussed in section [ ppq ] . throughout this paper , we use the convention that @xmath71 represents the finite group @xmath72 . [ pqp ] the group @xmath73 for @xmath74 , has a finitely generated subgroup @xmath9 with distortion at least @xmath75 . some of the techniques to be introduced in this paper include some computations with polynomials . we will use the theory of modules over principal ideal domains in section [ mt ] to reduce the problem of subgroup distortion in @xmath33 to the consideration of certain @xmath3-generated subgroups in @xmath47 . every such subgroup is associated with a polynomial , and therefore we need to define and compute the distortion of arbitrary polynomial , as in theorem [ tgs ] . all of these techniques are used in conjunction with theorem [ t6 ] , which provides a formula for computing the word length in arbitrary wreath product and makes computing subgroup distortion more tangible in the examples we consider . here we provide some examples of distortion as well as some basic facts to be used later on . \1 . consider the three - dimensional heisenberg group @xmath76 , [ a , c]=[b , c ] = 1 \rangle.$ ] it has cyclic subgroup @xmath77 with quadratic distortion , which follows from the equation @xmath78 $ ] . the baumslag - solitar group @xmath79 has cyclic subgroup @xmath80 with at least exponential distortion , because @xmath81 however , there are no similar mechanisms distorting subgroups in @xmath47 . therefore , a natural conjecture would be that free metabelian groups or the group @xmath47 do not contain distorted subgroups . this conjecture was brought to the attention of the authors by denis osin . the result of theorem [ x ] shows that the conjecture is not true . the following facts are well - known and easily verified . when we discuss distortion functions , it is assumed that the groups under consideration are finitely generated . [ wkf ] \1 . if @xmath82 and @xmath83 < \infty$ ] then @xmath84 . \2 . if @xmath85 then @xmath86 \3 . if @xmath85 then @xmath87 \4 . if @xmath9 is a retract of @xmath7 then @xmath84 \5 . if @xmath7 is a finitely generated abelian group , and @xmath82 , then @xmath84 . we consider the wreath products @xmath35 of finitely generated groups @xmath88 and @xmath89 . we introduce the notation that @xmath35 is the semidirect product @xmath90 , where @xmath91 is the direct product @xmath92 of isomorphic copies @xmath93 of the group @xmath94 we view elements of @xmath91 as functions from @xmath41 to @xmath1 with finite support , where for any @xmath95 , the support of @xmath14 is @xmath96 . the ( left ) action @xmath97 of @xmath41 on @xmath91 by automorphisms is given by the following formula : for any @xmath98 and @xmath99 we have that @xmath100 . any element of the group @xmath35 may be written uniquely as @xmath101 where @xmath102 . the formula for multiplication in the group @xmath35 is given as follows . for @xmath103 we have that @xmath104 in particular , @xmath41 acts by conjugation on @xmath91 in the wreath product : @xmath105 therefore the wreath product is generated by the subgroups @xmath41 and @xmath106 where non - trivial functions from @xmath107 have support @xmath108 in what follows , the subgroup @xmath107 is identified with @xmath109 and so @xmath110 and @xmath111 is a finite set of generators in @xmath112 in particular , @xmath47 is generated by @xmath113 and @xmath114 where @xmath113 generates the left ( passive ) infinite cyclic group and @xmath114 generates the right ( active ) one . here we observe that a finitely generated abelian subgroup of @xmath115 with finitely generated abelian @xmath1 and @xmath41 is undistorted . it should be remarked that the authors are aware that the proof of the fact that abelian subgroups of @xmath33 are undistorted is available in @xcite . in that paper it is shown that @xmath33 is a subgroup of the thompson group @xmath69 , and that every finitely generated abelian subgroup of @xmath69 is undistorted . however , our observation is elementary and so we include it . [ abel ] let @xmath1 and @xmath41 be finitely generated abelian groups . then every finitely generated abelian subgroup @xmath9 of @xmath116 is undistorted . it follows from the classification of finitely generated abelian groups @xmath7 that every subgroup @xmath10 is a retract of a subgroup of finite index in @xmath7 , and so we are done if @xmath9 is a subgroup of @xmath1 or @xmath41 , or if @xmath117 , by lemma [ wkf ] . therefore we assume that @xmath118 since @xmath9 is abelian , this implies that the the factor - group @xmath119 is finite . then it suffices to prove the lemma for @xmath120 since @xmath121\le \infty.$ ] because @xmath122 is finitely generated , it is contained in a finite product of conjugate copies of @xmath1 . that is to say , @xmath123 for a wreath product @xmath124 where @xmath125 has finite index in @xmath41 . we are now reduced to our earlier argument , thus completing the proof . in fact , under the assumptions of lemma [ abel ] , @xmath9 is a retract of a subgroup having finite index in @xmath112 we now return to one of the motivating ideas of this paper , and complete the explanation of remark [ mmm ] . [ t2 ] the group @xmath47 is the smallest metabelian group which embeds to itself as a normal distorted subgroup in the following sense . for any metabelian group @xmath7 , if there is an embedding @xmath126 such that @xmath127 and @xmath128 is a distorted subgroup in @xmath7 , then there exists some subgroup @xmath9 of @xmath7 for which @xmath129 by lemma [ wkf ] , we have that the group @xmath130 is infinite , else @xmath128 would be undistorted . being a finitely generated solvable group , @xmath130 must have a subnormal factor isomorphic to @xmath49 . because @xmath131 , one may repeat this argument to obtain a subnormal series in @xmath7 with arbitrarily many infinite cyclic factors . therefore , the derived subgroup @xmath132 has infinite ( rational ) rank . since the group @xmath133 is finitely presented , the action of @xmath41 by conjugation makes @xmath132 a finitely generated left @xmath41 module . hence , @xmath134 for some finitely generated @xmath135 . because it is a finitely generated abelian group , @xmath136 is a product of cyclic groups . therefore for some @xmath137 we have a subgroup @xmath138 of finite rank in @xmath132 but @xmath139 has infinite rank . then @xmath1 has an element @xmath113 such that the @xmath140-submodule generated by @xmath113 has infinite rank , and so it is a free @xmath140-module . it follows that @xmath113 and @xmath114 , where @xmath141 , generate a subgroup of the form @xmath47 . in @xcite , magnus shows that if @xmath142 is an absolutely free group of rank @xmath43 with normal subgroup @xmath143 , then the group @xmath45 $ ] embeds into @xmath144 . this wreath product is a semidirect product @xmath145 where the action of @xmath7 by conjugation turns @xmath91 into a free left @xmath146$]-module with @xmath43 generators . for more information in an easy to read exposition , refer to @xcite . the monomorphism @xmath147 \rightarrow { { \mathbb z}}^k { \textrm { wr } } g$ ] is called the magnus embedding . we let @xmath148 denote the @xmath43-generated derived length @xmath149 free solvable group . [ y ] if @xmath150 , then the group @xmath148 contains a subgroup isomorphic to @xmath47 . it is well known ( and follows from the magnus embedding ) that any nontrivial @xmath151 and @xmath152 generate @xmath47 . it should be noted that by results of @xcite , the group @xmath153 can not be embedded into any free metabelian or free solvable groups . subgroup distortion has connections with the membership problem . it was observed in @xcite and proved in @xcite that for a finitely generated subgroup @xmath9 of a finitely generated group @xmath7 with solvable word problem , the membership problem is solvable in @xmath9 if and only if the distortion function @xmath154 is bounded by a recursive function . by theorem 2 of @xcite , the membership problem for free solvable groups of length greater than two is undecidable . therefore , because of the connections between subgroup distortion and the membership problem just mentioned , we restrict our primary attention to the case of free metabelian groups . it is worthwhile to note that the membership problem for free metabelian groups is solvable ( see @xcite ) . lemma [ y ] motivates us to study distortion in @xmath47 in order to better understand distortion in free metabelian groups . distortion in free metabelian groups is similar to distortion in wreath products of free abelian groups , by lemma [ y ] and the magnus embedding . in particular , if @xmath155 then @xmath156 thus by lemma [ wkf ] , given @xmath157 we have @xmath158 this explains corollary [ ot ] . on the other hand , given @xmath159 then we have @xmath160 based on this discussion , we ask the following . an answer would be helpful in order to more fully understand subgroup distortion in free metabelian groups . what effects of subgroup distortion are possible in @xmath161 for @xmath162 ? here we aim to further understand how the length of an element of a wreath product @xmath163 depends on the canonical form of this element . let us start with @xmath164 where @xmath165 because the subgroup @xmath91 of @xmath166 is abelian , we also use additive notation to represent elements of @xmath91 . [ yyy ] in the case of @xmath167 , we use module language to write any element as @xmath168 is a laurent polynomial in @xmath169 , and the sums are finite . similarly , if @xmath1 is a finitely generated abelian group , then by the definition of @xmath170 arbitrary element in @xmath0 is of the form @xmath171 where @xmath172 are laurent polynomials and @xmath173-s generate @xmath1 . this form is unique if @xmath174 is a free abelian with basis @xmath175 . we will use the notation that @xmath176 equals the conjugate @xmath177 for @xmath178 and @xmath179 the normal form described in remark [ yyy ] for elements of @xmath180 is necessary to obtain a general formula for computing the word length . [ yy ] arbitrary element of @xmath0 may be written in a normal form , following @xcite , as @xmath181 where @xmath182 , and @xmath183 are elements in @xmath184 the following formula for the word length in @xmath0 is given in @xcite . [ t5 ] given an element in @xmath0 having normal form as in remark [ yy ] , its length is given by the formula @xmath185 where @xmath186 is the length in the group @xmath94 the formula from lemma [ t5 ] becomes more intelligible if one extends it to wreath products @xmath35 of arbitrary finitely generated groups . we want to obtain such a generalization in this section since we consider non - cyclic active groups in section [ ppq ] . we fix the notation that , with respect to the symmetric generating set @xmath187 , the cayley graph @xmath188 is defined as follows . the set of vertices is all elements of @xmath7 . for any @xmath189 , @xmath15 and @xmath190 are joined by an edge pointing from @xmath15 to @xmath190 whose label is @xmath191 . any @xmath192 can be expressed as follows : @xmath193 where @xmath194 and for @xmath195 we have @xmath196 . the expression @xmath197 is unique , up to a rearrangement of the ( commuting ) factors @xmath198 . for any @xmath199 with canonical form as in equation @xmath197 we consider the set @xmath200 of paths in the cayley graph @xmath188 which start at @xmath201 , go through every vertex @xmath202 and end at @xmath15 . we introduce the notation that @xmath203 @xmath204 we also define the norm of any such representative @xmath205 of @xmath91 by @xmath206 we have the following formula for word length , which generalizes that given for the case where @xmath207 in the paper @xcite . ( * caution * : the _ right_-action definition of wreath product would be incompatible with the standard definition of cayley graph in the proof of theorem [ t6 ] . ) [ t6 ] for any element @xmath199 , we have that @xmath208 where @xmath209 is the canonical form of equation @xmath197 . we will use the following pseudo - canonical ( non - unique ) form in the proof . this is just the expression of equation @xmath197 but without the assumption that all @xmath210 are distinct or @xmath211-s are non - trivial . for any element @xmath192 which is expressed in pseudo - canonical form we may define a quantity depending on the given factorization by @xmath212 first we show that for @xmath213 in canonical form @xmath214 it holds that @xmath215 . by the choice of generating set @xmath216 of @xmath35 , we have that any element @xmath192 may be written as @xmath217 where @xmath218 and @xmath219 can be trivial , but all other factors are non - trivial . we may choose the expression so that @xmath220 observe that we may use the expression from equation @xmath221 to write @xmath222 where @xmath223 and @xmath224 , for @xmath225 . then we have by definition that for the pseudo - canonical form @xmath226 , @xmath227 it is possible that in the form of equation @xmath226 , some @xmath228 for @xmath229 . when taking @xmath213 to the canonical form @xmath230 of equation @xmath197 , we claim that @xmath231 obtaining the canonical form requires a finite number of steps of the following nature . we take an expression such as @xmath232 and replace it with @xmath233 the assertion of equation @xmath234 follows because @xmath235 equation @xmath236 is true because @xmath237 which implies that @xmath238 @xmath239 finally , we have that @xmath240 because the right hand side is the length of a particular path in @xmath200 : the path which travels from @xmath201 to @xmath241 to @xmath242 to @xmath243 to @xmath15 . it follows that the length of this path is at least as large as the length of @xmath244 . thus for a canonical form @xmath209 we see by equations @xmath245 , @xmath234 and @xmath236 that @xmath246 to obtain the reverse inequality , take @xmath247 in @xmath35 in canonical form . by the definition , @xmath244 will be a path that starts at @xmath201 , goes in some order directly through all of @xmath202 , and ends at @xmath248 we may rephrase this to say that for some @xmath249 there is a path @xmath250 in @xmath200 such that @xmath251 . in other words , @xmath252 moreover , in the wreath product we have that @xmath253 @xmath254 this implies that @xmath255 @xmath256 we begin with the following result , the proof of which exploits the formula of theorem [ t6 ] . [ t3 ] the group @xmath5 contains distorted subgroups . this is interesting in contrast to the case of @xmath258 which has no effects of subgroup distortion . the essence in the difference comes from the fact that the cayley graph of @xmath49 is one - dimensional , and that of @xmath259 is asymptotically two - dimensional , which gives us more room to create distortion using theorem [ t6 ] . we will use the following notation in the case of @xmath260 : @xmath113 generates the passive group of order @xmath3 while @xmath114 and @xmath261 generate the active group @xmath262 the canonical form of equation @xmath197 will be denoted by @xmath263 for @xmath264 distinct elements of @xmath259 and @xmath265 . we may do this because any nontrivial element of @xmath266 is just equal to the generator @xmath113 . [ ss ] let @xmath82 be generated by a nontrivial element @xmath267 as well as the generators @xmath268 of @xmath259 . then @xmath269 . we know that @xmath270 is a free module over @xmath271 $ ] therefore , we may think of @xmath91 as being the laurent polynomial ring in two variables , say , @xmath169 for @xmath114 and @xmath272 for @xmath261 . we can use the module language to express any element as @xmath273 , where for @xmath274 we have that @xmath275 . this corresponds to the canonical form @xmath276 where @xmath277 for @xmath278 . we now have all the required facts to prove proposition [ t3 ] . of proposition [ t3 ] : let @xmath279 and @xmath280 where @xmath281=(1+x)a$ ] . by lemma [ ss ] we have that @xmath269 . let @xmath282 the element @xmath283 is in canonical form , when written in the additive group notation as @xmath284 by theorem [ t6 ] , we have that its length in @xmath9 is at least @xmath285 since the support of it has cardinality @xmath286 , and the length of arbitrary loop going through @xmath286 different vertices is at least @xmath286 . + now we compute the length of @xmath287 in @xmath7 . we have that @xmath288a.\ ] ] theorem [ t6 ] shows that @xmath289 . this is because the shortest path in @xmath290 starting at @xmath201 , passing through @xmath291 and @xmath292 and ending at @xmath201 is given by traversing the perimeter of the rectangle , and so gives the length of @xmath293 . therefore the subgroup @xmath9 is at least quadratically distorted . the subgroup @xmath9 is not normal in @xmath7 because the element @xmath294 is not in @xmath9 . the proof of proposition [ t3 ] can be generalized as follows . consider the group @xmath295 for @xmath296 and @xmath162 . then the subgroup @xmath297 where @xmath298,b_2], ... b_{k-1}]$ ] has distortion at least @xmath75 . this is a restatement of proposition [ pqp ] . by ( the analogue of ) lemma [ ss ] we have that @xmath269 and so we can compute lengths using theorem [ t6 ] . consider the element @xmath299 in @xmath9 . then it has length in @xmath9 at least equal to @xmath300 because the path in @xmath301 arising from theorem [ t6 ] would need to pass through at least @xmath2 vertices : @xmath302 for @xmath303 . in the group @xmath7 , @xmath304 this has linear length , which follows because the vertices of the support are placed along the edges of a @xmath43-dimensional parallelotope , such that the length of any edge of the parallelotope is at most @xmath149 . although the notion of equivalence has only been defined for functions from @xmath16 to @xmath16 , we would like to define a notion of equivalence for functions on a finitely generated group . we say that two functions @xmath305 are equivalent if there exists @xmath18 such that for any @xmath306 we have @xmath307 if there is a function @xmath308 such that @xmath309 , then for any subgroup @xmath9 of @xmath7 , @xmath310 . we need to establish a looser way of estimating lengths in @xmath47 , than the formula introduced in lemma [ t5 ] . recall that this group has standard generators @xmath311 ( passive ) and @xmath114 ( active ) . here we call _ exemplary _ any sugroup @xmath312 where @xmath313 we have @xmath314 where @xmath315 and @xmath316 this follows without loss of generality by conjugating of @xmath205 by a power of @xmath114 . thus we associate a polynomial @xmath317 $ ] with any exemplary subgroup @xmath318 [ xxx ] the mapping @xmath319 extends to a monomorphism of the wreath product @xmath47 onto the exemplary subgroup this follows because in this case @xmath91 is a free module with one generator @xmath113 over the domain @xmath320,$ ] @xmath314 and the mapping @xmath321 ( @xmath322 ) is an injective module homomorphism . then for any element @xmath323 , we may write @xmath324 where @xmath325 is a laurent polynomial . denote by @xmath326 the sum @xmath327 for this element , consider the norms @xmath328 letting @xmath329 @xmath330 we define @xmath331 consider the function @xmath332 the following lemma shows that we may simplify computations of word length in exemplary subgroups . [ r2 ] let @xmath333 be an exemplary subgroup . then we have that @xmath334 recall that by lemma [ t5 ] , we have the following formulas . for @xmath323 with the notation established above , we have that : @xmath335 and @xmath336 the following inequality follows from the definitions : @xmath337 similarly , we have that @xmath338 we have that @xmath339 thus , @xmath340 @xmath341 the first inequality follows from equation ( [ bb ] ) , the second from equation ( [ bbb ] ) . on the other hand , we have that @xmath342 @xmath343 the first inequality follows from equation ( [ bbb ] ) , the second from equation ( [ c ] ) , and the third from equation ( [ cc ] ) . thus @xmath344 and @xmath345 are equivalent . in order to understand distortion in exemplary subgroups of @xmath47 , we will introduce the notion of distortion of a polynomial . [ pd ] let @xmath346 be a subring of a field with a real valuation , and consider the polynomial ring @xmath347 $ ] . we will define the norm function @xmath348 \rightarrow \mathbb{r}^+$ ] which takes any @xmath349 $ ] to @xmath350 for any @xmath351 $ ] and @xmath29 , we define the distortion of the polynomial @xmath352 from @xmath16 to @xmath16 by @xmath353 taking into account the inequality @xmath354 , one can easily find some explicit upper boundes @xmath355 for the modules of the coefficient at @xmath356 of @xmath357 in formula , starting with the lowest coefficients . therefore the supremum in equation is finite . furthermore , if @xmath358 or @xmath359 then the supremum is taken over a compact set of polynomials of bounded degree with bounded coefficients , and since @xmath10 is a continuous function , one may replace @xmath360 by @xmath361 in definition [ pd ] . note that the distortion does not depend on the constant @xmath261 , up to equivalence , and so we will consider @xmath362 . the following fact makes concrete our motivation for studying distortion of polynomials . [ hpu ] let @xmath9 be an exemplary subgroup @xmath363 and @xmath364 for @xmath365 $ ] . then @xmath366 by lemma [ r2 ] , we have that @xmath367 let @xmath368 be so that @xmath369 . there exists @xmath370 so that @xmath371 and @xmath372 where @xmath373 is a regular polynomial . it is easy to check that @xmath374 and @xmath375 now observe that @xmath376 and @xmath377 . therefore , @xmath378 on the other hand , let us choose any polynomials @xmath379 such that @xmath380 @xmath381 and @xmath382 then by lemma [ t5 ] , @xmath383 while @xmath384 it follows that @xmath385 and the lemma is proved . given any polynomial @xmath386 , d_0,d_t \ne 0 $ ] with complex , real or integer coefficients , we are able to compute the equivalence class of its distortion function . [ hnc ] the distortion @xmath362 of @xmath352 with respect to the ring of polynomials over @xmath387 or @xmath359 is bounded from below by @xmath388 , up to equivalence , where @xmath261 is a complex root of @xmath352 of multiplicity @xmath389 and modulus one . let @xmath261 be a complex root of @xmath352 of multiplicity @xmath389 and modulus @xmath201 . that is , @xmath390 over @xmath359 . let @xmath391 then the product @xmath392 satisfies @xmath393 , because @xmath394 . on the other hand , because @xmath395 , we have that @xmath396 this implies that if @xmath397 ; i.e. @xmath398 , then @xmath399 , where the distortion is considered over @xmath359 , @xmath400 or over @xmath401 . we will show that a similar computation holds over @xmath400 and over @xmath49 even in the case when @xmath402 . let @xmath403 be the complex conjugate of @xmath261 . by hypothesis that @xmath404 we know that @xmath405 then @xmath406 is a root of @xmath407 of multiplicity @xmath389 as well , and @xmath408 where @xmath409 has real coefficients . consider the product @xmath410 where @xmath411 a simple calculation shows that each of the coefficients of this product is a sum of the form @xmath412 this is a geometric progression with common ratio @xmath413 . therefore , the modulus of every such coefficient is at most @xmath414 and so @xmath415 is @xmath416 . this computation implies that the products @xmath417 have the sum of the modules of their coefficients which are @xmath416 . the polynomial @xmath418 has real coefficients . there is a polynomial @xmath419 with integer coefficients such that each coefficient of @xmath420 has modulus at most one . thus @xmath421 is also @xmath416 . we will show that the sums of modules of coefficients of @xmath419 grow at least as @xmath388 on a subsequence from remark [ poa ] . it suffices to obtain the same property for @xmath422 since @xmath395 , we have that the sum of the modules of the coefficients of @xmath423 is at least @xmath424 we will show that there exists a subsequence @xmath425 so that on this sequence , @xmath426 we have that @xmath427 because @xmath406 . therefore @xmath428 and similarly , @xmath429 one of these two numbers must be at least one half because @xmath430 . thus either @xmath149 or @xmath431 can be included in the sequence @xmath425 , and all required properties are shown . in order to obtain upper bounds on distortion of polynomials we require some facts from linear algebra . fix an integer @xmath432 and let @xmath433 be arbitrary . [ ob ] let @xmath434 be @xmath435 column vectors . suppose that the sum of the modules of all coordinates of @xmath436 is bounded by some constant @xmath261 , and that the modulus of each coordinate of @xmath437 and @xmath438 is also bounded by @xmath261 . suppose further we have the relationship @xmath439 where @xmath1 is a @xmath440 matrix , in jordan normal form , having only one jordan block . then the modulus of each coordinate of arbitrary @xmath441 is bounded by @xmath442 where @xmath50 depends on @xmath1 only . in the case where the eigenvalue of @xmath1 does not have modulus one , the modulus of each coordinate of arbitrary @xmath441 is bounded by @xmath443 , where @xmath50 depends on @xmath1 only . all matrix entries are assumed to be complex . let @xmath444 be the eigenvalue of @xmath1 , so that @xmath445 we will consider cases . * first suppose that @xmath446 . from formula we derive : @xmath447 the following formula for @xmath448 is well - known because @xmath1 is assumed to be a jordan block ; it may also be checked easily using induction . we have that @xmath449 with the understanding that if @xmath450 , any terms of the form @xmath451 where @xmath452 are @xmath453 . arbitrary nonzero element of the matrix @xmath448 is of the form @xmath451 for some @xmath454 . let @xmath455 denote the @xmath456 entry of @xmath448 . then @xmath455 is either zero or of the form @xmath451 for some @xmath457 depending on @xmath458 and @xmath191 . then @xmath459 which is a constant depending on @xmath1 and not on @xmath137 , because the series on the right is convergent when @xmath460 . let @xmath461 let @xmath462 be the @xmath440 matrix whose @xmath456 entry is @xmath463 , and the column @xmath464 be obtained by placing in the @xmath465 row the sum of the modules of the entries of the @xmath465 row of all @xmath466 and @xmath437 . then every entry of @xmath464 is bounded by @xmath467 . observe that the modulus of every entry in the right side of is bounded by an entry of @xmath468 , which is in turn bounded by @xmath469 , which does not include any power of @xmath57 at all . * let @xmath470 . because @xmath471 is an eigenvalue of @xmath472 , there exists a decomposition @xmath473 where @xmath474 letting @xmath475 and @xmath476 we see by equation that @xmath477 for @xmath478 . observe that the sum of modules of coordinates of @xmath479 is less than or equal to @xmath480 , where @xmath458 depends on @xmath10 ( and hence on @xmath1 ) only . similarly , the sum of all modules of all coordinates of @xmath481 is bounded above by @xmath480 . this case now follows just as the previous one to obtain constant upper bounds on the modules of the entries in @xmath482 . finally , the modulus of any coordinate of @xmath483 is bounded by @xmath484 times the modulus of a coordinate of @xmath479 . * let @xmath485 . in this case , we have that @xmath486 @xmath487 it follows from equation @xmath488 that every entry of @xmath489 is bounded above by @xmath490 . [ yx ] let @xmath434 be @xmath435 column vectors . suppose that the sum of the modules of all coordinates of @xmath436 is bounded by some constant @xmath261 , and that the modulus of each coordinate of @xmath437 and @xmath438 is also bounded by @xmath261 . suppose further we have the relationship @xmath491where @xmath1 is a @xmath440 matrix . then the modulus of each coordinate of arbitrary @xmath441 is bounded by @xmath492 where @xmath50 depends on @xmath1 only , and @xmath493 is the maximal size of any jordan block of the jordan form of @xmath1 having eigenvalue with modulus one . there exists a jordan decomposition , @xmath494 . let @xmath495 and let @xmath496 . then for @xmath476 and @xmath475 we have that @xmath497 by hypothesis , the sum of the modules of all coordinates of @xmath481 is bounded by @xmath498 and the coordinates of @xmath499 and @xmath500 are bounded by @xmath501 as well . as we will explain , our problem can be reduced to the similar problem for @xmath502 in ( [ s ] ) . suppose that the modules of coordinates of every @xmath502 are bounded by @xmath503 where @xmath50 depends on @xmath1 only . then , letting @xmath504 and @xmath505 we have by definition of @xmath502 that arbitrary element of @xmath489 has modulus bounded above by @xmath506 where @xmath507 only depends on @xmath508 , as required . lemma [ ob ] says that if @xmath508 has only one jordan block , then the bound is constant if the eigenvalue does not have modulus one . otherwise , we have in this case that @xmath509 and the claim is true . if there is more than one jordan block present in @xmath508 , the problem is decomposed into at most @xmath43 subproblems , each with only one jordan block of size smaller than @xmath43 . again , we are done by lemma [ ob ] . we will use lemma [ yx ] to prove the following fact , which requires establishing some notation prior to being introduced . let @xmath510 where @xmath511 . let the @xmath512 matrix @xmath513 have @xmath465 column , for @xmath514 , given by @xmath515^{t}$ ] , where @xmath516 first appears as the @xmath465 entry in this @xmath465 column . given the matrix @xmath513 , we may also construct the matrix @xmath517 where @xmath518 , for @xmath519 . let @xmath389 be the maximal size of a jordan block of the jordan form of @xmath1 having eigenvalue with modulus one . [ r9 ] suppose that @xmath520^t$ ] is a solution to the system of equations @xmath521 , where @xmath522^t$ ] . then it is possible to bound the modules of all coordinates @xmath523 of the vector @xmath524 such that @xmath525 where @xmath526 for @xmath527 and the constant @xmath261 depends upon @xmath528 only . prior to proving lemma [ r9 ] we prove an easier special case . [ a ] it is possible to bound the coordinates @xmath529 of the vector @xmath524 from lemma [ r9 ] from above by @xmath530 where @xmath526 and @xmath531 . by cramer s rule , we have the explicit formula that @xmath532 where @xmath533 is the @xmath440 upper left submatrix of @xmath513 corresponding to the first @xmath43 equations , and @xmath534 is obtained by replacing column @xmath137 in @xmath533 with @xmath535^t$ ] . because @xmath536 , it suffices to show that the desired bounds exist for @xmath537 that is , we must show that there exists a constant @xmath538 depending on @xmath539 only such that @xmath540 for @xmath541 . by expanding along the @xmath542 column in @xmath534 , we find that @xmath543 where for each @xmath541 , @xmath544 is a function of @xmath539 only obtained as the determinant of a submatrix containing none of @xmath545 . the proof is complete by the triangle inequality . note that the @xmath546 for @xmath547 are similarly bounded by @xmath548 for the same @xmath114 and some @xmath549 as in lemma [ a ] . it is clear according to lemma [ a ] that we may assume that @xmath550 for the same @xmath551 for all @xmath552 . we proceed with the proof of lemma [ r9 ] . it suffices to obtain upper bounds for @xmath553 when @xmath554 . for such indices , we have that @xmath555 in other words , @xmath556 where @xmath557 and @xmath518 . let @xmath558^{t}$ ] and let @xmath559^t$ ] . then for the matrix @xmath1 of equation we have the recursive relationship @xmath560 for @xmath561 . observe that the matrix @xmath1 depends on @xmath528 only , and that the sum of modules of the entries in all @xmath562 are bounded by @xmath563 . we see by lemma [ a ] that lemma [ yx ] applies to our situation . therefore , the modules of coordinates of arbitrary @xmath564 , @xmath565 are bounded by @xmath566 , where @xmath50 depends only on @xmath528 , @xmath567 . [ cz ] let @xmath568 , where @xmath569 . then the distortion of @xmath352 is at most @xmath388 where @xmath389 is the maximal size of a jordan block of the jordan form of @xmath570 of equation with eigenvalue having modulus one . consider any @xmath571 as in definition [ pd ] . then consider @xmath572 . the coefficients @xmath573 are given by the matrix equation @xmath574 where @xmath575^t , y=[y_{s } , \dots , y_{s+p+t}]^t$ ] and @xmath576 is an @xmath577 matrix . by lemma [ r9 ] we have that for each @xmath578 that @xmath579 where @xmath580 therefore , @xmath581 the following theorem shows that the upper and lower bounds are the same , and so we can compute exactly the distortion of a polynomial . [ tgs ] let @xmath568 be a polynomial in @xmath582 $ ] . then the distortion of @xmath352 is equivalent to a polynomial . further , the degree of this polynomial is exactly one plus the maximal multiplicity of a ( complex ) root of @xmath407 having modulus one . on the one hand , lemma [ hnc ] shows that the distortion is bounded from below by the polynomial of degree one plus the maximal multiplicity @xmath389 of a root of @xmath407 having modulus one . on the other hand , the characteristic polynomial @xmath583 of the matrix @xmath1 in lemma [ cz ] equals @xmath584 . and so the real polynomials @xmath583 and @xmath407 have the same roots with modulus @xmath201 ( and with the same multiplicities ) . since the size of a jordan block does not exceed the multiplicity of the root of the characteristic polynomial , we have @xmath585 by lemma [ cz ] . the theorem is proved . theorem [ tgs ] will be used here for polynomials with integer coefficients , but it is valid ( with the same proof ) for polynomials with complex or real coefficients . theorem [ tgs ] and lemma [ cz ] imply the following . [ exempl ] the distortion of any exemplary subgroup @xmath9 of @xmath47 is equivalent to a polynomial . the degree of this polynomial is exactly one plus the maximal multiplicity of a ( complex ) root having modulus one of the polynomial @xmath407 associated with @xmath9 . for every @xmath586 the wreath product @xmath47 has subgroups @xmath587 isomorphic to @xmath588 and so we are forced to study distortion in the groups @xmath33 even we are interested in @xmath47 only . let @xmath589 be canonical generators of @xmath590 if a subgroup @xmath9 of @xmath591 is generated by @xmath592 where every @xmath593 belongs to the normal closure @xmath594 of @xmath173 ( @xmath594 = the submodule @xmath320a_i$ ] of @xmath91 ) then we say that @xmath9 is a _ tame _ subgroup of @xmath590 if @xmath595 then the subgroup @xmath596 is an exemplary subgroup of the wreath product @xmath597 [ lcc ] for the tame subgroup @xmath9 , we have that @xmath598 observe that @xmath599 is an undistorted embedding , due to that fact that @xmath596 is a retract of @xmath9 ( and similarly for @xmath600 ) . therefore , by lemma [ wkf ] we have that @xmath601 for every @xmath137 , and therefore @xmath602 to prove the other inequality , we consider an element @xmath603 with @xmath604 then there is a unique decomposition @xmath605 where @xmath606 and for @xmath607 we have @xmath608 since @xmath9 is tame . then we have @xmath609 since @xmath610 is a retract of @xmath40 therefore the required inequality will follow from the inequality @xmath611 this inequality is true indeed by theorem [ t6 ] because @xmath612 since @xmath613 and @xmath614 since @xmath9 is a tame subgroup of @xmath40 [ tame ] every tame subgroup of @xmath615 has a polynomial distortion . the statement follows from corollary [ exempl ] and lemma [ lcc ] . to get rid of the word ` tame ' in the formulation of lemma [ tame ] , we will need few remarks about modules . the following is well known ( see also @xcite ) . [ r1 ] the ring @xmath616 $ ] is a principal ideal domain if @xmath69 is a field . [ ccc ] suppose that @xmath617 is a submodule of a free module @xmath618 of rank @xmath43 over a ( commutative ) principal ideal ring @xmath619 then @xmath620 is a free module of rank @xmath621 , and modules @xmath618 and @xmath617 have bases @xmath622 and @xmath623 respectively such that for some @xmath624 @xmath625 at first we apply this statement to the following special case of theorem [ x ] part @xmath70 . [ ta ] if @xmath626 is a prime , then any finitely generated subgroup @xmath9 of @xmath627 containing the generator @xmath114 is undistorted . by lemma [ wkf ] , it suffices to show that @xmath9 has finite index in a retract @xmath628 of @xmath40 since @xmath626 is a prime , @xmath629 is a field . this implies by lemma [ r1 ] , that the ring @xmath630 $ ] is a principal ideal ring . let @xmath631 then @xmath632 is a free @xmath346-module by lemma [ ccc ] , and we have that @xmath632 and @xmath91 have bases @xmath633 and @xmath634 respectively , for @xmath635 such that @xmath636 for some polynomials @xmath637 . thus we can choose the generators for @xmath7 and @xmath9 to be @xmath638 and @xmath639 , respectively , and @xmath9 is a subgroup of the retract @xmath628 of @xmath7 , where @xmath628 is isomorphic to @xmath640 and is generated by @xmath641 now @xmath632 is a submodule of the @xmath642$]-module @xmath643 generated by @xmath644 and the factor - module @xmath645 is a direct sum of cyclic modules @xmath646 hence @xmath645 is finite since it is easy to see that each @xmath647 has finite order at most @xmath648 since the subgroup @xmath9 contains @xmath649 the index of @xmath9 in @xmath628 is also finite . we return to our discussion of module theory . let @xmath650 be generated by @xmath114 , as well as any elements @xmath651 . let @xmath632 be the normal closure of @xmath652 in @xmath653 i.e. , the @xmath320$]-submodule of @xmath91 generated by @xmath654 let @xmath655 and @xmath656 . observe @xmath617 and @xmath618 are free modules over @xmath657}$ ] of respective ranks @xmath43 and @xmath621 . [ cb ] it follows from lemma [ ccc ] that there exist @xmath658 with @xmath659 where @xmath660}$ ] . moreover , the modules generated by @xmath661 and @xmath662 are free . [ ca ] there is a bijective correspondence between the set of finitely generated @xmath663}$ ] submodules @xmath143 of @xmath663}^k$ ] and the set of subgroups @xmath664 of @xmath33 such that the finite set of generators of @xmath628 is of the form @xmath665 , @xmath666 . [ bfz ] let @xmath667 and @xmath668 be generated as submodules over @xmath663}$ ] by the elements from remark [ cb ] : @xmath669 and @xmath634 respectively . let @xmath122 and @xmath670 be subgroups of @xmath671 generated by @xmath672 and @xmath673 respectively . it follows by remark [ cb ] that that @xmath674 and @xmath675 . [ bc ] observe that under the correspondence of remark [ ca ] each generator @xmath676 of the group @xmath122 is in the normal closure of only one generator @xmath544 of @xmath670 , i.e. , @xmath122 is a tame subgroup of @xmath677 [ ba ] there exists @xmath678 so that @xmath679 and @xmath680 by remark [ ca ] we have that @xmath632 is a finitely generated @xmath663}$ ] module with generators @xmath652 . for each @xmath593 , we have that the element @xmath681 . therefore , by lemma [ ccc ] , there are @xmath682}$ ] so that @xmath683 first observe that @xmath684 because @xmath685 . next , there exists @xmath686 so that @xmath687 where @xmath688}$ ] . let @xmath689 then for any @xmath690 , we have that @xmath691 where @xmath692},$ ] and therefore , @xmath693 as required . a similar argument works for obtaining @xmath694 . [ ab ] let @xmath695 and let @xmath696 be the normal closure of elements @xmath666 . suppose that there exists @xmath697 and a finitely generated subgroup @xmath698 so that @xmath699 then @xmath700 we will use the notation that @xmath701 observe that the mapping @xmath702 is an injective homomorphism which restricts to an isomorphism @xmath703 . an easy computation which uses lemma [ t5 ] and the definition of @xmath704 shows that for any @xmath705 , we have that latexmath:[\[\label{e1 } computed in @xmath7 with respect to the usual generating set @xmath707 . observe that under the map @xmath704 we have that @xmath708 where the lengths in @xmath709 are computed with respect to the images under @xmath704 of a fixed generating set of @xmath710 . by their definitions , we have the embeddings @xmath711 by equation ( [ ac ] ) there exists @xmath712 depending only on the chosen generating sets of @xmath710 and @xmath713 so that @xmath714 it also follows by equation ( [ ac ] ) that there exists a constant @xmath715 depending only on the chosen generating sets of @xmath709 and @xmath713 so that @xmath716 first we show that @xmath717 let @xmath718 be such that @xmath719 and @xmath720 . then there exists @xmath721 such that @xmath722 . therefore , it follows that @xmath723 the first and second equalities follow by definition , the third by equation ( [ e2 ] ) , and the inequality is true because by equation ( [ e1 ] ) we have that @xmath724 we claim that @xmath725 let @xmath705 be such that @xmath726 . then @xmath727 which follows from equations ( [ e1 ] ) , ( [ e3 ] ) and by definition . on the other hand , we will show that @xmath728 let @xmath729 be such that @xmath730 . then @xmath731 which follows from equations ( [ e1 ] ) , ( [ xa ] ) and by definition . therefore , we have that @xmath732 we say that @xmath9 is a _ subgroup with @xmath114 _ in a wreath product @xmath733 if @xmath734 where @xmath735 . [ addm ] let @xmath9 be a subgroup with @xmath114 in @xmath591 . then the distortion of @xmath9 in @xmath33 is equivalent to the distortion of a tame subgroup @xmath122 of a wreath product @xmath736 @xmath737 this follows from the results of section [ mt ] . recall that the tame subgroup @xmath122 of the group @xmath670 was defined in lemma [ bfz ] , and these groups were associated to the given @xmath82 . it follows from lemmas [ ba ] and [ ab ] that @xmath738 and therefore @xmath739 [ withb ] the distortion of every subgroup with @xmath114 in @xmath33 is polynomial . this follows from corollary [ tame ] and lemma [ addm ] . in this section , we will reduce distortion in subgroups of @xmath0 where @xmath1 is finitely generated abelian to that in subgroups of @xmath33 only . [ grga ] let @xmath1 be a finitely generated abelian group and consider @xmath740 assume that @xmath43 is the torsion - free rank of @xmath741 if @xmath9 is a subgroup with @xmath114 in @xmath7 then the distortion of @xmath9 in @xmath7 is equivalent to that of a subgroup with @xmath114 in @xmath33 . there exists a series of subgroups @xmath742 for @xmath743 where @xmath744 has prime order for @xmath745 . we induct on @xmath746 . if @xmath747 , then @xmath748 and the claim holds . now let @xmath749 . observe that @xmath107 is a finitely generated abelian group with a series @xmath750 of length @xmath63 . therefore , by induction , any subgroup with @xmath114 in @xmath751 has distortion equivalent to that of a subgroup with @xmath114 in @xmath33 , for some @xmath43 . by lemma [ ta ] , all subgroups with @xmath114 of @xmath752 are undistorted . denote the natural homomorphism by @xmath753 let @xmath754 observe that @xmath755 the product is semidirect because @xmath756 is a normal subgroup which meets @xmath757 trivially , and it is isomorphic to the wreath product by definition : the action of @xmath114 on the module @xmath758 is the same . let @xmath759 $ ] . observe that @xmath346 is a noetherian ring . this follows from basic algebra because @xmath49 is a commutative noetherian ring . therefore , @xmath91 is a finitely generated module over the noetherian ring @xmath346 , hence is noetherian itself . thus , the @xmath346-submodule @xmath760 is finitely generated . let @xmath761 generate @xmath760 as a @xmath346-module . let @xmath762 be a set of generators of @xmath9 modulo @xmath756 ; that is , the canonical images of these elements generate the subgroup @xmath763 of @xmath670 . then the set @xmath764 generates @xmath9 . furthermore , the collection @xmath765 generates the subgroup @xmath766 of @xmath767 . let @xmath323 have @xmath768 . then the image @xmath769 in @xmath670 belongs to @xmath122 , because @xmath323 , and has length @xmath770 it follows by lemma [ ta ] that @xmath122 is undistorted in @xmath677 therefore , there exists a linear function @xmath771 ( which does not depend on the choice of @xmath15 ) such that @xmath772 that is to say , there exists a product @xmath200 of at most @xmath773 of the chosen generators @xmath762 of @xmath122 such that @xmath774 in @xmath122 . taking preimages , we obtain that @xmath775 . because @xmath9 is a subgroup of @xmath7 , there exists a constant @xmath261 depending only on the choice of finite generating set of @xmath9 such that for any @xmath776 we have that @xmath777 ) that @xmath778 this follows because @xmath775 by construction , and @xmath323 by choice . further , @xmath779 because it is a product of some of the generators of @xmath9 . since @xmath780 we see that @xmath781 . using the fact that @xmath7 and @xmath767 are wreath products together with the length formula in lemma [ t5 ] , we have that for any @xmath782 @xmath783 by induction , the finitely generated subgroup @xmath784 of @xmath767 has distortion function @xmath785 equivalent to that of a subgroup @xmath786 with @xmath114 in @xmath33 for some @xmath43 . that is , @xmath787 . in particular , for any @xmath788 , @xmath789 the first inequality follows from equation ( [ ax ] ) , the second from equation ( [ cx ] ) , and the last from equation ( [ xc ] ) . because @xmath790 there is a constant @xmath43 such that for any @xmath791 combining all previous estimates , we compute that @xmath792 thus , at this point we have shown that @xmath793 since @xmath14 is linear . on the other hand , @xmath794 by lemma [ ab ] . by lemma [ wkf ] we have that @xmath795 and so @xmath796 . [ awrz ] for any finitely generated abelian group @xmath1 , the distortion of every subgroup @xmath9 with @xmath114 in @xmath0 is polynomial . @xmath9 is undistorted if @xmath1 is finite . this follows from lemma [ grga ] and corollary [ withb ] . [ r4 ] let @xmath7 be a group having normal subgroup @xmath91 and cyclic @xmath797 . then any finitely generated subgroup @xmath9 of @xmath7 may be generated by elements of the form @xmath798 where @xmath666 . [ r5 ] let @xmath9 be a finitely generated subgroup of @xmath0 not contained in @xmath91 , where @xmath1 is finitely generated abelian . then the distortion of @xmath9 in @xmath0 is equivalent to the distortion of a subgroup @xmath812 with @xmath114 in @xmath813 where @xmath814 is also finitely generated abelian . by lemma [ r4 ] the generators of @xmath9 may be chosen to have the form @xmath815 where @xmath666 . therefore , for this value of @xmath191 we have that @xmath9 is a subgroup of @xmath802 . using the isomorphisms of lemmas [ zz ] and [ zzz ] we have that @xmath9 is a subgroup of @xmath816 generated by the image of @xmath817 under the two isomorphisms : elements @xmath818 . finally , because @xmath819<\infty$ ] we have by lemma [ wkf ] that the distortion of @xmath9 in @xmath0 is equivalent to the distortion of its image in @xmath820 . theorem [ x ] parts @xmath821 and @xmath70 follow from lemma [ abel ] if the subgroup @xmath9 is abelian . otherwise they follow from corollary [ awrz ] and lemma [ r5 ] . now we complete the proof of theorem [ x ] , part @xmath822 . let @xmath1 be a finitely generated abelian group of rank @xmath43 . consider the @xmath3-generated subgroup @xmath157 constructed as follows . let @xmath54 . consider @xmath823 . then the distortion of the polynomial @xmath352 is seen to be equivalent to @xmath64 , by lemma [ tgs ] . by lemma [ hpu ] , this means that the @xmath3-generated subgroup @xmath824 has distortion @xmath825 . the subgroup @xmath47 is a retract of @xmath0 if @xmath1 is infinite . therefore , the distortion of @xmath67 in @xmath47 and in @xmath0 are equivalent by lemma [ wkf ] . if we adopt the notation that the commutator @xmath826=aba^{-1}b^{-1}$ ] , then we see that in @xmath47 , the element of @xmath91 corresponding to the polynomial @xmath827 is @xmath828,b ] , \cdots , b]$ ] where the commutator is @xmath62-fold . this explains corollary [ xx ] .

i clarify the differences between various approaches in the literature which attempt to link gravity and thermodynamics . i then describe a new perspective based on the following features : ( 1 ) as in the case of any other matter field , the gravitational field equations should also remain unchanged if a constant is added to the lagrangian ; in other words , the field equations of gravity should remain invariant under the transformation(constant ) . ( 2 ) each event of spacetime has a certain number ( ) of microscopic degrees of freedom ( ` atoms of spacetime ' ) . this quantity is proportional to the area measure of an equi - geodesic surface , centered at that event , when the geodesic distance tends to zero . the spacetime should have a zero - point length in order for to remain finite . ( 3 ) the dynamics is determined by extremizing the heat density at all events of the spacetime . the heat density is the sum of a part contributed by matter and a part contributed by the atoms of spacetime , with the latter being . the implications of this approach are discussed .

1602.01474

the idea that gravitational field equations could be interpreted using ( or derived from ) thermodynamic arguments has been explored by many people from widely different perspectives . ( see e.g.,@xcite ) . there is a tendency in the literature to club together these very different attempts as essentially the same or , at least , as very similar . such a point of view is technically incorrect , and given this tendency , it is useful to clarify the differences between the various approaches , as regards their assumptions , physical motivation and the generality of the results . i will begin with a series of comments aimed at this task : * to begin with , one must sharply distinguish between ( i ) the attempts concerned with the _ derivation _ of the field equations by thermodynamic arguments ( like e.g. , @xcite ) and ( ii ) the attempts related to the _ interpretation _ of the field equations in a thermodynamic language ( like e.g. , @xcite ) . the latter is as important as the former because the existence of a purely thermodynamic interpretation for the field equations is vital for the overall consistency of the programme . it is rather self - defeating to derive the field equations @xmath3 from thermodynamic arguments and then interpret them in the usual geometrical language ! if gravity is thermodynamic in nature , then the gravitational field equations must be expressible in a _ thermodynamic _ language . this crucial feature has not been given due recognition in the literature . unless the final result has an interpretation in thermodynamic language , such a derivation of the field equations is conceptually rather incongruous . + as an example of what i mean by such an interpretation , let me recall the following result . it can be shown @xcite that the evolution of geometry can be interpreted in thermodynamic terms , as the heating and cooling of null surfaces , through the equation : @xmath4 where @xmath5 are the degrees of freedom in the surface @xmath6 and bulk @xmath7 of a 3-dimensional region and @xmath8 is the average davies - unruh temperature @xcite of the boundary . the @xmath9 is the induced metric on the @xmath10 constant surface , @xmath11 , and @xmath12 is the proper - time evolution vector corresponding to observers moving with four - velocity @xmath13 . the factor @xmath14 ensures the correct result for either sign of the komar energy @xmath15 . the time evolution of the metric in a region ( described by the left hand side of ) , can be interpreted @xcite as the heating / cooling of the spacetime and arises because @xmath16 . in any _ static _ spacetime @xcite , on the other hand , @xmath17 , leading to `` holographic equipartition '' : @xmath18 . this result translates gravitational dynamics into the thermal evolution of the spacetime . the validity of for all observers ( i.e. , foliations ) ensures the validity of einstein s equations . + in fact , _ no _ thermodynamic derivation of the field equations in the literature actually obtains the _ tensorial _ form of field equation @xmath19 . what is always done _ is to obtain an equation of the form @xmath20 ( where @xmath21 is either a timelike or null vector ) and postulate its validity for all @xmath21 . so it is important to understand the physical meaning of such an equation , especially the left hand side , for a given class of @xmath21 . this will be a recurrent theme which i will elaborate on later sections . * many thermodynamic derivations of field equations available in the literature , work with the assumption that the entropy of a horizon is proportional to its area ( e.g. , @xcite ) and attempt to introduce thermodynamic arguments centered around it . _ it is likely that such derivations miss some essential physics . _ the connection between gravity and thermodynamics , motivated historically from the laws of black hole mechanics @xcite and the membrane paradigm @xcite , transcends einstein s theory . in a more general class of theories , the ( wald ) entropy @xcite of the horizon is _ not _ proportional to its area . one should therefore distinguish approaches in this subject which are specially tuned to einstein gravity ( and uses the entropy - area proportionality ) from a broader class of approaches ( like e.g. @xcite ) because the latter ones , being more general , probably capture the underlying physics better . the above criticism is also valid for approaches based on entanglement entropy when it is assumed to be proportional to the horizon area . * another feature which distinguishes different approaches in the literature is whether the field equations are derived from a variational principle or from some other procedure . i am personally in favour of approaches which use a variational principle because they could offer a better window into microscopic physics . what is more , the approaches which does _ not _ use a variational principle are very limited in their scope . for example , it is virtually impossible to generalize such models beyond einstein s theory . ( in contrast , the very first approach which used a thermodynamic variational principle to derive the field equations @xcite , obtained the field equations for all models at one go . ) some of these approaches like , for example , those which use the raychaudhuri equation also have non - trivial technical issues @xcite . + even amongst the approaches which use variational principles , we need to distinguish between ( i ) those which vary the geometry ( viz . , the metric in some form , sometimes in a rather disguised manner ) and ( ii ) those which vary some auxiliary vector field , keeping the metric fixed . many approaches involving holographic concepts and entanglement entropy do vary the geometry in some form ; however , i prefer approaches which vary an auxiliary vector . ( after all , if you are going to vary the metric / geometry in an extremum principle , why not just use the einstein - hilbert action and be done with it ? ! ) an example @xcite of an extremum principle which does _ not _ vary the metric , is given by the functional @xmath22\right ) \label{qtotyy}\ ] ] here , @xmath23 and @xmath24 are the shear and expansion of a null congruence @xmath25 , @xmath26 are the shear and bulk viscous coefficients of a null fluid @xcite and the integrand can be interpreted as the rate of generation of heat ( ` dissipation without dissipation ' ; see @xcite ) due to matter and gravity on a null surface . varying @xmath27 with respect to @xmath28 and demanding that the extremum should hold for all @xmath28 ( i.e. , for all null surfaces ) will lead to einstein s equations . ( we will say more about this in sec . [ sec : dwd ] . ) such a variational principle and others of a similar genre which we will discuss later treat the geometry as fixed and does not vary the metric . * at a more fundamental level , the horizon entropy can not be finite unless some kind of discreteness exists in the spacetime near planck scales . this is clear in the case of entanglement entropy , which is a manifestly divergent quantity ( see e.g. , @xcite ) and needs to be regularized by some ad - hoc cut - off ; but it is implicit in all approaches . so , unless we have a model which captures _ at least some of _ the quantum gravitational effects on the spacetime , any derivation of the field equations using a finite value for entropy is , at best , incomplete . * finally , let me emphasize that _ gravity can not be an entropic force_. this was ably demonstrated by matt visser @xcite by an argument which uses ( essentially ) elementary vector analysis . it is trivial to prove , in the newtonian limit , that a conservative force @xmath29 can not , in general , be expressed in the entropic form @xmath30 if @xmath31 is the davies - unruh temperature that depends on the magnitude of the acceleration @xmath32 . the relation @xmath33 implies that the level surfaces of @xmath34 coincide with those of @xmath35 , allowing us to introduce a function @xmath36 . this , in turn , implies that @xmath37 and hence the level surface of @xmath34 coincide with the level surfaces of @xmath31 . but since @xmath31 depends only on @xmath32 , this requires the level surfaces of @xmath32 to coincide with those of @xmath34 . this condition is , in general , impossible to satisfy and can happen only in situations of high symmetry ( for example , spherical , cylindrical , planar etc . ) . it would be preferable if the phrase `` entropic gravity '' is _ not used as a rather generic term _ to describe the different approaches in this subject , for the simple reason that gravity can not be an entropic force . to summarize , there exist many different attempts in the literature to link gravity and thermodynamics . all of these are _ not _ equivalent either conceptually or technically and it is also likely that at least some of them are fundamentally flawed or incomplete . the approach i have been pursuing which i will describe here is marked by the following features : ( 1 ) much of it works for a wide class of theories , more general than einstein s gravity . in particular , the results hold for theories in which entropy is _ not _ proportional to horizon area . ( 2 ) the field equations are derived from a thermodynamic extremum principle in which the geometry is not varied but some other auxiliary vector field is varied . ( 3 ) the resulting field equations are interpreted in a thermodynamic language and not in a geometric language . ( 4 ) the introduction of a zero - point length to the spacetime by quantum gravitational effects allows us to provide a microscopic basis for the variational principle which is used . here , i will concentrate on developing this perspective _ from first principles _ in a streamlined manner . obviously this will require us to make some educated guesses but i shall argue that these guesses are well - motivated and the results are quite rewarding . in particular , i will describe the following two aspects : * i will demonstrate @xcite a deep connection between two aspects of gravity which are usually considered in the literature to be quite distinct . the first is the fact that gravity seems to be immune to the shift in the zero level of the energy , i.e , to the shift in the value of cosmological constant . second is the feature i mentioned above , viz . , gravitational dynamics can be reinterpreted in a purely thermodynamic language . i will show how the first feature _ leads to _ the second and , in fact , provides a simple and natural motivation to consider the heat density of the null surfaces as a key physical entity . * much of the previous work treated the spacetime as analogous to a fluid and investigated its properties in the _ thermodynamic limit_. the next , deeper , level of description of a fluid will be the _ kinetic theory _ which recognizes the discreteness and quantifies it in terms of a distribution function for its molecules . i will describe an attempt @xcite to do the same for the spacetime by introducing a distribution function for the atoms of spacetime ( which will count the microscopic degrees of freedom ) and relating it to the extremum principle which , in turn , will lead to the field equations . the principle of equivalence , along with principle of general covariance , strongly suggest that gravity is the manifestation of a curved spacetime and will set @xmath38 so that @xmath39 . occasionally , i will also set @xmath40 when no confusion is likely to arise . ] , described by a non - trivial metric @xmath41 . the _ kinematics _ of gravity , viz . how a given gravitational field affects matter , can then be determined by postulating the validity of special relativistic dynamics in all freely falling frames . this will lead to the condition @xmath42 for the energy momentum tensor of matter , which encodes the influence of gravity on matter . unfortunately , we do not have any equally elegant guiding principle to determine the _ dynamics _ of gravity , viz . how matter determines the evolution of the spacetime metric . the dynamics is contained in the gravitational field equation , which in einstein s theory is assumed to be given by @xmath3 . ( in a more general class of theories , like e.g , models , the left hand side will be replaced by a more complicated second rank , symmetric , divergence - free tensor . ) one can obtain this equation , as einstein did , by ( i ) assuming that the right hand side _ must _ be @xmath43 and ( ii ) by constructing a second rank , symmetric , divergence - free tensor from the metric containing upto second derivatives . alternatively , as hilbert did , one can write down a suitable scalar lagrangian and vary it with respect to the metric and obtain the field equations . in either procedure , one tacitly assumes that the spacetime metric is a dynamical variable with a status similar to , say , that of the gauge potential @xmath44 in electromagnetism . this belief is based on the fact that einstein s equation is a second order differential equation for the metric just as maxwell s equation is a second order differential equation for @xmath44 . it is in the same spirit that we justify varying the metric in the hilbert action ( as analogous to varying @xmath44 in the electromagnetic action ) to obtain einstein s equation . further , once we have a classical action @xmath45 $ ] in which @xmath46 is varied to get the field equations , it is tempting to think of a quantum theory , defined through a path integral over @xmath47 $ ] ( or in some other equivalent manner ) with the metric playing the role of a quantum variable . but , given the fact that spacetime geometry is conceptually very different from an external field propagating in it , this assumption viz . , that the metric is a dynamical variable similar to other fields is indeed nontrivial . further , if varying the metric in the hilbert action is _ not _ the appropriate way to obtain the classical theory , then one is forced to think afresh about all the quantum gravity programmes . interestingly enough , this textbook procedure of treating the metric as a dynamical variable , accepted without a second thought is by no means a unique way to obtain einstein s equation . in fact , it is probably _ not _ the most natural or efficient procedure . one can come up with alternative approaches _ and physically motivated extremum principles , _ leading to einstein s equation , in which the metric is _ not _ a dynamical variable . let me describe one such approach . the field equations we seek should be a relativistic generalization of newton s law of gravity @xmath48 . a natural way of generalizing this law is to begin by noticing that : ( i ) the energy density in the right hand side @xmath49 is foliation / observer dependent where @xmath50 is the four velocity of an observer . there is no way we can keep @xmath50 out of it . ( ii ) we know from the principle of equivalence that @xmath46 plays the role of @xmath51 . so a covariant , scalar generalization of the left hand side , @xmath52 , could come from the curvature tensor which contains the second derivatives of the metric . any such generalization _ must depend on the four - velocity @xmath50 of the observer _ since the right hand side does . ( iii ) it is perfectly acceptable for the left hand side _ not _ to have second _ time _ derivatives of the metric , in the rest frame of the observer , since they do not occur in @xmath52 . to obtain a scalar analogous to @xmath52 , having _ spatial _ second derivatives , we first project the indices of @xmath53 to the space orthogonal to @xmath50 , using the projection tensor @xmath54 , thereby obtaining the tensor @xmath55 . the only scalar we can get from @xmath56 is @xmath57 where @xmath58 can be thought of as the radius of curvature of the space . and @xmath58 should _ not _ to be confused with the curvature tensor @xmath59 and the curvature scalar @xmath60 of the 3-space orthogonal to @xmath50 . ] the natural generalization of newton s law @xmath48 is then given by @xmath61 . working out the left hand side ( see e.g. , p. 259 of @xcite ) and fixing the proportionality constant from the newtonian limit , one finds that @xmath62 if this scalar equation should hold for all observers ( general covariance ) then we need @xmath63 which is the standard result . demanding that @xmath64 holds for each observer , captures the geometric statement viz . that energy density curves space as viewed by any observer in a nice manner and is indeed the most natural generalization of newton s law : @xmath48 . so , in this approach , the fundamental equation determining the geometry is which should hold for all normalized , timelike vectors @xmath50 at each event of spacetime rather than the standard equation @xmath63 . while the two formulations are algebraically equivalent , they are conceptually rather different . in the conventional approach to derive @xmath63 , we do not invoke any special class of observers instead , we _ assume _ that the right hand side of the field equation _ must _ be @xmath43 and look for a generally covariant , divergence - free , second - rank tensor built from geometry to put on the left hand side . ( alternatively , we look for a scalar lagrangian made from geometrical variables ) . but the source in newtonian gravity is actually @xmath65 which _ does _ involve an extra four - velocity for its definition . if we introduce observers with four - velocity @xmath50 and in the end demand that the equation should hold for all @xmath50 we obtain the same gravitational field equations by a different route . this approach to _ dynamics _ brings it closer to the way we handled the _ kinematics _ by introducing the freely falling observers . the real importance of this approach stems from the fact that it allows us to construct a different kind of extremum principles which will lead to the gravitational field equations , _ without treating the metric as a dynamical variable ! _ since this approach introduces an extra vector field @xmath66 into the fray , one can consider an extremum principle in which we vary @xmath50 which makes physical sense in terms of changing the observer instead of the metric . it is now possible , for example , to obtain the field equations by varying @xmath50 in a variational principle with the lagrangian @xmath67 and demanding that the extremum must hold for all @xmath50 . in fact , we can also use the lagrangian @xmath68 . varying @xmath66 in the resulting action , after imposing the constraint @xmath69 and demanding that the extremum should hold for all @xmath50 , will lead to the equation @xmath70 where @xmath71 is the lagrange multiplier . using the bianchi identity and @xmath42 , we will recover the field equations except for an undetermined cosmological constant @xcite . removing a total divergence from @xmath72 , we see that this is equivalent to a variational principle based on the functional term ) but , of course , here we are varying the vector field @xmath50 and _ not _ the metric @xmath73 . in fact , one can add any functional of the metric to the lagrangian and it would make no difference since the metric is not varied . ] @xmath74 = \int d^4 x \ , \sqrt{-g}\ , \left[(\nabla_iu^i)^2 - \nabla_ju^i\nabla_iu^j - 8\pi \rho\right ] \label{tracekaction}\ ] ] varying @xmath75 in @xmath76 $ ] and demanding the extremum to hold for all @xmath50 will lead to except for an undetermined cosmological constant . in an extremum principle , you get an evolution equation for @xmath77 . here we vary @xmath75 in but get the equation constraining @xmath46 ! this comes about because , after varying @xmath75 , we demand that the equation hold for all @xmath75 to take care of all observers . ( recall that this is also done in _ all _ attempts to derive field equations by thermodynamic arguments . ) while conceptually different from the usual extremum principles , it is perfectly well - defined and makes physical sense . ] so , one can indeed obtain the classical field equations for gravity _ without varying the metric _ in any action principle . the existence of such alternative variational principles takes away the motivation to treat the metric as a dynamical variable either classically or quantum mechanically . if this alternative procedure or a variant of it is the correct interpretation of classical gravity , then the hilbert action has no meaning classically ( and hence in a quantum mechanical path integral ) . in the classical theory , what ultimately matters is the field equation and the rest is just window dressing . but the distinction between these two approaches is vital when we want to bring together the principles of quantum theory and gravity . if the metric is not a dynamical variable in the classical theory and the correct classical variational principle involves varying some other auxiliary variable like @xmath50 rather than the metric it makes no sense to quantise the metric . such an attempt will , at best , be similar to quantizing the velocity or density field of a material medium . while gravitons will emerge with the same conceptual status as , say , phonons , the attempt will not lead to a complete quantum description of the spacetime . so , the alternative paradigm suggests a completely different picture about the quantum nature of spacetime . this point of view , that the metric is not a dynamical variable , receives independent support from the tantalizing relationship between gravitational dynamics and the thermodynamics of null surfaces . as i mentioned earlier , one can provide a purely thermodynamic interpretation to quite easily but _ not _ to @xmath3 . we need the extra vector field @xmath66 for this interpretation , just as we need it to define an energy density . we will see later that the situation is still better when we use a null vector in place of the timelike vector . one major problem with the conventional approach to gravitational dynamics is that we lack a good guiding principle to determine the field equations . my first task is to take care of this . i will begin by postulating a guiding principle @xcite , which turns out to be as powerful as the principle of equivalence , in obtaining the gravitational dynamics . let us recall that the equations of motion for matter , derived from an action principle , remain invariant if we add a constant to the matter lagrangian , _ i.e. _ , under the change @xmath78 constant . this encodes the principle that the zero level of energy density does not affect the dynamics . motivated by this fact , it seems reasonable to postulate that the gravitational field equations should not break this symmetry , which is already present in the matter sector . since @xmath79 will occur , in one form or another , as the source for gravity ( as can be argued from the principle of equivalence and considerations of the newtonian limit ) , we postulate that : * the extremum principle that determines the dynamics of spacetime must be invariant under the change @xmath80 ( constant ) @xmath81 . this principle leads to two useful results : first , this principle rules out the possibility of varying the metric tensor @xmath46 in a covariant , local , action principle to obtain the field equations ! it is easy to prove @xcite that if ( i ) the action is obtained from a local , covariant lagrangian integrated over a region of spacetime with the covariant measure @xmath82 and ( ii ) the field equations are obtained through unrestricted variation fixed ; i do not think we have a sound physical motivation for this approach . ] of the metric in the action , then the field equations _ can not _ remain invariant under @xmath80 ( constant ) @xmath81 . in fact , @xmath78 constant is no longer a symmetry transformation of the action if the metric is treated as the dynamical variable . therefore , any variational principle we come up with can not have @xmath46 as the dynamical variable . as i highlighted in the last section , this need not scare us ; it is certainly possible to come up with variational principles leading to the field equations in which the metric is not a dynamical variable . this brings us to the second result : the most natural structure , built from @xmath43 , which maintains the required invariance under @xmath80 ( constant ) @xmath81 , is given by @xmath83 \equiv t_{ab } \ell^a \ell^b \label{qtot}\ ] ] where @xmath84 is a null vector .- dimensional spacetime , the null vector with @xmath85 degrees of freedom is the minimum we need . in contrast , if we use , say , a combination @xmath86 with a symmetric traceless tensor @xmath87 , in order to maintain the invariance under @xmath80 ( constant ) @xmath81 , then we need to introduce @xmath88 degrees of freedom ; in @xmath89 , this introduces nine degrees of freedom , which is equivalent to introducing three null vectors rather than one . ] this is very similar to the right hand side of except that we are now using a _ null _ vector @xmath84 rather than a _ timelike _ vector @xmath50 to implement our guiding principle . demanding that the equation @xmath90 holds for all null vectors @xmath84 at each event will also lead to @xmath91 ( where @xmath92 is the undetermined ) and our guiding principle points towards such an approach . while the right hand side of ( @xmath93 ) changes under @xmath80 ( constant ) @xmath81 , the right hand side of ( @xmath94 ) remains invariant , which is what we want . but , @xmath95 has a clear physical meaning as the energy density measured by an observer with velocity @xmath50 , but the physical meaning of @xmath96 \equiv t_{ab } \ell^a \ell^b$ ] is not obvious . our next task is to clarify that . in the case of an ideal fluid , with @xmath97 , the combination @xmath94 is actually the _ heat density _ @xmath98 where @xmath31 is the temperature and @xmath99 is the entropy density of the fluid . ( the last equality follows from gibbs - duhem relation and we have chosen the null vector with @xmath100 for simplicity . ) the invariance of @xmath101 under @xmath80 ( constant ) @xmath81 reflects the fact that the cosmological constant , with the equation of state @xmath102 , has zero heat density . our guiding principle , as well as , suggests that _ it is the heat density rather than the energy density which is the source of gravity . but @xmath103 is the energy density for _ any _ kind of @xmath43 , not just for that of an ideal fluid . how do we interpret @xmath94 in a general context when @xmath43 could describe any kind of source not necessarily a fluid for which concepts like temperature and entropy do not exist intrinsically ? _ remarkably enough , this can be done_. in any spacetime , around any event , there exists a class of observers ( called a local rindler observers ) who will interpret @xmath94 as the heat density contributed by the matter to a null surface which they perceive as a horizon . this _ motivates _ us to introduce @xcite the concept of local rindler frame ( lrf ) and local rindler observers which will allow us to provide a thermodynamic interpretation of @xmath94 for any @xmath43 . this arises as follows : . light rays travelling at 45 degrees in the local inertial frame define the light cones at @xmath104 . ( b ) right : a local rindler observer who is accelerating with respect to the inertial observer . for a sufficiently large acceleration , the trajectory of such an observer will be close to the light cones emanating from @xmath104 . the local rindler observer will perceive the light cone as a local rindler horizon and attribute to it a temperature given by . in other words , the vacuum fluctuations of the local inertial frame will appear as thermal fluctuations in the local rindler frame . ] in a region around any event @xmath104 , we first introduce the freely falling frame ( fff ) with coordinates @xmath105 . next , we boost from the fff to a local rindler frame ( lrf ) with coordinates @xmath106 constructed using some acceleration @xmath107 , through the transformations : @xmath108 when @xmath109 and similarly for other wedges . one of the null surfaces passing though @xmath104 , will get mapped to the @xmath110 surface in the fff and will act as a patch of horizon to the @xmath111 constant rindler observers . this construction leads to a beautiful result @xcite in quantum field theory . the local vacuum state , defined by the freely - falling observers around an event , will appear as a thermal state to the local rindler observers with the temperature : @xmath112 where @xmath107 is the acceleration of the local rindler observer , which can be related to other geometrical variables of the spacetime in different contexts [ see fig . [ fig : daviesunruh ] ] . the existence of the davies unruh temperature tells us that around _ any _ event , in _ any _ spacetime , there exists a class of observers who will perceive the spacetime as hot . let us now consider the flow of energy associated with the matter that crosses the null surface . nothing unusual happens when this is viewed in the fff by the locally inertial observer . but the local rindler observer attributes a temperature @xmath31 to the horizon and views it as a hot surface . such an observer will interpret the energy @xmath113 , dumped on the horizon , by the matter that crosses the null surface , as energy added to a _ hot _ surface , thereby contributing a _ heat _ content @xmath114 . ( recall that , as seen by the outside observer , matter actually takes an infinite amount of time to cross a _ black hole _ horizon , thereby allowing for thermalization to take place . similarly , a local rindler observer will find that the matter takes a very long time to cross the horizon . ) to compute @xmath113 in terms of @xmath43 , note that the lrf provides us with an approximate killing vector field @xmath115 , generating the lorentz boosts , which coincides with a suitably defined null normal @xmath116 at the null surface . the heat current arises from the energy current @xmath117 of matter and hence the total heat energy dumped on the null surface will be : @xmath118 where we have used the result that @xmath119 on the null surface . so we find that constant and treating the light cone as the degenerate limit @xmath120 of the hyperboloids . we set @xmath121 and take the corresponding limit . the motivation for this choice will become clearer later on . ] \frac{dq_{m}}{\sqrt{\gamma}d^{2}xd\lambda}=t_{ab } \ell^a\ell^b \label{hmatter}\ ] ] can indeed be interpreted as the heat density ( energy per unit area per unit affine time ) of the null surface , contributed by matter crossing a local rindler horizon , as interpreted by the local rindler observer . this interpretation works in the lrf irrespective of the nature of @xmath43 . so , the need to work with @xmath122 , forced on us by our guiding principle , _ leads to _ the introduction of local rindler observers in order to interpret this quantity as the heat density . there is an alternative interpretation of @xmath122 which will prove to be useful . since the parameter @xmath123 ( defined through @xmath124 ) is similar to a time coordinate , we can also think of @xmath96 $ ] in as the heat generated per unit area of null surface per unit time . but since there are null surfaces through any event in the spacetime , we will always have observers who see the matter heating up these surfaces ! this is something we probably do _ not _ want and we will see later on that gravity comes to our rescue . as we will see , contribution to the heating from the microscopic degrees of freedom of the spacetime precisely cancels out @xmath96 $ ] on all the null surfaces .. i do _ not _ introduce the notion of entropy for the rindler horizon ( as proportional to its area ) or work with its variation . ] let us get back to the task of constructing an extremum principle from which we can obtain the field equations . we have argued that the matter sector appears in the extremum principle through the combination @xmath125 which has the interpretation of the heat density ( or the heating rate ) contributed to a null surface by the matter crossing it . we also saw that we can not vary the metric in the extremum principle . but in any variational principle constructed from @xmath122 , we now have the option of varying @xmath84 and leaving the metric alone . such a variational principle can take the form : @xmath126+\mathcal{h}_g[x^i , \ell_a]\right ) ; \qquad \mathcal{h}_m[x^i , \ell_a ] \equiv t_{ab}(x ) \ell^a \ell^b \label{qtot1}\ ] ] where we will interpret @xmath127 as the contribution to the heat density from the microscopic degrees of freedom of geometry ( ` atoms of space ' ; i will use these two phrases interchangeably ) . this should depend on both @xmath128 and @xmath84 for the variational principle to be well defined . the success of this approach depends on our coming up with a candidate for @xmath129 $ ] which is physically well - motivated and also depends on the null vectors at each event . i will now show how this can be achieved . since @xmath127 has the dimension of energy density , it is convenient to write it as @xmath130 where we have introduced a length scale @xmath131 to set the dimensions and @xmath1 denotes the number of atoms of space at an event @xmath128 with an extra attribute characterized by a null vector @xmath84 . so the total heat density now becomes : @xmath132\right ) \label{qtot2}\ ] ] our next task is to determine @xmath1 . it appears natural to assume that the number of atoms of space , @xmath1 , ( _ i.e. _ , the microscopic degrees of freedom , contributing to the heat density ) at an event @xmath104 should be proportional to either the area or volume ( which are the most primitive constructs ) we can `` associate with '' the event @xmath104 . so we need to give precise meaning to the phrase , `` area or volume associated with '' the event @xmath104 . to do this we first introduce the notion of equi - geodesic surface , which can be done either in the euclidean sector or in the lorentzian sector ; we will work in the euclidean sector . an equi - geodesic surface @xmath133 is the set of all events at the same geodesic distance @xmath134 from some event , which we take to be the origin @xcite . a natural coordinate system to describe such a surface is given by @xmath135 where @xmath134 , the geodesic distance from the origin , acts as the `` radial '' coordinate and @xmath136 are the angular coordinates on the equi - geodesic surfaces corresponding to @xmath137 constant . the metric in this coordinate system becomes : @xmath138 where @xmath139 is the induced metric being the angular coordinates . ] on @xmath133 . the most basic quantities we can now introduce are the volume element @xmath140 in the bulk , and the area element for @xmath133 given by @xmath141 . for the metric in , @xmath142 , and hence , both these measures are identical . using standard differential geometry , one can show @xcite that , in the limit of @xmath143 , either of these is given by : @xmath144 where @xmath145 is the normal to @xmath133 and @xmath146 arises from the standard metric determinant of the angular part of a unit sphere . the second term containing @xmath147 gives the curvature correction to the area of ( or the volume enclosed by ) an equi - geodesic surface . is a standard result in differential geometry and is often presented as a measure of the curvature at any event . we can now `` associate '' an area ( or volume ) with an event @xmath148 in a natural way by the following limiting procedure : ( i ) construct an equi - geodesic surface @xmath133 centered on @xmath148 at a geodesic distance @xmath134 ; ( ii ) compute the volume enclosed by @xmath133 and the surface area of @xmath133 ; and ( iii ) take the limit of @xmath120 to define the area ( or volume ) associated with @xmath148 . however , as we can readily see from , these measures vanish in the limit of @xmath143 . this is , however , to be expected . the existence of microscopic degrees of freedom requires some form of discrete structure in the spacetime and they can not be meaningfully defined if the spacetime is treated as a continuum all the way , just as we can not have a finite number of molecules in a fluid if it is treated as a continuum all the way . classical differential geometry , which is what we have used so far , knows nothing about any discrete spacetime structure and hence can not give us a nonzero @xmath127 . to obtain a nonzero @xmath127 from the above considerations , we need to ask how the geodesic interval and the spacetime metric get modified in a quantum description of spacetime and whether such a modified description will have a @xmath149 ( or @xmath150 ) which does not vanish in the coincidence limit . i will now turn to this task . there is a fair amount of evidence ( see e.g. , @xcite ) to suggest that a primary effect of quantum gravity will be to introduce into the spacetime a zero - point length , by modifying the geodesic interval @xmath151 between any two events @xmath152 and @xmath153 ( in a euclidean spacetime ) to a form like @xmath154 where @xmath155 is a length scale of the order of planck length . where the function @xmath156 satisfies the constraint @xmath157 with @xmath158 finite . our results are insensitive to the explicit functional form of @xmath156 . so , for the sake of illustration , we will use @xmath159 . ] while we do not know how quantum gravity modifies the classical metric , we have an indirect handle on it if we assume that quantum gravity introduces a zero point length to the spacetime . this arises as follows : just as the original @xmath160 can be obtained from the original metric @xmath46 , we would expect the quantum gravity - corrected geodesic interval @xmath156 to arise from a corresponding quantum gravity - corrected metric @xcite , which we will call the q - metric @xmath161 . but no such local , non - singular @xmath161 can exist because , for any such @xmath161 , the resulting geodesic interval will vanish in the coincidence limit , by definition of the integral . therefore , we expect @xmath162 to be a bitensor , which should be singular at all events in the coincidence limit @xmath163 . we can determine @xcite its form by using two conditions : ( i ) it should lead to a geodesic interval @xmath156 and ( ii ) a green s function describing small metric perturbations should have a finite coincidence limit . ( alternatively , we can use the criterion that the flat metric should lead to a flat q - metric . ) these conditions determine @xmath161 uniquely @xcite in terms of @xmath46 ( and its associated geodesic interval @xmath160 ) . we find that : @xmath164 with @xmath165 where @xmath166 is the spacetime dimension , @xmath167 . the @xmath168 is the van - vleck determinant related to the geodesic interval @xmath169 by : @xmath170 and @xmath171 is the corresponding quantity computed by replacing @xmath172 by @xmath173 ( and @xmath46 by @xmath161 in the covariant derivatives ) in the above definition . for our purpose of determining @xmath127 , we will compute the area element ( @xmath174 ) of an equi - geodesic surface and the volume element ( @xmath175 ) for the region enclosed by it , using the renormalized q - metric . ( for the q - metric in , corresponding to the @xmath46 in , these two measures will not be equal , because @xmath176 . ) if our ideas are correct , we should get a non - zero limit and there must be a valid mathematical reason to prefer one of these measures over the other . the computation is straightforward and we find ( for in @xmath177 , though similar results @xcite hold in the more general case in @xmath166 dimensions ) that : @xmath178\sqrt{h_\omega } \label{qfinal}\end{aligned}\ ] ] and : ) arises from the standard result , by the simple replacement of @xmath179 . but note that this replacement does _ not _ work for the result in ( which is @xmath180 ) due to the @xmath181 factor which has the limiting form @xmath182 when @xmath120 . this is why each event has zero volume , but finite area associated with it!. a possible insight into this curious feature is provided by the following fact : the leading order dependence of @xmath183 shows that the volumes scale as @xmath160 while the area measure is finite . this , in turn , is related to the fact @xcite that _ the effective dimension of the renormalized spacetime becomes @xmath184 close to planck scales , _ independent of the original @xmath166 . this result has been noticed by several people ( @xcite ; also see @xcite ) in different , but specific , models of quantum gravity . our approach leads to this result in a _ model - independent _ manner , which , in turn , is connected with the result that events have zero volume , but non - zero area . ] @xmath185\sqrt{h_\omega } \label{hfinal}\end{aligned}\ ] ] when @xmath186 , we recover the standard result in , as we should . our interest , however , is in the limit @xmath187 at finite @xmath155 . something remarkable happens when we take this limit . the volume measure @xmath188 vanishes ( just as for the original metric ) but @xmath149 has a non - zero limit : @xmath189\sqrt{h_\omega } \label{hlimit}\end{aligned}\ ] ] the q - metric ( which we interpret as representing the renormalized / dressed spacetime ) attributes to every point in the spacetime a finite area measure , but a zero volume measure ! since @xmath190 is the volume measure of the @xmath191 surface , we define @xcite the dimensionless density of the atoms of spacetime , contributing to the gravitational heat density , as : @xmath192 so far we have been working in the euclidean sector with @xmath145 being normal to the equi - geodesic surface . the limit @xmath120 in the euclidean sector makes the equi - geodesic surface shrink to the origin . _ but , in the lorentzian sector this leads to the null surface which acts as the local rindler horizon around this event . _ so , in this limit , we can identify @xmath193 with the normal to the null surface @xmath84 and express the gravitational heat density as @xmath194 to see this in some detail , consider the euclidean version of the local rindler frame . one can extend the null line ( 45 degree line in fig . [ fig : daviesunruh ] ) to the null _ plane _ @xmath195 in spacetime and similarly for the hyperboloid . alternatively , one can extend the null line to the null _ cone _ by @xmath196 with @xmath197 and the hyperboloid @xmath198 constant will go ` around ' the null cone in the spacetime ( see the left part of fig . [ fig : lightcones ] ) . observers living on this hyperboloid will use their respective ( rotated ) @xmath199 axis to study the rindler frame physics . ] the local rindler observers , living on the hyperboloid @xmath200 perceive local patches of the light cone @xmath201 as their horizon ( see the left half of fig . [ fig : lightcones ] ) . if we now analytically continue to the euclidean sector , the hyperboloid @xmath200 will become a sphere @xmath202 ( see the right half of fig . [ fig : lightcones ] ) . the light cones @xmath201 goes over to @xmath203 and collapses into the origin . so , taking the limit @xmath204 in the euclidean sector corresponds to approaching the local rindler horizons in the lorentzian sector . this is the limit in which the hyperboloid degenerates into the light cones emanating from @xmath104 . the normal @xmath193 to the euclidean sphere can then be identified with the normal to the null surface @xmath84 . the dependence of @xmath1 on @xmath193 in the euclidean equi - geodesic surface is what translates into its dependence on the null normal @xmath84 in the lorentzian sector . , with a normal @xmath84 . the local rindler observers live on the hyperboloid @xmath205 constant around these light cones and perceive a patch of the light cone as a local rindler horizon with a non - zero temperature . the arrow denotes the normal to the hyperbola . ( b ) right : in the euclidean sector , the hyperboloid @xmath206 will become a sphere @xmath202 and the normal to the hyperboloid becomes the normal to the sphere . the light cone @xmath201 will go over to @xmath203 which collapses into the origin . the limit @xmath204 in the euclidean sector corresponds to approaching the rindler horizon in the lorentzian sector . in this limit the hyperboloid degenerates into the light cones emanating from @xmath104 . the direction of the normal to the sphere becomes ill - defined in the euclidean sector when the radius of the sphere shrinks to zero . in the lorentzian sector , we can take it to be the normal to the null surface in the limit when the hyperboloid degenerates to the light cone . the dependence of @xmath1 on @xmath193 in the euclidean equi - geodesic surface is what translates into its dependence on the null normal @xmath84 in the lorentzian sector . ] so the contribution to the _ gravitational _ heat density on a null surface in is obtained by integrating @xmath207 over the volume : @xmath208 \label{corres}\ ] ] the numerical factor in front of the second term depends on the ratio @xmath209 which we expect to be of order unity , if @xmath131 is the standard planck length . this ratio can not be determined without better knowledge of quantum gravity . but we get the correct equations with the choice @xmath210 , which we will make . where @xmath211 is a factor of order unity , the numerical coefficients will change . we take @xmath212 as a natural choice . ] the full variational principle is then based on the functional : @xmath213)\nonumber\\ & = & \int \sqrt{\gamma}\ , d^2x \ , d\lambda\,\left [ \frac{1}{l_p^4}+\left\{t^a_b-\frac{1}{8\pi l_p^2}r^a_b\right\}\ell_a\ell^b \right ] \label{qtot3}\end{aligned}\ ] ] extremising this functional with respect to @xmath84 after introducing a lagrange multiplier to keep @xmath214 , and demanding the extremum holds for all @xmath84 at an event again leads to the result @xmath215 where @xmath71 is the lagrange multiplier . using the bianchi identity and @xmath42 , we will recover the field equations except for a cosmological constant . ( there is actually a way to determine its value , which i will discuss in sec . [ sec : cc ] . ) there are several points which are noteworthy about this result which i will now comment upon : we defined the heat density of gravity as @xmath216 with the number of atoms of space @xmath1 being given by the limit : @xmath217 in a renormalized spacetime with a zero - point length . the result had the combination @xmath218 , at the relevant order , which is crucial . further , this term came with a _ minus sign _ without which the programme would have failed . these results bring to the center - stage the geodesic interval @xmath151 ( rather than the metric ) as the proper variable to describe spacetime geometry @xcite . in a classical spacetime , @xmath151 and @xmath41 contain the same amount of information and each is derivable from the other . but @xmath151 seems to be better suited to take into account quantum gravitational effects to a certain extent . we mentioned earlier that one could have also thought of @xmath122 as the heating _ rate _ of the null surface by matter . we see that the corresponding heating rate of the null surface by the atoms of space precisely compensates this on - shell leaving an unobservable constant factor @xmath219 . ( since we can add any quantity independent of @xmath84 to the integrand of @xmath220 , we can even renormalize this away ; we leave it there because it has an interesting implication for the ; see sec . [ sec : cc ] ) . so the field equation , which is now in the form of , has a clear physical meaning ; both sides represent the heating rate of null surfaces and einstein s equation arises as a heat balance equation . this interpretation is reinforced by the fact that the term @xmath221 is related to the `` dissipation without dissipation '' @xcite of the null surfaces , which can be described as follows : introduce the second null vector @xmath222 and define the 2-metric on the cross - section of the null surface in the standard manner , @xmath223 . next define the expansion @xmath224 and shear @xmath225 where @xmath226 . ( we take the null congruence to be affinely parametrized . ) one can then show that ( see e.g. , eq(a60 ) of @xcite ) : @xmath227 + \frac{1}{8\pi l_p^2}\frac{1}{\sqrt{\gamma}}\frac{d}{d\lambda}(\sqrt{\gamma}\theta)\equiv\mathcal{d}+\frac{1}{8\pi l_p^2}\frac{1}{\sqrt{\gamma}}\frac{d}{d\lambda}(\sqrt{\gamma}\theta ) \label{rai}\ ] ] where @xmath228 are the shear and bulk viscous coefficients @xcite and @xmath229 is the viscous dissipation rate . so , ignoring the total divergence term and the constant , the relevant part of @xmath220 in can be expressed as @xmath230\right ) = \int \sqrt{\gamma}\ , d^2x \ , d\lambda\ , \left(t^a_b(x ) \ell_a\ell^b + \mathcal{d}\right ) \label{qtoty}\ ] ] both terms now have an interpretation of the rate of heating ( due to matter or atoms of spacetime ) . ) is just a restatement of the raychaudhuri equation . what is relevant in the extremum principle are the _ terms in shear and expansion , while the term giving the change in the cross - sectional area of the congruence is a total divergence and is irrelevant . this tells us that ignoring the quadratic terms of the raychaudhuri equation can miss a key element of physics ; see also @xcite . ] our extremum principle can be thought of extremising the rate of production of heat on the null surface . the combination @xmath231 , which is crucial to our results , arises in several other geometrical quantities just as it came up in the area measure @xmath149 in . in particular , this combination has some relevance to entanglement entropy , in the following sense . in the literature , it is usually claimed that the entanglement entropy @xmath35 of a field partitioned by , say , a horizon is proportional to the area of the horizon . _ to be precise , this statement is meaningless because the entanglement entropy @xmath35 is a divergent quantity . _ for example , for a free , massless , scalar field in @xmath166 dimensional spacetime , @xmath35 is given by the expression : @xmath232 where @xmath233 is the transverse area and @xmath234 is the schwinger proper time kernel @xcite . in the coincidence limit , the @xmath235 and the integral in diverges as @xmath236 at the lower limit , where @xmath237 is a lower cutoff length scale . in @xmath177 spacetime , this gives @xmath238 , which diverges quadratically , which is the standard result . the introduction of a zero - point length into the spacetime corresponds to introducing a regulator @xmath239 into the heat kernel ( with some cutoff scale @xmath240 ) _ which will render the entanglement entropy finite _ @xcite and these calculations meaningful . what is more , the kernel will also pick up a van vleck determinant in a curved spacetime which also contains the factor @xmath231 over and above the flat spacetime result . _ this suggests that one may be able to relate @xmath241 in to the entanglement entropy which could provide an alternative interpretation to our results . _ this idea could be tested by computing the entanglement entropy of a field in the renormalized spacetime with zero - point length , which renders it finite in a systematic manner . the coupling between spacetime geometry and matter is now through a vector field @xmath84 and at the lowest order they couple to @xmath84 through the terms @xmath242 and @xmath101 respectively . the physical origin of these two couplings is quite distinct . the @xmath101 arises from the behaviour of matter crossing the local rindler horizon and the @xmath84 in this expression represents the normal to the local rindler horizon . the @xmath242 term , however , arose from the limit of the area measure @xmath149 in a spacetime with a zero - point length . the @xmath193 in this case is identified with the normal @xmath84 to the null surface through a limiting process when one takes the limit @xmath143 in the euclidean sector . this , in turn , depends on the fact that the condition @xmath243 will lead to @xmath244 in the euclidean space while it will describe all events connected by a null ray in the lorentzian space . more generally , we can introduce a vector @xmath245 that couples to spacetime curvature through @xmath246 and to matter through @xmath247 , thereby providing an indirect coupling between matter and geometry . this procedure is conceptually rather satisfactory . in the conventional approach to einstein s theory we actually do not have a _ mechanism _ which tells us how @xmath43 ends up curving the spacetime . the relation @xmath3 equates apples and oranges ; the left hand side is purely geometrical while the right hand side is made of matter with a large number of discrete ( quantum ) degrees of freedom . an equation of the kind , @xmath248 , on the other hand , does better in this regard . we can hope to interpret both sides independently ( say , as the heat densities ) and think of this equation as a balancing act performed by spacetime . such a description is reinforced by the extremum principle in in which both @xmath246 and @xmath249 can be thought of as distorting the value of @xmath1 from unity , and the gravitational field equations restore the value @xmath250 on - shell . ( i will say a little more about this in the last section . ) since classical gravity is obtained by extremising @xmath27 in with respect to @xmath193 , we could ask whether it is meaningful to attempt a quantum theory for @xmath193 using a path integral . this would require evaluating the path integral @xmath251 \ , \exp i q_{\rm tot } \left [ n_a\right]\ ] ] incorporating @xmath252 $ ] through a functional fourier transform with respect to a lagrange multiplier field @xmath253 $ ] , we can reduce this to the form : @xmath254_x \propto \prod_x\int \mathcal{d } \lambda(x)\ ; { \rm det}^{-1/2 } \left [ m^a_b + \lambda \delta^a_b \right]_x\ ] ] where @xmath255 . this is a rather intriguing theory which will lead to einstein s equations in the saddle point limit if we demand that the saddle point condition should hold for all @xmath193 . the euclidean version of this theory suggests working with @xmath256 and it will be interesting to explore this further . where @xmath257 is an affinely parametrized null normal , then we can upgrade just @xmath258 to a dynamical field with the lagrangian @xmath259 $ ] . integrating over @xmath258 , treating it as a rapidly varying fluctuation ( with @xmath260 $ ] approximately constant ) will lead to an effective action @xmath261 for @xmath260 $ ] such that @xmath262 is the same as @xmath263 . i will describe this in detail elsewhere . ] on shell , when the equations of motion hold , the two terms in the curly brackets in cancel each other and the net heat density has the planckian value @xmath264 , which , of course , has no gravitational effect . but it tells us that there _ is _ a zero - point contribution to the degrees of freedom in spacetime , which , in dimensionless form , is just unity . therefore , it makes sense to ascribe @xmath265 degrees of freedom to an area @xmath266 , which is consistent with what we know from earlier results in this subject . so a two - sphere @xmath267of radius @xmath131 has @xmath268 , which was the crucial input that was used in a previous work to determine the numerical value of the for our universe . ( this is similar to assigning @xmath269 molecules to a phase volume . in kinetic theory , we do not worry about the fact that @xmath1 is not always an integer . in the same spirit , we are not concerned by the fact that @xmath270 is not an integer . ) . thus , the microscopic description does allow us to determine @xcite the value of the , ( which arose as an integration constant ) , as it should in any complete description . let me elaborate a little bit on this aspect , since it can provide a solution to what is usually considered the most challenging problem of theoretical physics today . observations suggest that our universe has three distinct phases of expansion : ( i ) an inflationary phase with an approximately constant density @xmath271 , fairly early on . ( ii ) a phase dominated , first by radiation and then by matter , with @xmath272 $ ] , where @xmath273 is the epoch at which the matter and radiation densities were equal , @xmath274 and the @xmath275 is a constant equal to the energy density of either matter or radiation at @xmath276 . ( iii ) an accelerated phase of expansion at late stages , driven by the energy density @xmath277 of the cosmological constant . these three constants @xmath278 $ ] completely specify the dynamics of our universe and act as its signature . of these , @xmath271 and @xmath275 can , in principle , be determined by standard high energy physics . but we need a _ new _ principle to fix the value of @xmath277 , which is related to the integration constant that appears in the field equations in our approach . it turns out that such a universe , with these three phases , harbors a new _ conserved _ quantity , which is the number @xmath279 of length scales ( or radial geodesics ) , that cross the hubble radius during any of these phases @xcite . any physical principle which can determine the value of @xmath279 during the radiation - matter dominated phase , say , will fix the value of @xmath277 in terms of @xmath280 $ ] . taking the modes into the very early phase , we can fix the value of this conserved quantity @xmath279 at the planck scale , as the degrees of freedom in a two - sphere of radius @xmath131 . in other words , we take @xmath281 . this , in turn , leads to a remarkable prediction relating the three densities @xcite : @xmath282 from cosmological observations , we know that @xmath283 ; if we assume that the range of the inflationary energy scale is @xmath284 gev , we get @xmath285 , which is consistent with the observations ! this approach for solving the cosmological constant problem provides a unified view of cosmic evolution , connecting the three phases through in contrast with standard cosmology in which the three phases are joined together in an unrelated , _ ad hoc _ manner . moreover , this approach to the cosmological constant problem _ makes a falsifiable prediction _ , unlike any other approach i am aware of . from the observed values of @xmath277 and @xmath286 , we can _ predict _ the energy scale of inflation within a very narrow band to within a factor of about five if we include the ambiguities associated with reheating . if future observations show that inflation occurred at energy scales outside the band of @xmath287 gev , our model for explaining the value of is ruled out . let me reiterate the logical sequence described in this work which leads to a completely different perspective on gravity and the derivation of its field equations . * we postulate that the gravitational field equations should arise from an extremum principle which remains invariant under the transformation @xmath0(constant ) . * this leads to two conclusions : ( a ) the metric tensor can not be a dynamical variable which is varied in the extremum principle . ( b ) the @xmath43 should appear in the extremum principle though the combination @xmath288 where @xmath84 is a null vector . * we next look for a physical interpretation of @xmath289 for an arbitrary @xmath43 and find that , in any spacetime , the local rindler observers will interpret @xmath289 as the heat density contributed to a null surface by the matter crossing it . this interpretation works for _ any _ @xmath43 and provides a strong _ motivation _ for introducing local rindler observers in the spacetime . * since ( i ) the metric can not be a dynamical variable and ( ii ) we now have the auxiliary null vector field @xmath84 arising through @xmath289 , we look for an extremum principle in which @xmath84 is varied . the extremum should hold for all null vectors at any event and constrain the background metric . * we take the integrand of the extremum principle to be @xmath290 where @xmath291 is interpreted as the heat density due to the microscopic spacetime degrees of freedom . we have introduced a length scale @xmath131 from dimensional considerations and @xmath1 is the dimensionless count of the microscopic degrees of freedom . * a discrete count for the microscopic degrees of freedom implies a discrete nature for spacetime at planck scales . we incorporate this fact by an effective , renormalized / dressed metric @xmath161 which ensures that the geodesic distance of the effective metric has a zero - point length and is given by @xmath292 where @xmath293 is the geodesic distance of the classical spacetime . we take @xmath294 . * the number of spacetime degrees of freedom , at an event , is taken to be proportional to the area of an equi - geodesic surface centered at that event in the limit of vanishing geodesic distance ( see ) . with this choice , one obtains an extremum principle based on . varying this with respect to all null vectors @xmath84 and demanding the equation to hold for all @xmath84 at an event leads to einstein s equation with an undetermined cosmological constant . * when equations of motion hold , we can assign @xmath265 degrees of freedom with every area element @xmath266 in spacetime . this , in turn , allows us to fix the value of the undetermined correctly and _ provides a solution to the cosmological constant problem_. the approach outlined here is based on the idea that gravity is the thermodynamic limit of the statistical mechanics of certain microscopic degrees of freedom ( ` atoms of space ' ) . in the thermodynamic limit , deriving the field equations of classical gravity is algebraically straightforward one might even say trivial , but let us not shun simplicity ! it is obtained from an extremum principle based on the functional : @xmath295 varying @xmath84 in @xmath296 with the constraint that @xmath214 , and demanding that the result holds for all @xmath84 will lead to einstein s equation with an arbitrary cosmological constant . the approach works for a wild class of gravitational theories including the models . as far as classical gravity goes , that is the end of the story . but we could enquire about the physical meaning of this extremum principle . the combination @xmath297 is invariant under the shift @xmath298 ( constant ) which was the original reason to put it in the variational principle . postulating that the extremum principle must be invariant under the transformation @xmath0(constant ) naturally _ leads to _ the introduction of @xmath94 in the variational principle and to the local rindler observers for its interpretation . as i mentioned earlier , this by itself is a valuable insight and shows the connection between two features viz . the immunity of gravity to shifts in the cosmological constant , and the thermodynamic interpretation of gravity which were considered as quite distinct . further , @xmath94 _ does _ have the interpretation as the heat density of matter on any null surface . it is also possible to interpret @xmath299 as the heating rate ( `` dissipation without dissipation '' ; see sec . [ sec : dwd ] ) of the null surface @xcite , thereby providing a purely thermodynamic underpinning for classical gravity . none of this poses any conceptual or technical problem . the real issues arise when we try to go beyond the classical theory and provide a semi - classical or quantum gravitational interpretation of this result . the description in terms of the expression in is approximate and only valid when @xmath300 . the entire description , based on the q - metric acting as a proxy for the effective spacetime metric , can not be trusted too close to planck scales . we expect it to capture the quantum gravitational effects to a certain extent , but at present we have no way of quantifying the accuracy of this approach . ( it should also be noted that the identification @xmath301 might have further corrections close to planck scales . ) we also do not know how to deal with matter fields in a quantum spacetime and it is not clear how to introduce @xmath43 in a systematic way . in fact , the _ only _ reason to vary the metric tensor in an extremum principle a procedure which i have argued against is to obtain @xmath43 classically and @xmath302 semi - classically . none of the thermodynamic derivations , which leads to the exact ( rather than linearized ) field equations , obtains @xmath43 ( or @xmath302 ) from fundamental considerations based on a matter action . it is ironic that the problem arises from the matter sector rather than from the gravity sector ! the ideas outlined in this work suggest a more radical solution . matter , as we understand it , is quantum mechanical ; which essentially means that it is made of _ discrete _ degrees of freedom . how does such a _ discrete _ structure end up curving the _ continuum _ geometry ? that is , what is the actual _ mechanism _ by which @xmath43 produces @xmath303 ? the approach developed here suggests that one needs to introduce certain `` hidden variables '' , viz . , the auxiliary vector field @xmath193 , which encodes the discrete nature of geometry , and couple it to @xmath43 ( which is also fundamentally discrete in nature ) . the continuum geometry has to be related to @xmath193 , at scales much larger than the planck scale , by a suitable approximation just as the continuum density or pressure of a fluid arises when we average over the discreteness of the molecules . i have outlined one possible way in which this idea could be implemented , but it is by no means unique . further exploration of this approach could lead to a better understanding of how matter _ really _ ends up curving the spacetime . my research work is partially supported by j.c.bose fellowship of department of science and technology , india . i thank sumanta chakraborty , sunu engineer , dawood kothawala and kinjalk lochan for discussions and comments on the first draft . parattu k. , b. r. majhi , and t. padmanabhan , 2013 `` the structure of the gravitational action and its relation with horizon thermodynamics and emergent gravity paradigm '' _ phys . * 87 * , 124011 [ arxiv:1303.1535 ] . chakraborty s. , t. padmanabhan , `` evolution of spacetime arises due to the departure from holographic equipartition in all lanczos - lovelock theories of gravity '' , 2014 _ phys . , _ * d 90 * , 124017 [ arxiv:1408.4679 ]

the dynamical discrete web ( ddw ) , introduced in recent work of howitt and warren , is a system of coalescing simple symmetric one - dimensional random walks which evolve in an extra continuous dynamical time parameter . the evolution is by independent updating of the underlying bernoulli variables indexed by discrete space - time that define the discrete web at any fixed . in this paper , we study the existence of exceptional ( random ) values of where the paths of the web do not behave like usual random walks and the hausdorff dimension of the set of exceptional such . our results are motivated by those about exceptional times for dynamical percolation in high dimension by hggstrom , peres and steif , and in dimension two by schramm and steif . the exceptional behavior of the walks in the ddw is rather different from the situation for the dynamical random walks of benjamini , hggstrom , peres and steif . in particular , we prove that there are exceptional values of for which the walk from the origin has with a nontrivial dependence of the hausdorff dimension on . we also discuss how these and other results extend to the dynamical brownian web , a natural scaling limit of the ddw . the scaling limit is the focus of a paper in preparation ; it was also studied by howitt and warren and is related to the brownian net of sun and swart .

0704.2706

in this paper , we present a number of results concerning a dynamical version of coalescing random walks , which was recently introduced in @xcite . our results concern times of hausdorff dimension less than one where the system of coalescing walks behaves exceptionally . the results are analogous to and were motivated by the model of dynamical percolation and its exceptional times @xcite . in this section , we define the basic model treated in this paper , which we call the dynamical discrete web ( ddw ) , recall some facts about dynamical percolation , and then briefly describe our main results . the justification for calling this model a discrete web is that there is a natural scaling limit , which is one of our main motivations for analyzing the discrete web ( as it is in @xcite ) ; we also discuss in this section that scaling limit , which is a dynamical version of the brownian web ( see @xcite ) . a paper is in preparation @xcite on the construction of that model , which is closely related to the brownian net of sun and swart @xcite . we note that conjectures conerning ways to construct scaling limits of dynamical percolation ( in two - dimensional space ) appear in @xcite . we further note that exceptional times for dynamical versions of random walks in various spatial dimensions have been studied in @xcite and elsewhere , but these are quite different from the random walks of the ddw , as we note in subsection [ mainresults ] below . let @xmath4 for @xmath5 denote a simple symmetric random walk on @xmath6 starting at @xmath7 , i.e. at @xmath8 at @xmath9 . ( for real @xmath10 , we set @xmath11)$ ] , where @xmath12 $ ] denotes the integer part of @xmath13 . ) if we also consider other simple symmetric random walks starting from arbitrary points on the even space - time sublattice @xmath14 , which are independent of each other except that they coalesce when they meet , that is the system of ( one - dimensional ) coalescing random walks that is closely related to the one - dimensional ( discrete time ) voter model ( see @xcite ) and may be thought of as a one plus one dimensional directed percolation model . the percolation structure is highlighted by defining @xmath15 for @xmath16 to be the increment between times @xmath17 and @xmath18 of the random walker at location @xmath19 at time @xmath17 . these bernoulli variables are symmetric and independent and the paths of all the coalescing random walks can be reconstructed by assigning to any point @xmath20 an arrow pointing from @xmath20 to @xmath21 and considering all the paths starting from arbitrary points in @xmath22 and following the arrows . we note that there is also a set of dual ( or backward ) paths defined by the same @xmath15 s with arrows from @xmath23 to @xmath24 . the collection of all dual paths is a system of backward ( in time ) coalescing random walks that do not cross any of the forward paths . the ddw is a very simple stochastic process @xmath25 in a new dynamical time parameter @xmath0 whose distribution at any deterministic @xmath0 is exactly that of the static coalescing random walk model just described . specifically , let @xmath26 be a family of independent continuous time cadlag markov processes with state space @xmath27 and rate @xmath28 for changing state in either direction , with the initial condition that @xmath29 is a family of independent bernoulli random variables with @xmath30 . although this dynamical version of coalescing random walks sounds quite trivial at first hearing , it turns out that it can have interesting behavior at exceptional values of the dynamical time parameter s. this is a feature that it shares in common with dynamical percolation . static percolation models are defined also in terms of independent bernoulli variables @xmath31 , indexed by points @xmath32 in some @xmath33-dimensional lattice , which in general are asymmetric with parameter @xmath34 . there is a critical value @xmath35 when the system has a transition from having infinite clusters ( connected components ) with probability zero to having them with probability one . it is expected that at @xmath36 there are no infinite clusters and this is proved for @xmath37 and for high @xmath33 ( see , e.g. , @xcite ) . in dynamical percolation one extends @xmath31 to time varying functions @xmath38 , as in the case of coalescing walks , except that the transition rates for the jump processes @xmath38 are chosen to have the critical asymmetric @xmath39 distribution to be invariant . the question raised in @xcite was whether there were exceptional times when an infinite cluster ( say , one containing the origin ) occurs , even though this does not occur at deterministic times . this was answered negatively in @xcite for large @xmath33 and , more remarkably , was answered positively by schramm and steiff for @xmath37 in @xcite , where they further obtained upper and lower bounds on the hausdorff dimension ( as a subset of the dynamical time axis ) of these exceptional times . we apply in this paper the approaches used for dynamical percolation to the dynamical discrete web . although we restrict attention to one - dimensional random walks whose paths are in two - dimensional space - time and hence analogous to @xmath37 dynamical percolation , by considering different possible exceptional phenomena , we use both the high @xmath33 and @xmath37 approaches of @xcite . a natural initial question was whether there might be exceptional dynamical times @xmath0 for which the walk from the origin @xmath40 is transient ( say to @xmath41 ) . our first main result ( see theorem [ th1 ] in section [ tameness ] below ) , modeled after the high-@xmath33 dynamical percolation results , is that there are no such exceptional times . as we explain in remark [ remtame ] in section [ tameness ] , a small modification of the proof of theorem [ th1 ] shows that there are also no exceptional times where some pair of walks avoids eventually coalescing . our other two main results are modelled after the @xmath37 dynamical percolation results . one of them ( see theorem [ violation ] below ) concerns a kind of violation of the central limit theorem , or more accurately a kind of weak subdiffusivity , by the random walk @xmath40 for exceptional dynamical times @xmath0 ; namely , that @xmath42 for all @xmath43 . the other ( see theorem [ dimension ] ) gives upper and lower bounds on the hausdorff dimension of these exceptional times , that depend nontrivially on the constant @xmath3 so that the dimension tends to zero ( respectively , one ) as @xmath44 ( respectively , @xmath45 ) . this is strikingly in contrast with the dynamical random walks of @xcite where there are no exceptional times for which the law of the iterated logarithm fails . to explain why the walks of @xcite can behave so differently from those of the discrete web , we note that a single switch in the former case affects only a single increment of the walk while some switches in the discrete web change the path of the walker by a `` macroscopic '' amount , as discussed in the next subsection on scaling limits see also figure 1 where switching has changed one of the paths macroscopically . by an obvious symmetry argument , there are also exceptional dynamical times @xmath0 for which @xmath46 . one may ask whether there are exceptional @xmath0 for which @xmath47 . as discussed in remark [ twosided ] below , it can be shown , at least for small @xmath3 , that there are no such exceptional times . the case of large @xmath3 is unresolved . [ fixeds ] be the random walk at dynamical time @xmath0 starting from @xmath48 . this graph , with @xmath13 the horizontal and @xmath49 the vertical coordinate , represents simultaneously the family of functions @xmath50 ( @xmath51 ) . the lowest path , say for @xmath13 greater than about @xmath52 , differs `` macroscopically '' from the others.,title="fig : " ] there is a natural scaling limit of the ( static ) coalescing random walks model , the brownian web ( see @xcite ) . here one does a usual diffusive scaling in which the random walk time @xmath13 is scaled by @xmath53 , and space by @xmath54 so that the random walk path starting from @xmath55 $ ] at time @xmath56 $ ] scales to a brownian motion starting from @xmath57 at time @xmath58 . the collection of all random walk paths from all space - time starting points scales to a collection of coalescing brownian motion paths starting from all points of continuum space - time . now taking the rate of switching to be of order @xmath59 , rescaling time and space respectively by @xmath53 and @xmath54 , and then letting @xmath60 go to @xmath8 leads to a nontrivial limit @xmath61 , the dynamical brownian web . the idea of taking a scaling limit of the dynamical discrete web to obtain a dynamical continuum model is a natural one , which is at the heart of @xcite , although their approach appears to be somewhat different than the one we had already been taking . both approaches are closely related to the brownian net construction of sun and swart @xcite as will be extensively explored in @xcite . as we shall discuss in the next subsection , our approach is based on the construction of a certain poissonian marking of special space - time points of the ( static ) brownian web . these are the so - called @xmath62 points where a single brownian web path enters the point from earlier times and then two paths leave to later times , one to the left and one to the right with exactly one of the those two paths the continuation of the path from earlier time and the other one `` newly - born '' ; see figure 2 . neither the idea of doing a poissonian marking of special points for the brownian web nor the idea of using those marked points to construct a scaling limit of a dynamical discrete model is completely new . in particular , we note that a different type of marking ( of @xmath63 points ) was used in @xcite to study the scaling limits of noisy voter models . also the idea of using marked double points of @xmath64 to construct the scaling limit of two - dimensional dynamical percolation is discussed in @xcite . one motivation for the proposed marking in the @xmath64 context was the analogy with markings of @xmath62 as well as of @xmath63 points of the brownian web . in the dynamical brownian web @xmath65 , one can also consider exceptional dynamical times @xmath0 where the path @xmath66 starting from the origin at continuous time @xmath9 behaves differently than an ordinary brownian motion path . the results of @xcite are very similar to those of this paper for the discrete web . indeed , in some respects , the proofs are simpler since calculations with brownian motions are often easier than those with random walks . there is however one substantial complication , which is the main focus of @xcite and the reason we do not present the dynamical brownian web exceptional time results already in this paper . that complication is the actual construction of the dynamical brownian web a construction that is considerably less trivial than that of the dynamical discrete web , as we explain in the next subsection . [ changedirections ] is the vertical and @xmath49 the horizontal coordinate . if a @xmath67 ( left ) point of the original web ( left side of the figure ) is marked to switch at some @xmath68 $ ] , then the direction of that @xmath62 point in the web @xmath69 is changed at @xmath70 so that it becomes a @xmath71 ( right ) point as on the right side of the figure with the incoming path joined to the rightmost of the two paths starting from that point.,title="fig : " ] at the discrete level the scaling is chosen in such a way that between the dynamical times @xmath8 and @xmath0 , in a macroscopic box ( i.e. , one with size of order @xmath72 in the original lattice ) , the number of arrows that change direction will be of order @xmath53 . the situation can be simplified by focusing on switchings with `` macroscopic '' effects ( i.e. , switchings that will lead to a macroscopic alteration of a walker s trajectory in the initial web @xmath73 ) . a priori , one should also consider combinations of switchings that have macroscopic effects , but it turns out ( this will be proved rigorously in @xcite ) that the probability of macroscopic effects from switching two or more arrows is negligible compared to switching single arrows , and can be neglected . there is a natural way of characterizing those critical switchings . for example , let us consider the forward ( rescaled ) path @xmath74 starting from the origin in @xmath73 and assume that the arrow located at some @xmath75 is orginally oriented to the left . now we ask whether a switching of this single arrow will alter the path in such a way that the altered path will be to the right of @xmath76 , where @xmath77 are both positive macroscopic quantities . this will happen if and only if the backward path @xmath78 starting from @xmath79 hits @xmath74 at time @xmath13 ( more precisely , hits @xmath80 at time @xmath81 ) . more generally , the critical arrows leading to similar alterations are the `` contact '' points between @xmath74 and the backward path @xmath78 at which a switching occurs on @xmath82 $ ] . but it is now fairly easy to see what the statistics of such a set of points are . in fact , let @xmath83 be the random variable counting the number of such switchings up to the macroscopic time @xmath13 . the distribution of @xmath83 is simply given by : @xmath84 where @xmath85 are i.i.d . bernoulli random variables with @xmath86 , which is @xmath87 as @xmath88 . as @xmath88 , @xmath89 converges to the `` local time '' @xmath90 of the forward brownian path @xmath91 starting from the origin along a backward brownian path @xmath92 starting on the right of the path @xmath93 ( the joint distribution of @xmath93 and @xmath92 is analysed in @xcite and this `` local time '' will be defined precisely in @xcite ) . further , it is a standard fact that @xmath94 converges to a poisson process . hence , @xmath83 will converge to a poisson process run by the random clock @xmath95 . in other words , this set of points will consist of a two - dimensional ( @xmath13 and @xmath0 ) poisson point process with intensity measure @xmath96 , where @xmath97 is lebesgue measure and @xmath98 is the local time measure ( note that @xmath99 will be a locally finite measure so that the poisson process is well defined ) . so far , we have only selected the critical switchings inducing a specific type of macroscopic effect . namely , the ones altering the path @xmath93 in such a way that a point originally on one side of @xmath93 will be on the opposite side after switching occurs . but in order to select all the critical arrows leading to any kind of macroscopic changes , we should not only consider a poisson process run by the local time of a single forward path @xmath93 against a single backward path @xmath92 , but rather a poisson process run by the `` local time of the entire forward web along the entire backward web '' multiplied by the intensity @xmath100 . in other words , the set of marked points will be a three - dimensional poisson point process with intensity measure @xmath101 , where @xmath97 is lebesgue measure ( in the variable @xmath0 ) and @xmath102 is the local time measure of the forward web along the backward web . would not be a locally finite measure i.e. , the set of marked points in space - time is actually dense in @xmath103 . however , like what is presented in @xcite ( see p. 11 there ) , one can add an extra coordinate and lift @xmath102 to be a @xmath104-finite measure , or equivalently approximate @xmath102 by a sequence of locally finite measures @xmath105 , do the markings using @xmath105 , and then let @xmath106 . ] since the @xmath62 points of the continuum web are precisely those at which a forward and a backward path meet ( see , e.g. , @xcite ) , the measure @xmath107 will be supported by this set of points . from our previous description of them , it should be clear that each @xmath62 point has a preferred left or right `` direction '' . for example , a @xmath71 ( right ) point is one for which the continuing path ( coming in from earlier time ) is to the right of the `` newly - born '' path . hence , at the continuum level , the analog of an arrow switching will simply be a change of direction of all marked @xmath62 points ( see figure 2 ) . the web @xmath69 at time @xmath108 will be `` simply '' deduced from @xmath109 by switching the direction of all marked @xmath62 points whose @xmath0-coordinate is in @xmath110 $ ] . a last comment concerns the nature of the dependence of the two continuum paths @xmath111 and @xmath112 . these turn out to be a pair of `` sticky '' brownian motions , which are independent except when they touch each other . this is one of the major observations in @xcite ; we give a brief derivation of this fact in section [ pairsofpaths ] by analyzing pairs of paths in the discrete setting to see what must occur in the continuum scaling limit . in section [ tameness ] , we state our main theorem about tameness ; i.e. , that there are no exceptional dynamical times when the random walkers are transient . we also give there some other results about tameness in two extended remarks one about non - coalescence and the other about two - sided bounds of order @xmath113 . then in section [ existence ] , we show that there are exceptional dynamical times when the walkers are ( weakly ) subdiffusive i.e. , have one - sided bounds of order @xmath113 . in section [ hausdorff ] we derive upper and lower bounds on the hausdorff dimension of the set of such exceptional dynamical times . some estimates for random walks that are needed for our arguments are given in appendix [ appendix ] . the dynamics can be described , equivalently to the definition in section [ intro ] , in the following manner . the initial configuration is set to @xmath115 , but now we place independent poisson clocks at each site @xmath20 that ring at rate @xmath116 . every time clocks ring we toss independent fair coins to decide on the values of @xmath117 after the ring . statistically the two descriptions are equivalent . the motivation for this second description is that it leads to a useful representation of the interaction between the discrete webs at different dynamical times @xmath0 and @xmath118 . in particular , let @xmath119 and @xmath120 be the walks starting at @xmath7 and defined for times @xmath121 , belonging to @xmath25 and @xmath114 . if @xmath122 , then @xmath123 and @xmath124 are independent since the directions of the arrows at two distinct sites are independent . on the other hand , if @xmath125 , then the next steps of the two walks are now correlated . if the clock at the site @xmath126 did not ring on @xmath127 , then the two paths will coincide at time @xmath128 . if it rang at least once , then with probability @xmath129 they will coincide , and with probabily @xmath129 they wo nt . let us now define inductively a sequence of pairs of stopping times @xmath130 with @xmath131 and : @xmath132 on the interval of integer time @xmath133 $ ] , the paths @xmath134 coincide and at time @xmath135 they decide to separate with probability @xmath129 . in other words , from time @xmath136 , the walkers @xmath137 move independently until the next meeting time @xmath138 . hence , if we skip the intervals of time @xmath139 , @xmath140 behave as two independent random walks @xmath141 , while if we skip the intervals @xmath142 , the two walks coincide with a single random walk @xmath143 . furthermore , since @xmath143 is constucted from the arrow configuration at different sites than the ones used to construct @xmath141 , it is independent of @xmath141 ; and @xmath144 are i.i.d . random variables with @xmath145 . now , skipping the intervals @xmath139 corresponds to making the random time change @xmath146 with @xmath147 , and 1 . @xmath148 , 2 . @xmath149 , while skipping @xmath142 corresponds to making the time change @xmath150 ; i.e. , @xmath151 where @xmath152 are three independent standard random walks . in the following @xmath153 , distributed as @xmath137 , will be referred to as a pair of _ sticky random walks_. [ varyings ] fixed , this graph represents @xmath154 starting from @xmath48 ( with @xmath155 and @xmath156).,title="fig : " ] time and space are respectively rescaled by @xmath53 and @xmath157 and the rate of switching is taken as @xmath158 . from the previous section , the pair of rescaled processes @xmath159 is statistically equivalent to @xmath160 @xmath161 where @xmath162 are three independent rescaled random walks , and @xmath163 with 1 . @xmath164 2 . @xmath165 3 . @xmath166 are i.i.d random variables taking values in @xmath167 , with @xmath168 . as @xmath88 , @xmath169 converges in distribution to three independent brownian motions @xmath170 ; and @xmath171 converges in distribution to the local time @xmath172 at the origin of @xmath173 . moreover , as a consequence of the law of large numbers , if we take @xmath174 , with @xmath175 of order @xmath176 , then @xmath177 converges to @xmath178 . hence , it should come as no surprise that @xmath179 converges to @xmath180 where now @xmath181 , which is identical in distribution to a pair of sticky brownian motions with stickiness parameter @xmath182 ( see , e.g. , [ ss06 ] ) . we note that for small @xmath60 , the location @xmath183 , of the path starting from some @xmath184 is , for fixed @xmath13 , quite discontinuous in @xmath0 see figure 3 . [ th1 ] almost surely , all the paths are recurrent for every @xmath0 . in section 3 of @xcite , it is proved that for any homogeneous graph with critical probability @xmath35 for percolation and such that @xmath185 , the probability that ( with parameter @xmath34 ) the origin belongs to an infinite cluster , satisfies @xmath186 for @xmath187 , there is almost surely no dynamical time @xmath0 at which percolation occurs . in our setting , an entirely parallel argument can be used to show tameness of the dynamical discrete web with respect to recurrence . we discuss this briefly below , pointing to the relevant parts of @xcite . we consider the event @xmath188 that the walker starting from @xmath20 does not visit the site to the left of its starting position , that is , that the path @xmath189 started at @xmath20 does not contain any @xmath190 with @xmath191 . let @xmath192 be the probability of that event under @xmath34 i.e. , when the random walk increments are @xmath193 ( resp . , @xmath194 ) with probability @xmath34 ( resp . , @xmath195 ) . under the usual coupled construction of the model for @xmath196 $ ] , this event is increasing with @xmath34 in @xmath197 $ ] . for @xmath198 ( resp . @xmath199 ) , @xmath189 is distributed as a right ( resp . left ) drifting random walk . in particular , it is well known that for @xmath200 $ ] @xmath201 we now describe the parallel argument alluded to above . let @xmath202 denote the cardinality of the set @xmath203:\,a_{i , j}\mbox { occurs in } ( \xi)_s\}$ ] . @xmath192 and @xmath202 are the analogues of @xmath204 and @xmath205 in section 3 of @xcite . an analogue of lemma 3.1 there also holds here with the same proof , where here @xmath206 is the analogue of @xmath35 there , and the analogue of @xmath207 is the number @xmath208 of @xmath209 such that @xmath188 occurs in @xmath210 } \xi^s_{i',j'},\,(i',j')\in{\mathbb z}_{even}\right\},\ ] ] and we conclude from ( [ tame2 ] ) that @xmath211 . analogues of lemmas 3.2 and 3.4 also hold with the same proofs for the analogue quantities , and with @xmath188 replacing the event \{_v percolates_}. we then have that @xmath212 , and thus almost surely for every @xmath0 every walker eventually visits the site to the left of its starting position . the same is of course true of the site to the right of the starting position by symmetry . we conclude that almost surely for every @xmath13 every walker eventually visits every site in @xmath213 ( infinitely often ) . [ remtame ] another property of the static discrete web with respect to which the dynamical one is tame is the almost sure coalescence of all of its paths . it is enough to consider the case of two paths . for those , a similar argument as that for recurrence holds . the analogue objects to be considered in this case are as follows . given @xmath214 ( let us assume that @xmath215 ) , let @xmath216 be the event that the paths starting from @xmath217 do not eventually coalesce , and let @xmath218 be the cardinality of the set @xmath203:\,c_{v , v'}\mbox { occurs in } ( \xi)_s\}$ ] . for @xmath219 , let also @xmath220 be the event that the path of @xmath221 starting from @xmath222 and that of @xmath223 starting from @xmath224 do not coalesce eventually , where @xmath225 } \xi^s_{i , j},\,(i , j)\in{\mathbb z}_{even}\right\}.\ ] ] let now @xmath226 be the number of @xmath209 such that @xmath220 occurs . to show that @xmath227 , we analyze @xmath226 and its related quantities analogously to the analysis of @xmath208 and its related quantities to show that @xmath228 . in particular , the fact that @xmath229 follows from standard random walk estimates , using the fact that the difference of two random walk paths is another random walk path . [ twosided ] a main result of this paper is the existence of exceptional @xmath0 such that for all @xmath13 , @xmath230 ( see theorem [ violation ] in section [ existence ] ) , and of course there are then also exceptional @xmath0 such that @xmath231 . however , it can be shown that there are no exceptional @xmath0 for the two - sided bound @xmath47 , at least for small enough @xmath3 . the precise condition on @xmath3 under which we can prove this result is that @xmath232 , where @xmath233 is defined in proposition [ upper ] below . note that this condition implies according to proposition [ upper ] that the hausdorff dimension of the set of exceptional @xmath0 for either of the corresponding one - sided bounds does not exceed @xmath206 . the proof of this tameness claim combines arguments like those of theorem [ th1 ] and remark [ remtame ] with the estimates of lemma [ cont ] and proposition [ principal ] and with an application of the fkg inequalities . the specific fkg inequality , for the two events @xmath234 that for some @xmath235 $ ] and all @xmath13 , @xmath236 , is that @xmath237 . this is so because @xmath238 ( resp . , @xmath239 ) is an increasing ( resp . , decreasing ) event with respect to the basic @xmath240 processes see , e.g. , lemma 3.3 of @xcite for more details . we finally note that by essentially the same arguments one obtains tameness for two - sided bounds of the form @xmath241 provided that @xmath242 are small enough that @xmath243 . let @xmath244 be a sequence of positive integers divisible by @xmath245 . we construct inductively a sequence of `` diffusive '' boxes @xmath246 in the following manner : * @xmath247 is the rectangle with vertices @xmath248 , @xmath249 , @xmath250 and @xmath251 . * let @xmath252 and @xmath253 be respectively the middle point of the lower edge and the upper right vertex of @xmath254 . @xmath255 is the rectangle of height @xmath256 and width @xmath257 such that @xmath258 equals @xmath253 ( see figure 4 ) . let @xmath259 be the event that the path of @xmath25 starting at @xmath260 is at or to the right of @xmath261 at time @xmath262 and that it is never to the left of the left edge @xmath263 of the box @xmath246 . we would like to prove that for a certain choice of @xmath264 , there exist some exceptional times @xmath0 at which @xmath265 occurs for every @xmath266 . at those times , this would imply that the path starting from the origin stays to the right of the graphs obtained by patching together the left edges , @xmath263 , ( see figure 4 ) . by the same kind of reasoning used in dynamical percolation @xcite , to prove that ordinary diffusive behavior does not occur at certain exceptional dynamical times @xmath0 , it suffices to derive the following lemma , which we do later in this section of the paper . with @xmath13 the vertical and @xmath49 the horizontal coordinate . the solid curves represent segments of the paths starting from @xmath267 , @xmath268 and @xmath269 for which the events @xmath270 , @xmath271 and @xmath272 occur . the leftmost curve represents @xmath273 . ] [ propp ] there exists @xmath274 such that if @xmath275 + 1)$ ] for @xmath276 ( where @xmath277 $ ] is the integer part of @xmath49 ) with @xmath278 , then @xmath279 where @xmath34 is bounded away from @xmath8 when @xmath116 is bounded away from @xmath280 . let @xmath281 be the set of times @xmath0 on @xmath282 $ ] such that @xmath283 occurs . the previous lemma implies that @xmath284 . since @xmath285 is obviously decreasing in @xmath286 , if the @xmath281 were closed subsets of @xmath282 $ ] it would follow that @xmath287 . as explained in the proof of the next theorem , for @xmath288 , @xmath289 for some @xmath290 ( depending on @xmath291 ) and all @xmath292 . unfortunately , the set of times at which one arrow is ( or any finitely many are ) oriented to the right ( resp . , to the left ) is not a closed subset of @xmath282 $ ] since we have a right continuous process , and thus @xmath281 is not a closed set . this extra technicality is handled like in lemma 5.1 in @xcite , as follows . let @xmath293 denote the ( random ) set of all switching times for all @xmath294 s . by modifying every @xmath294 so that for @xmath295 , @xmath296 ( rather than being right - continuous ) , each @xmath281 is replaced by a closed @xmath297 . on the other hand , @xmath298 as a consequence of the fact that @xmath293 is countable and by independence of the @xmath117 s no @xmath295 can be exceptional . the uniformity with respect to small @xmath116 in lemma [ propp ] means that once space and time are diffusively rescaled by @xmath299 and @xmath53 and @xmath116 is rescaled by @xmath300 , the inequality ( [ esd ] ) is still valid , with @xmath34 fixed as @xmath88 . as a consequence of lemma [ propp ] , we will obtain the following . ( * violation of the clt * ) [ violation ] for @xmath301 , let @xmath302)$ ] where @xmath303 is the path starting at @xmath7 of the dynamical discrete web with switching rate @xmath304 . there exists @xmath305 such that @xmath306 , the probability to have a nonempty set of exceptional times s in @xmath282 $ ] for which @xmath307 for all @xmath308 satisfies the following : 1 . for any @xmath309 , there exist @xmath310 such that @xmath311 2 . similarly , for any @xmath312 , @xmath313 , there exists @xmath314 such that ( [ gg ] ) is valid . 3 . for any fixed @xmath60 , @xmath315 , @xmath316 . in the unrescaled coordinates , we take boxes @xmath246 as in lemma [ propp ] with @xmath317 + 1 ) \in ( \gamma^k/\lambda,4+\gamma^k/\lambda]$ ] . then in rescaled coordinates we have boxes @xmath318 with ( spatial ) width @xmath319 $ ] and ( temporal ) height @xmath320 . let @xmath321 denote the right - continuous function obtained by joining together the left boundaries @xmath322 of @xmath318 . on @xmath323 with @xmath324 , we have @xmath325 @xmath326 . if @xmath327 are such that @xmath328 then we will have @xmath329 for all @xmath10 as desired . the inequality ( [ boundary ] ) can be rewritten as @xmath330^{1/2 } \ , .\ ] ] using the bound @xmath331 on the left - hand side of ( [ gggg ] ) and the bounds @xmath332 on the right - hand side , it follows that in order to verify ( [ gggg ] ) it suffices to have , for @xmath333 , @xmath334 using the elementary bound @xmath335 ( for @xmath336 ) , we see that in order to verify ( [ anotherineq ] ) , it suffices to have , for @xmath333 , @xmath337 choosing @xmath338 yields this inequality provided @xmath339 . it is easy to see that for any @xmath340 , this will be valid provided @xmath341 is small enough and @xmath312 is large enough so that @xmath342 . this and lemma [ propp ] prove the first claim of the theorem ; the second claim , in which @xmath312 and @xmath341 are given , follows similarly . we now turn to the proof of the final claim . we set @xmath343 since essentially the same proof works for any @xmath344 . let @xmath345}$ ] denote the set of @xmath346 $ ] such that @xmath347 for @xmath348 and let @xmath17 be an integer so large that ( by the second claim of the theorem ) @xmath349 . first , @xmath350 } \supset \hat t_{j}^{[0,1]}\bigcap \{s\in [ 0,1]:\xi_{m , m}^s = + 1 \ \ \textrm{for } \ \ m < j\}$ ] where @xmath351}$ ] is the set of @xmath346 $ ] such that @xmath352 for @xmath353 . furthermore , @xmath351 } \supset \bar t_{j}^{[0,1]}$ ] , where @xmath354}$ ] is the set of @xmath346 $ ] such that @xmath355 for @xmath353 . but @xmath354}$ ] is just the translation ( from @xmath7 to @xmath356 ) of @xmath357}$ ] . since @xmath358 : \forall \ k < j , \ \xi_{k , k}^s=+1\}$ ] and @xmath359}$ ] are independent , it follows that @xmath360 , \ \forall \ k < n , \ \ \xi_{k , k}^s=+1)>0 \ , .\ ] ] in particular , if we assume that @xmath361 converges to the dynamical brownian web ( see subsection [ continuum ] ) in some appropriate sense as @xmath88 , this shows that the analogue of theorem [ tameness ] ( except for the final claim with @xmath60 fixed and @xmath362 ) will be valid for the continuum model as well . we now turn to : _ proof of lemma [ propp ] . _ by the cauchy - schwarz inequality , @xmath363\right)^2 } { \mathbb{e}\left[\left(\int_0 ^ 1 \prod_{k=0}^n 1_{a_k^s } \ ds \right)^2\right ] } \\ & = & \left(\left[\int_0 ^ 1 \int_0 ^ 1 \prod_{k=0}^n \frac{\mathbb{p}(a_k^s \bigcap a_k^{s'})}{\mathbb{p}(a_k)^2 } \ ds \ ds ' \right ] \right)^{-1 } \label{integrand}\end{aligned}\ ] ] where @xmath364 and the equality is a consequence of the stationarity of @xmath365 and the independence between the different boxes @xmath246 . it is enough to show that the integrand in the last expression of ( [ integrand ] ) is bounded above by a integrable function on @xmath282\times[0,1]$ ] , uniformly in @xmath286 . the rest of the proof will verify this property . now , for fixed @xmath266 and two deterministic times @xmath366 , let us rescale space and time respectively by @xmath367 and @xmath368 . also , let @xmath369 be the paths starting at @xmath206 at time @xmath8 , defined as the rescaled and translated version of the paths @xmath370 starting at @xmath260 , the middle point of the lower segment of the box @xmath246 . @xmath179 is a pair of sticky rescaled random walks starting at @xmath206 at time @xmath9 whose statistics ( up to a translation of starting point ) are described in equations ( [ stat1])-([stat2 ] ) . by definition , @xmath371 } \ \tilde s_i^\delta ( t ) > 0)$ ] . to complete the proof of lemma [ propp ] , we will use the following lemma , in which @xmath157 may be taken as an integer divisible by @xmath245 . [ inequality ] let @xmath369 be a pair of sticky random walks starting from @xmath206 at time @xmath9 as defined in ( [ stat1])-([stat2 ] ) . let @xmath372 be the event that for @xmath373 , @xmath374 and @xmath375 } \ \tilde s_i^\delta ( t ) \geq0 $ ] . if @xmath376 , then for @xmath377 small enough , @xmath378 where @xmath379 and @xmath380 are positive constants ( independent of @xmath381 and @xmath60 ) . in this proof we set @xmath382 . for any positive @xmath383 , let @xmath384 be as in ( [ stat1])-([stat2 ] ) and let @xmath385 } s_i^\delta(t)$ ] . then @xmath386 we start by dealing with the first term of the right - hand side of ( [ equation1 ] ) . first , @xmath387 ) + 2 \mathbb{p}(\inf s_1^\delta\in[-\delta^\alpha,0])\end{aligned}\ ] ] using the independence of the walks @xmath388 and the equidistribution of @xmath389 . the last two terms can be dealt with in a number of ways . for example , in @xcite , it is proved that a sequence of rescaled standard random walks @xmath390 and a brownian motion @xmath92 can be constructed on the same probability space in such way that for for any @xmath391 the quantity @xmath392 goes to @xmath8 faster than any power of @xmath60 . on this probability space , @xmath393 ) \leq \mathbb{p } ( { \hat b}(1 ) \in [ 1- 2 \delta^\alpha,1+\delta^\alpha])+ \mathbb{p}(\sup|\hat b - s_i^\delta|>\delta^\alpha ) \ , , \\ \mathbb{p}(\inf s_i^\delta\in[-\delta^\alpha,0])\leq \mathbb{p } ( \inf { \hat b } \in [ - 2 \delta^\alpha,\delta^\alpha])+ \mathbb{p}(\sup|\hat b - s_i^\delta|>\delta^\alpha ) \ , , \label{es2}\end{aligned}\ ] ] where the sup ( and inf ) are over @xmath394 $ ] . since @xmath395 , we have @xmath396 , implying that for @xmath397 and @xmath60 small enough the last terms on the right - hand side of ( [ 91 ] ) and ( [ es2 ] ) are bounded by @xmath398 , and consequently by @xmath399 . finally , ( [ 90 ] ) , ( [ 91 ] ) and ( [ es2 ] ) yield : @xmath400 where @xmath379 is a positive constant and @xmath397 . it only remains to deal with the rest of the terms on the right - hand side of equation ( [ equation1 ] ) . we will prove that @xmath401 ; the other terms can be treated in a similar fashion . for any @xmath402 , we have @xmath403 lemma [ lemm1 ] below takes care of the second term on the right - hand side of the inequality when @xmath404 . on the other hand , since @xmath405 we have that @xmath406 now , on the event @xmath407 , by definition of @xmath408 , we have for any @xmath409 $ ] : @xmath410 since @xmath411 and @xmath412 is an increasing function , it follows that @xmath413 implying @xmath414}|s_3^\delta ( t ) |\geq \frac{\delta^\alpha}{2 } ) + \mathbb{p}(\sup_{t \in [ 1-\delta^{\beta},1]}|s_1^\delta(1)-s_1^\delta(t)| \geq\frac{\delta^{\alpha}}{2 } ) \ , .\end{aligned}\ ] ] by ( the @xmath415 version of ) doob s inequality , we then have @xmath416 therefore , taking @xmath417 with @xmath418 gives the desired bound for the second term of the right - hand side of inequality ( [ equation1 ] ) . for the first term of this inequality , we only needed @xmath419 and the conclusion follows . [ lemm1 ] for any @xmath420 and @xmath421 small enough @xmath422 where @xmath423 . by the markov inequality , @xmath424 since @xmath425 converges in distribution to the local time of @xmath426 , where @xmath93 is a standard brownian motion , @xmath427 is uniformly bounded in @xmath60 . furthermore , @xmath428 , implying that @xmath429 * completion of proof of lemma [ propp ] . * + recall that @xmath430 + 1 ) \geq \gamma^k / \lambda$ ] , where @xmath431 with @xmath432 to be fixed later . by lemma [ inequality ] , there exists @xmath433 small enough such that ( [ lemmainequality ] ) is valid for @xmath434 . we define @xmath435 $ ] so that for @xmath436 , ( [ lemmainequality ] ) is valid for @xmath259 and @xmath437 . @xmath438 is independent of @xmath116 and since @xmath439 , @xmath440 where @xmath380 and @xmath379 are as in lemma [ inequality ] . the right - hand side of ( [ final1 ] ) is independent of @xmath116 and @xmath441 and is finite . indeed , @xmath442 since the boxes @xmath246 have diffusively scaled sizes and therefore @xmath443 as @xmath444 , where @xmath445 is the event that a brownian motion @xmath446 starting at @xmath129 at time 0 has @xmath447 and @xmath448 } { \hat b}(t)>0 $ ] . on the other hand , @xmath449 where @xmath450 . taking @xmath451 , by ( [ final1 ] ) and ( [ final2 ] ) we have that for every @xmath286 @xmath452 with @xmath453 and @xmath454 . since @xmath455\times[0,1])$ ] ) and ( [ energy ] ) is uniform in @xmath286 , this concludes the proof of lemma [ propp ] . in this section , we derive some lower and upper bounds for the set of exceptional dynamical times @xmath456 . to simplify notation , the rate of switching @xmath116 and the scaling parameter @xmath60 will both be taken equal to @xmath176 from now on . however , as in the previous section , it can easily be checked that essentially all the results stated below are again uniform in @xmath457 once space and time and @xmath116 are properly rescaled according to @xmath60 ( see subsection [ scalinglimits ] ) . the result that is not uniform as stated is proposition [ constant ] ; to have uniformity , @xmath458 should be replaced by @xmath459 for any @xmath460 . [ def1 ] we say that @xmath0 is a @xmath3-exceptional time if the path @xmath461 in @xmath25 starting from the origin at time @xmath9 does not cross the moving boundary @xmath462 . @xmath463 is then defined as the set of all @xmath3-exceptional times @xmath464 . clearly , the set consisting of all the k - exceptional times in @xmath465 is a non - decreasing function of @xmath3 . note that in definition [ def1 ] , the constant term for the moving boundary is fixed at @xmath176 . the next propostion asserts that for fixed @xmath3 the hausdorff dimension @xmath466 of the set of exceptional times is unchanged if @xmath176 is replaced by any @xmath467 . ( the remark following the proof of the proposition points out that more can be proved by essentially the same arguments . ) we note that as in dynamical percolation ( see sec . 6 of @xcite ) , @xmath468 is a.s . a constant by the ergodicity in @xmath0 of the dynamical discrete web . [ constant ] the hausdorff dimension @xmath466 of the set @xmath469 of exceptional times @xmath470 such that @xmath119 does not cross the moving boundary @xmath471 does not depend on @xmath472 ( for fixed @xmath3 ) . by monotonicity in @xmath266 , it is enough to prove that @xmath473 for @xmath266 any positive integer . by the same reasoning used to prove the last claim of theorem [ violation ] , the hausdorff dimension of @xmath474 is @xmath475 the hausdorff dimension of the set of times @xmath476 where @xmath477 is the translation ( from @xmath7 to @xmath478 ) of @xmath479 . by ergodicity in @xmath0 , the a.s . constant @xmath480 is the essential supremum of the random variable @xmath481)$ ] . on the other hand , since @xmath482 $ ] and @xmath203 : \forall \ m < k , \ \xi_{m , m}^s=+1\}$ ] are independent and the probability to have @xmath483 , \ \forall \ m < k , \ \xi_{m , m}^s=+1\}$ ] is strictly positive , it follows that @xmath484 : \forall \ m < k , \ \xi_{m , m}^s=+1\ } \bigcap \bar t_k(k))$ ] has the same essential sup as @xmath485)$ ] . hence @xmath486 and the conclusion follows . define @xmath487 to be the set of @xmath0 such that for some @xmath488 and some @xmath489 , the infimum over @xmath348 of @xmath490 ( or equivalently of @xmath491 ) is @xmath492 . it is not hard to see by arguments like those of proposition [ constant ] that @xmath493 . [ lower ] @xmath494 converges to @xmath176 as @xmath45 . let @xmath495 be fixed . since @xmath463 increases with @xmath3 it is enough to show that for @xmath3 large enough the hausdorff dimension is at least @xmath383 . consider the random measure @xmath496 , defined as @xmath497 , for any borel set @xmath498 in @xmath282 $ ] ( where @xmath265 are defined as in section [ existence ] ) . we define the @xmath383-energy of @xmath496 as @xmath499 by identical arguments as in section 6 of [ ss05 ] , if the expected value of @xmath500 is bounded above as @xmath106 , then the hausdorff dimension of the set of exceptional @xmath0 ( in @xmath465 ) for which @xmath501 occurs is at least @xmath383 . by fubini s theorem , @xmath502 and by ( [ energy ] ) , we have that @xmath503 \right ) \ ] ] with @xmath504 . in particular , taking @xmath505 as in the proof of theorem [ violation ] with @xmath291 large enough , @xmath506 can be made smaller than @xmath176 , and the right - hand side of ( [ haus ] ) is finite . combining the proofs of lemma [ propp ] and proposition [ lower ] with the remark following proposition [ constant ] , we can obtain a more explicit lower bound on @xmath507 as follows . set @xmath508 , where @xmath445 is the event that a brownian motion @xmath446 starting at @xmath206 at time @xmath9 has @xmath447 and @xmath509 , and define @xmath510 as the solution in @xmath511 of @xmath512 for @xmath513 . letting @xmath514 , we then have @xmath515 we will prove the following proposition . [ upper ] for any @xmath516 , @xmath517 where @xmath518 is the real solution @xmath519 of the equation @xmath520 furthermore , @xmath521 and more significantly @xmath522 . as a consequence of propositions [ constant ] , [ lower ] and [ upper ] , we immediately have the following . [ dimension ] the limits as @xmath523 and @xmath524 of @xmath468 are @xmath525 for any continuous function @xmath526 starting at @xmath527 , @xmath528 such that @xmath529 , the set of exceptional times for which the path starting from the origin at time @xmath8 does not cross @xmath526 has hausdorff dimension zero . to prove proposition [ upper ] we need the two following lemmas . ( sato [ s77 ] ) [ samere ] let @xmath530 , where @xmath531 are both positive , and @xmath93 is a standard brownian motion . then there exists @xmath532 such that @xmath533 where @xmath534 is the real solution in @xmath535 of ( [ sato ] ) . [ cont ] for any @xmath536 , let @xmath537 where @xmath538 and @xmath93 is a standard brownian motion . then for some @xmath539 and @xmath540 , @xmath541 let @xmath542 be the density of @xmath543 . by the girsanov theorem , @xmath544 where @xmath545 is the density corresponding to a standard brownian motion . therefore , since @xmath546 , we have @xmath547 integrating by parts , we get that @xmath548 on the one hand , by lemma [ samere ] , @xmath549 on the other hand , @xmath550 the last three displayed equations together easily imply ( [ exitbound ] ) . _ proof of proposition [ upper ] . _ we are now ready to obtain an upper bound for the hausdorff dimension of the set of k - exceptional times @xmath463 . let us partition @xmath282 $ ] into intervals of equal length @xmath551 , and select the intervals containing a k - exceptional time . the union of those is a cover of @xmath463 and we now estimate the number @xmath552 of intervals in the cover . let @xmath553 be the event that there is a time @xmath0 in @xmath554 $ ] such that @xmath555 . from the full dynamical arrow configuration for all @xmath556 $ ] , we construct a static arrow configuration as follows . we declare the static arrow at @xmath20 to be right oriented if and only the dynamical arrow is right oriented ( i.e. , @xmath557 ) at some @xmath235 $ ] ( a similar construction was used in section [ tameness ] ) . in this configuration , the path @xmath558 starting from the origin and following the arrows is a slightly right - drifting random walk with @xmath559 . clearly , @xmath560 proposition [ principal ] of appendix [ appendix ] implies that for any @xmath561 @xmath562 and by lemma [ cont ] it follows that @xmath563 hence @xmath564 so that @xmath565 by fatou s lemma , @xmath566 is almost surely bounded , which implies that the hausdorff dimension of @xmath463 is bounded above by @xmath567 and completes the proof of proposition [ upper ] . we will prove the following proposition . [ principal ] let @xmath568 , where @xmath93 is a standard brownian motion , and let @xmath558 be a discrete time simple random walk with drift given by @xmath569 for @xmath513 there exists @xmath570 such that for any @xmath516 , @xmath571 we consider @xmath572 the discrete time random walk embedded in the drifting brownian motion @xmath573 . namely , we define inductively a sequence of stopping times @xmath574 and their increments @xmath575 , with @xmath576 and @xmath577 and then we define @xmath578 . the proof of proposition [ principal ] will be done by coupling @xmath572 and @xmath573 in this particular way . note that @xmath572 has a different drift than @xmath579 since @xmath580 is not @xmath581 , but rather is @xmath582 . but proving ( [ theineq ] ) with @xmath558 replaced by @xmath583 suffices since @xmath583 has a larger positive drift than @xmath558 . now , let us consider some variants of @xmath573 and @xmath583 . define @xmath584 , where @xmath585 , as explained later , and @xmath586 is defined as @xmath587 this is the diffusion obtained by turning on a small drift of magnitude @xmath588 after the stopping time @xmath589 . in exactly the same way , we define @xmath590 and @xmath591 for @xmath586 ( but without the @xmath551 superscript ) and define @xmath592 . in particular @xmath593 and @xmath594 coincide for @xmath595 . 1 . the first part is to show that except on a set of probability @xmath606 , @xmath607 can be replaced by @xmath608 in the last expression of ( [ fff ] ) , with @xmath609 so that @xmath610 as @xmath611 . this will be done essentially by an application of the law of large numbers . once the above replacement has been made , the desired conclusion follows directly from the correlation inequality of lemma [ third ] and the inequality , @xmath612 where the @xmath613 were defined in ( [ tautau ] ) , with @xmath614 . noting that @xmath615 and hence @xmath616 , the last displayed inequality is a standard fact about sums of i.i.d . positive mean random variables . in gambling terms , it says that a gambler with a slight advantage has a strictly positive probability of never falling behind . it remains to justify the first part , for which it is enough to prove that , up to an error of at most @xmath606 , on the event @xmath617 , the inequality @xmath618 is actually valid for all @xmath286 . for @xmath619 , the @xmath591 are the exit times @xmath620 of brownian motion with a small drift @xmath551 . clearly , @xmath621 as @xmath622 . by the law of large numbers and standard large deviation estimates , we can assume that @xmath552 is in @xmath623 $ ] and show that the event @xmath624 occurs , except on a set of probability @xmath606 . hence , up to this error , on the event @xmath625 , the inequality @xmath626 can be extended from all @xmath627 also to all @xmath628 . hence , it only remains to control the indices @xmath286 in @xmath629\}$ ] . since @xmath630 , and since we can assume that @xmath631 , we get that for any such @xmath286 , @xmath632 this completes the proof of lemma [ firstlemma ] _ proof of lemma [ third ] . _ this result is a consequence of the fkg inequality for independent random variables . the variables @xmath634 ( see ( [ tautau ] ) ) and @xmath635 are completely independent , since in the case of a standard brownian motion , knowing the exit time from the interval @xmath636 $ ] does not give any information about the exit location . hence , given @xmath552 , @xmath637 behaves as a usual symmetric simple random walk for @xmath627 . thereafter the walk has a small positive drift to the right . this suggests that the indicator of the event @xmath638 can be expressed as a nondecreasing function of @xmath639 ( and some other variables to be determined ) since the larger @xmath639 is , the smaller @xmath552 will be , inducing more drift for @xmath596 . to make this more precise , we will couple @xmath640 and @xmath641 . the coupling involves the mutually independent @xmath642-valued @xmath639 , @xmath27-valued @xmath643 and @xmath644-valued @xmath645 , with @xmath646 the coupling is not via a brownian motion but rather is given in terms of our independent variables by @xmath647 and @xmath648 it is clear now that the above suggestion about the nondecreasing nature of the event in question is indeed valid . since the other event , @xmath649 is clearly nondecreasing with @xmath650 , the claim of lemma [ third ] follows by the fkg inequality . _ proof of lemma [ secondlemma ] . let us condition on @xmath552 . we use the coupling of @xmath651 and @xmath583 just discussed . now @xmath652 and we need to prove that the last term is of order @xmath551 for a suitable choice of the exponent @xmath380 ( where @xmath653 is the time threshold at which @xmath596 starts drifting ) . first , @xmath654 denoting by @xmath655 the solution of @xmath656 , we will show by induction on @xmath19 that as @xmath622 , @xmath657 , s_{\epsilon}'(n)-\bar s_{\epsilon}(n ) \geq k(1-l)\sqrt{n } ) \ , = \ , \nonumber \\ o(\epsilon k_1 + \epsilon^2 ( k_2-k_1)^2 + ... + \epsilon^2 ( k_{i}-k_{i-1})^2).\end{aligned}\ ] ] assuming this has been proved , we let @xmath658 be such that @xmath659 , and ( [ ps ] ) then implies that @xmath660 , \ , s_{\epsilon}'(n)- \bar s_{\epsilon}(n)\geq k(1-l ) n^\frac{1}{2 } ) \ , = \nonumber \\ o(\epsilon k_1 + \epsilon^2 ( k_2-k_1)^2 + ... + \epsilon^2 ( k_{n}-k_{n-1})^2 ) \ , .\end{aligned}\ ] ] for large @xmath286 , @xmath661 and therefore @xmath662 , implying that @xmath663 , \ , s_{\epsilon}'(n)- \bar s_{\epsilon}(n)\geq k(1-l ) n^\frac{1}{2})= o(\epsilon^2 n(\epsilon)^{\frac{3}{2 } } ) \ , .\ ] ] since we can assume by the law of large numbers that @xmath664 , taking @xmath665 implies that @xmath666 , and we get that the last term of ( [ reduce ] ) is @xmath606 . note that everything was done independently of @xmath552 ( except that @xmath667 ) . therefore , summing over the possible values of @xmath552 would finish the proof . it remains to prove ( [ ps ] ) , which we do by induction . first , for @xmath668 , since @xmath669 , @xmath670 , \ , s_{\epsilon}'(n)-\bar s_{\epsilon}(n ) \geq k(1-l ) \sqrt{n } ) & = & \mathbb{p}(s_{\epsilon}'(\cdot)-\bar s_{\epsilon}(\cdot ) \ \textrm{jumps on $ [ 0,k_1]$ } ) \\ & = & k_1 o(\epsilon ) .\end{aligned}\ ] ] next , assuming that ( [ ps ] ) is valid up to @xmath19 , we have @xmath671 , \ , \ s_{\epsilon}'(n)-\bar s_{\epsilon}(n ) \geq k(1-l)\sqrt{n } ) & \\ & \leq \mathbb{p } ( { \textrm { for some}}\ n\in(0,k_i ] , \ s_{\epsilon}'(n)- \bar s_{\epsilon}(n)\geq k(1-l)\sqrt{n } ) & \\ & + \mathbb{p}(\forall n \in ( 0,k_i ] , \ s_{\epsilon}'(n)- \bar s_{\epsilon}(n)<k(1-l)\sqrt{n } & \\ & \ { \textrm { and for some } } \ n \in ( k_{i},k_{i+1 } ] , \ s_{\epsilon}'(n)- \bar s_{\epsilon}(n)\geq k(1-l)\sqrt{n } ) \ , . & \end{aligned}\ ] ] we need to bound the last term of this inequality . since on @xmath672 $ ] we have @xmath673 $ ] ( by the definition of @xmath655 ) , and since @xmath674 only takes integer value , if @xmath675 , then @xmath676 . on the other hand , @xmath677 , and then our process has to jump at least twice on @xmath672 $ ] . but this probability is bounded by a term of order @xmath678 and ( [ ps ] ) follows . this completes the proof of lemma [ secondlemma ] . _ acknowledgements . _ the research of l.r.g . fontes was supported in part by fapesp grant 2004/07276 - 2 and cnpq grants 307978/2004 - 4 and 484351/2006 - 0 ; the research of the other authors was supported in part by n.s.f . grants dms-01 - 04278 and dms-06 - 06696 .

let be a-dimensional kleinian punctured torus group with accidental parabolic transformations . the deformation space of in the group of mbius transformations on the-sphere is well - known as the maskit slice of punctured torus groups . in this paper , we study deformations of in the group of mbius transformations on the-sphere such that does not contain screw parabolic transformations . we will show that the space of the deformations is realized as a domain of-space , which contains the maskit slice as a slice through a plane . furthermore , we will show that the space also contains the maskit slice of fourth - punctured sphere groups as a slice through another plane . some of another slices of the space will be also studied .

0707.2427

let @xmath0 be a @xmath1-dimensional kleinian once - punctured ( or simply punctured ) torus group with accidental parabolic transformations , which acts on the riemann sphere @xmath7 as mbius transformations . in this paper , we regard @xmath0 as a @xmath8-dimensional kleinian group and study deformations @xmath4 of @xmath0 in the group of mbius transformations on the @xmath1-sphere @xmath9 ; a kleinian group @xmath4 is called a _ deformation _ of @xmath0 if there is an isomorphism @xmath10 which takes a parabolic transformation to a parabolic transformation . especially , we focus on all deformations @xmath11 of @xmath0 such that @xmath12 takes a pure parabolic transformation to a pure parabolic transformation ; the definition of a pure / screw parabolic transformation will be described below ( see also 2.5 ) . the @xmath13-dimensional sphere @xmath14 is naturally identified with the ideal boundary of the @xmath15-dimensional hyperbolic space @xmath16 . therefore , the group @xmath17 of orientation preserving mbius transformations of @xmath18 is identified with the group @xmath19 of orientation preserving isomorphisms of @xmath16 . a discrete subgroup of @xmath20 is said to be a @xmath15-dimensional _ kleinian group_. @xmath1-dimensional kleinian groups are well studied in various contexts ; for example , hyperbolic geometry , @xmath1-dimensional topology , complex analysis , etc . we refer the reader to kapovich @xcite and marden @xcite for overviews of the theory of @xmath1-dimensional kleinian groups . on the other hand , the study of kleinian groups in higher dimensions is more wild and is far from getting the whole picture . we refer the reader to apanasov @xcite and kapovich @xcite for more information of this area . among higher dimensional kleinian groups , the study of @xmath8-dimensional kleinian groups should be interesting in its own light because they have fractal " limit sets in the @xmath1-sphere @xmath21 into which our geometric insight work well . some examples of @xmath8-dimensional kleinian groups with computer graphics of their limit sets can be found in ahara araki @xcite . we now move into more detailed argument . in what follows , we regard @xmath22 as a subgroup of @xmath23 by embedding @xmath24 into @xmath21 by @xmath25 . recall that the group @xmath22 is naturally identified with the group @xmath26 of conformal automorphisms of the riemann sphere @xmath7 , or the group of linear fractional transformations . let @xmath27 be a _ kleinian punctured torus group with accidental parabolic transformations _ ; that is , @xmath0 is a @xmath1-dimensional kleinian group such that the quotient manifold @xmath28 is homeomorphic to the trivial interval bundle over a punctured torus , and the ideal conformal boundary of @xmath28 is a union of a punctured torus and a thrice - punctured sphere . then @xmath0 is a rank-@xmath2 free group and a pair of generators @xmath29 of @xmath0 can be chosen so that @xmath30 and the commutator @xmath31 $ ] of the generators are parabolic . the following normalization of such a group @xmath0 up to conjugation is fundamental . let @xmath32 be a ( not necessarily discrete ) rank-@xmath2 free group in @xmath22 such that @xmath30 and @xmath31 $ ] are parabolic . then there is a complex number @xmath33 such that @xmath0 is conjugate in @xmath22 to the group @xmath34 generated by two mbius transformations @xmath35 we define : @xmath36 this set is known as the _ maskit slice _ of @xmath1-dimensional kleinian punctured torus groups ( see @xcite for more information ) . in this paper we want to consider deformations of a @xmath1-dimensional kleinian group @xmath37 with @xmath38 in the group @xmath23 . to explain our results , we need to recall a classification of elements of @xmath23 ; see section 2 for more details . similar to the case of @xmath22 , elements of @xmath23 are classified into three types : elliptic , parabolic and loxodromic transformations . especially a transformation in @xmath23 is _ parabolic _ if it has exactly one fixed point in @xmath21 . those parabolic elements are further classified into two types : a parabolic transformation @xmath39 is said to be _ pure parabolic _ if it is conjugate in @xmath23 to a translation , and _ screw parabolic _ otherwise . a screw parabolic transformation is conjugate to a composition of a rotation and a translation along a common axis . the first goal of this paper is to show the following theorem which is an analogue of lemma 1.1 for groups in @xmath23 : ( theorem 4.5 ) let @xmath32 be a ( not necessarily discrete ) rank-@xmath2 free group in @xmath23 such that @xmath30 and @xmath31 $ ] are pure parabolic . then there is a point @xmath40 in @xmath41 such that @xmath0 is conjugate in @xmath23 to the group @xmath42 generated by @xmath43 it is just a technical reason why we assume in the theorem above that @xmath30 and @xmath31 $ ] are pure parabolic " instead of parabolic , " and it is more natural to consider the case where @xmath30 and @xmath44 $ ] could be screw parabolic . in fact , suppose that a homeomorphism @xmath45 conjugates a @xmath1-dimensional kleinian group @xmath46 with a parabolic transformation @xmath47 to a @xmath8-dimensional kleinian group @xmath48 . then the transformation @xmath49 could be screw parabolic in general ( see 2.5 ) . unfortunately , we do not know the result of theorem 1.2 with pure parabolic " replaced by parabolic . " we also remark that , recently , y. kim and k. sakugawa independently announced that the level @xmath2 congruence subgroup of the modular group , the rank-@xmath2 free group @xmath50 in @xmath22 such that @xmath51 and @xmath52 are parabolic , has a continuous family of deformations @xmath53 in @xmath23 such that @xmath4 are kleinian groups containing screw parabolic transformations ( see also theorem 4.2 and its remark ) . we define : @xmath54 the next purpose of this paper is to understand the shape of the set @xmath55 . we remark that if @xmath56 , the group @xmath57 is nothing but the poincar extension of the group @xmath58 with @xmath59 . therefore @xmath55 contains @xmath3 as a slice through the plane @xmath60 in the parameter space @xmath61 . on the other hand , we will show in section 5 the slice of @xmath55 through the plane @xmath62 is the maskit slice @xmath6 of @xmath1-dimensional fourth - punctured sphere groups which is defined by @xmath63 where @xmath64 , @xmath65 and @xmath66 ( see section 3 ) . ( theorems 5.4 and 5.7 ) the maskit slice @xmath6 appears as the slice of @xmath55 through the plane @xmath62 . moreover , there is a constant @xmath67 such that for every @xmath68 $ ] , @xmath6 also appears as the slice through the plane @xmath62 with angle @xmath69 rotated along the @xmath70-axis . although the boundary of the slices of @xmath55 through the planes @xmath60 and @xmath62 are so called fractal , " we will also obtain in section 5 the following theorem ( see also figure 1 ) . ( corollary 5.11 ) the boundary of the slice of @xmath55 through the plane @xmath71 is a union of countably many analytic arcs . through the planes @xmath71 , @xmath62 and @xmath60.,height=264 ] one can find in figure 2 some computer graphics of the limit sets of the groups @xmath72 for parameters @xmath73 which lie ( or seem to be lie ) in the set @xmath55 . these figures can be seen as @xmath1-dimensional extensions of the beautiful patterns in the complex plane included in the book @xcite by mumford , series and wright . of @xmath72 restricted to the domain @xmath74 . the brightness of a point @xmath75 depends on its @xmath76-value . these are the views form the same point.,height=718 ] we remark that mbius transformations in @xmath23 can be related to two - by - two matrices whose entries lie in the quaternion algebra , see cao . however , in this paper , we take a geometric approach and make use of geometric techniques in the theory of kleinian groups , instead of calculating in the quaternion algebra . this paper is organized as follows : in section 2 , we recall some basic facts of mbius transformations . in section 3 , we recall the definition of the maskit slice @xmath3 ( resp . @xmath6 ) of @xmath1-dimensional punctured torus groups ( resp . fourth - punctured sphere groups ) . in section 4 , we prove theorem 1.2 on a normalization of @xmath8-dimensional punctured torus groups with accidental parabolics . in section 5 , we study the shape of the space @xmath55 and prove theorems 1.3 and 1.4 . as an application of theorem 1.2 , we give in appendix an example of a family of @xmath8-dimensional kleinian groups with @xmath1-generators , which contain punctured torus groups as @xmath2-generator subgroups . the authors would like to thank shin nayatani and yasushi yamashita for their interest and useful comments , yohei komori for his valuable information and discussions , and keita sakugawa for bringing to our attention to deformations of kleinian groups with screw parabolic transformations . we also grateful for the referee s valuable comments and suggestions . in this section , we recall from beardon @xcite and matsumoto @xcite the basic facts of mbius transformations on @xmath13-dimensional sphere @xmath14 . in the succeeding sections , we are mainly concerned with the cases of @xmath77 . for @xmath78 , @xmath79 denotes the euclidean norm of @xmath80 . a @xmath81-dimensional _ sphere _ @xmath82 in @xmath18 is either an euclidean sphere @xmath83 with @xmath84 , @xmath85 , or a @xmath81-dimensional euclidean plane @xmath86 plus @xmath87 . the _ inversion _ @xmath88 in the sphere @xmath82 is defined as follows : if @xmath89 then @xmath90 and if @xmath91 then @xmath92 is the reflection in the plane @xmath86 . especially we denote the inversion in the unit sphere at the origin by @xmath93 then the inversion @xmath92 in @xmath94 is written as @xmath95 a _ mbius transformation _ on @xmath18 is a product of finite numbers of inversions in codimension one spheres in @xmath18 . we denote by @xmath17 the group of orientation preserving mbius transformations . it is known that the group @xmath17 is equal to the group of all orientation preserving conformal automorphism of @xmath18 . especially , the group @xmath22 is naturally identified with the group @xmath96 of orientation preserving conformal automorphism of @xmath97 and with the lie group @xmath98 . the followings are typical examples of transformations @xmath99 in @xmath17 : * a _ translation _ : @xmath100 , where @xmath84 . we denote by @xmath101 . * a _ magnification _ : @xmath102 , where @xmath103 . * an _ orthonormal transformation _ : @xmath104 , where @xmath105 . general mbius transformations can be written as follows . let @xmath106 . 1 . if @xmath107 then @xmath108 , where @xmath109 , @xmath105 and @xmath110 . 2 . if @xmath111 then @xmath112 , where @xmath109 , @xmath113 and @xmath114 . let @xmath106 be the same as in lemma 2.1 ( 2 ) . note that @xmath115 and @xmath116 . the _ isometric sphere _ @xmath117 of @xmath118 is defined by @xmath119 it follows from the equality @xmath120 that the inversion in @xmath117 is written as @xmath121 . thus the transformation @xmath122 can be written as @xmath123 from this , one can observe that @xmath118 maps the interior of @xmath124 onto the exterior of @xmath125 . especially , the map @xmath118 restricted to @xmath117 is an isometry . we embed @xmath18 in @xmath126 in the natural way by @xmath127 and @xmath128 . then for every @xmath106 , there is a unique @xmath129 , called the _ poincar extension _ of @xmath118 , such that @xmath130 . in fact , if @xmath118 is a composition of inversions in @xmath81-dimensional spheres @xmath131 in @xmath18 , @xmath132 is obtained by the composition of inversions in @xmath13-dimensional spheres @xmath133 in @xmath126 , where @xmath134 is the sphere which is orthogonal to @xmath18 at @xmath135 . in this way , @xmath17 can be regarded as a subgroup of @xmath136 . let @xmath137 denote the upper half - space of @xmath126 equipped with the hyperbolic metric @xmath138 . it is the upper half - space model of the @xmath15-dimensional hyperbolic space . for every @xmath106 , its poincar extension @xmath129 induces an orientation preserving isomorphism of @xmath16 . in this way , @xmath17 can be identified with the group @xmath19 of orientation preserving isomorphism of @xmath16 . let @xmath139 . we say that @xmath118 is _ elliptic _ if @xmath118 has a fixed point in @xmath16 . if @xmath118 is not elliptic , @xmath118 is said to be _ parabolic _ if @xmath118 has exactly one fixed point in @xmath18 , and _ loxodromic _ if @xmath118 has exactly two fixed point in @xmath18 . it is known that every element @xmath106 is either elliptic , parabolic or loxodromic ( see ( 2.23 ) in @xcite ) . to describe a standard form of elliptic transformations , it is convenient to consider in the unit ball model @xmath140 of the hyperbolic space : in this setting , if @xmath141 is elliptic with @xmath142 , then @xmath104 for some @xmath143 . in the rest of the paper , we always consider in the upper half - space model of the hyperbolic space . standard forms of parabolic and loxodromic transformations are given in the following : let @xmath106 . 1 . if @xmath118 is a parabolic transformation with a fixed point @xmath144 , then @xmath145 for some @xmath146 and @xmath105 with @xmath147 . 2 . if @xmath118 is a loxodromic transformation with fixed points @xmath148 , then @xmath149 for some @xmath109 with @xmath150 and @xmath105 . a parabolic transformation @xmath106 is said to be _ pure parabolic _ if it is conjugate to a translation @xmath151 , and _ screw parabolic _ otherwise . we now observe the case of @xmath152 . since every @xmath153 has an eigenvalue of @xmath154 , we have the following : let @xmath39 . 1 . if @xmath118 is parabolic , it is conjugate in @xmath23 to a transformation given by @xmath155 where @xmath156 . ( then @xmath118 is pure parabolic if and only if @xmath157 . ) 2 . if @xmath118 is loxodromic , it is conjugate in @xmath23 to a transformation given by @xmath158 where @xmath159 and @xmath160 . it is worth to note that in @xmath23 a pure parabolic transformation is topologically conjugate to any screw parabolic transformation . more precisely , a translation @xmath161 is conjugate to a screw parabolic transformation of the form @xmath162 by an orientation - preserving homeomorphism @xmath45 defined by @xmath163 a discrete subgroup @xmath0 of @xmath164 is called a @xmath15-dimensional _ kleinian group_. it is known that @xmath0 is a kleinian group if and only if @xmath0 acts properly discontinuously on @xmath16 ; that is , given any compact subset @xmath165 , the set @xmath166 is finite . let @xmath0 be a @xmath15-dimensional kleinian group . the _ domain of discontinuity _ @xmath167 of @xmath0 is defined to be the set of points @xmath168 such that there is a neighborhood @xmath169 of @xmath80 such that the set @xmath170 is finite . the complement of @xmath171 in @xmath18 is called the _ limit set _ of @xmath0 , and denoted by @xmath172 . throughout this paper , we use the following notations : * if @xmath173 is a subset of a group @xmath174 , we denote by @xmath175 the subgroup generated by @xmath173 . * according to the usual convention , a mbius transformation @xmath176 is written in a linear fractional form @xmath177 by identifying @xmath24 with the rieman sphere @xmath7 . * a mbius transformation @xmath45 in @xmath23 is written as a function @xmath178 of @xmath179 . * we denote by @xmath180 the plane @xmath181 and by @xmath182 the plane @xmath183 and so on . in addition , we write @xmath184 and @xmath185 and so on . in this section , we recall some basic facts of kleinian groups in @xmath22 which are isomorphic to the fundamental groups of a thrice - punctured sphere , a fourth - punctured sphere or a punctured torus . we refer the reader to kra @xcite and keen series @xcite . a surface @xmath186 of _ type _ @xmath187 is an oriented closed surface of genus @xmath188 with @xmath13 punctures . if @xmath189 , the fundamental group @xmath190 of @xmath186 is a free group . a representation @xmath191 is said to be _ type - preserving _ if it takes a loop surrounding a puncture to a parabolic transformation . in this section , we consider images of faithful type - preserving representations @xmath191 for @xmath192 and @xmath193 . we choose generators for @xmath194 , @xmath195 and @xmath196 as in figure 3 . , @xmath197 and @xmath198 , and free homotopy classes of elements of their fundamental groups . , height=113 ] we define groups of type @xmath199 and @xmath193 in @xmath22 as follows : let @xmath0 be a subgroup of @xmath22 . * @xmath0 is said to be of _ type _ @xmath200 if it is the image of a faithful , type - preserving representation @xmath201 , or equivalently , if it is a rank-@xmath2 free group @xmath202 such that @xmath203 and @xmath204 are parabolic . * @xmath0 is said to be of _ type _ @xmath205 if it is the image of a faithful , type - preserving representation @xmath206 such that @xmath207 is parabolic , or equivalently , if it is a rank-@xmath1 free group @xmath208 such that @xmath209 and @xmath210 are parabolic . * @xmath0 is said to be of _ type _ @xmath193 if it is the image of a faithful , type - preserving representation @xmath211 such that @xmath207 is parabolic , or equivalently , if it is a rank-@xmath2 free group @xmath212 such that @xmath30 and @xmath31 $ ] is parabolic . observe that the subgroup @xmath213 of @xmath22 generated by @xmath214 is of type @xmath200 . it is known that any group of type @xmath200 in @xmath22 is conjugate to this group : let @xmath215 be a rank-@xmath2 free subgroup of @xmath22 such that @xmath203 and @xmath216 are parabolic . then @xmath0 is conjugate in @xmath22 to the group @xmath217 defined above . for a given @xmath33 , let @xmath218 be the group in @xmath22 generated by @xmath219 it is known that any group of type @xmath205 in @xmath22 is normalized in this form : let @xmath220 be a rank-@xmath1 free subgroup of @xmath22 such that @xmath209 and @xmath221 are parabolic . then @xmath0 is conjugate in @xmath22 to @xmath222 for some @xmath33 . a fundamental domain for @xmath223 with @xmath224 is given in the left of figure 4 . maskit slice of groups of type @xmath205 in @xmath22 _ is defined by @xmath225 some basic properties of @xmath6 can be found in the next subsection . ( left ) and @xmath226 ( right ) . , height=264 ] for a given @xmath33 , let @xmath227 be the group in @xmath22 generated by @xmath228 it is known that any group of type @xmath193 in @xmath22 is normalized in this form : let @xmath50 be a rank-@xmath2 free subgroup of @xmath22 such that @xmath30 and @xmath44 $ ] are parabolic . then @xmath0 is conjugate in @xmath22 to @xmath58 for some @xmath33 . a fundamental domain for @xmath226 with @xmath229 is given in the right of figure 4 . maskit slice of groups of type @xmath193 in @xmath22 _ is defined by @xmath230 ( see also figure 5 ) . it is easy to see that @xmath3 is invariant under the maps @xmath231 and @xmath232 , and that @xmath3 contains the set @xmath233 . it is also known that the maskit slice @xmath3 is contained in the set @xmath234 ( cf . @xcite ) . furthermore , minsky @xcite showed that @xmath3consists of two connected component and that each connected component of @xmath3 plus @xmath87 is a closed topological disc in @xmath97 . we mention here the relationship between the maskit slices @xmath6 and @xmath3 . first observe for every @xmath33 that @xmath235 thus @xmath236 is always a subgroup of @xmath237 . this implies that @xmath238 . furthermore , it is known by kra that @xmath6 is similar to @xmath3 : the map @xmath239 of @xmath240 onto itself induces a bijective map from @xmath6 onto @xmath3 . ( black part ) restricted to the domain @xmath241.,height=188 ] we define @xmath8-dimensional analogues of @xmath1-dimensional groups of types @xmath199 and @xmath193 as follows : let @xmath0 be a subgroup of @xmath23 . * @xmath0 is said to be of _ type _ @xmath200 if it is a rank-@xmath2 free group @xmath202 such that @xmath203 and @xmath204 are pure parabolic . * @xmath0 is said to be of _ type _ @xmath205 if it is a rank-@xmath1 free group @xmath208 such that @xmath209 and @xmath210 are pure parabolic . * @xmath0 is said to be of _ type _ @xmath193 if it is a rank-@xmath2 free group @xmath212 such that @xmath30 and @xmath31 $ ] are pure parabolic . as mentioned in introduction , it is just a technical reason why we assume some elements of these groups are ( not only parabolic but also ) pure parabolic . in this section , we study normalizations of these groups in @xmath23 up to conjugation . observe that the poincar extension of the group @xmath242 acting on @xmath7 is given by @xmath243 where @xmath7 is identified with @xmath244 via the map @xmath245 , and @xmath246 is the inversion in the unit sphere . ( here and hereafter , we use the same symbols @xmath247 for the poincar extensions of @xmath248 by abuse of notation . ) any group of type @xmath200 in @xmath23 is conjugate to this group : let @xmath215 be a rank-@xmath2 free subgroup of @xmath23 such that @xmath30 , @xmath249 and @xmath216 are pure parabolic . then @xmath0 is conjugate in @xmath23 to the group @xmath250 defined above . observe first that @xmath251 . in fact , if @xmath252 , @xmath30 commutes with @xmath249 since both @xmath253 are pure parabolic . this contradicts that @xmath254 is free . therefore , after conjugating the group @xmath202 in @xmath23 if necessary , we may assume that @xmath255 and @xmath256 . observe that there exist lines @xmath257 and @xmath258 through the origin @xmath259 which are invariant under the actions of @xmath260 and @xmath261 , respectively . after conjugating by an orthonormal translation if necessary , we may assume that both @xmath262 lie in the sphere @xmath263 . then one can see that the group @xmath254 preserves the sphere @xmath263 and its orientation . thus the result follows form lemma 3.2 . recently , y. kim and k. sakugawa independently announced ( oral communication ) that there is a continuous family of rank-@xmath2 free kleinian groups @xmath264 in @xmath23 such that @xmath203 and @xmath204 are ( screw ) parabolic . in the next subsection , we need the following : let @xmath202 be a rank-@xmath2 free subgroup of @xmath23 such that @xmath30 , @xmath249 and @xmath216 are pure parabolic . if @xmath265 and @xmath256 then @xmath266 , that is , @xmath267 . by theorem 4.2 , there is a transformation @xmath39 which conjugates @xmath268 to @xmath213 . since @xmath269 conjugates @xmath270 and @xmath271 each other , we may assume that @xmath272 and @xmath273 . this implies that @xmath142 and @xmath107 . then by lemma 2.1 , we have @xmath274 , where @xmath103 and @xmath153 . moreover , it follows from the condition @xmath272 that @xmath118 is a rotation along the @xmath275-axis . since such @xmath118 conjugates @xmath271 to itself , we have @xmath276 . in this subsection , we will show that every group of type @xmath205 in @xmath23 is conjugate to a group of type @xmath205 in @xmath22 . for a given @xmath277 , let @xmath278 be the subgroup of @xmath23 generated by @xmath279 observe that if @xmath280 lies in the plane @xmath182 , the group @xmath281 for @xmath56 is the poincar extension of the group @xmath222 for @xmath59 defined in 3.2 . let @xmath220 be a rank-@xmath1 free subgroup of @xmath23 such that @xmath282 and @xmath210 are pure parabolic . then @xmath0 is conjugate in @xmath23 to @xmath283 for some @xmath284 . note that the subgroups @xmath254 , @xmath285 of @xmath286 are of type @xmath200 in @xmath23 . therefore , after conjugating @xmath286 in @xmath23 if necessary , we may assume that @xmath265 and @xmath287 , and hence that @xmath288 by corollary 4.3 . since a rotation along the @xmath275-axis conjugates the group @xmath213 to itself , we may also assume that the fixed point @xmath289 of @xmath290 lies in the plane @xmath180 . to obtain the result , it suffices to show that @xmath291 . now let us conjugate the group @xmath292 by the translation @xmath293 to the group @xmath294 . since @xmath295 fixes @xmath259 , we have @xmath296 from corollary 4.3 . thus we conclude that @xmath297 . in this subsection , we will obtain a normalization of groups of type @xmath193 in @xmath23 . in contrast to the case of groups of type @xmath205 , we will see that the space of groups of type @xmath193 in @xmath23 is strictly larger than the space of groups of type @xmath193 in @xmath22 . for a given @xmath73 , let @xmath298 be the subgroup of @xmath23 generated by @xmath299 where @xmath300 is the inversion in the sphere @xmath301 . observe for every @xmath73 that @xmath302 thus @xmath303 is always a subgroup of @xmath304 . in addition , observe that if @xmath280 lies in the plane @xmath182 , the group @xmath72 for @xmath56 is the poincar extension of the group @xmath37 for @xmath59 defined in 3.3 . the next theorem reveals that every group of type @xmath193 in @xmath23 is conjugate to the group @xmath72 for some @xmath73 . ( we will prove a slightly stronger statement . ) let @xmath32 be a rank-@xmath2 free group in @xmath23 such that @xmath30 is pure parabolic and @xmath31 $ ] is ( not necessarily pure ) parabolic . then @xmath0 is conjugate in @xmath23 to @xmath57 for some @xmath277 . we first show that @xmath305 . if not , we may assume that @xmath306 . it then follows from lemma 2.1 that @xmath307 and @xmath308 for some @xmath103 , @xmath153 and @xmath309 . then a calculation yields @xmath31({{{\boldsymbol{x}}}})={{{\boldsymbol{x}}}}+\lambda p({{{\boldsymbol{v}}}})-{{{\boldsymbol{v}}}}$ ] . thus @xmath31 $ ] is pure parabolic with the fixed point @xmath144 . this implies that @xmath30 commutes with @xmath31 $ ] , which contradicts that @xmath310 is free . therefore we may assume that @xmath311 and @xmath312 . it then follows from lemma 2.1 that @xmath313 , @xmath30 are of the forms @xmath314 and @xmath315 for some @xmath103 , @xmath316 and @xmath317 . we may further conjugate the group @xmath212 by a transformation @xmath39 fixing @xmath259 and @xmath144 without changing our assumptions @xmath311 , @xmath312 . such a transformation is of the form @xmath318 with @xmath319 by lemma 2.3 . therefore , after conjugating @xmath310 by a magnification if necessary , we may assume that the radius of the isometric sphere of @xmath313 equals @xmath154 , and hence that @xmath320 are of the forms @xmath321 and @xmath315 . in addition , we claim that we may also assume that @xmath322 for some @xmath323 after conjugating @xmath310 by a suitable orthonormal transformation if necessary . in fact , choose @xmath324 so that @xmath325 and @xmath326 . then @xmath327 conjugates @xmath328 to @xmath329 and @xmath330 to @xmath331 . now observe that @xmath332 takes @xmath325 to @xmath326 . therefore , replacing @xmath333 , @xmath334 and @xmath335 by @xmath86 , @xmath280 and @xmath336 , respectively , we obtain the claim . next we will show that @xmath337 by using the condition that @xmath31 $ ] is parabolic . recall that @xmath338 denotes the translation @xmath339 with @xmath340 . using this notation , we have @xmath341 and @xmath342 . then the transformation @xmath343\beta\alpha=\alpha^{-1}\beta^{-1}\alpha\beta$ ] is calculated as @xmath344 where the last equality follows from @xmath345 . since @xmath346 and @xmath347 lie in the plane @xmath180 , the transformation @xmath348 preserves the sphere @xmath263 and its orientation . by letting @xmath349 and identifying @xmath263 with @xmath7 via the map @xmath350 , we see that the action of @xmath351 restricted to the sphere @xmath352 is a mbius transformation @xmath353 whose matrix representation is @xmath354 since @xmath351 is parabolic and @xmath355 , we conclude that @xmath356 , and hence that @xmath357 . now let us denote by @xmath358 the rotation of angle @xmath359 along the @xmath275-axis . since @xmath316 and @xmath147 , we have @xmath360 for some @xmath361 . noting that @xmath362 and @xmath363 for every @xmath359 , we obtain @xmath364 therefore , after conjugating @xmath310 by @xmath365 with @xmath366 @xmath367 if necessary , we may assume that @xmath368 are of the forms @xmath369 and @xmath370 . thus we obtain the desired normalization . the condition that the group @xmath50 is free is only used to show that @xmath305 . we define the _ space of kleinian groups of type @xmath193 in @xmath23 _ by @xmath371 the aim of this section is to study the shape of @xmath372 . especially , the slices of @xmath55 along the planes @xmath373 and @xmath374 will be studied . although @xmath375 \in g_{{{\boldsymbol{p}}}}$ ] are pure parabolic for every @xmath280 , it remains the possibility that for some @xmath376 , the group @xmath72 has ( accidental ) screw parabolic transformations . we do not know whether it does happen or not . observe that by definition we have @xmath377 where the set @xmath378 is regarded as a subset of the plane @xmath182 via the identification @xmath379 . we next observe in theorem 5.1 below that there are two subsets @xmath380 of the parameter space @xmath61 such that @xmath381 and that both of them are invariant under the action of the rotation along the @xmath70-axis . recall that @xmath382 denotes the rotation of angle @xmath359 around the @xmath275-axis . we also regard @xmath365 acts on the parameter space @xmath61 as a rotation around the @xmath70-axis . furthermore , identifying @xmath7 with @xmath383 via the map @xmath384 , we define a map @xmath385 by @xmath386 we have @xmath387 suppose first that @xmath376 . then @xmath72 is discrete , and thus the subgroup @xmath388 is also discrete . choose @xmath389 and @xmath390 such that @xmath391 . then we have @xmath392 , where @xmath223 is naturally regarded as a subgroup of @xmath23 . it then follows from the discreteness of @xmath281 that @xmath393 , and hence that @xmath394 . next suppose that @xmath395 for some @xmath389 . then the interiors of the isometric spheres @xmath396 of @xmath397 and @xmath398 of @xmath399 are disjoint . therefore we can find a fundamental domain for @xmath304 in @xmath21 as a @xmath1-dimensional analogue of the fundamental domain for @xmath226 in @xmath7 as in figure 4 . then by poincar s polyhedron theorem ( see maskit ( * ? ? ? * iv , h ) and the remark below ) , @xmath72 is a rank-@xmath2 free kleinian group in @xmath23 . thus we conclude that @xmath376 . poincar s polyhedron theorem for @xmath1-dimensional polyhedra in @xmath21 with side - pairing maps in @xmath23 can be deduced from the theorem for @xmath8-dimensional polyhedra in @xmath400 with side - pairing maps in @xmath401 ; the precise statement can be found in maskit @xcite . we remark that , in general , a polyhedron in @xmath21 need not be convex . see also epstein petronio @xcite . the goal of this subsection is to show theorem 5.4 , which states that the maskit slice @xmath6 of groups of type @xmath205 in @xmath22 appears as the slice of @xmath55 through the plane @xmath402 . before starting the proof of theorem 5.4 , we will make some observations on groups @xmath403 with @xmath404 . observe that if @xmath404 , the action of @xmath405 preserves the sphere @xmath301 . more precisely , let @xmath406 and @xmath407 . then the action of @xmath304 restricted to the sphere @xmath301 is given by @xmath408 where @xmath301 is identified with @xmath7 via the map @xmath409 . ( here we use the notations @xmath410 and @xmath411 to distinguish them from @xmath226 and @xmath412 defined in 3.3 . ) note that @xmath411 is an orientation reversing conformal automorphism of @xmath97 and that @xmath413 . let @xmath414 be the index two subgroup of @xmath410 of orientation preserving transformations . then we have @xmath415 in fact , one can check that @xmath416 and @xmath417=2 $ ] . now observe that @xmath418 it then follows that @xmath236 is a subgroup of @xmath419 and that @xmath420 . the statement of theorem 5.4 can be rephrase that @xmath414 is discrete provided that @xmath223 is discrete . to prove theorem 5.4 , we need to recall some basic facts of the action of a group @xmath222 for @xmath393 on @xmath7 . for simplicity we assume that @xmath421 , but the argument for the case @xmath422 is parallel . let @xmath423 and @xmath424 be the subgroups of @xmath223 of type @xmath200 , and let @xmath425 and @xmath426 . in this notation , we have the following : 1 . let @xmath427 . the disc @xmath428 is a component of @xmath429 , and is @xmath430-invariant ; that is , @xmath431 for every @xmath432 and @xmath433 for every @xmath434 . @xmath435 for every @xmath436 . the set @xmath437 is either an empty set or a connected component of @xmath429 which is @xmath223-invariant and simply connected . we state below the second klein maskit combination theorem for kleinian groups in @xmath23 to what extent we need in the proof of theorem 5.4 ( and its extension , theorem 5.7 ) . we refer the reader to maskit ( * ? ? ? * vii ) for more information . let @xmath438 be a torsion - free kleinian group and @xmath439 . let @xmath440 be subgroups of @xmath441 , and let @xmath442 be closed topological balls . if they satisfy the following conditions @xmath443@xmath444 then @xmath445 is discrete and isomorphic to the hnn - extension @xmath446 of @xmath441 by @xmath447 : 1 . let @xmath427 . the interior @xmath448 of @xmath449 is @xmath450-invariant ; that is , @xmath451 for every @xmath432 and @xmath452 for every @xmath434 . 2 . @xmath453 for every @xmath454 . the complement @xmath455 has an interior point . the transformation @xmath447 takes the interior of @xmath456 onto the exterior of @xmath457 ; that is , @xmath458 and @xmath459 . 5 . @xmath460 . the conditions @xmath443 , @xmath461 and @xmath462 guarantee that @xmath463 and @xmath464 can be embedded disjointly into @xmath465 , and that the complement @xmath466 has an interior point . the conditions @xmath467 and @xmath444 guarantee that the action of @xmath447 descends to a pairing map of the resulting boundary of @xmath466 . for the convenience of the reader , we give here a sketch of the proof of theorem 5.3 ; see ( * ? ? ? d and e ) for more details . recall from @xcite that the group @xmath468 is the free group of words in @xmath447 and the elements of @xmath441 modulo equivalence induced from the relation @xmath460 . therefore each element of @xmath446 is equivalent to a word of the form @xmath469 ( @xmath470 , @xmath471 ) which satisfy the following conditions : ( 1 ) @xmath472 for @xmath473 , ( 2 ) @xmath474 for @xmath475 , ( 3 ) @xmath476 if @xmath477 and @xmath478 , and ( 4 ) @xmath479 if @xmath480 and @xmath481 . we denote by @xmath482 the natural projection . now let us take an open set @xmath169 in @xmath483 . since @xmath441 is a torsion - free kleinian group , we may assume that @xmath484 for every @xmath485 . we will show that @xmath486 for every @xmath487 . this implies that @xmath488 is an isomorphism and @xmath489 is discrete . we first assume that @xmath490 . if @xmath491 then @xmath492 and thus @xmath493 . when @xmath494 , @xmath495 lies in @xmath496 if @xmath497 and in @xmath498 if @xmath499 . in both cases we have @xmath493 . next assume that @xmath500 . if @xmath501 then @xmath502 and @xmath503 , which implies that @xmath504 . when @xmath505 , we can show that @xmath506 . in fact , if @xmath507 , one see that @xmath508 and @xmath497 . this implies that @xmath509 and that @xmath510 . similarly , if @xmath511 we have @xmath512 . finally , if @xmath513 , @xmath514 lies in @xmath496 if @xmath509 and in @xmath498 if @xmath515 . thus we have @xmath516 , which implies that @xmath504 also in this case . the proof for the case of @xmath517 is obtained by induction . applying lemma 5.2 and theorem 5.3 , we can now prove the following : we have @xmath518 it follows from theorem 5.1 that if @xmath519 then @xmath520 . conversely , suppose that @xmath521 , and hence that the subgroup @xmath522 of @xmath523 is discrete . we assume for simplicity that the third coordinate @xmath524 of @xmath525 is positive , but the argument for the case @xmath526 is almost parallel . let @xmath527 and @xmath528 we will check that @xmath529 and @xmath530 satisfy the conditions @xmath443@xmath444 in theorem 5.3 . now take @xmath393 such that @xmath531 . then @xmath421 by our assumption @xmath85 . the map @xmath532 conjugates the action of @xmath223 on @xmath7 to the action of @xmath281 on the sphere @xmath301 . therefore the group @xmath281 can be regarded as the poincar extension of the group @xmath223 acting on the sphere @xmath533 ( see the left of figure 6 ) . thus the conditions @xmath443 and @xmath461 directly follows from lemma 5.2 . now observe that , for @xmath534 , every orbit @xmath535 ( @xmath536 ) of @xmath449 except for @xmath449 is a ball of radius @xmath537 with center in @xmath538 . therefore one can take an open domain , say , @xmath539 which does not intersect @xmath540 . thus the condition @xmath462 follows . it is easy to check the condition @xmath467 . finally , the condition @xmath444 @xmath541 follows from the facts that @xmath542 and @xmath543 . therefore theorem 5.3 asserts that @xmath544 is discrete and isomorphic to the group @xmath545 . thus we conclude that @xmath376 . , @xmath457 , @xmath546 and @xmath547 and the planes @xmath538 and @xmath548 in the proofs of theorems 5.2 ( left ) and 5.5 ( right ) . these are the views along the @xmath275-axis . , height=264 ] recall from theorem 5.4 that the maskit slice @xmath6 appears as the slice of @xmath55 through the plane @xmath549 . in this subsection , we will extend this result to theorem 5.7 , which states that there is a constant @xmath550 such that @xmath6 also appears as the slice of @xmath55 through the plane @xmath551 for every @xmath552 $ ] . we first fix our terminology : the intersection @xmath553 of two spherical balls @xmath554 , @xmath555 in @xmath21 is called a _ lens_. the _ inner angle _ of lens @xmath553 is the interior angle @xmath556 formed by two faces of @xmath553 . the following lemma is essential in the proof of theorem 5.7 ; this is a special case of the collar lemma of basmajian ( * ? ? ? * theorem 1.1 ) . for a given @xmath393 , let @xmath557 be a non - invariant component of @xmath429 . let @xmath558 be the convex hull of @xmath557 in @xmath559 , and let @xmath560 denote the stabilizer of @xmath557 in @xmath223 . then there is a constant @xmath561 , which does not depend on the choices of @xmath393 and @xmath562 , such that the @xmath563-neighborhood of @xmath558 in @xmath559 with respect to the hyperbolic metric is @xmath564-invariant . note that @xmath565 can be regarded as a lens in @xmath21 with inner angle @xmath566 . similarly , the @xmath563-neighborhood of @xmath558 in @xmath559 is a lens in @xmath21 with inner angle @xmath567 , where @xmath568 is the constant determined by @xmath563 . we give here a proof of lemma 5.6 for the convenience of the reader . since @xmath557 is non - invariant component of @xmath429 , it follows form lemma 5.2 that the stabilizer @xmath560 of @xmath557 is a group of type @xmath200 , and that @xmath557 is @xmath564-invariant . therefore we only need to show that there exists a constant @xmath561 , which does not depend on the choices of @xmath569 and @xmath557 , such that the hyperbolic distance from @xmath558 to @xmath570 is greater than @xmath571 for every @xmath572 . now let @xmath573 denote the relative boundary of @xmath558 in @xmath559 and let @xmath574 be the nearest point retraction . for a given @xmath572 , let @xmath575 denote the image of @xmath570 via the map @xmath524 . then one can check that @xmath493 for every @xmath576 . therefore the hyperbolic area of @xmath169 is bounded above by the hyperbolic area ( @xmath577 ) of the thrice - punctured sphere @xmath578 . from this , we can deduce the existence of a desired constant @xmath563 . using lemma 5.6 , we can now prove the following : there is a constant @xmath579 such that @xmath580 for every @xmath552 $ ] . the argument is similar to the argument of the proof of theorem 5.4 . let @xmath579 be the constant as in remark of lemma 5.6 . we will show below that for every @xmath552 $ ] we have @xmath581 . by symmetry , we may assume that @xmath582 . it follows from theorem 5.1 that if @xmath583 then @xmath584 . conversely , suppose that @xmath585 , and hence that the subgroup @xmath303 of @xmath72 is discrete . we again assume that @xmath85 . let @xmath456 , @xmath586 and @xmath587 be the same as in the proof of theorem 5.4 . we will check that @xmath529 and @xmath530 satisfy the conditions @xmath443@xmath444 in theorem 5.3 . since the conditions @xmath467 and @xmath444 are similarly satisfied , we only need to check the conditions @xmath443@xmath462 . now take @xmath393 such that @xmath588 . then @xmath421 by our assumption @xmath85 . observe that the action of group @xmath281 preserves the sphere @xmath589 , and that the map @xmath590 conjugates the action of @xmath223 on @xmath7 to the action of @xmath281 on the sphere @xmath551 . therefore the group @xmath281 can be regarded as the poincar extension of the group @xmath223 acting on the sphere @xmath591 ( see the right of figure 6 ) . observe that @xmath365 takes the components @xmath592 , @xmath593 of @xmath429 as in lemma 5.2 to the intersections @xmath594 of the balls @xmath595 with the sphere @xmath589 , respectively . therefore @xmath596 is @xmath597-invariant for @xmath427 , and @xmath598 for every @xmath536 . now let @xmath599 denote @xmath600 . then our assumption @xmath582 can be written as @xmath601 . let @xmath602 , @xmath603 be the two components of @xmath604 such that the inner angles of the lenses @xmath605 and @xmath606 are @xmath607 and @xmath608 , respectively . \(1 ) we first show that @xmath496 is @xmath609-invariant . ( the same argument reveals that @xmath457 is @xmath610-invariant . ) it is easy to see that @xmath611 for every @xmath612 . to show that @xmath613 for every @xmath614 , it suffices to show that @xmath615 . in fact , @xmath616 takes the lens @xmath617 with inner angle @xmath607 to the lens @xmath618 with the same inner angle . since we are assuming that @xmath601 , it follows from lemma 5.6 that @xmath615 . \(2 ) we next show that @xmath619 for every @xmath536 . by symmetry , it suffices to show that @xmath620 . this follows from the facts that @xmath621 , that the lens @xmath622 has the inner angle @xmath607 , and that the lens @xmath623 has the inner angle @xmath608 . \(3 ) finally we show that @xmath624 has an interior point . this follows from the fact that , for @xmath534 , every orbit @xmath535 ( @xmath536 ) of @xmath449 except for @xmath449 is a ball of radius @xmath537 which intersects with the plane @xmath589 . therefore theorem 5.3 asserts that @xmath544 is discrete and isomorphic to the group @xmath545 . thus we conclude that @xmath376 . in this subsection , we consider the case of @xmath625 . in this case the cyclic subgroup @xmath626 preserves the sphere @xmath627 and its orientation . one of the main result of this subsection is theorem 5.9 , which gives a necessary and sufficient condition for @xmath625 to be contained in @xmath55 . as a consequence , we will show in corollary 5.11 that the boundary of the slice of @xmath55 through the plane @xmath374 is a union of countably many analytic arcs , although the boundaries of the slices of @xmath55 through the planes @xmath182 and @xmath373 are so called fractal . " we begin with the notion of a ford domain : let @xmath628 be a kleinian group and suppose that @xmath629 for every @xmath630 . given @xmath47 we denote by @xmath631 the exterior of the isometric sphere @xmath632 of @xmath249 ; that is , @xmath631 is the connected component of @xmath633 containing @xmath144 . then the _ ford domain _ for @xmath0 is defined by @xmath634 which turns out to be a fundamental domain for @xmath0 . below , we denote by @xmath635 the radius of the isometric sphere @xmath117 of a transformation @xmath39 with @xmath111 . we will need the following lemma in the proof of theorem 5.9 . let @xmath39 be a loxodromic transformation with @xmath636 , and suppose that @xmath118 preserves the sphere @xmath637 and its orientation . let @xmath638 . then @xmath639 $ ] is loxodromic , pure parabolic or elliptic if and only if @xmath640 or @xmath641 , respectively . note that the points @xmath642 , @xmath643 lie in the plane @xmath627 . we may assume that @xmath644 after conjugating by a translation if necessary . we write @xmath116 . then one see from 2.3 that @xmath645 for some @xmath316 . moreover , since @xmath118 preserves the sphere @xmath637 and its orientation , one can see that @xmath646 , where @xmath647 . now write @xmath648 and suppose first that @xmath649 . then we see that the transformation @xmath639=fbf^{-1}b^{-1}$ ] maps the exterior of the sphere @xmath650 onto the interior of the sphere @xmath651 of radius @xmath652 with center at @xmath653 ( see figure 7 ) . note that @xmath654 touches to @xmath651 at @xmath655 , and that the transformation @xmath639 $ ] fixes the point @xmath655 . furthermore , we see that the differential of the map @xmath639 $ ] at this point is the identity . thus we conclude that @xmath639 $ ] is pure parabolic . by the same argument , we see that if @xmath656 then @xmath657 and thus @xmath639 $ ] is loxodromic , and if @xmath658 then @xmath639 $ ] fixes every points of @xmath659 and thus @xmath639 $ ] is elliptic . and @xmath651 in the case of @xmath649 . this is the view along the @xmath76-axis.,height=226 ] let @xmath625 . then @xmath376 if and only if @xmath660 for all @xmath661 . moreover , for a given @xmath661 , @xmath662 if and only if @xmath663 $ ] is pure parabolic . let @xmath625 . we first remark that @xmath664 for every @xmath665 , because @xmath666 lies in the interiors of @xmath667 or @xmath668 . therefore the ford domain @xmath669 for the cyclic group @xmath670 can be defined . in addition , observe that the center @xmath666 of @xmath671 lies in the plane @xmath627 for every @xmath665 , and by definition that @xmath672 for every @xmath661 . suppose first that @xmath660 for all @xmath661 . then the ford domain @xmath673 contains the set @xmath674 . therefore we can apply the klein maskit combination theorem ( theorem 5.3 ) to show that @xmath304 is discrete and isomorphic to the rank-@xmath2 free group @xmath675 by letting @xmath676 , @xmath677 and @xmath678 . conversely , suppose that @xmath679 for some @xmath13 . then by lemma 5.8 , @xmath663 $ ] is elliptic . if the order of this elliptic element is infinite then the group @xmath72 is non - discrete , and if the order is finite then @xmath72 is not free . in both cases , we conclude @xmath680 . the second statement also follows from lemma 5.8 . now let @xmath681 and @xmath682 . in this notation , the map @xmath683 restricted to the sphere @xmath637 is a mbius transformation given by @xmath684 where @xmath637 is identified with @xmath7 via the map @xmath685 . ( here we use the notation @xmath686 to distinguish it from @xmath412 and @xmath411 . ) the matrix representation of @xmath686 is @xmath687 we denote the entries of @xmath688 for @xmath661 by @xmath689 then the radius of the isometric sphere @xmath690 of @xmath688 is given by @xmath691 . in other words , we have @xmath692 for every @xmath661 . since @xmath693 we have @xmath694 and @xmath695 . therefore we obtain the following recurrence equations : @xmath696 especially @xmath697 is a monic @xmath569-polynomial of degree @xmath81 ; for example @xmath698 and so on . using this notation and identifying @xmath374 with @xmath240 , we can rephrase theorem 5.9 as follows . we have @xmath699 is a one - dimensional real analytic variety for every @xmath661 , we have the following corollary . the boundary of @xmath700 in the plane @xmath374 is a union of countably many analytic arcs . one can find in figures 8 and 9 the numerical graphics of the loci of @xmath33 such that @xmath701 or @xmath702 for some @xmath13 . such that @xmath701 for @xmath703 and @xmath704 . the arcs without labels correspond to @xmath705.,height=170 ] such that @xmath706 for @xmath707 . the darkness increases as the number of the intersection increases.,height=170 ] in this section we consider a family of @xmath1-generator kleinian groups in @xmath23 , which can be viewed as an analogue of the family of groups of type @xmath193 in @xmath22 . these groups contain groups of type @xmath193 in @xmath23 and their limit sets are union of round spheres . for a given @xmath708 with @xmath709 , we define an ideal hexahedron @xmath710 as follows ( see figure 10 ) : @xmath711 observe that eight edges of @xmath710 have dihedral angle @xmath712 and the rest four edges have dihedral angle @xmath566 . moreover , the following three transformations in @xmath23 pair the faces of @xmath710 : @xmath713 ( although the notation @xmath397 is common with the previous sections , we remark that @xmath270 , @xmath271 are different from those in the previous sections . ) therefore , it follows from poincar s polyhedron theorem ( cf . @xcite ) that the group @xmath714 generated by @xmath715 and @xmath271 in @xmath23 is discrete , and that @xmath716 is isomorphic to the abstract group @xmath717 ^ 2=[a , c]^2=[b , c]={{\mathrm{id}}}{\rangle}.\ ] ] for arbitrary @xmath73 , we also denote by @xmath718 the group in @xmath23 generated by the transformations @xmath397 , @xmath270 and @xmath271 as in @xmath719 . note that if @xmath716 is isomorphic to @xmath720 , the subgroup @xmath721 of @xmath716 is a group of type @xmath193 in @xmath23 because @xmath722 is a translation of length @xmath2 . for a given @xmath725 with @xmath709 , we will study deformations of @xmath726 in the next subsection . as a preparation for this purpose , in this subsection , we study deformations of the subgroup @xmath727 of @xmath728 . let @xmath729 denote a torus with a singular point of cone - angle @xmath730 . the fundamental group @xmath731 of the orbifold @xmath729 is isomorphic to an abstract group @xmath732 ^ 2={{\mathrm{id}}}{\rangle}$ ] . we say that a group @xmath0 in @xmath23 is of _ type _ @xmath724 if it is the image of a faithful representation @xmath733 ^ 2={{\mathrm{id}}}{\rangle}\to { { \mathrm{m\ddot{o}b}}({\hat { \mathbb{r}}}^3)}$ ] such that @xmath207 is pure parabolic . note that if @xmath734 is isomorphic to @xmath720 , the subgroups @xmath735 , @xmath736 of @xmath716 are of type @xmath724 in @xmath23 . similar to theorem 4.5 , we have the following normalization of groups of type @xmath724 in @xmath23 . let @xmath32 be a subgroup of @xmath23 which is isomorphic to the abstract group @xmath732 ^ 2={{\mathrm{id}}}{\rangle}$ ] and suppose that @xmath30 is pure parabolic . then @xmath0 is conjugate in @xmath23 to @xmath737 for some @xmath277 , where @xmath738 are as in @xmath719 . the proof is almost parallel to the proof of theorem 4.5 . we can first show that @xmath305 . in fact , if not , the same argument as in the proof of theorem 4.5 implies that @xmath739={{\mathrm{id}}}$ ] , which contradicts to the assumption that @xmath32 is isomorphic to the abstract group @xmath732 ^ 2={{\mathrm{id}}}{\rangle}$ ] . therefore , by using the same normalization as in the proof of theorem 4.5 , we may assume that @xmath51 are of the forms @xmath328 , @xmath740 where @xmath316 , @xmath73 and @xmath323 such that @xmath347 . then @xmath351 preserves the sphere @xmath352 and its orientation . furthermore , identifying @xmath263 with @xmath7 via the map @xmath741 and letting @xmath349 , one see that the map @xmath351 restricted to the sphere @xmath352 is a mbius transformation whose matrix representation is @xmath742 it then follows from the condition @xmath31 ^ 2={{\mathrm{id}}}$ ] that @xmath743 , and hence , that @xmath744 . the remaining argument is the same to the argument of the proof of theorem 4.5 . for a given @xmath725 with @xmath709 , we now consider deformations @xmath745 of @xmath726 in @xmath23 such that the isomorphism @xmath12 takes a pure parabolic transformation to a pure parabolic transformation . the following theorem reveals that every such deformation @xmath0 of @xmath726 is conjugate to @xmath716 for some @xmath73 . let @xmath746 be a subgroup of @xmath23 which is isomorphic to the abstract group @xmath747 ^ 2=[a , c]^2=[b , c]={{\mathrm{id}}}{\rangle}$ ] , and suppose that @xmath253 and @xmath748 $ ] are pure parabolic . then @xmath0 is conjugate in @xmath23 to @xmath734 for some @xmath73 . note that the subgroups @xmath212 , @xmath749 of @xmath0 are of type @xmath724 in @xmath23 , and @xmath750 is of type @xmath193 . by theorem 6.1 , we may assume that @xmath751 for some @xmath73 after conjugating @xmath752 if necessary . we will show that @xmath266 below . since @xmath249 is pure parabolic and commutes with @xmath270 , we have @xmath753 . since we are normalizing so that the radius of the isometric sphere of @xmath754 equals @xmath154 , it follows from theorem 6.1 that @xmath249 is a translation of length @xmath755 , and from theorem 4.5 that @xmath756 is a translation of length @xmath2 . therefore the direction of translations of @xmath270 and @xmath249 are perpendicular . this implies that @xmath757 . thus we obtain @xmath758 . we now define : @xmath759 since the group @xmath716 for @xmath760 contains the groups @xmath761 , @xmath721 of type @xmath193 in @xmath23 , the set @xmath762 lies in the rotations of the set @xmath55 of angles @xmath763 along the @xmath764-axis . one can also see that @xmath765 is invariant under the action of the translations @xmath766 , @xmath767 . the following lemma shows that the set @xmath765 contains a @xmath1-dimensional domain of @xmath41 . for a given @xmath277 with @xmath769 , we let @xmath770 . we suppose that @xmath771 for simplicity , but the argument for the case of @xmath772 is almost parallel . deforming the ideal hexahedron @xmath773 in a similar way as in figure 4 , we can obtain an ideal polyhedron which can be applied poincar s polyhedron theorem ( cf . @xcite ) to show the group @xmath718 is discrete and isomorphic to the abstract group @xmath720 . thus we conclude that @xmath760 . for a given @xmath73 , one can see that @xmath774 and @xmath775 hold . therefore the group @xmath776 always contains the group @xmath777 note that the action of @xmath778 preserves the sphere @xmath301 and its orientation . now we take a ideal pentahedron @xmath779 as follows : @xmath780 is a fundamental domain for the action of @xmath778 acting on the half - space @xmath781 . therefore @xmath778 can be regarded as a kleinian group in @xmath22 of the first - kind , that is , the limit set @xmath782 of @xmath778 is the whole sphere @xmath301 . thus the limit set @xmath783 of a kleinian group @xmath716 for @xmath760 is the closure of a union of round spheres : @xmath784 one can find in figure 11 some computer graphics of the limit sets of @xmath716 for parameters @xmath73 which lie ( or seem to be lie ) in @xmath765 . by using the observation of the subgroup @xmath785 above , we obtain an alternative proof of lemma 6.3 . in fact , a kleinian group @xmath716 for @xmath40 with @xmath769 can be regarded as an amalgamation of the kleinian group @xmath778 with the cyclic group @xmath670 as follows : suppose @xmath771 for simplicity . then we can apply the second klein maskit combination theorem ( theorem 5.3 ) to show that @xmath786 is discrete and @xmath787 by letting @xmath788 @xmath789 and @xmath790 . of groups @xmath716 : these are the images @xmath791 of @xmath783 by the mbius transformation @xmath792 , @xmath793 , which takes the half - space @xmath794 to the unit ball @xmath795 and the point @xmath796 to the infinity . these are the views along the @xmath76-axis.,height=453 ] m. kapovich . _ kleinian groups in higher dimensions . _ in `` geometry and dynamics of groups and spaces . in memory of alexander reznikov '' , m. kapranov et al ( eds ) . birkhauser , progress in mathematics , vol . 265 , 2007 , p. 485 - 562 .

in the sub - riemannian heisenberg group equipped with its car- not - carathodory metric and with a haar measure , we consider _ isodiametric sets _ , i.e. sets maximizing the measure among all sets with a given diameter . in particular , given an isodiametric set , and up to negligible sets , we prove that its boundary is given by the graphs of two locally lipschitz functions . moreover , in the restricted class of _ rotationally invariant sets _ , we give a quite complete characterization of any compact ( rotationally invariant ) isodiametric set . more specifically , its steiner symmetrization with respect to the-plane is shown to coincide with the euclidean convex hull of a cc - ball . at the same time , we also prove quite unexpected non - uniqueness results .

1010.1133

the classical isodiametric inequality in the euclidean space says that balls maximize the volume among all sets with a given diameter . this was originally proved by bieberbach @xcite in @xmath1 and by urysohn @xcite in @xmath2 , see also @xcite . in this paper we are interested in the case of the heisenberg group @xmath3 equipped with its carnot - carathodory distance @xmath4 and with the haar measure @xmath5 ( see section [ sec : heisenberg ] for the definitions ) . our aim is to study _ isodiametric sets _ , i.e. sets maximizing the measure among sets with a given diameter . recalling that the homogeneous dimension of @xmath3 is @xmath6 , we define the isodiametric constant @xmath7 by @xmath8 where the supremum is taken among all sets @xmath9 with positive and finite diameter . sets realizing the supremum do exist , see @xcite or theorem [ thm : existence ] below . since the closure of any such set is a compact set that still realizes the supremum , we consider the class @xmath10 of compact isodiametric sets , @xmath11 in other words , due to the presence of dilations in @xmath3 , @xmath10 denotes the class of compact sets that maximize the @xmath5-measure among all sets with the same diameter . in contrast to the euclidean case , balls in @xmath12 are not isodiametric ( @xcite ) and we shall give in this paper some further and refined evidence that the situation is indeed quite different from the euclidean one . before describing our main results let us recall some classical motivations and consequences coming from the study of isodiametric type problems . first the isodiametric constant @xmath7 coincides with the ratio between the measure @xmath5 and the @xmath13-dimensional hausdorff measure @xmath14 in @xmath12 , namely , @xmath15 where @xmath16 . this can actually be generalized to any carnot group equipped with a homogeneous distance ( @xcite ) , and for abelian carnot groups one recovers the well - known euclidean situation . as a consequence , the knowledge of the numerical value of the isodiametric constant @xmath7 , or equivalently the explicit description of isodiametric sets , gives non trivial information about the geometry of the metric space @xmath12 and about the measure @xmath14 which may be considered as a natural measure from the metric point of view . let us also mention that there are some links with the besicovitch 1/2-problem which is in turn related to the study of the connections between densities and rectifiability . see @xcite for an introduction and known results about the besicovitch 1/2-problem and @xcite for the connection between the isodiametric problem in carnot groups and the besicovitch 1/2-problem . our main results in the present paper are a regularity property for sets in @xmath10 and a rather complete solution to a restricted isodiametric problem within the class of so - called rotationally invariant sets . let us first describe our regularity result . we shall prove that given @xmath17 then @xmath18 is still a compact isodiametric set with the same diameter as @xmath19 and with locally lipschitz boundary . more precisely , identifying @xmath3 with @xmath20 ( see section [ sec : heisenberg ] ) , we prove that @xmath21\in { { \mathbb{h}^n}}~;~z\in \overline u,~f^-(z)\leq t \leq f^+(z)\}\ ] ] for some open set @xmath22 in @xmath23 and some continuous maps @xmath24 , @xmath25 that are locally lipschitz continuous on @xmath22 . see theorem [ thm : regisodiam ] for a complete statement . this regularity property will actually follow from a slightly more general result . we will prove that a set @xmath17 must satisfy the following necessary condition , @xmath26 see proposition [ prop : neccdtrefined ] , and is @xmath27-convex , see subsection [ subsec : geometric - transf ] for the definition of @xmath27-convexity and proposition [ prop : tconv - isodiam ] . independently from the isodiametric problem , the property together with @xmath27-convexity turn out to imply the regularity properties sketched above . see theorem [ thm : regboundary ] . as already mentioned , one knows that balls in @xmath12 are not isodiametric and isodiametric sets in @xmath12 are actually not explicitly known so far . this question turns out to be a challenging and rather delicate one . however , restricting ourselves to the family @xmath28 of so - called rotationally invariant sets , we are able to give a rather complete picture of the situation for compact isodiametric sets within this class . as we shall explain below , this picture will give some further information about the class @xmath10 . this may also hopefully give some insight towards a complete solution of the general isodiametric problem in @xmath12 . we shall denote by @xmath29 the class of compact sets in @xmath28 that are isodiametric within the class @xmath28 . see section [ sec : isodiampb ] for the definition of the class @xmath28 and of this restricted isodiametric problem . first it is not hard to check that sets in @xmath29 satisfy and are @xmath27-convex , see proposition [ prop : neccdtrefined ] and proposition [ prop : tconv - isodiam ] , and hence satisfy the regularity properties of theorem [ thm : regboundary ] . next our main specific result concerning sets @xmath30 is the characterization of their steiner symmetrization @xmath31 with respect to the @xmath23-plane ( see subsection [ subsec : geometric - transf ] for the definition of @xmath31 ) . we prove that if @xmath30 then @xmath31 belongs to @xmath29 and has the same diameter as @xmath19 . moreover we prove that @xmath31 is actually uniquely determined once the diameter of @xmath19 is fixed , i.e. , @xmath32 for some peculiar set @xmath33 , see theorem [ thm : rotisodiamsets ] . given @xmath34 , the set @xmath35 can be guessed via the following argument . one starts with the ball in @xmath12 centered at the origin and with diameter @xmath36 . as already said it does not satisfy . thus one can enlarge it around points where fails and get a set with still the same diameter but with greater measure . one can actually try to enlarge it as much as possible without increasing the diameter , remaining in the class @xmath28 , and preserving the property that it coincides with its steiner symmetrization with respect to the @xmath23-plane . in such a way , one ends up with a maximal set @xmath35 that satisfies . it turns out that this set is the closed convex hull ( in the euclidean sense when identifying @xmath3 with @xmath37 ) of the ball in @xmath12 centered at the origin and with diameter @xmath36 . see figure [ ccball ] . of the unit ball ( left ) and of the boundary @xmath38 of its euclidean convex hull ( right ) with a plane containing the vertical @xmath27-axis.,title="fig:",width=188 ] of the unit ball ( left ) and of the boundary @xmath38 of its euclidean convex hull ( right ) with a plane containing the vertical @xmath27-axis.,title="fig:",width=188 ] we also construct small suitable perturbations of the set @xmath35 that preserve its lebesgue measure and its diameter , see proposition [ prop : aphi ] . considering rotationally invariant perturbations , this gives the non uniqueness of sets in @xmath29 . this non uniqueness has to be understood in an `` essential '' sense , i.e. , also up to left translations and dilations . see corollary [ cor : nonuniqueness ] . finally , considering non rotationally invariant pertubations , one gets the existence of sets in @xmath10 that are not rotationally invariant , even modulo left translations , see corollary [ cor : nonrotinvariant ] . this gives one more significant difference with the euclidean situation . see figure [ a2pert ] . with a plane containing the vertical @xmath27-axis.,width=188 ] the paper is organized as follows . in section [ sec : heisenberg ] we introduce the heisenberg group @xmath3 and recall basic facts about the carnot - carathodory distance @xmath4 and balls in @xmath12 . we also introduce @xmath27-convexification and the steiner symmetrization with respect to the @xmath23-plane . section [ sec : isodiampb ] is devoted to existence and regularity results , while in section [ sec : rotisodiamsets ] we prove our more specific results about the isodiametric problem restricted to the class @xmath28 . the heisenberg group @xmath3 is a connected and simply connected lie group with stratified lie algebra ( see e.g. @xcite ) . we identify it with @xmath20 and denote points in @xmath3 by @xmath39 $ ] , where @xmath40 and @xmath41 . the group law is @xmath42\cdot [ z',t ' ] = [ z + z ' , t+t'+ 2 { \operatorname{im}}z \overline z ' ] \ ] ] where @xmath43 . the unit element is @xmath44 $ ] . there is a natural family of dilations @xmath45 on @xmath3 defined by @xmath46)= [ \lambda z , \lambda^2 t]$ ] for @xmath47 . we define the canonical projection @xmath48 as @xmath49 ) = z\ ] ] for any @xmath39\in { { \mathbb{h}^n}}$ ] . given @xmath50 , we shall frequently denote by @xmath51 $ ] its coordinates in @xmath52 . we also represent @xmath3 as @xmath53 through the identification @xmath39\approx ( x , y , t)$ ] , where @xmath54 , @xmath55 , @xmath56 and @xmath57 with @xmath58 . the @xmath59-dimensional lebesgue measure @xmath5 on @xmath60 is a haar measure of the group . it is @xmath13-homogeneous with respect to dilations , @xmath61 for all measurable @xmath9 and @xmath47 . the horizontal subbundle of the tangent bundle is defined by @xmath62 where the left invariant vectors fields @xmath63 and @xmath64 are given by @xmath65 setting @xmath66 the only non trivial bracket relations are @xmath67 = -4t$ ] , hence the lie algebra of @xmath3 admits the stratification @xmath68 . we fix a left invariant riemannian metric @xmath69 on @xmath3 that makes @xmath70 an orthonormal basis . the carnot - carathodory distance between any two points @xmath71 and @xmath72 is then defined by @xmath73 where a curve is said to be horizontal if it is absolutely continuous and such that at a.e . every point its tangent vector belongs to the horizontal subbundle @xmath74 of the tangent bundle . recall that by chow - rashevsky theorem any two points can be joined by a horizontal curve of finite length . therefore the function @xmath4 turns out to be a distance . it induces the original topology of the group , it is left - invariant , i.e. , @xmath75 for all @xmath71 , @xmath76 , @xmath77 , and one - homogeneous with respect to dilations , i.e. , @xmath78 for all @xmath71 , @xmath72 and @xmath47 . equipped with this distance , @xmath3 is a separable and complete metric space in which closed bounded sets are compact . we will denote by @xmath79 , respectively @xmath80 , the open , respectively closed , ball with center @xmath50 and radius @xmath81 . note that the diameter of any ball in @xmath12 is given by twice its radius . [ lem : distopen ] let @xmath50 . the distance function @xmath82 from @xmath71 defined by @xmath83 is an open map . we prove that @xmath84 is open for any open ball @xmath85 . since balls in @xmath12 are connected ( this is more generally true in any length space ) it follows that @xmath84 is a bounded interval . setting @xmath86 and @xmath87 it is thus enough to prove that @xmath88 and @xmath89 . if @xmath90 then @xmath91 because @xmath92 . if @xmath93 we assume by contradiction that @xmath94 then we can find @xmath95 such that @xmath96 . the map @xmath97 being continuous , one has @xmath98 for all @xmath36 close enough to 1 . it follows that @xmath99 provided @xmath100 is close enough to 1 , which gives a contradiction . the fact that @xmath89 can be proved in a similar way and this concludes the proof . [ rmk : distopen ] as an immediate consequence of lemma [ lem : distopen ] we get that for any set @xmath9 and any @xmath101 , @xmath102 moreover , if @xmath103 is bounded then @xmath104 . although the carnot - carathodory distance between any two points is in general hardly explicitly computable , we recall for further reference the following well - known peculiar cases . one has @xmath105,[z',t ] ) = \|z'-z\|\ ] ] for all @xmath106 , @xmath107 such that @xmath108 and all @xmath41 . here @xmath109 for @xmath110 . we have also @xmath111,[z , t ' ] ) = ( \pi |t'-t|)^{1/2}\ ] ] for all @xmath112 and @xmath27 , @xmath113 . explicit descriptions of balls in @xmath12 are well - known , see @xcite , @xcite , @xcite . one has @xmath114\in{{\mathbb{h}^n}}~;~\chi\in{{\mathbb c}^n}~,~\|\chi\|\leq 1~,~{\varphi}\in[-\pi,\pi]\big\}~.\end{gathered}\ ] ] we set @xmath115 the function @xmath69 admits a maximum at @xmath116 with @xmath117 . it is increasing from @xmath118 $ ] onto @xmath119 $ ] and decreasing from @xmath120 $ ] onto @xmath121 $ ] . we set @xmath122 the function @xmath123 is decreasing from @xmath124 $ ] onto @xmath125 $ ] . let @xmath126 \rightarrow [ 0,\pi]$ ] denote its inverse and set @xmath127 we have the following other description of the closed unit ball @xmath128\in{{\mathbb{h}^n}}~;~ \|z\|\leq 1,~|t| \leq h(\|z\|)\}~.\ ] ] using dilations we have @xmath129\in{{\mathbb{h}^n}}~;~ \|z\|\leq \lambda,~|t| \leq h_\lambda(\|z\|)\}\ ] ] for all @xmath34 , where @xmath130 .,width=188 ] we list some properties of the function @xmath131 that will be needed in the sequel . see figure [ grafacca ] for a picture . * the map @xmath131 is increasing from @xmath119 $ ] onto @xmath121 $ ] and decreasing from @xmath132 $ ] onto @xmath119 $ ] . * there exists @xmath133 such that @xmath134 on @xmath135 , @xmath136 , and @xmath137 on @xmath138 . indeed we have @xmath139 and @xmath140 for all @xmath141 and where @xmath142 $ ] . then statement ( i ) follows from the expression of @xmath143 together with the properties of @xmath144 . statement ( ii ) follows noting that there exists a unique @xmath145 such that @xmath146 and that one has @xmath147 if and only if @xmath148 . we call vertical segment any set of the form @xmath149\in{{\mathbb{h}^n}}~;~ t^-\leq t\leq t^+\}$ ] for some @xmath150 . given @xmath71 , @xmath151 with @xmath152 , we denote by @xmath153 the vertical segment joining @xmath71 to @xmath76 , @xmath154\in{{\mathbb{h}^n}}~;~\min(t_p , t_q)\leq t \leq \max(t_p , t_q)\}~.\ ] ] in the next proposition we state an elementary geometric property of balls in @xmath3 for further reference . when not specified , by ball we mean a ball that can be indifferently taken as open or closed . [ prop : ballsproperties ] the following statements hold . * let @xmath155 denote a ball in @xmath3 . for any @xmath71 , @xmath95 such that @xmath156 , we have @xmath157 . * for any @xmath158 and any @xmath159 , @xmath160 such that @xmath161 , we have @xmath162 for all @xmath163 . property ( i ) holds for the ( closed or open ) unit ball by . then this property follows for any ball using dilations and translations and noting that these maps are bijective maps that send vertical segments onto vertical segments . to prove property ( ii ) set @xmath164 . we have @xmath159 , @xmath165 . hence @xmath166 by ( i ) and thus ( ii ) follows . in the next lemma we deal with an outer vertical cone property for balls centered at the origin . its proof ( that we provide for the reader s convenience ) follows from the local lipschitz continuity of the profile function @xmath131 on @xmath167 . [ coneball ] let @xmath168 and @xmath169 be fixed . there exists @xmath170 such that the following holds . if @xmath171 is such that @xmath172 , respectively @xmath173 , and @xmath174 \in { { \mathbb{h}^n}}$ ] is such that @xmath175 then @xmath174\notin \overline b(0,d)$ ] . let @xmath176 to be chosen later . let us consider the case where @xmath171 and @xmath174 \in { { \mathbb{h}^n}}$ ] are such that @xmath172 and @xmath177 , the other case being analogous . if @xmath178 then we obviously have @xmath174\notin \overline b(0,d)$ ] . if @xmath179 we want to prove that @xmath180 . we have @xmath181 and thus it will be sufficient to show that @xmath182 since @xmath183 , one can find @xmath184 such that @xmath185 for all @xmath186 $ ] . in particular @xmath187 because @xmath188 . we choose @xmath189 to be the lipschitz constant of @xmath190 on @xmath191 $ ] . if @xmath192 it follows @xmath193 as wanted . whereas @xmath194 if @xmath195 $ ] which concludes the proof [ rem : alpha ] we note that , for any fixed @xmath168 , the function @xmath196 can be taken continuous w.r.t . the variable @xmath197 . this follows from the definition of @xmath198)}$ ] together with the fact that @xmath199 can be chosen to be continuous and the fact that the map @xmath200)}$ ] is continuous . in this subsection we introduce two geometric transformations , namely the convexification along the vertical @xmath27-axis and the steiner symmetrization with respect to the @xmath23-plane . they will play a crucial role in the sequel . given @xmath201 we define its @xmath27-convex hull @xmath202 by @xmath203 we say that @xmath103 is @xmath27-convex if @xmath204 . [ lem : tco ] let @xmath9 . we have @xmath205 and @xmath206 . we obviously have @xmath205 and in particular @xmath207 . conversely let @xmath71 , @xmath208 . one can find @xmath159 , @xmath209 with @xmath161 such that @xmath210 . then it follows from proposition [ prop : ballsproperties](ii ) that @xmath211 similarly one can find @xmath212 , @xmath213 with @xmath214 such that @xmath215 . then it follows once again from proposition [ prop : ballsproperties](ii ) that @xmath216 which concludes the proof . given @xmath201 measurable , its steiner symmetrization @xmath217 with respect to the @xmath23-plane is defined by @xmath218\in{{\mathbb{h}^n}}~;~z\in \pi(f),~2|t|\leq { \mathcal{l}^{1}}(\{s\in{\mathbb{r}}~;~[z , s]\in f\})\}\ ] ] where @xmath219 denotes the one - dimensional lebesgue measure and @xmath220 is the canonical projection defined in . we define the reflection map @xmath221 as @xmath222)=[\overline z , t].\ ] ] for the sake of simplicity , the following lemma is stated for compact sets . this will be the only case needed in this paper . it can however be easily generalized to non compact sets . [ lem : diamste ] let @xmath9 be compact and such that @xmath223 . then @xmath224 . since @xmath103 is a compact subset of @xmath3 , then @xmath202 is compact and is obviously @xmath27-convex . we have @xmath225 as soon as @xmath223 . since @xmath226 and @xmath206 by lemma [ lem : tco ] , it is thus sufficient to consider @xmath27-convex compact sets @xmath103 such that @xmath223 . then we can describe @xmath103 as @xmath227\in{{\mathbb{h}^n}}~;~z\in \pi(f),~ a(z ) - \mathcal{l}(z ) \leq t \leq a(z ) + \mathcal{l}(z)\}\ ] ] for some map @xmath228 which satisfies @xmath229 for all @xmath230 and where @xmath231\in f\})/2 $ ] . note that we also have @xmath232 for any @xmath230 . let @xmath233\in { \operatorname{st}}f$ ] and @xmath234 \in { \operatorname{st}}f$ ] . set @xmath235\in f~;~z = z_1~\text{or}~z=\overline z_1\}~ , \\ f_2 = \{[z , t]\in f~;~z = z_2~\text{or}~z=\overline z_2\}~.\end{gathered}\ ] ] we will prove that @xmath236 since @xmath237 this will imply @xmath238 as wanted . set @xmath239\cdot f_i$ ] , @xmath240 , i.e. , @xmath241\\ \hat f_2 = \{z_2,\overline z_2\ } \times [ b-\mathcal{l}(z_2),b+\mathcal{l}(z_2)]\end{gathered}\ ] ] where @xmath242 . the distance being left invariant , we have @xmath243 next set @xmath244 . we have @xmath245 $ ] hence it follows from proposition [ prop : ballsproperties](ii ) that @xmath246 ) , d(p_1,[z_2 , -t_2]))~.\ ] ] assume that @xmath247 . then @xmath248\in \hat f_2 $ ] and since @xmath249 we get @xmath250 ) \leq \max(d(q_1,q_2)~;~q_1\in \hat f_1,~q_2\in \hat f_2)~.\ ] ] let @xmath251 denote the isometry in @xmath12 defined by @xmath252 ) = [ \overline z , -t].\ ] ] we have @xmath253 and @xmath254 ) = [ \overline z_2,t_2 ] \in \hat f_2 $ ] , hence @xmath255 ) = d(\iota(p_1),\iota([z_2,-t_2 ] ) ) \leq \max(d(q_1,q_2)~;~q_1\in \hat f_1,~q_2\in \hat f_2)~.\ ] ] inequalities , and together with give . the case where @xmath256 can be traited in a similar way and this concludes the proof . we recall the definitions of the isodiametric constant @xmath257 and of the class of compact isodiametric sets @xmath11 recall that it is not restrictive to ask isodiametric sets to be compact as the closure of any set which realizes the supremum in the right - hand side of the definition of @xmath7 is a compact set that still realizes the supremum . we also introduce the class of so - called rotationally invariant sets . given @xmath258 , we define the rotation @xmath259 around the @xmath260-axis by @xmath261 ) = [ ( e^{i\theta_1}z_1,\dots , e^{i\theta_n}z_n),t]~.\ ] ] any such @xmath262 is an isometry in @xmath12 . we denote by @xmath28 the class of rotationally invariant sets , @xmath263 we set @xmath264 and denote by @xmath29 the family of compact rotationally invariant sets that are isodiametric within the class of rotationally invariant sets , @xmath265 in other words , @xmath10 , resp . @xmath29 , denotes the class of compact sets , resp . compact sets in @xmath28 , that maximize the @xmath5-measure among all subsets of @xmath3 , resp . among all sets in @xmath28 , with the same diameter . we first prove the existence of sets in @xmath10 and @xmath29 . [ thm : existence ] both families @xmath10 and @xmath29 are nonempty . the proof that @xmath10 is non empty relies on the compactness of equibounded sequences of non empty compact sets with respect to the hausdorff metric ( see 2.10.21 in @xcite ) together with the upper - semicontinuity of the lebesgue measure ( see theorem 3.2 in @xcite ) . the fact that @xmath266 as well can be proved in a similar way noting that @xmath28 is closed with respect to the convergence of sets in the hausdorff metric . the rest of this section is devoted to the study of the regularity of sets in @xmath10 and @xmath29 . the necessary condition introduced in section [ sec : intro ] will be one of the key ingredients in this study and we prove in the next proposition that sets in @xmath10 and @xmath29 do satisfy this condition . [ prop : neccdtrefined ] let @xmath267 . then @xmath19 satisfies the necessary condition . we argue by contradiction . assume that @xmath17 and @xmath268 is such that @xmath269 for all @xmath270 . the distance from @xmath71 being a continuous and open map and @xmath19 being compact , we have @xmath271 ( see lemma [ lem : distopen ] and remark [ rmk : distopen ] ) and hence @xmath272 choosing @xmath273 , it follows that @xmath274 on the other hand , since @xmath19 is closed and @xmath275 , we have @xmath276 . this implies in particular that @xmath277 which contradicts the fact that @xmath17 . when @xmath278 we modify the argument as follows . we set @xmath279 where @xmath280 as before . we have @xmath281 and @xmath282 indeed , to prove we fix @xmath283 . if @xmath284 then @xmath285 . if @xmath286 , @xmath287 then there exist @xmath288 , @xmath289 such that @xmath290 , @xmath240 . recalling that any rotation @xmath262 is an isometry in @xmath12 and that @xmath291 , it follows that @xmath292 if @xmath293 and @xmath294 with @xmath295 for some @xmath296 , we have @xmath297 on the other hand , similarly as before , we have @xmath298 and this contradicts the fact that @xmath278 . let us introduce some notations . given a compact set @xmath19 , we define @xmath299 and @xmath300 as follows . we set @xmath301\in e)~,\\ f^-(z)&=\min(t\in{\mathbb{r}}~;~[z , t]\in e)\end{aligned}\ ] ] for all @xmath302 . clearly , @xmath303 \in { \partial}e$ ] for all @xmath304 . recalling the definition of @xmath305 given in subsection [ subsec : geometric - transf ] , one has @xmath306\in { { \mathbb{h}^n}}~;~ z\in \pi(e),~ f^-(z)\leq t\leq f^+(z)\},\ ] ] and @xmath305 is itself compact . we set @xmath307 and @xmath308\in { { \mathbb{h}^n}}~;~z\in\overline u,~f^-(z)\leq t\leq f^+(z)\}~.\ ] ] since @xmath309 is closed , we have @xmath310 . in particular @xmath311 and @xmath24 are well defined on @xmath312 . moreover @xmath313 is compact and contained in @xmath305 . we are now ready to state our key regularity result . it concerns @xmath27-convex and compact sets satisfying . [ thm : regboundary ] for any @xmath27-convex and compact set @xmath19 satisfying the following properties hold . * the set @xmath22 is open in @xmath23 and the maps @xmath24 and @xmath311 are locally lipschitz on @xmath22 and continuous on @xmath309 . * @xmath314 . * we have @xmath315\in { { \mathbb{h}^n}}~;~z\in u,~f^-(z)<t < f^+(z)\}~,\\ { \partial}\hat e & = \{[z , f^{\pm}(z)]\in{{\mathbb{h}^n}}~;~ z\in \overline u\}~,\\ \hat e & = \overline{{\operatorname{int}}e}~.\end{aligned}\ ] ] before starting with the proof of theorem [ thm : regboundary ] , we introduce some notations and give a technical lemma . given @xmath81 and @xmath159 , @xmath160 with @xmath316 , @xmath317 $ ] and @xmath318 $ ] for some @xmath319 and @xmath320 , @xmath321 , we set @xmath322 and @xmath323\in{{\mathbb{h}^n}}~;~\|w - z_{12}\|\leq r,~\big|s-\dfrac{t_1+t_2}{2}\big|\leq \delta_{12}\ : \big(1-\dfrac{\|w - z_{12}\|}{r}\big)\big\}\ ] ] whenever @xmath324 . clearly , @xmath325 provided @xmath326 . [ lem : fisinb ] let @xmath327 , @xmath168 and @xmath169 be fixed . there exists @xmath328 such that the following holds . for any @xmath329 $ ] , @xmath317 $ ] and @xmath318\in { { \mathbb{h}^n}}$ ] such that @xmath330 and @xmath331 , we have @xmath332 for all @xmath50 such that @xmath333 . set @xmath334 $ ] . after a left translation by @xmath335 we need to prove that for all @xmath81 small enough we have @xmath336 for all @xmath50 such that @xmath337 and @xmath338 . for such a @xmath339\in{{\mathbb{h}^n}}$ ] we have @xmath340\in\overline b(0,d)$ ] for @xmath240 , hence @xmath341 and @xmath342 . it follows that @xmath343 . recalling that @xmath344 and considering @xmath345 such that @xmath346 whenever @xmath347 $ ] , we get that @xmath348 . we have @xmath349\in{{\mathbb{h}^n}}~;~\|w\|\leq r,~|s+2 { \operatorname{im}}z_{12}\overline w|\leq \delta\big(1-\dfrac{\|w\|}{r}\big)\big\}~.\ ] ] let @xmath174 \in p_0\cdot f_{p_1,p_2}(r)$ ] and let us show that @xmath350 \in \overline b(0,d)$ ] or equivalently that @xmath351 and @xmath352 provided @xmath81 is small enough . first note that @xmath353 is less than @xmath4 provided @xmath354 . next we have @xmath355 where the last inequality follows from the fact that @xmath356=[z , t-\delta]^{-1}\in\overline b(0,d)$ ] . on the other hand if @xmath357 , we have @xmath358 . let @xmath359 denote the lipschitz constant of @xmath190 on @xmath360 $ ] . then we have @xmath361 it follows that @xmath362 provided @xmath363 . similarly we have @xmath364 provided @xmath363 and where the second inequality follows from the fact that @xmath365=[z , t+\delta]^{-1}\in\overline b(0,d)$ ] . hence the lemma follows with @xmath366 [ rmk : monotonicitygamma3 ] note that , @xmath367 and @xmath4 being fixed , the function @xmath368 can be taken to be continuous w.r.t . the variable @xmath197 . this is a consequence of and the fact that @xmath369 can be chosen continuous w.r.t . @xmath197 . we turn now to the proof of theorem [ thm : regboundary ] . we fix a @xmath27-convex and compact set @xmath19 satisfying and set @xmath370 we begin with two lemmata . the first one is a consequence of lemma [ lem : fisinb ] . [ lem : fine ] let @xmath19 be as above , @xmath169 be fixed and let @xmath371 be as in lemma [ lem : fisinb ] . then , for any @xmath372 $ ] , one has @xmath373 for all @xmath159 , @xmath374 such that @xmath316 and @xmath331 . since @xmath19 is closed , it is enough to show that @xmath375 assume by contradiction that @xmath376 is nonempty . by @xmath27-convexity of @xmath19 we know that also @xmath377 is nonempty since it contains @xmath378 . therefore there exists @xmath379 . then by there exists @xmath380 such that @xmath381 . thanks to lemma [ lem : fisinb ] , we have @xmath382 thus @xmath383 which is a contradiction . [ lem : onesidelip ] let @xmath19 be as before and let @xmath169 be fixed . there exists @xmath384 such that the following holds . if @xmath385 is such that @xmath386 , then @xmath387 for all @xmath388 . we claim that the lemma holds for @xmath389 where @xmath196 is given by lemma [ coneball ] . assume by contradiction that @xmath388 is such that @xmath390 the case where @xmath391 being analogous . since @xmath392\in{\partial}e$ ] , by there exists @xmath270 such that @xmath393,q)=d$ ] . set @xmath394 $ ] and @xmath395=q^{-1}\cdot[w , f^+(w)]$ ] . we have @xmath396\ ] ] and @xmath397=[w - z_q , f^+(w)-t_q-2{\operatorname{im}}z_q\overline{w}]~.\ ] ] it follows that @xmath398 where the inequality follows by the choice of @xmath388 . on the other hand set @xmath399 $ ] . we have @xmath400 . we also have @xmath401\in e$ ] and @xmath402 , hence @xmath403)\leq d$ ] or equivalently @xmath404 . therefore @xmath405 recalling that @xmath171 we get @xmath406 then , taking lemma [ coneball ] into account , we infer from and that @xmath395\not \in \overline b(0,d)$ ] , i.e. , @xmath407)>d$ ] . this implies that @xmath408 , a contradiction . [ rmk : monotonicitybeta ] taking into account remark [ rem : alpha ] , one can take the function @xmath409 to be continuous w.r.t . the variable @xmath197 . we prove now the continuity of @xmath24 and @xmath311 on @xmath309 . [ lem : continuity ] the functions @xmath24 and @xmath311 are continuous on @xmath309 . let @xmath302 and let us prove that @xmath311 is continuous at @xmath106 , the case of the function @xmath24 being similar . let @xmath410 be a sequence of points in @xmath309 such that @xmath411 as @xmath412 . since @xmath19 is compact , @xmath311 is bounded and to prove the continuity of @xmath311 at @xmath106 it is sufficient to prove that any possible limit of the sequence @xmath413 coincides with @xmath414 . by contradiction assume that @xmath415 as @xmath412 for some @xmath416 . since @xmath19 is compact , hence closed , and @xmath417\in e$ ] , we have @xmath39\in e$ ] . it follows in particular that , by definition of @xmath311 and @xmath24 , we must have @xmath418 . setting @xmath419 $ ] and @xmath420 $ ] , we thus have @xmath421 . owing to lemma [ lem : fine ] , one obtains that @xmath422 where @xmath423 is given by lemma [ lem : fisinb ] . therefore , recalling once again the definition of @xmath311 , we get in particular that @xmath424 , a contradiction . we turn now to the proof of the fact that @xmath22 is open and that @xmath24 , @xmath311 are locally lipschitz continuous on @xmath22 . [ lem : uopenflip ] the set @xmath425 is open and the maps @xmath24 , @xmath311 are locally lipschitz continuous on @xmath22 . let us introduce some notations . we will denote by @xmath426 the open ball in @xmath23 with center @xmath112 and radius @xmath81 . given @xmath385 , we set @xmath427 and @xmath428 where @xmath429 and @xmath430 are given , respectively , by lemma [ lem : fisinb ] and lemma [ lem : onesidelip ] . we first prove that @xmath425 is open . let @xmath385 . set @xmath431 , @xmath432 and let @xmath433 $ ] , @xmath434\in e$ ] . it follows from lemma [ lem : fine ] that @xmath435 . in particular for any @xmath436 with @xmath437 , we have @xmath174\in e$ ] for all @xmath438 \\ & = \big[t_1+\delta_z \dfrac{\|w - z\|}{\gamma_z } , t_2-\delta_z \dfrac{\|w - z\|}{\gamma_z}\big]~. \end{split}\ ] ] since @xmath439 as soon as @xmath437 , it follows that @xmath440 . hence @xmath425 is open . we prove now that @xmath24 and @xmath311 are locally lipschitz continuous on @xmath22 . let @xmath385 . by the previous argument we already know that @xmath441 for all @xmath442 . next , lemma [ lem : onesidelip ] implies that @xmath443 for all @xmath444 . we fix a compact set @xmath445 and define @xmath446 and @xmath447 . owing to the continuity of @xmath448 , @xmath449 and @xmath450 ( see lemma [ lem : continuity ] and remarks [ rmk : monotonicitygamma3 ] and [ rmk : monotonicitybeta ] ) , one has @xmath451 and @xmath452 . therefore , for any @xmath453 satisfying @xmath454 we conclude that latexmath:[\[\begin{gathered } |f^+(w)-f^+(z)|\leq l\,\|w - z\|~,\\ and @xmath24 are lipschitz continuous on @xmath456 , as desired . we are now ready to conclude the proof of theorem [ thm : regboundary ] . statement ( i ) follows from lemmata [ lem : continuity ] and [ lem : uopenflip ] . statement ( ii ) is a consequence of @xmath457\in { { \mathbb{h}^n}}~;~ z\in \pi(e)\setminus\overline u,~t = f^+(z)\}\ ] ] and of the continuity of @xmath311 . finally ( iii ) is straightforward and follows from ( i ) by standard arguments . we are going to apply theorem [ thm : regboundary ] to sets in @xmath10 and @xmath29 . in order to do this , we first need to prove that such sets are @xmath27-convex . [ prop : tconv - isodiam ] any set @xmath458 is @xmath27-convex . assume by contradiction that one can find @xmath459\in { t\textup{-co}\:}e\setminus e$ ] . by definition of @xmath305 , one must have @xmath460 and hence @xmath385 . we have @xmath461 and @xmath462 by lemma [ lem : tco ] . since @xmath463 whenever @xmath291 , this implies that @xmath464 . hence @xmath305 is a compact set that satisfies ( see proposition [ prop : neccdtrefined ] ) and is obviously @xmath27-convex . then theorem [ thm : regboundary ] applies to @xmath305 . noting that the maps @xmath465 , and consequently the set @xmath22 , associated to @xmath19 and @xmath305 coincide , we get from theorem [ thm : regboundary](iii ) that @xmath466 . since @xmath19 is closed and @xmath467 it follows that @xmath468 . in particular @xmath469 and @xmath470 . recalling that @xmath471 and that both @xmath19 and @xmath305 belong to @xmath10 , resp . @xmath29 , with @xmath472 , this gives a contradiction . noting that @xmath473 whenever @xmath19 is @xmath27-convex and that @xmath474 whenever @xmath291 , we get from theorem [ thm : regboundary](ii ) that @xmath475 , resp . @xmath476 , whenever @xmath477 , resp . @xmath30 , with @xmath478 . summing up let us gather in the next theorem properties of sets in @xmath479 proved in this section . recall that the notations used in the statement to follow are those introduced before theorem [ thm : regboundary ] . [ thm : regisodiam ] for any set @xmath267 , the following properties hold . * @xmath19 is @xmath27-convex . * @xmath480 and @xmath478 . * @xmath475 , resp . @xmath476 , whenever @xmath477 , resp . @xmath30 . * the set @xmath22 is open in @xmath23 and the maps @xmath24 and @xmath311 are locally lipschitz on @xmath22 and continuous on @xmath309 . * we have @xmath315\in { { \mathbb{h}^n}}~;~z\in u,~f^-(z)<t < f^+(z)\}~,\\ { \partial}\hat e & = \{[z , f^{\pm}(z)]\in{{\mathbb{h}^n}}~;~ z\in \overline u\}~,\\ \hat e & = \overline{{\operatorname{int}}e}~.\end{aligned}\ ] ] in this section we characterize the steiner symmetrization with respect to the @xmath23-plane of sets @xmath30 . our main result states that the set @xmath31 belongs to @xmath29 and is uniquely determined once the diameter of @xmath19 is fixed . it coincides with a peculiar set @xmath33 , defined below , that consequently also belongs to @xmath29 . constructing suitable pertubations of this set that preserve the diameter and the @xmath5-measure , see proposition [ prop : aphi ] , we also get two remarkable consequences . first , the essential non uniqueness of sets in @xmath481 , see corollary [ cor : nonuniqueness ] . second , the existence of sets in @xmath10 which are not rotationally invariant even up to isometries , see corollary [ cor : nonrotinvariant ] . given @xmath34 , we set @xmath482\\ h_{\frac{\lambda}{2}}(r ) \quad \text{if } r\in [ \lambda / \pi,\lambda /2]\\ \end{cases}\ ] ] ( see for the definition of @xmath483 ) and @xmath484\in{{\mathbb{h}^n}}~;~2\|z\|\leq \lambda,~|t|\leq l_\lambda(\|z\|)\}~.\ ] ] we have @xmath485 where @xmath486\in { { \mathbb{h}^n}}~;~\|z\|\leq \lambda/\pi,~2\pi|t|\leq \lambda^2\}\ ] ] and @xmath487\in { { \mathbb{h}^n}}~;~\lambda/\pi\leq \|z\|\leq \lambda/2,~|t|\leq h_{\frac{\lambda}{2}}(r)\ } \\ & = \overline b(0,\lambda/2 ) \cap \{[z , t]\in { { \mathbb{h}^n}}~;~\|z\|\geq \lambda/\pi\}~. \end{split}\ ] ] the set @xmath35 is the closed convex hull , in the euclidean sense when identifying @xmath3 with @xmath37 , of the ball @xmath488 in @xmath12 centered at the origin and with diameter @xmath36 . see figure [ ccball ] in section [ sec : intro ] for a picture . [ lem : diama ] let @xmath168 , @xmath492 and @xmath41 be fixed . set @xmath493\in { { \mathbb{h}^n}}~;~\|z\|=r\}\ ] ] and @xmath494\in { { \mathbb{h}^n}}~;~\|z\|\leq r\}~.\ ] ] let @xmath50 . assume that @xmath495 and @xmath496 . then @xmath497 and there exists @xmath498 such that @xmath499 and @xmath500 . we have @xmath501,[z',t])\leq d([z , t],[0,t ] ) + d([0,t],[z',t ] ) = \|z\| + \|z'\| \leq 2r$ ] for all @xmath39 $ ] , @xmath502 \in d$ ] and @xmath501,[-z , t ] ) = 2\|z\| = 2r$ ] for all @xmath39\in c$ ] ( see ) . it follows that @xmath503 hence @xmath504 for some @xmath505\in d\setminus c$ ] . set @xmath506 and @xmath507 so that @xmath508\in { { \mathbb{h}^n}}~;~\|z - z_p\|\leq \overline d,~ h^-_{p,\overline d}(z ) \leq s \leq h^+_{p,\overline d}(z)\}~.\ ] ] we have @xmath505\in { \partial}b(p,\overline d)$ ] . changing if necessary @xmath71 , @xmath367 and @xmath509 into @xmath510 , @xmath511 and @xmath512 ( see for the definition of @xmath251 ) one can assume with no loss of generality that @xmath513 . we have @xmath514 . first we note that this implies @xmath515 . otherwise , since @xmath516 we could find @xmath106 close enough to @xmath517 such that @xmath39\in d$ ] with @xmath518 . on the other hand we have @xmath519 which gives a contradiction . it follows that the map @xmath520 is well defined on an open neighbourhood of @xmath517 in @xmath23 . next since @xmath514 and @xmath76 is in the relative interior of @xmath509 in @xmath521 it follows that @xmath520 admits a local minimum at @xmath517 . assume now by contradiction that @xmath522 . then @xmath520 is differentiable at @xmath517 and we get that @xmath523 where @xmath524 . this implies that @xmath525 . indeed otherwise implies that @xmath526 and hence @xmath527 . next if @xmath528 we get from that @xmath529 with @xmath530 where the scalar product is that of @xmath531 after identifying points in @xmath23 with points in @xmath531 . it follows that @xmath532 which also holds true if @xmath526 . on the other hand , restricting to a segment @xmath533 for @xmath534 small enough we get that @xmath535 hence @xmath536 where @xmath537 $ ] is the interval where @xmath143 is increasing ( see subsection [ subsec : balls ] ) . since @xmath538 it follows that @xmath539 all together we finally get @xmath540 recalling that @xmath541 this gives a contradiction and concludes the proof . we go back now to the proof of proposition [ prop : diama ] . we have @xmath542 ( recall remark [ rmk : distopen ] ) . if we set @xmath543\in { { \mathbb{h}^n}}~;~\|z\|=1/\pi,~2\pi t=\pm 1\}~,\ ] ] @xmath544\in { { \mathbb{h}^n}}~;~\|z\|\leq 1/\pi,~2\pi t=\pm 1\}~,\ ] ] we have @xmath545 hence @xmath546 . let @xmath547 . we have @xmath548 hence @xmath549 . if @xmath550 it follows from lemma [ lem : diama ] that @xmath551 which contradicts the fact that @xmath547 . hence @xmath552 for all @xmath547 . since @xmath553 it follows that @xmath554 . similarly we have @xmath555 . it thus only remains to check that @xmath556 . let @xmath557 . we have @xmath558 hence @xmath549 by the previous argument . if @xmath550 it follows from lemma [ lem : diama ] that one can find @xmath559 such that @xmath499 and @xmath560 . on the other hand we have @xmath561 and it follows from that @xmath562 which gives a contradiction . hence @xmath552 for any @xmath557 and since @xmath563 we finally get @xmath556 as wanted . [ lem : isodiamsubsetcyl ] let @xmath564 . assume moreover that @xmath19 is symmetric with respect to the @xmath23-plane , i.e. , @xmath565\in e$ ] for all @xmath39\in e$ ] . then @xmath566\in { { \mathbb{h}^n}}~;~2\|z\|\leq { \operatorname{diam}}e,~2\pi |t|\leq ( { \operatorname{diam}}e)^2\}~.\ ] ] let @xmath39\in e$ ] . since @xmath291 we have @xmath567\in e$ ] and it follows from that @xmath568,[z , t ] ) \leq { \operatorname{diam}}e$ ] . we also have @xmath565\in e$ ] by assumption . it follows from that @xmath569,[z ,- t])\leq { \operatorname{diam}}e$ ] and this concludes the proof . first we note that since @xmath19 is compact @xmath31 is also compact . indeed , since @xmath19 is bounded , @xmath31 is also obviously bounded . next the fact that @xmath19 is compact implies that the map @xmath572\in e\})$ ] is upper semi - continuous . it follows that @xmath31 is closed and hence compact . since @xmath564 , we have @xmath573 and it follows from lemma [ lem : diamste ] that @xmath574 . on the other hand one has @xmath575 . since @xmath30 and @xmath576 , one actually has @xmath577 and @xmath570 . we set @xmath578 and we prove now that @xmath579 . noting that @xmath580 and taking into account the fact that @xmath570 is also symmetric with respect to the @xmath23-plane , it follows from theorem [ thm : regisodiam ] that @xmath581\in { { \mathbb{h}^n}}~;~ z\in\pi(e),~ |t|\leq f(z)\}\ ] ] for some continuous map @xmath582 and that the set @xmath583 is open in @xmath23 . we have @xmath584 by lemma [ lem : isodiamsubsetcyl ] and we prove now that @xmath585 for all @xmath304 . in case @xmath586 , we note that since @xmath22 is open we must have @xmath587 for any @xmath304 such that @xmath588 . next we know from lemma [ lem : isodiamsubsetcyl ] that @xmath589 for all @xmath302 such that @xmath590 . it thus only remains to prove that @xmath585 for @xmath302 such that @xmath591 . given such a @xmath106 we assume by contradiction that @xmath592 and set @xmath593 \in a_{\lambda } \cap { \partial}b(0,\lambda/2)$ ] . since @xmath594 is continuous and @xmath106 belongs to the open set @xmath22 , one can find an open set @xmath595 containing @xmath106 and such that @xmath596 for all @xmath597 . since @xmath598\in{{\mathbb{h}^n}}~;~w\in v,~|s|\leq f(\|w\|)\}\subset{\operatorname{st}}e$ ] it follows that @xmath599 . next , since @xmath600 with @xmath601 , we can write @xmath71 as @xmath602\ ] ] for some @xmath603 such that @xmath604 and some @xmath605 ( see e.g. @xcite ) . it follows from ( * ? ? ? * lemma 1.11 ) that @xmath606 where @xmath607~.\ ] ] since @xmath608 and since the distance function from @xmath76 is an open map ( see lemma [ lem : distopen ] ) we can find @xmath609 such that @xmath610 . on the other hand we have @xmath611.\ ] ] since @xmath612 is symmetric with respect to the @xmath23-plane , we get that @xmath613 , i.e. a contradiction . it follows that @xmath614 . since @xmath615 by proposition [ prop : diama ] , and since @xmath616 and @xmath570 , we get that @xmath617 . being @xmath31 closed , we obtain that @xmath618 . it then follows that @xmath619 and finally @xmath620 as wanted . we now show that @xmath35 can be perturbed near the @xmath27-axis in such a way that the resulting set has the same volume and diameter as @xmath35 . as we shall see , the class of such perturbations is quite rich . it contains in particular rotationally and non rotationally invariant sets . [ prop : aphi ] there exists @xmath621 such that for every @xmath34 and for any lipschitz function @xmath622 with compact support in @xmath623 and with lipschitz constant @xmath624 , the set @xmath625\in { { \mathbb{h}^n}};\ ; \|z\|\leq \lambda/\pi,\ 2\pi |t - f(z)| \leq \lambda^2\ } \cup a_\lambda^2\ ] ] satisfies @xmath626 and @xmath627 . the fact that @xmath626 is a consequence of the definition of @xmath628 and of fubini s theorem . to prove the last part of the lemma , we assume that @xmath629 without loss of generality . since @xmath630 and @xmath553 we have @xmath631 . to complete the proof , we will show that the inequality @xmath632 holds up to a suitable choice of @xmath633 . first , for a given @xmath634 , we define @xmath635 so that @xmath636\in { { \mathbb{h}^n}}~;~\|z - z_p\|\leq 1,~ h^-_{p}(z ) \leq s \leq h^+_{p}(z)\}$ ] . we set @xmath637 . indeed one can find @xmath639 such that @xmath643 for all @xmath644 . then we get @xmath645 for all @xmath634 such that @xmath640 and all @xmath112 such that @xmath646 . here the scalar product is that of @xmath531 after identifying points in @xmath23 with points in @xmath531 . this gives . the proof of is similar . to prove this claim , we set @xmath652\in { { \mathbb{h}^n}};\ \|z\|\geq \overline r\}\subset \overline b(p_0,1)\cap \overline b(p_0^{-1},1)\ ] ] ( recall that @xmath335 , @xmath653 and @xmath654 ) . since @xmath655 we actually have @xmath656 by compactness of @xmath456 and continuity of @xmath657 , one can then find @xmath650 such that @xmath658 for any @xmath659 and @xmath660 . this proves claim @xmath648 . fix @xmath661 such that and hold for all @xmath634 satisfying @xmath640 . then choose @xmath650 such that holds . set @xmath662 . let @xmath622 be a lipschitz function with compact support in @xmath663 and with lipschitz constant @xmath664 . we have @xmath665 hence @xmath666\in { { \mathbb{h}^n}}~;~ \|z\| < r~,~ 4\:|t- \dfrac{1}{2\pi}| < \pi r^2\big\ } \\ & \phantom{ .... }\cup \big\{[z , t]\in { { \mathbb{h}^n}}~;~ \|z\| < r~,~ 4\:|t+ \dfrac{1}{2\pi}| < \pi r^2\big\}\\ & \subset b(p_0,\pi r/2 ) \cup b(p_0^{-1},\pi r/2)\\ & \subset b(p_0,\hat r)\cup b(p_0^{-1},\hat r)~. \end{split}\ ] ] if @xmath671 then @xmath672 and by we get @xmath673 . if @xmath674 then @xmath675 if @xmath676 , since @xmath677 and @xmath678 , we have @xmath679 . therefore @xmath680 if @xmath681 we note that @xmath640 and @xmath682 . then by we get @xmath683 similarly , using , @xmath684 hence we have @xmath685 that is @xmath686 . in other words , although there is uniqueness modulo steiner symmetrization with respect to the @xmath23-plane for sets in @xmath29 with a given diameter , we have essential non uniqueness of sets in @xmath29 . consider a set @xmath628 given by proposition [ prop : aphi ] for some @xmath34 and where @xmath691 is moreover chosen in such a way that @xmath692 . by theorem [ thm : rotisodiamsets ] , we have @xmath693 . on the other hand , by proposition [ prop : aphi ] , we have @xmath626 and @xmath627 . therefore @xmath694 . let us prove that @xmath695 for all @xmath50 . first we note that if @xmath696 and @xmath103 is bounded then @xmath697 if and only if @xmath526 . next it is straightforward from the analytic description of @xmath35 and @xmath628 that @xmath695 for any @xmath634 such that @xmath526 . this concludes the proof . if @xmath699 then there is nothing to prove . if @xmath700 then @xmath701 . in particular it then follows from theorem [ thm : rotisodiamsets ] that @xmath702 for any @xmath34 . let @xmath628 be given by proposition [ prop : aphi ] where @xmath594 is moreover chosen in such a way that @xmath703 . by proposition [ prop : aphi ] , we have @xmath626 and @xmath627 , and hence @xmath704 . next we check that @xmath705 for all @xmath50 . assume by contradiction that @xmath706 for some @xmath50 . then we get in particular that @xmath707 for all @xmath708 where @xmath709 is the rotation in @xmath23 defined by @xmath710 . on the other hand we have @xmath711 . all together this implies that @xmath526 . then we get that @xmath712 is the vertical left translation by @xmath713 $ ] of @xmath714 and hence belongs to @xmath28 , a contradiction . l. capogna , d. danielli , s.d . pauls , j.t . tyson , _ an introduction to the heisenberg group and the sub - riemannian isoperimetric problem . _ progress in mathematics , 259 . birkhuser verlag , basel , 2007 . xvi+223 pp . stein , _ harmonic analysis : real - variable methods , orthogonality , and oscillatory integrals . _ princeton mathematical series , 43 . monographs in harmonic analysis , iii . princeton university press , princeton , nj , 1993 .

this paper is concerned with a generalized type of darboux transformations defined in terms of a twisted derivation satisfying where is a homomorphism . such twisted derivations include regular derivations , difference and-difference operators and superderivatives as special cases . remarkably , the formulae for the iteration of darboux transformations are identical with those in the standard case of a regular derivation and are expressed in terms of quasideterminants . as an example , we revisit the darboux transformations for the manin - radul super kdv equation , studied in q.p . liu and m. maas , physics letters b * 396 * 133140 , ( 1997 ) . the new approach we take enables us to derive a unified expression for solution formulae in terms of quasideterminants , covering all cases at once , rather than using several subcases . then , by using a known relationship between quasideterminants and superdeterminants , we obtain expressions for these solutions as ratios of superdeterminants . this coincides with the results of liu and maas in all the cases they considered but also deals with the one subcase in which they did not obtain such an expression . finally , we obtain another type of quasideterminant solutions to the main - radul super kdv equation constructed from its binary darboux transformations . these can also be expressed as ratios of superdeterminants and are a substantial generalization of the solutions constructed using binary darboux transformations in earlier work on this topic .

0911.1413

there have been many papers written recently on noncommutative versions of soliton equations , such as the kp equation , the kdv equation , the hirota - miwa equation and the two - dimensional toda lattice equation @xcite . it has been shown that noncommutative integrable systems often have solutions expressed in terms of quasideterminants @xcite . in @xcite for example , two families of solutions of the noncommutative kp equation were presented which were termed quasiwronskians and quasigrammians . the origin of these solutions was explained by darboux and binary darboux transformations . the quasideterminant solutions were then verified directly using formulae for derivatives of quasideterminants ( see also @xcite ) . supersymmetric integrable systems are a particular noncommutative extension of integrable systems , and have attracted much attention because of their applications in physics . some well - known example are supersymmetric versions of the kdv , kp , sine - gordon , nonlinear schrdinger equation , akns and harry dym equations @xcite . among these , the manin - radul super kdv equation is perhaps the best known and has been studied extensively and a number of interesting properties have been established . we mention here the existence of an infinite number of conservation laws , a bi - hamiltonian structure @xcite , a bilinear form @xcite and darboux transformations @xcite . partly motivated by the properties of superderivatives , which we describe in section 2 , we consider a generalized derivation which has regular derivations , difference operators , @xmath3-difference operators and superderivatives as some of its special cases . we call this a twisted derivation , following the terminology used in @xcite . we show that one can formulate darboux transformations for such twisted derivations and the iteration formulae are expressed in terms of quasideterminants in which one simply replaces the derivative with the twisted derivation . a special case of this result can be used to construct solutions to supersymmetric equations in terms of quasideterminants . in @xcite , solutions for the manin - radul super kdv equation were constructed by iterating darboux transformations by considering the cases of an even number and an odd number of iterations separately . all but one of the formulae obtained by the authors expressed the solutions in terms of superdeterminants . in this paper , we use an alternative approach to the darboux transformations using quasideterminants . this is successful in obtaining unified formulae for the solutions , not depending on the parity of the number of iterations . from these quasideterminant solutions , we are not only able to recover superdeterminant solutions given in @xcite but also get the superdeterminant representation in the one case they did not . the paper is organized as follows . in section 2 , we give a brief review on superdeterminants , quasideterminants and the relationship between them . in section 3 , we define a twisted derivation and related darboux transformation . we obtain a formula for iteration of this twisted darboux transformation in terms of quasideterminants . section 4 gives applications of both darboux and binary darboux transformations to the manin - radul super kdv system . in this section we obtain two general solution formulae in terms of quasideterminants , obtained using iterated darboux and binary darboux transformations respectively , and then show how these can be expressed in terms of superdeterminants . on the other hand , we present quasideterminant solutions constructed from its binary darboux transformations to the mrskdv system . in section 5 we present conclusions . in this section , we collect together some basic facts about supersymmetric objects such as superderivatives , supermatrices , supertranspose and superdeterminants @xcite , about quasideterminants @xcite and about the relationship between superdeterminants and quasideterminants @xcite . the reader is referred to the above mentioned literature for more details . let @xmath4 be a supercommutative , associative , unital superalgebra over a ( commutative ) ring @xmath5 . there is a standard @xmath6-grading @xmath7 such that @xmath8 . elements of @xmath4 that belong to either @xmath9 or @xmath10 are called _ homogeneous _ ; those in @xmath9 are called _ even _ and those in @xmath10 are called _ odd_. the _ parity _ @xmath11 of a homogeneous element @xmath12 is 0 if it is even and 1 if it is odd . it follows that if @xmath13 are homogeneous then @xmath14 . supercommutativity means that all homogeneous elements @xmath13 satisfy @xmath15 , i.e. even elements commute with all elements , and odd elements anticommute . in particular , this implies that @xmath16 , for all @xmath17 . [ [ grade - involution - and - superderivative ] ] grade involution and superderivative + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the homomorphism @xmath18 satisfying @xmath19 for @xmath20 is called the _ grade involution_. for general @xmath21 , expressed as @xmath22 where @xmath20 , we have @xmath23 . also for any matrix @xmath24 over @xmath4 , @xmath25 . it is easy to see that @xmath26 . a _ superderivative _ @xmath0 is a linear mapping @xmath27 such that @xmath28 and @xmath29 and satisfying @xmath30 . one way to obtain a superderivative is as @xmath31 where @xmath32 is an even variable and @xmath33 is an odd ( grassmann ) variable . for such a superderivative @xmath34 . note that since @xmath35 and @xmath36 , it follows that @xmath37 and so grade involution and superderivatives anticommute . [ [ even - and - odd - supermatrices ] ] even and odd supermatrices + + + + + + + + + + + + + + + + + + + + + + + + + + a block matrix @xmath38 over @xmath4 where @xmath39 is @xmath40 , @xmath41 is @xmath42 , @xmath43 is @xmath44 and @xmath45 is @xmath46 for integers @xmath47 , @xmath48 , @xmath49 and @xmath50 with @xmath51 and @xmath52 is called an @xmath53 _ supermatrix_. it is said to be a _ even _ , and has parity @xmath54 , if @xmath39 and @xmath45 ( if not empty ) have even entries and @xmath41 and @xmath43 ( if non - empty ) have odd entries . one the other hand , if @xmath39 and @xmath45 have odd entries and @xmath41 , @xmath43 have even entries then @xmath55 is said to be a _ odd _ , and has parity @xmath56 . it is said to be homogeneous if it is either even or odd . [ [ supertranspose ] ] supertranspose + + + + + + + + + + + + + + the _ supertranspose _ of a homogeneous supermatrix @xmath57 , is defined to be @xmath58 where @xmath59 denotes the normal matrix transpose . in particular , an even @xmath60-row vector has the form @xmath61 where @xmath62 , and its supertranspose is @xmath63 on the other hand , an odd @xmath60-row vector has the form @xmath64 and the supertranspose @xmath65 for homogenous supermatrices @xmath66 , @xmath57 and @xmath67 , it is known that @xmath68 and @xmath69 also , supertranspose commutes with the grade involution but not with a superderivative ; for a homogeneous matrix @xmath70 , @xmath71 [ [ superdeterminants ] ] superdeterminants + + + + + + + + + + + + + + + + + consider an even @xmath72 supermatrix @xmath73 in which @xmath39 and @xmath45 are non - singular . the _ superdeterminant _ , or _ berezinian _ , of @xmath57 is defined to be @xmath74 it is also convenient to define @xmath75 an @xmath76 matrix @xmath77 over a ring @xmath78 ( noncommutative , in general ) has @xmath79 _ quasideterminants _ written as @xmath80 for @xmath81 , which are also elements of @xmath78 . they are defined recursively by @xmath82 in the above @xmath83 represents the @xmath84th row of @xmath85 with the @xmath86th element removed , @xmath87 the @xmath86th column with the @xmath84th element removed and @xmath88 the submatrix obtained by removing the @xmath84th row and the @xmath86th column from @xmath85 . quasideterminants can be also denoted as shown below by boxing the entry about which the expansion is made @xmath89 note that if the entries in @xmath85 commute then @xmath90 invariance under row and column operations + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the quasideterminants of a matrix have invariance properties similar to those of determinants under elementary row and column operations applied to the matrix . consider the following quasideterminant of an @xmath91 matrix : @xmath92 the above formula can be used to understand the effect on a quasideterminant of certain elementary row operations involving multiplication on the left . this formula excludes those operations which add left - multiples of the row containing the expansion point to other rows since there is no simple way to describe the effect of these operations . for the allowed operations however , the results can be easily described ; left - multiplying the row containing the expansion point by @xmath93 has the effect of left - multiplying the quasideterminant by @xmath93 and all other operations leave the quasideterminant unchanged . there is analogous invariance under column operations involving multiplication on the right . [ [ noncommutative - jacobi - identity ] ] noncommutative jacobi identity + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + there is a quasideterminant version of jacobi identity for determinants , called the noncommutative sylvester s theorem by gelfand and retakh @xcite . the simplest version of this identity is given by @xmath94 where @xmath95 , @xmath96 is an @xmath76 matrix and @xmath97 ( resp . @xmath98 ) are column ( resp . row ) @xmath50-vectors over @xmath78 . [ [ quasi - plcker - coordinates ] ] quasi - plcker coordinates + + + + + + + + + + + + + + + + + + + + + + + + + given an @xmath99 matrix @xmath96 , denote the @xmath84th row of @xmath96 by @xmath100 , the submatrix of @xmath96 having rows with indices in a subset @xmath101 of @xmath102 by @xmath103 and @xmath104 by @xmath105 . given @xmath106 and @xmath101 such that @xmath107 and @xmath108 , one defines the _ ( right ) quasi - plcker coordinates _ @xmath109 for any column index @xmath110 . the final equality in comes from an identity of the form and proves that the definition is independent of the choice of @xmath48 . [ [ solving - linear - systems ] ] solving linear systems + + + + + + + + + + + + + + + + + + + + + + solutions of systems of linear systems over an arbitrary ring can be expressed in terms of quasideterminants . [ 1 ] let @xmath111 be an @xmath76 matrix over a ring @xmath78 . assume that all the quasideterminants @xmath112 are defined and invertible . then the system of equations @xmath113 has the unique solution @xmath114 let @xmath115 be the @xmath76 matrix obtained by replacing the @xmath116-th row of the matrix @xmath96 with the row @xmath117 . then we have the following noncommutative version of _ cramer s rule_. [ 2 ] in notation of theorem [ 1 ] , if the quasideterminants @xmath112 and @xmath118 are well defined , then @xmath119 the basic formulae connecting quasideterminants of even supermatrices with their berezinians are given in @xcite . let @xmath57 be an @xmath72-supermatrix . then @xmath120 ( -1)^{i+j}\dfrac{{\mathop{\mathrm { ber}^*}}(\mathcal{m})}{{\mathop{\mathrm { ber}^*}}(\mathcal{m}^{i , j } ) } & m+1\le i , j\le m+n , \end{cases}\ ] ] _ _ ( _ _ cf . . _ ) _ roughly speaking , a quasideterminant with indices in one of the even blocks of @xmath70 is given as a ratio of berezinians . a quasideterminant with its indices in the one of the odd blocks is not well - defined . consider now a more general setting in which @xmath4 is an associative , unital algebra over ring @xmath5 , not necessarily graded . suppose that there is a homomorphism @xmath121 ( i.e. for all @xmath122 , @xmath123 , @xmath124 , @xmath125 and @xmath126 ) and a _ twisted derivation _ or @xmath2-derivation @xcite @xmath27 satisfying @xmath28 and @xmath127 . simple examples arise in the case that elements @xmath21 depend on a variable @xmath32 , say . derivative : : here @xmath128 satisfies @xmath129 and @xmath2 is the identity mapping . forward difference : : the homomorphism is the shift operator @xmath45 , where @xmath130 and the twisted derivation is @xmath131 satisfying @xmath132 . jackson derivative : : the homomorphism is a @xmath3-shift operator defined by @xmath133 and the twisted derivation is @xmath134 satisfying @xmath135 . superderivative : : as described in section [ sec : super ] , for @xmath123 , a superalgebra , @xmath136 where @xmath137 is the grade involution . there are a number of simple properties of such a twisted derivation , which are summarized in the following lemma . [ lem : basic properties ] 1 . let @xmath138 be matrices over @xmath4 . whenever @xmath139 is defined , @xmath140 and @xmath141 , 2 . let @xmath96 be an invertible matrix over @xmath4 . then @xmath142 and @xmath143 , 3 . let @xmath144 be matrices over @xmath4 such that @xmath145 is well - defined . then @xmath146 let @xmath147 be a sequence in @xmath4 . consider the sequence @xmath148,\theta[1],\theta[2],\ldots$ ] in @xmath4 , generated from the first sequence by darboux transformations of the form @xmath149 where @xmath0 and @xmath2 are the twisted derivation and homomorphism defined above . to be specific , @xmath148=\theta_0 $ ] and @xmath150=g_{\theta[0]}$ ] , then let @xmath151 } \theta[1]=g[0](\theta_1)=d(\theta_1)-d(\theta_0)\theta_0^{-1}\theta_1\ ] ] and @xmath152=g_{\theta[1]}$ ] , @xmath153=g[1]\circ g[0](\theta_2)$ ] and @xmath154=g_{\theta[2]}$ ] and so on . in general , for @xmath155 , @xmath156 } \theta[k]=g[k-1]\circ g[k-2]\circ\dots\circ g[0](\theta_k),\quad g[k]=\sigma(\theta[k])d\theta[k]^{-1},\ ] ] and we require that each @xmath157 $ ] is invertible . in the standard case , @xmath158 and @xmath159 , it is well known that the terms in the sequence of darboux transformations have closed form expressions in terms of the original sequence . in the case that @xmath4 is commutative , they are expressed as ratios of wronskian determinants @xcite , @xmath160=\frac { \begin{vmatrix } \theta_0&\dots&\theta_{n-1}&\theta_n\\ \theta^{(1)}_0&\dots&\theta^{(1)}_{n-1}&\theta^{(1)}_n\\ \vdots&&\vdots\\ \theta^{(n-1)}_0&\dots&\theta^{(n-1)}_{n-1}&\theta^{(n-1)}_{n}\\ \theta^{(n)}_0&\dots&\theta^{(n)}_{n-1}&\theta^{(n)}_n \end{vmatrix } } { \begin{vmatrix } \theta_0&\dots&\theta_{n-1}\\ \theta^{(1)}_0&\dots&\theta^{(1)}_{n-1}\\ \vdots&&\vdots\\ \theta^{(n-1)}_0&\dots&\theta^{(n-1)}_{n-1 } \end{vmatrix } } , \quad n\in\mathbb{n},\ ] ] where @xmath161 denotes @xmath162 . in the case that @xmath4 is not commutative , the terms in the sequence are expressed as quasideterminants @xcite , @xmath160= \begin{vmatrix } \theta_0&\dots&\theta_{n-1}&\theta_n\\ \theta^{(1)}_0&\dots&\theta^{(1)}_{n-1}&\theta^{(1)}_n\\ \vdots&&\vdots\\ \theta^{(n-1)}_0&\dots&\theta^{(n-1)}_{n-1}&\theta^{(n-1)}_{n}\\ \theta^{(n)}_0&\dots&\theta^{(n)}_{n-1}&\fbox{$\theta^{(n)}_n$ } \end{vmatrix},\quad n\in\mathbb{n}.\ ] ] the following theorem gives a generalisation of this formula to the case of general @xmath0 and @xmath2 . note in particular that the expressions do not depend on @xmath2 and are obtained simply by replacing @xmath163 with @xmath0 . [ thm : dt ] let @xmath164=\phi$ ] and for @xmath165 let @xmath166=d(\phi[n-1])-d(\theta[n-1])\theta[n-1]^{-1}\phi[n-1],\ ] ] where @xmath167=\phi[n]|_{\phi\to\theta_n}$ ] . then , for @xmath165 , @xmath168 } \phi[n]= \begin{vmatrix } \theta_0&\cdots&\theta_{n-1}&\phi\\ d(\theta_0)&\cdots&d(\theta_{n-1})&d(\phi)\\ \vdots&&\vdots&\vdots\\ d^{n-1}(\theta_0)&\cdots&d^{n-1}(\theta_{n-1})&d^{n-1}(\phi)\\ d^{n}(\theta_0)&\cdots&d^{n}(\theta_{n-1})&\fbox{$d^{n}(\phi)$ } \end{vmatrix}.\ ] ] ( by induction . ) the case @xmath169 follows from the definition and . now assume that holds for @xmath170 . then @xmath171=d^k(\phi)- \begin{bmatrix } d^k(\theta_0)&\cdots & d^k(\theta_{k-1 } ) \end{bmatrix } \theta^{-1 } \begin{bmatrix } \phi\\d(\phi)\\\vdots\\d^{k-1}(\phi ) \end{bmatrix},\ ] ] where @xmath172 to complete the proof we must show that @xmath173=d(\phi[k])-d(\theta[k])\theta[k]^{-1}\phi[k]$ ] can be written in the form for @xmath174 . using lemma [ lem : basic properties ] , one obtains @xmath175)= \begin{vmatrix } \theta_0&\cdots&\theta_{k-1}&\phi\\ d(\theta_0)&\cdots&d(\theta_{k-1})&d(\phi)\\ \vdots&&\vdots&\vdots\\ d^{k-2}(\theta_0)&\cdots&d^{k-2}(\theta_{k-1})&d^{k-2}(\phi)\\ d^{k-1}(\theta_0)&\cdots&d^{k-1}(\theta_{k-1})&d^{k-1}(\phi)\\ d^{k+1}(\theta_0)&\cdots&d^{k+1}(\theta_{k-1})&\fbox{$d^{k+1}(\phi)$ } \end{vmatrix}+ \sigma\left(\begin{vmatrix } \theta_0&\cdots&\theta_{k-1}&0\\ d(\theta_0)&\cdots&d(\theta_{k-1})&0\\ \vdots&&\vdots&\vdots\\ d^{k-2}(\theta_0)&\cdots&d^{k-2}(\theta_{k-1})&0\\ d^{k-1}(\theta_0)&\cdots&d^{k-1}(\theta_{k-1})&1\\ d^{k}(\theta_0)&\cdots&d^{k}(\theta_{k-1})&\fbox{$0 $ } \end{vmatrix}\right ) \phi[k],\ ] ] and so , @xmath176)= \begin{vmatrix } \theta_0&\cdots&\theta_{k-1}&\theta_{k}\\ d(\theta_0)&\cdots&d(\theta_{k-1})&d(\theta_{k})\\ \vdots&&\vdots&\vdots\\ d^{k-2}(\theta_0)&\cdots&d^{k-2}(\theta_{k-1})&d^{k-2}(\theta_{k})\\ d^{k-1}(\theta_0)&\cdots&d^{k-1}(\theta_{k-1})&d^{k-1}(\theta_{k})\\ d^{k+1}(\theta_0)&\cdots&d^{k+1}(\theta_{k-1})&\fbox{$d^{k+1}(\theta_{k})$ } \end{vmatrix}+ \sigma\left(\begin{vmatrix } \theta_0&\cdots&\theta_{k-1}&0\\ d(\theta_0)&\cdots&d(\theta_{k-1})&0\\ \vdots&&\vdots&\vdots\\ d^{k-2}(\theta_0)&\cdots&d^{k-2}(\theta_{k-1})&0\\ d^{k-1}(\theta_0)&\cdots&d^{k-1}(\theta_{k-1})&1\\ d^{k}(\theta_0)&\cdots&d^{k}(\theta_{k-1})&\fbox{$0 $ } \end{vmatrix}\right ) \theta[k].\ ] ] then , using , @xmath177)-d(\theta[k])\theta[k]^{-1}\phi[k]$ ] can be expressed in the required form . as an application of this theorem , in the following section we will use it to construct solutions of the super kdv equation . the manin - radul supersymmetric kdv ( mrskdv ) system @xcite is @xmath178 where @xmath179 and @xmath180 are even and odd dependent variables respectively , @xmath181 are even independent variables variables and @xmath0 is the superderivative defined by @xmath31 , where @xmath33 is a grassmann odd variable , satisfying @xmath34 . this system has the lax pair @xmath182 in the sense that @xmath183=0 $ ] implies . eigenfunctions satisfy @xmath184 for eigenvalue @xmath185 . a darboux transformation for this system @xcite is @xmath186 where @xmath33 is an invertible , and hence necessarily even , solution of . note that it is an example of the general type of darboux transformation discussed in section [ sec : dt ] . as discussed there , this transformation may be iterated by taking solutions @xmath187 of to obtain @xmath188&=d(\phi[k])-d(\theta[k])\theta[k]^{-1}\phi[k],\\ \theta[k]&=\phi[k]|_{\phi\to\theta_k}.\end{aligned}\ ] ] the requirement that each @xmath157 $ ] is invertible means that it must be even and consequently that @xmath189 must have parity @xmath84 . the corresponding solutions of mrskdv are @xmath190=\alpha$ ] , @xmath191=u$ ] and @xmath192&=-\alpha[k]+2(d(\theta[k])\theta[k]^{-1})_x,\\ \label{u rec } u[k+1]&=u[k]+d(\alpha[k])-2d(\theta[k])\theta[k]^{-1}(\alpha[k]-(d(\theta[k])\theta[k]^{-1})_x).\end{aligned}\ ] ] theorem [ thm : dt ] give a closed - form expression for @xmath193 $ ] as a quasideterminant . we will obtain the corresponding expressions for @xmath194 $ ] and @xmath195 $ ] in terms quasideterminants of the form @xmath196 \label{qn } & = -\begin{vmatrix } \theta_0&\cdots&\theta_{n-1}\\ \vdots&\ddots&\vdots\\ d^{n - j-2}(\theta_0)&\cdots&d^{n - j-2}(\theta_{n-1})\\ d^{n - j}(\theta_0)&\cdots&d^{n - j}(\theta_{n-1})\\ \vdots&\ddots&\vdots\\ d^{n-1}(\theta_0)&\cdots&d^{n-1}(\theta_{n-1})\\ d^{n+i}(\theta_0)&\cdots&d^{n+i}(\theta_{n-1})\\ \end{vmatrix}_{n , s } \begin{vmatrix } \theta_0&\cdots&\theta_{n-1}\\ \vdots&\ddots&\vdots\\ d^{n-1}(\theta_0)&\cdots&d^{n-1}(\theta_{n-1 } ) \end{vmatrix}_{n - j , s}^{-1},\end{aligned}\ ] ] for any @xmath197 ( see ) . the following lemma records some useful properties of the @xmath198 . [ lem : qn ] _ ( _ i _ ) _ @xmath199 . i.e. , @xmath198 has the parity @xmath200 , + _ ( _ ii _ ) _ @xmath201 and @xmath202)\theta[k]^{-1}=-q_k(0,0)-q_{k+1}(0,0)$ ] for @xmath203 , + _ ( _ iii _ ) _ @xmath204 . \(i ) this follows from the facts that @xmath189 has parity @xmath84 and @xmath0 is odd , and from the invariance properties of quasideterminants . + ( ii ) it is obvious from the definition that @xmath205 . the second result follows by using the same method used in the proof of theorem [ thm : dt ] with @xmath206 . + ( iii ) expanding @xmath207 , written in the form , using gives the terms on the right hand side . [ thm : alpha u ] after @xmath50 repeated darboux transformations , the mrskdv system has new solutions @xmath194 $ ] and @xmath195 $ ] expressed in terms of @xmath208 and @xmath209 . @xmath210&=(-1)^n\alpha-2q_n(0,0)_x,\label{alpha s}\\ u[n]&=u-2q_n(0,1)_x-2q_n(0,0)((-1)^n\alpha - q_n(0,0)_x)+{1-(-1)^n\over 2}d(\alpha).\label{u s}\end{aligned}\ ] ] ( by induction . ) first consider . for @xmath169 we must show that @xmath211=-\alpha+2(d(\theta_0)\theta_0^{-1})_x.\ ] ] this is clear from for @xmath212 and lemma [ lem : qn](ii ) . by using in general case and lemma [ lem : qn](ii ) , assuming that holds for @xmath170 , we have @xmath213&=-\alpha[k]+2(d(\theta[k])\theta[k]^{-1})_x\\ & = ( -1)^{k+1}\alpha-2q_{k+1}(0,0)_x,\end{aligned}\ ] ] as required . the proof of is very similar and makes use of lemma [ lem : qn](iii ) . the quasideterminant solutions and could also be obtained by solving a noncommutative linear system . in @xcite , the solutions @xmath194 $ ] and @xmath195 $ ] for the mrskdv system were given as @xmath210&=(-1)^n\alpha-2\partial a_{n , n-1},\\ u[n]&=u-2\partial a_{n , n-2}-2a_{n , n-1}((-1)^n\alpha-\partial a_{n , n-1})+{1-(-1)^n\over 2}d\alpha,\end{aligned}\ ] ] where @xmath214 satisfy the linear system @xmath215 by using theorem [ 2 ] and , we can solve the above linear system to obtain @xmath216 it is obvious that solutions @xmath194 $ ] and @xmath195 $ ] obtained here coincide with those given by and . it is natural for a supersymmetric system to express the solutions in terms of superdeterminants and this was done , in all but one case in @xcite . in this section , we will show systematically that the solutions given in theorem [ thm : alpha u ] , expressed in terms of quasideterminants @xmath208 and @xmath209 , can be reexpressed in terms of superdeterminants . the expressions we will obtain coincide with the superdeterminant solutions found in @xcite and we also find the superdeterminant expressions in the case that they did not . recall the property that @xmath189 has parity @xmath84 . let us therefore introduce the relabeling @xmath217 also , we write @xmath218 and @xmath219 , where @xmath220 denotes the @xmath86th derivative with respect to @xmath32 . consider the matrix @xmath221 appearing in the definition of @xmath198 . there is a natural reordering of the rows and columns @xmath222 which gives an even matrix @xmath223 . this reordering does not change the value of any associated quasideterminant , as long as the expansion point in each refers to the same element . in the case that @xmath50 is even , @xmath224 and @xmath225 and @xmath226 are all @xmath227 matrices . in the case that @xmath50 is odd , @xmath228 is @xmath229 , @xmath230 is @xmath231 , @xmath232 is @xmath233 and @xmath234 is a @xmath227 matrix whose precise form can be easily deduced from the above description . similarly , consider the matrix @xmath235 appearing in the definition of @xmath209 . a similar reordering of this matrix @xmath236 gives another even matrix @xmath237 where , for example , @xmath238 [ thm : summary ] for @xmath165 , @xmath239 where @xmath240 if @xmath50 is even , and @xmath241 if @xmath50 is odd . using a straightforward generalization of the method of differentiating quasideterminants given in @xcite , one can show that @xmath242 note that @xmath243 and so it is even . if @xmath244 , then by @xmath245 similarly , if @xmath246 , then @xmath247 by repeated use of , noting that @xmath248 , it follows that when @xmath50 is even @xmath249 and when @xmath50 is odd @xmath250 as required . also , from , we have @xmath251 the two quasideterminant in these expressions can be expressed , by , as a ratio of berezinians in which a common factor cancels . the result is that @xmath252 \dfrac{{\mathop{\mathrm { ber}^*}}(\mathcal w'_{2k+1})}{{\mathop{\mathrm { ber}^*}}(\mathcal w_{2k+1})}&n=2k+1 \end{cases},\ ] ] as required . binary darboux transformations for the mrskdv system were discussed in @xcite . in these articles , solutions expressed in terms of determinants were obtained . as discussed in connection with darboux transformations it is to be expected that solutions for this supersymmetric system should be _ superdeterminants _ in general . in this section , we will construct a more general type of binary darboux transformation which will be shown to give these superdeterminants solutions and includes the solutions found in @xcite as a special case . first we recall the definition of the adjoint for supersymmetric linear operators . for a linear operator @xmath253 , @xmath254 denotes its parity . for example , @xmath255 and @xmath256 , where @xmath163 denotes any derivative with respect to an even variable , and the parity of multiplication by a homogeneous element is the parity of that element ( in the usual sense ) . the rules defining the superadjoint are @xmath257 where @xmath70 denotes any matrix over @xmath4 , together with the product rule @xmath258 where @xmath253 and @xmath259 are operators ( cf . for the case of matrices ) . in particular , this gives @xmath260 and , consistently , @xmath261 . for any @xmath21 , @xmath262 . the lax pair , has the adjoint form @xmath263 and adjoint eigenfunctions satisfy @xmath264 for eigenvalue @xmath265 . given an ( eigenfunction , adjoint ) eigenfunction pair @xmath266 , the binary darboux transformation @xcite is given by @xmath267 where eigenfunction @xmath33 and adjoint eigenfunction @xmath268 have opposite parities . since @xmath269 , @xmath270 is even and assumed to be invertible . when iterating this transformation , both previous papers @xcite on this topic considered the case that all eigenfunctions are even and all adjoint eigenfunctions are odd . we will show that this is not the most general possibility however . consider an even @xmath60-row vector eigenfunction @xmath271 and an odd @xmath60-row vector adjoint eigenfunction @xmath272 , where @xmath273 for @xmath274 and @xmath275 for @xmath276 are even and @xmath277 for @xmath274 and @xmath278 for @xmath276 are odd . these row vectors satisfy @xmath279 where @xmath280 and @xmath281 are constant @xmath282 diagonal matrices containing the eigenvalues . then @xmath283 is an even @xmath72-supermatrix defined up to a constant by @xmath284 the closed form expressions for the results of iterated binary darboux transformations are stated in the following theorems . [ thm : phi psi ] iterating the binary darboux transformations , for @xmath285 , one obtains @xmath286=\begin{vmatrix } \omega(\mathcal e,\mathcal o)&\omega(\phi,\mathcal o)\\ \mathcal e&\fbox{$\phi$ } \end{vmatrix},\quad \psi[m+n]=\begin{vmatrix } \omega(\mathcal e,\mathcal o)^\dagger&\omega(\mathcal e,\psi)^\dagger\\ \mathcal o&\fbox{$\psi$ } \end{vmatrix } = \begin{vmatrix } { \widehat}{\omega(\mathcal e,\mathcal o)}&\mathcal o^\dagger\\ { \widehat}{\omega(\mathcal e,\psi)}&\fbox{$\psi$ } \end{vmatrix}\label{phi psi}\ ] ] with @xmath287,\psi[m+n])= \begin{vmatrix } \omega(\mathcal e,\mathcal o)&\omega(\phi,\mathcal o)\\ \omega(\mathcal e,\psi)&\fbox{$\omega(\phi,\psi)$ } \end{vmatrix}.\ ] ] ( by induction . ) let @xmath288 . the formulae and evidently hold for @xmath289 . now suppose that they hold for given @xmath290 and consider another binary darboux transformation defined in terms of @xmath291=\begin{vmatrix } \omega(\mathcal e,\mathcal o)&\omega(\theta,\mathcal o)\\ \mathcal e&\fbox{$\theta$ } \end{vmatrix},\quad \rho[k]=\begin{vmatrix } { \widehat}{\omega(\mathcal e,\mathcal o)}&\mathcal o^\dagger\\ { \widehat}{\omega(\mathcal e,\rho)}&\fbox{$\rho$ } \end{vmatrix}\label{theta rho}.\ ] ] here @xmath266 be another eigenfunction , adjoint eigenfunction pair of opposite parity , so that @xmath292 and @xmath293 are even and odd row vectors respectively . then , using , @xmath294&=\phi[k]-\theta[k]\omega(\theta[k],\rho[k])^{-1}\omega(\phi[k],\rho[k])\\ & = \begin{vmatrix } \omega(\mathcal e,\mathcal o)&\omega(\theta,\mathcal o)&\omega(\phi,\mathcal o)\\ \omega(\mathcal e,\rho)&\omega(\theta,\rho)&\omega(\phi,\rho)\\ \mathcal e&\theta&\fbox{$\rho$ } \end{vmatrix},\end{aligned}\ ] ] verifying the first equation in for @xmath295 . the second equation is proved in a similar way . the alternative expression for @xmath296 $ ] follows from . finally , it is straightforward to show that @xmath0 applied to the rhs of gives @xmath297\phi[k],\end{aligned}\ ] ] as required . [ thm : alpha u ] let @xmath298 be a solution of mrskdv and let @xmath299 and @xmath300 respectively be even and odd @xmath60-row vectors satisfying and . then for any integers @xmath301 @xmath302=\alpha-2a[m+n]_x,\quad u[m+n]=u+2(\alpha - a[m+n]_x)a[m+n]-2u[m+n]_x,\ ] ] where @xmath303= \begin{vmatrix } \omega(\mathcal e,\mathcal o)&{\widehat}{\mathcal o}^\dagger\\ \mathcal e&\fbox{$0 $ } \end{vmatrix},\quad u[m+n]= \begin{vmatrix } \omega(\mathcal e,\mathcal o)&d(\mathcal o^\dagger)\\ \mathcal e&\fbox{$0 $ } \end{vmatrix},\ ] ] are also solutions of mrskdv . let @xmath288 . the formulae are true by definition for @xmath212 ; @xmath190=\alpha$ ] and @xmath191=u$ ] . the formulae will be proved by finding certain expression involving @xmath304 $ ] and @xmath305 $ ] that are independent of @xmath290 . as in the proof of theorem [ thm : phi psi ] , suppose that a binary darboux transformation is determined by @xmath291=\begin{vmatrix } \omega(\mathcal e,\mathcal o)&\omega(\theta,\mathcal o)\\ \mathcal e&\fbox{$\theta$ } \end{vmatrix},\quad \rho[k]=\begin{vmatrix } { \widehat}{\omega(\mathcal e,\mathcal o)}&\mathcal o^\dagger\\ { \widehat}{\omega(\mathcal e,\rho)}&\fbox{$\rho$ } \end{vmatrix},\ ] ] where @xmath266 are an eigenfunction , adjoint eigenfunction pair of opposite parity , so that @xmath292 and @xmath293 are even and odd row vectors respectively . then , by and , @xmath306 } \alpha[k+1]&=\alpha[k]+2(\theta[k]\omega(\theta[k],\rho[k])^{-1}{\widehat}{\rho[k]})_x,\\ \label{u[k+1 ] } u[k+1]&=u[k]-2(\alpha[k]+(\theta[k]\omega(\theta[k],\rho[k])^{-1}{\widehat}{\rho[k]})_x)\theta[k]\omega(\theta[k],\rho[k])^{-1}{\widehat}{\rho[k]}\nonumber\\&\qquad+2(\theta[k]\omega(\theta[k],\rho[k])^{-1}d(\rho[k]))_x.\end{aligned}\ ] ] using the notation introduced in , we have @xmath307= \begin{vmatrix } \omega(\mathcal e,\mathcal o)&\omega(\theta,\mathcal o)&{\widehat}{\mathcal o}^\dagger\\ \omega(\mathcal e,\rho)&\omega(\theta,\rho)&{\widehat}{\rho}\\ \mathcal e&\theta&\fbox{$0 $ } \end{vmatrix},\quad u[k+1]= \begin{vmatrix } \omega(\mathcal e,\mathcal o)&\omega(\theta,\mathcal o)&d(\mathcal o^\dagger)\\ \omega(\mathcal e,\rho)&\omega(\theta,\rho)&d(\rho)\\ \mathcal e&\theta&\fbox{$0 $ } \end{vmatrix}.\ ] ] then using , and noting that @xmath308 $ ] and @xmath309 $ ] are odd , it is straightforward to show that @xmath310 } \theta[k]\omega(\theta[k],\rho[k])^{-1}{\widehat}{\rho[k]}=a[k]-a[k+1],\ ] ] and @xmath311 } \theta[k]\omega(\theta[k],\rho[k])^{-1}d(\rho[k])=u[k]-u[k+1]-a[k+1]a[k].\ ] ] substituting these in and one obtains @xmath290 independent expressions @xmath312 + 2a[k+1]_x=\alpha[k]+2a[k]_x=\alpha,\ ] ] and @xmath313 - 2(\alpha - a[k+1]_x)a[k+1]+2u[k+1]_x = u[k]-2(\alpha -a[k]_x)a[k]+2u[k]_x = u,\ ] ] and hence hold for all @xmath290 . in this section we will show how the quasideterminant solutions @xmath314 , u[m+n])$ ] obtained using binary darboux transformations can be expressed in terms of superdeterminants . to do this , it is necessary to introduce a more detailed notation for row vector eigenfunctions and adjoint eigenfunctions . recall that for the general transformation we use @xmath60-row vectors @xmath299 and @xmath300 which are even and odd with entries @xmath189 and @xmath315 respectively . here we will also write @xmath316 and @xmath317 for the row vectors containing the first @xmath84 entries of @xmath299 and @xmath300 respectively , and denote by subscript @xmath54 and @xmath56 the even and odd element parts of @xmath299 and @xmath300 respectively . thus @xmath318 and @xmath319 . the expressions @xmath314 , u[m+n])$ ] can expressed as @xmath320=d\left(\log { \mathop{\mathrm { ber}}}(\mathcal g_{(m|n)})\right),\quad u[m+n]=\frac{{\mathop{\mathrm { ber}}}(\mathcal g'_{(m+1|n)})}{{\mathop{\mathrm { ber}}}(\mathcal g_{(m|n)})},\ ] ] where @xmath321 in an even @xmath72-supermatrix and @xmath322 in an even @xmath323-supermatrix . the formula for @xmath324 $ ] in follows directly from and . for any @xmath325 , @xmath326= \begin{vmatrix } \omega(\mathcal e^i,\mathcal o^i)&{\widehat}{\mathcal o}^i{}^\dagger\\ \mathcal e^i&\fbox{$0 $ } \end{vmatrix},\ ] ] and @xmath327= \begin{vmatrix } \omega(\mathcal e^i,\mathcal o^i)&\omega(\theta_i,\mathcal o^i)\\ \mathcal e^i&\fbox{$\theta_i$ } \end{vmatrix},\quad \rho[i]=\begin{vmatrix } { \widehat}{\omega(\mathcal e^i,\mathcal o^i)}&\mathcal o^i{}^\dagger\\ { \widehat}{\omega(\mathcal e^i,\rho_i)}&\fbox{$\rho_i$ } \end{vmatrix},\ ] ] where @xmath328=u[0]=0 $ ] and @xmath148=\theta_0 $ ] , @xmath329=\rho_0 $ ] . also define @xmath330=\omega(\theta[i],\rho[i])= \begin{vmatrix } \omega(\mathcal e^{i},\mathcal o^{i})&\omega(\theta_i,\mathcal o^i)\\ \omega(\mathcal e^i,\rho_i)&\fbox{$\omega(\theta_i,\rho_i)$ } \end{vmatrix}.\ ] ] using , @xmath330= \begin{cases } \dfrac{{\mathop{\mathrm { ber}}}(\omega(\mathcal e^{i+1},\mathcal o^{i+1}))}{{\mathop{\mathrm { ber}}}(\omega(\mathcal e^{i},\mathcal o^{i}))}&0\le i\le m-1 \\[6pt ] \dfrac{{\mathop{\mathrm { ber}^*}}(\omega(\mathcal e^{i+1},\mathcal o^{i+1}))}{{\mathop{\mathrm { ber}^*}}(\omega(\mathcal e^{i},\mathcal o^{i}))}&m\le i\le m+n-1 . \end{cases},\ ] ] where @xmath331 . if fact , for @xmath332 then @xmath333 , since all pairs @xmath334 have opposite parity , @xmath335 $ ] and @xmath336 $ ] also have opposite parity and so commute with each other . hence may be written for arbitrary @xmath337 as @xmath338=a[i]-\frac{{\widehat}{\rho[i]}\theta[i]}{\omega[i]}.\ ] ] now , the parity of @xmath336 $ ] is the same as that of @xmath315 and hence @xmath339}= \begin{cases } -\rho[i]&0\le i\le m-1\\ \rho[i]&m\le i\le m+n-1 \end{cases}.\ ] ] finally , this gives the recurrence relation @xmath338=a[i]+\begin{cases } d(\log\omega[i])&0\le i\le m-1\\ -d(\log\omega[i])&m\le i\le m+n-1 \end{cases},\quad a[0]=0,\ ] ] and so @xmath340=d\left(\log\frac{\omega[0]\omega[1]\cdots\omega[m-1]}{\omega[m]\omega[m+1]\cdots\omega[m+n-1]}\right)=d\bigl(\log{\mathop{\mathrm { ber}}}(\omega(\mathcal e,\mathcal o))\bigr),\ ] ] as required . the earlier papers on this topic @xcite take @xmath341 only . in this case , @xmath342 and @xmath343 and we obtain solutions expressed in terms of determinants not superdeterminants @xmath344=d(\log\det\omega(\mathcal e_0,\mathcal o_1)),\ ] ] and @xmath345=\frac{\det\begin{pmatrix } \omega(\mathcal e_0,\mathcal o_1)&d(\mathcal o_1^t)\\ \mathcal e_0&0 \end{pmatrix}}{\det(\omega(\mathcal e_0,\mathcal o_1))}.\ ] ] in this paper , we have considered a twisted derivation which includes normal derivative , forward difference operator , q - difference operator and superderivatives as special cases . we showed that a darboux transformation defined in terms of such a twisted derivation has an iteration formula written in terms of a quasideterminant . this results opens the opportunity for an unified approach to darboux transformations for differential , superdifferential difference and @xmath3-difference operators . in this paper we showed how this was achieved for one example , the manin - radul super kdv equation . in the darboux transformation approach to this equation given in @xcite , the authors were forced to consider two different cases , for an odd and an even number of iterations , in order to obtain solutions expressed in terms of superdeterminants . using the same darboux transformation , we have obtained quasideterminant solutions in a unified manner , irrespective of the parity of the number of iterations , and then we were able in all cases to reexpress the solutions in terms of superdeterminants using the known relationship between quasideterminants and superdeterminants . this illustrates the advantage of using quasideterminants over using superdeterminants from the start and comes about because quasideterminants need no assumption about the nature of the noncommutativity whereas for superdeterminants restrictive assumptions about parity of matrix elements are required . the same advantage holds in connection with iterated binary darboux transformations . in this case we were able to construct a much more general types of solution as well . in earlier work @xcite only solutions expressed in terms of determinants were found . however , it is to be expected that for a supersymmetric integrable system the most general solutions have expressions in terms of superdeterminants rather than determinants . this deficiency was remedied in this paper and we have obtained a much wider class of solutions expressed in terms of both quasideterminants and superdeterminants . this work was carried out during the authors visit to the newton institute for mathematical sciences as part of the programme on discrete integrable systems ( january - july 2009 ) . the authors are grateful to the organizers for their invitation and to the isaac newton institute for financial support and hospitality . this work was supported by the national natural science foundation of china ( grant no . 10601028 ) and the project - sponsored by srf for rocs , sem . kupershmidt , b. a. kp or mkp , vol . 78 of mathematical surveys and monographs . american mathematical society , providence , ri ( 2000 ) . noncommutative mathematics of lagrangian , hamiltonian , and integrable systems . paniak , l. d. exact noncommutative kp and kdv multi - solitons ( 2001 ) . hep - th/0105185 . sakakibara , m. _ j. phys . * 37*(2004 ) l599l604 . wang , n. and wadati , m. _ j. phys . soc . japan _ * 72*(2003 ) 13661373 . wang , n. and wadati , m. _ j. phys . soc . japan _ * 72*(2003 ) 18811888 . wang , n. and wadati , m. j. phys . japan * 73*(2004 ) 16891698 . hamanaka , m. noncommutative solitons and d - branes . ph.d . thesis ( 2003 ) . hep - th/0303256 . hamanaka , m. and toda , k. _ phys . lett . a _ * 316*(2003 ) 7783 . dimakis , a. and mller - hoissen , f. _ j. phys . a _ * 38*(2005 ) 54535505 . nimmo , j. j. c. _ j. phys . * 39*(2006 ) 50535065 . li , c. x. , nimmo , j. j. c. and tamizhmani , k. m. _ proc . r. soc . a _ * 465*(2009 ) 14411451 . gilson c.r . and nimmo j.j.c . _ j. phys . a : math . theor . _ * 40*(2007 ) 38393850 . gilson , c. r. , nimmo , j. j. c. and ohta , y. _ j. phys . _ * 40*(2007 ) 1260712617 . li , c. x. and nimmo , j. j. c. , _ proc . * 464*(2008 ) 951966 . gelfand i.m . and retakh v.s . , determinants of matrices over noncommutative rings . appl . _ * 25*(1991 ) 1325 . manin y.i . and radul a.o . , _ comm . math . phys . _ * 98*(1985 ) 6577 . mathieu p. , _ phys . _ * 29*(1988 ) 24992506 . kupershmidt b. a. _ phys . 102a*(1984 ) 213215 . chaichian , m. and kulish , p. p. , _ phys . b _ * 77*(1978 ) 413416 . siddiq m. , hassan m. and saleem u. , _ j. phys . a : math . gen . _ * 39*(2006 ) 73137318 . roelofs g.h.m . and kersten p.h.m . , _ j. math . phys . _ * 33*(1992 ) 21852206 . morosi c. and pizzocchero l. , _ commun . phys . _ * 176*(1996 ) 353381 . liu q.p . , _ j. phys . a : math . gen . _ * 28*(1995 ) l245l248 . oevel w. , popowicz z. , _ commun . phys . _ * 139*(1991 ) 441460 ; figueroa - ofarril j. , mas j. , ramos e. , _ rev . phys . _ * 3*(1991 ) 479501 . carstea a.s . , ramani a. , grammaticos b. , _ nonlinearity _ * 14*(2001 ) 14191423 . mcarthur i.n . , yung c.m . , * 8*(1993 ) 17391745 . liu q.p . and manas m. , _ phys . lett . b _ * 396*(1997 ) 133140 . liu q.p . , manas m. , _ phys . b _ * 394*(1997 ) 337342 . hartwig j. t. , larsson d. and silvestrov s. d. , _ j. algebra _ * 295*(2006 ) 314361 . de concini s. and procesi c. , _ quantum groups _ , lecture notes in mathematics * 1565*(1993 ) 31140 . berezin f.a . , introduction to superanalysis ( d. reidel publishing company , dordrecht , 1987 ) . dewitt b. , supermanifolds ( cambridge university press , 1984 ) . gelfand i. , gelfand s. , retakh v. and wilson r.l . , _ adv . in math . _ * 193*(2005 ) 56141 . etingof p. , gelfand i. and retakh v. , _ math . * 4*(1997 ) 413425 . bergvelt m.j . and rabin j.m . , super curves , their jacobians and super kp equations . arxiv : alg - geom/9601012v1 . crum m.m . , _ q , j. math . _ * 6*(1955 ) 121127 . and tu m.h . , _ j. math . phys . _ * 39*(1998 ) 47734784 . liu q.p . and maas m. , _ phys . lett . a _ * 394*(1997 ) 337342 .

a charge - density - like covariance is expected to characterize the transition radiation energy spectrum of a electron bunch as far as the charge form factor is intended to account for bunch - density effects in the radiation emission . the beam charge passing from a single electron to a high density electron bunch , the covariance of the transition radiation energy is expected to evolve from a charge - point - like to a charge - density - like one . besides covariance , the radiation energy spectrum is expected to conform to the temporal causality principle : the single electron amplitudes composing the radiation field are expected to propagate from the metallic screen with relative emission phases causally correlated with the temporal sequence of the particle collisions onto the metallic screen . in the present paper , the case of a electron bunch hitting at a normal angle of incidence a flat metallic surface with arbitrary size and shape will be considered . for such an experimental situation , the distribution function of the electron longitudinal coordinates rules the temporal causality constraint into the transition radiation energy spectrum . the covariance feature of the transition radiation energy spectrum deals instead with the lorentz invariance of the projection of the electron spatial density in the transverse plane with respect to the direction of motion of the electron beam . because of the invariance of the electron transverse density under a lorentz transformation with respect to the direction of motion of the electron beam , the single electron radiation amplitudes composing the radiation field show a covariant dependence on the distribution function of the electron transverse coordinates , the relative emission phases of the single electron radiation amplitudes being indeed only a function of the electron longitudinal coordinates because of the temporal causality constraint . as a consequence of the temporal causality and the covariance , both the temporal coherent and incoherent components of the radiation energy spectrum bear the covariant imprinting of the distribution function of the electron transverse coordinates as in the following argued . virtual quanta , coherence , fourier transform , collective effects 41.60.-m , 41.75.-i , 42.25.kb , 42.30.kq

1207.0644

a relativistic charge in a rectilinear and uniform motion can originate an electromagnetic radiative mechanism crossing the dielectric interface between two different media . the fast dipolar oscillation of the polarization charge , induced on the dielectric interface by the relativistic charge , generates indeed an instantaneous , broad spectral band , radially polarized and highly directional radiation emission , the so called transition radiation @xcite . a photon pulse propagates backward and forward from the dielectric interface according to a double conical spatial distribution whose angular aperture scales down with the lorentz @xmath1 factor of the charge . the higher and steeper the discontinuity of the dielectric constant across the interface , the more intense the radiation emission . thanks to the relativistic origin , transition radiation can be fruitfully exploited for energy detection or mass identification of high energy particles @xcite . transition radiation based diagnostics is commonly used in a particle accelerator to monitor the transverse @xcite or the longitudinal profile of a charged beam @xcite . transition radiators , typically made of a thin metallic foil or a thin polished aluminium coating on dielectric substrates , respond as an ideal conductor over a wide spectral range , from the long wavelength to far beyond the visible optical region @xcite . this experimental situation is supposed in the present work . in particular , the case of a @xmath0 electron bunch colliding at a normal angle of incidence onto a flat ideal conductor surface will be considered in the following . an ideal conductor surface being assimilable to a double layer of charge , transition radiation emission can be thus schematized as the result of the interaction of an incident charge with the conduction electrons of the ideal conductor surface . as the relativistic charge approaches the metallic surface , the charge induced conduction electrons freely move in the transverse plane to maintain the metallic surface equipotential . the sudden and fast dipolar oscillation of the double layer of charge is thus responsible for the emission of the radiation pulse as confirmed by the emission of backward radiation . in the relativistic limit , the electromagnetic field travelling with a relativistic electron can be assimilated to a transverse electromagnetic wave , the so called _ virtual quanta _ field @xcite . in the case of a single relativistic electron , the harmonic component of the electric field at a given wavelength @xmath2 extends indeed in the transverse plane over an efficacious range in the order of @xmath3 , the ratio of the longitudinal to the transverse component being @xmath4 . the _ virtual quanta _ field showing the nature of a quasi - plane wave front , the radiation emission can be formally described as the result of the wave propagation of the _ virtual quanta _ scattered by the metallic surface according to the huygens - fresnel principle @xcite . under the far field approximation @xcite , the harmonic components of the transition radiation field at the observation point can be calculated as the fourier transform of the _ virtual quanta _ field with respect to the spatial coordinates of the radiator surface . the transition radiation energy spectrum can be finally obtained as the flux of the poynting vector . a single and individually radiating electron and a high density electron beam are in general different from the point of view of the features of the spectral and the angular distribution of the emitted radiation . compared to the case of a single and individually radiating electron , bunch - density effects may strongly affect the electromagnetic radiative mechanism by an electron beam . for a fixed energy of the beam , the number and the angular distribution of the photons radiated at a given wavelength may indeed change as a function of the density of the electron beam . in the formula of the radiation energy spectrum , the charge form factor accounts for bunch collective effects . this is , in principle , defined as the square module of the fourier transform of the distribution function of the particle density of the charged beam . aim of the present paper is to verify how well such a theoretical definition of the charge form factor fits into a covariant formulation of the radiation energy spectrum or if , instead , under the covariance and the temporal causality constraints , the radiation energy spectrum shows a dependence on the charge density that goes beyond the above mentioned formal definition of the charge form factor . in fact , the transverse density of the electron beam is a relativistic invariant under a lorentz transformation with respect to the direction of motion of the beam . the invariance of the projection of the electron beam density onto the transverse plane with respect to the direction of motion is expected to leave , on the _ virtual quanta _ field and , consequently , on the radiation field , a covariant mark whose observability , for a given harmonic component of the radiation field , only depends on the transverse dimension of the electron beam compared to the transverse component of the wave - vector which is itself a lorentz invariant with respect to the direction of motion of the beam . in the following , the charge - density - like covariance of the transition radiation energy spectrum by an electron beam will be investigated . first , the formal steps leading from the electromagnetic field of a @xmath0 electron bunch to the transition radiation energy spectrum will be analyzed with regard to the temporal causality constraint . finally , the covariance of the given formula of the electromagnetic field of the @xmath0 electron bunch will be checked by verifying that , under a lorentz transformation from the laboratory to the rest reference frame , the expected formula of the electric field of a charge distribution at rest is obtained . in the following , @xmath0 electrons in motion with a common rectilinear and uniform velocity @xmath5 along the @xmath6 axis of the laboratory reference frame are supposed to strike , at a normal angle of incidence , onto a flat ideal conductor surface @xmath7 . the radiator , being in vacuum and placed in the plane @xmath8 of the laboratory reference frame , is supposed to have an arbitrary size and shape . compared to the synchrotron motion of a single electron in the bunch , the time scale of the emission of the radiation pulse by the entire electron bunch can be considered instantaneous . consequently , in modeling the radiation emission from the electron bunch , collisions between electrons in the bunch can be neglected and the @xmath0 electron bunch can be thus described in terms of a frozen - in - time " spatial distribution function , i.e. , a spatial distribution function that is invariant under a time - space translation in the laboratory reference frame ( see also [ cov ] ) . with reference to such an experimental context , the temporal sequence of the @xmath0 electron collisions onto the metallic screen is only ruled by the distribution function of the @xmath0 electron longitudinal coordinates @xmath9 ( @xmath10 ) which , for the sake of ease , are taken at the time @xmath11 when the center of mass of the electron bunch is supposed to strike the metallic surface . on the radiator surface @xmath7 ( @xmath8 ) , the harmonic components of the transverse electric field of both the relativistic charge ( @xmath12 ) and the charge induced conduction electrons ( @xmath13 ) satisfy the following boundary constraint : @xmath14 the boundary condition above rules the dipolar oscillation of the the conduction electrons , which is induced on the ideal conductor surface by the incident relativistic charge , and explains why a radiation field needs to fly away from the boundary surface in order to maintain it equipotential . the transition radiation emission can be thus interpreted as the result of the scattering of an electromagnetic wave from a metallic surface and described , on the basis of the huygens - fresnel principle , as the wave propagation of the transverse component ( @xmath12 ) of the _ virtual quanta _ field @xcite from the surface @xmath7 ( @xmath8 ) to the observation point . the transition radiation field ( @xmath15 ) can be calculated by means of the helmholtz - kirchhoff integral theorem @xcite . under the far - field approximation , this integral can be finally expressed as the fourier transform of the harmonic component of the _ virtual quanta _ field with respect to the spatial coordinates @xmath16 of the radiator surface @xmath7 @xcite : @xmath17 where @xmath18 is the wave - number , @xmath19 is the transverse component of the wave - vector and @xmath20 is the distance from the radiator surface to the observation point that , in the present context , is supposed to be on the @xmath6-axis of the laboratory reference frame . in the case of @xmath0 electrons in a rectilinear and uniform motion with a common velocity @xmath21 , the harmonic component of the _ virtual quanta _ electric field at the radiator surface @xmath7 ( @xmath8 ) reads , see @xcite and also [ vq - lab ] : @xmath22 where @xmath23 , while @xmath24 and @xmath9 @xmath25 are , respectively , the transverse and the longitudinal coordinates of the electrons in the laboratory reference frame at the time @xmath11 when the center of mass of the electron bunch is supposed to strike the metallic surface . taking into account eqs.([due],[prima - eq ] ) , the transition radiation electric field by a @xmath0 electron bunch reads : @xmath26 where @xmath27 with @xmath28 . with reference to eqs.([tre],[quattro1 ] ) , the transition radiation energy spectrum by a @xmath0 electron beam can be finally calculated as the flux of the poynting vector : @xmath29 where in previous equation the radiator surface @xmath7 has , in general , an arbitrary shape and size ( either infinite @xmath30 or finite @xmath31 ) . the _ virtual quanta _ field - see eq.([prima - eq ] ) - shows a field structure as a train of @xmath0 travelling transverse waves hitting the metallic surface with a relative phase delay only dependent on the difference between the longitudinal coordinates of the @xmath0 electrons @xmath9 @xmath25 . in the same way , the radiation field - eqs.([tre],[quattro1 ] ) - results from the linear addition of @xmath0 single particle field amplitudes @xmath32 whose relative emission phases - @xmath33 , @xmath25 - from the metallic surface causally depend on the temporal sequence of the @xmath0 electron collisions onto the metallic screen . the temporal causal structure characterizing the transition radiation field of the @xmath0 electron bunch - eqs.([tre],[quattro1 ] ) - is also a feature of the radiation energy spectrum - see eq.([cento - uno ] ) - where the @xmath0 single particle field amplitudes @xmath32 interfere indeed each other via a relative phase factor only dependent on the relative difference of the electron longitudinal coordinates @xmath9 @xmath25 . in fact , @xmath34 the radiator surface @xmath7 being in principle arbitrary and @xmath35 the formal expression of the transition radiation energy spectrum being expected to be invariant whether the radiator surface @xmath7 is finite ( @xmath31 ) or infinite ( @xmath30 ) and whatever is its shape , the @xmath0 single electron radiation field amplitudes @xmath32 @xmath25 - eqs.([tre],[quattro1 ] ) - can be formulated in terms of an implicit integral form and treated as a sort of special function whose numerical value can be only formally and implicitly stated in view of the general formulation of the radiation energy spectrum , eq.([cento - uno ] ) . therefore , thanks to the arbitrariness of the radiator surface @xmath7 and to the consequent implicit integral formulation of the @xmath0 single electron radiation field amplitudes , the formula of the transition radiation energy spectrum - eq.([cento - uno ] ) - explicitly shows the temporal causal feature : the interference between the different single electron field amplitudes @xmath32 is indeed ruled by the distribution function of the electron longitudinal coordinates @xmath9 @xmath25 via the phase factor @xmath36 @xmath37 . about the electron transverse coordinates @xmath24 @xmath25 , they play the role to contribute , as a function of the electron displacement from the @xmath6-axis where the radiation field is supposed to be observed , not only to the global phase factor of the single electron field amplitude @xmath32 @xmath25 at the observation point but also to the single electron field amplitude itself . it can be indeed demonstrated @xcite that , in the case of a round screen with a finite radius , the integral representing the @xmath0 single electron amplitudes composing the radiation field - eqs.([tre],[quattro1 ] ) - can be solved showing the structure of the product of a real amplitude and a phase factor both depending on the electron transverse coordinates : @xmath38 on the basis of eq.([cov - end - milleeuno ] ) , the formula of the radiation energy spectrum given in eq.([cento - uno ] ) shows the well known dependence on the three - dimensional form factor as well as an intrinsic dependence of the @xmath0 single electron radiation field amplitudes on the electron transverse coordinates . the transverse density of the @xmath0 electrons is indeed an invariant under a lorentz transformation with respect to the direction of motion of the beam . the transverse coordinates of the @xmath0 electrons are thus expected : ( @xmath39 ) to contribute to the relative phase delay of the @xmath0 electron field amplitudes at the observation point as a function of the transverse displacement of the @xmath0 electrons with respect to the beam axis where the radiation field is supposed to be observed ; ( @xmath40 ) because of the lorentz invariance of the transverse projection of the @xmath0 electron density , to leave a covariant mark on the @xmath0 electron radiation field amplitudes - eqs.([tre],[quattro1 ] ) - and , consequently , on both the temporal coherent and incoherent components of the radiation energy spectrum , eq.([cento - uno ] ) . in conclusion , in the most general case of an electron bunch at a normal angle of incidence onto a radiator surface @xmath7 with an arbitrary size and shape ( finite @xmath31 or infinite @xmath30 ) , eq.([cento - uno ] ) represents a temporal causal and covariance consistent formulation of the transition radiation energy spectrum . the covariance of the charge form factor and , in general , of the transition radiation energy spectrum - as formulated in eqs.([tre],[quattro1],[cento - uno ] ) - will be analyzed in the following paragraph . for such a purpose , the fourier transformations of fields and charge distributions as normally defined in the ordinary @xmath41-dimensional space will be reformulated in an explicitly covariant form as @xmath42-dimensional fourier representations in the conjugate fourier wavevector - frequency @xmath42-space @xmath43 of the space - time @xmath44 in the laboratory reference frame @xmath45 or in the conjugate fourier @xmath42-space @xmath46 of the space - time @xmath42-vector @xmath47 in the rest reference frame @xmath48 , see [ lorentz ] , [ cov ] , [ cov - rest ] and [ vq - lab ] . the case of the covariant transformation of the spatial density of a @xmath0 electron bunch , as in the following described , will exemplify the above mentioned explicitly covariant extension of a fourier representation of a field or a charge distribution from the ordinary @xmath41-dimensional space to the lorentz @xmath42-space . in the present context , the @xmath0 electrons of the bunch are described by a frozen " in time spatial distribution function : all the electrons are supposed to move with a rectilinear and uniform velocity @xmath21 along the @xmath6 axis of the laboratory reference frame @xmath45 . in the reference frame @xmath48 , where the @xmath0 electrons are at rest , the distribution function of the spatial density of the electron bunch reads : @xmath49 where @xmath50 @xmath25 are the spatial coordinates of the @xmath0 electrons at the time @xmath51 in @xmath48 . in the ordinary @xmath41-space @xmath52 and the conjugate fourier @xmath41-space @xmath53 of @xmath48 , the spatial density of the @xmath0 electrons - eq.([quattro ] ) - reads @xmath54 the above fourier transformation of the spatial density of the @xmath0 electron bunch is defined in the space @xmath55 and in the conjugate fourier space @xmath56 of the rest frame of reference @xmath48 , which are subspaces of the space - time @xmath47 and of the wavevector - frequency @xmath46 , respectively . consequently , the charge density , as expressed by eq.([cinque ] ) in the ordinary @xmath41-dimensional space of @xmath48 , can not explicitly show the expected covariance under a lorentz transformation from the rest to the laboratory reference frame @xmath57 . an explicitly covariant fourier representation of the spatial density of the @xmath0 electrons can be obtained by extending the fourier transformation as given in eq.([cinque ] ) into the fourier conjugate @xmath42-space @xmath46 of the reference frame @xmath48 , for details see [ cov ] . with reference to [ cov ] , the spatial density of the @xmath0 electron bunch - as given in eq.([cinque ] ) - can be represented in an explicitly covariant form in the space - time and wavevector - frequency @xmath42-spaces of the rest reference frame @xmath48 as follows : @xmath58 where @xmath59 and @xmath60 where @xmath61 @xmath25 are the space - time @xmath42-vectors of the electron coordinates , while @xmath62 and @xmath63 are the space - time and the conjugate fourier wavevector - frequency 4-vectors in the rest reference frame @xmath48 , respectively . the covariance of the fourier representation of the spatial density of @xmath0 electron bunch - as represented in eq.([cinque ] ) or , equivalently , in eqs.([sei],[sette ] ) - can be now easily checked , see also [ cov ] . with reference to eqs.([sei],[sette ] ) , @xmath64 is the only non - lorentz - invariant quantity in the integrand of eq.([sei ] ) . moreover , according to eqs.([sette ] ) , the covariance of @xmath64 only depends on how the delta dirac function @xmath65 transforms under a lorentz transformation . under a lorentz transformation from the rest to the laboratory reference frame @xmath57 - see [ lorentz ] and [ cov ] - the delta dirac function transforms indeed as @xmath66=\frac{\delta(\omega-\vec{w}\cdot\vec{k})}{\gamma}.\label{sette - bis}\ ] ] in conclusion , under a lorentz transformation @xmath57 , the @xmath0 electron spatial density - as represented by eq.([cinque ] ) in the ordinary @xmath41-space or , equivalently , by eqs.([sei],[sette ] ) in the lorentz @xmath42-space - transforms in agreement with the expected time - like covariance as @xmath67 where @xmath68 is the @xmath42-dimensional fourier transformation of the @xmath0 electron spatial density in the laboratory reference frame @xmath45 , see also [ cov ] . in the present context , a bunch of @xmath0 electrons moving with a common rectilinear and uniform velocity @xmath21 along the @xmath6-axis of the laboratory reference frame is supposed to strike , at a normal angle of incidence , a flat metallic screen placed in the plane @xmath8 . for such an experimental situation , equations ( [ tre],[quattro1 ] ) and ( [ cento - uno ] ) formulate the transition radiation energy spectrum in the most general case of a radiator surface @xmath7 having an arbitrary shape and size ( either finite @xmath31 or infinite @xmath30 ) . this formulation of the radiation energy spectrum is very general . it is indeed expressed in terms of the original discrete distribution function of the @xmath0 electron coordinates : no continuous limit approximation is indeed applied to the distribution function of the electron spatial coordinates . furthermore , the size and the shape of the radiator surface being indeed arbitrary , eqs.([tre],[quattro1],[cento - uno ] ) represent a very general and implicit formulation of the transition radiation energy spectrum of a @xmath0 electron bunch which is invariant in relation to the size and shape of the radiator . this formulation of the radiation energy spectrum also meets the temporal causality constraint . for the electromagnetic radiative mechanism of the @xmath0 electron bunch that is described in the present paper , the causality constraint rules the temporal correlation between the single electron collision onto the metallic screen and the consequent emission from the metallic surface of the corresponding single electron radiation field amplitude @xmath32 @xmath25 , see eqs.([tre],[quattro1 ] ) . in the considered experimental context , the temporal sequence of the @xmath0 electron collisions onto the metallic screen is only ruled by the distribution function of the longitudinal coordinates @xmath9 @xmath25 of the @xmath0 electrons . thanks to the implicit integral formulation of the @xmath0 single electron amplitudes @xmath32 @xmath25 composing the radiation field - see eqs.([tre],[quattro1 ] ) - the general formula of the transition radiation energy spectrum - eq.([cento - uno ] ) - explicitly meet the temporal causality constraint . in eq.([cento - uno ] ) indeed , the dependence of the @xmath0 single electron field amplitudes on the electron transverse coordinates @xmath24 @xmath25 being encoded in the implicit integral formulation given in eqs.([tre],[quattro1 ] ) , the interference between the @xmath0 single electron field amplitudes @xmath32 @xmath25 is a direct function of the corresponding emission phases from the metallic surface , @xmath36 @xmath37 , in agreement with the temporal causal principle . about the covariance of the transition radiation energy spectrum of the @xmath0 electron bunch as formulated in eqs.([tre],[quattro1],[cento - uno ] ) , this can be argued on the basis of the following analysis : first , verify the covariance of the formal steps leading from the _ virtual quanta _ field of the @xmath0 electron bunch to the resultant radiation energy spectrum , see eq.([prima - eq ] ) and eqs.([tre],[quattro1],[cento - uno ] ) ; finally , verify the covariance of the given expression of the _ virtual quanta _ field under a lorentz transformation from the laboratory reference frame @xmath45 to the reference frame @xmath48 of rest of the @xmath0 electrons . about the covariance of the formal procedure leading from the _ virtual quanta _ field to the radiation field , the following observation can be done . on the basis of the far - field implementation of the helmholtz - kirchhoff integral theorem @xcite , a given harmonic component of the transition radiation field of the @xmath0 electron bunch can be obtained by fourier transforming the corresponding _ virtual quanta _ field with respect to the spatial coordinates of the radiator surface @xmath7 , see eq.([due ] ) . the calculation of such a fourier transformation only involves the coordinates @xmath69 of the radiator surface @xmath7 and the related conjugate fourier coordinates @xmath70 which are transverse with respect to the direction of motion of the electron bunch and thus invariant under a lorentz transformation . therefore , the covariance properties of the _ virtual quanta _ field are entirely and unalterably transferred into the transition radiation field - see eq.([prima - eq ] ) and eq.([tre ] ) - via the fourier transformation as defined in eq.([due ] ) . finally , about the covariance of the charge form factor formulation in the transition radiation energy spectrum - eqs.([tre],[quattro1],[cento - uno ] ) - this can be verified by analyzing how the _ virtual quanta _ field of the @xmath0 electron bunch transforms under a lorentz transformation from the laboratory to the rest reference frame ( @xmath71 ) , see also [ lorentz ] . in the present context , a frozen - in - time " electron bunch is considered , i.e. , the @xmath0 electrons in motion with the same velocity @xmath21 are described by a distribution function of the spatial coordinates that is invariant under a time - space translation in the laboratory reference frame @xmath45 , see also [ cov ] @xmath72 where @xmath73 @xmath25 are the @xmath0 electron spatial coordinates at the time @xmath11 when , for instance , the center of mass of the @xmath0 electron coordinates is supposed to cross the plane @xmath8 . for such an experimental situation , with reference to @xcite , the expression of the charge electric field reads , see also [ vq - lab ] : @xmath74}{[k^2-(\omega / c)^2 ] } \left(\sum_{j=1}^{n}e^{-i\vec{k}\cdot\vec{r}_{0j}}\right)\delta(\omega-\vec{w}\cdot\vec{k}).\label{cov - bisbis}\end{aligned}\ ] ] the expression above of the electric field of the @xmath0 electron bunch does not explicitly show the expected covariance as the argument of the phase factor in eq.([cov - bisbis ] ) clearly indicates . following the procedure already described in previous subsection [ parag - cov - bunch - density ] and in [ cov ] , [ cov - rest ] and [ vq - lab ] , the explicit covariance of the electric field of the @xmath0 electrons can be retrieved by extending the fourier representation of the charge electric field from the ordinary @xmath41-dimensional spaces of the spatial coordinates @xmath75 and conjugate wave - vector coordinates @xmath76 into the lorentz @xmath42-space of the space - time @xmath77 and the conjugate fourier @xmath42-space of the wavevector - frequency @xmath78 . with reference to the time - space translation of the spatial coordinates of the @xmath0 electrons in the laboratory reference frame @xmath45 , see equations above , the fourier representation of the electric field of the @xmath0 electrons - given in eq.([cov - bisbis ] ) - can be reformulated in the following explicitly covariant form , see also eqs.([vq - lab - uno],[vq - lab - due ] ) in [ vq - lab ] : @xmath79 where @xmath80 and , taking into account the following time - space translation @xmath81 ( see coordinates transformation above ) , @xmath82}{[k^2-(\omega / c)^2 ] } \left(\sum_{j=1}^{n}e^{-i\vec{k}\cdot(\vec{r}_{j}-\vec{w}t)}\right)\delta(\omega-\vec{w}\cdot\vec{k})\nonumber\\ & & = -i(8\pi^2e)\frac{[\vec{k}-(\omega\vec{w}/c^2 ) ] } { \kappa^2}\left(\sum_{j=1}^{n}e^{-i\kappa\cdot\xi_j}\right)\delta(\omega-\vec{w}\cdot\vec{k}),\label{cov - due}\end{aligned}\ ] ] where the @xmath42-vector @xmath83 @xmath25 represents the space - time coordinates of the @xmath0 electrons in the laboratory reference frame . the covariance of the electric field of the @xmath0 electron bunch - see eq.([cov - due ] ) or equivalently eq.([cov - bisbis ] ) - can be finally verified , as in following argued . in fact , looking at the integrand of eq.([cov - uno ] ) and at eq.([cov - due ] ) , the term @xmath84\delta(\omega-\vec{w}\cdot\vec{k})$ ] is the only non - lorentz - invariant quantity whose covariance has to be checked . under a lorentz transformation from the laboratory to the rest reference frame ( @xmath71 ) , such a quantity transforms indeed as , see also [ lorentz ] and [ vq - lab ] : @xmath85 = \delta(\frac{\omega'}{\gamma})=\gamma\delta(\omega')\\ ( k_z-\omega w / c^2)\longrightarrow\gamma(k'_z+\beta\omega'/c)-\gamma(\omega'+\beta ck'_z)w / c^2= \frac{k'_z}{\gamma } \end{array } \right.\ ] ] finally , under a lorentz transformation @xmath71 , eqs.([cov - uno],[cov - due ] ) transform as : @xmath86\delta(\omega-\vec{w}\cdot\vec{k})=\nonumber\\ & & = -\frac{iec}{2\pi^2}\int d^4\kappa'e^{i\kappa'\cdot\xi'}\frac{\sum_{j=1}^{n}e^{-i\kappa'\cdot\xi'_j}}{\kappa'^2 } \left ( \begin{array}{c } \gamma k'_x\\ \gamma k'_y\\ k'_z\\ \end{array}\right)\delta(\omega')=\nonumber\\ & & = \left ( \begin{array}{c } \gamma e'_x(\vec{r'},t ' ) \\ \gamma e'_y(\vec{r'},t ' ) \\ e'_z(\vec{r'},t')\\ \end{array}\right),\label{cov - tre}\end{aligned}\ ] ] where , in the last term of eq.([cov - tre ] ) , the three components of the charge electric field @xmath87 in the rest reference frame @xmath48 can be identified , see also [ cov - rest ] : @xmath88 equation([cov - cinque ] ) is the explicitly covariant fourier representation of the electric field of a @xmath0 electron bunch in the rest reference frame @xmath48 , where @xmath47 and @xmath46 are , respectively , the space - time and the conjugate fourier wavevector - frequency @xmath42-vectors of @xmath48 and @xmath89 @xmath25 is the space - time @xmath42-vectors representing the @xmath0 electron coordinates in @xmath48 , see also [ cov - rest ] and [ vq - lab ] . in conclusion , the expression of the _ virtual quanta _ field , see eq.([cov - due ] ) or equivalently eq.([cov - bisbis ] ) , which in the present and other works @xcite is the starting point to the formulation of the radiation energy spectrum of a @xmath0 electron bunch , is demonstrated to meet the covariant constraint . the formulation of the radiation energy spectrum - see eqs.([tre],[quattro1],[cento - uno ] ) - is also demonstrated to be covariant and temporal causality consistent . equations ( [ tre],[quattro1],[cento - uno ] ) are indeed the final result of a series of temporal - causality - consistent formal steps which preserve and transmit the original covariance of the _ virtual quanta _ field of the @xmath0 electron beam into the radiation energy spectrum . finally , about the invariance of the transverse projection of the @xmath0 electron density with respect to the direction of motion and about the covariant imprinting of it on the charge electric field and , consequently , on the radiation field and on the radiation energy spectrum , the following observation can be done . the effect of the transverse electron coordinates @xmath24 @xmath25 on the fourier transformation of the charge electric field is driven by a phase factor , see eqs.([cov - uno],[cov - due ] ) and eqs.([cov - tre],[cov - cinque ] ) . this is a function of the lorentz invariant quantity @xmath90 @xmath25 , where @xmath70 are the transverse component of the wave - vector . with reference to eqs.([cov - uno],[cov - due ] ) and eqs.([cov - tre],[cov - cinque ] ) , the charge electric field shows a dependence on the transverse electron coordinates which , under a lorentz transformation from the laboratory to the rest reference frame , transfers and transforms in a covariant way and whose observability at a given wavelength remains unchanged , this being determined by the lorentz - invariant scalar product of the electron transverse coordinates and the transverse components of the wave - vector . in the light of such a meaning , the statement that , see also end of section [ pseudo - virtual ] , the lorentz invariance of the transverse projection of the @xmath0 electron density is expected to leave a covariant mark on the @xmath0 electron radiation field amplitudes - eqs.([tre],[quattro1 ] ) - and , consequently , on both the temporal coherent and incoherent components of the radiation energy spectrum - see eq.([cento - uno ] ) - has to be read and integrated with the statement that the observability of the covariant effect of the transverse coordinates on the charge electric field remains unchanged , passing from an inertial reference frame to the other , this being a function of a lorentz - invariant quantity . in the present paper , the issue of the covariant formulation of the radiation energy spectrum and , consequently , of the charge form factor of a @xmath0 electron bunch is argued . the @xmath0 electron bunch is supposed to collide , at a normal angle of incidence , onto a metallic screen @xmath7 which has an arbitrary shape and size ( either finite @xmath31 or infinite @xmath30 ) and is supposed to behave as an ideal conductor in the wavelength region of interest . as the covariance of a charged distribution is expected to evolve from a charge - point - like into a charge - density - like one when passing from a single electron to a @xmath0 electron bunch , the covariance of the radiation energy spectrum as well is expected to behave in the same way when a @xmath0 electron bunch is considered instead of a single electron . in order to verify the covariant formulation of the radiation energy spectrum , the formal steps leading from the virtual quanta field to the transition radiation energy spectrum of a @xmath0 electron bunch have been explicitly derived and checked to meet , first , the temporal causality constraint and , then , to be covariance consistent . finally , the expression of the _ virtual quanta _ field , which , in the present paper and in other papers @xcite , is the starting point to achieve the formula of the radiation energy spectrum , has been checked to transform in a covariant way under a lorentz transformation from the laboratory reference frame ( @xmath45 ) to the reference frame of rest of the @xmath0 electron bunch ( @xmath48 ) . in order to perform such a covariance check , the fourier representations of fields and charged distributions have been suitably extended from the ordinary @xmath41-dimensional space of the spatial coordinates and the conjugate fourier wave - vectors into the lorentz @xmath42-spaces of the space - time and of the conjugate fourier wavevector - frequency in both @xmath45 and @xmath48 . in a temporal causal and covariant formulation of the transition radiation energy spectrum of a @xmath0 electron bunch , the invariance of the projection of the electron density in the transverse plane with respect to the direction of motion manifests itself as a covariant feature of the @xmath0 single electron amplitude composing the radiation field and , consequently , of both the temporal coherent and incoherent components of the transition radiation energy spectrum . in the present work , the case of a bunch of @xmath0 electrons in a rectilinear and uniform motion with a common velocity @xmath21 along the positive direction of the z - axis of the laboratory reference frame @xmath45 is considered . in the present paper , lorentz transformations from the laboratory reference frame @xmath45 to the reference frame of rest @xmath48 of the @xmath0 electron bunch - and viceversa - are applied to fields and spatial distributions of charges . therefore , for the sake of easy reading , the main results are below reported . the @xmath42-vectors of @xmath45 , the space - time @xmath44 and the conjugate fourier wavevector - frequency @xmath43 , lorentz - transform into the respective counterparts @xmath47 and @xmath46 of @xmath48 according to @xcite ( @xmath91 ) : @xmath92 @xmath93 where @xmath94 and @xmath95 . as for the electromagnetic fields @xmath96 and @xmath97 , lorentz transformations @xmath91 read : @xmath98 and @xmath99 the inverse lorentz transformations @xmath71 for coordinates and fields can be obtained from previous equations by substituting @xmath100 and by exchanging the prime in field components and coordinates . @xmath0 electrons in a bunch are supposed to move with a common rectilinear and uniform velocity in the laboratory reference frame @xmath45 . in the frame of reference @xmath48 where each electron is at rest , the @xmath0 electrons can be described in terms of a static " distribution function of the particle density : @xmath101 where @xmath52 and @xmath50 @xmath25 are the @xmath0 electron coordinates in the rest reference frame @xmath48 . in the conjugate fourier space @xmath56 of the ordinary @xmath41-space of the spatial coordinates @xmath55 of @xmath48 , the equation above reads @xmath102 where @xmath103 is the fourier transform of the distribution function of the @xmath0 particle density . since the bunch density - as represented by eq.([cov2 ] ) - is only defined in a subspace @xmath56 of the wavevector - frequency 4-vector @xmath46 of @xmath48 , it can not explicitly show the expected lorentz - covariance . the explicit covariance of the spatial density of the @xmath0 electrons can be directly retrieved by extending the fourier representation of eqs.([cov2],[cov3 ] ) into the wavevector - frequency @xmath46 and space - time @xmath47 @xmath42-spaces of the rest reference frame @xmath48 . such a result can be achieved by extending the integral of eq.([cov2 ] ) into the @xmath42-space @xmath104 and the integrand into itself times @xmath105 where @xmath65 is a delta dirac function . following such a formal procedure , eq.([cov2 ] ) can be reformulated as follows : @xmath106 where @xmath107 with @xmath63 and @xmath62 and where @xmath61 @xmath25 are the space - time coordinate @xmath42-vectors of the electrons in the rest reference frame @xmath48 . thanks to the extension of the fourier representation of the spatial density of the @xmath0 electron bunch into the @xmath42-space of the rest reference frame @xmath48 - see eqs.([cov4],[cov5 ] ) - the expected covariance of the charge density can be explicitly recovered . the following quantities are indeed lorentz - invariant @xmath108 consequently , under a lorentz transformation from the rest to the laboratory reference frame ( @xmath91 ) , eq.([cov5 ] ) transforms as follows , see also [ lorentz ] : @xmath109=\nonumber\\ & & = 2\pi\left(\sum_{j=1}^{n}e^{-i\kappa\cdot\xi_j}\right)\frac{\delta(\omega-\vec{w}\cdot\vec{k})}{\gamma}= \frac{1}{\gamma}\rho(\vec{k},\omega),\label{cov6}\end{aligned}\ ] ] where @xmath110 represents the @xmath42-dimensional fourier transformation of the distribution function of the spatial density of the @xmath0 electron bunch in the laboratory reference frame @xmath45 . in the equation above , @xmath111 @xmath25 are the @xmath42-vector space - time coordinates of the electrons in @xmath45 as well as @xmath112 and @xmath113 are , respectively , the space - time and the conjugate fourier wavevector - frequency lorentz @xmath42-vectors of @xmath45 . in conclusion , under a lorentz transformation @xmath91 , the distribution function of the spatial density of the @xmath0 electrons - see eqs.([cov1],[cov2 ] ) and eqs.([cov6],[cov6-bis ] ) - transforms according to the expected covariance : @xmath114 where @xmath115 represents the distribution function of the @xmath0 electron density in the laboratory reference frame @xmath45 , see also eqs.(@xmath116,@xmath117 ) in @xcite . the distribution function of the spatial density of the @xmath0 electrons - as represented by eq.([cov8 ] ) - is invariant under a time - space translation in the laboratory reference frame @xmath45 @xmath72 or , in other words , it is frozen - in - time " as already mentioned in the present paper where the case of a @xmath0 electron bunch in a rectilinear and uniform motion is considered . looking at both integrand of eqs.([cov4],[cov5],[cov6-bis],[cov8 ] ) , the presence of a delta dirac function and , in particular , the argument of this delta are instructive about the attribute given to the charge distribution function to be static " in the rest reference frame @xmath48 or frozen - in - time " in the laboratory reference frame @xmath45 . in the reference frame @xmath48 , where the @xmath0 electrons are at rest , @xmath65 in eq.([cov5 ] ) is indicating that only a wave with @xmath118 constitutes the relevant harmonic contribution to the fourier transformation of the charge density . conversely , in the laboratory reference frame @xmath45 , where all the @xmath0 electrons move with a common rectilinear and uniform velocity @xmath5 , the delta dirac @xmath119 in eq.([cov8 ] ) is indicating that a plane wave traveling with the same velocity @xmath5 as the electron bunch constitutes the only relevant harmonic contribution to the fourier transformation of the charge density . in the rest reference frame @xmath48 of the @xmath0 electron bunch , the magnetic field @xmath120 . the electric field @xmath121 can be obtained from the poisson equation for the scalar potential @xmath122 : @xmath123 in the fourier @xmath41-space @xmath53 of @xmath48 , where the scalar potential and the charge density read @xmath124 the poisson equation reads @xmath125 and the electric field reads @xmath126 where the explicit expression of @xmath127 is given in eq.([cov3 ] ) . the fourier representation of the electric field of a @xmath0 electron bunch in the rest reference frame @xmath48 - as given in eq.([rest4 ] ) - is evidently non - explicitly covariant being restricted to the subspaces @xmath53 and @xmath128 , respectively , of the wavevector - frequency @xmath129 and space - time @xmath130 lorentz @xmath42-vectors of @xmath48 . the covariance of the electric field can be explicitly retrieved by suitably extending the fourier transformation in eq.([rest4 ] ) into the fourier @xmath42-space @xmath46 . following the same formal procedure already described in [ cov ] for the charge density , the fourier transformation of the electric field of the @xmath0 electron bunch in the rest reference frame @xmath48 - eq.([rest4 ] ) - can be reformulated as : @xmath131}=\nonumber\\ & & = -\frac{iec}{2\pi^2}\int d^4\kappa'e^{i\kappa'\cdot\xi ' } \frac{(\sum_{j=1}^{n}e^{-i\kappa'\cdot\xi'_j})}{\kappa'^2}\vec{k'}\delta(\omega')=\nonumber\\ & & = \frac{c}{(2\pi)^4}\int d^4\kappa ' \ , e^{i\kappa'\cdot\xi'}\vec{e'}(\vec{k'},\omega'),\label{rest5}\end{aligned}\ ] ] where @xmath132 are the space - time coordinates of the @xmath0 electrons @xmath25 in @xmath48 . in the integrand of the equation above , the only non - lorentz invariant quantity is the fourier transform of the electric field @xmath133 : @xmath134 the term @xmath135 is the only quantity in eqs.([rest5],[rest6 ] ) whose covariance has to be checked . under a lorentz transformation from the rest to the laboratory reference frame @xmath57 - see also [ lorentz ] - such a quantity transforms as : @xmath136= \frac{k_{x , y}}{\gamma}\delta(\omega - wk_z)\\ k'_z\delta(\omega')\longrightarrow \gamma(k_z-\frac{w\omega}{c^2})\delta[\gamma(\omega - wk_z)]=(k_z-\frac{w\omega}{c^2})\delta(\omega - wk_z ) \end{array } \right.\ ] ] where @xmath21 is the velocity of the electrons in the laboratory reference frame @xmath45 . finally , under a lorentz transformation @xmath91 , the electric field of a @xmath0 electron bunch in the rest reference frame @xmath48 - see eqs.([rest4],[rest5],[rest6 ] ) - transforms into @xmath137 where , in the equation above , the @xmath41 components of the electric field of the @xmath0 electron bunch in the laboratory reference frame @xmath45 can be recognized , see eqs.([cov - uno],[cov - due ] ) and also [ vq - lab ] : @xmath138 in the case of a bunch of @xmath0 electrons in motion in vacuum with a common uniform and rectilinear velocity @xmath5 in the laboratory reference frame @xmath45 , the fourier representation of the propagation equations of the @xmath42-potential @xmath139 in the @xmath42-space @xmath140 of @xmath45 reads in the gauge of lorentz as @xmath141 where @xmath142 is the fourier transformation of the current density vector . with reference to the equations above , the fourier transformation of the charge electric field reads : @xmath143}\rho(\vec{k},\omega).\label{vq - lab - uno}\end{aligned}\ ] ] with reference to eq.([cov6-bis ] ) in [ cov ] , the fourier representation of the electric field of the @xmath0 electron bunch - as given in eq.([vq - lab - uno ] ) - can be expressed in an explicitly covariant form in the space - time @xmath77 and in the @xmath42-space of the conjugate fourier wavevector - frequency @xmath144 of the laboratory reference frame @xmath45 , see also eqs.([cov - uno],[cov - due ] ) and eq.([rest8 ] ) in [ cov - rest ] : @xmath145 where @xmath146 and @xmath83 @xmath25 are the space - time @xmath42-vectors of the @xmath0 electron coordinates in the laboratory reference frame @xmath45 . under a lorentz transformation from the laboratory to the rest reference frame @xmath147 , the non - lorentz - invariant quantities in eq.([vq - lab - due ] ) transforms as follows : @xmath148 see also [ lorentz ] . taking into account the equations above , in the laboratory reference frame @xmath45 , the covariance of the @xmath42-dimensional fourier representation of the electric field of the @xmath0 electron bunch - eq.([vq - lab - due ] ) - can be finally checked . under a lorentz transformation @xmath71 , eq.([vq - lab - due ] ) transforms indeed as @xmath149 where , in the second term of previous equation , the explicitly covariant fourier representation of the charge electric field @xmath133 in the rest reference frame @xmath48 can be recognized , see eq.([cov - cinque ] ) and also eq.([rest6 ] ) in [ cov - rest ] . for the sake of completeness , the electric field of @xmath0 electrons in the rest reference frame @xmath48 - see second term of eq.([vq - lab - duecento ] ) or , equivalently , eq.([cov - cinque ] ) or eq.([rest6 ] ) in [ cov - rest ] - will be in the following obtained by lorentz transforming the electric and magnetic fields in the laboratory frame of reference @xmath45 . in the laboratory reference frame @xmath45 , the fourier transformation of the magnetic field by a @xmath0 electron bunch reads : @xmath150 the @xmath42-dimensional fourier representation of the transverse components of the charge electric field in @xmath48 can be obtained from the corresponding components of the electromagnetic field in @xmath45 as ( see also [ lorentz ] ) : @xmath151=\nonumber\\ & & = \gamma(-i8\pi^2 e)\frac{\left(\sum_{j=1}^{n}e^{-i\kappa\cdot\xi_j}\right)}{\kappa^2}\delta(\omega-\vec{w}\cdot\vec{k})[\left ( \begin{array}{c } k_x \\ k_y \\ \end{array}\right)\pm\beta^2\left ( \begin{array}{c } -k_x \\ k_y \\ \end{array}\right)]=\nonumber\\ & & = \frac{-i8\pi^2e}{\gamma } \frac{\left(\sum_{j=1}^{n}e^{-i\kappa\cdot\xi_j}\right)}{\kappa^2}\delta(\omega - wk_z)\left ( \begin{array}{c } k_x \\ k_y \\ \end{array}\right)=\nonumber\\ & & = -i8\pi^2e\frac{\left(\sum_{j=1}^{n}e^{-i\kappa'\cdot\xi'_j}\right)}{\kappa'^2}\delta(\omega')\left ( \begin{array}{c } k'_x \\ k'_y \\ \end{array}\right)\label{vq - lab - trecento}\end{aligned}\ ] ] where in the last term of previous equation the transverse component of the fourier transform of the charge electric field in the rest reference frame @xmath48 can be recognized ( see [ cov - rest ] , eq.([rest6 ] ) in particular ) . concerning the longitudinal component @xmath152 , the corresponding covariant transformation is trivial , see eq.([vq - lab - duecento ] ) or eq.([rest6 ] ) . 00 v. l. ginzburg and i. m. frank , soviet phys . jetp * 16 * ( 1946 ) 15 . garibian , sov . jetp , * 6 * ( 33 ) , 6 ( 1958 ) 1079 - 1085 garibyan , sov . jetp , * 37 * ( 10 ) , 2 ( 1960 ) 372 - 376 . frank , sov . , * 4 * , 5 , ( 1962 ) 740 - 746 . f.g . bass , v. m. yakovenko , sov . , * 8 * , 3 ( 1965 ) 420 - 444 frank , sov . , * 8 * , 5 ( 1966 ) 729 - 741 . ter - mikaelian , _ high - energy electromagnetic processes in condensed media _ , wiley , new york ( 1972 ) . ginzburg , v.n . tsytovich , _ transition radiation and transition scattering _ , adam hilger , bristol ( 1990 ) . s. prnster , c.l . wang , l.c.l . yuan , j. oostens , phys . letters * 28b * , 1 ( 1968 ) 47 - 50 . alikhanyan , k.a . ispiryan , a.g . oganesyan , sov . jetp * 29 * , 6 ( 1969 ) 964 - 967 . yuan , c.l . wang , s. prnster , phys . * 23 * , 9 ( 1969 ) 496 - 498 . alikhanyan , k.a . ispirian , a.g . oganessian , a.g . tamanian , nucl . * 89 * ( 1970 ) 147 - 153 . l. wartski , j. marcou , a. papiernik , s. roland , m. bolore , g. filippi , nucl . * 106 * ( 1973 ) 103 - 107 . cherry , g. hartmann , d. mller , t.a . prince , phys . d * 10 * , 11 ( 1974 ) 3594 - 3607 . fabjan , w. struczinkski , phys . letters * 57b * , 5 ( 1975 ) 483 - 486 . chu , m.a . piestrup , t.w . barbee , jr . pantell , f.r . buskirk , rev . instrum . * 51 * , 5 ( 1980 ) 597 - 601 . piestrup , d.g . boyers , c.i . pincus , q. li , g.d . hallewell , m.j . moran , d.m . skopik , r.m . silzer , x.k . maruyama , d.d snyder , g.b . rothbart , phys . a * 45 * , 2 ( 1992 ) 1183 . fiorito , d.w . rule , m.a . piestrup , x.k . maruyama , r.m . silzer , d.m . skopik , a.v . shchagin , phys . e * 51 * , 4 ( 1995 ) r2759 . wakely , s. plewnia , d. mller , j.r . hrandel , f. gahbauer , nucl . instr . and meth . res . a * 531 * ( 2004 ) 435 - 444 . a. andronic et al . , instr . and meth . res . a * 558 * ( 2006 ) 516 - 525 . l. wartski , s. roland , j. lasalle , m. bolore , g. filippi , j. appl . phys . * 46 * ( 1975 ) 8 . j. bosser , j. mann , g. ferioli , l. wartski , nucl . . meth . in phys . res . a * 238 * ( 1985 ) 45 - 52 . rule , nucl . instr . and meth . b * 24/25 * ( 1987 ) 901 - 904 . lumpkin , r.b . feldman , d.w . feldman , s.a . apgar , b.e . carlsten , r.b . fiorito , d.w . rule , nucl . instr . and meth . res . a * 285 * ( 1989 ) 343 - 348 . rule , r.b . fiorito , a.h . lumpkin , r.b . feldman , b.e . carlsten , nucl . meth . in phys . res . a * 296 * ( 1990 ) 739 - 743 . de loos , s.b . van der geer , w.p . leemans proc . of the 1994 european particle accelerator conference 1679 - 1681 . j .- denard , p. piot , k. capek , f.feldl , proc . pac 97 , vancouver , canada , ( 1997 ) 2198 - 2200 . p. catravas , w. p. leemans , e. esarey , m. zolotorev , d. whittum , r. iverson , m. hogan , d. walz , proc . of the 1999 particle accelerator conference , new york , ( 1999 ) 2111 - 2113 . lebedev , nucl . instr . meth . in phys . res . a * 372 * ( 1996 ) 344 - 348 . x. artru , r. chehab , k. honkavaara , a. variola , nucl . instr . meth . in phys . b * 145 * ( 1998 ) 160 - 168 . m. castellano , v.a . verzilov , phys . st - ab * 1 * ( 1998 ) 062801 . potylitsyn , nucl . instr . and meth . b * 145 * ( 1998 ) 169 - 179 . n.f . shulga , s.n . dobrovolsky , v.g . syshchenko , nucl . instr . and meth . b * 145 * 1998 180 - 184 . e. bravin , t. lefvre , proc . dipac 2003 , mainz , germany , 92 - 94 . f. sakamoto , h. iijima , k. dobashi , t. imai , t. ueda , t. watanabe , m. uesaka . * 44 * , 3 ( 2005 ) 1485 - 1491 . welsch , e. bravin , t. lefvre , rev . ( 2007 ) 73 - 79 . holloway , r.b . fiorito , a.g . shkvarunets , p.g . oshea , s.v . benson , d. douglas , p. evtushenko , k. jordan , phys . st - ab * 11 * , 082801 ( 2008 ) . u. happek , a.j . sievers , e.b . blum , phys . lett . * 67 * , 21 ( 1991 ) 2962 - 2965 . y. shibata , k. ishi , t. takahashi , t. kanai , m. ikezawa , k. takami , t. matsuyama , k. kobayashi , y. fujita , phys . rev . a , * 45 * , 12 ( 1992 ) r8340 . w. barry , cebaf preprint pr-91 - 012 ( 1991 ) . t. takahashi , y. shibata , f. arai , t. ohsaka , m. ikezawa , y. kondo , t. nakazato , s. urasawa , r. kato , s. niwano , m. oyamada , phys . e * 48 * , 6 ( 1993 ) 4674 - 4677 . p. kung , h.lihn , h. wiedemann , d. bocek , phys . lett . * 73 * , 7 ( 1994 ) 967 - 970 . y. shibata , t. takahashi , t. kanai , k. ishi , m. ikezawa , j. ohkuma , s. okuda , t. okada , phys . e * 50 * , 2 ( 1994 ) 1479 - 1484 . r. lai , u. happek , a.j . sievers , phys . e * 50 * , 6 ( 1994 ) r4294 . j. rosenzweig , g. travish , a. tremaine , nucl . instr . and meth . res . a * 365 * ( 1995 ) 255 - 259 . h. lihn , p. kung , c. settakorn , h. wiedemann , d. bocek , phys . e * 53 * , 6 ( 1996 ) 6413 - 6418 . v. schlott , h. loos , h. genz , h .- grf , r. hahn , a. richter , m. thomas , t. wesp , m. wiencken , part . * 52 * ( 1996 ) 45 - 53 . schneider , r. lai , w. walecki , a.j . sievers , nucl . instr . and meth . res . a * 396 * ( 1997 ) 283 - 292 . t. watanabe , m. uesaka , j. sugahara , t. ueda , k. yoshii , y. shibata , f. sakai , s. kondo , m. kando , h. kotaki , k. nakajima , nucl . instr . and meth . a * 437 * ( 1999 ) 1 - 11 . m. geitz , g. schmidt , p. schmser , g. v.walter , nucl . instr . and meth . in phys . res . a * 445 * ( 2000 ) 343 - 347 . d. mihalcea , c.l . bohn , u. happek , p. piot , phys . st - ab * 9 * , 082801 ( 2006 ) . j. van tilborg , c.b . schroeder , c.v . filip , cs . tth , c.g.r . geddes , g. fubiani , r. huber , r.a . kaindl , e. esarey , w.p . leemans , phys . lett . * 96 * , 014801 ( 2006 ) . d. stterlin , d. erni , m. dehler , h. jckel , h. sigg , v. schlott , nucl . instr . and meth . b * 264 * ( 2007 ) 361 - 370 . r.a . marsh , a.s . kesar , r.j . temkin , phys . st - ab * 10 * , 082801 ( 2007 ) . s. casalbuoni , b. schmidt , p . schmser , v. arsov , s. wesch , phys . st - ab * 12 * , 030705 ( 2009 ) . , ed . palik , academic press ( 1985 ) . jackson , _ classical electrodynamics _ , wiley , new york ( 1975 ) . m. born , e. wolf , _ principles of optics _ , pergamon press , oxford ( 1965 ) . orlandi , _ covariant formulation of the transition radiation energy spectrum of an electron beam at a normal angle of incidence onto a round metallic screen _ paul scherrer institut preprint psi - pr-12 - 02 . orlandi , opt . commun . * 211 * ( 2002 ) 109 - 119 . orlandi , opt . commun . * 267 * ( 2006 ) 322 - 334 . orlandi , proc . fel 2005 , stanford university ( 2005 ) , jacow / econf c0508213 , 576 - 579 . orlandi , proc . epac 08 , genoa , italy , ( 2008 ) 1221 - 1223 . orlandi , b. beutner , r. ischebeck , v. schlott , b. steffen , proc . fel 2009 , liverpool , uk , 153 ( 2009 ) . orlandi , r. ischebeck , v. schlott , b. steffen , proc . fel 2010 , malm , sweden , ( 2010 ) .

we present ubvri surface photometry for 16 dwarf elliptical galaxies in the virgo cluster with previously measured kinematic properties . the global optical colors are red , with median values for the sample of 0.24 0.03 in ( u b ) , 0.77 0.02 in ( b v ) , and 1.02 0.03 in ( v i ) . we recover the well known color - magnitude relation for cluster galaxies , but find no significant difference in dominant stellar population between rotating and non - rotating dwarf elliptical galaxies ; the average age of the dominant stellar population is 5 - 7 gyr in all 16 galaxies in this sample . analysis of optical spectra confirm these age estimates and indicate fe and mg abundances in the range of 1/20th to 1/3 of solar , as expected for low luminosity galaxies . based on lick indices and simple stellar population models , the derived [/fe ] ratios are sub - solar to solar , indicating a more gradual chemical enrichment history for des as compared to giant elliptical galaxies in the virgo cluster . these observations confirm the marked difference in stellar population and stellar distribution between dwarf and giant elliptical galaxies and further substantiate the need for alternative evolutionary scenarios for the lowest mass cluster galaxies . we argue that it is likely that several different physical mechanisms played a significant role in the production of the virgo cluster de galaxies including _ in situ _ formation , infall of des that were once part of local group analogs , and transformation of dwarf irregular galaxies by the cluster environment . the observations support the hypothesis that a large fraction of the virgo cluster des are formed by ram pressure stripping of gas from infalling dis .

astro-ph0409346

dwarf elliptical galaxies are ubiquitous in high density regions . for example , in the local group , dwarf elliptical galaxies out number high luminosity galaxies by a factor of 6 @xcite , and more than 50% of the galaxies in the virgo cluster are des @xcite . however , while the clustering properties of dwarf elliptical galaxies appear to be an extension of the well known morphology - density relation for giant elliptical galaxies ( e.g. , * ? ? ? * ; * ? ? ? * ) , their structural parameters suggest that dwarf ellipticals are not merely low mass versions of elliptical galaxies ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? unlike their giant counterparts , gas - poor dwarf elliptical galaxies and gas - rich dwarf irregular galaxies have similar stellar distributions which suggests that passive evolution of the stellar population could result in morphological evolution from one dwarf classification to the other ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? in fact , many of the key morphological differences between dwarf elliptical and dwarf irregular galaxies can be traced to the cessation of star formation activity rather than to dynamical differences between the two classes . in support of this , @xcite note that the star formation histories of dwarf elliptical galaxies may be very similar to those of dwarf irregular galaxies right up to the point of the loss of their gas . these similarities led to the hypothesis that dwarf elliptical galaxies are evolved versions of dis whose ism has been disrupted by either internal ( e.g. , blow - out , * ? ? ? * ) or external ( e.g. , ram pressure stripping , * ? ? ? * ) processes . in addition to similar stellar distributions , recent kinematic studies of des in the virgo cluster indicate that a large fraction of des and ds0s have a significant rotational component @xcite . furthermore , the amplitudes of the de rotation curves are comparable to those of similar luminosity dwarf irregular galaxies @xcite . these results support the possibility that some cluster des are evolved dwarf irregular galaxies which have lost their ism and thus have little - to - no star formation activity at the present epoch . the possibility that some dis evolve into des is almost tautological since des must have been gas - rich , star forming low mass galaxies at some point in the past . thus , the key question for dwarf galaxy evolution is not _ whether _ des evolve from dis , but , rather , _ when _ and _ why _ do some dwarf irregular galaxies lose their ism and stop forming stars . the strong clustering properties of des ( e.g. , * ? ? ? * ) suggest that environmental influences may play a key role in the gas - loss process . possible evolution mechanisms include triggered bursts of star formation with subsequent blow - out of the ism ( e.g. , * ? ? ? * ) , or ram pressure stripping of the ism as the galaxy plunges through the hot intra - cluster gas ( e.g. , * ? ? ? * ) . both of these scenarios will result in an overabundance of dwarf elliptical galaxies in high density regions , but the resulting star formation histories are quite different . in a triggered starburst scenario , most of the star formation activity will take place in discrete events which may be correlated with passage through the cluster . on the other hand , if the ism is lost through ram pressure stripping , most of the star formation activity will take place prior to the passage through the cluster , and may occur in a more quiescent manner , typical of dis @xcite . thus , examination of the stellar populations and relative chemical enrichments in dwarf elliptical galaxies may provide insight into which of these mechanisms is the dominant process for the evolution of dis to des in high density regions . within the local supercluster , the virgo cluster is the nearest high density region and thus provides an excellent environment in which to explore stellar populations and kinematic properties of low luminosity galaxies . at the same time , virgo is dynamically young ( e.g. , * ? ? ? * ) and contains several substructural units @xcite ; thus , with a well selected sample it may be possible to study the evolution of dwarf elliptical galaxies within the cluster itself . recent observations of surface brightness fluctuations @xcite have shown that the de population is stretched in depth similarly to the bright elliptical population @xcite but not as extended as claimed previously by @xcite . furthermore , nucleated des appear to be more strongly clustered than non - nucleated des in the cluster ( e.g. , * ? ? ? * ) . in this paper , we examine broad band optical colors and optical spectroscopy of 16 dwarf elliptical galaxies in the virgo cluster to investigate their star formation histories and to search for correlations between kinematic properties and past star formation activity . the sixteen galaxies in this sample were selected as part of a kinematic study of dwarf elliptical galaxies in virgo @xcite ; approximately half of the sample have a significant rotational component . as discussed in @xcite , it is likely that the rotating dwarf elliptical galaxies are remnants of dis that have been stripped of their gas ; the rotating dwarf elliptical galaxies tend to be located near the outskirts of the cluster and their rotation curves and maximum amplitude velocities are comparable to similar luminosity dwarf irregular galaxies . at the same time , the origin of the non - rotating dwarf ellipticals is unclear . are these low luminosity galaxies which began with little - to - no angular momentum ? were they formed through the collision of even lower mass galaxies and thus ended up with no net angular momentum ? did they transfer their angular momentum as they moved through the dense cluster ? in support of the latter process , @xcite have shown that `` tidal stirring '' or repeated tidal shocks suffered by dwarf satellite galaxies can remove the kinematic signature from a low mass galaxy without significant alteration of its stellar distribution . while it is difficult to distinguish between these scenarios based on the data at hand , the common feature of all of these scenarios is that non - rotating dwarf elliptical galaxies may have been significantly altered by their interaction with the cluster environment . thus , it is possible that the galaxies which still retain their kinematic signature have been accreted relatively recently while the non - rotating dwarf elliptical galaxy sample may contain galaxies with a wide range of cluster accretion times , from the very recent to several gyr in the past . since it is likely that the age of the dominant stellar population is indicative of the time of accretion , observations of the stellar populations in these galaxies could provide insight into the accretion timescales . in this paper , we analyze the past star formation activity in virgo dwarf elliptical galaxies based on the combination of stellar population synthesis and spectral synthesis models . the imaging and spectroscopic observations are described in section 2 . we discuss the derived optical structural parameters of the dwarf elliptical galaxies in section 3 . in section 4 , we analyze the optical colors in the context of simple stellar populations . in section 5 , optical spectra are analyzed in the context of simple stellar population models to investigate typical timescales for chemical enrichment . the implications for the observed stellar population ages and enrichment histories are discussed in section 6 . the conclusions of the paper are summarized in section 7 . optical images were obtained of 16 dwarf elliptical galaxies with the vatt 1.8 m and the wiyn 3.5 m telescopes . optical spectra were obtained with the palomar 5 m telescope . the kinematic results of the spectroscopic observations are presented in @xcite ; here , we analyze the spectra in terms of stellar population models . the sample selection and observations are described in this section . as described in @xcite , the sample was selected to optimize the detection of rotational motion in de galaxies . sixteen dwarf elliptical galaxies were selected from the virgo cluster catalog ( vcc , * ? ? ? * ) based on their apparent luminosity ( m@xmath2 ) , morphological classification ( dwarf elliptical or ds0 ) , and apparent ellipticity ( @xmath3 ) . the latter criterion was imposed to maximize the projected rotational velocity in the kinematic observations . in addition , the sample was restricted to those galaxies with measured recessional velocities that place them within the virgo cluster , at an approximate distance of 16.1 mpc ( e.g. , * ? ? ? * ) . further details of the sample selection and distribution of the galaxies within the virgo cluster are described in @xcite . ubvri images of 16 dwarf elliptical galaxies in the virgo cluster were obtained with the vatt 1.8 m on 2003 march 29 - 31 . the telescope was equipped with ccd26 , a thinned loral 3 2048 @xmath4 2048 ccd with 15 pixels . ccd26 has a read noise of 5.7 e@xmath5 and a gain of 1.9 e@xmath5 per adu was used . on - chip binning of 2 @xmath4 2 resulted in 0.375 per pixel and a field of view of 6.4 @xmath4 6.4 arcmin . the galaxies were centered in the field , but dithers of approximately 1 arcmin were taken between each exposure to assist with the flat fielding process . table [ tab : obs ] lists the basic parameters of the de sample and summarizes the observations . unless otherwise noted in table [ tab : obs ] , the vatt 1.8 m observations consisted of 2 exposures of 540 sec , 420 sec , 300 sec , 240 sec , and 300 sec with the u , b , v , r , and i filters , respectively . complete ubvri images were obtained for 9 galaxies in the sample ; bvri images were obtained for an additional 4 galaxies and 3 galaxies were imaged in only b and r. the observations were structured so that the b and r images were obtained on the same night under similar observing conditions ; similarly , the v and i images were taken sequentially . thus , the ( b - r ) and ( v - i ) colors for these observations are quite secure , while ( b - v ) and ( u - b ) colors have additional error propagation terms since the colors are based on the calibrated magnitudes rather than calibration of the instrumental colors . the final combined images were sufficiently deep to permit accurate photometry of the outer regions of the galaxies . typical surface brightness limits for the combined images are 25.8 , 26.8 , 26.0 , 25.6 , and 25.0 mag arcsec@xmath6 , in u , b , v , r , and i , respectively . all 3 nights were photometric . calibration coefficients were derived from observations of standard stars from @xcite interspersed with the galaxy observations . based on the derived photometric coefficients , the bvri images have 2% accuracy while the u - band images are accurate to 3% . the same sample of galaxies was imaged with the wiyn 3.5 m telescope on 2002 may 12 - 16 . poor weather hampered the usefulness of these observations , but the wiyn images provide confirmation for the structural parameters derived from the vatt 1.8 m data . in particular , the wiyn images have slightly better seeing characteristics than the vatt 1.8 m images . the wiyn 3.5 m telescope was equipped with the mini - mosaic camera , which consists of two site 4096 @xmath4 2048 ccds with 15 pixels . the two ccds are read - out by 4 amplifiers ( 2 per ccd ) , with typical read noises of 5.5 e@xmath5 and gains of 1.4 e@xmath5 per adu . the pixel scale of mini - mo is 0.14 per pixel , which results in a field of view of 9.6 @xmath4 9.6 arcmin with a 7.1 gap between the ccds . in all cases , the telescope pointing was offset so that the dwarf elliptical galaxy fit on a single ccd . the wiyn 3.5 m observations consisted of 2 exposures of 600 sec , 600 sec , 300 sec , and 600 sec with the b , v , r , and i filters , respectively . r - band images were obtained for 13 of the 16 des in the sample ; additional bvi images were obtained for 6 of the 13 galaxies . all 5 nights were non - photometric so the limiting surface brightness varied significantly from exposure to exposure . typical image quality for the r - band wiyn images is 1 - 2 . the optical images were reduced and analyzed with the iraf package . the image reduction included bias subtraction and flat fielding . the vatt images were flat fielded to high accuracy using twilight flats . the wiyn 3.5 m images were first flat fielded using dome flats taken during the afternoon ; since these images had a residual flat field problem , a secondary flat field was generated from a combination of the object frames using standard tasks in the mscred package . both the vatt and wiyn i - band images were fringe corrected based on a fringe frame generated by taking the median of all i - band galaxy observations obtained during the night . for the wiyn images , data from the two ccds must be projected onto a common coordinate system prior to further data processing . for each image , the wcs was updated using star lists from the usno catalog and the iraf tasks msczero and msccmatch . the multi - ccd images were then projected into single images using the mscimage task using a common coordinate system for all observations of the same galaxy . for both the vatt and wiyn images , sky values were measured as the mode of galaxy - free regions of the images prior to combination of multiple exposures . the images were then scaled , aligned , and averaged together to make a final combined image . for the non - photometric observations ( wiyn ) , the individual exposures were scaled to the image with the most counts ; scale factors were determined based on the relative intensity of several stars in each frame . the vatt data were photometric and required no additional scaling before combination . the vatt and wiyn images were processed independently in order to provide internal consistency checks for the derived structural parameters . since the vatt images are photometric and higher quality than the wiyn images , the subsequent analysis will focus on results from the vatt observations . the optical spectroscopic observations are described in detail in @xcite . briefly , optical spectra covering the wavelength ranges of 4800 5700 and 8250 8900 were obtained with the double spectrograph on the 5 m palomar telescope in 2001 and 2002 . here , we focus on the `` blue '' spectra , which were centered on the mg ib triplet and include a subset of the spectral regions appropriate for spectral synthesis using lick / ids indices . the observations were obtained on non - photometric , transparent nights . flux calibration was obtained from observations of several spectrophotometric standards @xcite and observations of standard stars from the lick / ids system . the long slit spectra were processed following standard practice [ see @xcite for complete data reduction details ] . after bias subtraction , scattered light correction , flat fielding , wavelength rectification , and night sky line subtraction , the 2-dimensional ( 2-d ) images were flux calibrated . one - dimensional ( 1-d ) spectra were then extracted from the 2-d spectra to create spectra of the integrated light from the galaxy . the 1-d spectra were doppler corrected for the systemic velocity of the galaxies . since the observed spectra have higher spectral resolution than the stellar spectra that define the lick / ids system , the 1-d spectra were then degraded from 1.6 resolution to 8.4 resolution . while the lick indices are based on stellar spectra with variable resolution , the spectral region covered by these de galaxy spectra correspond to a region of approximately constant spectral resolution in the lick / ids system @xcite . lick indices were then measured for the galaxies and the lick standard stars . the observed spectra include the spectral regions appropriate for the following spectral indices : h@xmath7 , fe5015 , mgb , fe5270 , fe5335 , and fe5406 . in addition , similar to @xcite , we define the average fe index , @xmath8 , as the average of fe5270 and fe5335 indices . the standard star observations were used to confirm that the indices measured here are consistent with previous measurements . however , since there were only a handful of comparison stars , and the stars observed comprise a restricted range of spectral types , it was not possible to determine a correction to the observed indices to place the measurements precisely on the lick system . nonetheless , for the few galaxies in common with the @xcite sample , the newly derived lick indices are consistent with the previous measurements . thus , while the derived indices may have some systematic errors , the amplitude of these errors should be minor . furthermore , the _ relative _ measurements for the de galaxies in this sample should be secure even if the absolute values may be less precise . the surface photometry for each galaxy was derived from ellipse fits to the final combined images . prior to ellipse fitting , the foreground stars were masked and replaced by the interpolated background of neighboring pixels . the surface brightness profiles , position angle , and ellipticity are shown as a function of semi - major axis in figure [ fig : surf ] . since the morphological parameters were similar for all filters , the apertures defined by the r - band fit were used for subsequent analysis of the surface brightness profiles and surface colors . the morphological parameters are summarized in table [ tab : results ] . listed in table [ tab : results ] is the semi - major axis at which the b - band surface brightness drops to 25 mag arcsec@xmath6 , r@xmath9 , and the average ellipticity and position angle from the isophotal fits of the r - band images for ellipses with semi - major axes between 5 and r@xmath9 . apertures larger than 5 were used for these average values since several of the galaxies in this sample have bright nuclear regions . also listed in table [ tab : results ] are the b - band apparent magnitudes at which the curves of growth level off and the observed optical colors as measured within the r@xmath9 aperture . as illustrated in figure [ fig : surf ] , the surface colors and aperture colors reach similar values near r@xmath9 so these values were chosen as the most representative galaxy color . also tabulated in table [ tab : results ] are estimates of the foreground galactic extinction as calculated from the dust maps of @xcite . the images were examined for residual disky or boxy isophotes by subtracting the model surface photometry from the image . eleven of the 16 galaxies had residual structure that was readily apparent in the residual images ; the presence of disky or boxy isophotes is tabulated in table [ tab : results ] . deviations from the best - fitting ellipse as measured by b@xmath10 are shown in figure [ fig : surf ] . as previously noted by @xcite , there appears to be little correlation between the presence of disky or boxy residuals and the kinematic properties of the galaxies . as an indication of the kinematic properties of these galaxies , the rotation curve slope from @xcite is listed in table [ tab : results ] . due to the limited spatial extent of the stellar kinematic observations , we list the rotation curve slope rather than a maximum velocity width ; galaxies with rotation curve slopes greater than 20 km s@xmath11 kpc@xmath11 have significant rotation . based on the kinematic observations , vcc 178 , 437 , 543 , 965 , 990 , 1036 , 1743 , 1857 , and 2019 form the subset of rotation dominated des in this sample . table [ tab : results2 ] summarizes the derived photometric properties of the dwarf elliptical galaxies in this sample . the b - band absolute magnitude was calculated from the apparent magnitude assuming all galaxies are at the virgo cluster distance of 16.1 mpc @xcite and correcting for galactic extinction . also listed are the effective radius and effective surface brightness in b - band . for all of the following except the exponential parameters , size parameters are quoted in geometric mean radii where r = @xmath12 ; for the exponential parameters , the semi - major axis is adopted as the radius . the effective radius tabulated in table [ tab : results2 ] is the half - light radius based on the apparent magnitude listed in table [ tab : results ] ; the effective surface brightness is then defined as the average surface brightness within the r@xmath13 elliptical aperture . we note that while these values are model independent , they can be sensitive to systematic errors introduced by shallow imaging observations since the half - light radii are defined based on the observed apparent magnitudes . several of the galaxies in this sample have surface photometry reported in the literature . recent observations with partially overlapping samples include vcc 543 , 917 , 1036 , 1261 , and 1308 reported in @xcite ; vcc 965 , 1036 , 1122 , and 1308 reported in @xcite ; vcc 1036 and 1261 reported in @xcite ; and vcc 543 and 1743 reported in @xcite . in general , there is reasonable agreement between the literature values and those reported here , particularly given the variety of filters and telescopes used by these studies . however , there are notable differences between the values reported here and the magnitudes and effective radii listed in @xcite . with the exception of vcc 1261 , it is likely that many of these differences can be attributed to differing image depths and to incomplete surface photometry as a result of the restricted field - of - view of hst imaging observations . for vcc 1261 , the large discrepancy between the effective radius reported here and that reported in @xcite ( more than a factor of 2 ) can not be reconciled by mere observational differences ; the present value agrees with the surface photometry of vcc 1261 reported in @xcite . to enable comparison with galaxies of other morphological types , the outer isophotes ( r @xmath14 10 ) of the b - band surface brightness profiles were fit to an exponential model : @xmath15 where @xmath16 is the observed surface brightness at semi - major axis @xmath17 , @xmath18 is the extrapolated central surface brightness , and @xmath1 is the exponential scale length . the face on central surface brightness , @xmath19 , was calculated by applying a galactic extinction correction and a line of sight correction of @xmath20 , where @xmath21 is the inclination derived from the observed axial ratios and a thin disk approximation . while an exponential distribution does not necessarily imply a disk - like stellar distribution , the simplicity of this functional form permits a robust analysis of the light distribution of the outer regions of a galaxy , regardless of the actual shape of the galaxy ( disk or spheroid ) . the exponential fits enable a direct comparison between the structural parameters of de and di galaxies ( figure [ fig : struct ] ) . the top panel of figure [ fig : struct ] shows the scale length as a function of absolute magnitude while the bottom panel shows the face on central surface brightness as a function of absolute magnitude . for comparison , the structural parameters for dwarf irregular galaxies from @xcite are also shown . despite the fact that this comparison sample may not be ideal , since the dis were selected to be isolated galaxies and the des are cluster members , these figures confirm that the structural parameters of dis and des are quite similar ( as first discussed in * ? ? ? in particular , both the di and de samples contain low surface brightness galaxies with typical central surface brightnesses between 22 to 24 mag arcsec@xmath6 and scale lengths of only a few kpc . while the parameters for an exponential fit are sufficient to represent the stellar distribution in the outer galaxy , a more generalized surface brightness profile , such as the srsic function @xcite , can provide additional insight into the stellar distribution . the surface brightness profiles were fit to a srsic function of the form : i = i@xmath22 e@xmath23 . in magnitude space , this corresponds to @xmath24 where @xmath25 is the central surface brightness , r@xmath26 is a size parameter , and @xmath27 is a shape parameter . in this form , n=0.25 corresponds to an r@xmath28 law distribution , while n=1 corresponds to an exponential ( disk ) distribution . the surface brightness profiles were fit for geometric mean radii between 1 and the radius at which the r - band surface brightness reached 25 mag arcsec@xmath6 . since the derived parameters were similar regardless of filter choice , only the r - band fits are listed in table [ tab : results2 ] and shown in figure [ fig : surf ] . in general , the surface brightness profiles were well fit by the srsic function , with shape parameters , n , between 0.5 and 1 . however , the srsic parameterization proved to be quite sensitive to choice of fitting radii . thus , while errors are listed in table [ tab : results2 ] , equally valid solutions can be found with parameterizations that are outside of the range apparently permitted by these formal errors . several of the galaxies in this sample have srsic parameters listed in the literature . the present results agree with the literature values for vcc 1036 and 1261 @xcite , vcc 1308 @xcite , and vcc 1743 @xcite . however , when overlayed with the observed surface brightness profile , the srsic parameters for vcc 543 , 917 , and 1035 from @xcite yielded good fits only to the inner regions of these galaxies ( as mentioned above , their values for vcc 1261 do not match the surface brightness profiles obtained here ) . similarly , the parameters listed in @xcite for vcc 543 are only applicable to the inner profile . the disagreement between the present study and the literature is not surprising since srsic fits are sensitive to the choice of fitting radii , and the previous analysis often restricted the fits to the inner 20 - 30 . the results of srsic fits for dwarf elliptical galaxies in the virgo cluster are shown in figure [ fig : sersic ] . figure [ fig : sersic ] includes all de galaxies fainter than m@xmath29 of -17.5 with published kinematic data ( e.g. , the sample includes galaxies compiled from * ? ? ? * ; * ? ? ? structural parameters for many of the galaxies in @xcite and @xcite were compiled from @xcite . for those galaxies in @xcite without published b - band effective surface brightnesses , the published v - band effective surface brightness was converted to an approximate b - band value by assuming that the typical ( b - v ) color of a virgo de is 0.77 ( see section 4.2 ) . while figure [ fig : sersic ] indicates that there is a general trend of larger shape parameters ( n ) for fainter dwarf ellipticals , and for brighter central surface brightnesses for brighter galaxies , the rotating and non - rotating dwarf elliptical galaxies can not be distinguished in any of these figures . in fact , the kinematic samples are well mixed in terms of shape , surface brightness , and size . in summary , there appear to be no strong morphological differences between rotating and non - rotating dwarf elliptical galaxies in the virgo cluster . underlying disky and boxy isophotes are seen in both kinematic samples ( see also * ? ? ? the structural parameters are similar as well . in the next section , we investigate whether there are stellar population differences between the rotating and non - rotating dwarf elliptical galaxies . as illustrated in figure [ fig : surf ] , many of the dwarf elliptical galaxies in this sample have a slight color gradient . however , this trend does not appear to be correlated with the presence of a nucleated region ; several of the nucleated des have no color gradients while several of the non - nucleated des do have a color gradient . color gradients are commonly seen in des , usually in the sense that the outer regions are redder than the inner regions @xcite . in this sample , vcc 965 , 1743 , 1857 , and 2019 have color gradients in this expected direction . however , approximately one half of the current sample have color gradients in the opposite sense ; the inner regions are redder than the outer regions of vcc 178 , 543 , 917 , 990 , 1036 , 1122 , 2050 . a color gradient implies differing star formation histories , or differing metal content , in the inner and outer regions of the galaxy . it is likely that both of these effects are relevant to the interpretation of observed colors and stellar populations of virgo des . for example , a galaxy may develop a color gradient if the outer gas is preferentially stripped off as it falls into the virgo cluster . in this scenario , the inner region of the galaxy may continue to have active star formation while the outer regions are reduced to an aging stellar population ( redder colors ) . such a scenario easily explains the observed color gradients in most dwarf elliptical galaxies ( redder in the outskirts ) . alternatively , a color gradient in the opposite sense can be created if the inner region retains a larger fraction of its enriched material ; if there is a metallicity gradient , the more metal - rich stars ( inner regions ) will be redder than the metal - poor stars ( outskirts ) . the relative importance of both of these effects will depend on the galaxy s detailed star formation history and the interface of the galaxy with the intracluster medium . as shown in figure [ fig : rot ] , the presence of a color gradient does not correlate with the kinematic properties of the des . galaxies with a blue core ( vcc 965 and 2019 ) or red core ( vcc 543 , 917 , 990 , 1036 , and 2050 ) are equally likely to be rotation dominated as galaxies with no strong color gradient . also shown in figure [ fig : rot ] is the luminosity - line width correlation for spiral galaxies @xcite and for field dwarf irregular galaxies compiled in @xcite . as discussed in @xcite , the observed de rotation velocities are lower limits due to observational constraints ; reasonable extrapolation of the observed rotation curves result in linewidths comparable to galaxies that are 1 - 2 magnitudes brighter than the current de population . this fading estimate is in reasonable agreement with that expected if des are evolved dis ( e.g. , * ? ? ? * ) . nonetheless , it is quite remarkable that the observed ( un - extrapolated ) rotation velocities show excellent agreement with the bright galaxy tully - fisher relation . the presence of a color gradient complicates the definition of a global galaxy color . to avoid systematic trends due to nuclear contamination of the colors , the colors tabulated in table [ tab : results ] are measured within the b - band 25 mag arcsec@xmath6 isophotal aperture ; this aperture color was selected to be representative of the true color of the galaxies since the aperture and surface colors are quite similar at this radius ( see figure [ fig : surf ] ) . table [ tab : results2 ] lists the observed optical colors corrected for galactic extinction . the optical colors are remarkably similar for all of the des in this sample . the median colors are 0.24 @xmath0 0.03 in ( u b ) , 0.77 @xmath0 0.02 in ( b v ) , 1.26 @xmath0 0.05 in ( b r ) , 0.48 @xmath0 0.01 in ( v r ) , and 1.02 @xmath0 0.03 in ( v i ) . observations of low luminosity elliptical galaxies yield similar colors for galaxies in the same absolute magnitude range @xcite . the optical colors are shown in figure [ fig : evol ] ; as expected , the dwarf elliptical galaxies are red systems in all color combinations . for comparison , the evolutionary tracks for 3 possible star formation histories are shown in figure [ fig : evol ] . the evolutionary tracks are obtained from the simple stellar population models of @xcite assuming a mean metallicity of 1/5 solar ; the low metallicity stellar population models were selected for this figure because low luminosity galaxies are usually low metallicity systems ( e.g. , * ? ? ? * ; * ? ? ? the @xcite galaxy evolution code is an improved version of the composite stellar population models originally described in @xcite . as in @xcite , the mono metallicity composite stellar populations are built through the synthesis of stellar evolutionary tracks ( e.g. , * ? ? ? the version of the galaxy evolution code run here includes options for sub and super solar metallicity stellar populations , in addition to allowing the user to vary the star formation history and the initial mass function . for simplicity , a @xcite imf was adopted for all of the models considered here ; a top or bottom heavy imf will result in minor variations in the evolutionary tracks , but relative conclusions should be robust with regard to the choice of imf . the three fiducial star formation histories shown in figure [ fig : evol ] include ( 1 ) a simple burst of star formation that lasts 10 myr , ( 2 ) an extended burst of star formation with an e - folding time of 1 gyr ( @xmath30 ) , and ( 3 ) a constant star formation rate model . as expected , the constant star formation rate model can be ruled out for the virgo des since these galaxies have no current star formation activity ; this model is appropriate for dwarf irregular galaxies which have quasi - continuous star formation activity @xcite . the observed colors are consistent with both fiducial star burst models ( short and extended starbursts ) . in both cases , the observed colors are indicative of an older stellar population with an age of approximately 6 gyr . while there is some ambiguity in this age estimate due both to the well known age - metallicity degeneracy and to possible different star formation histories , it is clear that the observed colors are indicative of a stellar population which has had little - to - no star formation activity within the last few gyr . thus , while the observed broad band colors do not permit a detailed analysis of the past star formation activity in these galaxies , they do indicate that none of the galaxies in this sample are _ recently _ stripped dwarf irregular galaxies . further , the small range of observed colors for the galaxies in this sample indicates that the dominant stellar population is quite similar for both the rotating and non - rotating dwarf elliptical galaxies . as illustrated in figure [ fig : colmag ] , much of the scatter in the observed colors can be attributed to the well known color - magnitude relation for dwarf elliptical galaxies ( e.g. , * ? ? ? in general , the more luminous galaxies in this sample are slightly redder , as would be expected if the more luminous galaxies in the sample are also more metal - rich ( e.g. , * ? ? ? the two significant outliers to this relation are vcc 1857 and vcc 1261 , both of which are bluer than one might expect for their absolute magnitude . interestingly , neither of these galaxies have a significant color gradient , so their anomalous colors can not be explained by nuclear contamination . we note also that the present results for vcc 1261 are consistent with the b - r color reported in @xcite for this galaxy . it is suggestive that vcc 1857 is not only bluer than expected , but also has slightly different structural parameters than the other virgo des ; this galaxy is the best candidate in this sample for a recently stripped di . nonetheless , its optical colors are still consistent with an older stellar population , indicating that the last star formation activity occurred several gyr ago . unfortunately , the kinematic observations of vcc 1857 were inconclusive ; while the derived rotation curve indicates that vcc 1857 may have a significant rotational component , the rotation curve was poorly constrained because of the galaxy s low surface brightness nature . in contrast , the other anomalous galaxy , vcc 1261 , is the one galaxy in the sample with clear absence of a rotational component along its major axis . using these two outliers as indicative examples , the relative colors of the virgo des do not appear to be related to their kinematic signatures . rather , the small range of observed colors for the galaxies in this sample indicates that the ages of the dominant stellar population are approximately the same for both rotating and non - rotating dwarf elliptical galaxies . we note , however , that broad band colors do not permit a detailed analysis of the past star formation activity in these galaxies since the observed colors are consistent with both a short burst and with a more extended star formation history . we explore additional constraints on the past star formation activity by examining the chemical signatures of the dominant stellar population in the next section . the results of the spectral analysis are shown in figure [ fig : alpha ] . the expected lick indices for three possible [ @xmath1/fe ] ratios are shown based on the models of @xcite . unlike giant elliptical galaxies , the majority of dwarf elliptical galaxies in virgo appear to have solar or sub - solar [ @xmath1/fe ] ratios . similar results are found by @xcite and @xcite for their samples of virgo dwarf elliptical galaxies ; in addition , similar [ @xmath1/fe ] ratios are seen in stellar abundances of nearby dwarf spheroidals ( e.g. , * ? ? ? * ; * ? ? ? * ) . since type ii supernova produce the majority of @xmath1rich elements , the observed [ @xmath1/fe ] ratios trace the timescale of star formation activity in each galaxy @xcite . in particular , super - solar [ @xmath1/fe ] indicates rapid enrichment from type ii supernova and thus implies that the galaxy has undergone a short burst of star formation activity ; most giant elliptical galaxies have super - solar [ @xmath1/fe ] ( e.g. , * ? ? ? in contrast , the solar and sub - solar abundance ratios for des indicate slow chemical enrichment , or a more quiescent star formation history . thus , the dominant stellar populations in virgo des may not be post - burst populations . rather , previous star formation activity appears to have occurred on more extended timescales in the low mass galaxies ; these and similar results ( e.g. , * ? ? ? * ; * ? ? ? * ) rule out various theoretical models for de formation , including the possibility that des and dsphs form all stars prior to reionization ( e.g. , * ? ? ? the observed [ @xmath1/fe ] ratios are consistent with the idea that virgo des are formed through the stripping of dis ; if star formation activity ceases when the ism is removed from the low mass system , the galaxy will evolve into a quiescent , red galaxy with structural and kinematic parameters similar to the progenitor di , and with a chemical enrichment history representative of a more continuous star formation history . comparison of the h@xmath7 and mgb indices provides an indication of the age of the dominant stellar population in the des ( figure [ fig : metals ] ) . also shown in figure [ fig : metals ] are the stellar population models of @xcite for solar [ @xmath1/fe ] . as expected from the optical colors ( section 4 ) , the stellar populations are both metal poor and evolved . based on these observations and stellar population models , virgo des have typical ages of 5 - 7 gyr and metal abundances between 1/20 and 1/3 of solar . similar results for virgo des were also obtained by @xcite . the derived age and metallicity estimates for these galaxies are also in remarkably good agreement with those obtained from the broad band optical images . as found with the optical colors , there is no clear separation of the kinematic samples in either age or metallicity of the dominant stellar population . thus , both stellar population models and spectral synthesis indicate that virgo des contain evolved low metallicity stellar populations with typical ages of a several gyr and solar or sub - solar [ @xmath1/fe ] . the derived structural parameters and metal enrichment histories indicate that the virgo des are similar in nature to field dwarf irregular galaxies with the exception of the age of the dominant stellar population . thus , the current observations are consistent with the idea that some cluster des are the remnants of dis which have been stripped of their ism . taken at face value , the derived mean stellar population ages indicate that the last major star formation episode in the virgo des occurred 5 - 7 gyr ago , or at @xmath31 . however , due to the inherent ambiguities in stellar population models , it is difficult to determine exactly how long it has been since star formation occurred in these galaxies . for example , the observed colors are not only consistent with the two burst models discussed above ( extended starburst and short burst ) , but also with a galaxy which has had a constant star formation rate which is abruptly truncated 3 gyr ago ( @xmath32 ) . unfortunately , the virgo dwarfs are too distant to produce detailed star formation histories as is possible within the local group ( e.g. , * ? ? ? nonetheless , all of these models indicate that the virgo des have not had substantial star formation activity within the last few gyr . furthermore , the observed [ @xmath1/fe ] ratios are solar or sub - solar , indicating that the stellar mass of the cluster des has been built gradually , rather than through a major burst of star formation . thus , not only do low mass cluster galaxies have different structural properties than their giant cousins , their stellar populations and star formation histories appear to be substantially different . while both giant and dwarf elliptical galaxies appear to have strong environmental dependences on their formation and evolution , it is likely that low mass cluster galaxies follow different evolutionary paths than their giant counterparts . for example , interactions with the cluster potential and the icm may be the key to formation of des , while galaxy - galaxy interactions may play a major role in the evolution of giant ellipticals . since the early recognition of similar structural parameters between de and di galaxies ( e.g. , * ? ? ? * ; * ? ? ? * ) , it has often been suggested that des could simply be dis stripped of their gas and left to fade . the extensive study of the virgo cluster resulting in the vcc ( c.f . * ; * ? ? ? * ; * ? ? ? * ) confirmed the strong overlap between these two populations , but led @xcite to conclude that the des are not formed from dis due to the apparent lack of a bright progenitor di population and to the presence of bright nuclei in a large fraction of des . while @xcite was careful to note that his arguments applied only to the brightest des , he argued that if one could not explain the entire de population with stripping then it did not make sense to invoke this process since it is inelegant to explain the well defined sequence in l , @xmath1 , @xmath33 space with more than one process . in @xcite , we proposed that virgo des are formed through multiple processes . first , it is likely that the most massive galaxies near the core of the cluster formed with a coterie of satellite galaxies , which soon became de galaxies . however , we also know that the virgo cluster is still forming today , and that field galaxies and groups of galaxies are falling into the cluster @xcite . if the groups of galaxies that later fell into the virgo cluster were similar to our local group , then they likely consisted of roughly equal numbers of de and di galaxies before entering the cluster environment . in addition , the infalling di galaxies are likely to be affected by the cluster environment and may lose their ism through ram pressure stripping or through interactions with other galaxies . it therefore seems unavoidable that there are at least three different origins for the virgo des ( _ in situ _ formation near massive elliptical galaxies , infalling des that were once part of local group analogs , and converted infalling dis ) . thus , the narrow locus formed by des in structural parameter space may better be interpreted as the result of an inevitable structural form as opposed to a uniform process of creation . in other words , the uniformity of de / di properties may be due more to the uniformity of their dark matter halos @xcite , rather than a single formation mechanism . additional arguments have been made against the stripping of infalling dis to form the virgo de population . @xcite reason that the scarcity of dwarf galaxies with blue colors and low hi content argues against extensive stripping of di systems in the last @xmath34gyr . however , recent observations have identified several likely candidates as recently stripped dis . the best case for stripping of a dwarf galaxy in process is that of the hi cloud between ngc 4472 and ugc 7636 discovered by @xcite . subsequent observations of this system have shown that the morphologies of the optical galaxy and the hi cloud , although separated by 15 kpc , are `` remarkably similar '' @xcite and that the ism abundance of the cloud agrees with that expected for the luminosity of ugc 7636 @xcite - essentially clinching the proposal that this is a recent stripping event . @xcite found blue star clusters with ages up to 10@xmath35 yrs associated with both the optical galaxy and the hi cloud and this provides a minimum time since the stripping event . in addition to ugc 7636 , ic 3475 has been identified as a recently stripped irregular galaxy @xcite , ic 3365 appears to be a di caught in the act of being stripped @xcite , and vcc 882 is a nucleated de that from a long trail of dust appears to have recently been stripped of its ism @xcite . anecdotal evidence allows us to be sure that the process of ram pressure stripping is viable , but a statistical approach is required to address the point made by @xcite . from a large sample of hi observations of virgo dis , @xcite concluded that the hi depletion of dis is only moderate . however , this was based on comparing m@xmath36/l ratios of dis to those of spirals . the means of 0.2 for the bright virgo dis and 0.5 for the faint virgo dis are both significantly smaller than the typical m@xmath36/l of 1 for field dis ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) . from this , one can conclude that a significant fraction of virgo dis are showing the effects of the cluster environment . @xcite have revisited this question , and a comparison of virgo dis with field dis show the majority of virgo dis in their sample have hi deficiencies . finally , from our color and spectroscopic dating of virgo des , it is clear that the majority have been des for several gyr . if most of the conversions happened long ago , a low transformation rate today might only imply that the process is slowing down , rather than excluding it altogether . in fact , the argument is based on the assumption of a constant rate of creation of des over the hubble time . given the stochastic nature of infall , a variable rate of infall would also be consistent with a low rate of de creation in the present epoch and a large population of des formed from stripped dis . arguments against conversion from dis to des by ram pressure stripping have also been made on theoretical grounds . @xcite revisit this question and offer a rough calculation of the time scale for ram pressure stripping and note that it is unreasonably long for distances greater than 300 kpc from m87 ( which excludes only the very core of the virgo cluster ) . they conclude that `` stripping was probably not the dominant gas - removal mechanism '' in the virgo cluster . however , they note that a more realistic treatment of the ism in a di might yield significantly shorter gas removal times . the empirical counter example to this theoretical calculation comes in the form of hi observations of the spiral galaxy ngc 4522 @xcite . they show this 0.5 l@xmath37 galaxy to be undergoing significant stripping at a projected distance of @xmath34 800 kpc from m 87 . the stripping of the hi is complete to significantly higher mass surface densities than seen in most dwarf galaxies , and thus any gas rich dwarf in this environment is likely to be stripped . binggeli s ( 1985 ) argument concerning the lack of a progenitor population for nucleated des remains as a possible constraint on the number of virgo des which are created as a result of stripping of an infalling di . unfortunately , without a consensus view on the formation of the nuclei of nucleated des ( see , e.g. , * ? ? ? * ; * ? ? ? * ) , it is difficult to judge the strength of this constraint . the fact that the nucleated des are more strongly clustered than the non - nucleated des ( e.g. , * ? ? ? * ) may be an indication that they are part of an _ in situ _ population of galaxies formed as satellites of the original cluster galaxies . based on specific globular cluster frequencies , @xcite propose that the non - nucleated des are the more likely des to form as a result of stripping . however , in this regard , note that 3 of the 5 galaxies in the @xcite sample with strong rotational support ( v/@xmath38 @xmath14 1 ) are nucleated des . alternatively , it is possible that nucleated des could be formed from dis with large central starbursts @xcite . further study of their stellar populations and structure are needed to understand fully the origin of the nuclei and thus to determine if stripped dis can evolve into nucleated dwarf ellipticals . finally , a possible concern for di to de evolution through ram pressure stripping is a possible disparity in baryon - to - dark matter ratios in these low mass galaxies . while the characteristics of the dark matter halos of the local group dsph galaxies are currently a matter of debate ( cf . * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , it appears that dsph may have lower dark matter - to - luminosity ratios than similar luminosity dwarf irregular galaxies ( e.g. , * ? ? ? * ) . however , such results should be viewed with some caution since dark matter halos are measured in different ways for rotating and non - rotating systems @xcite . thus , at this time , possible discrepancies in the masses of dark matter halos and baryon - to - dark matter ratios of non - rotating virgo des are not sufficiently robust to exclude ram pressure stripping as a viable evolutionary pathway from di to de . further , as shown in figure [ fig : rot ] , the rotating virgo des have maximum rotation velocities comparable to similar luminosity dis , suggesting that at least this subset of de galaxies may have similar baryon - to - dark matter ratios as field dis . in summary , after reviewing the historical comparisons of the de and di populations in virgo which have been used to rule out the creation of the virgo des by infalling dis , we find no observational evidence which rules out the creation of a majority of the virgo des by the stripping of gas from dis . we note , however , that if ram pressure stripping is a common mechanism for transforming dis to des , there remains a fundamental difficulty of identifying the population of recently stripped dis . in all likelihood , the recently stripped dis are not ( yet ) classified as des ( ugc 7636 is di ) . in fact , any young stellar population must have already faded by the time a galaxy is classified as a de since the primary morphological criterion is that des have smooth and regular isophotes . thus , it should not be surprising that all the des in this sample have similar optical colors . by the time a galaxy is classified as a de , its stellar population has aged sufficiently ( @xmath342 gyr ) that the observed colors are red and it is no longer possible to determine a precise time since the last star formation episode . in analogy with the local group dsph , however , it is likely that the detailed star formation histories of the virgo des ( as revealed by color magnitude diagrams ) will differ substantially even though the integrated colors are similar . the optical images and spectroscopy presented here indicate that there are no obvious differences between the stellar populations of rotating and non - rotating dwarf elliptical galaxies . in @xcite , we proposed the possibility that non - rotating des may have entered the cluster environment as des , while rotating des may have been converted from dis while entering the cluster environment . if this scenario is correct , then one would expect , on average , that the non - rotating des would have older stellar populations . unfortunately , the age resolution afforded by the present observations and the sample size are insufficient to provide a clear test . there is the additional complication that a rotationally supported de which is a result of a ram pressure stripping event may follow a trajectory in which a close encounter significantly alters its kinematics ( e.g. , * ? ? ? thus , even if the rotating and non - rotating des are entering the cluster environment in similar distributions , the distributions of the observed rotating and non - rotating des could be significantly different . the best way to sort out this question observationally may be to study the population of recently stripped galaxies and those galaxies being actively stripped today . an unbiased hi survey of the virgo cluster environment may reveal a large number of hi clouds similar to the one associated with ugc 7636 . if so , then the distribution ( in both real space and velocity space ) of these objects and their association with rotating or non - rotating galaxies will offer significant constraints on all scenarios accounting for the de population in virgo . the new seven feed l - band array ( alfa ) at arecibo observatory is well suited to such a blind neutral hydrogen survey . with a relatively small beam ( 3 arcmin ) , alfa observations of the virgo cluster will have sufficient spatial resolution and sensitivity to identify recently stripped hi clouds and gas poor dwarf irregular galaxies within this high density region ; such observations will improve significantly our understanding of the population and location of recently stripped dis and their connection to the dwarf elliptical population . the abundant de population in the local universe implies the existence of a large population of star forming dwarf galaxies at intermediate redshifts , corresponding to progenitors of the local de population . the detectability and the nature of this population depend entirely on the evolutionary histories of des . if des build stellar mass rapidly , they could appear as luminous objects ( @xmath39 ) during their formation and then fade by @xmath40 magnitudes to the present epoch . such `` starburst '' scenarios are favored to explain the rapidly evolving population of faint blue galaxies at intermediate redshift ( e.g. * ? ? ? however , the observed [ @xmath1/fe ] indicates that des build their stellar mass more slowly . if most des form from dis , with their roughly constant star formation histories punctuated by , at most , moderate bursts of star formation , the forming des may be only 1 - 2 magnitudes brighter than those at the present epoch . thus , with expected magnitudes m@xmath41 ( m@xmath42 at @xmath43 ) , the population of de progenitors will be rare in most extragalactic surveys . we caution , however , that integrated colors and elemental abundance ratios provide only modest constraints on the detailed star formation history of a galaxy since a modest star burst , or period of quiescence , can be masked by subsequent star formation episodes . thus , while the present observations are suggestive that the di to de transition does not require a starburst episode , observations of the resolved stellar populations will be necessary to determine the detailed star formation history of dwarf elliptical galaxies . such observations are beyond the capabilities of current optical telescopes , but the next generation of ground based telescopes may make this a realistic goal even at the distance of the virgo cluster . such observations are crucial for our understanding and interpretation of galaxy counts at moderate to high redshift . we present optical images and spectral synthesis of 16 dwarf elliptical galaxies in the virgo cluster . the major results of these observations are as follows : \(1 ) there appear to be no major morphological differences between rotating and non - rotating dwarf elliptical galaxies . underlying boxy and disky isophotes are seen in both kinematic samples . as first discussed by @xcite and @xcite , the cluster des have similar structural parameters as field dis . \(2 ) the global optical colors are red , with median values for the sample of 0.24 @xmath0 0.03 in ( u b ) , 0.77 @xmath0 0.02 in ( b v ) , 1.26 @xmath0 0.05 in ( b r ) , 0.48 @xmath0 0.01 in ( v r ) , and 1.02 @xmath0 0.03 in ( v i ) . \(3 ) stellar population models and spectral synthesis indicate that the dominant stellar populations have ages of approximately 5 - 7 gyr . based on lick indices and simple stellar population models , the derived [ @xmath1/fe ] ratios are sub - solar to solar , indicating a more gradual chemical enrichment history for des as compared to giant elliptical galaxies in the virgo cluster . \(4 ) we argue that it is likely that several different physical mechanisms played a significant role in the production of the virgo cluster de galaxies including _ in situ _ formation , infall of des that were once part of local group analogs , and the conversion of infalling dis . the present observations support the hypothesis that a large fraction of the virgo cluster des are formed by ram pressure stripping of gas from infalling dis . this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration . we thank the vatt time allocation committee for awarding tbs time to complete this project . lvz acknowledges partial support by indiana university . support for ejb was provided by nasa through hubble fellowship grant # hst - hf-01135.01 awarded by the space telescope science institute , which is operated by the association of universities for research in astronomy , inc . , for nasa , under contract nas 5 - 26555 . eds is grateful for partial support from a nasa ltsarp grant no . nag5 - 9221 and the university of minnesota . babul , a. , & rees , m. j. 1992 , , 255 , 346 barazza , f. d. , binggeli , b. , & jerjen , h. 2003 , , 407 , 121 barkana , r. , & loeb , a. 1999 , , 523 , 54 binggeli , b. 1985 , in star - forming dwarf galaxies and related objects . kunth , d. , thuan , t. x. , tran thanh van , j. ( eds ) ditions frontires , gif sur yvette , p.53 binggeli , b. , barazza , f. , & jerjen , h. 2000 , , 359 , 447 binggeli , b. , & cameron , l. m. 1991 , , 252 , 27 binggeli , b. , sandage , a. , & tammann , g. a. 1985 , , 90 , 1681 binggeli , b. , tammann , g. a. , & sandage , a. 1987 , , 94 , 251 binggeli , b. , tarenghi , m. , & sandage , a. 1990 , , 228 , 42 bothun , g. d. , mould , j. r. , caldwell , n. , & macgillivray , h. t. 1986 , , 92 , 1007 bruzual a. , g. , & charlot , s. 1993 , , 405 , 538 bruzual , g. , & charlot , s. 2003 , , 344 , 1000 caldwell , n. 1983 , , 88 , 804 caldwell , n. , & bothun , g. d. 1987 , , 94 , 1126 davies , j. i. , & phillipps , s. 1988 , , 233 , 553 dekel , a. , & silk , j. 1986 , , 303 , 39 dressler , a. 1980 , , 236 , 351 durrell , p. r. 1997 , , 113 , 531 elmegreen , d. m. , elmegreen , b. g. , chromey , f. r. , & fine , m. s. 2000 , , 120 , 733 ferguson , h. c. , & binggeli , b. 1994 , , 6 , 67 fisher , j. r. & tully , r. b. 1975 , , 44 , 151 gallagher , j. s. , & hunter , d. a. 1989 , , 98 , 806 gallagher , j. s. , iii . & wyse , r. f. g. 1994 , , 106 , 1225 geha , m. , guhathakurta , p. , & van der marel , r. p. 2003 , , 126 , 1794 gilmore , g. , & wyse , r. f. g. 1991 , , 367 , l55 hoffman , g. l. , helou , g. , salpeter , e. e. , & sandage , a. 1985 , , 289 , l15 ichikawa , s .- wakamatsu , k .- okamura , s. 1986 , , 60 , 475 impey , c. , bothun , g. , & malin , d. 1988 , , 330 , 634 jerjen , h. , binggeli , b. , & barazza , f.d . 2004 , , 127 , 771 jerjen , h. , binggeli , b. , & freeman , k. c. 2000 , , 119 , 593 kelson , d. d. et al . 2000 , , 529 , 768 kenney , j. d. p. , van gorkom , j. h. , & vollmer , b. 2004 , , 127 , 3361 kleyna , j. , wilkinson , m. i. , evans , n. w. , gilmore , g. , & frayn , c. 2002 , , 330 , 792 kormendy , j. 1985 , , 295 , 73 landolt , a. u. 1992 , , 104 , 340 lee , m. g. , kim , e. , & geisler , d. 1997 , , 114 , 1824 lee , h. , mccall , m. l. , & richer , m. g. 2003 , , 125 , 2975 lee , h. , richer , m. g. , & mccall , m. l. 2000 , , 530 , l17 lejeune , th . , cuisinier , f. , & buser , r. 1997 , , 125 , 229 lin , d. n. c. , & faber , s. m. 1983 , , 266 , l21 mateo , m. l. 1998 , , 36 , 435 mayer , l. , governato , f. , colpi , m. , moore , b. , quinn , t. , wadsley , j. , stadel , j. , & lake , g. 2001a , , 547 , l123 mayer , l. , governato , f. , colpi , m. , moore , b. , quinn , t. , wadsley , j. , stadel , j. , & lake , g. 2001b , , 559 , 754 mcnamara , b. r. , sancisi , r. , henning , p. a. , & junor , w. 1994 , , 108 , 844 miller , b. w. , lotz , j. m. , ferguson , h. c. , stiavelli , m. , & whitmore , b. c. 1998 , , 508 , l133 navarro , j. f. , frenk , c. s. , & white , s. d. m. 1997 , , 490 , 493 neilsen , e. h. jr . , & tsvetanov , z. i. 2000 , , 536 , 155 oh , k. s. & lin , d. n. c. 2000 , , 543 , 620 oke , j. b. 1990 , , 99 , 1621 palma , c. , majewski , s. r. , siegel , m. h. , patterson , r. j. , ostheimer , j. c. , & link , r. 2003 , , 125 , 1352 pedraz , s. , gorgas , j. , cardiel , n. , snchez - blzquez , p. , & guzmn , r. 2002 , , 332 , l59 piatek , s. , pryor , c. , armandroff , t. e. , & olszewski , e. w. 2002 , , 123 , 2511 pierini , d. 2002 , , 330 , 997 postman , m. , & geller , m. j. 1984 , , 281 , 95 prugniel , ph . , bica , e. , klotz , a. , & alloin , d. 1993 , , 98 , 229 richer , m. g. , & mccall , m. l. 1995 , , 445 , 642 salpeter , e. e. 1955 , , 121 , 161 sancisi , r. , thonnard , n. , & ekers , r. d. 1987 , , 315 , l39 sandage , a. , binggeli , b. , & tammann , g. a. 1985 , , 90 , 1759 schlegel , d. j. , finkbeiner , d. p. , & davis , m. 1998 , , 500 , 525 srsic , j .- l . 1968 , atlas de galaxias astrales , observatorio astronomico , cordoba shetrone , m. d. , ct , p. , sargent , w. l. w. 2001 , apj , 548 , 592 shetrone , m. , venn , k. a. , tolstoy , e. , primas , f. , hill , v. , & kaufer , a. 2003 , aj , 125 , 684 skillman , e. d. 1996 , asp conf . ser . 106 : the minnesota lectures on extragalactic neutral hydrogen , 208 skillman , e. d. , bothun , g. d. , murray , m. a. , & warmels , r. h. 1987 , , 185 , 61 skillman , e. d. , kennicutt , r. c. , & hodge , p. w. 1989 , , 347 , 875 skillman , e. d. , tolstoy , e. , cole , a. a. , dolphin , a. e. , saha , a. , gallagher , j. s. , dohm - palmer , r. c. , & mateo , m. 2003 , , 596 , 253 stoehr , f. , white , s. d. m. , tormen , g. , & springel , v. 2002 , , 335 , l84 thomas , d. , bender , r. , hopp , u. , maraston , c. , & greggio , l. 2003a , ap&ss , 284 , 599 thomas , d. , maraston , c. , & bender , r. 2003b , , 339 , 897 trager , s. c. , faber , s. m. , worthey , g. , & gonzlez , j. j. 2000 , , 119 , 1645 tully , r. b. , & pierce , m. j. 2000 , , 533 , 744 tully , r. b. , & shaya , e. j. 1984 , , 281 , 31 vader , j. p. , vigroux , l. , lachize - rey , m. , & souviron , j. 1988 , , 203 , 217 van zee , l. 2000 , 119 , 2757 van zee , l. 2001 , 121 , 2003 van zee , l. , haynes , m. p. , & giovanelli , r. 1995 , , 109 , 990 van zee , l. , skillman , e. d. , & haynes , m. p. 2004 , , 128 , 121 vigroux , l. , thuan , t. x. , vader , j. p. , & lachieze - rey , m. 1986 , , 91 , 70 west , m. j. , & blakeslee , j. p. 2000 , , 543 , l27 worthey , g. , & ottaviani , d. l. 1997 , , 111 , 377 young , c. k. , & currie , m. j. 1995 , , 273 , 1141

the fate of ionizing radiation in starburst galaxies is key to understanding cosmic reionization . however , the galactic parameters on which the escape fraction of ionizing radiation depend are not well understood . ionization - parameter mapping provides a simple , yet effective , way to study the radiative transfer in starburst galaxies . we obtain emission - line ratio maps of [ siii]/[sii ] for six , nearby , dwarf starbursts : ngc 178 , ngc 1482 , ngc 1705 , ngc 3125 , ngc 7126 , and he 2 - 10 . the narrow - band images are obtained with the maryland - magellan tunable filter at las campanas observatory . using these data , we previously reported the discovery of an optically thin ionization cone in ngc 5253 , and here we also discover a similar ionization cone in ngc 3125 . this latter cone has an opening angle of ( 0.4 ster ) , indicating that the passageways through which ionizing radiation may travel correspond to a small solid angle . additionally , there are three sample galaxies that have winds and/or superbubble activity , which should be conducive to escaping radiation , yet they are optically thick . these results support the scenario that an orientation bias limits our ability to directly detect escaping lyman continuum in many starburst galaxies . a comparison of the star - formation properties and histories of the optically thin and thick galaxies is consistent with the model that high escape fractions are limited to galaxies that are old enough ( 3 myr ) for mechanical feedback to have cleared optically thin passageways in the ism , but young enough ( 5 myr ) that the ionizing stars are still present .

1311.2227

between @xmath3 and @xmath4 , the universe experienced a change of state known as cosmic reionization ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) . the source for cosmic reionization is still unclear ; the primary contenders are star - forming galaxies and quasars . however , the space density of quasars peaks at @xmath5 and drops below the threshold needed to supply the required amount of ionizing radiation at @xmath6 @xcite . the massive stars found in high - redshift star - forming galaxies have the potential to provide enough ionizing radiation , but whether they contribute significantly to the intergalactic ionizing emissivity is not yet established ( e.g. , * ? ? ? in order to account for cosmic reionization , the mean fraction of ionizing radiation that escapes the galaxy population @xmath7 , needs to be at least 20% ( e.g. , * ? ? ? * ) . in theory , starburst galaxies should be porous to ionizing radiation ( e.g. , * ? ? ? * ; * ? ? ? stellar winds and supernovae , in conjunction with radiation pressure , will punch holes in the interstellar medium ( ism ) ( e.g. , * ? ? ? * ) , and thereby create low density passageways through which ionizing radiation may escape ( e.g. , * ? ? ? feedback from massive stars is strongest in galaxies with high star - formation rates , making starbursts attractive candidates for the source of cosmic reionization . the observational evidence for significant at high redshift is inconclusive . using a sample of 29 lyman break galaxies ( lbg ) , @xcite showed that at least some @xmath5 starbursts have large escape fractions . however , lyman continuum narrow - band imaging studies with larger galaxy samples found escaping lyman continuum in only 10% of the galaxies observed ( e.g. , * ? ? ? * ; * ? ? ? observations by @xcite , put the mean @xmath8 escape fraction at @xmath9 . is it important to note that @xcite recently illustrated that a majority of narrowband lyman continuum detections could actually be low @xmath10 interlopers , which would drive the observational even lower . at intermediate to low redshift , some evidence exists for significant from lgb analogs . a subset of these galaxies with extreme feedback have favorable conditions for escaping lyman continuum , such as low opacity in the ism @xcite . however , the fraction of galaxies with these conditions is still small , indicating a low mean ( e.g. , * ? ? ? looking at @xmath11 galaxies , the detection rate drops further , with only two examples of local starbursts that have measurable @xcite . most other studies find no direct evidence for escaping lyman continuum in local starbursts @xcite . in order to determine the mean galactic from observed escape fractions , we need to disentangle the effects of observational biases from trends with intrinsic galaxy properties . in the massive - star feedback - driven model , the escape fraction will depend on the star formation rate , the ism morphology and density @xcite , and distribution of star formation . however , the winds that clear the ism are not isotropic , which leads to a preferred direction for escape and finite opening angles @xcite . these effects create observational dependencies with the orientation of the host galaxy . finally , measuring the lyman continuum at any redshift is a challenge . most of the flux is absorbed by the intergalactic medium , thus the signal is low , which requires long integration times . for targets at higher redshift , low - redshift interlopers confuse the statistics @xcite . some studies measure the optical depth to uv lines in the ism in lieu of direct measurements of the lyman continuum . they obtain the escape fraction after making assumptions about the gas distribution and the relative optical depths in the lyman continuum and the observed ion ( e.g. , * ? ? ? * ) . in this study we use a method that is complementary to other established approaches for investigating the factors that affect the escape of ionizing radiation from starbursts . we map the ionization structure of extended ionized gas in six local starbursts using emission - line ratio images . spatial changes in emission - line ratios reveal the passage of ionizing radiation through the galaxy @xcite . thus , by studying the ionization structure of extended ionized gas in our sample , we can determine the optical depth of the galaxies and evaluate the likelihood of escaping lyman continuum . determining the fate of ionizing radiation in starburst galaxies is a challenging observational problem . here , we present an approach that uses the technique of ionization parameter mapping ( ipm ; * ? ? ? spatial changes to the ionization parameter , which evaluates the ionizing photon density relative to the gas density , trace the passage of ionizing radiation through the ism . in order to track these spatial changes , we create maps of the starbursts using the [ siii]/[sii ] line ratio , which acts as a proxy for the ionization parameter . optically thin and thick regions exhibit different ionization parameter morphologies . in regions that are optically thick , there is a transition from high to low ionization states at the interface of the ionized region and the neutral environment , whereas optically thin regions lack this transition zone @xcite . thus , by using ratio maps of two ions , one can distinguish between optically thick and thin regions @xcite . across an entire galaxy , one can thus use the ionization structure to evaluate whether and in what manner ionizing radiation escapes the galaxy ( e.g. , * ? ? ? * ; * ? ? ? this method has the advantage that it can be accomplished with narrow - band ground - based imaging , which makes it more observationally straightforward than other methods . the galaxies selected for this study are all nearby dwarf starbursts . they have star formation rates ( sfr ) of 0.055 /yr and stellar masses log(m@xmath12 ) @xmath13 . since massive star feedback is thought to play a key role in the escape of lyman continuum ( lyc ) photons @xcite , we select galaxies that have evidence in the literature for winds and expanding bubbles . table 1 shows the properties of our sample . we calculate the stellar masses ( column 2 ) using the @xmath14-band magnitudes from @xcite and the mass to light ( @xmath15/@xmath16 ) ratios following @xcite , which use the galaxies @xmath17 or @xmath18 colors to derive @xmath19 . we restrict this study to starbursts within 45 mpc , which corresponds to spatial scales of @xmath20 pc / arcsec , to ensure high enough spatial resolution to trace the ionization structure of the extended gas . we obtained narrow - band emission - line imaging in , [ sii]@xmath21 , and [ siii]@xmath22 on the nights of 2009 july 711 and 2010 february 1214 . the wavelengths quoted above are the rest frame wavelengths of the emission lines . we imaged the galaxies using the maryland - magellan tunable filter ( mmtf ; * ? ? ? * ) , which is mounted on the inamori - magellan areal camera and spectrograph ( imacs ) on the magellan baade telescope at las campanas observatory . for every emission - line image , we obtained paired continuum exposures of the galaxies . our observing log is presented in table [ t : obs ] . for each galaxy ( column 1 ) , we list the targeted emission line and central bandpass ( columns 2 , 3 ) , exposure times ( columns 4 ) , and the 3@xmath23 surface brightness limit in the reduced image ( columns 5 ) . we reduced the mmtf data using version 1.4 of the mmtf data reduction pipeline . this pipeline performs bias and flatfield corrections and then subtracts the sky using an azimuthal averaging . next , it corrects cosmic rays and bad ccd pixels before mosaicking the different ccd chips into one image . finally , it averages the individual emission - line and continuum images . we flux calibrated the images using standard stars ltt 1020 , 6248 and 7987 @xcite . we note that for the february 2010 observing run , our measured fluxes are a factor of 23 lower than the fluxes reported in the literature . a combination of shallower exposure times , combined with errors in the velocity offsets of our central bandpass , wind activity that shifts flux out the narrow mmtf bandpass , and uncertainty in the [ nii ] correction factors for the literature values , lead to this offset . however , a comparison with long - slit spectra confirms that the line ratios are sound . in order to cleanly subtract the continuum from the emission - line images , we first aligned the emission - line and continuum frames using _ iraf _ tasks wregister and imalign . next , we used psfmatch to convolve the image with the better seeing with the psf from the other . finally , we used several bright stars in the field and the task mscimatch to determine the proper scaling to match counts in the stars between the images . this last step was performed iteratively until we obtained a clean continuum subtraction . we present the final continuum - subtracted images and three - color composite images of the galaxies in our sample in section [ s : results ] . scattered light in the instrument causes an optical artifact that is not easily corrected for and that stretches across ccd chips 3 and 5 . for the most part , we avoid this issue by placing our targets on the other chips . however , for ngc 178 , most of our observations have the galaxy sitting on chips 3 or 5 . to mitigate the artifact , we mask the affected region in the flat field . while this increases the flux error on this galaxy , this effect is much less significant than the alteration in flux caused by the artifact on the flat . we also note that the data from the 2009 july observing run are bright time observations . as a result of scattered light in the f/2 camera and the mmtf instrument , these observations are challenging to reduce . the emission - line images of ngc 178 and ngc 7126 show many background sky features that are caused by these effects ( figure [ f : img_n178 ] and [ f : img_n7126 ] ) . l c c c c ngc 178 & 9.24 & 20.6 & @xmath24&(5),(8 ) + ngc 1482 & 10.56 & 22.6 & @xmath25 & ( 2 ) + ngc 1705 & 8.37 & 5.1 & @xmath26 & ( 3),(7 ) + ngc 3125 & 9.09 & 11.5 & @xmath27&(1),(7 ) + ngc 5253 & 9.05 & 3.8 & @xmath28 & ( 9),(7 ) + ngc 7126 & 10.44 & 45.5 & @xmath29 & ( 5),(8 ) + he 2 - 10 & 9.50 & 9 & @xmath30&(6),(4 ) l c c c c + + + ngc 178 & & 6594 & @xmath31 & @xmath32 + & [ sii]&6749 & @xmath33 & @xmath34 + & [ siii]&9113 & @xmath35 & @xmath36 + & & & @xmath37 & + ngc 7126&&6628 & @xmath31 & @xmath38 + & [ sii]&6783 & @xmath33 & @xmath39 + & [ siii]&9159 & @xmath33 & @xmath40 + + + + ngc 1705&&6572@xmath41 & @xmath42 & @xmath43 + & [ sii]&6730 & @xmath44 & @xmath45 + & [ siii]&9087 & @xmath31 & @xmath46 + he 2 - 10 & & 6582 & @xmath42 & @xmath47 + & [ sii]&6735 & @xmath31 & @xmath48 + & [ siii]&9095 & @xmath31 & @xmath49 + ngc 3125&&6587 & @xmath42 & @xmath50 + & [ sii]&6740 & @xmath31 & @xmath51 + & [ siii]&9102 & @xmath31 & @xmath52 + ngc 1482&&6602 & @xmath42 & @xmath53 + & [ sii]&6759 & @xmath42 & @xmath54 + & [ siii]&9121 & @xmath44 & @xmath55 in this section we present and discuss the ionization parameter maps of the galaxies individually . in section [ s : discuss ] , we discuss the results from the sample as a whole . ngc 5253 is a nearby dwarf galaxy currently undergoing an extreme burst of star formation . using [ siii]/[sii ] ratio maps , @xcite discovered an optically thin ionization cone in this amorphous elliptical galaxy ( figure 2 of * ? ? ? * ) . the ionization cone is narrow with an opening angle @xmath56 and is detected out to @xmath57 pc from the starburst . the measured opening angle implies a solid angle for the cone that spans @xmath58 of 4@xmath59 steradians on the sky , assuming axisymmetry about the cone axis @xcite . if many starbursts have similar geometry , this suggests that an orientation bias may limit our ability to detect escaping lyc in starburst and high - redshift galaxies . ngc 3125 is a dwarf galaxy that has a bursty star formation history with roughly three recent epochs of star formation @xcite . this galaxy has two prominent star forming knots , knot a to the northwest and knot b to the southeast @xcite ; both contain massive star clusters and are the main hosts for the wolf - rayet population in the galaxy . the ionized gas in ngc 3125 is concentrated around the star - forming knots towards the eastern side of the galaxy relative to the overall stellar distribution ( figure [ f : img_n3125 ] ) . figure [ f : ratios_3125 ] shows the emission - line ratio maps for ngc 3125 . @xmath22 , [ sii]@xmath21 , and continuum at @xmath60 in red , blue , and green , respectively . at 11.5 mpc , 10= 560 pc . in this figure , n is up and e is to the left . [ f : img_n3125],title="fig:",width=312 ] @xmath22 , [ sii]@xmath21 , and continuum at @xmath60 in red , blue , and green , respectively . at 11.5 mpc , 10= 560 pc . in this figure , n is up and e is to the left . [ f : img_n3125],title="fig:",width=312 ] + + we discover a bipolar ionization cone in the emission - line gas extending northeast and southwest of knot b. to the northeast this ionization cone has an opening angle of @xmath0 . we obtain this measurement using the lines parallel to the edges of the cone , which are shown by the red dotted lines on the image in figure [ f : img_n3125 ] . as can be seen in the [ siii]/[sii ] ratio map ( figure [ f : s3ds2_3125 ] ) , the ionization cone to the northeast of knot b exhibits high [ siii]/[sii ] throughout . the lack of transition to low values of [ siii]/[sii ] at the edge implies that the ionization cone is optically thin . this conclusion is supported by the low [ sii]/ observed throughout this portion of the cone ( figure [ f : s2dha_3125 ] ) , which is consistent with expectations for optically thin gas @xcite . to the southwest , the line ratios are less conclusive . while they transition to low [ siii]/[sii ] as one moves away from knot b , there seems to be another transition to high [ siii]/[sii ] near the outer edge . however , the line ratios are less certain in this region . in figure [ f : n3125_quant ] , we quantitatively compare the line ratios in the ionization cone to those of the rest of the galaxy . figure [ f : n3125_quant ] shows the line ratios from the binned ratio maps ( figure [ f : ratios_3125 ] ) as a function of @xmath10 , the projected distance above and below the major axis of the galaxy . the red points represent the line ratios from the ionization cone , selected from the region between the white lines shown in the bottom panel of figure [ f : n3125_quant ] . the blue points are the mean line ratios from the rest of the galaxy at that @xmath10 . the error bars on the blue points represent either the standard deviation of the line ratios at that @xmath10 or the error on the mean , whichever is more . all points originate from pixels that have a minimum 2@xmath23 detection in all three emission - line bands in the 3@xmath613 binned images . from the top and middle plots in figure [ f : n3125_quant ] , we see a clear excess in [ siii]/[sii ] and a deficiency of [ sii]/ in the ne side of the cone . this behavior is consistent with expectations for optically thin gas . the trends in the line ratios discussed above are robust , even after considering the effects of internal reddening on the ratio maps . the lines shown on the upper right of figure [ f : n3125_quant ] indicate how the line ratios would change with different levels of reddening . in order to explain the observed [ siii]/[sii ] excess , the reddening would have to have @xmath62 . we note that the measured reddening toward knots a and b are @xmath63 and 0.21 , respectively @xcite , which correspond roughly to @xmath64 . based on polarimetric observations most of the dust is located in the plane of the galaxy @xcite . thus it is reasonable to expect that the extinction in the cone is much less than that toward knots a and b. furthermore , while strong reddening would enhance the [ siii]/[sii ] ratio , correcting for it would drive the [ sii]/ ratio even lower , which strengthens the argument for optically thin gas . @xcite study the kinematics of the gas in ngc 3125 using echelle spectra . they detect a doppler ellipse along the ionization cone to the northeast , showing that it is a bubble expanding with @xmath65 km / s . this detected bubble is consistent with the idea that massive star feedback clears low - density passages in the ism through which ionizing radiation may escape . to add further support to this picture , low - resolution x - ray observations show that the x - ray distribution is elongated along the minor axis of the galaxy @xcite , suggestive of feedback related activity . furthermore , the ionization cone occurs at the edge of the galaxy , which may make it easier for ionizing radiation to escape @xcite . the kinematics towards the southwest are less clear , and @xcite do not detect a doppler ellipse in this region . we also see ionized gas extending to the northeast of knot a. figure [ f : s3ds2_3125 ] shows that while the gas near the cluster has high ionization parameter , it clearly transitions to a low ionization parameter as the distance from the cluster increases . the transition indicates that the galaxy is optically thick in that direction . the line ratio comparison in figure [ f : n3125_quant ] supports this analysis . , above and below the major axis of ngc 3125 for the ionization cone ( red points ) and the mean line ratio at that @xmath10 from rest of the galaxy ( blue points ) . _ top _ : [ siii]/[sii ] . the lines shown on the upper right indicate how the line ratios would change due to reddening of a@xmath66 = 1 , 2 , 5 , and 10 . all points represent pixels that have a minimum 2@xmath23 detection in all three emission - line bands . negative values are towards the northeast , positive values toward the southwest . _ middle _ : same as the top plot but for [ sii]/. _ bottom _ : 3-color emission - line image from figure [ f : img_n3125 ] with white lines delineating the region from which the red points are selected . this figure demonstrates that there is a clear excess of [ siii]/[sii ] and a deficiency of [ sii]/ in the ne side of the cone.[f : n3125_quant],title="fig:",width=331 ] , above and below the major axis of ngc 3125 for the ionization cone ( red points ) and the mean line ratio at that @xmath10 from rest of the galaxy ( blue points ) . _ top _ : [ siii]/[sii ] . the lines shown on the upper right indicate how the line ratios would change due to reddening of a@xmath66 = 1 , 2 , 5 , and 10 . all points represent pixels that have a minimum 2@xmath23 detection in all three emission - line bands . negative values are towards the northeast , positive values toward the southwest . _ middle _ : same as the top plot but for [ sii]/. _ bottom _ : 3-color emission - line image from figure [ f : img_n3125 ] with white lines delineating the region from which the red points are selected . this figure demonstrates that there is a clear excess of [ siii]/[sii ] and a deficiency of [ sii]/ in the ne side of the cone.[f : n3125_quant],title="fig:",width=331 ] , above and below the major axis of ngc 3125 for the ionization cone ( red points ) and the mean line ratio at that @xmath10 from rest of the galaxy ( blue points ) . _ top _ : [ siii]/[sii ] . the lines shown on the upper right indicate how the line ratios would change due to reddening of a@xmath66 = 1 , 2 , 5 , and 10 . all points represent pixels that have a minimum 2@xmath23 detection in all three emission - line bands . negative values are towards the northeast , positive values toward the southwest . _ middle _ : same as the top plot but for [ sii]/. _ bottom _ : 3-color emission - line image from figure [ f : img_n3125 ] with white lines delineating the region from which the red points are selected . this figure demonstrates that there is a clear excess of [ siii]/[sii ] and a deficiency of [ sii]/ in the ne side of the cone.[f : n3125_quant],title="fig:",width=240 ] as discussed in [ s : data ] , our observations are missing a significant amount of flux due to a combination of wavelength calibration uncertainties , velocity offsets , and shallow integration . for this reason , we use the observations of @xcite , to measure the flux in the ionization cone . after correcting for an expected 10% contribution to the flux from the nearby [ nii ] lines , we obtain @xmath67 erg s@xmath68 @xmath69 , which corresponds to a rate of ionizing photons of log()=51.02 . this means that the equivalent of at least 100 ( @xmath70 ) o7 stars ionize the cone , assuming = @xmath71 for an o7 star at the metallicity of ngc 3125 @xcite . knot b has the equivalent of @xmath72 o7 stars based on flux ( * ? ? ? * adjusted to the same per o7 star ) . based on these numbers , roughly 8% of the ionizing radiation generated by the cluster needs to escape into the ne cone . the ionization cone in ngc 3125 is narrow and appears somewhat collimated , with an opening angle of @xmath0 . this angle corresponds to a solid angle of @xmath73% of 4@xmath59 steradians , assuming axisymmetry about the cone axis . we note that this value is not inconsistent with the 8% escape fraction estimated from the flux , above . since knot b is smaller than the base of the cone , the covering fraction of the cone relative to the cluster is larger than that of the cone on the sky . we note that the 8% quoted here only refers to the cluster in knot b. using the total @xmath74()=3.4@xmath75 erg s@xmath68 @xmath69 @xcite , the cone represents @xmath76 of the ionizing flux of the galaxy . the presence of an optically thin ionization cone suggests that ionizing radiation is escaping this starburst . at the very least , we see that ionizing radiation is escaping the main body of the galaxy and traveling into the halo . the small implied solid angle , in conjunction with a preferred direction is similar to the geometry for ngc 5253 , again supporting the possibility that an orientation bias affects our ability to detect escaping lyman continuum in starbursts and lbgs . ngc 1705 is a nearby starburst galaxy with a stunning starburst - driven galactic wind @xcite . ngc 1705 has a combined stellar and gas mass of @xmath77 @xcite . the metallicity measurements range from 12+log(o / h ) = 8.21 to 8.46 @xcite . like similar dwarf starbursts , ngc 1705 has a bursty star formation history with two recent star forming epochs in the last 15 myr @xcite . one of these bursts formed the well known super - star cluster ( ssc ) at the center , which has log(@xmath15)@xmath78 and an age of @xmath79 myr @xcite . this ssc generates most of the uv luminosity and is consistent with a population dominated by b stars @xcite . that same 1015 myr old burst of star formation is thought to have launched the galactic wind @xcite . this burst was followed by a more recent epoch of star formation , @xmath80 myr ago that may have been triggered by the wind @xcite . @xmath22 , [ sii]@xmath21 , and continuum at @xmath60 in red , blue , and green , respectively . at 5.1 mpc , 10= 250 pc . in this figure , n is up and e is to the left . [ f : img_n1705],title="fig:",width=331 ] @xmath22 , [ sii]@xmath21 , and continuum at @xmath60 in red , blue , and green , respectively . at 5.1 mpc , 10= 250 pc . in this figure , n is up and e is to the left . [ f : img_n1705],title="fig:",width=312 ] + + in emission - line gas , ngc 1705 shows filamentary extended substructure as a result of its galactic wind @xcite . based on observations , the superbubble has an expansion velocity 50132 km / s @xcite . @xcite studied the kinematics using uv interstellar absorption lines and found further evidence for an expanding wind . x - ray observations show that the wind has similar size and morphology to the optically detected gas @xcite . the star - forming nucleus resides on the axis of symmetry for the superbubble , which further supports that the wind is driven by star formation @xcite . shocks contribute only a small amount to the ionization in the wind , as most of the emission - line gas is photo - ionized @xcite . our emission - line images and ratio maps for ngc 1705 are shown in figures [ f : img_n1705 ] , and [ f : ratios_1705 ] . the ionized gas associated with the wind is clearly detected in both [ siii ] and [ sii ] . the spatial changes in the [ siii]/[sii ] ratio map suggest that the wind in ngc 1705 is optically thick . close to the starburst , the [ siii]/[sii ] ratio is high , as expected from photoionized gas near the ionization source . however , as one looks further into the wind , there is a clear transition towards lower [ siii]/[sii ] ratios , and at the far edge of the shells , the line ratios indicate the [ sii ] emission is dominant . this is further supported by the [ sii]/ and [ siii]/ ratio maps in figures [ f : s2dha_1705 ] and [ f : s3dha_1705 ] . in these maps , the [ sii]/ ratio increases at the edges of the filament , and , as expected based on figure [ f : s3ds2_1705 ] , the [ siii]/ ratio decreases . while this behavior is consistent with optically thick gas , we note that if the radiation field is dominated by hot stars with softer ionizing spectra , as in the case of ngc 1705 , the gas may simultaneously exhibit a low - ionization transition zone and be optically thin @xcite . in addition to evaluating the optical depth in the plane of the sky , the ratio maps can be used to evaluate the optical depth along the line of sight . for gas with abundances similar to that of the lmc , [ sii]/ @xmath81 is a diagnostic for optically thin photoionized gas along the line of sight . of particular interest is the region around ngc 1705 - 1 , the super star cluster . @xcite estimated a very low absorbing hi column , @xmath82 , toward ngc 1705 - 1 . this column is an order of magnitude smaller than the column measured from 21 cm observations , which suggests that the cluster is either sitting in front of most of the hi or that we are observing the ssc through a hole in the ism @xcite . either scenario increases the likelihood for escaping lyman continuum . however , figure [ f : s2dha_1705 ] shows that [ sii]/ @xmath83 over the entire nuclear region , which indicates that the nuclear region is optically thick in the line of sight . the low hi column combined with optical [ sii]/ implies that even though the neutral gas may have large variations in column density , the majority of sight lines are optically thick . it is expected that strong starburst feedback , such as that present in ngc 1705 , is conducive to escaping lyman continuum . however , our observations indicate that the wind in ngc 1705 is optically thick . both the strength of the radiation field and the amount of gas available to absorb the ionizing radiation will determine whether the galactic wind is optically thick or thin . while there have been two recent epochs of star formation , one 10 - 15 myr ago and one @xmath80 myr ago @xcite , the ionizing population is dominated by b stars @xcite . this means that there are relatively fewer ionizing photons being emitted and the radiation field is softer than if there were significant numbers of o stars . in addition to a reduced radiation field , a ` spur ' of hi gas is associated with the outflow @xcite . this spur of neutral gas is likely co - spatial with the optically detected galactic wind based on its kinematics and orientation @xcite . the feature has an hi column density of @xmath84 , which is more than high enough to be optically thick to lyman continuum . thus , despite the presence of a strong galactic wind , ngc 1705 is likely optically thick due to a combination of population age and gas morphology . he 2 - 10 is one of the prototypical hii galaxies @xcite . it is 9 mpc away @xcite and has dynamical mass @xmath85 @xcite . in emission - line gas there are many notable loops and filaments ( figure [ f : img_he210 ] ) whose kinematics suggest expanding bubbles and possible outflows @xcite . these bubbles are centered on sites of recent star formation . he 2 - 10 is the host of nearly 80 sscs @xcite . the central starburst contains a handful of clusters in ultra - compact hii regions that contain the ionizing equivalent of 700 - 2600 o7 stars each @xcite . in addition to prodigiously forming stars , sfr @xmath86 0.87 - 2 /yr @xcite , this galaxy was recently found to be the host of an agn @xcite . / @xmath87 . bottom : three - color composite with rest frame [ siii]@xmath22 , [ sii]@xmath21 , and continuum at @xmath60 in red , blue , and green , respectively . at 9 mpc , 10= 460 pc . in this figure , n is up and e is to the left . [ f : img_he210],title="fig:",width=321 ] / @xmath87 . bottom : three - color composite with rest frame [ siii]@xmath22 , [ sii]@xmath21 , and continuum at @xmath60 in red , blue , and green , respectively . at 9 mpc , 10= 460 pc . in this figure , n is up and e is to the left . [ f : img_he210],title="fig:",width=312 ] + + in the [ siii]/[sii ] ratio map ( figure [ f : s3ds2_he210 ] ) , the central star forming region stands out with high [ siii]/[sii ] , as would be expected for a region hosting thousands of o stars . as the radius from the center increases , [ siii]/[sii ] drops considerably . there is a local maximum @xmath88 to the east around the location of the second ssc . however , even here the emission - line gas transitions to low [ siii]/[sii ] toward the edge of the expanding bubble . the contrast at the edges of the bubbles is seen more clearly in the [ sii]/ ratio map ( figure [ f : s2dha_he210 ] ) , where a clear enhancement of [ sii]/ is present . in contrast the [ siii]/ ratio ( figure [ f : s3dha_he210 ] ) falls off as the distance from the ionizing source increases , as would be expected for an optically thick nebula . the spatial changes in the line ratio maps are consistent with expanding superbubbles . high ionization parameter gas is present close to the ionizing source , while the edges of the bubbles have low ionization parameter and possible contribution to [ sii ] from shocks . studies of the kinematics indicate a potential doppler ellipse in the ne bubble and velocities differences of up to @xmath89 km / s across the galaxy @xcite . using uv interstellar lines , @xcite determined that the bulk motion of the ism indicates an outflow with @xmath90 km / s . the velocities reported by both works are in excess of the escape velocity @xmath91 km / s @xcite . however , the closed loops of the outflow seen in the images suggest that these bubbles have not yet broken out of the ism to form a galactic fountain . bubble walls severely hinder the passage of ionizing radiation , and can delay the escape of lyman continuum radiation until @xmath92 has dropped considerably @xcite . he 2 - 10 is unique in our sample , in that it is the only galaxy that contains a confirmed agn . this agn was discovered by @xcite as a radio and x - ray point source and is located near the star forming region of knot a. the accretion rate onto the agn is modest , but may have been greater in the past @xcite . thus , the presumed tracers for massive - star feedback , such as the large expanding bubbles , may be contaminated by contributions from the agn as opposed to being driven by star formation . the distribution of gas in the ism plays a critical role in the passage of ionizing radiation in a galaxy . the hi observations from @xcite show that the neutral ism extends significantly beyond the optically detected ionized gas . this observation is consistent with the observations of optically thick bubbles . however , if this overall hi envelope has a clumpy distribution , it is still possible to have small low - density passages through which ionizing radiation may escape the starburst . the [ sii]/ ratio map shows two locations that have [ sii]/@xmath81 and may be optically thin . we note these locations with small circles on figure [ f : img_he210 ] . one of these regions coincides with the location of the agn , and the other lies 4 to the se from the first . while our data can not establish the presence of an optically thin path from the galaxy conclusively , these observations are suggestive that feedback from the agn or star formation in the region have cleared out a small hole in the ism . the variable reddening measurements derived from different wavelengths and positions in the galaxy further suggest an inhomogeneous distribution of gas and dust throughout the galaxy ( e.g. , * ? ? ? * ) . the presence of small local holes is also supported by the fact that many of the probably young sscs do not show commensurate emission , which suggests that gas has been evacuated in the regions of these clusters @xcite . if this is the case , it supports the picture that ionizing radiation escapes galaxies through small holes and that an orientation bias affects our ability to detect escaping lyman continuum . we note that these line ratios have not been corrected for reddening . however , any reddening correction would drive the [ sii]/ line ratio lower , strengthening the argument for optically thin gas along the line of sight . ngc 1482 is an sa0 @xcite with a well known galactic wind @xcite . at a distance of 22.6 mpc @xcite , this starburst is a highly inclined disk with a prominent dust lane @xcite . the wind was first discovered by @xcite in optical emission - lines , with later optical and x - ray observations confirming its nature as a wind @xcite . @xmath22 , [ sii]@xmath21 , and continuum at @xmath60 in red , blue , and green , respectively . at 22.6 mpc , 10= 1.1 kpc . in this figure , n is up and e is to the left . [ f : img_n1482],title="fig:",width=331 ] @xmath22 , [ sii]@xmath21 , and continuum at @xmath60 in red , blue , and green , respectively . at 22.6 mpc , 10= 1.1 kpc . in this figure , n is up and e is to the left . [ f : img_n1482],title="fig:",width=312 ] + + in our emission - line images ( figure [ f : img_n1482 ] ) and ratio maps ( figure [ f : ratios_1482 ] ) , we primarily detect the wind in [ sii ] and . the [ siii ] emission is confined to the star forming region in the inner disk . the ionization parameter map shows strong evidence for radiation bounding in ngc 1482 . high [ siii]/[sii ] traces the disk star forming region , and the line ratio drops precipitously as one moves further into the wind ( figure [ f : s3ds2_1482 ] ) . @xcite found the wind to be dominated by shock ionization based on [ nii]/ observations . our [ sii]/ map ( figure [ f : s2dha_1482 ] ) confirms this picture with [ sii]/ @xmath93 in the wind . in reality , the emission observed in the wind will be a combination of shock- and photoionized gas . in order to put strong constraints on the [ siii]/[sii ] line ratio in the wind , we explore correcting the line ratios for the shock ionization as follows . as discussed by @xcite , the observed line ratio can be written as : @xmath94 where a and b correspond to the emission lines in question , and the subscripts @xmath95 , @xmath96 , and @xmath97 indicate the observed , shocked , and photoionized line strengths , respectively . to convert our observed line ratios to the line ratios of photoionized gas , we need to solve for the fraction of shock - ionized relative to photoionized gas , @xmath98 together with equation [ eq : photo ] , this yields : @xmath99 . in equation [ eq : step3 ] , @xmath100 can be replaced by the predicted line strengths from mappings iii shock models @xcite . since @xcite observed velocities of up to @xmath101 km / s in the wind , we explore shocks with @xmath102 km / s ( assuming strong shocks ) . for this calculation we use the mappings iii shock - only models , which provide the most optimistic correction factor for high [ siii]/[sii ] from photoionization . the precursor component to the shock would increase the relative contribution of shocks to the [ siii ] , thus moving the corrected line ratios more towards optically thick than the shock only models . to solve for @xmath103 , we write equation [ eq : step3 ] for each emission line ratio , [ siii]/[sii ] , [ siii]/ , and [ sii]/. by combining the resultant equations , we finally solve for @xmath103 : @xmath104}{{{\mbox{h{\footnotesize { \mbox{$\alpha$}}}}}}}_o\right ) + \it{d_2}\left(\frac{[siii]}{[sii]}_o\right ) + \it{d_3}}{\it{d_3d_2}}\ ] ] where @xmath105}{[sii]}_o - \frac{[siii]}{[sii]}_s\\\ ] ] and @xmath106 and @xmath107 are the corresponding values for [ sii]/ and [ siii]/ , respectively . after solving for @xmath103 , we use that value in equation [ eq : step3 ] to create a new [ siii]/[sii ] map ( figure [ f : n1482_shock ] ) . we find that the [ siii]/[sii ] maps remain strongly radiation bounded . this map is created assuming shock velocities of 950 km / s and the mappings iii shock - only model @xcite . even in the most optimistic correction , the [ siii]/[sii ] ratio transitions to values @xmath108 close to the disk of the galaxy . n is up and e is to the left . [ f : n1482_shock],width=384 ] since there is evidence that the wind has cleared gas from the ism of ngc 1482 , it is somewhat surprising that this object does not show evidence for an optically thin wind . hi observations reveal a hole in the hi distribution at the location of the wind @xcite . further evidence that the wind has cleared material from the center of the galaxy is found by @xcite . they generate a [ sii]@xmath109 ratio map and see a sharp gradient at the base of the wind , which suggests clearing of the inner disk by the wind @xcite . both of these lines of evidence suggest that ionizing radiation can escape along the wind . perhaps the massive star population in ngc 1482 has aged in the time since the wind launched , and , thus , @xmath92 has dropped significantly . another explanation could be that dust entrained in the wind @xcite provides additional material that absorbs radiation as it travels away from the galaxy . we note here that some of the emission may be shifted out of our very narrow bandpasses . for @xmath110 km / s , the emission will shift by 5 and 7 for [ sii ] and [ siii ] , respectively . this is comparable to half the bandpass in each of those filters . exacerbating this effect , our central bandpass in [ siii ] is shifted @xmath111 from the appropriate central bandpass for the redshift of ngc 1482 . this means that for ngc 1482 we do not detect some of the redshifted emission from [ siii ] in the wind , although we should detect all of the blueshifted component . that being said , the wind exhibits the transition from high to low ionization parameter gas relatively close to the disk of the galaxy . furthermore , the [ siii]/[sii ] line ratio is well below 0.5 for most of the wind . even if we assume that we are missing half of the [ siii ] flux in the wind because of the velocity offset , the difference does not push the [ siii]/[sii ] ratio high enough to match our expectations for optically thin emission . given the 5 kpc extent of the wind @xcite , the location of the transition zone makes it less likely that the transition is a spurious one . ngc 178 is forming stars at a rate of 0.55 /yr @xcite and lies 20.6 mpc away @xcite . in the emission - line images we see that ionized gas is split between two areas , one around the bulk of the observed continuum , and a separate extra - planar region ( figure [ f : img_n178 ] ) . since ionizing radiation emitted at the edge of galaxies is more likely to escape @xcite , it might be reasonable to expect to find some evidence of density bounding in this region of the ionization parameter map . @xmath22 , [ sii]@xmath21 , and continuum at @xmath60 in red , blue , and green , respectively . at 20.6 mpc , 10= 1 kpc . in this figure , n is up and e is to the left . [ f : img_n178],title="fig:",width=331 ] @xmath22 , [ sii]@xmath21 , and continuum at @xmath60 in red , blue , and green , respectively . at 20.6 mpc , 10= 1 kpc . in this figure , n is up and e is to the left . [ f : img_n178],title="fig:",width=312 ] however , the ionization parameter map shows that all nebular regions are optically thick . the ionized gas to the west is comprised of a few separate peaks in the [ siii]/[sii ] ratio map ( figure [ f : ratios_178 ] ) , all of which exhibit a clear transition to low ionization parameter gas at the edges . + + the [ sii]/ map ( figure [ f : ratios_178 ] ) is intriguing . there appears to be an envelope of high [ sii]/ surrounding the galaxy . this envelope could be caused by the well known [ sii]/ gradient in the warm , ionized medium of galaxies @xcite or could be a signature of shocked gas surrounding the galaxy . we note , however , that there are significant sky gradients and background features around the galaxy that are caused by scattered light within the instrument and by bright time observing conditions ( [ s : data ] ) . these sky features make it challenging to determine the nature of the surrounding low ionization parameter envelope . ngc 7126 is an sa(rs)c galaxy @xcite . at 45.5 mpc away @xcite , it is the most distant galaxy in our sample . the emission - line images reveal an extended spiral structure filled with hii regions ( figure [ f : img_n7126 ] ) . hi observations show that it is interacting with its nearby companion ngc 7125 @xcite . ngc 7126 does not have a catalogued wind , and , with its high star formation rate , acts as a control against this parameter in the sample @xcite . @xmath22 , [ sii]@xmath21 , and continuum at @xmath60 in red , blue , and green , respectively . at 45.5 mpc , 10= 2.3 kpc . in this figure , n is up and e is to the left . [ f : img_n7126],title="fig:",width=331 ] @xmath22 , [ sii]@xmath21 , and continuum at @xmath60 in red , blue , and green , respectively . at 45.5 mpc , 10= 2.3 kpc . in this figure , n is up and e is to the left . [ f : img_n7126],title="fig:",width=312 ] + + the emission - line ratio maps are shown in figure [ f : ratios_7126 ] . the [ siii]/[sii ] map shows a remarkably low ratio of [ siii]/[sii ] . while the spiral structure is detected , it does not show up as clearly as it does in the ratio maps . in the [ sii]/ ratio map , the hii regions themselves appear with line ratios consistent with photoionized gas . as one moves to the edges of the individual regions , [ sii]/ increases . this is consistent with expectations of the warm , ionized medium in galaxies @xcite . we note , however , that these line ratios are a bit high , and they approach values consistent with shock ionization . the [ siii]/ map exhibits relatively low line ratios throughout , even at the centers of the observed hii regions . this could be indicative that the ionizing stellar population is composed primarily of b or late o type stars . the full sample of seven galaxies provides examples of both optically thin and optically thick extended emission - line gas . by comparing the properties of these galaxies we can gain insight on the processes and properties that influence the escape of ionizing radiation . table [ t : sfr ] presents relevant properties for each galaxy in the sample . columns 1 and 2 are the galaxy name and hi mass , respectively ; columns 3 , 4 , and 5 present the derived sfr , sfr surface density ( ) , and the specific sfr ( ssfr ) , respectively . column 6 provides the references for the star formation rates , hi masses and radii . finally , column 7 indicates whether the galaxy shows wind and/or expanding bubble activity , and column 8 shows the ratio of hi mass to stellar mass ( table [ t : props ] ) . l c c c c c c c + + ngc 3125 & & 0.461 & 0.0826 & 0.374 & ( 1),n / a , ( 2 ) & y ( 3 ) & + ngc 5253 & 8.22 & 0.428 & 1.04 & 0.382 & ( 1),(4),(5 ) & y ( 3 ) & 0.006 + + + + ngc 1482 & 9.13&3.39 & 0.0271 & 0.093 & ( 1),(6),(7 ) & y ( 8) & 0.037 + ngc 1705 & 7.96 & 0.057 & 0.0091 & 0.241 & ( 1),(4),(9,7 ) & y ( 10 ) & 0.389 + he 2 - 10 & 8.49&0.872 & 0.340 & 0.276 & ( 1),(11),(12 ) & y ( 13 ) & 0.098 + + + + ngc 178 & 9.40 & 0.549 & 0.019 & 0.315 & ( 14),(4),(4 ) & n & 1.445 + ngc 7126 & 10.57@xmath112 & 5.21 & 0.0467 & 0.188 & ( 14),(4),(4 ) & n & 1.349 the interpretation of escaping lyman continuum observations in starbursts and lbgs is challenging because of the need to disentangle observational effects from intrinsic trends . one issue under debate is whether the relatively few ( @xmath113 ) galaxies that have significant are representative of the fraction of galaxies that have escaping lyman continuum . potential challenges that can modify the intrinsic distribution of include low redshift interlopers , a preferred direction of escape , and small covering fractions for the escaping lyman continuum ( e.g. , * ? ? ? * ; * ? ? ? examining the morphologies of the ionization cones observed in this work can shed light on this issue . two of the seven starburst galaxies studied here show strong evidence for optically thin ionization cones . in both ngc 5253 ( see also * ? ? ? * ) and ngc 3125 , the cones are aligned along the minor axis . this behavior is consistent with the expectations of massive - star feedback - driven models ( e.g. , * ? ? ? * ; * ? ? ? the observed alignment implies a preferred direction ; feedback will punch holes in the ism more easily along the minor axis ( e.g. , * ? ? ? in a study of low redshift lbg analogs , @xcite measured extreme wind velocities ( @xmath114 ) in all galaxies with low optical depth to lyman continuum . the high velocities indicate the direction of the wind must be nearly aligned with the line of sight to the galaxy . this relationship between low optical depth and high outflow velocities further strengthens the argument that lyman continuum escape has a preferred direction @xcite . furthermore , the ionization cones exhibit narrow morphologies and subtend an estimated 4% of @xmath115 steradians . this implies that the covering fraction of escaping radiation will be much less than unity , and is consistent with the results from studies of high redshift star - forming galaxies ( e.g. , * ? ? ? * ; * ? ? ? we note that the narrow ionization cones do not necessarily imply small . if the physical size of the cone s base is significantly larger than the size of the ionizing cluster , then large fractions of the ionizing flux produced by the cluster may still escape through the angularly narrow cone . the combination of small opening angles and preferred directions suggest that if one tried to directly detect escaping lyman continuum , the axis of escape would need to be close to the observer s line of sight in order to obtain a strong detection . if more galaxies are like ngc 3125 and ngc 5253 , then an orientation bias makes it more challenging to detect escaping lyman continuum in starbursts and lbgs . in light of the massive - star feedback - driven model discussed previously , the first galactic property to examine is the presence of a galactic wind . all of the galaxies that show evidence for extended emission also have evidence for either a galactic wind or superbubble expansion out of the galaxy . thus confirming the link between winds and the fate of ionizing radiation ( e.g. , however , ngc 1705 and ngc 1482 , the two galaxies that have the most clearly established winds , both appear optically thick . we note that both galaxies without extended emission , ngc 178 and ngc 7126 , do not have galactic winds . thus , as discussed in @xcite , we see that while wind activity is important for the escape of lyman continuum , other factors clearly play a determining role . another critical property to examine is star formation rate . models predict that escaping ionizing radiation will be associated with extreme feedback from strongly star forming systems ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) . from table [ t : sfr ] , it is immediately apparent that sfr ( ) is not directly correlated with the optical depth of extended emission . the galaxy with the highest sfr , ngc 7126 , is one that most clearly does not have extended optically thin emission . furthermore , the galaxy with the second highest sfr , ngc 1482 , is optically thick . a more meaningful comparison is to look at the and the specific sfr . these measures of the sfr take into account the sizes and masses of the galaxies . a more massive galaxy would need a higher sfr to produce enough feedback for ionizing radiation to escape . table [ t : sfr ] shows that ngc 178 and ngc 7126 , which show no evidence for extended emission , do have low , which is consistent with our expectations . however , distinguishing between the galaxies that have optically thin and thick emission is more challenging . ngc 5253 has the highest , but ngg 3125 has lower than he 2 - 10 , which is likely optically thick . one explanation is that the higher sfr in he 2 - 10 contains contamination from its agn . however , the agn in he 2 - 10 is not particularly luminous in @xcite making the significance of its contribution unclear . ngc 1705 has comparatively low . paired with the lower of ngc 1482 , the rates are consistent with the hypothesis that optical depth is correlated with . however , we can not extrapolate any conclusions on trends with based only on our findings for these individual galaxies . the comparison between ssfr is similarly fraught . ngc 5253 and ngc 3125 do have among the highest ssfr , but their ssfrs are not significantly higher than those of the galaxies with optically thick emission . furthermore , ngc 178 , which does nt have clear evidence for extended emission , has a rather high ssfr . thus , our small sample supports previous work in its conclusion that strong sfr is necessary but not sufficient for significant @xcite . we caution the reader that our sample is small , and the range of star - formation rates is also small . therefore , our sample does not provide the leverage to firmly establish any apparent correlation between optical depth and or ssfr . as an additional consequence of the small sample size , the galaxies that seem anomalous , such as ngc 178 with its high ssfr but lack of extended emission and he 2 - 10 , which is optically thick despite its high , may not be significant outliers . in addition to sfr , the recent star formation history will have an influential role on the escape of ionizing radiation . as a stellar population ages , the rate of ionizing photons produced ( ) decreases and the hardness of the radiation field softens . the ages of the clusters responsible for the ionization cones in both ngc 5253 and ngc 3125 are between 2 - 5 myr old @xcite . in contrast with this , ngc 1705 is in a post - burst state ( cluster ages @xmath116 myr ) in which most of the o stars in the main super star clusters have died off @xcite . thus , the softer and fainter b stars dominate the radiation field and do not provide enough ionizing radiation to allow for escape . this may similarly offer a partial explanation for the line ratios observed in he 2 - 10 . while the main star forming region , embedded in the center of the galaxy , contains many young clusters with ages ranging from 2 - 20 myr @xcite , the second star - forming region , further towards the eastern edge of the galaxy is likely much older with fewer of its massive stars remaining . furthermore , visual examination of the images of he 2 - 10 show closed bubbles , rather than fractured filaments ( figure [ f : img_he210 ] ) . we note that for radiation fields dominated by late type stars , the relationship between the ionization parameter map and the optical depth is complicated . in these situations , a transition to [ sii ] dominated gas does not necessarily indicate an optically thick nebula @xcite . if the radiation field is soft but sufficiently strong , it is possible that not enough gas is present to absorb the lower energy photons , yet , the ionization parameter map still exhibits a transition zone to [ sii ] . furthermore , whether the population is a single burst or an extended burst also influences the fate of ionizing radiation . while feedback generates low density passageways in the ism , simulations of superbubble evolution show that there is a delay between superbubble formation and the actual escape of ionizing radiation @xcite . during the phase prior to superbubble break - out , ionizing radiation gets trapped by the bubble walls @xcite . once the bubble breaks out of the disk , ionizing radiation may escape relatively unhindered . however , by this point in time , the rate of ionizing photons produced by the massive star population is considerably weaker than in the first few myr @xcite . if the starburst occurs over a short duration , essentially a single burst , then it is less likely to leak significant ionizing radiation . if instead the burst is more extended , the later episodes of star formation can capitalize on the passageways carved by prior star forming episodes . observationally , @xcite showed that extended bursts with typical timescales of hundreds of myr are common in dwarf starbursts . our data further support this scenario . in ngc 5253 , the current burst of star formation occurred within the last 35 myr @xcite . while it is unlikely that any superbubbles formed from this epoch will have broken from the disk , there is evidence for prior star formation episodes with ages on the order of 10100 myr @xcite . thus , the ionization cone in ngc 5253 probably formed through low density regions cleared out by feedback from earlier episodes of star formation . in ngc 3125 , the super star cluster at the base of the ionization cone has an age between 13 myr , while the region surrounding it has an overall mean age of between 810 myr @xcite . while the initial episode of star formation in this burst is younger than that shown for ngc 5253 , the dominant star formation is on the periphery of the galaxy , which makes it easier for ionizing radiation to escape @xcite . furthermore , @xcite stacked a sample of similar hii galaxies in which ngc 3125 was the dominant source and found that 65% of the stellar population is younger than 5 myr , 32% is between 5 and 100 myr , and the remaining 3% of the population us is older than 100 myr . while the sfh analysis is not based solely on ngc 3125 it does point in the direction of ngc 3125 having multiple recent episodes of star formation . the young ages of the ionizing populations in ngc 3125 and ngc 5253 may represent a special age range needed to find escaping ionizing radiation . in a recent paper , @xcite study a sample of @xmath117 galaxies whose extremely high [ oiii]/[oii ] suggest that they are likely optically thin . these galaxies all have ionizing populations that are between 3 - 5 myr @xcite , similar to the ages observed in galaxies with ionization cones studied in this work . furthermore , the two local galaxies that have direct measurements , both have significant populations of young stars @xcite . finally , it has been suggested that the mass of the galaxy is correlated with the escape fraction of ionizing radiation . there are some simulations that predict higher with larger galaxy mass ( e.g. , * ? ? ? others claim the opposite because the smaller galaxy s weaker potential well would make escape easier @xcite . based on our small sample here , we do not find evidence for or against either claim . the two galaxies for which the extended ionized gas is optically thin have masses that are in the middle to low range of masses in the sample ( tables [ t : props ] and [ t : sfr ] ) . interesting to note , these two galaxies have significantly higher fractions of hi relative to the stellar mass of the galaxy ( table [ t : sfr ] ) than the sample galaxies with extended ionized gas . meanwhile , ngc 5253 has an exceedingly low hi mass fraction , even when compared to the other galaxies with winds . this is consistent with the idea that the clearing , or consumption , of the neutral gas in the galaxy plays a role in regulating the escape of ionizing radiation . the passage of ionizing radiation through , and possibly out of , a galaxy has ramifications on our understanding of cosmic reionization . in this paper , we study the extended , low surface brightness , emission - line gas of seven dwarf starburst galaxies . using narrowband [ siii ] , [ sii ] , and images and the technique of ionization parameter mapping , we evaluate the optical depth of the extended emission . in two of the galaxies , we discover optically thin ionization cones extending along the minor axis . these cones suggest that ionizing radiation is escaping the main body of the galaxy . the narrow morphology suggests that ism morphology is a critical determinant of whether ionizing photons will escape a galaxy . in three of the galaxies , we find that the established galactic winds are most likely optically thick . these galaxies , despite the presence of strong feedback do not show clear evidence for escaping ionizing radiation . the two remaining galaxies , despite the strong star formation show little evidence of extended emission . our most convincing examples of optically thin emission are found in narrow ionization cones . these narrow cones illustrate a scenario in which the covering fraction of lyman continuum is significantly less than unity , as has been suggested by other work ( e.g. , * ? ? ? furthermore , both cones are aligned with the minor axis of the galaxy , indicating a preferred direction . the small opening angle and preferred direction suggest that unless the galaxy orientation is such that the axis of escape aligns with the line of sight , it will be challenging to directly detect escaping radiation . if other starbursts are similar to ngc 3125 and ngc 5253 , an orientation bias at least partially explains the low detection rate of ionizing radiation in starburst galaxies and lbgs . in addition to observational biases , we explore the galactic properties that may regulate the escape of ionizing radiation . while galactic winds and bubble activity encourage escaping radiation , not all starbursts with galactic wind have escaping lyman continuum . one explanation is that by the time the bubble or wind breaks out of the ism , the ionizing population has aged and is no longer producing ionizing radiation as prodigiously ( e.g. , * ? ? ? * ; * ? ? ? alternatively , the ionizing population is too young to have carved sufficient low - density paths out of the ism for escape . this suggests an optimal age between 3 - 5 myr for the escape of ionizing radiation . based on the recent star - formation histories of ngc 3125 and ngc 5253 , a starburst with a few recent star forming episodes strikes the balance between the time it takes for bubbles to burst and maximizing the number of ionizing photons produced that have clear passage out of the galaxy . another potential contributing factor is the concentration of star formation in the burst . in the examples of optically thin ionization cones , we see that they emanate from specific , concentrated star forming regions , rather than extended star - forming areas . in contrast , ngc 7126 shows no sign for feedback - driven extended emission , and has its copious star formation spread throughout the disk of the galaxy . this suggests that concentrated star formation will be more effective at clearing out the ism . however , we must be careful in drawing strong conclusions from so few objects . he 2 - 10 and ngg 1705 , which also have concentrated star formation , are most likely optically thick to ionizing radiation . thus , while the concentration of star formation is important , the interplay between the star formation and the ism morphology most likely regulates the escape of ionizing radiation from starbursts and lyman break galaxies . support for this work was provided by the national science foundation , grants ast-0806476 , ast-0907758 , and ast-1210285 to mso ; and ast-0606932 and ast-1009583 to sv and mm . we also acknowledge a margaret and herman sokol faculty award to mso . mm acknowledges support by nasa through a hubble fellowship grant hst - hf51308.01-a awarded by the space telescope science institute , which is operated by the association of universities for research in astronomy , inc . , for nasa , under contract nas 5 - 26555 . we are grateful to colin slater for help collecting observations and the um fang research group , eric pellegrini , amy reines , lee hartmann , and timothy mckay for helpful discussions . de vaucouleurs , g. , de vaucouleurs , a. , corwin , h. g. , jr . , et al . 1991 , third reference catalogue of bright galaxies . volume i : explanations and references . volume ii : data for galaxies between 0@xmath119 and 12@xmath119 . volume iii : data for galaxies between 12@xmath119 and 24@xmath119 . , by de vaucouleurs , g. ; de vaucouleurs , a. ; corwin , h. g. , jr . ; buta , r. j. ; paturel , g. ; fouqu , p .. springer , new york , ny ( usa ) , 1991 , 2091 p. , isbn 0 - 387 - 97552 - 7 , price us$ 198.00 . isbn 3 - 540 - 97552 - 7 , price dm 448.00 . isbn 0 - 387 - 97549 - 7 ( vol . i ) , isbn 0 - 387 - 97550 - 0 ( vol . ii ) , isbn 0 - 387 - 97551 - 9 ( vol .

the formalism of the relativistic ( or dirac- ) brueckner approach in infinite nuclear matter is described . as nucleon - nucleon interaction the one - boson exchange potentials bonn a , b , c and for comparison the walecka model are used . the t - matrix is determined from the thompson equation and is projected onto five covariant amplitudes . by the restriction to positive energy states an ambiguity arises in the relativistic brueckner approach which is discussed here in terms of the pseudo - scalar and the pseudo - vector projection . the influence of the coupling of the nucleon via the t - matrix as an effective two - nucleon interaction to the nuclear medium is expressed by the self - energy . in particular we investigate the scalar and vector components of the self - energy for the different one - boson exchange potentials and discuss their density and momentum dependence . we estimate the uncertainty of the self - energy due to the pseudo - scalar and the pseudo - vector choice . usually the momentum dependence of the self - energy is thought to be weak , however , we find that this depends on the one - boson exchange potentials . for the bonn potentials , in contrast to the-potential , the momentum dependence is strikingly strong above as well as below the fermi surface . we compare with the results of other groups and study the effects on the equation of state and the nucleon optical potential .

nucl-th9701060

in the relativistic ( or dirac- ) brueckner model the dynamical two - nucleon correlations in infinite nuclear matter are studied . the formalism is based on an effective quantum field theory for mesons and nucleons @xcite and thus ignores the underlying quark nature of the nucleon . in the brueckner model the t - matrix ( or brueckner g - matrix ) serves as an effective in - medium two - body interaction . it is determined by a self - consistent summation of the ladder diagrams in a quasipotential approximation ( thompson equation ) to the bethe - salpeter equation , see , e.g. , ref . the bare nucleon - nucleon ( ladder ) interaction is described by one - boson exchange potentials . in the present work we apply the bonn potentials a , b , c of refs . @xcite and for comparison also a simple @xmath0-potential which is well known from the walecka model @xcite . the nucleon inside the nuclear medium is treated as a dressed dirac particle where the influence of the coupling to the surrounding nucleons via the t - matrix serving as an effective interaction is expressed by the nucleon self - energy . the self - energy has a lorentz structure with large scalar and four - vector components and modifies the single - particle spectrum and the spinor wave functions . the relativistic brueckner model has been developed in the last decades by various authors . at the beginnig of the 80 s the relativistic problem was considered by the brooklyn group @xcite in first order perturbation theory . in the following the covariant formalism to a self - consistent treatment of the thompson equation was developed by horowitz and serot @xcite and is in detail outlined in ref . . however , in these works the nucleon - nucleon interactions was described within the framework of the @xmath0-model . calculations with realistic nucleon - nucleon interactions have been performed by brockmann and machleidt @xcite and later on by ter haar and malfliet @xcite . a more rigorous derivation of the brueckner approach in the framework of relativistic green s function can be found in ref . @xcite . in the present work we refer to the formalism described in refs . @xcite . the main difference between the relativistic brueckner approach of refs . @xcite and previous works is the use of medium dependent spinors and the requirement of self - consistency for both the spinor wave functions and the single - particle spectrum . the dirac spinors of the nucleon in the medium differ considerably from the free spinors due to the large scalar and vector self - energies . these introduce an additional density dependence into the spinor matrix elements of the one - boson exchange potentials and thus into the t - matrix . this dynamical effect is absent in non - relativistics . another new aspect of the modern brueckner approach is that the inertial frames where the t - matrix is determined and where it is used have to be distinguished carefully . due to the neglection of the retardation in the meson propagators the one - boson exchange potentials and the correspondingly t - matrix serve as quasi - potentials in the two - particle center - of - mass ( c.m . ) therefore the relativistic thompson equation is solved in this system which can be done by numerical standard techniques , as , e.g. , in the non - relativistic case of ref . @xcite . on the other hand single - particle quantities like the self - energy of the nucleon are calculated in the nuclear matter rest frame , where the single - particle distribution , the fermi sea , is naturally defined . thus a covariant representation of the t - matrix is required to connect both frames . even though , e.g. , the single particle potential can be calculated with some additional approximations avoiding this representation , as done in ref . @xcite , the covariant technique is unavoidable in order to calculate the self - energy components themselves . in the relativistic brueckner approach the self - consistent spinors are restricted to positive energy states . a different approach in the full dirac space which is also based on one - boson exchange potentials is used in refs . but in a full treatment one should further include renormalisation effects like the vacuum fluctuation contributions to the self - energy , as discussed in the context of the relativistic hartree approximation in refs . @xcite , and solve the full bethe - salpeter equation self - consistently . however , then the one - boson exchange potentials would be inappropriate . but these are still outstanding problems . in the standard brueckner model which we discuss in the present work one gets around the problem of renormalization . the masses and coupling constants in the model are physical renormalized quantities and as a consequence one neglects the effective negative - energy states . however , the effective positive - energy spinors still contain admixtures of free negative - energy states because they are a superposition of free positive- and negative - energy spinors . to lorentz transform the t - matrix as we have mentioned above the positive energy on - shell t - matrix is projected onto five invariant amplitudes . by the restriction to positive energy states an ambiguity arises @xcite which is discussed in terms of the pseudo - scalar and the pseudo - vector representation of the t - matrix in the dirac space . from analogy to the one - pion exchange coupling in the nucleon - nucleon potential the pseudo - vector choice is more natural and further it is known from meson theory analysis @xcite that this choice suppresses the coupling to the negative - energy states . however , the difference between both choices also states a measure for the uncertainty concerning the determination of the self - energy components due to the restriction on positive energy spinors . one issue of the present work is to estimate this uncertainty . thus , the aim of this work is to determine the self - energy components , to estimate their uncertainty due to different covariant representations of the t - operator and different nucleon - nucleon potentials and to examine the sensitivity of some commonly used observables , i.e. , the equation of state and the optical potential on the self - energies . we are investigating in detail the density and the momentum dependence of the scalar and the vector components and compare our results to those of other groups @xcite . thereby the momentum dependence can just be determined in a first approximation because it is neglected in the self - consistency treatment of the brueckner model . commonly this approximation is justified by the assumption of a weak momentum dependence , at least inside the fermi sea , but there have been also indications that this assumption depends on the chosen one - boson exchange potential @xcite . we show here for the first time the momentum dependence over its full range , i.e. , below and above the fermi surface , and we find it in the case of the bonn a , b , c potentials of the order of several 100 mev , i.e. , to be strikingly strong . in difference to the most publications in the field we focus our discussion to the self - energy components themselves . it is meanwhile well known that the nuclear matter saturation properties are much better reproduced in the relativistic than in the non - relativistic brueckner approach , i.e. , the predictions for the saturation point of a variety of nucleon - nucleon potentials are located along the _ coester band _ which meets in the relativistic case the empirical area @xcite . the same also holds , e.g. , for the nucleon optical potential . however , as we also find in the present analysis these commonly studied observables are not really sensitive to the self - energy components since they mainly depend on a cancellation of the components . the situation changes in the application of the brueckner theory beyond infinite nuclear matter . when applying relativistic brueckner results to the description of finite nuclei , e.g. , in the framework of an effective ( density dependent ) relativistic mean field theory @xcite the self - energies start to play a decisive role . this fact is , however , most pronounced in the investigation of heavy ion collisions . different from resting nuclear matter here the scalar and vector components are out of balance which is due to their different lorentz transformation character and diverse observables are in particular sensitive to the magnitude of the fields @xcite . furthermore the momentum space configuration can be highly anisotropic and rather corresponds to two counterstreaming currents of nuclear matter which can be described by two lorentz - elongated fermi ellipsoids . approximate determinations of relativistic brueckner mean fields for such configurations have been performed in @xcite and encouraging applications to heavy ion collisions in @xcite . thus the density and momentum dependence of the nuclear self - energy components as the quantity which most prominant characterizes the medium effects is a question of particular interest . this paper is organized in the following way : in the next section we review the relativistic brueckner approach . there are still some technical differences in the _ standard _ brueckner model between ref . @xcite and refs . here we follow the approach of horowitz and serot @xcite which allows the use of standard techniques for the solution of the thompson equation known from earlier works @xcite . however , we use the modern one - boson exchange potentials , treat the isospin channels and also the real and imaginary part of the t - matrix separately and project on both , the pseudo - scalar and the pseudo - vector representation . in the following section we discuss our numerical results . the calculations with the @xmath0-potential are compared with results from ref . we further discuss the density and momentum dependence of the scalar and vector self - energy , the influence of pseudo - scalar and pseudo - vector choice is analyzed and we compare to the results of brockmann and machleidt @xcite and the groningen group @xcite . further the equation of state and the optical potential are shown and their sensitivity to the self - energy components is analyzed . in the final section a summary is given . in the relativistic brueckner approach the nucleon inside the nuclear medium may be viewed as a dressed particle in conseqeunce of its two - body interaction with the surrounding nucleons . this problem is stated as a coupled set of three non - linear integral equations @xmath1 - g t ) . \label{eq3}\end{aligned}\ ] ] the t - matrix is calculated in the ladder approximation of the bethe - salpeter equation ( [ eq1 ] ) and the bare nucleon - nucleon interaction is described by one - boson exchange potentials @xmath2 . the intermediate nucleons are given by the two - body propagator @xmath3 which is , however , usually replaced by an effective propagator @xcite , i.e. , the blankenbecler - sugar propagator or the thompson propagator . in the present work the thompson choice is applied . the pauli operator @xmath4 accounts for the influence of the medium by the pauli principle and projects the intermediate scattering states outside the fermi sea . the dressed one - body green propagator @xmath5 is calculated via the dyson equation ( [ eq2 ] ) from the free propagator @xmath6 . the influence of the coupling to the surrounding nucleons is expressed by the self - energy @xmath7 of the nucleon . in the brueckner formalism this self - energy is determined by summing up the direct and exchange interactions with all the nucleons inside the fermi sea @xmath8 , see eq . ( [ eq3 ] ) . here the t - matrix plays the role of an effective medium dependent two - body interaction . in the relativistic brueckner approach one introduces additional approximations for solving the equations ( [ eq1]-[eq3 ] ) which are sketched in the following , for further reading see , e.g. , refs . @xcite . the most general form of the lorentz - structure of the self - energy compatible with translational invariance , hermiticity , parity conservation , time - reversal invariance and rotational invariance ( in the nuclear matter rest frame ) is @xmath9 with a scalar part @xmath10 and a vector part @xmath11 . due to the dyson equation ( [ eq2 ] ) the full green function has the formal solution @xmath12\ ] ] where we have introduced the effective mass and the kinetic momentum @xmath13 re and i m denote real and imaginary part since ( above the fermi surface ) the self - energy in general will be complex . here we adopt the _ quasiparticle approximation _ @xcite , i.e. , the i m @xmath14 $ ] will be neglected in eq . ( [ eq5 ] ) . this means that the decay width of the dressed nucleon is set equal to zero , resulting in an infinite lifetime of this quasiparticle. in the relativistic brueckner approach the full green function @xmath5 is replaced by its positive energy part . thus the self - energy in eq . ( [ eq3 ] ) is determined by the part proportional to the fermi sea @xmath8 , the so - called dirac green function @xmath15 @xcite @xmath162\pi i \delta(k^{\ast 2 } - m^{\ast 2}(k))\theta(k^{\ast 0 } ) \theta(k_f-|{\bf k}|)\ ] ] which eliminates the divergent contributions from the negative energy sea . however , in the form of eq . ( [ eq7 ] ) @xmath15 is restricted to the description of nuclear matter at rest . in order to achieve general convenience the fermi sphere @xmath17 has to be replaced by an fermi ellipsoid @xmath18 with the fermi energy @xmath19 and the boost 4-velocity @xmath20 @xcite . in eq . ( [ eq7 ] ) the momenta of the quasiparticles are put on mass - shell , which is expressed by the @xmath21-distribution and thus @xmath22 . hence the self - energy @xmath23 depends only on the three - momentum @xmath24 and , of course , on the respective density or @xmath25 . furthermore , in the relativistic brueckner model one applies a mean field approximation to @xmath7 , eq . ( [ eq4 ] ) , which allows a simple formulation of the self - consistency problem . the explicit momentum dependence of the self - energy which enters via the term @xmath26 can be dealt with by introducing the reduced kinetic momentum @xmath27 and the reduced effective mass @xmath28 . further one neglects the momentum dependence of the effective mass ( @xmath29 or @xmath30 ) . thus , the nucleons fulfill a quasi - free dirac equation @xmath31 u_{\lambda}(k ) = 0\ ] ] and using the normalization of ref . @xcite the self - consistent positive - energy spinors of helicity @xmath32 are defined as @xmath33 with @xmath34 being a pauli spinor . the dirac spinors depend on the effective mass and thus on the nuclear density . the matrix - elements of the t - matrix and the one - boson exchange potentials @xmath2 , eq . ( [ eq1 ] ) , are calculated with these spinors . in contrast to non - relativistic brueckner theory this fact introduces an additional density dependence into the interaction which is one major reason for the great success of the relativistic treatment . this density dependence is mediated by the additional parameter @xmath35 which is fixed at a reference point , usually at @xmath36 . @xmath35 is obtained as the solution of the non - linear equation ( _ reference spectrum approximation _ ) @xmath37 which follows from the definition given above . self - consistency is achieved by determining for a start value of @xmath30 the medium - dependent spinors of eq . ( [ eq9 ] ) , next the t - matrix from eq . ( [ eq1 ] ) and finally the self - energy and the new @xmath30 . ] . this procedure is repeated until convergence is reached . the t - matrix , in eq . ( [ eq1 ] ) , is most easily determined in the two - particle center - of - mass ( c.m . ) system while the self - energy , eq . ( [ eq3 ] ) , is calculated in the nuclear matter rest frame . therefore we will determine in a second step covariant amplitudes of the t - matrix . the thompson propagator ( and similar the blankenbecler sugar propagator ) projects the intermediate nucleons onto positive energy states and restricts the exchanged energy transfer by @xmath38 to zero . thus equation ( [ eq1 ] ) is reduced to a three dimensional integral equation of the lippmann - schwinger type , the so called thompson equation @xcite @xmath39 here the thompson equation is given in the two - nucleon c.m .- frame where @xmath40 is the relative momentum of the final states and similar @xmath41 are the relative momenta of the initial and intermediate states , respectively . the thompson propagator and similar the blankenbecler - sugar propagator imply that the timelike component of the momentum transfer in @xmath2 and @xmath42 is set equal to zero which is a natural constraint in the c.m .- frame , however , not a covariant one . hence we solve the thompson equation in the c.m .- frame in contrast to , e.g. , the approach of ref . @xcite . the starting energy in eq . ( [ eq11 ] ) is already fixed : @xmath43 . the pauli operator @xmath4 explicitely depends on the chosen frame , i.e. , on the boost 3-velocity @xmath44 into the c.m .- frame . with @xmath45 we denote the set of parameters @xmath46 on which the t - matrix actually depends . we solve the thompson equation ( [ eq11 ] ) for the on - shell t - matrix @xmath47 in the c.m .- system and thereby apply standard techniques which are described in detail by erkelenz @xcite . we construct the positive - energy helicity t - matrix elements from the @xmath48-scheme . in the on - shell case only five of the sixteen helicity matrix elements are independent which follows from general symmetries , see ref . @xcite . after a partial wave projection onto the @xmath48-states the integral reduces to a one - dimensional integral over the relative momentum @xmath24 and eq . ( [ eq11 ] ) decouples into three subsystems of integral equations for the uncoupled spin singlet , the uncoupled spin triplet and the coupled triplet states . the pauli operator @xmath4 is replaced by an angle averaged pauli operator @xmath49 @xcite . since the fermi sphere is deformed to a fermi ellipsoid in the two - nucleon c.m .- frame @xmath49 is evaluated for such a fermi - ellipsoid . @xmath50 with @xmath51 and @xmath52 . we are solving the integral equations by the matrix inversion techniques of haftel and tabakin @xcite . real and imaginary parts of the t - matrix are calculated separately by the principal - value treatment given in ref . @xcite . due to the antisymmetry of the two - fermion states we can restore the total isospin i of the two - nucleon system ( i=0,1 ) with the help of the selection rule : @xmath53 . from the five independent on - shell amplitudes in the @xmath48-representation the five independent partial wave amplitudes in the helicity representation ( for i=0,1 and real and imaginary part separately ) are obtained by inversion of eq . ( 3.32 ) and then of eq . ( 3.28 ) of ref . the summation over the total angular momentum @xmath54 ( we have taken @xmath55 and for comparison @xmath56 ) yields the full helicity matrix element @xmath57 d^j_{\lambda \lambda^{\prime } } ( \theta ) < |{\bf p}|\lambda^{\prime}_1 \lambda^{\prime}_2 |t^{j,{\rm i}}(x)| \ , |{\bf q}| \lambda_1 \lambda_2 > = < { \bf p } \lambda^{\prime}_1 \lambda^{\prime}_2 { \rm i } { \rm i}_3 | t(x ) @xmath58 is the scattering angle between @xmath59 and @xmath60 and @xmath61 . the reduced rotation matrices @xmath62 are those defined by rose @xcite . the matrix element ( [ eq12 ] ) is actually independent of the third component of the isospin @xmath63 . the next step is the projection of the five independent helicity amplitudes onto five covariant amplitudes . therefore the t - operator is expanded into the basis matrices in the dirac spinor space @xmath64 the subscripts @xmath65 denote the particle on which the matrix acts . this expansion , however , is not unique since we have specified the dirac matrix structure of the t operator by its action only on positive energy states . the expansion of eq . ( [ eq13 ] ) is given with @xmath66 a pseudo - vector interaction @xmath67 . it is defined as @xmath68 with the transferred momentum at the vertex @xmath69 where @xmath70 indicates final and initial momentum . we will , however , also adopt a pseudo - scalar interaction where the fourth term in eq . ( [ eq13 ] ) is replaced by @xmath71 as it was also done by horowitz and serot @xcite . the pseudo - vector choice is , e.g. , made by the groningen group . this choice is supported by the argument that it agrees with the one - pion exchange coupling in the nucleon - nucleon potential and furthermore suppresses the coupling to the negative energy states which are neglected in the brueckner approach @xcite . in the present work we apply both variants since we want to examine the influence of the different choices on the self - energy . next we take the helicity matrix elements of eq . ( [ eq13 ] ) . on the r.h.s . the helicity states are acting directly on the matrix operators which are now abbreviated by @xmath72 for i=1,2 and @xmath73 @xmath74 the helicity matrix elements on the r.h.s . are given explicitely , e.g. , in eqs . ( 2.11 - 2.17 ) of ref . for on - shell scattering between positive energy states the matrix elements of the pseudo - scalar ( ps ) and the pseudo - vector ( pv ) matrix operators are identical @xcite . this vertex equivalance follows immediately from the dirac eq . ( [ eq8 ] ) : @xmath75 . thus the covariant amplitudes @xmath76 are identical as well . the difference between both representations of the t - matrix in dirac space enters only at the one - body level when calculating the self - energy @xcite . for fixed parameters @xmath77 equation ( [ eq14 ] ) is a matrix relation between the five independent helicity amplitudes @xmath78 ( five of the sixteen amplitudes numbered with @xmath79 ) on the l.h.s . and the five unknown covariant amplitudes @xmath80 @xmath81 this corresponds to eq . ( 3.23 ) of ref . @xcite with respect to our different normalization of the spinor basis , eq . ( [ eq9 ] ) . the covariant amplitudes are determined by a matrix inversion of equation ( [ eq15 ] ) . this , however , only has to be done for two angles since for on - shell scattering in the two - particle c.m .- frame only two scattering angles appear , i.e. , @xmath82 for the direct and @xmath83 for the exchange interaction . however , now the matrix becomes singular because two of the five helicity amplitudes vanish . therefore it is useful to extract the leading angular dependence of the matrix @xmath84 as well as of the helicity amplitudes @xmath78 from the rotation matrices @xmath85 in eq . ( [ eq12 ] ) . thus , the limit @xmath86 can be performed analytically . this method has been developed in appendix c of ref . @xcite and formula ( [ eq15 ] ) can be found there for the limiting angles @xmath87 ( eqs . ( c.10,11 ) ) . proceeding this way we determine the real and the imaginary part of the five covariant amplitudes for the direct and exchange contribution in both , the isospin singlet and triplet channels of the t - matrix @xmath88 . now we are able to calculate the self - energy . inserting the green function , eq . ( [ eq7 ] ) , and the dirac representation of the t - matrix , eq . ( [ eq13 ] ) , into the definition of the self - energy , eq . ( [ eq3 ] ) , yields @xmath89 \right . \\ \label{eq16a } \left . + { \not q}^{\ast}_{\alpha\beta } \left [ 4 t^{\rm v}_d - t^{\rm s}_x + 2 t^{\rm v}_x + 2 t^{\rm a}_x - \frac{(k^{\ast\mu } - q^{\ast\mu})^2}{4m^{\ast 2 } } t^{\rm p}_x\right ] - ( { \not q}^{\ast } - { \not k}^{\ast})_{\alpha\beta } \frac{2 q^{\ast}_{\mu } ( k^{\ast\mu } - q^{\ast\mu})}{4m^{\ast 2 } } t^{\rm p}_x \right\ } \end{aligned}\ ] ] where the @xmath90 denote both , dirac ( spin ) and isospin indices . the direct and exchange parts of the t - matrix denoted by @xmath91 and @xmath92 are calculated from isospin sums of the isospin projection operators which yields @xmath93 and @xmath94 . the self - energy as a one - body operator in spinor space posseses components proportional to the unity matrix 1 and to @xmath95 as postulated in equation ( [ eq4 ] ) . from this relation the self - energy components @xmath96 follow by taking traces @xmath97 \qquad = \int \frac{d^3{\bf q}}{(2\pi)^3 } \theta(k_f-|{\bf q}| ) \frac{m^{\ast}}{e^{\ast}({\bf q } ) } t_s({\bf k , q};x ) , \\ \label{eq17 } \sigma_{0}(|{\bf k}|;m^{\ast},k_f ) & = & - \frac{1}{8 } tr[\gamma^{0 } \sigma ] \quad = \int \frac{d^3{\bf q}}{(2\pi)^3 } \theta(k_f-|{\bf q}| ) t_{(0)}({\bf k , q } ; x ) , \\ \label{eq18 } \sigma_{v}(|{\bf k}|;m^{\ast},k_f ) & = & - \frac{1}{8 } tr[\mbox{\boldmath$\gamma$\unboldmath } \cdot{\bf k } \sigma ] = \int \frac{d^3{\bf q}}{(2\pi)^3 } \theta(k_f-|{\bf q}| ) \frac{{\bf q}^{\ast}\cdot{\bf k}}{|{\bf k}|^2 e^{\ast}({\bf q } ) } t_{v}({\bf k , q};x ) \end{aligned}\ ] ] where we have used the abbreviations @xmath98 , \\ \label{eq20 } t_{(0)}({\bf k , q } ; x ) & = & \frac{1}{4}\left[- 4 t^{\rm v}_d + t^{\rm s}_x -2 t^{\rm v}_x - 2 t^{\rm a}_x + \frac{e^{\ast}({\bf k})}{e^{\ast}({\bf q } ) } \frac{m^{\ast 2 } - k^{\ast}_{\mu } q^{\ast\mu}}{2m^{\ast 2 } } t^{\rm p}_x\right ] , \\ \label{eq21 } t_{v}({\bf k , q } ; x ) & = & \frac{1}{4}\left[- 4 t^{\rm v}_d + t^{\rm s}_x -2 t^{\rm v}_x - 2 t^{\rm a}_x + \frac{|{\bf k}|}{q_z } \frac{m^{\ast 2 } - k^{\ast}_{\mu } q^{\ast\mu}}{2m^{\ast 2 } } t^{\rm p}_x\right ] .\end{aligned}\ ] ] in the pseudo - scalar case @xcite all the cofactors of @xmath99 are replaced just by -1 which yields @xmath100 . the self - energy of a nucleon is calculated in the nuclear matter rest frame by integrating the effective two - body interaction over the fermi sea . the 3-momenta @xmath101 of a nucleon pair are given in this frame and we define @xmath102 . on the other hand the covariant amplitudes are determined in the two - nucleon c.m . system and thus depend on the respective quantities in that frame @xmath103 from the invariant mass @xmath104 , the relative momentum in the c.m .- frame follows by @xmath105 . as already mentioned above , @xmath45 abbreviates the set of parameters @xmath106 , where the boost 3-velocity ( from the nuclear matter into the c.m .- frame ) is now given by @xmath107 . in the two - particle c.m .- frame there exist only two possible scattering angles , i.e. , @xmath82 for the direct and @xmath83 for the exchange amplitude . numerically we solve the thompson equation ( [ eq11 ] ) and determine the covariant amplitudes @xmath80 for a suitable two - dimensional range of the quantities @xmath108 . using the azimuthal symmetry we calculate the self - energy , eqs . ( [ eq16]-[eq18 ] ) , by a two - dimensional integral over the fermi sphere . therefore for each nucleon pair we have to interpolate the covariant amplitudes at the respective values of @xmath109 and @xmath110 . the resulting self - energy , taken at the fermi surface @xmath36 , yields the new value of the effective mass @xmath29 , eq . ( [ eq10 ] ) . this value serves as the input for the next iteration . this procedure is repeated until a convergence of the effective mass is achieved . for the solution of the thompson equation ( [ eq11 ] ) we apply the bonn a , b , c potentials of refs . @xcite as the nucleon - nucleon interaction @xmath2 . the potentials are calculated with the obnns code of r. machleidt @xcite . the bonn potentials are based on the exchange of the six nonstrange bosons with masses below 1 gev @xmath111 . for the pion a derivative pseudovector coupling is applied . the three parametrizations a , b and c differ essential in the @xmath112nn form - factor and as a consequence in the strength of the nuclear tensor force . actually we find only minor differences of a few mev in the resulting self - energies , see below , and thus we present mainly the bonn c results . for comparison with the results of horowitz and serot we also employ the simple @xmath0-meson exchange of ref . @xcite with @xmath113 and use the form - factor given therein . first we want to demonstrate the reliability of our approach by comparison with the results of ref . @xcite . in fig.1 we show the self - consistent effective mass @xmath29 , eq . ( [ eq10 ] ) , as a function of the fermi momentum @xmath25 which is a measure for the density @xmath114 . our results with the @xmath0-meson exchange potential and the pseudo - scalar decomposition of the t - matrix reproduce quite well the values taken from ref . @xcite . the slight deviation for densities above saturation density ( here @xmath115 is partly due to the different effective two - nucleon propagators . as mentioned above , we are using the thompson propagator , eq . ( [ eq11 ] ) , while in ref . @xcite the blankenbecler - sugar propagator is applied . furthermore , in ref . @xcite an additional approximation has been made . as discussed above , we are solving the thompson equation for a two - dimensional grid of the relative momentum @xmath116 and the c.m .- momentum @xmath117 respectively the boost - velocity @xmath118 and use this t - matrix for the determination of @xmath7 . in ref . @xcite the @xmath117-dependence is replaced by an `` average @xmath117 approximation '' for each @xmath116 which causes deviations mainly at high density systems . for the bonn c potential the effective mass is significantly different from that of the @xmath0-model . here we applied the pseudo - vector choice . in fig.2 we show the density dependence of the self - energy , i.e. , the scalar part @xmath10 as well as the time - like @xmath119 and the space - like component @xmath120 of the vector self - energy . the self - energies are determined for the three bonn potentials a , b , c and again we the pseudo - vector choice is applied . for the sake of a better comparison the sign of the vector self - energies is inversed and the dimensionless space - like component @xmath121 is multiplied by @xmath25 . the scalar and also the vector self - energy component @xmath122 differ only slightly for the three bonn potentials . the difference amounts about 5 mev at saturation density ( here @xmath123 ) and is still less than 10 mev at @xmath124 . in absolute magnitude bonn c yields the weakest and bonn a the strongest self - energies . the space - like vector self - energy @xmath120 is equal within line - width for the three bonn potentials and is small in comparison to @xmath125 , e.g. , about 7% of @xmath126 at saturation density . for further analysis we therefore concentrate on the large components @xmath125 calculated with the bonn c potential . the two different representations of the t - matrix , i.e. , the pseudo - scalar and the pseudo - vector case , represent an inherent ambiguity . this is , however , a common feature of all relativistic brueckner calculations which specify the dirac matrix structure of the t - matrix by using only positive energy t - matrix elements . in order to clarify the influence of this ambiguity in figs.3 . we show the self - energy components @xmath127 calculated exemplaryly with the bonn c potential for both choices . in the pseudo - scalar case , see fig.3 , the self - energies are larger by about 40 - 80 mev ( decreasing with increasing density ) than in the pseudo - vector case ( fig.4 ) . this difference may be interpreted as a measure of the general uncertainty of the self - energy components arising from any particular projection of the t - operator on positive energy matrix elements only . this uncertainty is larger at low densities . to get more insight into the origin of this difference we decompose the self - energy in the two contributions stemming from the direct and the exchange t - amplitude respectively . e.g. , in the pseudo - scalar case the direct part of the self - energy @xmath10 is determined by the integral over @xmath128 while the exchange part is given by the integral over the sum of all exchange - amplitudes @xmath129 , see eq . ( [ eq19 ] ) and the subsequent comment . we find that in the pseudo - scalar case direct and exchange parts of @xmath127 are identical within less than 0.5 mev , see fig.3 . however , in the pseudo - vector case the factor accompanying the @xmath99 amplitude in eqs . ( [ eq19],[eq20 ] ) reduces the exchange part by roughly 50% . for comparison in the @xmath0-model , which lacks of pion exchange , the @xmath99 amplitude is generally smaller and the effect , i.e. , the reduction of the exchange part , amounts to roughly 25% . of course also the direct part is different from that of the pseudo - scalar case since we consider the self - consistently determined fields . we conclude that the pseudo - vector choice in general reduces the total self - energy which is due to a considerable reduction of the exchange contributions while the direct contributions are even enhanced . in fig.5 we compare our results for @xmath122 obtained with the pseudo - vector choice and the bonn c potential with the results of two other groups , i.e. , the groningen group @xcite and the approach of brockmann and machleidt @xcite . in ref . @xcite the relativistic thompson equation is solved by standard methods @xcite directly in the nuclear matter rest frame instead of the two - particle c.m .- frame . therefore the thompson eq . ( [ eq11 ] ) is galilei transformed to the rest frame with the c.m .- momentum @xmath130 . the relativistic single particle potential @xmath131 in the nuclear matter rest frame can be evaluated from this t - matrix by @xmath132 with the definition of @xmath133 as in eq . ( [ eq3 ] ) . the advantage of this method is that one is not forced to lorentz transform the matrix elements between the nuclear matter rest frame and the two - particle c.m .- frame and thus a projection onto covariant amplitudes is obsolete which simplifies the task considerably . however , now the self - energy components @xmath122 can not be calculated directly as in the projection method of eqs . ( [ eq16]-[eq18 ] ) . instead , one has to extract them from the single particle potential @xmath131 . with the form of @xmath7 as in eq . ( [ eq4 ] ) , however , thereby neglecting @xmath121 one finds @xmath134 i.e. , the single particle potential results from a cancellation of the huge scalar and vector components . using a mean field ansatz for @xmath127 , i.e. , neglecting their momentum dependence , one is able to determine them by a fit procedure to @xmath135 . in ref . @xcite the self - energy components are constructed by this potential - fit method . in fig.5 we compare these results with our calculations both for the bonn c potential . in the density range considered the fitted self - energies are generally smaller in absolute values by 50 up to more than 150 mev , e.g. , at saturation density they are reduced by about 75 mev . in spite of this discrepancy both calculations have several features in common . e.g. , the single particle potential , defined in eq . ( [ eq24 ] ) , is almost identical and as we will see later on the same holds for the equation of state . concerning these two quantities the main difference between both treatments lies in the reference frame where the t - matrix is actually calculated while the projection method on covariant amplitudes is not essential . at moderate densities the t - matrix depends only weakly on the c.m .- momentum @xmath136 and respectively on the chosen frame . thus the numerical results show only minor differences . another feature in common is that also the potential - fit method yields self - energies for the three types of bonn potentials which are equal within a few mev @xcite . in fig.5 we further compare to the results of the groningen group @xcite . they also perform a full relativistic brueckner calculation solving the thompson equation in the two - particle c.m .- system and using a projection of the t - matrix onto a pseudo - vector representation . the difference to our method is on the one hand the application of a different one - boson exchange potential including different form - factors , the so called groningen potential . on the other hand as - a more technical point - in ref . @xcite the thompson equation is solved in full momentum - spin space by the use of pad approximants , for further details see , e.g. , ref . @xcite . for densities above saturation density the self - energies calculated with the groningen potential are roughly the same as the ones obtained by us , however , in the low density range they are much smaller in magnitude . up to now we have discussed the self - energy of a nucleon in nuclear matter obtained at the fermi surface ( @xmath36 ) , i.e. , the density dependence of the self - energy . in the relativistic brueckner theory the self - energy at the fermi surface and the t - matrix are determined self - consistently , as described above . the dressed nucleon propagators respectively the self - consistent spinor basis , eq . ( [ eq9 ] ) , entering into the calculation of the t - matrix are constructed with the effective mass , eq . ( [ eq10 ] ) . @xmath29 itself depends on the density dependent self - energies @xmath137 . thus only the effective mass acts as a self - consistency or iteration parameter . in a next step the self - energies above and below the fermi surface can be evaluated from eqs . ( [ eq16]-[eq18 ] ) . this has to be considered just as the first iteration step in the determination of self - consistent momentum dependent functions @xmath138 and stands in contrast to the non - relativistic brueckner treatment where self - consistency is required for the full positive single particle energy spectrum . however , these approximations are not based on the same footing . in the relativistic treatment we have an intrinsic momentum dependence due to the dirac structure of the nucleon which introduces a momentum dependence in the single particle potential @xmath131 even for constant values of @xmath96 , see eq . ( [ eq25 ] ) . a further argument often given in connection with the reference spectrum approximation @xcite is that the momentum dependence of the self - energies in the relativistic brueckner approach is found to be rather soft . however , in the present work we demonstrate that this is in general not the case but strongly depends on the special form of the one - boson exchange potential . in fig.6 we show the momentum dependence of the nucleon self - energies @xmath127 at saturation density calculated with the @xmath0-exchange potential . our results nicely agree with the results of ref . @xcite within 10 mev which might be due to different technical details already mentioned in the discussion of fig.1 . we find that @xmath10 decreases in the range from @xmath139 to @xmath25 by 14 mev and @xmath140 by 24 mev , i.e. , below the fermi surface the self - energies are constant within a few percent . in fig.7 we show the momentum dependence of the self - energies for calculations with the bonn c potential ( pseudo - vector choice ) and the @xmath0-potential ( pseudo - scalar choice ) , both at @xmath141 gev . more strictly this is the real part of the self - energy since for momenta greater than the fermi momentum the self - energy is complex . the results obtained with the bonn c potential show a strong momentum dependence : over the entire momentum range up to 0.8 gev corresponding to a single particle energy of around 300 mev the self - energies decrease by nearly 60% with respect to the central value . the most pronounced change occurs in the region around the fermi surface while deep inside the fermi sea and far above the fermi momentum the momentum dependence nearly saturates . for the @xmath0-potential the self - energies are rather weakly momentum dependent : over the full interval they decrease by about 20% . for comparison we also show two results obtained by the groningen group . in ref . @xcite the self - energies above the fermi surface are calculated in a similar way as in the present work , see discussion of fig.5 . the results obtained by the groningen group @xcite are in general softer k - dependent than those of the present work obtained with the bonn potentials . the results of ref . @xcite show a momentum dependence as strongly as ours . however , in this case a comparison is not fully appropriate since these calculations include an additional self - consistent treatment of the pion polarisation . in fig.8 we show the equation of state , i.e. , the energy per particle e / a as a function of the nuclear matter density @xmath142 , calculated with bonn a , b , c . in the relativistic brueckner theory the energy per particle is defined in analogy to the non - relativistic hartree - fock method as the kinetic plus half of the potential energy @xmath143 - m \end{aligned}\ ] ] with the self - consistent spinors @xmath144 given in eq . ( [ eq9 ] ) . by this way we have determined e / a by integrating the self - energy components of eqs . ( [ eq16]-[eq18 ] ) over the fermi sphere , see eq . ( [ eq27 ] ) . in fig.8 we compare to the results of ref . @xcite where e / a is calculated directly from the single particle potential , eq . ( [ eq24 ] ) , using the expression ( [ eq26 ] ) . for moderate densities the results agree with an accuracy better 0.5 mev and even for @xmath145 the deviation is less than 3 mev . firstly , this result is an additional test for the accuracy of our numerical treatment , i.e. , the projection of the t - matrix onto covariant amplitudes which are subsequently used to determine the self - energy components and the equation of state . secondly , these results again demonstrate that the equation of state is quite insensitive to the self - energy components themselves . actually , as can be seen clearly from the definition eq . ( [ eq26 ] ) , the equation of state depends directly on the single - particle potential , i.e. , the difference of @xmath10 and @xmath126 , eq . ( [ eq25 ] ) , and relatively weak on the effective mass via the kinetic energy part . nevertheless , the relativistic effects are responsible in contrast to non - relativistic calculations that the minima of the equations of state for the three different bonn potentials reveal a coester band which meets the empirically found saturation area of nuclear matter . in order to demonstrate the influence of the two different choices , i.e. , pseudo - vector and pseudo - scalar on the equation of state in fig.9 we show the respective equation of state , again for bonn c. actually the expectation value @xmath146 of the full self - energy operator , eq . ( [ eq16a ] ) , is identical for both choices @xcite . however , as can be seen from fig.9 the equation of state turns out to be rather sensitive on the different choices . this significant dependence is only due to the self - consistently iterated effective mass which differs in both approaches . this effect deepens the equation of state in the pseudo - scalar choice by around 1 to 2 mev and shifts the minimum towards higher densities . thus the saturation point is shifted away from the empirical point towards the results of non - relativistic brueckner calculations as , e.g. , published in ref . this observation additionally supports the pseudo - vector choice . in view of the strong momentum dependence of the self - energy components for the bonn potentials one might expect that the results depend sensitive on the actual momentum reference value where the selfconsistent effective mass is determined . as a consequence a selfconsistent treatment of the full momentum dependence might appear to be indispensable . therefore we have also examined the influence of a different choice of the selfconsistent effective mass . in the reference spectrum approximation , ( [ eq10 ] ) , @xmath29 is determined at the fermi surface which is the conventional treatment in the relativistic brueckner approach . however , this choice is by no means unique . thus we also determine @xmath29 by its mean value in analogy to eq . ( [ eq10 ] ) @xmath147 the self - energies @xmath148 are not simply taken at @xmath36 but replaced by their values averaged over the fermi sea @xmath149 these averages are lorentz scalars . the extension of the reference spectrum approximation , eq . ( [ eq28 ] ) , is especially meaningful with respect to the treatment of anisotropic nuclear matter , as done in ref . @xcite . here it serves as a test of the influence of the momentum dependence . due to the strong momentum dependence observed , e.g. , in fig.7 the mean value @xmath150 differs considerably from @xmath151 , i.e. , by about 57 mev at saturation density ( with bonn c ) . respectively for the different effective masses we obtain a mean value @xmath152 mev which is smaller than the selfconsistent mass @xmath153 mev . but if we use the alternatively defined effective mass , eq . ( [ eq28 ] ) , as iteration parameter we find for the final iterated mean value @xmath154 mev . thus the mean value is increased , i.e. , it is shifted towards the previous self - consistent result or , in other words , the selfconsistent value is rather stable . the comparison of the self - energy components @xmath155 in both iterations , see fig.2 , shows just slightly deviations , e.g. , about 10 mev at saturation density . and even more , the equation of state , shown in fig.9 , remains nearly unchanged . these results demonstrate that the strong momentum dependence has only minor influence on the density dependence of the self - energy components and that the equation of state remains unaffected . another quantity which is more sensitive on the momentum dependence of @xmath96 is the schroedinger equivalent optical potential which a nucleon feels inside the nuclear medium . the optical potential is given by @xmath156 it contains counteracting linear and quadratic terms in the self - energy components . for nucleons with @xmath157 the momentum space is not completely blocked for scattering processes and thus above the fermi surface the self - energies and the optical potential become complex . the single - particle energy @xmath158 is a monotonic function of @xmath24 which allows by inversion to determine @xmath159 as function of the single - particle energy @xmath160 . we want to stress that neglecting the momentum dependence of the fields @xmath161 and neglecting completely @xmath121 the optical potential @xmath162 would simply be a linear function of @xmath160 . however , this is not the case here . real and imaginary part of the optical potential calculated for bonn c ( pseudo - vector choice ) are shown in fig.10 . the real part agrees well with the groningen results of ref . while the imaginary part of the present calculations is about twice as large . further we compare to the empirical optical potential hama i , table 2 of ref . @xcite . for the full range of single - particle energies up to 700 mev , shown in fig.10 , our results deviate from the empirical re @xmath162 by less than 10 mev . up to the pion production threshold at around 300 mev we observe a remarkable agreement of the imaginary part of @xmath162 with the empirical data . for single - particle energies above this region the results appear to be no more fully reliable since then meson - nucleon resonances , e.g. the @xmath163(1232 ) resonance , should be taken into account . in this work in particular the self - energy of a nucleon inside the nuclear medium has been studied in the framework of the relativistic brueckner approach . therefore the dependence of the lorentz scalar and vector self - energy components on the nuclear density and the nucleon momentum have been examined . as bare nucleon - nucleon interaction various one - boson exchange potentials have been employed , i.e. , the bonn a , b , c potentials and for comparison the @xmath0-exchange potential . for the latter we are able to nicely reproduce the density dependence of the self - energies of ref . the self - energy components themselves deviate for the three different bonn potentials over a wide density range only by a few mev and thus we restricted our discussion mainly to the bonn c potential . from the comparison with our results we find that the potential - fit method of ref . @xcite underpredicts the magnitude of the self - energies by about 50 up to more than 150 mev . further we analyzed the influence of different possible covariant representations of the t - matrix . the positive energy on - shell t - matrix is projected on five invariant amplitudes and by the restriction to positive energy states an ambiguity arises which is discussed in terms of the pseudo - scalar and the pseudo - vector representation . although the t - amplitudes of both choices are equal on the one - body level , i.e. , concerning the respective self - energy components a significant influence of these representations is observed . this is due to the fact that in the pseudo - scalar case direct and exchange contributions to the self - energies are equal while in the pseudo - vector choice the exchange part is suppressed . for the bonn potentials this suppression lies between 40 - 80 mev and is most prominant at low densities . this difference between both choices has to be interpreted as an inherent measure for an uncertainty of the approach due to the restriction on positive energy spinors . commonly the momentum dependence of the self - energy components is supposed to be weak . this assumption is also reflected in the self - consistency treatment in the relativistic brueckner model which is managed by just one density dependent parameter , i.e. , the self - consistent effective mass . therefore we have carefully examined this momentum dependence which , however , can be done within the bruckner scheme just in a first approximation . as in ref . @xcite we find that taking the @xmath0-exchange potential the self - energies are found to be constant below the fermi surface within a few percent . however , this result does not hold for the more realistic bonn potentials . here we observe a strong momentum dependence which may be due to the strong pion exchange . the self - energies decrease from their central value , i.e. , at @xmath164 to the fermi surface by about one fourth and further to @xmath165 gev by more than one half . this striking momentum dependence makes the treatment of the self - consistency in the model somewhat questionable . but a test with a different defined self - consistent effective mass which is determined by a mean value over the fermi sea and not taken at @xmath25 as usually done shows that the self - energy components are rather stable at not too high densities . thus the self - consistency treatment in the model still appears to be justified up to two or three times saturation density . in this context the question raises which observables are sensitive to the self - energy components and their momentum dependence . it is remarkable that the full self - energy and , respectively , the single particle potential show ( absolutely ) a much weaker momentum dependence . as it is well known from mean field theory the large scalar and vector self - energy components contribute to these quantities with opposite signs and thus cancel each other to hight extent . the resulting observables , e.g. , the single particle potential are smaller by nearly one order of magnitude . here we found that this holds not only at @xmath25 but also over the full momentum range considered . thus the momentum dependence cancels to a large extent . also the equation of state depends mainly on the difference of the scalar and the vector self - energy and thus the absolute values of the single component are of minor importance . this can be seen most clearly by the use of just density dependent self - energy components determined with the potential - fit method of brockmann and machleidt @xcite which deviate from our results ( at @xmath25 ) by about 75 mev . the resulting binding energy , however , only deviates by about 1 mev . similarly the self - energy components obtained in the pseudo - scalar and the pseudo - vector choice differ considerable but the equations of state differ by less than one to two mev . on the other hand , small variations of the self - energy due to the different three bonn potentials produce a significant change in the equations of state . summarizing , the equation of state is not sensitive to the self - energy components themselves but to their difference . a bit more sensitive is another quantity , i.e. , the optical potential . it depends not only on the cancellation of the scalar and vector self - energy but also contains quadratic terms . we find a good agreement of our results with the experimental data of ref . @xcite for , both , the real and the imaginary part up to a single particle energy of several hundred mev . as a resum , the self - energy components in the relativistic brueckner model are still affected with some uncertainty . for realistic one - boson exchange potentials we find them to be strongly momentum dependent . however , for this attribute we have to be aware of the limits of the model . one major success of the relativistic approach was that the equation of state comes close to the empirical saturation point . the equation of state is rather insensitive to the self - energy components themselves . similarly , the nucleon optical potential is only little more sensitive . however , a precise knowledge of the size of each self - energy component is necessary , e.g. , in the much more complicate situation of heavy ion collisions . the authors acknowledge stimulating discussions with f. de jong and h. mther . 99 j. d. walecka , ann . ( n. y. ) * 83 * , 497 ( 1974 ) ; + b. d. serot , j. d. walecka , advances in nuclear physics , * 16 * , 1 , eds . j. w. negele , e. vogt , ( plenum , n.y . , 1986 ) g. e. brown , a. d. jackson , _ the nucleon nucleon interaction _ ( north holland , amsterdam , 1976 ) . r. machleidt , advances in nuclear physics , * 19 * , 189 , eds . j. w. negele , e. vogt , ( plenum , n.y . , 1989 ) . m. r. anastasio , l. s. celenza , c. m. shakin , phys . rev . * c23 * , 2258 ( 1981 ) ; phys . rev . * c23 * , 2273 ( 1981 ) . m. r. anastasio , l. s. celenza , w. s. pong , c. m. shakin , phys . rep . * 100 * , 327 ( 1983 ) . c. j. horowitz , b. d. serot , phys . * b137 * , 287 ( 1984 ) . c. j. horowitz , b. d. serot , nucl a464 * , 613 ( 1987 ) . r. brockmann , r. machleidt , phys . * b149 * , 283 ( 1984 ) . r. brockmann , r. machleidt , phys . rev . * c42 * , 1965 ( 1990 ) . b. t. haar , r. malfliet , phys . rep . * 149 * , 207 ( 1987 ) . b. t. haar , r. malfliet , phys . rev . * c36 * , 1611 ( 1987 ) . r. malfliet , progress in particle and nuclear physics , * 21 * , 207 , ed . a.faessler , ( oxford , 1988 ) ; + w. botermans , r. malfliet , phys . rep . * 198 * , 115 ( 1990 ) ; + f. d. jong , r. malfliet , phys . rev . * c44 * , 998 ( 1991 ) . m. i. haftel , f. tabakin , nucl . phys . * a158 * , 1 ( 1970 ) . p. poschenrieder , m. k. weigel , phys . rev . * c38 * , 471 ( 1988 ) . c. nuppenau , y. j. lee , a. d. mackellar , nucl a504 * , 839 ( 1989 ) . j. a. tjon , s. j. wallace , phys . rev . * c32 * , 267 ( 1985 ) ; phys . rev . * c32 * , 1667 ( 1985 ) . f. d. jong , b. t. haar , r. malfliet , phys . b220 * , 485 ( 1989 ) . c. fuchs , h. lenske , h. h. wolter , phys . rev . * c52 * , 3043 ( 1995 ) . c. fuchs , e. lehmann , l. sehn , f. scholz , t. kubo , j. zipprich , a. faessler , nucl . * a603 * , 471 ( 1996 ) . l. sehn , h. h. wolter , nucl . phys . * a601 * , 473 ( 1996 ) ; + c. fuchs , l. sehn , h. h. wolter , nucl . phys . * a601 * , 505 ( 1996 ) ; + h. elsenhans , l. sehn , a. faessler , h. mther , n. ohtzuka , h. h. wolter , nucl . * a536 * , 750 ( 1992 ) . c. fuchs , t. gaitanos , h. h. wolter , phys b381 * , 23 ( 1996 ) . k. erkelenz , phys . rep . * c13 * , 191 ( 1974 ) . j. d. bjorken , s. d. drell , relativistic quantum mechanics ( mcgraw - hill , n.y . , 1964 ) ; + j. d. bjorken , s. d. drell , relativistic quantum fields ( mcgraw - hill , n.y . , 1965 ) . m. trefz , a. faessler , w. h. dickhoff , nucl . a443 * , 499 ( 1985 ) . m. rose , elementary theory of angular momentum ( wiley , n.y . , 1957 ) . f. d. jong , ph.d . thesis , groningen , the netherlands ( 1992 ) ( unpublished ) . r. machleidt , in : computational nuclear physics 2 nuclear reactions , k. langanke , j. a. maruhn , and s. e. koonin , eds . ( springer , new york , 1993 ) , chapter 1 , p. 1 . s. hama , b. c. clark , e. d. cooper , h. s. sherif , r. l. mercer , phys . rev . * c41 * , 2737 ( 1990 ) . * self - consistent effective mass @xmath29 in units of the bare mass ( @xmath166=939 mev ) as a function of the fermi momentum @xmath25 . the relativistic brueckner results obtained with the @xmath0-exchange potential ( full line ) are compared to those of ref . @xcite ( diamonds ) . the effective mass for the bonn c potential ( dashed line ) differs significantly from that of the @xmath0-model . * self - energy components for the pseudo - vector choice , calculated with various bonn potentials . the dotted , dashed and full lines represent the bonn a , b and c potential . the scalar self - energy @xmath10 ( lower half - plane ) and the time- and space - like vector self - energies @xmath167 ( upper half - plane ) are shown as functions of @xmath25 . for the three bonn potentials the small space - like vector self - energies @xmath168 are equal within line width . the remaining curves ( dashed dotted ) involve a different definition of the self - consistent effective mass which is given in the text later on . * self - energy components @xmath122 calculated with the pseudo - scalar choice and the bonn c potential . the curves represent the total self - energy components ( full lines ) , their direct ( dir . ) and their exchange ( exc . ) parts which are practically equal ( dashed dotted lines ) . * self - energy components @xmath122 calculated with the pseudo - vector choice and the bonn c potential . the curves represent the total self - energy components ( full lines ) , their direct ( dir . ) parts ( dashed lines ) and their exchange ( exc . ) parts ( dotted lines ) . * self - energy components @xmath122 calculated with the pseudo - vector choice and the bonn c potential ( full lines ) . they are compared with the results of ref . @xcite ( dashed lines ) and of ref . @xcite ( dotted lines ) . * self - energy components @xmath122 at saturation density @xmath169 gev ) as function of the nucleon momentum @xmath170 . the self - energies are calculated with the @xmath0-exchange potential ( pseudo - scalar choice ) and compared with values ( symbols ) taken from ref . @xcite . * self - energy components @xmath122 as function of the momentum @xmath24 of the nucleon for @xmath141 gev . the self - energies are calculated with the bonn c ( pseudo - vector choice , solid line ) and the @xmath0-exchange potential ( pseudo - scalar choice , dotted line ) and compared with results of ref . @xcite ( dashed line ) and of ref . @xcite ( @xmath171 gev , dashed dotted line ) . * equation of state , i.e. , energy per particle @xmath172 as a function of the nuclear matter density @xmath142 , for the three bonn potentials a , b and c ( from below to top : dashed line , dotted line and solid line ) . the symbols represent the respective results of ref . @xcite . * equation of state for the bonn c potential calculated in the pseudo - vector ( solid line ) and the pseudo - scalar ( dashed line ) choice . the remaining curve ( dotted ) involves a different definition of the self - consistent effective mass given in the text . * real and imaginary part of the schroedinger equivalent optical potential @xmath162 as function of the single - paricle energy @xmath173 . the results for the bonn c potential ( real part : solid line , imaginary part : dotted line ) are compared to the results of ref . @xcite ( real part : dashed line , imaginary part : dashed dotted line ) and to the empirical optical potential taken from ref . @xcite ( real part : diamonds , imaginary part : circles ) .

the zero forcing number and the positive zero forcing number of a graph are two graph parameters that arise from two types of graph colourings . the zero forcing number is an upper bound on the minimum number of induced paths in the graph , while the positive zero forcing number is an upper bound on the minimum number of induced trees in the graph . we show that for a block - cycle graph the zero forcing number equals the path cover number . we also give a purely graph theoretical proof that the positive zero forcing number of any outerplanar graphs equals the tree cover number of the graph . these ideas are then extended to the setting of-trees , where the relationship between the positive zero forcing number and the tree cover number becomes more complex . zero forcing number , positive zero forcing number , path cover number , tree cover number

1311.7064

the zero forcing number of a graph was introduced in @xcite and the related terminology was extended in @xcite . since then this parameter has been considered by a wealth of researchers , see , for example , @xcite for additional sources on this topic . independently , physicists have studied this parameter , referring to it as the graph infection number , in conjunction with control of quantum systems @xcite . it also arises in computer science in the context of fast - mixed searching @xcite . in general , when determining the zero forcing number of a graph we start with a set of initial vertices of the graph ( which we say are coloured black , while all other vertices are white ) . then , using a particular colour change rule applied to these vertices , we change the colour of white vertices in the graph to black . the repeated application of this colour change rule partitions the graph into disjoint induced paths and each of the initial vertices is an end point of one of these paths . the challenge is to determine the smallest set of initial vertices so that by repeatedly applying the colour change rule will change the colour of every white vertex of the graph to black . recently a refinement of the colour change rule was introduced ( called the positive zero forcing colour change rule ) using this rule , the positive semi - definite zero forcing number was defined ( see , for example , @xcite ) . when the positive zero forcing colour change rule is applied to a set of initial vertices of a graph , the vertices are partitioned into disjoint induced trees , rather than paths . these parameters are both remarkable since they are graph parameters that provide an upper bound on the algebraic parameters of maximum nullity of both symmetric and positive semi - definite matrices associated with a graph ( see @xcite ) . that is , for a given graph @xmath1 , we define @xmath2 : a = a^{t } , \ ; { \rm for } \ ; i \neq j , a_{ij } \neq 0 \leftrightarrow \{i , j\ } \in e\},\ ] ] and let @xmath3 denote the subset of positive semi - definite matrices in @xmath4 . the _ maximum nullity _ of @xmath5 is @xmath6 and define @xmath7 is called the _ maximum positive semi - definite nullity of @xmath5_. ( here @xmath8 denotes the nullity of the matrix @xmath9 . ) in the next section , we define the first colour change rule , along with stating the definition of a zero forcing set and the zero forcing number of a graph . the relationship between these sets and induced paths in the graph will become clear with these definitions . in sections [ graphs with z(g)=p(g ) ] and [ doublepaths ] we prove the equality between the zero forcing number and the path cover number for the following three families of graphs : block - cycle graphs , double paths , and graphs that we call series of double paths . in section 5 we define the positive semi - definite colour change rule along with the positive semi - definite forcing number and forcing trees . the positive semi - definite forcing number for graphs that are formed by the graph operation called the vertex sum are considered in section [ vertex_sum_of_two_graphs ] . in sections [ doubletrees ] and [ outerplanar ] we establish results similar to those in sections [ graphs with z(g)=p(g ) ] and [ doublepaths ] for the positive semi - definite forcing number . specifically , we show that the parameters tree cover number and positive zero forcing number agree on double trees and outerplanar graphs . finally , in section [ k_trees ] we give some families of @xmath0-trees in which we can track both the tree cover number and the positive zero forcing number . let @xmath5 be a graph in which every vertex is initially coloured either black or white . if @xmath10 is a black vertex of @xmath5 and @xmath10 has exactly one white neighbour , say @xmath11 , then we change the colour of @xmath11 to black ; this rule is called the _ colour change rule_. in this case we say `` @xmath10 forces @xmath11 '' which is denoted by @xmath12 . the procedure of colouring a graph using the colour rule is called a _ zero forcing process _ or simply a _ forcing process_. given an initial colouring of @xmath5 , in which a set of the vertices is black and all other vertices are white , the _ derived set _ is the set of all black vertices resulting from repeatedly applying the colour - change rule until no more changes are possible . if the derived set for a given initial subset of black vertices is the entire vertex set of the graph , then the set of initial black vertices is called a _ zero forcing set_. the _ zero forcing number _ of a graph @xmath5 is the size of the smallest zero forcing set of @xmath5 ; it is denoted by @xmath13 . a zero forcing process is called _ minimal _ if the initial set of black vertices is a zero forcing set of the smallest possible size . for any non - empty graph @xmath5 @xmath14 the lower bound holds with equality if and only if @xmath5 is a path and the upper bound holds with equality if and only if @xmath5 is a complete graph . in fact , the parameter @xmath13 is also interesting since @xmath15 @xcite . if @xmath16 is a zero forcing set of a graph @xmath5 , then it is possible to produce a list of the forces in the order in which they are performed in the zero forcing process . this list is called the _ chronological list of forces_. a _ forcing chain _ is a sequence of vertices @xmath17 such that @xmath18 , for @xmath19 in the forcing process . the forcing chains are not unique for a zero forcing set , as different zero forcing processes can produce distinct sets of forcing chains . in every step of a forcing process , each vertex can force at most one other vertex ; conversely every vertex not in the zero forcing set is forced by exactly one vertex . thus the maximal forcing chains partition the vertices of a graph into disjoint induced paths . the number of these paths is equal to the size of the zero forcing set and the elements of the zero forcing set are the initial vertices of the forcing chains and hence end - points of these paths ( see ( * ? ? ? * proposition 2.10 ) for more details ) . a _ path covering _ of a graph is a family of induced disjoint paths in the graph that cover ( or include ) all vertices of the graph . the minimum number of such paths that cover the vertices of a graph @xmath5 is the _ path cover number of @xmath5 _ and is denoted by @xmath20 . since the forcing chains form a set of covering paths we have the following basic result . [ p(g)<=z(g ) ] let @xmath5 be a graph , then @xmath21 . this inequality can be strict , in fact , complete graphs are examples for which the difference between these two parameters can be arbitrarily large , since if @xmath22 we have @xmath23 conversely , there are many examples where this inequality holds with equality , for example a path on @xmath24 vertices . in the next section we consider families of graphs for which the zero forcing number equals the path cover number . the most famous family of graphs for which the path cover number agrees the zero forcing number is trees ( see ( * ? ? ? * proposition 4.2 ) ) . in this section , we establish this equality for the block - cycle graphs . we refer the readers to @xcite where some initial work comparing m and @xmath25 appeared for block - cycle graphs . a graph is called _ non - separable _ if it is connected and has no cut - vertices . block _ of a graph is a maximal non - separable induced subgraph . block - cycle _ graph is a graph in which every block is either an edge or a cycle . a block - cycle graph with only one cycle is a _ unicyclic _ graph . further , according to the definition , the only block - cycle graphs with no cut vertex is either a cycle or an edge . it is not hard to see that in a block - cycle graph each pair of cycles can intersect in at most one vertex , since otherwise there will be a block in the graph which is neither a cycle nor an edge . thus two blocks are said to be _ adjacent _ if they have exactly one vertex in common . a block in a block - cycle graph is a _ pendant block _ if it shares only one of its vertices with other blocks of the graph . the next result demonstrates that just as a tree must have at least two pendant vertices , a block - cycle graph must have at least two pendant blocks . [ pendant block ] any block - cycle graph has at least two pendant blocks . let @xmath5 be a block - cycle graph with exactly @xmath24 blocks . to prove this theorem , we will construct a graph with @xmath24 vertices that is a minor of @xmath5 . we will show that the end - points of the longest induced path in @xmath26 are associated to the pendant blocks in the original graph . let @xmath27 be the blocks in @xmath5 . note that @xmath28 is either an edge or a cycle , for any @xmath29 . the vertices of @xmath26 are @xmath30 and the vertex @xmath31 is associated to the block @xmath28 . vertices @xmath31 and @xmath32 are adjacent in @xmath26 if and only if the blocks @xmath28 and @xmath33 share a vertex . let @xmath34 be the longest induced path in @xmath26 and assume that each @xmath35 corresponds to the block @xmath28 . we will show that both @xmath36 and @xmath37 are pendant blocks . if @xmath36 is not a pendant block , then there is another block , @xmath9 , that shares a vertex with @xmath36 different from the vertex that @xmath36 shares with @xmath38 . since @xmath25 is a longest induced path , the block @xmath9 must also share a vertex with some @xmath28 with @xmath39 . however , in this case the blocks @xmath40 form a non - separable subgraph that is neither a cycle nor an edge . the following lemma is straightforward to prove . [ p(g - b ) , g is a block - cycle ] if @xmath9 is a pendant block in a block - cycle graph , then @xmath41 [ for block - cycle z(g)=p(g ) ] let @xmath5 be a block - cycle graph . then @xmath42 furthermore , the paths in any minimal path covering of @xmath5 are precisely the forcing chains in a minimal zero forcing process initiated by an appropriate selection of the end - points of the paths in this collection . we prove this theorem by induction on the number of blocks in @xmath5 along with lemma [ pendant block ] . the only block - cycle graphs with one block are either an edge or a cycle ; the theorem is clearly true for these two graphs . assume that the theorem is true for all block - cycle graphs @xmath26 with fewer than @xmath43 blocks . let @xmath5 be a block - cycle graph , then according to lemma [ pendant block ] , there is a pendant block , @xmath9 , in @xmath5 which is connected to the other blocks in @xmath5 only through the vertex @xmath10 . let @xmath26 be the graph formed by removing all the vertices of the block @xmath9 , except @xmath10 , from @xmath5 . the graph @xmath26 has @xmath44 blocks , so by the induction hypothesis @xmath45 and appropriately chosen end - points of the paths in a minimal path covering of @xmath26 forms a zero forcing set . by lemma [ p(g - b ) , g is a block - cycle ] , we have @xmath46 and there are two possible cases to consider . there is a minimal path - cover @xmath47 for the graph @xmath26 in which @xmath10 is the end - point of a path @xmath25 . first assume that @xmath9 is the edge @xmath48 . then @xmath26 is the graph formed by removing the pendant vertex @xmath11 from @xmath5 . since @xmath10 is an end - point of @xmath25 and @xmath11 is only connected to @xmath10 , returning @xmath9 to @xmath26 does not change the path cover number of the graph . by the induction hypothesis , the paths in the path - cover @xmath47 are the forcing chains of the forcing process initiated by the end - points of the paths in @xmath47 . also since @xmath10 is an end - point of @xmath25 , we can assume that it does not perform any forces . therefore , the zero forcing process will be continued by using @xmath10 to force @xmath11 . thus , @xmath49 which implies @xmath50 . next assume that @xmath9 is a pendant cycle . then @xmath51 since at least two paths are needed to cover the vertices of a cycle . let @xmath11 and @xmath52 be the two neighbours of @xmath10 in @xmath9 . then , since @xmath10 is an end - point of @xmath25 , the induced path @xmath53 covers all vertices in the cycle @xmath9 except @xmath11 . thus @xmath54 is a path cover for @xmath5 and @xmath55 . we also need that these paths are forcing chains . by assigning colour black to the vertex @xmath11 , all vertices in @xmath9 will be coloured by continuing the forcing process in @xmath25 through @xmath10 . thus , @xmath56 which implies @xmath50 . in every minimal path covering @xmath47 of @xmath26 , the vertex @xmath10 is an inner vertex of a path @xmath25 in @xmath47 . again , we first assume that @xmath9 is an edge @xmath48 . since @xmath57 covers all the vertices of @xmath5 , we have that @xmath58 . but if @xmath59 , then @xmath11 is covered in the same path as @xmath10 in a path covering of @xmath26 ; this contradicts the fact that @xmath10 is not an end - point of any path in any path covering of @xmath5 . thus @xmath55 . if the vertex @xmath11 is assigned the colour black , then we are able to colour the graph @xmath5 following the same forcing process which we followed to colour the graph @xmath26 . thus , @xmath56 which implies @xmath50 . if @xmath9 is a cycle , then @xmath60 , since @xmath47 along with a path covering the vertices of @xmath61 , covers all vertices of @xmath5 . let @xmath52 be a vertex in @xmath9 that is a neighbour of @xmath10 . the set of initial set of black vertices in the zero forcing set of @xmath26 along with @xmath52 forms a zero forcing set for @xmath5 . thus , @xmath56 which implies @xmath50 . the following corollary is obtained from the fact that any unicyclic graph is a block - cycle graph . this result was also recently shown to be true in @xcite . [ unicyclics ] if @xmath5 is a unicyclic graph , then @xmath50 . it seems that for general graphs it is rare to have the equality @xmath50 . to show this , along with the fact that the discrepancy between @xmath13 and @xmath20 can be arbitrarily large , we focus on the family of graphs with @xmath62 . [ prop : allkvalues ] let @xmath5 be a graph with @xmath62 and two covering paths @xmath63 and @xmath64 with @xmath65 and @xmath66 . then @xmath67 moreover , for any number @xmath0 in this interval , there is a graph @xmath5 satisfying @xmath62 with @xmath68 . assume that @xmath69 and note that the claim that @xmath70 is trivial . let @xmath9 be the set consisting of @xmath71 and an end - point of @xmath64 . obviously @xmath9 is a zero forcing set for @xmath5 . thus @xmath72 . let @xmath0 be any number in the interval @xmath73 and @xmath63 and @xmath64 be two paths with @xmath65 and @xmath66 . starting with an end - point of @xmath63 make each of the first @xmath0 consecutive vertices of @xmath63 adjacent to all of the vertices of @xmath64 . then , it is easy to observe that forcing number of this graph is @xmath74 ( the forcing set consists on the consecutive @xmath0 vertices of @xmath63 along with an end - point of @xmath64 ) . in the next section we discuss a family of graphs for which @xmath75 . a graph is _ outerplanar _ if there exists a planar embedding of the graph in which all the vertices are contained in a single face . if an outerplanar graph , that is not a path , has the property that its vertices can be covered with two induced paths , then the graph is called a _ double path _ or a _ parallel path_. such graphs were also called a _ graph of two parallel paths _ in @xcite . we refer to these two induced paths , naturally , as the _ covering paths_. clearly if @xmath5 is a double path , then @xmath62 . we will show that @xmath13 is also @xmath76 , so double paths are another family of graphs for which the path cover number equals the zero forcing number . [ zdoublepath ] if @xmath5 is a double path , then @xmath77 . assume that @xmath63 and @xmath64 are the two covering paths of @xmath5 in a given embedding of @xmath5 . thus the paths are fixed and we can talk about the left end points of the paths ( or , equivalently , the right end points ) . let @xmath10 be the left end point of @xmath63 and @xmath11 the left end point of @xmath64 . it is not difficult to deduce that since no edges are cross in this graph , the set @xmath78 forms a minimal zero forcing set for @xmath5 . note , if we choose the left endpoints @xmath78 of a double path , then it follows that @xmath63 and @xmath64 ( as in the proof above ) will also form the corresponding zero forcing chains for a double path . the concept of a double path can be generalized . if the vertices of a graph @xmath5 can be partitioned into paths @xmath79 so that : 1 . the only vertices not in the path @xmath80 that are adjacent to a vertex in @xmath80 are in either @xmath81 or @xmath82 ( assume @xmath83 and @xmath84 are the empty set ) , and 2 . the graph induced by @xmath85 and @xmath82 is a double path for @xmath86 , then @xmath5 is called a _ series of parallel paths_. [ pathseries ] if @xmath5 is a series of parallel paths , then @xmath87 . moreover , the left ( or right ) endpoints of the covering paths form a zero forcing set . we use induction on the number of paths . if there are two paths , then the result follows from theorem [ zdoublepath ] and assume the result holds for any series of parallel paths with @xmath0 paths . let @xmath5 be a graph that is the series of @xmath74 parallel paths and assume that @xmath84 is the final path . set @xmath88 . the set of left end points of the path covering of @xmath26 forms a zero forcing set of @xmath26 . then these end points , together with the left end point of @xmath84 forms a zero forcing set for @xmath5 and the forcing chains are the paths , @xmath89 . we note here that theorem [ pathseries ] may be used to yield the zero forcing number of the grid , namely the cartesian product of two paths ( see also @xcite ) . that is , the zero forcing number of the @xmath90-by-@xmath24 grid is given by @xmath91 . theorem [ zdoublepath ] can also be obtained from ( * ? ? ? * theorem 5.1 ) , although zero forcing is not considered in @xcite . in fact , they show that among all the graphs with @xmath62 , only those that are also outerplanar satisfy m@xmath92 . there are outerplanar graphs for which the path cover number and the zero forcing number are arbitrarily far apart ( see , for example , ( * ? ? ? . 2.11 ) ) . motivated by this , in the next section we consider positive zero forcing sets , and the positive zero forcing number . in 2010 , a variant of the zero forcing number , called positive semi - definite zero forcing or the positive zero forcing number , was introduced in @xcite , and a collection of its properties were discussed in @xcite and @xcite . the positive zero forcing number is also based on a colour change rule that is very similar to the zero forcing colour change rule . let @xmath5 be a graph and @xmath9 a set of vertices ; we will initially colour of the vertices of @xmath9 black and all other vertices white . let @xmath93 be the set of vertices of the connected components of @xmath94 . if @xmath10 is a vertex in @xmath9 and @xmath52 is the only white neighbour of @xmath10 in the graph induced by @xmath95 , then @xmath10 can force the colour of @xmath52 to black . this is the _ positive colour change rule_. the definitions and terminology for the positive zero forcing process , such as , colouring , derived set , positive zero forcing number etc . , are identical to those for the zero forcing number , except we use the positive semi - definite colour change rule . the size of the smallest positive zero forcing set of a graph @xmath5 is denoted by @xmath96 . note that for any non - empty graph @xmath97 the lower bound holds with equality if and only if @xmath5 is a tree and the upper bound only for complete graphs . also for all graphs @xmath5 , since a zero forcing set is also a positive zero forcing set we have that @xmath98 . moreover , in @xcite it was observed that @xmath99 , for any graph @xmath5 . we have seen that applying the zero forcing colour change rule to the vertices of a graph produces a path covering for the graph . analogously , applying the positive colour change rule produces a set of induced trees in the graph , and we refer to these trees as _ forcing trees_. to define these trees , let @xmath5 be a graph and @xmath100 be a positive zero forcing set of @xmath5 . construct the derived set , recording the forces in the order in which they are performed ; this is the chronological list of forces . note that in applying the colour change rule once , two or more vertices can perform forces at the same time and a vertex can force multiple vertices from different components at the same time . for any chronological list of forces , a forcing tree is an induced rooted tree , @xmath101 , formed by a sequence of sets of vertices @xmath102 , where @xmath103 is the root and the vertices in @xmath104 are at distance @xmath105 from @xmath106 in the tree . the vertices of @xmath104 for @xmath107 , are forced by applying the positive semi - definite colour change rule with the vertices in @xmath108 ; so for any @xmath109 there is a @xmath110 , such that @xmath10 forces @xmath11 if and only if @xmath11 is a neighbour of @xmath10 . in a forcing tree , the vertices in @xmath104 are said to be the vertices of the @xmath105-th level in the tree . note that the vertices in a specific level may have been forced in different steps of the positive semi - definite colour change procedure and they may also perform forces in different steps . the positive zero forcing number of the graph @xmath5 in figure [ ex.forcing trees ] is three and @xmath111 is a positive zero forcing set for the graph . the forcing trees in this colouring procedure ( as depicted in figure [ ex.forcing trees ] ) are as follows : @xmath112 , where @xmath113 , @xmath114 , @xmath115 , + @xmath116 . @xmath117 , where @xmath118 . + + our next concept is analogous to the path cover number of a graph . a _ tree covering of a graph _ is a family of induced vertex disjoint trees in the graph that cover all vertices of the graph . the minimum number of such trees that cover the vertices of a graph @xmath5 is the _ tree cover number _ of @xmath5 and is denoted by @xmath119 . any set of forcing trees corresponding to a minimal positive zero forcing set is of size @xmath96 and covers all vertices of the graph . this implies the following . [ t(g)&z_+(g ) ] for any graph @xmath5 , we have @xmath120 . this bound is clearly tight for trees and cycles . but there are graphs , such as complete bipartite graphs , for which the discrepancy between these parameters can be arbitrarily large . it is , hence , an interesting question to ask for which families of graphs does equality hold between these two parameters . one way to approach this problem is to find a graph operation which preserves the equality in graphs for which these parameters agree . using this approach , in the next section we will prove that equality between these two parameters holds for block - cyclic graphs . in section [ doubletrees ] we will define a family of graphs called _ double trees _ , these are analogous to the double paths of section [ doublepaths ] and we will show that the positive zero forcing number and the tree cover number of these graphs coincide . we will also show these parameters are equal for a much larger family of graphs in section [ outerplanar ] . let @xmath5 and @xmath121 be two graphs and assume that @xmath11 is a vertex of both @xmath5 and @xmath121 , then the _ vertex sum of @xmath5 and @xmath121 over @xmath11 _ is the graph formed by identifying @xmath11 in the two graphs . the vertex - sum is denoted by @xmath122[l]{$+$ } \raisebox{-.75ex}{\makebox[2ex]{$_{v}$ } } } h$ ] . a block - cycle graph can be recursively defined as the vertex sum of a block - cycle graph with a cycle or a path . the next result shows how to calculate the tree cover number of the vertex sum of two graphs . [ tofvertexsum ] for any graphs @xmath5 and @xmath121 , both with an identified vertex @xmath11 , we have @xmath123[l]{$+$ } \raisebox{-.75ex}{\makebox[2ex]{$_{v}$}}}\,h)=t(g)+t(h)-1.\ ] ] let @xmath124 and @xmath125 be minimal tree coverings of @xmath5 and @xmath121 , respectively , and suppose @xmath126 and @xmath127 are the trees covering @xmath11 . let @xmath128[l]{$+$ } \raisebox{-.75ex}{\makebox[2ex]{$_{v}$}}}\ , t_2 $ ] . observe that @xmath129 is an induced tree in @xmath130[l]{$+$ } \raisebox{-.75ex}{\makebox[2ex]{$_{v}$}}}\,h$ ] that also covers @xmath11 . therefore , @xmath131 is a tree covering of @xmath130[l]{$+$ } \raisebox{-.75ex}{\makebox[2ex]{$_{v}$}}}\,h$ ] . next we show , in fact , the vertices of @xmath130[l]{$+$ } \raisebox{-.75ex}{\makebox[2ex]{$_{v}$}}}\,h$ ] can not be covered with fewer trees . suppose that the vertices of @xmath130[l]{$+$ } \raisebox{-.75ex}{\makebox[2ex]{$_{v}$}}}\,h$ ] can be covered with @xmath132 induced disjoint trees . let @xmath133 be the tree that covers @xmath11 in such a tree covering of @xmath130[l]{$+$ } \raisebox{-.75ex}{\makebox[2ex]{$_{v}$}}}\,h$ ] . at least @xmath134 trees are needed to cover the vertices of @xmath135 . since the number of trees in the covering of @xmath130[l]{$+$ } \raisebox{-.75ex}{\makebox[2ex]{$_{v}$}}}\,h$ ] is @xmath132 , the vertices of @xmath136 are covered by at most ( @xmath137 ) trees , but this contradicts the fact that @xmath138 is the least number of trees that cover the vertices of @xmath121 . [ z+ofvertexsum ] for any graphs @xmath5 and @xmath121 , both with an identified vertex @xmath11 , we have @xmath139[l]{$+$ } \raisebox{-.75ex}{\makebox[2ex]{$_{v}$}}}\,h)=z_+(g)+z_+(h)-1.\ ] ] to prove this equality , let @xmath124 and @xmath125 be the sets of forcing trees for a minimal positive zero forcing set in @xmath5 and @xmath121 respectively . let @xmath126 and @xmath127 be the trees that contain @xmath11 . then @xmath140[l]{$+$ } \raisebox{-.75ex}{\makebox[2ex]{$_{v}$}}}\,t_2 $ ] is a forcing tree in @xmath130[l]{$+$ } \raisebox{-.75ex}{\makebox[2ex]{$_{v}$}}}\,h$ ] covering @xmath11 . then similar reasoning as in the previous lemma applies . [ z+vertexsum ] if @xmath5 and @xmath121 are two graphs that satisfy @xmath141 and @xmath142 , then @xmath139[l]{$+$ } \raisebox{-.75ex}{\makebox[2ex]{$_{v}$}}}\,h)=t(g\,{\makebox[0ex][l]{$+$ } \raisebox{-.75ex}{\makebox[2ex]{$_{v}$}}}\,h),\ ] ] where @xmath11 is an identified vertex in both @xmath5 and in @xmath121 . since a block - cycle graph is the vertex sum of a block - cycle graph and either a cycle or a path we have the following result . if @xmath5 is a block - cycle graph , then @xmath143 . in the next section , we will show that @xmath144 for every outerplanar graph . the first step towards verifying this claim is to show that it holds for a subset of these graphs called _ double trees_. recall that in section [ doublepaths ] we defined a double path , whereas a double tree can be viewed as an extension of the concept of a double path . if the vertices of a connected outerplanar graph , which is not a tree , can be covered with two induced trees , then the graph is called a _ double tree_. our first step will be to show that if @xmath5 is a double path , then @xmath145 . then any double tree can be constructed by applying an appropriate series of vertex sums of trees with an appropriate double path . thus , from corollary [ z+vertexsum ] , we will be able to conclude that @xmath145 holds for any double tree . in the following three lemmas we find different positive zero forcing sets for double paths . in all of these lemmas we will assume that @xmath5 is a double path with a specific planar embedding of @xmath5 with covering paths @xmath63 and @xmath64 . since this planar embedding of @xmath5 is fixed , we can refer to the end points of a covering path as the right end point and the left end point . [ lemma1 ] if @xmath10 and @xmath11 are both right ( or both left ) end points of @xmath63 and @xmath64 , respectively , then @xmath78 is a positive zero forcing set of @xmath5 . moreover , if @xmath78 are initially coloured black , then there is a positive zero - forcing process in which the forcing trees are @xmath63 and @xmath64 . we prove this lemma by induction on the number of vertices . the result is true for @xmath146 , which is a connected double path on the fewest number of vertices . assume that it is true for all graphs @xmath121 with @xmath147 . let @xmath5 be a graph on @xmath24 vertices . assume that @xmath10 and @xmath11 are left end points of @xmath63 and @xmath64 , respectively . by assigning the colour black to each of these vertices we claim that @xmath78 is a positive zero forcing set of @xmath5 . if @xmath10 is a pendant vertex , then it forces its only neighbour , say @xmath52 ( which must be in @xmath148 , which is a left end point of a covering path in @xmath149 . thus by the induction hypothesis @xmath150 is a positive zero forcing set of @xmath5 and there is a positive zero forcing process in which the forcing trees are @xmath63 and @xmath64 . similarly if @xmath11 is a pendant vertex , using a similar reasoning , the lemma follows . if neither @xmath10 nor @xmath11 are pendant , then @xmath10 and @xmath11 are adjacent and since both are on the same side , at least one of them , say @xmath10 , is of degree two . let @xmath52 be the only neighbour of @xmath10 in @xmath63 . thus @xmath10 can force @xmath52 ( its only white neighbour ) and again by the induction hypothesis @xmath150 is a positive zero forcing set of @xmath5 and there is a positive zero forcing process in which the forcing trees are @xmath63 and @xmath64 . the same reasoning applies when @xmath10 and @xmath11 are the right end points of @xmath63 and @xmath64 , respectively . [ lemma2 ] if @xmath10 and @xmath11 are two vertices of @xmath63 and @xmath64 , respectively , which form a cut set for @xmath5 , then @xmath78 is a positive zero forcing set for @xmath5 . moreover , there is a positive zero forcing forcing process in which @xmath63 and @xmath64 are the forcing trees ( with @xmath10 and @xmath11 the roots of the tree ) . let @xmath151 and @xmath152 be the vertices of the left hand side and the right hand side components of @xmath153 , respectively , and let @xmath154 and @xmath155 be the subgraphs induced by @xmath156 and @xmath157 , respectively . then according to lemma [ lemma1 ] , @xmath78 is a positive zero forcing set for both @xmath154 and @xmath155 and thus a positive zero forcing set for @xmath5 and there is a positive zero forcing process in which @xmath63 and @xmath64 are the forcing trees . [ lemma3 ] if @xmath10 is a vertex in @xmath63 which is not an end point , then there always is a vertex @xmath11 in @xmath64 such that @xmath78 is a cut set of @xmath5 . if @xmath64 contains at most two vertices this result is clear since @xmath10 is not an end point . suppose there is a vertex @xmath10 in @xmath63 for which there is no vertex @xmath11 in @xmath64 such that @xmath78 is a cut set for @xmath5 . this implies that @xmath158 since if @xmath159 , then at least one of @xmath160 and @xmath161 is a cut set . let @xmath11 be any non - pendant vertex of @xmath64 . obviously @xmath10 is a cut vertex of @xmath63 and @xmath11 is a cut vertex of @xmath64 . since @xmath78 is not a cut set of @xmath5 , there is a vertex in the left hand side ( or right hand side ) of @xmath10 that is adjacent to a vertex in right hand side ( or left hand side ) of @xmath11 . assume that @xmath52 is the farthest vertex from @xmath11 in @xmath64 having this property . since @xmath52 is the farthest vertex from @xmath11 with the described property and @xmath5 is an outerplanar graph , @xmath162 is a cut set of @xmath5 which contradicts with the fact that , there is no vertex in @xmath64 that forms a cut set along with @xmath10 for @xmath5 . combining lemmas [ lemma1 ] , [ lemma2 ] and [ lemma3 ] along with the fact that in the proof of all these three lemmas forces are performed along the covering paths , we have the following . [ pzfs of double path ] let @xmath5 be a double path with covering paths @xmath63 and @xmath64 . then for any vertex @xmath11 in @xmath63 , there is always another vertex @xmath10 in @xmath64 such that @xmath78 is a positive zero forcing set for @xmath5 . moreover there is a positive zero forcing process in which the two paths @xmath63 and @xmath64 are a minimal set of forcing trees in @xmath5 . the following result is a consequence of corollary [ pzfs of double path ] . [ pzfs of double tree ] let @xmath5 be a double tree with covering trees @xmath163 and @xmath164 . then for any vertex @xmath11 in @xmath163 , there is always another vertex @xmath10 in @xmath164 such that @xmath78 is a positive zero forcing set for @xmath5 . moreover @xmath165 coincides with a minimal collection of forcing trees in @xmath5 . in @xcite it is shown that the maximum positive semi - definite nullity is equal to the tree cover number for any outerplanar graph . since the positive zero forcing number is an upper bound on the maximum positive semi - definite nullity , this implies that @xmath166 holds for any outerplanar graph . this result was shown to be true in @xcite and @xcite ( where the proof is generalized to 2-trees ) , in this section we give a different proof of this fact that does not rely on schur - complements or orthogonal removal . moreover , we show that any minimum tree covering of an outerplanar graph coincides with a minimum collection of zero forcing trees . in a fixed tree covering of a graph , two trees , @xmath163 and @xmath164 , are said to be _ adjacent _ if there is at least one edge @xmath167 such that @xmath168 and @xmath169 . a tree is called _ pendant _ if it is adjacent to only one other tree from this given tree covering . throughout this section , @xmath5 will be an outerplanar graph with a planar embedding in which all the vertices are on the same face . an edge of @xmath5 is called _ outer _ if it lies on the face containing all of the vertices ; it an edge is not outer , then it is called _ inner_. further , let @xmath170 be a minimal tree covering for @xmath5 . define @xmath171 to be the graph whose vertices correspond to the trees in @xmath170 and two vertices in @xmath171 are adjacent if there is an outer edge between the corresponding trees in the graph @xmath5 . two trees of @xmath170 are called _ consecutive _ , if their corresponding vertices in @xmath171 are adjacent vertices each of degree two . [ consecutive trees ] let @xmath5 be an outerplanar graph and @xmath170 a minimum tree covering for @xmath5 . if there is no pendant tree in @xmath170 , then there are at least one pair of consecutive trees in @xmath170 . assume that @xmath170 is a minimum tree covering for @xmath5 in which there is no pendant tree . then , two cases are possible : * case 1 . * there is no tree in @xmath170 with at least one of the inner edges of @xmath5 in its edge set ( see figure [ fig_for_h_t ] for an example of such a graph ) . therefore @xmath171 is a cycle . accordingly , any adjacent pair of trees in @xmath170 are consecutive . * case 2 . * there is at least one tree in @xmath170 that has an inner edge of @xmath5 in its edge set ( figure [ fig_for_h_t_2 ] gives an example of such a graph ) . is a forcing tree with an inner edge in its edge set ] the idea in this case is that we will select `` left - most '' such inner edge of @xmath5 . then the subgraph induced by this edge , and all the vertices to the left of the edge , form an outerplanar graph in which no tree includes an inner edge of @xmath5 . in case 1 , we showed that such a subgraph will have a consecutive pair of trees , and thus so will @xmath5 . first , let @xmath172 be the set of all vertices which are the end - points of an inner edge of @xmath5 that is also included in the edge set of a tree in @xmath170 . second , note that any inner edge @xmath173 of @xmath5 partitions the plane into two parts and , consequently , partitions the set of vertices of @xmath174 into two subsets @xmath175 and @xmath176 . since @xmath5 is outerplanar and finite , there exists an inner edge , @xmath177 in @xmath172 such that at least one of @xmath175 or @xmath176 does not contain any of the vertices in @xmath172 . we will assume that @xmath175 is the vertex set that is disjoint from @xmath172 . the subgraph of @xmath5 induced by the vertices @xmath178 is an outerplanar graph ; call this @xmath121 . the trees in @xmath170 that intersect with @xmath121 form a minimal tree covering of @xmath121 , which we will call @xmath179 . none of the trees in @xmath179 can include an inner edge of @xmath121 ( this follows as there are no vertices that are both in @xmath121 and in @xmath172 ) . by case 1 , there is a pair of adjacent trees in @xmath179 and therefore there is at least one pair of consecutive trees in @xmath170 . the following theorem shows that for any outerplanar graph , it is possible to find a minimal tree covering that has a pendant tree ; this plays a key role in the proof of the fact that outerplanar graphs satisfy @xmath141 . [ pendant tree ] let @xmath5 be an outerplanar graph . then there is a minimum tree covering for @xmath5 in which there is a pendant tree . assume that @xmath170 is a minimum tree covering for @xmath5 in which there is no pendant tree . we use theorem [ consecutive trees ] to construct a new tree covering @xmath180 of @xmath170 with @xmath181 in which there is a pendant tree . by theorem [ consecutive trees ] there are two trees @xmath163 and @xmath164 in @xmath170 , which are consecutive . let @xmath121 be the outerplanar graph induced by @xmath182 . there are two outer edges in @xmath121 that have an end - point from each of trees @xmath163 and @xmath164 ( otherwise @xmath183 would be a tree and @xmath170 would not be a minimum tree covering ) . one of these outer edges of @xmath121 , call it @xmath184 , is an inner edge in @xmath5 ; we will assume that @xmath185 and @xmath186 . similarly , if @xmath10 has any neighbour in @xmath164 , then @xmath11 has no other neighbours in @xmath163 . thus we will assume that @xmath11 is the only neighbour of @xmath10 in @xmath164 . in fact , the subgraph @xmath187 is a forest and exactly one of the trees in the forest has vertices which are adjacent to a vertex in @xmath164 . call this tree @xmath188 . define a second new tree by @xmath189[l]{$+$ } \raisebox{-.75ex}{\makebox[2ex]{$_{v}$ } } } \{u , v\ } \right ) { \makebox[0ex][l]{$+$ } \raisebox{-.75ex}{\makebox[2ex]{$_{u}$ } } } ( t_1 \backslash s_1).\ ] ] by replacing @xmath163 and @xmath164 in @xmath170 with @xmath188 and @xmath190 we can construct a new minimum tree covering for @xmath5 in which @xmath188 is a pendant tree . a similar argument applies when @xmath10 has another neighbour in @xmath164 . if neither @xmath10 nor @xmath11 has any other neighbour in @xmath163 and @xmath164 , respectively , then either case mentioned above are applicable . we now have all the necessary tools to prove the main result of this section . [ outerplanars satisfy z_+=t ] let @xmath5 be an outerplanar graph . then @xmath191 moreover , any minimal tree covering of the graph @xmath170 coincides with a collection of forcing trees with @xmath192 . we prove the claim by induction on the tree cover number and using theorem [ pendant tree ] . it is obviously true for @xmath193 . assume that it is true for any outerplanar graph @xmath26 with @xmath194 . now let @xmath5 be an outerplanar graph with @xmath195 . by proposition [ t(g)&z_+(g ) ] , we have @xmath196 . let @xmath197 be a minimum tree covering of @xmath5 . we first consider the case when @xmath170 contains a pendant tree . * assume that @xmath163 is a pendant tree and that @xmath164 is the only tree adjacent to @xmath163 . let @xmath26 be the graph induced by the vertex set @xmath198 , then the induction hypothesis holds , so @xmath199 . further , @xmath200 are forcing trees in a positive zero forcing process whose initial set of black vertices , @xmath201 , has a vertex from each tree in @xmath202 . a positive zero forcing process in @xmath5 starting with the black vertices in @xmath201 can proceed as it does in @xmath26 until the first vertex of @xmath203 , say @xmath204 , that is adjacent to some vertex in @xmath163 gets forced . since the graph induced by @xmath182 is a double tree , according to corollary [ pzfs of double tree ] , the vertex @xmath204 determines a vertex @xmath205 in @xmath163 such that @xmath206 is a positive zero forcing set for the subgraph induced by @xmath207 . since the induction hypothesis holds for this subgraph , the tree @xmath164 is a forcing tree in this subgraph as well . thus the vertices of @xmath164 get forced in the same order as they were forced in @xmath26 . therefore we can complete the colouring of @xmath5 by adding the vertex @xmath205 to the initial set of black vertices . thus , @xmath208 is a positive zero forcing set of @xmath5 with @xmath209 as the forcing trees in this positive zero forcing process . thus , @xmath210 . * case 2 . * if @xmath170 does not contain a pendant tree , then by theorem [ pendant tree ] , it is possible to build a new tree covering that does have a pendant tree . by case 1 , this new tree covering has exactly @xmath96 trees and the trees of this new tree covering are forcing trees . now we need to show that the original minimal tree covering of the graph @xmath5 also coincides with a collection of forcing trees associated with a positive zero forcing set of @xmath5 . we will assume that @xmath163 and @xmath164 are a pair of consecutive trees in @xmath170 ( from theorem [ consecutive trees ] we know that such a pair exists ) . using the same notation as in theorem [ pendant tree ] we assume that @xmath211 has a neighbour , other than @xmath10 , in @xmath163 ( the other cases are similar ) . in the procedure of constructing a pendant tree in the minimal tree covering of @xmath5 , the pair of consecutive trees @xmath163 and @xmath164 were modified to obtain two new trees called @xmath188 and @xmath190 . the tree @xmath188 is pendant and only adjacent to the tree @xmath190 in the new minimal tree covering . let @xmath212 and define @xmath213 we will use a similar decomposition of @xmath164 . let @xmath214 be the set of all the trees in the forest @xmath215 that have a vertex which is adjacent to some vertex in @xmath216 . define @xmath217 . let @xmath201 be a positive zero forcing set for @xmath5 for which @xmath218 is a set of zero forcing trees . assume that @xmath219 and @xmath220 are the two vertices in @xmath221 from the trees @xmath188 and @xmath190 ( from case 1 such the zero forcing set must have two such vertices ) . we will consider two cases , the first is when @xmath204 is a vertex in @xmath164 and second is when @xmath222 . in the first case @xmath223 and @xmath224 . we claim that @xmath221 is a positive zero forcing set and there is positive zero forcing process in which the trees of @xmath170 are the zero forcing trees . to see this we will describe the positive zero forcing process . the positive zero forcing process proceeds along the forcing trees @xmath188 and @xmath190 until the vertex @xmath11 is forced . note that in the original process , @xmath11 must force @xmath10 . if the vertices @xmath11 and @xmath205 are removed then one of the connected components will include all of @xmath214 and some of the vertices from @xmath216 . starting with @xmath11 and @xmath205 it is possible to force all the vertices in this component following the trees @xmath214 and the portion of @xmath216 in the component . once all the vertices of @xmath214 are black , @xmath205 can force the remaining vertices along the tree @xmath216 . then the vertex in @xmath216 that is adjacent to @xmath10 will force @xmath10 . then @xmath10 can force the remaining vertices of @xmath225 . in the second case both @xmath204 and @xmath205 are vertices in @xmath163 . we claim that @xmath226 is a positive zero forcing set and there is positive zero forcing process in which the trees of @xmath170 are the zero forcing trees . just as in the previous case , using @xmath11 and @xmath205 all the vertices of @xmath214 will be forced . then , starting with @xmath205 , all the vertices for @xmath216 will be forced with @xmath216 the forcing tree . then the unique vertex in @xmath216 adjacent to @xmath10 will force @xmath10 and the positive zero forcing process will continue along @xmath225 . finally , starting with @xmath11 , the vertices along @xmath227 will be forced . a _ @xmath0-tree _ is constructed inductively by starting with a @xmath228 and at each step a new vertex is added and this vertex is adjacent to exactly @xmath0 vertices in an existing @xmath229 . a _ partial @xmath0-tree _ is any graph that is the subgraph of a @xmath0-tree . in particular , a graph is a partial @xmath76-tree if and only if it does not have a @xmath230 minor ( see @xcite ) . since outerplanar graphs are exactly the graphs with no @xmath230 and @xmath231 minors ( see @xcite , it is easy to see that every outerplanar graph is a partial @xmath76-tree . in @xcite it is shown that the proof that the maximum positive semi - definite nullity for outerplanar graphs is equal to the tree cover number from @xcite , can be extended to include any partial @xmath76-tree . from this it follows that if @xmath5 is a partial @xmath76-tree , then @xmath144 . in this section , we will give a purely graph theoretical version of this result for a subset of @xmath76-trees . we also try to track the variations between the positive zero forcing number and the tree cover number in this subset of @xmath0-trees with @xmath232 . this demonstrates that @xmath76-trees are rather special when it comes to comparing @xmath96 and @xmath233 . we will define a type @xmath0-tree that we call a _ @xmath0-cluster_. these @xmath0-trees are constructed recursively starting with a @xmath234 . at each step a new vertex is added to the graph and this new vertex is adjacent to exactly @xmath0 of the vertices in @xmath121 . in a general @xmath0-tree the new vertices are adjacent to any @xmath0-clique in the graph , but in a @xmath0-cluster the new vertices must be adjacent to a @xmath0-clique in @xmath121 . observe that for each vertex @xmath11 not in @xmath121 , there is exactly one vertex in @xmath121 that is not adjacent to it . if @xmath5 is a @xmath0-cluster , then define @xmath235 to be the set of all distinct @xmath0-cliques @xmath236 with the property that @xmath237 forms a clique of size @xmath74 in @xmath5 , for some @xmath238 . the size of @xmath235 can be no more than @xmath74 . [ z_+&t of k - trees ] suppose @xmath5 is a @xmath0-cluster and let @xmath235 be as defined above . 1 . if @xmath239 , then @xmath240 . 2 . if @xmath241 , then @xmath242 . if @xmath243 and @xmath0 is even , then @xmath244 . if @xmath245 and @xmath0 is even , then @xmath246 . since the minimum degree of @xmath5 is @xmath0 , it is clear that @xmath247 . suppose @xmath121 is the initial @xmath228 in the @xmath0-cluster . obviously , the set @xmath248 is a positive zero forcing set for @xmath5 and @xmath249 . to prove the first statement suppose @xmath9 is a positive zero forcing set of @xmath5 with @xmath250 . since @xmath251 , there are three vertices @xmath252 that are adjacent to distinct @xmath0-sets in @xmath121 . thus any vertex @xmath253 is adjacent to at least two of these vertices . so no positive zero forcing set can be contained in @xmath248 . thus there must be a vertex @xmath254 such that @xmath255 . if @xmath256 has only one white neighbour , then this neighbour must be in @xmath121 and @xmath256 can force it and we return to the first case . if @xmath256 has two or more white neighbours , then no vertex of @xmath9 can perform a force and again we reach a contradiction . to verify the second statement observe that if @xmath241 , then there is at least one vertex @xmath257 , that has at most one neighbour @xmath258 . the set @xmath259 forms a positive zero forcing set . to see this , first note that after removing this set , all the vertices in @xmath260 are disjoint . finally , there is one vertex in @xmath248 that is not adjacent to @xmath10 , this vertex can force @xmath11 . for the third statement assume @xmath0 is even and @xmath243 . since @xmath228 is a subgraph of @xmath5 , the tree cover number of @xmath5 is no less than @xmath261 . suppose that @xmath262 is a minimal tree covering for @xmath5 with @xmath263 . since no tree in @xmath262 can contain more than two vertices of @xmath121 and @xmath264 , each tree in @xmath262 , except one , contains exactly two vertices of @xmath121 . assume that @xmath163 is the tree that contains only a single vertex of @xmath121 . the size of @xmath235 is @xmath74 , so for any vertex , @xmath265 , there is a corresponding vertex in @xmath266 which is adjacent to all of the vertices of @xmath121 except @xmath52 . in particular , if @xmath52 is the single vertex of @xmath121 in the tree @xmath163 , then there is a vertex @xmath10 which is adjacent to all vertices in @xmath121 except @xmath52 . since @xmath267 , the vertex @xmath10 can not be covered by extending any of the trees of @xmath262 ( as @xmath10 is adjacent to both vertices in any tree from @xmath262 ) . this contradicts @xmath262 being a tree covering for @xmath5 . thus @xmath268 finally we will show that we can construct a tree covering of this size . let @xmath269 be a minimal tree covering for @xmath121 . thus @xmath270 and exactly one tree in @xmath269 contains only one vertex ( all other trees contain exactly two vertices ) . call this tree @xmath271 . extend @xmath133 to include every vertex in @xmath266 that is adjacent to @xmath10 . now the only vertices in @xmath5 that are not covered by a tree in @xmath269 are the vertices in @xmath266 that are adjacent to every vertex in @xmath121 , except @xmath10 , call these vertices @xmath272 . take any tree in @xmath269 , except @xmath273 . then this tree will have two vertices , say @xmath206 . remove this tree from @xmath269 and replace it with the two trees , @xmath274 and @xmath275 . this gives a tree covering of size @xmath276 . for the fourth statement , assume @xmath245 and @xmath0 is even . thus there is a vertex , @xmath257 that is adjacent to all of the vertices in the graph @xmath5 . let @xmath269 be a tree covering of @xmath121 with @xmath261 trees in which @xmath277 is a covering tree . then @xmath133 can be extended to cover all the vertices of @xmath278 . thus @xmath279 . note that if in theorem [ z_+&t of k - trees ] we have @xmath280 , then the positive zero forcing number and the tree cover number coincide ( this result is also proved in @xcite ) . [ k - tree with k odd ] let @xmath5 be a @xmath0-tree with @xmath0 is odd , then @xmath281 . we prove this by induction on the number of vertices in @xmath5 . the result is clearly true for @xmath230 ( the smallest @xmath282-tree ) . assume that the statement is true for all @xmath26 with @xmath283 . let @xmath5 be a @xmath0-tree with @xmath284 . let @xmath285 be a vertex of degree @xmath0 ( such a vertex always exists ) . by the induction hypothesis there exists a tree covering @xmath262 of @xmath286 with exactly @xmath287 trees . since neighbours of @xmath11 form a @xmath0-clique , any tree in @xmath262 covers at most two of the neighbours of @xmath11 . moreover , @xmath11 has an odd number of neighbours , thus there is exactly one tree @xmath133 in @xmath262 that covers only one of the neighbours of @xmath11 . therefore we can extend @xmath133 to cover @xmath11 , and conclude @xmath281 . recall that a graph is called _ chordal _ if it contains no induced cycles on four or more vertices . for instance , all @xmath0-trees are examples of chordal graphs . in general it is known for any chordal graph @xmath5 , that @xmath288 , where @xmath289 denotes the fewest number of cliques needed to cover ( or to include ) all the edges in @xmath5 ( see @xcite ) . this number , @xmath289 , is often called the _ clique cover number _ of the graph @xmath5 . further inspection of the work in @xcite actually reveals that , in fact , for any chordal graph , @xmath289 is equal to the _ ordered set number _ ( @xmath290 ) of @xmath5 . in @xcite , it was proved that for any graph @xmath5 , the ordered set number of @xmath5 and the positive zero forcing number of @xmath5 are related and satisfy , @xmath291 . as a consequence , we have that @xmath292 for any chordal graph @xmath5 , and , in particular , @xmath293 . so studies of the positive zero forcing number of chordal graphs , including @xmath0-trees , boils down to determining the clique cover number and vice - versa . in section [ graphs with z(g)=p(g ) ] we introduced families of graphs for which the zero forcing number and the path cover number coincide . in fact , we showed that for the family of block - cycle graphs this is true . however , there are additional families for which equality holds between these two parameters . for example , the graph @xmath294 , where @xmath177 is an edge of @xmath230 , has @xmath50 . it is , therefore , natural to propose characterizing all the graphs @xmath5 for which @xmath50 . in @xcite it is conjectured the result analogous to corollary [ z+vertexsum ] holds for zero forcing sets ; if this conjecture is confirmed , then there would be a much large family of graphs for which the path number and zero forcing number coincide . to this end , we state the following problem as a beginning to this study . [ conj_z&p_vertex_sum ] let @xmath5 and @xmath121 be two graphs , both with an identified vertex @xmath11 , and both satisfy @xmath50 and @xmath295 . then @xmath296[l]{$+$ } \raisebox{-.75ex}{\makebox[2ex]{$_{v}$}}}\,h)=p(g\,{\makebox[0ex][l]{$+$ } \raisebox{-.75ex}{\makebox[2ex]{$_{v}$}}}\,h).\ ] ] it is not difficult to verify that in any tree , any minimal path cover coincides with a collection of forcing chains . we conjecture that this is also the case for the block - cycle graphs ( and refer the reader to ( * ? ? ? * section 5.2 ) for more details ) . in general , it is an interesting question if for a graph @xmath5 with @xmath50 , is it true that any minimal path cover of @xmath5 coincides with a collection of forcing chains of @xmath5 ? in section [ forcing_trees ] , we proved the equality @xmath297 where @xmath5 is an outerplanar graph . the structure of a planar embedding of outerplanar graphs was the key point to establishing the equality . there are many non - outerplanar graphs with a similar structure ; generalizing this structure will lead to discovering more graphs that satisfy @xmath141 . in general , we are interested in characterizing all the graphs @xmath5 for which @xmath143 . e. almodovar , l. deloss , l. hogben , k. hogenson , k. myrphy , t. peters , c. ramez , c. minimum rank , maximum nullity and zero forcing number , and spreads of these parameters for selected graph families . , 3 : 371392 , 2010 . f. barioli , w. barrett , s. fallat , h. t. hall , l. hogben , b. shader , p. van den driessche , h. van der holst . parameters related to tree - width , zero forcing , and maximum nullity of a graph . , 72 : 146 - 177 , 2013 .

a precessing source frame , constructed using the newtonian orbital angular momentum , can be invoked to model inspiral gravitational waves from generic spinning compact binaries . an attractive feature of such a precessing convention is its ability to remove all spin precession induced modulations from the orbital phase evolution . however , this convention usually employs a post - newtonian ( pn ) accurate precessional equation , appropriate for the pn accurate orbital angular momentum , to evolve the-based precessing source frame . this influenced us to develop inspiral waveforms for spinning compact binaries in a precessing convention that explicitly employ to describe the binary orbits . our approach introduces certain additional 3pn order terms in the evolution equations for the orbital phase and frequency with respect to the usual-based implementation of the precessing convention . we examine the practical implications of these additional terms by computing the match between inspiral waveforms that employ and-based precessing conventions . the match estimates are found to be smaller than the optimal value , namely , for a non - negligible fraction of unequal mass spinning compact binaries .

1507.00406

inspiralling compact binaries containing spinning neutron stars and ( or ) black holes ( bhs ) are key sources for the network of second generation interferometric gravitational wave ( gw ) detectors @xcite . these instruments include the two advanced ligo ( aligo ) observatories @xcite , the advanced virgo @xcite , the kagra @xcite , the geo - hf @xcite and the planned ligo - india @xcite . in contrast , massive spinning bh binaries are one of the most exciting sources for the space - based gw observatory like the planned elisa @xcite . gws from such inspiralling compact binaries , whose components are specified by their masses and spins , can be accurately modeled using perturbative approaches to tackle the underlying einstein field equations @xcite . therefore , the optimal detection technique of _ matched filtering _ can be employed to detect and characterize inspiral gws from such binaries @xcite . this involves cross correlating the interferometric output data with a bank of templates that theoretically model inspiral gws from spinning compact binaries . a successful detection demands that at least one template should remain in phase as much as possible with the buried weak gw signals in the frequency windows of various gw observatories @xcite . during the gw emission induced inspiral , dynamics and associated gws from compact binaries can be accurately described using the post - newtonian ( pn ) approximation to general relativity @xcite . the pn description provides interesting quantities , required for various template constructions , as an asymptotic series in terms of certain dimensionless parameter . for binaries in quasi - circular orbits , it is usual to use @xmath3 as the pn expansion parameter while constructing inspiral templates , where @xmath4 and @xmath5 stand for the total mass and orbital ( angular ) frequency of the binary @xcite . currently , gw frequency and associated phase evolution , crucial inputs to construct various inspiral template families , are known to 3.5pn order for non - spinning compact binaries @xcite . in other words , pn corrections to the above two quantities are computed to the @xmath6 order beyond the leading quadrupolar ( newtonian ) order for such binaries . moreover , the amplitudes of the two gw polarization states , @xmath7 and @xmath8 , for non - spinning binaries are available to 3pn order @xcite . very recently , detailed computations led to the determination of the dynamics of such binaries to the 4pn order @xcite . binaries that contain compact objects with intrinsic rotations , the spin effects enter the dynamics and gw emission via spin - orbit and spin - spin interactions @xcite . in binaries containing maximally spinning bhs , the spin - orbit coupling ( linear in the spins ) first appears at the 1.5pn order , while the spin - spin interaction ( which is quadratic in spins ) first occurs at the 2pn order @xcite . additionally , @xmath9 , @xmath10 and @xmath1 , the two spin and orbital angular momenta , for generic spinning compact binaries precess around the total angular momentum @xmath11 due to spin - orbit and spin - spin interactions . this forces substantial modulations of the emitted gws from inspiralling generic spinning compact binaries compared to their non - spinning counterparts @xcite . therefore , it is important to incorporate various effects due to the intrinsic rotations of compact objects while constructing inspiral gw templates for spinning compact binaries . at present , gw frequency evolution and amplitudes of @xmath7 and @xmath8 for bh binaries having maximally spinning components are fully determined to 2.5pn and 2pn orders , respectively , while incorporating all the relevant spin induced effects @xcite . moreover , the on - going detailed computations are providing various higher pn order spin - orbit and spin - spin contributions to the dynamics of spinning compact binaries in general orbits and to the orbital frequency evolution for quasi - circular inspiral . at present , the spin - orbit contributions to binary dynamics and gw frequency evolution are available up to the next - to - next - to - leading order ( 2pn order ) beyond the leading order @xcite while adapting the mpm ( multipolar post - minkowskian ) approach @xcite . in contrast , the higher order spin - spin contributions to the binary dynamics are usually tackled in the arnowitt - deser - misner canonical formalism @xcite and in the effective field theory formalism @xcite ( note that the spin - orbit effects in the effective field theory formalism are computed , for example , in @xcite ) . these approaches provide various spin(1)-spin(2 ) and spin - squared contributions to the orbital dynamics @xcite . moreover , various source multipole moments needed to obtain the spin contributions to gw luminosity at the 3pn order and gw polarization states to the 2.5pn order were computed in @xcite . the construction of time - domain @xmath7 and @xmath12 associated with inspiralling generic spinning compact binaries requires us to numerically solve a set of pn accurate differential equations for @xmath13 and the orbital phase @xcite , where @xmath0 is the newtonian orbital angular momentum . the numerical integration provides temporal evolutions for the orbital phase , the associated angular frequency and the two angles that specify the orientation of the orbital plane in an inertial frame associated with the direction of @xmath14 at the initial epoch . these variations are incorporated into the pn accurate expressions for @xmath7 and @xmath8 to obtain pn accurate time - domain inspiral waveforms from such binaries @xcite . in this approach , the differential equation for the orbital phase explicitly depends on the precessional motion of the orbital plane @xcite . therefore , it is not possible to express the orbital phase as an integral of the orbital frequency as usually done in the case of non - spinning compact binaries @xcite . a decade ago , buonanno , chen and vallisneri proposed an approach , referred to as the precessing convention , that factorizes the generic spinning binary waveform into a _ carrier signal _ and a _ modulated amplitude _ term @xcite . in this approach , the phase of the carrier signal ( @xmath15 ) essentially coincides with the accumulated orbital phase such that @xmath16 . moreover , the precessional dynamics of the orbital plane only influences the modulated amplitude part of inspiral waveform even for generic compact binaries . therefore , the precessing convention disentangles the precessional effects from its non - precessional counterparts while modeling both the amplitude and the phase evolutions of inspiral gws from such astrophysical systems . this convention was employed to model inspiral gw signals from compact binaries containing misaligned single - spin and to probe its data analysis benefits @xcite . very recently , inspiral - merger - ringdown waveforms for generic spinning bh binaries , invoking the effective - one - body approach @xcite , also adapted the precessing convention to model gws from the inspiral part @xcite . we note that this convention requires a _ precessing source frame _ which is usually based on the newtonian orbital angular momentum @xmath0 . however , @xcite employed both @xmath17 and @xmath0 to model gws during the late part of the binary inspiral just prior to the plunge . in this paper , we develop a prescription to compute pn accurate inspiral waveforms for generic spinning compact binaries while using the pn accurate orbital angular momentum @xmath17 to construct the precessing source frame . this is influenced by the usual practice of employing precessional equation appropriate for @xmath1 to evolve @xmath0 and the associated precessing source frame while constructing inspiral waveforms via the precessing convention of @xcite . we show that the use of such an adiabatic approximation , namely employing an orbital averaged differential equation for @xmath0 , can lead to pn corrections to @xmath18 . these observations motivated us to provide a set of pn accurate equations to obtain temporally evolving quadrupolar order inspiral gw polarization states for generic spinning compact binaries in the @xmath17-based precessing convention . in our approach , the spin precession induced modulations enter the differential equation for the orbital phase only at the 3pn order . moreover , the @xmath17-based convention requires us to include additional 3pn order terms in the differential equation for @xmath19 , compared to the usual @xmath0-based approach . we explore the practical implications of these additional terms with the help of _ match _ computations , detailed in @xcite . the match computations involve two inspiral families where one is constructed via our @xmath1-based precessing convention and therefore incorporate the above mentioned 3pn order terms . the other family is based on the lalsuite spintaylort4 code , developed by the ligo scientific collaboration ( lsc ) @xcite , that implemented the @xmath0-based precessing convention of @xcite . in this code , the precessional dynamics is fully 2pn accurate while the differential equation for @xmath19 incorporates spin - orbit contributions to 3.5pn order . for match computations , we employ pn accurate relation connecting @xmath0 and @xmath1 to construct two waveform families with physically identical initial orbital and spin orientations . we terminate the two inspiral waveform families when their respective @xmath19 parameter reaches @xmath20 which roughly corresponds to orbital separations @xmath21 . these considerations allow us to attribute the reported match estimates to the above mentioned additional 3pn order terms present in our differential equations for the orbital phase and frequency . we find that the match estimates are less than the optimal 0.97 value for a non - negligible fraction of unequal mass spinning compact binaries . it may be recalled that such an optimal match value roughly corresponds to a @xmath22 loss in the ideal event rate . in what follows , we briefly summarize the usual implementation of the @xmath0-based precessing convention and explore the consequence of employing an orbital averaged precessional equation for @xmath0 . in section [ sec_l ] , we detail the construction of quadrupolar order gw polarization states in our @xmath1-based precessing convention . the match estimates involving these two inspiral families with physically equivalent initial configurations and associated discussions are listed in section [ result ] while section [ sec_dis ] provides a brief summary . we begin by summarizing the usual implementation of the @xmath0-based precessing convention and explore the consequence of employing an orbital averaged differential equation for @xmath0 . the precessing convention , introduced in @xcite , aims to remove all the spin precession induced modulations from the orbital phase evolution . in this approach , the orbital phase @xmath23 is written as an integral of the orbital frequency @xmath24 , namely @xmath25 , even for generic spinning compact binaries . this feature is crucial to ensure that the inspiral waveform for a precessing spinning compact binary can be written as the product of a non - precessing carrier waveform and a modulation term that contains all the precessional effects . this is how the approach disentangles the precessional effects from their non - precessional counterparts both in the amplitude and phase of inspiral waveforms . it should be noted that in the absence of precessing convention , the orbital phase of a generic spinning binary is given by @xmath26 \ , dt'$ ] , where @xmath27 and @xmath28 specify the orientation of @xmath0 in an inertial frame associated with the initial direction of @xmath14 @xcite . to obtain inspiral waveforms for spinning compact binaries in their precessing convention , @xcite employed certain _ precessing source frame _ @xmath29 , where @xmath30 is the unit vector along @xmath0 . the basis vectors of this triad satisfy the evolution equations @xmath31 and @xmath32 . the angular frequency @xmath33 is constructed in such a manner that these three basis vectors always form an orthonormal triad . this is possible with the following expression for @xmath33 , namely @xmath34 , where @xmath35 is the usually employed precessional frequency for @xmath30 . the expression for @xmath36 that includes the dominant order spin - orbit and spin - spin contributions can be obtained by collecting the terms that multiply @xmath37 @xmath38 in equation ( 9 ) of @xcite . with the help of @xmath39 and @xmath40 , an orbital phase @xmath23 may be defined such that @xmath41 , where @xmath42 is the unit vector along the binary separation vector @xmath43 . additionally , one may define a co - moving frame ( @xmath44 ) such that the time derivative of @xmath42 is given by @xmath45 . it was argued in @xcite that @xmath33 should only be proportional to @xmath42 which ensures that @xmath46 . this leads to the desirable expression for @xmath47 , namely @xmath48 , while employing the adiabatic condition for the sequence of circular orbits : @xmath49 . this adiabatic condition is equivalent to @xmath50 that provides another version of the pn independent relation connecting the linear and orbital angular velocities , @xmath51 . a close inspection reveals that this approach usually employs an orbital averaged differential equation for @xmath0 to evolve the precessing source frame while constructing pn accurate inspiral waveforms . it turns out that the differential equation for @xmath0 in such an adiabatic approximation is identical to the evolution equation for the pn accurate orbital angular momentum @xmath1 @xcite . in what follows , we explore the effect of such an adiabatic approximation on the equation for @xmath52 in the ( @xmath53 ) co - moving frame and on the derivation of @xmath47 equation . the usually employed expression for @xmath54 while considering only the leading order spin - orbit interactions may be written as @xmath55 where @xmath56 ( @xmath57 ) is the mass ratio and @xmath58 while @xmath59 is the symmetric mass ratio . the kerr parameters @xmath60 and @xmath61 of the two compact objects of mass @xmath62 and @xmath63 specify their spin angular momenta by @xmath64 , where @xmath65 and @xmath66 are the unit vectors along @xmath9 and @xmath10 . it is straightforward to show that the above equation is identical to @xmath67 terms that multiply @xmath68 on the right hand side of equation ( 9 ) in @xcite . to explore the implication of using the above expression for @xmath54 to construct @xmath69 , we revisit the arguments detailed in the appendix b of @xcite . these arguments , crucial to obtain @xmath48 , require that @xmath70 , appearing in the equation @xmath45 , should be zero . a closer look reveals that @xcite did not use the explicit expression for @xmath33 to show that @xmath33 lies along @xmath42 . instead , the authors arrived at such a conclusion with the help of the following two steps . first , it was noted that @xmath71 should be @xmath72 ( see lines around equations ( b3 ) , ( b4 ) and ( b5 ) in @xcite ) . this was inferred by employing the definition for @xmath30 ( @xmath73 ) , the resulting time derivative for @xmath30 ( @xmath74 ) , the adiabatic condition for circular orbits ( @xmath75 and the time derivative for @xmath76 in the co - moving frame ( @xmath77 ) . in the second step , @xcite invoked the requirement that @xmath71 should also be given by @xmath78 in the precessing source frame . with the help of the above two arguments , namely @xmath79 and @xmath80 , @xcite concluded that @xmath33 should lie along @xmath42 as @xmath81 . we would like to emphasize that @xcite never invoked their explicit expression for @xmath33 to show that @xmath33 can have components only along @xmath42 which is essential to obtain the relation @xmath48 . in fact , it is straightforward to show with the help of equation ( [ eq_l_omega_k ] ) that @xmath33 can have components along @xmath76 since @xmath82 , in general . this results in the following 1.5pn accurate expression for @xmath83 @xmath84 + \frac{\delta_2}{q}\ , \chi_2\ , \bigl [ { \boldsymbol{s}}_2 \times { \boldsymbol{n } } - ( { \boldsymbol{s}}_2 \cdot { \boldsymbol{l}})\ , { \boldsymbol{\lambda } } \bigr ] \bigg\ } \,.\end{aligned}\ ] ] clearly , the 1.5pn order terms that arise from @xmath70 in the above equation will not be zero for generic spinning compact binaries . interestingly , it is still possible to equate @xmath47 to @xmath85 by employing the adiabatic condition @xmath49 as @xmath86 even in the presence of non - vanishing 1.5pn order @xmath70 term . however , the equivalent version of the adiabatic condition , namely @xmath50 , forces the differential equation for @xmath15 to contain 3pn order corrections in addition to the standard @xmath5 term . in other words , two equivalent versions of the same adiabatic condition for circular orbits , namely @xmath49 and @xmath50 , provide different evolution equations for @xmath15 . this is the unexpected consequence of employing precessional equation appropriate for @xmath1 to evolve the @xmath0-based precessing source frame . it is not difficult to deduce that this essentially arises from the non - vanishing 1.5pn order @xmath70 contributions to @xmath83 listed in the above equation . the following arguments can also be used to clarify why the above two versions of the same adiabatic condition for the circular orbits result in two different expressions for @xmath87 . in our opinion , this arises by identifying @xmath88 to be @xmath89 , where @xmath90 is the unit vector along @xmath1 . strictly speaking , the precessing source frame of @xcite is based on @xmath1 rather than @xmath0 due to the use of @xmath54 that provides the precessional equation for @xmath1 @xcite . this implies that their co - moving triad is rather @xmath1-based and the expression for @xmath91 got components along @xmath88 instead of @xmath81 . therefore , @xmath92 results in pn corrections to @xmath87 as @xmath93 is unity only at the leading order ( this may be deduced from our equation ( [ eq_l_k ] ) listed below ) . this leads to a differential equation for @xmath87 that involves pn corrections to @xmath5 . the use of precessional equation appropriate for @xmath1 while implementing the precessing convention of @xcite motivated us to develop a @xmath90-based precessing convention for constructing inspiral waveforms for spinning compact binaries . this should also allow to explore the practical implications of using the adiabatic approximation to evolve @xmath0 in the usual implementation of the precessing convention . influenced by the above arguments and @xcite , we first introduce a @xmath90-based precessing source frame : ( @xmath94,@xmath95 , @xmath96 ) . the precessional dynamics of @xmath97 and @xmath98 are provided by @xmath99 , where @xmath100 and @xmath101 is the precessional frequency of @xmath90 . it should be obvious that @xmath102 is identical to @xmath103 as @xmath104 . it is possible to construct a @xmath90-based co - moving triad ( @xmath105 ) and introduce an orbital phase @xmath106 such that [ eq_n_xi ] @xmath107 it is fairly straightforward to obtain following expressions for the time derivatives of @xmath42 and @xmath108 @xmath109 we are now in a position to obtain the differential equation for @xmath106 . this is derived with the help of the frame independent adiabatic condition for circular orbits , namely @xmath110 . employing the above expression for @xmath111 in such an adiabatic condition leads to @xmath112 where @xmath113 is given by @xmath114 this results in the following 3pn accurate differential equation for @xmath106 @xmath115 ^ 2\biggr\ } \,.\ ] ] these additional terms appear at the 3pn order as we employ equation ( [ eq_l_omega_k ] ) for @xmath54 that only incorporates the leading order spin - orbit interactions appearing at 1.5pn order . it is not difficult to deduce that the inclusion of 2pn order spin - spin interaction terms in @xmath54 can lead to certain 4pn order contributions to the @xmath116 equation . additionally , the presence of the above 3pn contributions to @xmath116 equation may be attributed to the fact that the orbital velocity @xmath117 can have non - vanishing pn order components along @xmath1 while considering generic spinning compact binaries @xcite . the use of @xmath1 to describe binary orbits also modifies the evolution equation for @xmath5 ( or @xmath19 ) . this is because the spin - orbit interactions are usually incorporated in terms of @xmath118 and @xmath119 , in the literature @xcite . these terms require modifications due to the following 1.5pn order relation connecting @xmath30 and @xmath90 @xmath120\ , { \boldsymbol{n } } \nonumber \\ & & + \bigl [ 2\ , x_1 ^ 2 \ , \chi_1 \ , ( { \boldsymbol{s}}_1 \cdot { \boldsymbol{\xi}})+ \eta\ , \chi_1\ , ( { \boldsymbol{s}}_1 \cdot { \boldsymbol{\xi } } ) \nonumber \\ & & + 2\ , x_2 ^ 2 \ , \chi_2 \ , ( { \boldsymbol{s}}_2 \cdot { \boldsymbol{\xi}})+ \eta\ , \chi_2\ , ( { \boldsymbol{s}}_2 \cdot { \boldsymbol{\xi } } ) \bigr]\ , { \boldsymbol{\xi } } \biggr\ } \,,\end{aligned}\ ] ] where @xmath121 and @xmath122 . the above relation can easily be extracted , for example , from equations ( 6.10 ) and ( 7.10 ) in @xcite . this relation leads to certain additional 3pn order contributions to @xmath123 while describing the binary orbits with @xmath90 . these additional terms are , for example , with respect to equation ( 3.16 ) of @xcite that provides pn accurate expression for @xmath124 while invoking @xmath30 to describe binary orbits . our additional contributions to @xmath125 appear at 3pn order as the dominant spin - orbit interactions , in terms of @xmath118 and @xmath119 , contribute to the @xmath19 evolution equation at 1.5pn order . the differential equation for @xmath19 in our approach may be written as @xmath126 \nonumber \\ & \quad-\frac{47}{3}\ , x_2 ^ 2\ , \chi_2\ , \bigl[-\frac{1}{2}\ , \eta\ , \bigl(\chi_1\ , ( { \boldsymbol{s}}_1 \cdot { \boldsymbol{n } } ) + \chi_2\ , ( { \boldsymbol{s}}_2 \cdot { \boldsymbol{n}})\bigr)({\boldsymbol{s}}_2 \cdot { \boldsymbol{n } } ) \nonumber \\ & \quad+\bigl(2\ , x_1 ^ 2\ , \chi_1\ , ( { \boldsymbol{s}}_1\cdot { \boldsymbol{\xi } } ) + \eta\ , \chi_1\ , ( { \boldsymbol{s}}_1 \cdot { \boldsymbol{\xi } } ) + 2\ , x_2 ^ 2\ , \chi_2\ , ( { \boldsymbol{s}}_2\cdot { \boldsymbol{\xi } } ) + \eta\ , \chi_2\ , ( { \boldsymbol{s}}_2 \cdot { \boldsymbol{\xi } } ) \bigr ) ( { \boldsymbol{s}}_2 \cdot { \boldsymbol{\xi}})\bigr ] \nonumber \\ & \quad-\frac{25}{4}\ , ( x_1-x_2)\ , x_2\ , \chi_2\ , \bigl[-\frac{1}{2}\ , \eta\ , \bigl(\chi_1\ , ( { \boldsymbol{s}}_1 \cdot { \boldsymbol{n } } ) + \chi_2\ , ( { \boldsymbol{s}}_2 \cdot { \boldsymbol{n}})\bigr)({\boldsymbol{s}}_2 \cdot { \boldsymbol{n } } ) \nonumber \\ & \quad+\bigl(2\ , x_1 ^ 2\ , \chi_1\ , ( { \boldsymbol{s}}_1\cdot { \boldsymbol{\xi}})+\eta\ , \chi_1\ , ( { \boldsymbol{s}}_1 \cdot { \boldsymbol{\xi } } ) + 2\ , x_2 ^ 2\ , \chi_2\ , ( { \boldsymbol{s}}_2\cdot { \boldsymbol{\xi } } ) + \eta\ , \chi_2\ , ( { \boldsymbol{s}}_2 \cdot { \boldsymbol{\xi } } ) \bigr ) ( { \boldsymbol{s}}_2 \cdot { \boldsymbol{\xi}})\bigr ] \nonumber \\ & \quad+\frac{25}{4}\ , ( x_1-x_2)\ , x_1\ , \chi_1\ , \bigl[-\frac{1}{2}\ , \eta\ , \bigl(\chi_1\ , ( { \boldsymbol{s}}_1 \cdot { \boldsymbol{n}})+\chi_2\ , ( { \boldsymbol{s}}_2 \cdot { \boldsymbol{n}})\bigr)({\boldsymbol{s}}_1 \cdot { \boldsymbol{n } } ) \nonumber \\ & \quad+\bigl(2\ , x_1 ^ 2\ , \chi_1\ , ( { \boldsymbol{s}}_1\cdot { \boldsymbol{\xi}})+\eta\ , \chi_1\ , ( { \boldsymbol{s}}_1 \cdot { \boldsymbol{\xi } } ) + 2\ , x_2 ^ 2\ , \chi_2\ , ( { \boldsymbol{s}}_2\cdot { \boldsymbol{\xi } } ) + \eta\ , \chi_2\ , ( { \boldsymbol{s}}_2 \cdot { \boldsymbol{\xi } } ) \bigr ) ( { \boldsymbol{s}}_1 \cdot { \boldsymbol{\xi}})\bigr ] \biggr ] \biggr \ } \,,\end{aligned}\ ] ] where the first term is adapted from equation ( 3.16 ) in @xcite by replacing its @xmath30 vectors by our @xmath90 vectors . clearly , these additional _ spin_-squared terms contribute to @xmath124 at 3pn order . note that the use of @xmath90 for @xmath30 in the leading order spin - spin contributions to @xmath124 , as listed in equation ( 1 ) of @xcite , results in additional 3.5pn order _ spin_-cubed terms . such contributions to our @xmath124 equation are neglected as the lalsuite spintaylort4 code that implement the @xmath0-based precessing convention does not include any _ spin_-cubed terms in its differential equation for @xmath19 . in what follows , we model inspiral gws from spinning compact binaries in our @xmath90-based precessing convention . we begin by displaying the following quadrupolar order expressions for the two gw polarization states : [ eq_hp_hx ] @xmath127 where @xmath128 and @xmath129 are the @xmath19 and @xmath130 components of @xmath108 and @xmath42 in an inertial frame associated with @xmath131 , the unit vector that points from the source to the detector , while @xmath132 is the distance to the binary . these @xmath19 and @xmath130 components can be expressed in terms of the cartesian components of @xmath94 and @xmath95 via equations ( [ eq_n_xi ] ) . we note that the above expressions for the quadrupolar order @xmath7 and @xmath12 are written in the so - called frame - less convention @xcite . these expressions emerge from the following standard definitions for the quadrupolar order gw polarization states : @xmath133 where @xmath134 and @xmath135 are the two polarization vectors forming , along with @xmath131 , an orthonormal right - handed triad . to obtain equation ( [ eq_hp_hx ] ) , we used the following expression for the quadrupolar order transverse - traceless part of the far - zone field @xmath136 @xmath137 in the frame - less convention , we let the components of @xmath134 and @xmath135 in the @xmath131-based inertial frame to be @xmath138 and @xmath139 . clearly , we need to specify how the cartesian components of @xmath108 and @xmath42 vary in time to obtain temporally evolving gw polarization states for inspiralling generic spinning compact binaries . therefore , we require to solve numerically the differential equations for @xmath140 and @xmath95 to obtain @xmath141 and @xmath142 . we use equation ( [ eq_phidot ] ) for @xmath106 while the differential equation for @xmath19 is given by equation ( [ eq_dxdt ] ) that contains all the non - spinning contributions accurate up to 3.5pn order and the usual spin contributions that are fully 2pn accurate . these contributions are provided , for example , by equation ( 3.16 ) in @xcite and are listed in the [ appendix ] , where we have replaced @xmath30 by @xmath90 . these specific pn order choices are influenced by the fact that the lalsuite spintaylort4 code also employs an evolution equation for @xmath5 that incorporates such pn contributions . we note that this routine implements the @xmath30-based precessing convention of @xcite to construct inspiral templates to search for gws from generic spinning binaries . the differential equations for @xmath94 and @xmath95 in our approach are given by [ eq_e1_e2_dot ] @xmath143 where we use the following expression for @xmath54 that incorporates the leading order spin - orbit and spin - spin interactions @xmath144 \bigg\ } \,.\end{aligned}\ ] ] it is fairly straightforward to verify that this expression is identical to the coefficient of @xmath68 appearing on the right hand side of equation ( 9 ) in @xcite . the above equations imply that we also need to invoke the precessional equations for @xmath65 , @xmath66 and @xmath90 ( or @xmath98 ) to tackle numerically the dynamics of such binaries . the three coupled equations for @xmath65 , @xmath66 and @xmath90 that include the leading order spin - orbit and spin - spin interactions read [ eq : s1_s2_dot ] @xmath145 \bigg\ } \ , , \\ { \dot { { \boldsymbol{s}}}_{2 } } & = \frac{c^3}{gm}\,x^{5/2}\ , \bigg\{\delta_2 \left({\boldsymbol{k}}\times { \boldsymbol{s}}_2\right ) + \frac{1}{2}\,x^{1/2}\,\bigg [ x_1 ^ 2\,\chi_1\,({\boldsymbol{s}}_1 \times { \boldsymbol{s}}_2 ) -3\ , x_1 ^ 2\,\chi_1\,({\boldsymbol{k } } \cdot { \boldsymbol{s}}_1 ) \ , ( { \boldsymbol{k } } \times { \boldsymbol{s}}_2 ) \bigg ] \bigg\ } \ , , \\ { \dot { { \boldsymbol{k } } } } & = \frac{c^3}{gm}\ , x^{3}\,\bigg\ { \delta_1\,q\ , \chi_1\ , \left ( { \boldsymbol{s}}_1\times{\boldsymbol{k}}\right ) + \frac{\delta_2}{q}\ , \chi_2\ , \left ( { \boldsymbol{s}}_2\times{\boldsymbol{k}}\right ) \nonumber \\ & \quad-\frac{3}{2}\,x^{1/2}\,\eta\ , \chi_1\ , \chi_2\ , \biggl [ ( { \boldsymbol{k } } \cdot { \boldsymbol{s}}_1)\ , ( { \boldsymbol{s}}_2 \times { \boldsymbol{k } } ) + ( { \boldsymbol{k } } \cdot { \boldsymbol{s}}_2 ) \ , ( { \boldsymbol{s}}_1 \times { \boldsymbol{k } } ) \biggr ] \bigg\ } \,.\end{aligned}\ ] ] it is not very difficult to verify that the above equations for @xmath146 and @xmath147 are identical to equations ( 2 ) and ( 3 ) in @xcite while the differential equation for @xmath90 , as expected , arises from the usual conservation of total angular momentum @xmath14 . this conservation implies that @xmath148 . we would like to state again that the equations ( [ eq_phidot ] ) , ( [ eq_dxdt_l ] ) , ( [ eq_sk ] ) and ( [ eq_dxdt ] ) provide the differential equations for @xmath106 and @xmath19 in the present implementation of @xmath90-based precessing convention . strictly speaking , the use of equation ( [ eq_omega_k ] ) for @xmath54 requires us to include the additional 4pn and 3.5pn contribution to @xmath149 and @xmath124 , respectively . however , we do not incorporate such spin - quartic and spin - cubic terms in our present work . in practice , we numerically solve simultaneously the differential equations for @xmath150 , @xmath65 , @xmath66 , @xmath106 and @xmath19 to obtain temporally evolving cartesian components of @xmath42 and @xmath108 . the resulting variations in these cartesian components are imposed on the expressions for @xmath151 , given by equations ( [ eq_hp_hx ] ) . this leads to inspiral waveforms for generic spinning compact binaries in our @xmath90-based precessing convention . note that we do not solve the differential equation for @xmath95 . this is because the temporal evolution of @xmath95 can be estimated using the relation @xmath152 . this implies that we solve 12 differential equations for the cartesian components of @xmath94 , @xmath90 , @xmath65 and @xmath66 along with the differential equations for @xmath106 and @xmath19 to track the time evolution for the cartesian components of @xmath42 and @xmath108 . the required initial values for the cartesian components of @xmath94 , @xmath90 , @xmath65 and @xmath66 are given by freely choosing the following five angles : @xmath153 , @xmath154 , @xmath155 , @xmath156 and @xmath157 . these angles specify the above four unit vectors in the @xmath131-based inertial frame at the initial epoch such that [ eq:16 ] @xmath158 this choice is also influenced by the lalsuite spintaylort4 code of lsc . additionally , we let the initial @xmath19 value to be @xmath159 where @xmath160 hz and the initial phase @xmath161 to be zero . we move on to explain how to specify initial conditions , physically identical to equations ( [ eq:16 ] ) , while constructing inspiral waveforms based on the @xmath30-based precessing convention . clearly , the initial orientations of two spin vectors in the inertial @xmath131 frame should be identical in the two approaches . however , the orientation of @xmath30 in such an inertial frame is different from that of @xmath90 . we compute @xmath162 using equation ( [ eq_l_k ] ) that provide 1.5pn accurate relation connecting @xmath30 and @xmath90 . this pn relation makes the value of @xmath162 at @xmath163 to depend on @xmath164 and the four angles that specify @xmath65 and @xmath66 in the inertial @xmath131 frame . we observe that [ @xmath165 is maximum for equal mass maximally spinning compact binaries and the difference is usually less than @xmath166 . the value of @xmath167 specifies the initial orientation of @xmath30 in the inertial @xmath131 frame as we usually let its azimuthal angle to be zero along with @xmath168 ( we have verified that the changes in these angles play no role in our match computations ) . the difference in @xmath169 and @xmath170 values leads to slightly different values for @xmath171 and @xmath172 at the initial epoch . therefore , the orbital frequency and phase evolutions are slightly different in the two approaches even in the absence of our 3pn order additional terms . we observe that the differences in these two dot products , namely @xmath171 and @xmath172 , have their maximum values for equal mass maximally spinning compact binaries . in the next section , we purse the _ match _ computations to probe the implications of additional 3pn order terms present in the frequency and phase evolution equations while constructing our @xmath90-based inspiral waveforms . we employ the _ match _ , detailed in @xcite , to compare inspiral waveforms constructed via the above described @xmath30 and @xmath90-based precessing conventions . our comparison is influenced ( and justified ) by the fact that the precessing source frames of the above two conventions are functionally identical . this should be evident from the use of _ the same _ precessional frequency , appropriate for @xmath90 , to obtain pn accurate expressions for both the @xmath30-based @xmath173 and @xmath90-based @xmath174 . therefore , the _ match _ estimates probe influences of the additional 3pn order terms present in the differential equations for @xmath106 and @xmath19 in our approach . additionally , we have verified that these 3pn order terms are not present in the usual implementation of the precessing convention as provided by the lalsuite spintaylort4 code . our match @xmath175 computations involve @xmath176 and @xmath177 , the two families of inspiral waveforms arising from the @xmath30 and @xmath90-based precessing conventions . the @xmath178 inspiral waveform families are adapted from the lalsuite spintaylort4 code of lsc while @xmath177 families , as expected , arise from our approach ( equations ( [ eq_hp_hx ] ) ) . it should be noted that we employ the quadrupolar ( newtonian ) order expressions for @xmath179 while computing @xmath180 and @xmath181 in the present analysis . however , the lalsuite spintaylort4 code can provide waveforms that include 1.5pn order corrections to their amplitudes . we would like to stress that the two families involved in our @xmath175 computations are characterized by identical values of @xmath182 and @xmath183 . additionally , the initial orientations of the two spins in the @xmath131-based inertial frame were also chosen to be identical . the computation of @xmath162 from @xmath184 with the help of equation ( [ eq_l_k ] ) ensures that @xmath30 and @xmath90 orientations at the initial epoch are physically equivalent . therefore , our @xmath175 computations indeed compare two waveform families with physically equivalent orbital and spin configurations at the initial epoch . to obtain a specific @xmath175 estimate , we first compute an overlap function between the relevant @xmath176 and @xmath177 inspiral waveforms : @xmath185 where @xmath186 and @xmath187 stand for the normalized @xmath188 and @xmath189 waveforms , respectively . the angular bracket between @xmath180 and @xmath181 defines certain noise weighted inner product , namely @xmath190 in the above equation , @xmath191 and @xmath192 stand for the fourier transforms of @xmath188 and @xmath189 , while @xmath193 denotes the one - sided power spectral density . we have invoked the zero - detuned , high power sensitivity curve of aligo @xcite in our @xmath175 computations . the upper cut - off frequency @xmath194 is chosen to be @xmath195 while the lower cut - off frequency @xmath196 , associated with the gw detector , equals @xmath197 hz . the match @xmath198 is computed by maximizing the @xmath199 over two extrinsic variables , namely the time of arrival @xmath200 and the overall phase @xmath201 of gw at time @xmath200 @xcite . this leads to @xmath202 the maximizations of over @xmath200 and @xmath201 are performed by following @xcite . we perform the maximization over @xmath200 via the fft algorithm while the maximization over @xmath201 requires us to deploy two orthogonal templates . let us emphasize that we terminate @xmath176 and @xmath177 inspiral waveform families when their respective @xmath19 parameters reach @xmath20 which roughly corresponds to orbital separations @xmath203 . this choice arguably ensures the validity of pn approximation to describe the temporal evolutions of the above two families in our @xmath204-@xmath205 $ ] frequency window . therefore , it is reasonable to associate the departure of @xmath206 estimates from unity to the additional 3pn order contributions to the differential equations for the orbital phase and the associated angular frequency . we move on to list results of some of our @xmath206 computations . it is not very difficult to realize that extensive @xmath206 computations that deal with all aligo relevant spin and binary configurations will be rather difficult to achieve . therefore , we restrict our attention to a selected number of binaries to compare the @xmath188 and @xmath189 inspiral waveform families . in our match computations , we mainly consider binaries having total mass @xmath207 due to the following two reasons . first , it is comparatively expensive ( computationally ) to generate lengthy inspiral waveforms for low mass binaries in the aligo frequency window . additionally , we are interested in to explore the dependence of our @xmath206 estimates on the mass ratio @xmath208 in the @xmath209 $ ] range . clearly , low @xmath4 binaries can lead to secondary bhs having masses lower than the usual neutron star masses for high @xmath208 cases . in figure [ figure : q_m_phi ] , we consider maximally spinning bh binaries , characterized by the total mass @xmath210 , while varying the mass ratio @xmath208 from unity to @xmath197 . the left and right panel plots are for configurations having initial dominant spin - orbit misalignments @xmath211 @xmath212 given by @xmath213 and @xmath214 , respectively . additionally , we let the initial orbital plane orientation in the @xmath131-based inertial frame to take two values leading to edge - on ( @xmath215 ) and face - on ( @xmath216 ) binary orientations . these binary configurations , characterized by two different spin - orbit misalignments and orbital plane orientations in the @xmath131-based inertial frame , are obtained by choosing appropriately different values for the more massive bh initial spin orientation , namely @xmath217 in equation ( [ eq : s1_s2_1 ] ) . for example , we choose @xmath153 to be @xmath213 and @xmath218 to get @xmath211 equals @xmath213 and @xmath218 , respectively , for face - on binaries . however , in the case of edge - on binaries , we require to choose @xmath153 to be @xmath218 and @xmath213 , respectively . all other initial angular variables , appearing in equations ( [ eq : s1_s2_1 ] ) and ( [ eq : s1_s2_2 ] ) , were chosen to be @xmath219 , @xmath220 , @xmath221 ( we have verified that the @xmath206 estimates are rather insensitive to the choice of these initial angular variables ) . let us note that @xmath30 orientations ( from @xmath131 ) for these configurations will be slightly different from @xmath222 or @xmath223 due to the 1.5pn accurate relation between @xmath30 and @xmath90 . @xmath224{plots_m_phi_q_ks1_30_10m_t4_14feb15.eps } & \includegraphics[width=62mm , height=60mm]{plots_m_phi_q_ks1_60_10m_t4_14feb15.eps } \end{array}$ ] the upper and lower row plots in figure [ figure : q_m_phi ] are for @xmath225 , the accumulated orbital phase differences in the @xmath204-@xmath205 $ ] frequency interval while dealing with the @xmath176 and @xmath177 inspiral waveforms and the associated @xmath198 estimates , respectively . we find that the variations in @xmath206 estimates are quite independent of the initial orbital plane orientations . however , variation in match estimates do depend on the initial dominant spin - orbit misalignment and the mass ratio @xmath208 . the left panel plots show gradual decrease in @xmath206 values as we increase the @xmath208 value and this variation is reflected in the gradual increase of @xmath225 . incidentally , this pattern is also observed for configurations having somewhat smaller initial dominant spin - orbit misalignments . however , the @xmath206 estimates are close to unity for tiny @xmath226 values and this is expected as precessional effects are minimal for such binaries . therefore , the effect of the above discussed additional 3pn order terms are more pronounced for high @xmath208 compact binaries having _ moderate _ dominant spin - orbit misalignments . this picture is modified for binaries having substantial dominant spin - orbit misalignments as evident from the @xmath206 plots in the right panel of figure [ figure : q_m_phi ] . for such binaries , the @xmath206 estimates dip to a minimum and recover as we vary @xmath208 from @xmath227 to @xmath197 for both edge - on and face - on orbital plane orientations . in contrast , we observe a gradual increase in @xmath228 . the monotonic increment in @xmath229 plots is essentially due to the presence of @xmath230 terms in the additional 3pn order contributions to @xmath124 , given by equation ( [ eq_dxdt ] ) . this is because @xmath231 terms are absent for equal mass binaries which leads to smaller @xmath232 estimates for smaller @xmath208 value binaries . however , it is not possible to explain the observed @xmath233 variations purely in terms the displayed @xmath228 values , especially when precessional effects are substantial as in the @xmath234 cases . we observe that the initial values of angles like @xmath235 and @xmath236 also influence the precessional modulations in the waveforms , where @xmath237 is the unit vector along @xmath14 . therefore , we examined how the initial values of these angles vary as functions of @xmath208 . we find monotonic increase ( decrease ) in the initial values of @xmath235 @xmath238 when @xmath208 value is varied from 1 to 10 . therefore , we speculate that the combined effect of such variations and the non - negligible @xmath229 values may provide a possible explanation for the dip in match estimates around q=5 . this is because initial values of the above mentioned angles do define the way various dot products , involving @xmath42 , @xmath108 , @xmath65 and @xmath66 , vary in time . it should be noted that these dot products are present in the additional 3pn order terms in the @xmath124 expression given by equation ( [ eq_dxdt ] ) . this plausible explanation is tested in figure [ figure : th1_m_phi ] , where we plot @xmath229 and @xmath239 estimates for binary configurations having three different @xmath4 values . it is not difficult to infer that the above listed arguments should also hold for such compact binaries . the initial dominant spin - orbit misalignment is again chosen to be @xmath218 while considering only edge - on configurations . all other spin and orbital orientations at the initial epoch are identical to the cases displayed in the right panel plots of figure [ figure : q_m_phi ] . clearly , the plots in figure [ figure : th1_m_phi ] are qualitatively similar to those in the right panel plots of figure [ figure : q_m_phi ] . however , the dip in @xmath206 estimates shifts to higher @xmath208 values for higher @xmath4 compact binaries . we conclude from the above two figures that gw data analysis relevant differences between the above two inspiral families are more pronounced for unequal mass bh binaries . incidentally , we also find similar behavior while plotting @xmath233 and @xmath240 for maximally spinning @xmath241 bh binaries . in this case , the minimum @xmath206 value occurs around @xmath242 and this is consistent with the trend observed in figure [ figure : th1_m_phi ] . it is reasonable to suspect that the spin - squared additional terms are influential only for maximally spinning bh binaries . however , we observe qualitatively similar @xmath239 estimates for binary configurations having moderately spinning bhs ( @xmath243 ) as well as having mildly spinning bhs ( @xmath244 ) . moreover , the @xmath239 computations indicate that the effect of additional 3pn order terms in @xmath149 and @xmath124 equations are non - negligible even for single - spin compact binaries . in figure [ figure : th1_m_phi_ln ] , we plot @xmath229 and @xmath239 while considering @xmath30 and @xmath90-based single - spin waveforms for @xmath245 binaries in edge - on orientations . we find that these plots fairly resemble their double spin counterparts shown in the right panel plots of figure [ figure : q_m_phi ] . additionally , the variations in the initial values of @xmath235 and @xmath246 are similar to the double spin binaries , while varying @xmath208 value from @xmath227 to @xmath197 . therefore , we conclude that it may be beneficial to keep the additional 3pn order terms in the differential equations for @xmath106 and @xmath19 while modeling inspiral gws even from single - spin binaries via the precessing convention . and @xmath247 as functions of @xmath208 for edge - on maximally spinning bh binaries having @xmath248 and @xmath249 while keeping @xmath211 to be @xmath218 . all other initial angular parameters are identical to those used in the right panel plots of figure [ figure : q_m_phi ] . the position of the dip in @xmath250 plots is shifted towards higher @xmath208 values for higher @xmath4 binaries . a possible explanation relies on the more influential contributions from the various dot products that appear in equation ( [ eq_dxdt ] ) for such binaries having @xmath208 roughly in the @xmath251 to @xmath252 range . , width=309,height=279 ] and @xmath90-based approaches . these @xmath253 and @xmath250 plots are for edge - on @xmath254 binaries containing maximally spinning dominant bh having @xmath255 . the qualitative and quantitative nature of these plots are essentially identical to the double spin binaries of figure [ figure : q_m_phi ] . , width=309,height=279 ] the lsc also developed lalsuite spintaylort2 code that employs a slightly different version of @xmath124 equation while implementing the @xmath30-based precessing convention . note that it is possible to construct various pn approximants to model inspiral gws by using the same pn accurate expressions for the conserved energy , far - zone energy flux and the _ energy balance argument _ @xcite . the differential equation for @xmath19 in the spintaylort2 approximant arises from the following considerations . we begin by displaying ( symbolically ) the 3.5pn accurate expression for @xmath124 , employed in the lalsuite spintaylort4 code , as @xmath256 where the coefficients @xmath257 are functions of @xmath258 and @xmath259 ( euler s constant ) when including only the non - spinning contributions . however , the coefficients from @xmath260 to @xmath261 additionally depend on @xmath262 and @xmath61 to incorporate various spin contributions . the explicit expressions for these @xmath257 coefficients may be obtained either from equation ( 3.16 ) in @xcite or from the lalsuite spintaylort4 code itself . with these inputs , one defines the differential equation for @xmath19 in the spintaylort2 approximant to be @xmath263 this construction ensures that the coefficients @xmath264 are going to depend on various @xmath257 coefficients due to the binomial expansion of the denominator of equation ( [ eq_t2 ] ) that includes all the @xmath6 contributions . for example , the @xmath265 coefficient is going to depend explicitly on all the @xmath257 coefficients whereas the @xmath266 coefficient depends only on the @xmath267 and @xmath268 coefficients . we observe that the resulting spin - squared terms are usually not incorporated into the @xmath269 and @xmath265 coefficients in the lalsuite spintaylort2 code of lsc . and @xmath206 as functions of @xmath208 that invoke @xmath30 and @xmath90-based spintaylort2 approximants . we are dealing with edge - on maximally spinning bh binaries having @xmath270 and @xmath271 . the @xmath225 and @xmath206 estimates are similar to those obtained using spintaylort4 approximant as shown in the right panel plots of figure [ figure : q_m_phi ] . , width=309,height=279 ] we construct the differential equation for @xmath19 in our spintaylort2 approximant in a similar manner . therefore , the various @xmath264 coefficients are given in terms of @xmath257 coefficients which can be extracted from our equations ( [ eq_dxdt ] ) , ( [ eq_dxdt_l ] ) and ( [ eq_sk ] ) for @xmath124 . for example , the expressions for @xmath268 and @xmath272 in our approach are given by @xmath273 + \left [ -\frac{743}{336}-\frac{11\eta}{4}\right ] ^2 \ , . \end{aligned}\ ] ] in the @xmath90-based spintaylort2 approximant , we incorporate our additional 3pn accurate spin - spin terms which contribute _ only _ to the @xmath274 coefficient . in other words , @xmath275 contains our additional 3pn order spin - squared terms via the spintaylort4 based @xmath276 coefficient . we would like to emphasize that we are not incorporating possible 3pn and 3.5pn order spin - spin terms arising due the binomial expansion . this is to make sure that the @xmath30 and @xmath90-based spintaylort2 approximants differ only by our additional 3pn order spin - spin terms . we explore the implications of these additional 3pn order terms again with the help of @xmath175 computations , where @xmath180 and @xmath181 now stand for the @xmath30-based and @xmath90-based waveform families that employ the above described pn accurate expressions for @xmath124 . in figure [ figure : m_phi_q_t2 ] , we plot @xmath233 and @xmath253 for edge - on @xmath277 maximally spinning bh binaries having dominant spin - orbit misalignment equal to @xmath218 at @xmath163 . the @xmath206 and @xmath225 estimates are similar to those obtained using spintaylort4 approximant as shown in the right panel plots of figure [ figure : q_m_phi ] . it may be argued that the inclusion of our 3pn order spin - squared terms to the differential equation for @xmath19 is not desirable . this is because we do not incorporate 3pn order contributions to @xmath124 that arise from the next - to - leading order spin - spin interactions . these terms are not included as they are yet to be computed in the literature . clearly , the present match computations should be repeated when these 3pn order contributions are explicitly available . however , this should not prevent one from probing the data analysis implications of the adiabatic approximation employed in the @xmath0-based precessing convention of @xcite . this could be especially useful while invoking pn accurate differential equation for @xmath19 that incorporate higher order spin contributions as pursued in lalsuite spintaylort4 and lalsuite spintaylort2 codes , developed by the lsc to implement the @xmath0-based precessing convention . hopefully , the data analysis implications of our 3pn order terms should provide additional motivation to explicitly obtain the next - to - leading order spin - spin contributions to @xmath124 . in contrast , it should be noted that the additional pn contribution to @xmath278 , given by equation ( [ eq_phidot ] ) , are fully accurate to 3pn order . the fact that such additional precessional effects enter the expression for @xmath278 only at the 3pn order should justify naming our approach as @xmath90-based precessing convention . it is worthwhile to remember that precession induced modulations enter the differential equation for the orbital phase at 1.5pn order in the absence of precessing convention . we developed a precessing convention to model inspiral gws from generic spinning compact binaries that employs pn accurate orbital angular momentum @xmath1 to describe binary orbits and to construct the required precessing source frame . the main motivation for our approach is the usual practice of using pn accurate precessional equation , appropriate for @xmath1 , to evolve the newtonian orbital angular momentum @xmath0 while constructing inspiral waveforms . we showed that this practice leads to higher order pn corrections to @xmath18 equation . a set of differential equations and the quadrupolar order gw polarization states in certain frame - less convention are developed to model inspiral gws in our @xmath1-based approach . we explained why the differential equations for the orbital phase and frequency will have additional 3pn order terms in our approach compared to the usual @xmath0-based implementation of precessing convention . the influence of these additional 3pn order terms were explored with the help of _ match _ computations involving @xmath1 and @xmath0-based inspiral waveforms for spinning compact binaries with physically equivalent orbital and spin configurations at the initial epoch . we adapted both lalsuite spintaylort4 and lalsuite spintaylort2 codes , developed by the lsc , while implementing @xmath0-based precessing convention . the resulting match estimates indicate that our additional 3pn order terms should not be neglected for a substantial fraction of unequal mass bh binaries . it will be useful to pursue our match computations for an extended range of the relevant parameter space . the present computations should also be extended by invoking 1.5pn order amplitude corrections to both families of inspiral waveforms in the frame - less convention ( the lalsuite spintaylort4 and lalsuite spintaylort2 codes do incorporate such amplitude corrections ) . this requires us to add certain additional 1.5pn order amplitude corrections to the usual expressions for @xmath7 and @xmath8 as noted in @xcite . investigating the influence of these additional 3pn ( 1.5pn ) order terms in phase ( amplitude ) while estimating the gw measurement accuracies of compact binary parameters will be interesting . it should also be worthwhile to probe the influence of these additional terms during the construction of inspiral - merger - ringdown waveforms from generic spinning bh binaries with the help of the effective - one - body approach . we list below the regular contributions to @xmath124 that we employ while implementing our version of the spintaylort4 approximant . these contributions , adapted from @xcite , read @xmath279 + x^{3/2}\,\biggl [ 4\ , \pi -\frac{47}{3}\ , s_{k } -\frac{25}{4}\ , ( x_1-x_2)\,\sigma_{k } \biggr ] \nonumber \\ & \quad+x^2\,\biggl[\bigl ( \frac{34103}{18144}+\frac{13661}{2016}\eta + \frac{59}{18}\eta^2 \bigr ) - \frac{1}{48 } \eta \chi_1 \chi_2 \,\bigl ( 247 ( { \boldsymbol{s}}_1 \cdot { \boldsymbol{s}}_2 ) -721 ( { \boldsymbol{s}}_1 \cdot { \boldsymbol{k } } ) ( { \boldsymbol{s}}_2 \cdot { \boldsymbol{k } } ) \bigr ) \nonumber \\ & \quad+x_1 ^ 2\ , \chi_1 ^ 2\ , \bigl ( \frac{5}{2}\ , ( 3\ , ( { \boldsymbol{s}}_1 \cdot { \boldsymbol{k}})^2 - 1)+\frac{1}{96}\ , ( 7-({\boldsymbol{s}}_1 \cdot { \boldsymbol{k}})^2)\bigr ) \nonumber \\ & \quad+x_2 ^ 2\ , \chi_2 ^ 2\ , \bigl ( \frac{5}{2}\ , ( 3\ , ( { \boldsymbol{s}}_2 \cdot { \boldsymbol{k}})^2 - 1 ) + \frac{1}{96}\ , ( 7-({\boldsymbol{s}}_2 \cdot { \boldsymbol{k}})^2)\bigr)\biggr ] \nonumber \\ & \quad+ x^{5/2}\,\biggl[-\frac{4159}{672}\ , \pi-\frac{5861}{144}\ , s_{k}-\frac{809}{84}\ , ( x_1-x_2)\ , \sigma_{k } \nonumber \\ & \quad+\eta\ , \bigl ( -\frac{189}{8}\ , \pi + \frac{1001}{12}\ , s_{k } + \frac{281}{8}\ , ( x_1-x_2)\ , \sigma_{k}\bigr ) \biggr ] \nonumber \\ & \quad + x^3\ , \biggl [ \frac{16447322263}{139708800 } + \frac{16}{3}\ , \pi^2 -\frac{1712}{105}\ , \gamma_{\rm e}-\frac{856}{105}\ , \ln[16x ] -\frac{188}{3}\ , \pi\ , s_{k } \nonumber \\ & \quad-\frac{151}{6}\ , \pi\ , ( x_1-x_2)\ , \sigma_{k}+\eta\ , \bigl ( -\frac{56198689}{217728}+\frac{451}{48}\ , \pi^2\bigr ) + \frac{541}{896}\ , \eta^2-\frac{5605}{2592}\ , \eta^3 \biggr ] \nonumber \\ & \quad+x^{7/2}\ , \biggl [ \bigl(-\frac{4323559}{18144}+\frac{436705}{672}\ , \eta - \frac{5575}{27}\ , \eta^2\bigr)\ , s_{k } \nonumber \\ & \quad+(x_1-x_2)\ , \bigl ( -\frac{1195759}{18144}+\frac{257023}{1008}\ , \eta-\frac{2903}{32}\ , \eta^2 \bigr)\ , \sigma_{k } \nonumber \\ & \quad+\pi\ , \bigl ( -\frac{4415}{4032}+\frac{358675}{6048}\ , \eta + \frac{91495}{1512}\ , \eta^2\bigr)\biggr ] \biggr \ } \,,\end{aligned}\ ] ] where @xmath280 and @xmath281 are given by we would like to stress that in the above equations we have merely replaced @xmath30 appearing in equation ( 3.16 ) of @xcite by @xmath90 . this differential equation for @xmath19 is also employed in the usual implementation of the spintaylort4 approximant , as provided by lalsuite spintaylort4 code , while using @xmath30 to describe the binary orbits . jaranowski p and schfer g 2012 _ phys . d _ * 86 * 061503 + foffa s and sturani r 2013 _ phys . * 87 * 064011 + jaranowski p and schaefer g 2013 _ phys . d _ * 87 * 081503 + damour t , jaranowski p and schfer g 2014 _ phys . * 89 * 064058 steinhoff j , schfer g and hergt s 2008 _ phys . d _ * 77 * 104018 + steinhoff j , hergt s and schfer g 2008 _ phys . d _ * 78 * 101503 + hergt s and schfer g 2008 _ phys . d _ * 78 * 124004 + hergt s , steinhoff j and schfer g 2010 _ class . quantum grav . _ * 27 * 135007 + hartung j and steinhoff j 2011 _ annalen der physik _ * 523 * 919 + porto r a and rothstein i z 2008 _ phys d _ * 78 * 044012 + porto r a and rothstein i z 2008 _ phys . d _ * 78 * 044013 + levi m and steinhoff j 2014 _ j. cosmol . astropart . phys . _ jcap12(2014)003 and references therein