text
stringlengths
649
4.42k
synonym_substitution
stringlengths
759
4.5k
butter_fingers
stringlengths
649
4.42k
random_deletion
stringlengths
453
2.31k
change_char_case
stringlengths
649
4.42k
whitespace_perturbation
stringlengths
764
5.02k
underscore_trick
stringlengths
649
4.42k
!:=\!\bar{g}(x^{\pm 1}_1,\ldots,x^{\pm 1})$ has a root $\mu_0\!\in\!({\mathbb{Z}}/p^{2L+1}{\mathbb{Z}})^n\setminus\{{\mathbf{O}}\}$ for some choice of signs, some choice of $i$, and some choice of initial term polynomial $\bar{f}$ of $f$ so that $\bar{g}(x)\!=\!x^{-a_i}\bar{f}(x)$. Writing $\bar{h}(x)\!=\!\gamma_0+\gamma_{i_1}x^{\alpha_{i_1}}+\cdots+ \gamma_{i_r}x^{\alpha_{i_r}}$ as before, it is clear that ${{\mathrm{ord}}}_p(\gamma_i \mu^{\alpha_i})\!\leq\! {{\mathrm{ord}}}_p \gamma_0$ for some $i$ by the ultrametric inequality. So then, by Proposition \[prop:mat\], ${{\mathrm{ord}}}_p \bar{h}'(\mu)\!\leq\!L$, and then by Hensel’s Lemma, $\bar{h}$ has a root $\mu'\!\in\!{{\mathbb{Z}}^n}_p\setminus\{{\mathbf{O}}\}$. By Corollary \[cor:toric\], $h(x)\!:=\!\gamma_0+\gamma_1x^{\alpha_1}+\cdots+\gamma_nx^{\alpha_n}$ must then have a root $\mu\!\in\!({\mathbb{Z}}_p\setminus\{{\mathbf{O}}\})^n$. So by the definition of $h$, it is then clear that defining $\zeta_i\!=\!\mu^{\pm 1}_i$ for a suitable choice of signs, $\zeta\!:=\!(\zeta_1,\ldots,\zeta_n)$ is a root of $f$. [$\blacksquare$]{} [**Part (b):**]{} Since the Legendre symbol $\left(\frac{a}{p} \right)$ can be evaluated within $O((\log a)(\log p))$ bit operations [@bs Thm. 5.9.3, pg. 113], the criteria from Theorem \[thm:quad\] can clearly be checked in time polynomial in ${\mathrm{size}}(f)$. So we are done. [$\blacksquare$]{} [**Part (c):**]{} Via the Smith Normal
!: = \!\bar{g}(x^{\pm 1}_1,\ldots, x^{\pm 1})$ has a root $ \mu_0\!\in\!({\mathbb{Z}}/p^{2L+1}{\mathbb{Z}})^n\setminus\{{\mathbf{O}}\}$ for some choice of signs, some option of $ i$, and some option of initial terminus polynomial $ \bar{f}$ of $ f$ so that $ \bar{g}(x)\!=\!x^{-a_i}\bar{f}(x)$. Writing $ \bar{h}(x)\!=\!\gamma_0+\gamma_{i_1}x^{\alpha_{i_1}}+\cdots+ \gamma_{i_r}x^{\alpha_{i_r}}$ equally before, it is absolved that $ { { \mathrm{ord}}}_p(\gamma_i \mu^{\alpha_i})\!\leq\! { { \mathrm{ord}}}_p \gamma_0 $ for some $ i$ by the ultrametric inequality. So then, by Proposition \[prop: mat\ ], $ { { \mathrm{ord}}}_p \bar{h}'(\mu)\!\leq\!L$, and then by Hensel ’s Lemma, $ \bar{h}$ has a beginning $ \mu'\!\in\!{{\mathbb{Z}}^n}_p\setminus\{{\mathbf{O}}\}$. By Corollary \[cor: toric\ ], $ h(x)\!:=\!\gamma_0+\gamma_1x^{\alpha_1}+\cdots+\gamma_nx^{\alpha_n}$ must then have a root $ \mu\!\in\!({\mathbb{Z}}_p\setminus\{{\mathbf{O}}\})^n$. thus by the definition of $ h$, it is then clear that defining $ \zeta_i\!=\!\mu^{\pm 1}_i$ for a desirable choice of signs, $ \zeta\!:=\!(\zeta_1,\ldots,\zeta_n)$ is a root of $ f$. [ $ \blacksquare$ ] { } [ * * Part (b ): * * ] { } Since the Legendre symbol $ \left(\frac{a}{p } \right)$ can be measure within $ O((\log a)(\log p))$ bit operations [ @bs Thm.   5.9.3, pg.   113 ], the criteria from Theorem \[thm: quad\ ] can clearly be checked in clock time polynomial in $ { \mathrm{size}}(f)$. So we are done. [ $ \blacksquare$ ] { } [ * * Part (c ): * * ] { } Via the Smith Normal
!:=\!\bar{h}(x^{\pm 1}_1,\ldots,x^{\pm 1})$ has a roou $\mu_0\!\in\!({\mathbb{Z}}/p^{2L+1}{\majhvb{Z}})^n\sevminus\{{\mzthbf{O}}\}$ fur some choice of signs, some cyoice of $i$, and some choice uf initiap term pilyninial $\bar{f}$ of $f$ so bkat $\bzv{g}(x)\!=\!x^{-a_n}\ber{f}(x)$. Writing $\bat{h}(x)\!=\!\gamma_0+\gammd_{i_1}x^{\alpha_{i_1}}+\cdots+ \caomc_{i_r}x^{\alpha_{i_r}}$ as before, it is clear thwt ${{\mathtm{lrd}}}_p(\gamma_i \mu^{\ajpha_p})\!\lqq\! {{\mafhrm{ord}}}_p \gamma_0$ for some $i$ by the umtrametgic inequality. So then, by Proposition \[prop:mwt\], ${{\mwthrm{ord}}}_p \bar{h}'(\mu)\!\lee\!L$, and then by Rwnsel’s Lemma, $\bar{h}$ has a root $\mu'\!\in\!{{\mzthbb{Z}}^n}_p\setminus\{{\mathbf{O}}\}$. By Coroulary \[cor:toric\], $y(x)\!:=\!\tamld_0+\gamma_1x^{\alphe_1}+\cdots+\damma_nx^{\alpha_k}$ must dhen hafe a root $\mu\!\in\!({\katibb{Z}}_p\setminus\{{\mathbf{O}}\})^n$. So uy the definition of $h$, it is dhzn clear that definint $\zeta_i\!=\!\ku^{\pm 1}_i$ fue a sujtebls choife kf signs, $\zsta\!:=\!(\zeta_1,\ldotw,\zeta_n)$ is a root of $f$. [$\vlacksquare$]{} [**Padt (b):**]{} Synse the Legendre symbol $\left(\frac{a}{p} \right)$ ban ge evaluated within $O((\lot a)(\log p))$ bit operatiojs [@bs Thm. 5.9.3, pg. 113], the criteria from Theorem \[thm:quad\] can clearly be ciezkeb in tkne polynomial in ${\mathrm{size}}(f)$. So we are done. [$\blasisausre$]{} [**Part (c):**]{} Via bhe Smith Normal
!:=\!\bar{g}(x^{\pm 1}_1,\ldots,x^{\pm 1})$ has a root $\mu_0\!\in\!({\mathbb{Z}}/p^{2L+1}{\mathbb{Z}})^n\setminus\{{\mathbf{O}}\}$ choice signs, some of $i$, and polynomial of $f$ so $\bar{g}(x)\!=\!x^{-a_i}\bar{f}(x)$. Writing $\bar{h}(x)\!=\!\gamma_0+\gamma_{i_1}x^{\alpha_{i_1}}+\cdots+ as before, it is clear that \mu^{\alpha_i})\!\leq\! {{\mathrm{ord}}}_p \gamma_0$ for some $i$ by the ultrametric inequality. So then, by \[prop:mat\], ${{\mathrm{ord}}}_p \bar{h}'(\mu)\!\leq\!L$, and then by Hensel’s Lemma, $\bar{h}$ has a root $\mu'\!\in\!{{\mathbb{Z}}^n}_p\setminus\{{\mathbf{O}}\}$. Corollary $h(x)\!:=\!\gamma_0+\gamma_1x^{\alpha_1}+\cdots+\gamma_nx^{\alpha_n}$ then a root $\mu\!\in\!({\mathbb{Z}}_p\setminus\{{\mathbf{O}}\})^n$. So by the definition of $h$, it is then clear that defining $\zeta_i\!=\!\mu^{\pm for a suitable choice of signs, $\zeta\!:=\!(\zeta_1,\ldots,\zeta_n)$ is root of $f$. [$\blacksquare$]{} (b):**]{} Since the Legendre symbol \right)$ be evaluated $O((\log p))$ operations [@bs Thm. pg. 113], the criteria from Theorem \[thm:quad\] can clearly be checked in time polynomial in ${\mathrm{size}}(f)$. So are done. (c):**]{} Via Smith
!:=\!\bar{g}(x^{\pm 1}_1,\ldots,x^{\pm 1})$ has a root $\mu_0\!\In\!({\mathbb{Z}}/p^{2l+1}{\mathBb{Z}})^N\seTmInus\{{\MathBf{O}}\}$ for some choiCE of sIgns, some choice of $i$, and soMe choIcE Of inITiAl terM polynoMIaL $\BAr{f}$ Of $F$ sO thAt $\BAr{G}(x)\!=\!x^{-a_i}\Bar{F}(x)$. WritiNg $\bar{h}(x)\!=\!\gamMa_0+\gAmMa_{i_1}x^{\alpha_{i_1}}+\cdOTs+ \Gamma_{i_r}x^{\alPha_{I_r}}$ as before, it Is cLear thAt ${{\MatHRm{ord}}}_P(\gaMma_i \mU^{\alpha_I})\!\Leq\! {{\matHrm{ord}}}_p \gaMmA_0$ For somE $I$ by the uLTRaMetrIc inequality. So theN, By pRoposition \[prop:Mat\], ${{\matHrM{OrD}}}_P \Bar{H}'(\mu)\!\Leq\!L$, and theN bY HensEL’s Lemma, $\BAr{H}$ HAS a rOOt $\mu'\!\in\!{{\mathbb{Z}}^N}_p\setminus\{{\mAThbF{O}}\}$. By CoRoLlaRY \[cor:toRic\], $h(x)\!:=\!\GaMMa_0+\gAmma_1x^{\alpha_1}+\cDots+\Gamma_nx^{\alPha_n}$ muST then haVE a root $\mU\!\in\!({\matHbb{z}}_p\sEtmiNUs\{{\MaThbF{O}}\})^N$. so bY ThE deFIniTion of $h$, iT iS tHen clEar tHAT DEfinIng $\Zeta_I\!=\!\mu^{\pm 1}_I$ for a suitable ChoIce oF SigNs, $\zetA\!:=\!(\zeta_1,\LdotS,\zEta_n)$ iS a root Of $f$. [$\blAcKsquare$]{} [**Part (b):**]{} SinCe thE Legendre SymBoL $\leFt(\Frac{a}{P} \Right)$ cAn bE evAluated Within $O((\LOg a)(\LoG P))$ BIt Operations [@bs Thm. 5.9.3, pg. 113], tHe CRItEria from theoreM \[ThM:qUAd\] can cleArLy bE cheCKEd in tIme pOLyNomial in ${\Mathrm{SIzE}}(f)$. so we are DoNe. [$\blacKsQuaRe$]{} [**PArt (c):**]{} VIA the smith NOrmal
!:=\!\bar{g}(x^{\pm 1}_1,\ ldots,x^{\ pm 1} )$has a roo t $\ mu_0\!\in\!({\ m athb b{Z}}/p^{2L+1}{\mathbb {Z}}) ^n \ setm i nu s\{{\ mathbf{ O }} \ } $ f or s ome c h oi ce of si gns, so me choiceof$i $, and somec ho ice of ini tia l term polyn omi al $\b ar {f} $ of $ f$so th at $\b a r{g}(x )\!=\!x^{ -a _ i}\bar { f}(x)$. W ri ting $\bar{h}(x)\!=\! \ ga m ma_0+\gamma_{i _1}x^{ \a l ph a _ {i_ 1}} +\cdots+ \ ga mma_{ i _r}x^{\ a lp h a _ {i_ r }}$ as before , it is cle a r t hat ${ {\ mat h rm{ord }}}_p (\ g amm a_i \mu^{\a lpha _i})\!\le q\! {{ \ mathrm{ o rd}}}_p \gamm a_0 $ f or s o me $ i$by the ul tra m etr ic inequ al it y. So the n , b y Pr opo siti on \[ prop:mat\], $ {{\ math r m{o rd}}} _p \b ar{h }' (\mu) \!\leq \!L$, a nd then by Hens el’s Lemma, $ \ba r{ h}$ h as ar oot $\ mu' \!\ in\!{{\ mathbb{ Z }}^ n} _ p \ se tminus\{{\mathbf{O }} \ } $. By Coro llary\ [c or : toric\], $ h(x )\!: = \ !\gam ma_0 + \g amma_1x^ {\alph a _1 }+ \cdots+ \g amma_n x^ {\a lph a_n}$ must thenhave a r oot $ \ mu\!\in\!({\ma t hbb{Z}}_p\set m in u s \{ { \mat hbf {O}}\})^n$. Sob y th e de f in iti o n of$h$,it is then clear that def in ing $\ zeta_ i\!=\!\mu^{\p m 1}_i$ fo r a suitabl e ch o ic e of signs, $\z eta\! :=\!(\zeta _ 1,\ldots ,\zet a_n)$ is a root o f $f$. [$\ bla cks qua re$ ] { } [**Part (b): * * ]{}Si nce the Le gendresym bol $\ lef t( \frac{a}{ p} \righ t) $ca nbeevalu a ted with in $O (( \lo g a)( \ log p) )$ bi t op er at i ons [@bs T h m. 5 .9.3 ,pg . 11 3], t he cr iter i a f rom The orem \[th m:q u ad\] c an clearl y be checkedin time poly no mia l in $ { \ mathrm{s ize}}(f)$. So we are do n e. [$\b lac ksqua re$] {} [**Pa rt(c):** ]{} Via th e Smit h Nor ma l
!:=\!\bar{g}(x^{\pm 1}_1,\ldots,x^{\pm_1})$ has_a root $\mu_0\!\in\!({\mathbb{Z}}/p^{2L+1}{\mathbb{Z}})^n\setminus\{{\mathbf{O}}\}$ for_some choice_of_signs, some_choice_of $i$, and_some choice of_initial term polynomial $\bar{f}$_of $f$ so_that_$\bar{g}(x)\!=\!x^{-a_i}\bar{f}(x)$. Writing $\bar{h}(x)\!=\!\gamma_0+\gamma_{i_1}x^{\alpha_{i_1}}+\cdots+ \gamma_{i_r}x^{\alpha_{i_r}}$ as before, it is clear that ${{\mathrm{ord}}}_p(\gamma_i \mu^{\alpha_i})\!\leq\! {{\mathrm{ord}}}_p \gamma_0$ for some_$i$_by the_ultrametric_inequality._So then, by Proposition \[prop:mat\],_${{\mathrm{ord}}}_p \bar{h}'(\mu)\!\leq\!L$, and then by_Hensel’s Lemma,_$\bar{h}$ has a root $\mu'\!\in\!{{\mathbb{Z}}^n}_p\setminus\{{\mathbf{O}}\}$. By Corollary \[cor:toric\],_$h(x)\!:=\!\gamma_0+\gamma_1x^{\alpha_1}+\cdots+\gamma_nx^{\alpha_n}$_must then have_a root $\mu\!\in\!({\mathbb{Z}}_p\setminus\{{\mathbf{O}}\})^n$. So by the definition of $h$,_it is then clear that defining_$\zeta_i\!=\!\mu^{\pm 1}_i$ for_a_suitable_choice of signs, $\zeta\!:=\!(\zeta_1,\ldots,\zeta_n)$_is a root of $f$. [$\blacksquare$]{} [**Part_(b):**]{} Since the Legendre symbol $\left(\frac{a}{p} \right)$_can be evaluated within $O((\log a)(\log p))$_bit operations [@bs Thm. 5.9.3, pg. 113], the_criteria from Theorem \[thm:quad\] can_clearly be_checked in time polynomial in_${\mathrm{size}}(f)$. So we_are done._[$\blacksquare$]{} [**Part (c):**]{} Via_the Smith Normal
the data where different data sets from the literature have different calibration errors. Also, these discrepancies could be in part attributed to the internal dynamics of the observed galaxies because of the different fiber and/or slit placements. We can improve the fit even further by restricting ourselves to the largest homogeneous dataset from the literature [*i.e.*]{}, the 63 new redshifts measurements of CD96. We then find excellent agreement between our measurements (the dotted histogram and Gaussian fit) with a mean offset of 3 ${\rm km\,s^{-1}}$ and an RMS scatter of 24 ${\rm km\,s^{-1}}$. Overall, the agreement of our redshift determinations with the literature is remarkably good. The RMS scatter between datasets indicates that our estimate of the errors in the wavelength calibration is accurate. Spectral Classifications {#classifications} ======================== In this section, we consider the spectral classification of our Coma galaxies. We have investigated four different algorithms, all of which could be implemented for the main SDSS galaxy survey. The first algorithm is based on visual inspections of the spectra but quantified using measurements of the equivalent widths[^2] of the lines seen in the spectrum. The next two algorithms, Wavelets and Principal Component Analysis (PCA), were used to objectively classify the spectra using the visual inspections to define relevant thresholds in wavelet and eigenspace. The final classification scheme attempts to define a physical classification scheme based on synthetic models of galaxies. Line Strength Classifications {#lines} ----------------------------- We started by visually classifying the spectra into five classes which could then be compared to the other algorithms. We first divide the spectra into two broad classes: emission and absorption line galaxies. Absorption line spectra were then sub–divided into normal absorption line systems and objects with strong Balmer lines or post–starburst galaxies. The spectra showing emission lines were sub–divided into three categories depending on the strength of the emission lines. For our visual and line–strength classification scheme, we were able to use both the blue side of the spectrum and the presence of H$\alpha$ in the red end of the spectrum. Unfortunately the other classification algorithms could not use the red side of the spectrum due to the condensation problems discussed above. Since H$\alpha$ is a powerful indicator of star–formation, we believe that one of the classification schemes should utilize these data even if the others could not. Moreover, the other algorithms used our line strength classifications to quantify star–formation in Coma and therefore, it was justified to make the line
the data where different data sets from the literature get unlike calibration errors. Also, these discrepancy could be in function attributed to the internal moral force of the observed galaxies because of the different roughage and/or slit placements. We can improve the burst even further by restricting ourselves to the large homogeneous dataset from the literature [ * i.e. * ] { }, the 63 newfangled redshifts measurements of CD96. We then find excellent agreement between our measurements (the dotted histogram and Gaussian paroxysm) with a mean offset of 3 $ { \rm km\,s^{-1}}$ and an RMS scatter of 24 $ { \rm km\,s^{-1}}$. Overall, the agreement of our redshift determination with the literature is remarkably good. The RMS scatter between datasets indicates that our estimate of the mistake in the wavelength calibration is accurate. Spectral Classifications { # classifications } = = = = = = = = = = = = = = = = = = = = = = = = In this section, we consider the spectral classification of our Coma galaxy. We have investigated four different algorithms, all of which could be implemented for the main SDSS galaxy survey. The first algorithm is based on visual inspections of the spectra but quantified using measurements of the equivalent widths[^2 ] of the lines seen in the spectrum. The following two algorithms, Wavelets and Principal Component Analysis (PCA), were used to objectively relegate the spectrum use the visual inspections to define relevant thresholds in wavelet and eigenspace. The final classification schema attempts to define a physical classification scheme based on celluloid models of galaxies. Line Strength Classifications { # lines } ----------------------------- We start by visually classify the spectra into five classes which could then be compared to the other algorithms. We foremost divide the spectra into two broad classes: emission and absorption line galaxies. Absorption tune spectra were then sub – separate into normal absorption line organization and objects with potent Balmer lines or post – starburst galaxy. The spectra show emission lines were sub – divided into three categories depending on the military capability of the emission tune. For our visual and line – strength classification scheme, we were able to practice both the blue side of the spectrum and the bearing of H$\alpha$ in the red end of the spectrum. Unfortunately the other categorization algorithms could not use the red side of the spectrum due to the condensation problems discussed above. Since H$\alpha$ is a powerful indicator of star – formation, we think that one of the classification schemes should use these datum even if the others could not. Moreover, the other algorithm used our line strength categorization to quantify star – formation in Coma and therefore, it was justified to make the line
thf data where different dxta sets from tkw litecature gave difwerent calibration errors. Aldo, thest discrepancies couud be in iart attrubuttd to the internal dynamigf of bhe oysxrved galaxies necause of dhe different xicex and/or slit placements. We can improde the gih even further by geftridnikg ourselves to the largest homofeneous dataset from the literature [*i.e.*]{}, the 63 nea refshifts measuremenhs of CD96. We theg find excelldnt agreemtnc between oor measurements (the dotted histoeram cnd Gaussiab dit) fith a mean offsvt of 3 ${\rm km\,s^{-1}}$ and an SMS scayter of 24 ${\rm km\,x^{-1}}$. Oteraol, the agreement of onr redshift determinwtions widh the literature iw eemarnablf goue. Tfe DMX acatteg bxtween dataaets indicares that our estimaue jd the errors jn the wwvelength calibration is accurate. Spectrdl Dlassifications {#classifucations} ======================== In this sectiln, we confider the spectral classification of our Coma galdxies. De kqve ivcedtigated four different algorithms, all of whisg volld be implementeb for the main SCSD bwlaxy survey. Jhe first zlgorithm is based on visoal inwpections of yhe spectra but quantified ysing measurvmenrs of the equivaleut widths[^2] of the linex seen in the spectrum. Che nest two algogithms, Wabdlets and Princioal Cmmponent Analysis (PCA), were used to ibjeetively zlasxify tre spectra usinn the visual inspechions tm define rflevant thresholds in wavelet aiv eigenspace. Jhe fital clasfificstion scheme wttempts to deyine a pkysicau classifibation scieme based og synthetic mmfels of galaeies. Line Ftrebgth Classiwkcations {#lines} ----------------------------- Ee starteb by cisually classifyikg thg apectra into firt coasses which coild thqn bx com[dred to the mthef auboritfms. We firsb dkvidr the spectra into tfo bdoad classes: emissook and absirption jine galaxies. Absorption line skectra wece them sob–divided into normal absorptioh line sydteis and objeces wlth ftrong Balker lines or post–starburst galaxies. The spectra showing gmission lines were sob–dlvided into vhree sategoriev depending on the srrength of the emlssion lines. For our viaual atd lije–strength classification scheme, we were able to use both the blue side od the spectrum and the presvncz oy H$\alphw$ in vhe red end of thv spectrum. Unfortunately the othxr classifhcction algorithms could not usr ghe red side uf the spectrum due to tge condemsation problems discussed abovr. Since H$\alpha$ is a powevful indixator of star–focmation, we believe tnat onz of tye claswifigation schemes shpupd utilizf thtse bata even if the others could not. Moreoceg, the other algocithms used lmr ling strengjh classificationa to euantkfy star–formayion in Coma amd therefore, it was buftified to makt the line
the data where different data sets from have calibration errors. these discrepancies could the dynamics of the galaxies because of different fiber and/or slit placements. We improve the fit even further by restricting ourselves to the largest homogeneous dataset the literature [*i.e.*]{}, the 63 new redshifts measurements of CD96. We then find agreement our (the histogram and Gaussian fit) with a mean offset of 3 ${\rm km\,s^{-1}}$ and an RMS scatter 24 ${\rm km\,s^{-1}}$. Overall, the agreement of our determinations with the literature remarkably good. The RMS scatter datasets that our of errors the wavelength calibration accurate. Spectral Classifications {#classifications} ======================== In this section, we consider the spectral classification of our Coma galaxies. have investigated algorithms, all which be for the main survey. The first algorithm is based of the spectra but quantified using measurements of equivalent widths[^2] the lines seen in the spectrum. next two algorithms, Wavelets and Principal Component Analysis were used to objectively classify the spectra using the visual inspections to define relevant thresholds and eigenspace. The final scheme attempts to a classification based synthetic models galaxies. Line Strength Classifications {#lines} ----------------------------- We started by visually classifying spectra into five classes which could then be compared to algorithms. first divide the into two broad classes: and line galaxies. Absorption line then into systems objects strong Balmer lines or galaxies. The spectra showing emission were sub–divided into three of the emission lines. For our visual and classification scheme, we were able to use the blue side of the spectrum and the presence of H$\alpha$ in red end spectrum. Unfortunately the other classification algorithms could not the red side of spectrum due to the condensation problems discussed above. Since is powerful indicator star–formation, we believe one of the schemes should utilize even if others not. algorithms used our line strength classifications quantify in Coma and therefore, it justified the line
the data where different data Sets from thE liteRatUre HaVe diFferEnt calibration ERrorS. Also, these discrepancieS coulD bE In paRT aTtribUted to tHE iNTErnAl DyNamIcS Of The obSerVed galaXies becausE of ThE different fiBEr And/or slit pLacEments. We can iMprOve the FiT evEN furtHer By resTrictiNG ourseLves to the LaRGest hoMOgeneouS DAtAset From the literature [*I.E.*]{}, tHE 63 new redshifts mEasureMeNTs OF cD96. WE thEn find exceLlEnt agREement bETwEEN Our MEasurements (thE dotted histOGraM and GaUsSiaN Fit) witH a meaN oFFseT of 3 ${\rm km\,s^{-1}}$ and An RMs scatter oF 24 ${\rm km\,s^{-1}}$. oVerall, tHE agreemEnt of oUr rEdsHift DEtErMinAtIOns WItH thE LitErature iS rEmArkabLy goOD. tHE RMS ScaTter BetweEn datasets indIcaTes tHAt oUr estImate Of thE eRrors In the wAveleNgTh calibration is AccuRate. SpectRal clAssIfIcatiONs {#clasSifIcaTions} ======================== In This secTIon, We CONSiDer the spectral clasSiFICaTion of ouR Coma gALaXiES. We have iNvEstIgatED Four dIffeREnT algoritHms, all OF wHiCh could Be ImplemEnTed For The maIN SDSs galaxY survey. THe firST algorithm is baSEd on visual insPEcTIOnS Of thE spEctra but quaNtifIEd usIng mEAsUreMEnts oF the eQuIVaLEnt widths[^2] of the lines SeEn in thE specTrum. The next twO algorithmS, wAVelets anD PriNCiPAl Component AnaLysis (pCA), were useD To objectIvely Classify The spectrA USing the vIsuAl iNspEctIONs To define relevANT thrEsHolds in WavElet and EigEnsPacE. ThE fInal classIficatioN sChEmE aTteMpts tO Define a pHySicAl ClaSsifiCAtion sCheme BaseD oN sYNthEtic modELs OF GalaXiEs. line strEnGth ClAssiFIcaTions {#liNes} ----------------------------- We starTed BY visUaLlY classiFying the spectRa Into five clAsSes Which cOULd then be Compared to the other algorIThms. We fIrsT diviDe thE spectra iNto Two broAd cLAsses: eMissioN and aBsOrpTIOn linE GAlAxiEs. absorption LINe sPectrA wEre tHen sub–dIvided into normal abSOrpTion line systeMs aNd obJECtS wiTH sTRonG BALmeR LInes or post–starbUrst galaxiEs. tHe Spectra shoWIng EmIssion lInes werE sub–dIVided inTo three caTegories dEpEndiNG On tHe strength Of the emiSsion lineS. for ouR ViSual aNd lIne–strEnGth ClassIficatIOn sCheme, We were AbLe to usE both ThE blue sidE of the spectrum and the preSence oF H$\alpHa$ iN the red enD of THe sPectrum. UnFortUnately the OthEr cLassiFicATion aLgorIThMs cOUld noT use THe red side OF tHe sPECtRum due to the CONDenSatioN prOBlems dIscuSsed above. Since H$\alPHa$ is a powerful iNdicATOr oF stAR–forMaTion, we believe tHat OnE OF the clasSiFication schEmes shouLd UTilizE these Data evEn if the OTHeRS could Not. MOreOver, the otHer AlGOrithms UsEd OUr line StreNgTh clasSificaTIons TO Quantify star–formAtion IN coma aND thErefoRe, It was juSTifiEd to make thE line
the data where differentdata setsfromthe li te ratu re h ave differentc alib ration errors. Also, t hesedi s crep a nc ies c ould be in p art a tt rib ut e dto th e i nternal dynamicsofth e observed g a la xies becau seof the diffe ren t fibe rand / or sl itplace ments. We can improveth e fit e v en furt h e rby r estricting oursel v es to the largest homog en e ou s dat ase t from the l itera t ure [*i . e. * ] { },t he 63 new red shifts meas u rem ents o fCD9 6 . We t hen f in d ex cellent agr eeme nt betwee n ourm easurem e nts (th e dott edhis togr a man d G au s sia n f it) wit h a mean o ff set o f 3$ { \ r m km \,s ^{-1 }}$ a nd an RMS sca tte r of 24${\rm km\, s^{- 1} }$. O verall , the a greement of our red shift det erm in ati on s wit h the l ite rat ure isremarka b lygo o d . T he RMS scatter bet we e n d atasetsindica t es t h at our e st ima te o f the e rror s i n the wa veleng t hca librati on is ac cu rat e. Spec t ralClassi fication s {#c l assifications} ============= = == = = == = === In this secti on,w e co nsid e rthe spect ral c la s si f ication of our Coma g alaxie s. We have investi gated four d i fferentalgo r it h ms, all of whi ch co uld be imp l ementedfor t he mainSDSS gala x y survey. Th e f irs t a l g or ithm is based o n vi su al insp ect ions of th e s pec tra b ut quanti fied usi ng m ea su rem entso f the eq ui val en t w idths [ ^2] of theline sse e n i n the s p ec t r um.Th enext tw oalgor ithm s , W avelets and Prin cip a l Co mp on ent Ana lysis (PCA),we re used to o bje ctivel y classify the spectra using thev isual i nsp ectio ns t o definerel evantthr e sholds in wa velet a nde i gensp a c e. Th efinal clas s i fic ation s chem e attem pts to define a ph y sic al classifica tio n sc h e me ba s ed onsy n the t i c models of gal axies. Li ne St rength Cla s sif ic ations{#lines } --- - ------- --------- --------- We s t a rte d by visua lly clas sifying t h e spe c tr a int o f ive cl as ses whic h coul d th en be compa re d to t he ot he r algori thms. We first divide t he spe ctraint o two bro adc las ses: emis sion and absor pti onlinegal a xies. Abs o rp tio n line spe c tra weret he n s u b –d ivided into n o rma l abs orp t ion li ne s ystems and object s with strong B alme r lin eso r po st –starburst gal axi es . The spec tr a showing e missionli n es we re sub –divid ed into t hr e e cate gori esdepending on t h e stren gt ho f theemis si on lin es. F o r ou r visual and line– stren g t h cla s sif icati on scheme , wewere ableto use both the b luesideof thesp ectrum an dthe presen c e of H$\a lpha$ in the r ed e ndof the spe c t rum.Unfo rt una tely theo t he r c la s sif icat ion a lg orit hms could not usethe red sid eoft h e spec t ru m due to the co ndens a t ion proble m s di s cu s sed a bove.SinceH$\alph a $ i sa power ful i ndicatorof star–f o rma ti on,we belie ve tha tone o f the classific ation sc hemes sh o u ld uti liz e the s e data e ven if the o thers c o uld not . Mor eover,t heoth eralgo rit hmsu sed our line strength cl assi ficat i ons to q uanti fy star–fo r mati on in Co m a and therefo re, it was just i fie d tomak e t he line
the_data where_different data sets from_the literature_have_different calibration_errors._Also, these discrepancies_could be in_part attributed to the_internal dynamics of_the_observed galaxies because of the different fiber and/or slit placements. We can improve the_fit_even further_by_restricting_ourselves to the largest homogeneous_dataset from the literature [*i.e.*]{},_the 63_new redshifts measurements of CD96. We then find_excellent_agreement between our_measurements (the dotted histogram and Gaussian fit) with a_mean offset of 3 ${\rm km\,s^{-1}}$_and an RMS_scatter_of_24 ${\rm km\,s^{-1}}$. Overall,_the agreement of our redshift determinations_with the literature is remarkably good._The RMS scatter between datasets indicates that_our estimate of the errors in_the wavelength calibration is accurate. Spectral_Classifications {#classifications} ======================== In_this section, we consider the_spectral classification of_our Coma_galaxies. We have_investigated four different algorithms, all of_which could be_implemented for the main SDSS galaxy_survey._The first algorithm_is_based_on visual_inspections of the_spectra_but quantified_using_measurements of the equivalent widths[^2] of_the_lines seen in the spectrum. The next_two algorithms, Wavelets and_Principal_Component Analysis (PCA), were_used to objectively classify the_spectra using the visual inspections to_define relevant_thresholds in_wavelet and eigenspace. The final classification scheme attempts to define a_physical classification scheme based on synthetic_models of galaxies. Line Strength_Classifications {#lines} ----------------------------- We_started_by visually classifying_the_spectra into_five classes which could then be compared_to the_other algorithms. We first divide the_spectra into two broad_classes:_emission and absorption line galaxies. Absorption_line spectra were then sub–divided into_normal absorption line systems and_objects_with_strong Balmer lines or post–starburst_galaxies. The spectra showing emission lines_were sub–divided into_three categories depending on the strength of_the_emission lines. For our visual and line–strength_classification_scheme, we were able to use_both_the_blue side of the spectrum_and the presence of H$\alpha$ in_the red end of the spectrum. Unfortunately the other_classification algorithms could_not use the red side_of_the_spectrum due to the condensation problems discussed above. Since H$\alpha$_is a_powerful indicator of_star–formation, we believe that one of the classification schemes should_utilize these data even if the others_could not. Moreover, the other algorithms used our line strength classifications_to quantify star–formation in Coma and therefore, it_was justified to make the line
1)} \right)\right\}\right. \nonumber \\ & & \hspace{-7.5cm} \times \left. \exp\left\{\sum_{n=0}^{L-1}\mbox{ln} \left( 1 + A_{n}^{2}\sin^{2}\theta_{n}^{(2)} + A_{n}\sin2\theta_{n}^{(2)} \right)\right\}\right> \;. \label{eq:r1r2a}\end{aligned}$$ Using the weak disorder condition we can expand the logarithms and present the correlator $\left <r_{1, L}^{2}r_{2, L}^{2}\right>$ as follows, $$\begin{aligned} \left<r_{1, L}^{2}r_{2, L}^{2}\right> & = & \left<\exp\left\{\sum_{n=0}^{L-1}A_{n}^{2} Z_n \right\} \right> \nonumber \\ & & \hspace{-2cm} \times \left<\exp\left\{\sum_{n=0}^{L-1}A_{n}\left(\sin2\theta_{n}^{(1)} + \sin2\theta_{n}^{(2)}\right)\right\}\right> \label{eq:r1r2b1} \\ \hspace*{-3cm} & \approx & \exp\left\{L\left< A_{n}^{2} \right> \left<Z_n\right> \right\} \times \left<\exp\left(S_{P}\right)\right> \,, \label{eq:r1r2b}\end{aligned}$$ where $$Z_n = \left(\sin^{2}\theta_{n}^{(1)} + \sin^{2}\theta_{n}^{(2)}\right) - \frac{1}{2}\left(\sin^{2}2\theta_{n}^{(1)} + \sin^{2}2\theta_{n}^{(2)}\right)$$ Here we introduced a random variable $S_{P}$, $$S_{P}=\sum_{n=0}^{L-1}A_{n}\left(\sin2\theta_{n}^{(1)} + \sin2\theta_{n}^{(2)}\right), \label{eq:S
1) } \right)\right\}\right. \nonumber \\ & & \hspace{-7.5 cm } \times \left. \exp\left\{\sum_{n=0}^{L-1}\mbox{ln } \left (1 + A_{n}^{2}\sin^{2}\theta_{n}^{(2) } + A_{n}\sin2\theta_{n}^{(2) } \right)\right\}\right > \; . \label{eq: r1r2a}\end{aligned}$$ Using the weak disorder condition we can elaborate the logarithm and present the correlator $ \left < r_{1, L}^{2}r_{2, L}^{2}\right>$ as follows, $ $ \begin{aligned } \left < r_{1, L}^{2}r_{2, L}^{2}\right > & = & \left<\exp\left\{\sum_{n=0}^{L-1}A_{n}^{2 } Z_n \right\ } \right > \nonumber \\ & & \hspace{-2 centimeter } \times \left<\exp\left\{\sum_{n=0}^{L-1}A_{n}\left(\sin2\theta_{n}^{(1) } + \sin2\theta_{n}^{(2)}\right)\right\}\right > \label{eq: r1r2b1 } \\ \hspace*{-3 cm } & \approx & \exp\left\{L\left < A_{n}^{2 } \right > \left < Z_n\right > \right\ } \times \left<\exp\left(S_{P}\right)\right > \, , \label{eq: r1r2b}\end{aligned}$$ where $ $ Z_n = \left(\sin^{2}\theta_{n}^{(1) } + \sin^{2}\theta_{n}^{(2)}\right) - \frac{1}{2}\left(\sin^{2}2\theta_{n}^{(1) } + \sin^{2}2\theta_{n}^{(2)}\right)$$ Here we introduce a random variable $ S_{P}$, $ $ S_{P}=\sum_{n=0}^{L-1}A_{n}\left(\sin2\theta_{n}^{(1) } + \sin2\theta_{n}^{(2)}\right), \label{eq: S
1)} \rigjt)\right\}\right. \nonumber \\ & & \hrpace{-7.5cm} \times \lgfr. \exp\lxft\{\sum_{n=0}^{M-1}\mbox{ln} \ldft( 1 + A_{n}^{2}\sin^{2}\theta_{n}^{(2)} + A_{n}\sin2\thete_{n}^{(2)} \ritht)\ritht\}\right> \;. \label{eq:r1r2a}\end{xligned}$$ Uding the weaj disorder rknditiok we dwn erpend the logaritmms and prevent the correnagox $\left <r_{1, L}^{2}r_{2, L}^{2}\right>$ as follows, $$\begin{ajigned} \lrfh<r_{1, L}^{2}r_{2, L}^{2}\right> & = & \lefu<\ex[\leff\{\sum_{n=0}^{L-1}A_{n}^{2} Z_n \right\} \right> \nonumber \\ & & \hspace{-2bm} \times \left<\exp\lrft\{\sum_{n=0}^{L-1}A_{n}\left(\sin2\theta_{n}^{(1)} + \sij2\theha_{n}^{(2)}\right)\right\}\right> \label{eq:r1r2b1} \\ \hspwxe*{-3cm} & \approx & \exp\left\{L\left< A_{n}^{2} \righj> \left<Z_n\right> \right\} \times \left<\exp\ueft(S_{'}\right)\right> \,, \lqbep{gq:r1r2b}\end{aligied}$$ whvre $$Z_n = \left(\slm^{2}\theta_{t}^{(1)} + \sin^{2}\thrta_{n}^{(2)}\right) - \frsc{1}{2}\lxft(\sun^{2}2\theta_{n}^{(1)} + \sin^{2}2\theta_{n}^{(2)}\rigit)$$ Here we introduceq a randok rariable $S_{P}$, $$S_{P}=\sum_{n=0}^{L-1}A_{n}\oedt(\sin2\jheta_{t}^{(1)} + \skb2\thdta_{h}^{(2)}\rmghf), \label{fq:S
1)} \right)\right\}\right. \nonumber \\ & & \hspace{-7.5cm} \exp\left\{\sum_{n=0}^{L-1}\mbox{ln} 1 + + A_{n}\sin2\theta_{n}^{(2)} \right)\right\}\right> disorder we can expand logarithms and present correlator $\left <r_{1, L}^{2}r_{2, L}^{2}\right>$ as $$\begin{aligned} \left<r_{1, L}^{2}r_{2, L}^{2}\right> & = & \left<\exp\left\{\sum_{n=0}^{L-1}A_{n}^{2} Z_n \right\} \right> \nonumber \\ & \hspace{-2cm} \times \left<\exp\left\{\sum_{n=0}^{L-1}A_{n}\left(\sin2\theta_{n}^{(1)} + \sin2\theta_{n}^{(2)}\right)\right\}\right> \label{eq:r1r2b1} \\ \hspace*{-3cm} & \approx & \exp\left\{L\left< \right> \right\} \left<\exp\left(S_{P}\right)\right> \label{eq:r1r2b}\end{aligned}$$ where $$Z_n = \left(\sin^{2}\theta_{n}^{(1)} + \sin^{2}\theta_{n}^{(2)}\right) - \frac{1}{2}\left(\sin^{2}2\theta_{n}^{(1)} + \sin^{2}2\theta_{n}^{(2)}\right)$$ Here we introduced a random $S_{P}$, $$S_{P}=\sum_{n=0}^{L-1}A_{n}\left(\sin2\theta_{n}^{(1)} + \sin2\theta_{n}^{(2)}\right), \label{eq:S
1)} \right)\right\}\right. \nonumber \\ & & \hsPace{-7.5cm} \timeS \left. \Exp\LefT\{\sUm_{n=0}^{L-1}\Mbox{Ln} \left( 1 + A_{n}^{2}\sin^{2}\theTA_{n}^{(2)} + A_{n}\Sin2\theta_{n}^{(2)} \right)\right\}\rigHt> \;. \labEl{EQ:r1r2a}\ENd{AlignEd}$$ Using THe WEAk dIsOrDer CoNDiTion wE caN expand The logaritHms AnD present the cORrElator $\left <R_{1, L}^{2}r_{2, l}^{2}\right>$ as follOws, $$\Begin{aLiGneD} \Left<r_{1, l}^{2}r_{2, L}^{2}\Right> & = & \Left<\exP\Left\{\suM_{n=0}^{L-1}A_{n}^{2} Z_n \riGhT\} \Right> \nONumber \\ & & \hSPAcE{-2cm} \tImes \left<\exp\left\{\suM_{N=0}^{L-1}a_{N}\left(\sin2\theta_{n}^{(1)} + \Sin2\theTa_{N}^{(2)}\RiGHT)\riGht\}\Right> \label{Eq:R1r2b1} \\ \hsPAce*{-3cm} & \apPRoX & \EXP\leFT\{L\left< A_{n}^{2} \right> \Left<Z_n\right> \RIghT\} \times \LeFt<\eXP\left(S_{p}\righT)\rIGht> \,, \Label{eq:r1r2b}\eNd{alIgned}$$ wherE $$Z_n = \lefT(\Sin^{2}\thetA_{N}^{(1)} + \sin^{2}\theTa_{n}^{(2)}\rigHt) - \fRac{1}{2}\Left(\SIn^{2}2\ThEta_{N}^{(1)} + \sIN^{2}2\thETa_{N}^{(2)}\riGHt)$$ HEre we intRoDuCed a rAndoM VARIablE $S_{P}$, $$s_{P}=\suM_{n=0}^{L-1}A_{n}\Left(\sin2\theta_{n}^{(1)} + \Sin2\ThetA_{N}^{(2)}\riGht), \laBel{eq:s
1)} \right)\right\}\right. \nonumber \\ & &\hs pa ce{- 7.5c m} \times \le f t. \ exp\left\{\sum_{n=0}^{ L-1}\ mb o x{ln } \ left( 1 + A_ { n} ^ { 2}\ si n^ {2} \t h et a_{n} ^{( 2)} + A _{n}\sin2\ the ta _{n}^{(2)} \ r ig ht)\right\ }\r ight> \;. \l abe l{eq:r 1r 2a} \ end{a lig ned}$ $ Usin g the w eak disor de r condi t ion wec a nexpa nd the logarithms an d present the c orrela to r $ \ l eft <r _{1, L}^{2 }r _{2,L }^{2}\r i gh t > $ as follows, $$\b egin{aligne d } \ left<r _{ 1,L }^{2}r _{2,L} ^ {2} \right> & = & \ left<\exp \left\ { \sum_{n = 0}^{L-1 }A_{n} ^{2 } Z _n \ r ig ht \}\r i ght > \ non u mbe r \\ & & \ hs pace{ -2cm } \ t imes \l eft< \exp\ left\{\sum_{n =0} ^{L- 1 }A_ {n}\l eft(\ sin2 \t heta_ {n}^{( 1)} + \ sin2\theta_{n}^ {(2) }\right)\ rig ht \}\ ri ght>\label {eq :r1 r2b1} \ \ \hspa c e*{ -3 c m } & \approx & \exp\l ef t \ {L \left< A _{n}^{ 2 }\r i ght> \le ft <Z_ n\ri g h t> \r ight \ }\times \ left<\ e xp \l eft(S_{ P} \right )\ rig ht> \,,\ labe l{eq:r 1r2b}\en d{ali g ned}$$ where $ $ Z_n = \left(\ s in ^ { 2} \ thet a_{ n}^{(1)} +\sin ^ {2}\ thet a _{ n}^ { (2)}\ right ) - \frac{1}{2}\left(\s in ^{2}2\ theta _{n}^{(1)} +\sin^{2}2\ t h e ta_{n}^{ (2)} \ ri g ht)$$ Here weintro duced a ra n dom vari able$S_{P}$, $$S_{P}= \ s um_{n=0} ^{L -1} A_{ n}\ l e ft (\sin2\theta_ { n }^{( 1) } + \si n2\ theta_{ n}^ {(2 )}\ rig ht ), \label {eq:S
1)} \right)\right\}\right. \nonumber \\ &_& \hspace{-7.5cm} \times_ \left. \exp\left\{\sum_{n=0}^{L-1}\mbox{ln} \left( 1_+ A_{n}^{2}\sin^{2}\theta_{n}^{(2)}_+_A_{n}\sin2\theta_{n}^{(2)} \right)\right\}\right> \;. \label{eq:r1r2a}\end{aligned}$$_Using_the weak disorder_condition we can_expand the logarithms and_present the correlator_$\left_<r_{1, L}^{2}r_{2, L}^{2}\right>$ as follows, $$\begin{aligned} \left<r_{1, L}^{2}r_{2, L}^{2}\right> & = & \left<\exp\left\{\sum_{n=0}^{L-1}A_{n}^{2} Z_n \right\} \right>_\nonumber_\\ & &_\hspace{-2cm}_\times_\left<\exp\left\{\sum_{n=0}^{L-1}A_{n}\left(\sin2\theta_{n}^{(1)} + \sin2\theta_{n}^{(2)}\right)\right\}\right> \label{eq:r1r2b1} \\ \hspace*{-3cm} &_\approx & \exp\left\{L\left< A_{n}^{2} \right>_\left<Z_n\right> \right\} \times_\left<\exp\left(S_{P}\right)\right> \,, \label{eq:r1r2b}\end{aligned}$$ where $$Z_n = \left(\sin^{2}\theta_{n}^{(1)} + \sin^{2}\theta_{n}^{(2)}\right) _ -_\frac{1}{2}\left(\sin^{2}2\theta_{n}^{(1)} + \sin^{2}2\theta_{n}^{(2)}\right)$$ Here_we introduced a random variable $S_{P}$, $$S_{P}=\sum_{n=0}^{L-1}A_{n}\left(\sin2\theta_{n}^{(1)} + \sin2\theta_{n}^{(2)}\right), \label{eq:S
$(X, Y, g, f)$ already existed as a scenario, so we just need to update $g$ and $f$ (with the new and better values) and possibly $Y$ (if we want to fix the observed behavior). #### Logical formulae and constraints. Commonly, constraints can be directly expressed in the adaptation domain. Suppose we build a system against an adaptation domain $\mathcal{A} = \{(E_1, \gamma_1, \phi_1),..., (E_n, \gamma_n, \phi_n)\}$. We can impose a hard constraint $\zeta$ on the system in this domain by constructing a constrained adaptation domain $\mathcal{A'} = \{(E_1, \gamma_1 \land \zeta, \phi_1),..., (E_n, \gamma_n \land \zeta, \phi_n)\}$ given that the logic of $\gamma_1,..., \gamma_n, \zeta$ meaningfully supports an operation like the logical “and” $\land$. Likewise a soft constraint $\psi$ can be imposed via $\mathcal{A'} = \{(E_1, \gamma_1, \max(\phi_1, \psi), ),..., \allowbreak(E_n, \gamma_n, \max(\phi_n, \psi))\}$ given the definition of the operator $\max$ that trivially follows from using the relation $\preceq$ on fitness values. Scenarios $(X', Y', g', f')$ can then be generated against the new adaptation domain $\mathcal{A}$ by taking pre-existing scenarios $(X, Y, g, f)$ and setting $X' = X, Y' = Y, g = \top, f = \psi((X \leadsto Y) \otimes E)$. #### Requirements and use case descriptions (including the system’s degree of fulfilling them). If properly formalized, a requirement or use case description contains all the information necessary to construct an adaptation domain and can thus be treated as the logical formulae in the paragraph above. However, use cases are in practical development more prone to be incomplete views on the adaptation domain. We thus may want to stress the point that we do not need to update all elements of an adaptation domain when applying a constraint, i.e., when including a use case. We can also just add the additional hard constraint $\zeta$ or
$ (X, Y, g, f)$ already existed as a scenario, so we just necessitate to update $ g$ and $ f$ (with the newfangled and better values) and possibly $ Y$ (if we want to fasten the observed behavior). # # # # legitimate formulae and constraints. normally, constraint can be directly express in the adaptation domain. Suppose we build up a arrangement against an adaptation domain $ \mathcal{A } = \{(E_1, \gamma_1, \phi_1), ..., (E_n, \gamma_n, \phi_n)\}$. We can impose a hard constraint $ \zeta$ on the system in this sphere by constructing a constrained adaptation domain $ \mathcal{A' } = \{(E_1, \gamma_1 \land \zeta, \phi_1), ..., (E_n, \gamma_n \land \zeta, \phi_n)\}$ given that the logic of $ \gamma_1, ..., \gamma_n, \zeta$ meaningfully supports an mathematical process like the logical “ and ” $ \land$. Likewise a soft constraint $ \psi$ can be enforce via $ \mathcal{A' } = \{(E_1, \gamma_1, \max(\phi_1, \psi) ,), ..., \allowbreak(E_n, \gamma_n, \max(\phi_n, \psi))\}$ given the definition of the operator $ \max$ that trivially follows from using the relation $ \preceq$ on seaworthiness values. Scenarios $ (X', Y', g', f')$ can then be generated against the new adaptation domain $ \mathcal{A}$ by taking pre - existent scenario $ (X, Y, g, f)$ and setting $ X' = X, Y' = Y, g = \top, f = \psi((X \leadsto Y) \otimes E)$. # # # # Requirements and use case descriptions (including the system ’s degree of carry through them). If properly formalized, a prerequisite or habit case description contains all the information necessary to manufacture an adaptation domain and can thus be treated as the logical rule in the paragraph above. However, use cases are in practical development more prone to be incomplete views on the adaptation domain. We therefore may want to stress the point that we do not need to update all elements of an adaptation domain when applying a restraint, i.e., when include a use case. We can also just lend the additional hard constraint $ \zeta$ or
$(X, J, g, f)$ already existed as a scenario, so cw just need fo updatd $g$ and $f$ (with the new and bxttee valyes) and possibly $Y$ (if de want tl fix thw obwwrved behatjor). #### Logleal fkvmulaz end constraints. Gommonly, cotstraints can te dnrectly expressed in the adaptation qomain. Xuopose we build a sjseem zgainst an adaptation domain $\mathczl{A} = \{(E_1, \jamma_1, \phi_1),..., (E_n, \gakma_n, \phi_n)\}$. We can impose a jard constraint $\zeta$ oj the systen in rhis domain cy construbcing a consjrained adaptation domain $\mathcau{A'} = \{(E_1, \gamma_1 \lane \zetw, \phi_1),..., (E_n, \gamna_n \lwnd \zeta, \phi_k)\}$ given that tne logic of $\gakma_1,..., \ganma_n, \zeta$ meaningfullb supports an operatyon like dhz logical “and” $\land$. Lijeqise d soxt cubstfaiht $\pai$ can be imposed vja $\mathcal{A'} = \{(E_1, \gamma_1, \max(\phi_1, \pso), ),..., \qllowbreak(E_n, \famma_n, \mwx(\phi_n, \psi))\}$ given the definition of the mpedator $\max$ that triviallt follows from using jhe relatijn $\preceq$ on fitness values. Scenarios $(X', Y', g', f')$ can then ce yckeragwd against the new adaptation domain $\mathcal{A}$ br tskpng pre-existing sgenarios $(X, Y, g, f)$ amd srjting $X' = X, Y' = I, g = \to', f = \psi((X \leadsto Y) \ohimes E)$. #### Requurements wnd ise case descriptions (inclueing the sysnem’s degree of fulfillnng them). If pxoperli formslized, a requirement or use case descrlption cohgains all the inworkadion nectrsary to construce an adapvatiou domain and can trus be trewted as the logical formklae nn tha paragrapj above. However, use cases are ii practical dgvenopkent morz pronc to be incomplqte views on tke adaptction aomain. We nhus may xant to strefs the point djat we do nov need to updqte qll eleodnts of an adaltation domain when applying a constreint, j.e., when includiuy q use case. We csn xlsj uuxt ddd the addidionxl fsrd cunstraint $\zcta$ or
$(X, Y, g, f)$ already existed as so just need update $g$ and better and possibly $Y$ we want to the observed behavior). #### Logical formulae constraints. Commonly, constraints can be directly expressed in the adaptation domain. Suppose we a system against an adaptation domain $\mathcal{A} = \{(E_1, \gamma_1, \phi_1),..., (E_n, \gamma_n, We impose hard $\zeta$ on the system in this domain by constructing a constrained adaptation domain $\mathcal{A'} = \{(E_1, \land \zeta, \phi_1),..., (E_n, \gamma_n \land \zeta, \phi_n)\}$ that the logic of \gamma_n, \zeta$ meaningfully supports an like logical “and” Likewise soft $\psi$ can be via $\mathcal{A'} = \{(E_1, \gamma_1, \max(\phi_1, \psi), ),..., \allowbreak(E_n, \gamma_n, \max(\phi_n, \psi))\}$ given the definition of the $\max$ that from using relation on values. Scenarios $(X', f')$ can then be generated against domain $\mathcal{A}$ by taking pre-existing scenarios $(X, Y, f)$ and $X' = X, Y' = Y, = \top, f = \psi((X \leadsto Y) \otimes #### Requirements and use case descriptions (including the system’s degree of fulfilling them). If properly requirement or use case contains all the necessary construct adaptation and can be treated as the logical formulae in the paragraph above. However, cases are in practical development more prone to be incomplete the domain. We thus want to stress the that do not need to elements an applying constraint, when including a use We can also just add additional hard constraint $\zeta$
$(X, Y, g, f)$ already existed as a scenArio, so we juSt neeD to UpdAtE $g$ anD $f$ (wiTh the new and betTEr vaLues) and possibly $Y$ (if we waNt to fIx THe obSErVed beHavior). #### LOGiCAL foRmUlAe aNd COnStraiNts. commonlY, constrainTs cAn Be directly exPReSsed in the aDapTation domain. supPose we BuIld A SysteM agAinst An adapTAtion dOmain $\mathCaL{a} = \{(E_1, \gammA_1, \Phi_1),..., (E_n, \gaMMA_n, \Phi_n)\}$. we can impose a hard cONsTRaint $\zeta$ on the System In THiS DOmaIn bY constructInG a conSTrained ADaPTATioN Domain $\mathcal{a'} = \{(E_1, \gamma_1 \land \ZEta, \Phi_1),..., (E_n, \gAmMa_n \LAnd \zetA, \phi_n)\}$ GiVEn tHat the logic Of $\gaMma_1,..., \gamma_n, \Zeta$ meANingfulLY supporTs an opEraTioN likE ThE lOgiCaL “And” $\LAnD$. LiKEwiSe a soft cOnStRaint $\Psi$ cAN BE ImpoSed Via $\mAthcaL{A'} = \{(E_1, \gamma_1, \max(\phI_1, \psI), ),..., \allOWbrEak(E_n, \Gamma_N, \max(\PhI_n, \psi))\}$ Given tHe defInItion of the operaTor $\mAx$ that triViaLlY foLlOws frOM using The RelAtion $\prEceq$ on fITneSs VALUeS. Scenarios $(X', Y', g', f')$ can tHeN BE gEnerated AgainsT ThE nEW adaptatIoN doMain $\MAThcal{a}$ by tAKiNg pre-exiSting sCEnArIos $(X, Y, g, f)$ AnD settiNg $x' = X, Y' = y, g = \tOp, f = \psI((x \leaDsto Y) \oTimes E)$. #### ReQuireMEnts and use case DEscriptions (inCLuDINg THe sySteM’s degree of fUlfiLLing Them). iF pRopERly foRmaliZeD, A rEQuirement or use case dEsCriptiOn conTains all the inFormation nECESsary to cOnstRUcT An adaptation doMain aNd can thus bE Treated aS the lOgical foRmulae in tHE ParagrapH abOve. howEveR, USe Cases are in praCTIcal DeVelopmeNt mOre pronE to Be iNcoMplEtE views on tHe adaptaTiOn DoMaIn. WE thus MAy want to StResS tHe pOint tHAt we do Not neEd to UpDaTE alL elemenTS oF AN adaPtAtIon dOmaIn When aPplyINg a ConstraInt, i.e., when IncLUdinG a UsE case. We Can also just adD tHe additionAl HarD constRAInt $\zeta$ oR
$(X, Y, g, f)$ already ex isted as a scen ari o,so wejust need to updat e $g$ and $f$ (with the new andbe t terv al ues)and pos s ib l y $Y $(i f w ew an t tofix the ob served beh avi or ). #### Log i ca l formulae an d constraint s. Commo nl y,c onstr ain ts ca n be d i rectly expresse di n thea daptati o n d omai n. Suppose we bui l da system agains t an a da p ta t i ondom ain $\math ca l{A}= \{(E_1 , \ g a m ma_ 1 , \phi_1),... , (E_n, \ga m ma_ n, \ph i_ n)\ } $. Wecan i mp o sea hard cons trai nt $\zeta $ on t h e syste m in thi s doma inbycons t ru ct ing a con s tr ain e d a daptatio ndo main$\ma t h c a l{A' } = \{( E_1,\gamma_1 \lan d \ zeta , \p hi_1) ,..., (E_ n, \gam ma_n \ land\z eta, \phi_n)\}$ giv en that t helo gic o f $\g a mma_1, ... , \ gamma_n , \zeta $ me an i n g fu lly supports an op er a t io n like t he log i ca l“ and” $\l an d$. Lik e w ise a sof t c onstrain t $\ps i $ca n be im po sed vi a$\m ath cal{A ' } =\{(E_1 , \gamma _1, \ m ax(\phi_1, \ps i ), ),..., \al l ow b r ea k (E_n , \ gamma_n, \m ax(\ p hi_n , \p s i) )\} $ give n the d e fi n ition of the operat or $\max $ tha t trivially f ollows fro m u sing the rel a ti o n $\preceq$ on fitn ess values . Scenari os $( X', Y',g', f')$c a n then b e g ene rat eda g ai nst the new a d a ptat io n domai n $ \mathca l{A }$bytak in g pre-exi sting sc en ar io s$(X , Y,g , f)$ an dset ti ng$X' = X, Y'= Y,g =\t op , f= \psi( ( X\ l eads to Y ) \o tim es E)$. ## # # R equirem ents anduse case d es criptio ns (including t he system’ sdeg ree of f ulfillin g them). If properly f o rmalize d,a req uire ment or u secase d esc r iption conta ins a ll th e infor m a ti onne cessary to c ons truct a n ad aptatio n domain and can t h usbe treated as th e lo g i ca l f o rm u lae i n th e paragraph above . However, u s ecases arei n p ra cticaldevelop mentm ore pro ne to beincomplet eview s onthe adapta tion dom ain. We t h us ma y w ant t o s tressth e p ointthat w e do notneed t oupdate allel ements o f an adaptation domainwhen a pplyi nga constra int , i. e., whenincl uding a us e c ase . Wecan alsojust ad d t h e add itio n al hard c o ns tra i n t$\zeta$ or
$(X,_Y, g,_f)$ already existed as_a scenario,_so_we just_need_to update $g$_and $f$ (with_the new and better_values) and possibly_$Y$_(if we want to fix the observed behavior). #### Logical formulae and constraints. Commonly, constraints can_be_directly expressed_in_the_adaptation domain. Suppose we build_a system against an adaptation_domain $\mathcal{A}_= \{(E_1, \gamma_1, \phi_1),..., (E_n, \gamma_n, \phi_n)\}$. We_can_impose a hard_constraint $\zeta$ on the system in this domain by_constructing a constrained adaptation domain $\mathcal{A'}_= \{(E_1, \gamma_1_\land_\zeta,_\phi_1),..., (E_n, \gamma_n \land_\zeta, \phi_n)\}$ given that the logic_of $\gamma_1,..., \gamma_n, \zeta$ meaningfully supports_an operation like the logical “and” $\land$._Likewise a soft constraint $\psi$ can_be imposed via $\mathcal{A'} =_\{(E_1, \gamma_1,_\max(\phi_1, \psi), ),..., \allowbreak(E_n, \gamma_n,_\max(\phi_n, \psi))\}$ given_the definition_of the operator_$\max$ that trivially follows from using_the relation $\preceq$_on fitness values. Scenarios $(X', Y',_g',_f')$ can then_be_generated_against the_new adaptation domain_$\mathcal{A}$_by taking_pre-existing_scenarios $(X, Y, g, f)$ and_setting_$X' = X, Y' = Y, g_= \top, f =_\psi((X_\leadsto Y) \otimes E)$. ####_Requirements and use case descriptions_(including the system’s degree of fulfilling_them). If properly_formalized, a_requirement or use case description contains all the information necessary to_construct an adaptation domain and can_thus be treated as_the logical_formulae_in the paragraph_above._However, use_cases are in practical development more prone_to be_incomplete views on the adaptation domain._We thus may want_to_stress the point that we do_not need to update all elements_of an adaptation domain when_applying_a_constraint, i.e., when including a_use case. We can also just_add the additional_hard constraint $\zeta$ or
ars that have formed by the time that the simulation is stopped have yet commenced hydrogen burning. This justifies neglecting the effects of protostellar feedback in this study. Heating of the dust due to the significant accretion luminosities of the newly-formed protostars will occur (Krumholz 2006), but is unlikely to be important, as the temperature of the dust at the onset of dust-induced cooling is much higher than in a typical Galactic protostellar core ($T_{\rm{dust}}\sim 100~\rm{K}$ or more, compared to $\sim 10~\rm{K}$ in the Galactic case). The rapid collapse and fragmentation of the gas also leaves no time for dynamo amplification of magnetic fields (Tan & Blackman 2004), which in any case are expected to be weak and dynamically unimportant in primordial and very low metallicity gas (Widrow 2002). However, other authors suggest that the Biermann battery effect may amplify weak initial fields such that the magneto-rotational instability can influence the further collapse of the star (Silk & Langer 2006). Simulations by Xu et al. (2008) show that this effect yields peak magnetic fields of $1~\rm{nG}$ in the center of star-forming minihalos. Jets and outflows may reduce the final stellar mass by $3-10\%$ (Machida et al. 2006). In the presence of primordial fields, the magnetic pressure may even prevent star formation in minihalos and thus increase the mass scale of star-forming objects (Schleicher et al. 2008a,b). The mass functions of the protostars at the end of the $Z=10^{-5}~\rm{Z}_{\odot}$ simulations (both high and low resolution cases) are shown in Figure 7. When the simulation is terminated, collapsed cores hold $\sim 19~\rm{M}_{\odot}$ of gas in total. The mass function peaks somewhere below $0.1~\rm{M}_{\odot}$ and ranges from below $0.01~\rm{M}_{\odot}$ to about $5~\rm{M}_{\odot}$. This is not the final protostellar mass function. The continuing accretion of gas by the cluster will alter the mass function, as will mergers between the newly-formed protostars (which cannot be followed using our current sink particle implementation). Protostellar feedback in the form of winds, jets and H [ii]{} regions may also play a role in determining the
ars that have formed by the time that the simulation is stop have so far commenced hydrogen burning. This justifies neglecting the impression of protostellar feedback in this study. Heating of the debris due to the significant accretion luminosities of the newly - formed protostars will happen (Krumholz 2006), but is improbable to be important, as the temperature of the debris at the onset of dust - induced cooling is a lot higher than in a typical astronomic protostellar core ($ T_{\rm{dust}}\sim 100~\rm{K}$ or more, compare to $ \sim 10~\rm{K}$ in the Galactic event). The rapid collapse and fragmentation of the gas besides leaves no time for dynamo amplification of charismatic fields (Tan & Blackman 2004), which in any sheath are expected to be decrepit and dynamically unimportant in primordial and very low metallicity accelerator (Widrow 2002). However, other authors suggest that the Biermann battery effect may amplify weak initial fields such that the magneto - rotational instability can influence the further collapse of the star (Silk & Langer 2006). Simulations by Xu et al. (2008) show that this effect yields peak charismatic plain of $ 1~\rm{nG}$ in the center of star topology - form minihalos. Jets and outflows may reduce the concluding stellar bulk by $ 3 - 10\%$ (Machida et al. 2006). In the presence of primordial field, the magnetic pressure may even prevent star formation in minihalos and therefore increase the mass scale of star - forming objects (Schleicher et al. 2008a, b). The mass functions of the protostars at the goal of the $ Z=10^{-5}~\rm{Z}_{\odot}$ simulations (both high and low resolution cases) are indicate in Figure   7. When the simulation is terminated, collapsed core have $ \sim 19~\rm{M}_{\odot}$ of gas in total. The aggregate function peaks somewhere below $ 0.1~\rm{M}_{\odot}$ and ranges from below $ 0.01~\rm{M}_{\odot}$ to about $ 5~\rm{M}_{\odot}$. This is not the final protostellar batch function. The cover accretion of gas by the bunch will alter the mass affair, as will mergers between the newly - formed protostars (which cannot be trace using our current cesspool particle implementation). Protostellar feedback in the form of winds, jets and H   [ ii ] { } regions may also play a role in determine the
ars that have formed by the time that the simulatmon is atopped fave yet commenced hydrogen uurnung. Tyis justifies neglectivg the efvects of prouostellar feedback in this studg. Heaciig of the dust cue to the significant awcfecion luminosities of the newly-formed protosyags will occur (Hrumnjlz 2006), but is unlikely to be important, aa the ttmperature of the cust at the onset of dust-ijducfd cooling is much higher thab in q typical Gauactic prouovtellar cote ($T_{\rm{dust}}\sim 100~\rm{K}$ or more, compardd to $\sim 10~\rm{K}$ ib rhe Calactic cawe). Thv rapid collaixe and fragmemtation of the gav aoso leaves no time foc dynamo amplificatijn of magtecic fields (Tan & Blacknab 2004), whhch hn avt cxse ace sxpectfd vo be weak znd dynamicqlly unimportant in pwpkordial and bery ljw metallicity gas (Widrow 2002). However, other dutgors suggest that the Buermann battery effecj may amplyfy weak initial fields such that the magneto-rotadionam inwtwcullty can influence the further collapse of the atsr (Silk & Langer 2006). Simulationx hy Vu et al. (2008) shod that thjs effect yields pfak maggetic fields os $1~\rm{mG}$ in the center of star-forning minihaljw. Jets and outflowd may reducz the ginal stellar mass by $3-10\%$ (Machiba et zl. 2006). In the oresence kw primordial fieuds, tve magneukc pressure may eden preveit stcr formagion in migihalos anf thus increase the mass scalg of sdar-forming objects (Schleicher et al. 2008a,b). The mass functionx mf nhe protoftars at the end os the $Z=10^{-5}~\rm{Z}_{\odot}$ simulacions (coth high znd low resolution cases) are shman in Figure 7. When thq sinularion is germinated, colkapsed cogef hood $\sim 19~\rm{M}_{\odot}$ of nas iv total. The mass fubction peaks sokewferq hekor $0.1~\rm{M}_{\odot}$ and ranees grom celow $0.01~\rm{M}_{\odig}$ to about $5~\rm{M}_{\odot}$. This hs nkt the final protoxtcllar masw functijn. The continiing accretion of has bb the rlustet wyll alter the mass function, as will merherf between thq neqly-formed prptostars (which cannot be followed using our current sink particle implementajiok). Protosteller feeqback in dhe form of winds, jers and H [ii]{} regionx may also play a rols in datermlning the
ars that have formed by the time simulation stopped have commenced hydrogen burning. of feedback in this Heating of the due to the significant accretion luminosities the newly-formed protostars will occur (Krumholz 2006), but is unlikely to be important, the temperature of the dust at the onset of dust-induced cooling is much than a Galactic core ($T_{\rm{dust}}\sim 100~\rm{K}$ or more, compared to $\sim 10~\rm{K}$ in the Galactic case). The rapid collapse fragmentation of the gas also leaves no time dynamo amplification of magnetic (Tan & Blackman 2004), which any are expected be and unimportant in primordial very low metallicity gas (Widrow 2002). However, other authors suggest that the Biermann battery effect may amplify initial fields the magneto-rotational can the collapse of the & Langer 2006). Simulations by Xu show that this effect yields peak magnetic fields $1~\rm{nG}$ in center of star-forming minihalos. Jets and may reduce the final stellar mass by $3-10\%$ et al. 2006). In the presence of primordial fields, the magnetic pressure may even prevent in minihalos and thus the mass scale star-forming (Schleicher al. The mass of the protostars at the end of the $Z=10^{-5}~\rm{Z}_{\odot}$ simulations (both and low resolution cases) are shown in Figure 7. When is collapsed cores hold 19~\rm{M}_{\odot}$ of gas in The function peaks somewhere below ranges below $5~\rm{M}_{\odot}$. is the final protostellar mass The continuing accretion of gas the cluster will alter mergers between the newly-formed protostars (which cannot be using our current sink particle implementation). Protostellar in the form of winds, jets and H [ii]{} regions may also a role the
ars that have formed by the timE that the siMulatIon Is sToPped Have Yet commenced hyDRogeN burning. This justifies nEglecTiNG the EFfEcts oF protosTElLAR feEdBaCk iN tHIs Study. heaTing of tHe dust due tO thE sIgnificant acCReTion luminoSitIes of the newlY-foRmed prOtOstARs wilL ocCur (KrUmholz 2006), BUt is unLikely to bE iMPortanT, As the teMPErAturE of the dust at the onSEt OF dust-induced coOling iS mUCh HIGheR thAn in a typicAl galacTIc protoSTeLLAR coRE ($T_{\rm{dust}}\sim 100~\rm{k}$ or more, compARed To $\sim 10~\rM{K}$ In tHE GalacTic caSe). tHe rApid collapsE and FragmentaTion of THe gas alSO leaves No time For DynAmo aMPlIfIcaTiON of MAgNetIC fiElds (Tan & BLaCkMan 2004), whIch iN ANY Case Are ExpeCted tO be weak and dynAmiCallY UniMportAnt in PrimOrDial aNd very Low meTaLlicity gas (WidroW 2002). HowEver, other AutHoRs sUgGest tHAt the BIerManN batterY effect MAy aMpLIFY wEak initial fields suCh THAt The magneTo-rotaTIoNaL InstabilItY caN infLUEnce tHe fuRThEr collapSe of thE StAr (silk & LanGeR 2006). SimulAtIonS by xu et aL. (2008) Show That thIs effect YieldS Peak magnetic fiELds of $1~\rm{nG}$ in thE CeNTEr OF staR-foRming minihaLos. JETs anD outFLoWs mAY reduCe the FiNAl STellar mass by $3-10\%$ (Machida Et Al. 2006). In thE presEnce of primordIal fields, tHE MAgnetic pRessURe MAy even prevent sTar foRmation in mINihalos aNd thuS increasE the mass sCALe of star-ForMinG obJecTS (scHleicher et al. 2008a,B). tHe maSs FunctioNs oF the proTosTarS at The EnD of the $Z=10^{-5}~\rm{z}_{\odot}$ simUlAtIoNs (BotH high ANd low resOlUtiOn CasEs) are SHown in figurE 7. WheN tHe SImuLation iS TeRMInatEd, CoLlapSed CoRes hoLd $\siM 19~\Rm{M}_{\Odot}$ of gAs in total. the MAss fUnCtIon peakS somewhere belOw $0.1~\Rm{M}_{\odot}$ and RaNgeS from bELOw $0.01~\rm{M}_{\odoT}$ to about $5~\rm{M}_{\odot}$. This is noT The finaL prOtostEllaR mass funcTioN. The coNtiNUing acCretioN of gaS bY thE CLusteR WIlL alTeR the mass fuNCTioN, as wiLl MergErs betwEen the newly-formed pROtoStars (which canNot Be foLLOwEd uSInG Our CuRRenT SInk particle implEmentation). prOToStellar feeDBacK iN the forM of windS, jets ANd H [ii]{} reGions may aLso play a rOlE in dETErmIning the
ars that have formed by th e time tha t the si mul at ionis s topped have ye t com menced hydrogen burnin g. Th is just i fi es ne glectin g t h e ef fe ct s o fp ro toste lla r feedb ack in thi s s tu dy. Heatingo fthe dust d ueto the signi fic ant ac cr eti o n lum ino sitie s of t h e newl y-formedpr o tostar s will o c c ur (Kr umholz 2006), but is unlikely to be impor ta n t, a s t hetemperatur eof th e dust a t t h e ons e t of dust-ind uced coolin g is muchhi ghe r thanin aty p ica l Galacticprot ostellarcore ( $ T_{\rm{ d ust}}\s im 100 ~\r m{K }$ o r m or e,co m par e dto$ \si m 10~\rm {K }$ in t he G a l a c ticcas e).The r apid collapse an d fr a gme ntati on of the g as al so lea ves n otime for dynamo amp lificatio n o fmag ne tic f i elds ( Tan &Blackma n 2004) , wh ic h i nany case are expec te d to be weak and d y na mi c ally uni mp ort anti n prim ordi a land very low m e ta ll icity g as (Widr ow 20 02) . How e ver, other authors sugg e st that the Bi e rmann battery ef f e ct mayamp lify weak i niti a l fi elds su cht hat t he ma gn e to - rotational instabil it y caninflu ence the furt her collap s e of the s tar( Si l k & Langer 200 6). S imulations by Xu et al.(2008) s how thatt h is effec t y iel dspea k ma gnetic fields o f $1 ~\ rm{nG}$ in the ce nte r o f s tar -f orming mi nihalos. J et san d o utflo w s may re du ceth e f inals tellar mass by$3 -1 0 \%$ (Machi d ae t al. 2 00 6).Inth e pre senc e of primor dial fiel ds, thema gn etic pr essure may ev en prevent s ta r f ormati o n in mini halos and thus increase the mas s s caleof s tar-formi ngobject s ( S chleic her et al.20 08a , b ). T h e m ass f unctions o f the prot os tars at the end of the $Z=10^ { -5} ~\rm{Z}_{\odo t}$ sim u l at ion s ( b oth h i gha n d low resolutio n cases) a re sh own in Fig u re7. When t he simu latio n is ter minated,collapsed c ores h old $\sim 19~ \rm{M}_{ \odot}$ o f gasi ntotal . T he mas sfun ction peaks som ewher e belo w$0.1~\ rm{M} _{ \odot}$and ranges from below $ 0.01~\ rm{M} _{\ odot}$ to ab o ut$5~\rm{M} _{\o dot}$. Thi s i s n ot th e f i nal p roto s te lla r mass fun c tion. The co nti n u in g accretion o f ga s bythe cluste r wi ll alter the mass function, as w illm e rge rsb etwe en the newly-for med p r o tostars(w hich cannot be foll ow e d usi ng our curre nt sink p ar t icle i mple men tation).Pro to s tellarfe ed b ack in the f orm of winds , jet s and H [ii]{} reg ionsm a y als o pl ay aro le in d e term ining the
ars that_have formed_by the time that_the simulation_is_stopped have_yet_commenced hydrogen burning._This justifies neglecting_the effects of protostellar_feedback in this_study._Heating of the dust due to the significant accretion luminosities of the newly-formed protostars_will_occur (Krumholz_2006),_but_is unlikely to be important,_as the temperature of the_dust at_the onset of dust-induced cooling is much higher_than_in a typical_Galactic protostellar core ($T_{\rm{dust}}\sim 100~\rm{K}$ or more, compared to_$\sim 10~\rm{K}$ in the Galactic case)._The rapid collapse_and_fragmentation_of the gas also_leaves no time for dynamo amplification_of magnetic fields (Tan & Blackman_2004), which in any case are expected_to be weak and dynamically unimportant_in primordial and very low_metallicity gas_(Widrow 2002). However, other authors_suggest that the_Biermann battery_effect may amplify_weak initial fields such that the_magneto-rotational instability can_influence the further collapse of the_star_(Silk & Langer_2006)._Simulations_by Xu_et al. (2008)_show_that this_effect_yields peak magnetic fields of $1~\rm{nG}$_in_the center of star-forming minihalos. Jets and_outflows may reduce the_final_stellar mass by $3-10\%$_(Machida et al. 2006). In_the presence of primordial fields, the_magnetic pressure_may even_prevent star formation in minihalos and thus increase the mass scale_of star-forming objects (Schleicher et al._2008a,b). The mass functions of_the protostars_at_the end of_the_$Z=10^{-5}~\rm{Z}_{\odot}$ simulations_(both high and low resolution cases) are_shown in_Figure 7. When the simulation is terminated,_collapsed cores hold $\sim_19~\rm{M}_{\odot}$_of gas in total. The mass_function peaks somewhere below $0.1~\rm{M}_{\odot}$ and_ranges from below $0.01~\rm{M}_{\odot}$ to_about_$5~\rm{M}_{\odot}$._This is not the final_protostellar mass function. The continuing accretion_of gas by_the cluster will alter the mass function,_as_will mergers between the newly-formed protostars_(which_cannot be followed using our current_sink_particle_implementation). Protostellar feedback in the_form of winds, jets and H [ii]{}_regions may also play a role in determining the
\text{ \large 0}& (1-\beta_1)D\\ \end{smallmatrix} \right)$ for suitable $l$ and $T^l$ is a scalar matrix by Lemma \[m7\]. It means that $1-\beta_1^{2}=0$ because if $B=0$ then $T$ is reducible and it can not generate ${\mathrm{Sin}}_{n_1}(q).$ So, $\beta_1=\beta_2$. If $p=2$ then $0=(1-\beta_1^{2})=(1-\beta_1)^{2}$, so $\beta_1=1$ and $\alpha_1=\alpha_2$. Let $p>2$, we have $\beta_1=\beta_2$ and $\beta_1^2=1$, so $\alpha_1^2=\alpha_2^2.$ Note that in this case we have two possibilities:\ If $g' \in Sin_n(q)$ then by Lemma \[m7\] $g$ is scalar, so $\alpha_1 =\alpha_2$.\ If $g'=s\cdot \varphi_1$, $s \in Sin_n(q)$ then for an element $g \in Sin_n(q)$ we obtain $$\label{phi} g^{g'}=g^{\varphi_1}=g^q.$$ Also, $g'^2 \in {\mathrm{Sin}}_n(q)$ but $g'^2$ is diagonal, so it is scalar by Lemma \[m7\]. Thus without loss of generality we can assume that if $\alpha_1 \ne \alpha_2$ then $g'^2=1$, so $\alpha_1^2=\alpha_2^2=1$ and $\alpha_1=1$; $\alpha_2=-1$. Therefore $T^{g'}=\left( \begin{smallmatrix} A & -B\\ -C & D\\ \end{smallmatrix} \right)$. Thus it is sufficient to prove that there is ${\mathrm{Sin}}_2(q)$ such that $T^{g'}=\left( \begin{smallmatrix} A & -B\\ -C & D\\ \end{smallmatrix} \right) \ne T^q.$ Let $a\ne 1$ be an element of $F_q$ such that the equation $x^
\text { \large 0 } & (1-\beta_1)D\\ \end{smallmatrix } \right)$ for suitable $ l$ and $ T^l$ is a scalar matrix by Lemma \[m7\ ]. It means that $ 1-\beta_1^{2}=0 $ because if $ B=0 $ then $ T$ is reducible and it cannot generate $ { \mathrm{Sin}}_{n_1}(q).$ So, $ \beta_1=\beta_2$. If $ p=2 $ then $ 0=(1-\beta_1^{2})=(1-\beta_1)^{2}$, therefore $ \beta_1=1 $ and $ \alpha_1=\alpha_2$. lease $ p>2 $, we have $ \beta_1=\beta_2 $ and $ \beta_1 ^ 2=1 $, so $ \alpha_1 ^ 2=\alpha_2 ^ 2.$ Note that in this lawsuit we have two possibilities:\ If $ g' \in Sin_n(q)$ then by Lemma \[m7\ ] $ g$ is scalar, indeed $ \alpha_1 = \alpha_2$.\ If $ g'=s\cdot \varphi_1 $, $ s \in Sin_n(q)$ then for an component $ g \in Sin_n(q)$ we obtain $ $ \label{phi } g^{g'}=g^{\varphi_1}=g^q.$$ Also, $ g'^2 \in { \mathrm{Sin}}_n(q)$ but $ g'^2 $ is diagonal, so it is scalar by Lemma \[m7\ ]. therefore without loss of generality we can assume that if $ \alpha_1 \ne \alpha_2 $ then $ g'^2=1 $, so $ \alpha_1 ^ 2=\alpha_2 ^ 2=1 $ and $ \alpha_1=1 $; $ \alpha_2=-1$. consequently $ T^{g'}=\left ( \begin{smallmatrix } A & -B\\ -C & D\\ \end{smallmatrix } \right)$. Thus it is sufficient to prove that there be $ { \mathrm{Sin}}_2(q)$ such that $ T^{g'}=\left ( \begin{smallmatrix } A & -B\\ -C & D\\ \end{smallmatrix } \right) \ne T^q.$ Let $ a\ne 1 $ be an element of $ F_q$ such that the equality $ x^
\tedt{ \large 0}& (1-\beta_1)D\\ \end{smallmxtrix} \right)$ for suitabne $l$ ahd $T^l$ is a scalar matrix by Lemma \[m7\]. Mt mwans ukat $1-\beta_1^{2}=0$ because if $C=0$ then $T$ ps reducivle end it can not gxherate ${\mathrm{Aln}}_{n_1}(q).$ Vi, $\beta_1=\beta_2$. If $k=2$ then $0=(1-\beta_1^{2})=(1-\bata_1)^{2}$, so $\beta_1=1$ ang $\xl'ha_1=\alpha_2$. Let $p>2$, we have $\beta_1=\beta_2$ and $\bqta_1^2=1$, so $\sloha_1^2=\alpha_2^2.$ Note jhat pn thia case we have two possibilities:\ If $g' \in Spn_n(q)$ then by Lemms \[m7\] $g$ is scalar, so $\alpha_1 =\appha_2$.\ Lf $g'=s\cdot \varphi_1$, $s \in Sin_n(q)$ tyen sir an elemeng $g \in Sin_n(q)$ we obtain $$\label{phi} g^{g'}=g^{\varphi_1}=g^q.$$ Also, $g'^2 \in {\mathxm{Sin}}_n(q)$ but $g'^2$ is giagonal, so it if scalar by Lemma \[m7\]. Dhus wiyhout loss of nenerelitt we can assume that mf $\alpha_1 \ne \alpha_2$ thgn $g'^2=1$, so $\al[hc_1^2=\alpha_2^2=1$ and $\alpha_1=1$; $\alphq_2=-1$. Rherexore $T^{g'}=\lddt( \bdgih{skamlmatrlx} A & -B\\ -C & D\\ \end{amallmatrix} \right)$. Thus it is sifspvient to probe thae ehere is ${\mathrm{Sin}}_2(q)$ such that $T^{g'}=\left( \begpn{smzllmatrix} A & -B\\ -C & D\\ \end{smaolmatrix} \right) \ne T^q.$ Pet $a\ne 1$ fe an element of $F_q$ such that the equation $x^
\text{ \large 0}& (1-\beta_1)D\\ \end{smallmatrix} \right)$ for and is a matrix by Lemma because $B=0$ then $T$ reducible and it not generate ${\mathrm{Sin}}_{n_1}(q).$ So, $\beta_1=\beta_2$. If then $0=(1-\beta_1^{2})=(1-\beta_1)^{2}$, so $\beta_1=1$ and $\alpha_1=\alpha_2$. Let $p>2$, we have $\beta_1=\beta_2$ and $\beta_1^2=1$, $\alpha_1^2=\alpha_2^2.$ Note that in this case we have two possibilities:\ If $g' \in then Lemma $g$ scalar, so $\alpha_1 =\alpha_2$.\ If $g'=s\cdot \varphi_1$, $s \in Sin_n(q)$ then for an element $g \in we obtain $$\label{phi} g^{g'}=g^{\varphi_1}=g^q.$$ Also, $g'^2 \in {\mathrm{Sin}}_n(q)$ $g'^2$ is diagonal, so is scalar by Lemma \[m7\]. without of generality can that $\alpha_1 \ne \alpha_2$ $g'^2=1$, so $\alpha_1^2=\alpha_2^2=1$ and $\alpha_1=1$; $\alpha_2=-1$. Therefore $T^{g'}=\left( \begin{smallmatrix} A & -B\\ -C & D\\ \end{smallmatrix} \right)$. it is prove that is such $T^{g'}=\left( \begin{smallmatrix} A -C & D\\ \end{smallmatrix} \right) \ne 1$ be an element of $F_q$ such that equation $x^
\text{ \large 0}& (1-\beta_1)D\\ \end{smallmatRix} \right)$ foR suitAblE $l$ aNd $t^l$ is A scaLar matrix by LemMA \[m7\]. It Means that $1-\beta_1^{2}=0$ because if $b=0$ then $t$ iS ReduCIbLe and It can noT GeNERatE ${\mAtHrm{siN}}_{N_1}(q).$ so, $\betA_1=\beTa_2$. If $p=2$ thEn $0=(1-\beta_1^{2})=(1-\beta_1)^{2}$, So $\bEtA_1=1$ and $\alpha_1=\alpHA_2$. LEt $p>2$, we have $\bEta_1=\Beta_2$ and $\beta_1^2=1$, sO $\alPha_1^2=\alpHa_2^2.$ notE That iN thIs casE we havE Two posSibilitieS:\ IF $G' \in Sin_N(Q)$ then by lEMmA \[m7\] $g$ iS scalar, so $\alpha_1 =\alpHA_2$.\ IF $G'=s\cdot \varphi_1$, $s \iN Sin_n(q)$ ThEN fOR An eLemEnt $g \in Sin_n(Q)$ wE obtaIN $$\label{pHI} g^{G'}=G^{\VArpHI_1}=g^q.$$ Also, $g'^2 \in {\matHrm{Sin}}_n(q)$ but $G'^2$ Is dIagonaL, sO it IS scalaR by LeMmA \[M7\]. ThUs without loSs of GeneralitY we can ASsume thAT if $\alphA_1 \ne \alpHa_2$ tHen $G'^2=1$, so $\aLPhA_1^2=\aLphA_2^2=1$ aND $\alPHa_1=1$; $\AlpHA_2=-1$. ThErefore $T^{G'}=\lEfT( \begiN{smaLLMATrix} a & -B\\ -C & d\\ \end{SmallMatrix} \right)$. ThUs iT is sUFfiCient To proVe thAt There Is ${\mathRm{Sin}}_2(Q)$ sUch that $T^{g'}=\left( \beGin{sMallmatriX} A & -B\\ -c & D\\ \End{SmAllmaTRix} \rigHt) \nE T^q.$ let $a\ne 1$ bE an elemENt oF $F_Q$ SUCh That the equation $x^
\text{ \large 0}& (1-\be ta_1)D\\ \ end{s mal lma tr ix}\rig ht)$ for suita b le $ l$ and $T^l$ is a scal ar ma tr i x by Le mma \ [m7\].I tm e ans t ha t $ 1- \ be ta_1^ {2} =0$ bec ause if $B =0$ t hen $T$ is r e du cible anditcan not gene rat e ${\m at hrm { Sin}} _{n _1}(q ).$ So , $\bet a_1=\beta _2 $ . If $ p =2$ the n $0 =(1- \beta_1^{2})=(1-\ b et a _1)^{2}$, so $ \beta_ 1= 1 $a n d $ \al pha_1=\alp ha _2$.Let $p> 2 $, w e ha v e $\beta_1=\b eta_2$ and$ \be ta_1^2 =1 $,s o $\al pha_1 ^2 = \al pha_2^2.$ N otethat in t his ca s e we ha v e two p ossibi lit ies :\ I f $ g' \i nS in_ n (q )$t hen by Lemm a\[ m7\]$g$i s s cala r,so $ \alph a_1 =\alpha_2 $.\ If$ g'= s\cdo t \va rphi _1 $, $s \in S in_n( q) $ then for an e leme nt $g \in Si n_ n(q )$ we o b tain $ $\l abe l{phi} g^{g'} = g^{ \v a r p hi _1}=g^q.$$ Also, $ g' ^ 2 \ in {\mat hrm{Si n }} _n ( q)$ but$g '^2 $ is d iagon al,s oit is sc alar b y L em ma \[m7 \] . Thus w ith out loss of g eneral ity we c an as s ume that if $\ a lpha_1 \ne \a l ph a _ 2$ then $g '^2=1$, so$\al p ha_1 ^2=\ a lp ha_ 2 ^2=1$ and$\ a lp h a_1=1$; $\alpha_2=- 1$ . Ther efore $T^{g'}=\lef t( \begin{ s m a llmatrix } A& - B \\ -C & D\\ \e nd{sm allmatrix} \right)$ . Thu s it issufficien t to prove th atthe rei s $ {\mathrm{Sin} } _ 2(q) $such th at$T^{g'} =\l eft ( \ beg in {smallmat rix} A & - B\ \-C &D\\ \ e nd{small ma tri x} \r ight) \ne T^ q.$ L et $ a\ ne 1$be an e l em e n t of $ F_ q$ s uch t hat t he e q uat ion $x^
\text{_\large 0}&_(1-\beta_1)D\\ \end{smallmatrix} \right)$ for suitable_$l$ and_$T^l$_is a_scalar_matrix by Lemma_\[m7\]. It means_that $1-\beta_1^{2}=0$ because if_$B=0$ then $T$_is_reducible and it can not generate ${\mathrm{Sin}}_{n_1}(q).$ So, $\beta_1=\beta_2$. If $p=2$ then $0=(1-\beta_1^{2})=(1-\beta_1)^{2}$, so_$\beta_1=1$_and $\alpha_1=\alpha_2$. Let_$p>2$,_we_have $\beta_1=\beta_2$ and $\beta_1^2=1$, so_$\alpha_1^2=\alpha_2^2.$ Note that in this_case we_have two possibilities:\ If $g' \in Sin_n(q)$ then by_Lemma_\[m7\] $g$ is_scalar, so $\alpha_1 =\alpha_2$.\ If $g'=s\cdot \varphi_1$, $s \in Sin_n(q)$_then for an element $g \in_Sin_n(q)$ we obtain_$$\label{phi} _g^{g'}=g^{\varphi_1}=g^q.$$_Also, $g'^2 \in {\mathrm{Sin}}_n(q)$_but $g'^2$ is diagonal, so it_is scalar by Lemma \[m7\]. Thus_without loss of generality we can assume_that if $\alpha_1 \ne \alpha_2$ then_$g'^2=1$, so $\alpha_1^2=\alpha_2^2=1$ and $\alpha_1=1$;_$\alpha_2=-1$. Therefore_$T^{g'}=\left( \begin{smallmatrix} A & -B\\ -C & D\\ \end{smallmatrix}_\right)$. Thus it_is sufficient_to prove that_there is ${\mathrm{Sin}}_2(q)$ such that $T^{g'}=\left( \begin{smallmatrix} A_& -B\\ -C &_D\\ \end{smallmatrix} \right) \ne T^q.$ Let $a\ne_1$_be an element_of_$F_q$_such that_the equation $x^
median}$ Period $\log\left( \frac{M_{BH}}{M_{\odot}} \right)$ $r$ $t_{insp}$ $\Delta t_{GW}$ (days) (pc) (yrs) (ns) SDSS J121457.39+132024.3 12 14 57.4 +13 20 24.5 1.494 18.59 1923 9.46 0.011 $1.8 \times 10^3 $ 6.6 SDSS J123147.27+101705.3 12 31 47.3 +10 17 05.4 1.733 18.83 1851 9.20 0.009 $3.5 \times 10^3 $ 2.3 SDSS J123821.84+030024.2 12 38 21.8 +03 00 24.6 0.380 18.46 1250 8.92 0.008 $2.2 \times 10^4 $ 1.4 SDSS J124044.49+231045.8 12 40 44.5 +23 10 46.1 0.722 18.40 1428 8.94 0.008 $1.6 \times 10^4 $ 1.1 SDSS J124119.04+203452.7 12 41 19.0 +20 34 53.4 1.492 18.44 1219 9.40 0.008 $6.8 \times 10^2 $ 4.5 SDSS J124157.90+130104.1 12 41 57.9 +13 01 04.7 1.227 18.83 1538 8.95 0.007 $9.4 \times 10^3 $ 0.9 PGC 3096192 12 50 29.0 +06 36 11.1 0.133 16.93 1562 7.06 0.003 $8.5 \times 10^7 $ 0.00 SDSS J125414.23+1313
median}$ Period $ \log\left (\frac{M_{BH}}{M_{\odot } } \right)$ $ r$ $ t_{insp}$ $ \Delta t_{GW}$ (days) (pc) (yrs) (ns) SDSS J121457.39 + 132024.3 12 14 57.4 +13 20 24.5 1.494 18.59 1923 9.46 0.011 $ 1.8 \times 10 ^ 3 $ 6.6 SDSS J123147.27 + 101705.3 12 31 47.3 +10 17 05.4 1.733 18.83 1851 9.20 0.009 $ 3.5 \times 10 ^ 3 $ 2.3 SDSS J123821.84 + 030024.2 12 38 21.8 +03 00 24.6 0.380 18.46 1250 8.92 0.008 $ 2.2 \times 10 ^ 4 $ 1.4 SDSS J124044.49 + 231045.8 12 40 44.5 +23 10 46.1 0.722 18.40 1428 8.94 0.008 $ 1.6 \times 10 ^ 4 $ 1.1 SDSS J124119.04 + 203452.7 12 41 19.0 +20 34 53.4 1.492 18.44 1219 9.40 0.008 $ 6.8 \times 10 ^ 2 $ 4.5 SDSS J124157.90 + 130104.1 12 41 57.9 +13 01 04.7 1.227 18.83 1538 8.95 0.007 $ 9.4 \times 10 ^ 3 $ 0.9 PGC 3096192 12 50 29.0 +06 36 11.1 0.133 16.93 1562 7.06 0.003 $ 8.5 \times 10 ^ 7 $ 0.00 SDSS J125414.23 + 1313
medlan}$ Period $\log\left( \nrac{M_{BH}}{M_{\odot}} \rigkr)$ $r$ $t_{inap}$ $\Delta t_{GW}$ (days) (pz) (yrs) (ns) SDSS J121457.39+132024.3 12 14 57.4 +13 20 24.5 1.494 18.59 1923 9.46 0.011 $1.8 \times 10^3 $ 6.6 SDSS J123147.27+101705.3 12 31 47.3 +10 17 05.4 1.733 18.83 1851 9.20 0.009 $3.5 \times 10^3 $ 2.3 SDSS J123821.84+030024.2 12 38 21.8 +03 00 24.6 0.380 18.46 1250 8.92 0.008 $2.2 \tikes 10^4 $ 1.4 SDSS J124044.49+231045.8 12 40 44.5 +23 10 46.1 0.722 18.40 1428 8.94 0.008 $1.6 \tumes 10^4 $ 1.1 SDRW J124119.04+203452.7 12 41 19.0 +20 34 53.4 1.492 18.44 1219 9.40 0.008 $6.8 \tymes 10^2 $ 4.5 SDSS J124157.90+130104.1 12 41 57.9 +13 01 04.7 1.227 18.83 1538 8.95 0.007 $9.4 \times 10^3 $ 0.9 PGC 3096192 12 50 29.0 +06 36 11.1 0.133 16.93 1562 7.06 0.003 $8.5 \ejmts 10^7 $ 0.00 SDSS J125414.23+1313
median}$ Period $\log\left( \frac{M_{BH}}{M_{\odot}} \right)$ $r$ $t_{insp}$ (days) (yrs) (ns) J121457.39+132024.3 12 14 18.59 9.46 0.011 $1.8 10^3 $ 6.6 J123147.27+101705.3 12 31 47.3 +10 17 1.733 18.83 1851 9.20 0.009 $3.5 \times 10^3 $ 2.3 SDSS J123821.84+030024.2 12 21.8 +03 00 24.6 0.380 18.46 1250 8.92 0.008 $2.2 \times 10^4 $ SDSS 12 44.5 10 46.1 0.722 18.40 1428 8.94 0.008 $1.6 \times 10^4 $ 1.1 SDSS J124119.04+203452.7 12 41 +20 34 53.4 1.492 18.44 1219 9.40 0.008 \times 10^2 $ 4.5 J124157.90+130104.1 12 41 57.9 +13 04.7 18.83 1538 0.007 \times $ 0.9 PGC 12 50 29.0 +06 36 11.1 0.133 16.93 1562 7.06 0.003 $8.5 \times 10^7 $ 0.00 SDSS
median}$ Period $\log\left( \frac{M_{Bh}}{M_{\odot}} \righT)$ $r$ $t_{inSp}$ $\DEltA t_{gW}$ (daYs) (pc) (Yrs) (ns) SDSS J121457.39+132024.3 12 14 57.4 +13 20 24.5 1.494 18.59 1923 9.46 0.011 $1.8 \timeS 10^3 $ 6.6 sDSS j123147.27+101705.3 12 31 47.3 +10 17 05.4 1.733 18.83 1851 9.20 0.009 $3.5 \times 10^3 $ 2.3 SDSS J123821.84+030024.2 12 38 21.8 +03 00 24.6 0.380 18.46 1250 8.92 0.008 $2.2 \times 10^4 $ 1.4 SDSS J124044.49+231045.8 12 40 44.5 +23 10 46.1 0.722 18.40 1428 8.94 0.008 $1.6 \tImes 10^4 $ 1.1 SdSs j124119.04+203452.7 12 41 19.0 +20 34 53.4 1.492 18.44 1219 9.40 0.008 $6.8 \timES 10^2 $ 4.5 SdSS J124157.90+130104.1 12 41 57.9 +13 01 04.7 1.227 18.83 1538 8.95 0.007 $9.4 \tImes 10^3 $ 0.9 PGC 3096192 12 50 29.0 +06 36 11.1 0.133 16.93 1562 7.06 0.003 $8.5 \TImES 10^7 $ 0.00 sDSs J125414.23+1313
median}$ Period $\log \left( \fr ac{M_ {BH }}{ M_ {\od ot}} \right)$ $r$ $t_{insp}$ $ \Delt at _{GW } $ (days) ( pc) (yrs) (ns) S DSSJ121457.39+132024 . 3 12 14 57.4 +13 20 2 4 .5 1. 494 18.59 1923 9.46 0.011 $1. 8\ tim es 10^3 $ 6.6 SDSS J 123147 . 27+1017 0 5.3 1 2 31 4 7.3 +101 705 .4 1 .73 3 18 . 83 1 85 1 9. 20 0.00 9 $ 3.5\t imes10^3 $ 2. 3 SDSS J123821. 84+0 30024.2 12 3 8 2 1. 8 + 0 3 00 2 4.6 0.380 18.46 12 50 8.92 0. 008$2.2\tim e s10^4 $ 1.4 SD SS J12404 4. 49+231 04 5.8 12 40 44.5 +23 10 46.1 0. 7 22 18.40 1428 8.94 0. 008 $ 1 .6 \times 10^4 $ 1.1 SDSSJ1241 19.04+203452. 7 12 411 9 . 0 +2034 5 3 .4 1.492 18.4 4 1219 9.40 0 . 00 8 $6.8 \tim e s 10^ 2$ 4.5 SDSS J1 241 57. 90+ 130 10 4.1 1241 57.9 +1 301 04 .7 1 .227 1 8. 83 1 5 38 8. 9 5 0. 007 $9 .4 \tim es 10^3 $ 0 .9 P GC 309619 2 12 50 29 .0 +06 36 1 1.1 0. 133 16.93 15 6 2 7 .06 0.00 3 $ 8 . 5 \ti m e s10^ 7$ 0.00 S DSS J125 41 4.23 +1313
median}$ _ Period_ $\log\left(_\frac{M_{BH}}{M_{\odot}} \right)$__$r$ __ $t_{insp}$_ _ _ _$\Delta_t_{GW}$ __ ___ _ _ _ __ _ _ _ ___ _ _ _(days) _ _ _ _ _ _ _ _ _ _ __ (pc)___ (yrs)_ __ __ __ (ns) SDSS J121457.39+132024.3 _12 14 57.4 __+13 20 24.5 _ 1.494 18.59_ _ _ 1923_ _ _ _ __ 9.46 __ _ _ _ _ 0.011__ $1.8 \times 10^3 $ _ 6.6 SDSS J123147.27+101705.3 _ 12 31 47.3 __+10_17 05.4 1.733_ 18.83 _ _ 1851 __ __ ___ _ 9.20 _ _ _ __0.009_ $3.5 \times 10^3 $ 2.3 _SDSS J123821.84+030024.2_ 12_38 21.8 +03 00 24.6 0.380_ 18.46 _ 1250 _ _ _ __8.92 ____ _ _ _ _ 0.008_ _$2.2 \times 10^4_$_ 1.4 SDSS J124044.49+231045.8 12 40 44.5 +23 10 46.1 _0.722 18.40 _ __ 1428 _ _ _ __ _8.94 _ _ _ _ _ _ _ 0.008__ $1.6 \times 10^4 $ 1.1 _SDSS_J124119.04+203452.7__ 12 41_19.0 _+20 34 53.4_ 1.492__ 18.44 _ _ 1219 _ _ 9.40 _ _ _ __ _ _ 0.008 $6.8 \times_10^2_$ 4.5 SDSS J124157.90+130104.1 12 41 57.9___+13 01 04.7__ 1.227 18.83 _____ 1538 _ __ _ __ _ _8.95 __ ____ _ 0.007 $9.4 \times 10^3 $ 0.9 PGC 3096192__ _ __ 12 50_29.0 +06 36 11.1 _0.133 16.93 _ _ 1562 _ _ 7.06_ __ _ _ _ 0.003 $8.5 \times 10^7 $ 0.00 SDSS J125414.23+1313
$. While it is clear that the sequence $\{p_n\}\/$ cannot possibly capture all the information contained in $\phi(z)\/$ as all the off-diagonal matrix elements $\langle m|\hat{\rho}|n\rangle \;,\;m\neq n\/$, are ignored, we can easily show that the PND $\{p_n\}\/$ and the distribution ${\cal P}(I)\/$ determine each other uniquely. To recover ${\cal P}(I)\/$ from $\{p_n\}\/$ we define a generating function $\Lambda(K),\;0\leq K<\infty\/$, to represent the latter: $$\begin{aligned} \Lambda(K) &=& \sum\limits^{\infty}_{n=0} (-K)^n p_n /n!\nonumber\\ &=& \int\limits^{\infty}_{0} dI {\cal P}(I)\; e^{-I} J_0(2\sqrt{IK})\;.\end{aligned}$$ There is a great deal of freedom in the way we set up $\Lambda(K)\/$; we have chosen it so that, among other things, on the basis of eqn.(2.8) it is entire analytic in $K\/$. One can then invert eqn.(2.14) by using the Fourier - Bessel integral theorem to get $$\begin{aligned} {\cal P}(I) = e^{I}\int\limits^{\infty}_{0} dK\;\Lambda(K) J_0(2\sqrt{KI})\;.\end{aligned}$$ Several signatures of nonclassicality of the PND are well known. The most familiar is the Mandel Q-parameter criterion which distinguishes (in a global sense) between super and subpoissonian PND’s: $$\begin{aligned} Q &=& ((\Delta n)^2 -\langle n\rangle )/\langle n\rangle \\ &=& (\langle n^2\rangle - \langle n\rangle ^2 - \langle n\rangle )/\langle n\rangle \;,\\ \langle n\rangle &=&\sum\limits^{\infty}_{n=0} n\;p_n = \int\limits^{\infty}_{0} dI{\cal P}(I) I\;,\\ \langle n^2\rangle &=& \sum\limits^{\infty}_{n=0} n^2 p_n = \int\limits
$. While it is clear that the sequence $ \{p_n\}\/$ cannot possibly capture all the information control in $ \phi(z)\/$ as all the away - diagonal matrix elements $ \langle m|\hat{\rho}|n\rangle \;,\;m\neq n\/$, are ignore, we can well show that the PND $ \{p_n\}\/$ and the distribution $ { \cal P}(I)\/$ determine each other uniquely. To recuperate $ { \cal P}(I)\/$ from $ \{p_n\}\/$ we define a generating function $ \Lambda(K),\;0\leq K<\infty\/$, to typify the latter: $ $ \begin{aligned } \Lambda(K) & = & \sum\limits^{\infty}_{n=0 } (-K)^n p_n /n!\nonumber\\ & = & \int\limits^{\infty}_{0 } dI { \cal P}(I)\; e^{-I } J_0(2\sqrt{IK})\;.\end{aligned}$$ There is a great hand of freedom in the way we fructify up $ \Lambda(K)\/$; we have choose it so that, among other things, on the basis of eqn.(2.8) it is entire analytic in $ K\/$. One can then invert eqn.(2.14) by use the Fourier - Bessel integral theorem to get $ $ \begin{aligned } { \cal P}(I) = e^{I}\int\limits^{\infty}_{0 } dK\;\Lambda(K) J_0(2\sqrt{KI})\;.\end{aligned}$$ Several key signature of nonclassicality of the PND are well known. The most familiar is the Mandel Q - parameter criterion which distinguish (in a global sense) between super and subpoissonian PND ’s: $ $ \begin{aligned } Q & = & (( \Delta n)^2 -\langle n\rangle) /\langle n\rangle \\ & = & (\langle n^2\rangle - \langle n\rangle ^2 - \langle n\rangle) /\langle n\rangle \;,\\ \langle n\rangle & = & \sum\limits^{\infty}_{n=0 } n\;p_n = \int\limits^{\infty}_{0 } dI{\cal P}(I) I\;,\\ \langle n^2\rangle & = & \sum\limits^{\infty}_{n=0 } n^2 p_n = \int\limits
$. Whlle it is clear that the sequence $\{p_n\}\/$ caubot povsibly capture all the information containxd ib $\phi(e)\/$ as all the off-diaeonal matgix elemebts $\oqngle m|\hat{\cgo}|n\rangle \;,\;m\nes n\/$, axe ignored, we cak easily shmw that the PNG $\{o_n\}\/$ and the distribution ${\cal P}(I)\/$ determyne eacn lther uniquely. To gesoved ${\cal P}(I)\/$ from $\{p_n\}\/$ we define a generzting flnction $\Lambda(K),\;0\lea K<\infty\/$, to represent the pattfr: $$\begin{aligned} \Lamhda(K) &=& \sum\limuts^{\igdty}_{n=0} (-K)^n p_n /n!\vonumber\\ &=& \int\limits^{\infti}_{0} dI {\cal P}(I)\; e^{-I} J_0(2\sqrt{IK})\;.\end{aligned}$$ Ghere is a greaj sewn of freedon in nhe way we seb up $\Lakbda(K)\/$; wr have chosen lt so thqt, among other things, on the basis of ezn.(2.8) it is ancire analytic in $K\/$. Onw xan tven hnvefr edn.(2.14) gy uaing tje Hourier - Beasel integrql theorem to get $$\btgig{qligned} {\cal P}(J) = e^{I}\ine\lymits^{\infty}_{0} dK\;\Lambda(K) J_0(2\sqrt{KI})\;.\end{aligned}$$ Seneram signatures of nonclaswicality of the PND ate well knjwn. The most familiar is the Mandel Q-parameter crhterikv wklgh dkwtlnguishes (in a global sense) between super and aunppissonian PND’s: $$\begin{alignrd} Q &=& ((\Qelta n)^2 -\langlg n\rangle )/\mangle n\rangle \\ &=& (\lwngle n^2\tangle - \langle n\ramgle ^2 - \langle n\rangle )/\langle n\rangle \;,\\ \lauglw n\rangle &=&\sum\limics^{\infty}_{n=0} n\;p_n = \int\kimitx^{\infty}_{0} dI{\cal P}(I) I\;,\\ \langlz n^2\ranfle &=& \sum\lilits^{\infty}_{h=0} v^2 p_n = \int\limits
$. While it is clear that the cannot capture all information contained in matrix $\langle m|\hat{\rho}|n\rangle \;,\;m\neq are ignored, we easily show that the PND $\{p_n\}\/$ the distribution ${\cal P}(I)\/$ determine each other uniquely. To recover ${\cal P}(I)\/$ from we define a generating function $\Lambda(K),\;0\leq K<\infty\/$, to represent the latter: $$\begin{aligned} \Lambda(K) \sum\limits^{\infty}_{n=0} p_n &=& dI {\cal P}(I)\; e^{-I} J_0(2\sqrt{IK})\;.\end{aligned}$$ There is a great deal of freedom in the way we up $\Lambda(K)\/$; we have chosen it so that, other things, on the of eqn.(2.8) it is entire in One can invert by the Fourier - integral theorem to get $$\begin{aligned} {\cal P}(I) = e^{I}\int\limits^{\infty}_{0} dK\;\Lambda(K) J_0(2\sqrt{KI})\;.\end{aligned}$$ Several signatures of nonclassicality of the are well most familiar the Q-parameter which distinguishes (in sense) between super and subpoissonian PND’s: ((\Delta n)^2 -\langle n\rangle )/\langle n\rangle \\ &=& n^2\rangle - n\rangle ^2 - \langle n\rangle )/\langle \;,\\ \langle n\rangle &=&\sum\limits^{\infty}_{n=0} n\;p_n = \int\limits^{\infty}_{0} dI{\cal I\;,\\ \langle n^2\rangle &=& \sum\limits^{\infty}_{n=0} n^2 p_n = \int\limits
$. While it is clear that the sequEnce $\{p_n\}\/$ cannOt posSibLy cApTure All tHe information cONtaiNed in $\phi(z)\/$ as all the off-diAgonaL mATrix ELeMents $\Langle m|\HAt{\RHO}|n\rAnGlE \;,\;m\nEq N\/$, ArE ignoRed, We can eaSily show thAt tHe pND $\{p_n\}\/$ and the dIStRibution ${\caL P}(I)\/$ Determine eacH otHer uniQuEly. tO recoVer ${\Cal P}(I)\/$ From $\{p_n\}\/$ WE definE a generatInG FunctiON $\Lambda(k),\;0\LEq k<\infTy\/$, to represent the lATtER: $$\begin{aligned} \LAmbda(K) &=& \SuM\LiMITs^{\iNftY}_{n=0} (-K)^n p_n /n!\nonUmBer\\ &=& \inT\Limits^{\iNFtY}_{0} Di {\Cal p}(i)\; e^{-I} J_0(2\sqrt{IK})\;.\end{Aligned}$$ TherE Is a Great dEaL of FReedom In the WaY We sEt up $\Lambda(K)\/$; We haVe chosen iT so thaT, Among otHEr thingS, on the BasIs oF eqn.(2.8) IT iS eNtiRe ANalYTiC in $k\/$. one Can then iNvErT eqn.(2.14) bY usiNG THE FouRieR - BesSel inTegral theorem To gEt $$\beGIn{aLigneD} {\cal P}(i) = e^{I}\iNt\LimitS^{\infty}_{0} DK\;\LamBdA(K) J_0(2\sqrt{KI})\;.\end{aliGned}$$ several siGnaTuRes Of NonclASsicalIty Of tHe PND arE well knOWn. THe MOST fAmiliar is the Mandel q-pARAmEter critErion wHIcH dIStinguisHeS (in A gloBAL sensE) betWEeN super anD subpoISsOnIan PND’s: $$\BeGin{aliGnEd} Q &=& ((\delTa n)^2 -\laNGle n\Rangle )/\Langle n\rAngle \\ &=& (\LAngle n^2\rangle - \laNGle n\rangle ^2 - \lanGLe N\RAnGLe )/\laNglE n\rangle \;,\\ \lanGle n\RAnglE &=&\sum\LImIts^{\INfty}_{n=0} N\;p_n = \inT\lIMiTS^{\infty}_{0} dI{\cal P}(I) I\;,\\ \langlE n^2\Rangle &=& \Sum\liMits^{\infty}_{n=0} n^2 p_n = \Int\limits
$. While it is clear thatthe sequen ce $\ {p_ n\} \/ $ ca nnot possibly capt u re a ll the information con taine di n $\ p hi (z)\/ $ as al l t h e of f- di ago na l m atrix el ements$\langle m |\h at {\rho}|n\ran g le \;,\;m\ne q n \/$, are ign ore d, weca n e a silysho w tha t theP ND $\{ p_n\}\/$an d the d i stribut i o n${\c al P}(I)\/$ deter m in e each other un iquely .T or e cov er${\cal P}( I) \/$ f r om $\{p _ n\ } \ / $ w e define a gen erating fun c tio n $\La mb da( K ),\;0\ leq K <\ i nft y\/$, to re pres ent the l atter: $$\begi n {aligne d} \La mbd a(K ) &= & \ su m\l im i ts^ { \i nft y }_{ n=0} (-K )^ np_n / n!\n o n u m ber\ \ & =& \ int\l imits^{\infty }_{ 0} d I {\ cal P }(I)\ ; e^ {- I} J_ 0(2\sq rt{IK }) \;.\end{aligned }$$There isa g re atde al of freedo m i n t he waywe setu p $ \L a m b da (K)\/$; we have ch os e n i t so tha t, amo n got h er thing s, on the b asisof e q n. (2.8) it is en t ir eanalyti cin $K\ /$ . O necan t h en i nverteqn.(2.1 4) by using the Four i er - Bessel i n te g r al theo rem to get $$\ begi n {ali gned } {\c a l P}( I) =e^ { I} \ int\limits^{\infty} _{ 0} dK\ ;\Lam bda(K) J_0(2\ sqrt{KI})\ ; . \ end{alig ned} $ $Several signat uresof nonclas s icalityof th e PND ar e well kn o w n. The m ost fa mil iar i sthe Mandel Q- p a rame te r crite rio n which di sti ngu ish es (in a gl obal sen se )be tw een supe r and sub po iss on ian PND’ s : $$\b egin{ alig ne d} Q&=& ((\ D el t a n)^ 2-\ lang len\ rangl e )/ \ lan gle n\r angle \\&=& (\la ng le n^2\ra ngle - \lang le n\rangle^2 -\langl e n\rangle )/\langle n\rangle \;, \ \ \lang len\ran gle &=&\sum\ lim its^{\ inf t y}_{n= 0} n\; p_n = \ int \ l imits ^ { \i nft y} _{0} dI{\ c a l P }(I)I\ ;,\\ \langl e n^2\rangle &=&\ sum \limits^{\inf ty} _{n= 0 } n ^2p _n = \ in t \li m i ts
$. While_it is_clear that the sequence_$\{p_n\}\/$ cannot_possibly_capture all_the_information contained in_$\phi(z)\/$ as all_the off-diagonal matrix elements_$\langle m|\hat{\rho}|n\rangle \;,\;m\neq_n\/$,_are ignored, we can easily show that the PND $\{p_n\}\/$ and the distribution ${\cal_P}(I)\/$_determine each_other_uniquely._To recover ${\cal P}(I)\/$ from_$\{p_n\}\/$ we define a generating_function $\Lambda(K),\;0\leq_K<\infty\/$, to represent the latter: $$\begin{aligned} \Lambda(K) &=& \sum\limits^{\infty}_{n=0} (-K)^n_p_n_/n!\nonumber\\ &=& \int\limits^{\infty}_{0} dI_{\cal P}(I)\; e^{-I} J_0(2\sqrt{IK})\;.\end{aligned}$$ There is a great deal of_freedom in the way we set_up $\Lambda(K)\/$; we_have_chosen_it so that, among_other things, on the basis of_eqn.(2.8) it is entire analytic in_$K\/$. One can then invert eqn.(2.14) by_using the Fourier - Bessel integral_theorem to get $$\begin{aligned} {\cal_P}(I) = e^{I}\int\limits^{\infty}_{0}_dK\;\Lambda(K) J_0(2\sqrt{KI})\;.\end{aligned}$$ Several signatures of nonclassicality of_the PND are_well known._The most familiar_is the Mandel Q-parameter criterion which_distinguishes (in a_global sense) between super and subpoissonian_PND’s:_$$\begin{aligned} Q &=& ((\Delta_n)^2_-\langle_n\rangle )/\langle_n\rangle \\ &=&_(\langle_n^2\rangle _-_\langle n\rangle ^2 - \langle n\rangle )/\langle_n\rangle_\;,\\ \langle n\rangle &=&\sum\limits^{\infty}_{n=0} n\;p_n = \int\limits^{\infty}_{0}_ dI{\cal P}(I) I\;,\\ \langle_n^2\rangle_ &=& \sum\limits^{\infty}_{n=0} n^2 p_n_= \int\limits
(x + y)) \big( Q_j(\sigma, \sigma) - Q_j(\sigma, \tau) \big)} ~. \end{aligned}$$ It is easy to check that when $\mu = M$, $A_j(s/M^2, \eta)$ coincides with the amplitude $\tilde{A}_j(s, M^2)$ obtained in Ref. [@egt]. Eqs. (\[aleptf\], \[solutionaj\]) describe all invariant amplitudes for $e^+e^-$ -annihilation into a quark or a lepton pair in the collinear kinematics (\[tmu\], \[umu\]). Scattering amplitudes at large values of $t$ and $u$ {#FB} ===================================================== In this section we calculate the scattering amplitudes $A$ when the restriction of Eqs. (\[tmu\], \[umu\]) for the kinematical configurations (\[tkin\], \[ukin\]) are replaced by $$\label{tm} s \gg M^2 \geq -t \gg \mu^2$$ and $$\label{um} s \gg M^2 \geq -u \gg \mu^2 ~.$$ In this kinematical regions it is more convenient to study the scattering amplitudes $A$ directly, rather than using the invariant amplitudes $A_j$. In order to unify the discussion for both kinematics (\[tm\], \[um\]), let us introduce $$\label{kappat} \kappa = -t ~,$$ when (\[tm\]) is considered and $$\label{kappau} \kappa = -u$$ for the other case (\[um\]). Using this notation, the same parameterization $A = A(s, \mu^2, M^2, \kappa)$ holds for both kinematics (\[tm\], \[um\]). Let us discuss now the evolution equations for $A$. As in the previous case, it is convenient to consider separately the purely QED part, $A^{(QED)}$ and the mixed part, $A'$: $$\label{separ} A(s, \kappa, \mu^2, M^2) = A^{(QED)}(s, \kappa, \mu^2) + A'(s, \kappa, \mu^2, M^2 ).$$ Generalizing Eq. (\[param\]), we can parameterize them as follows: $$A^{(QED
(x + y) ) \big (Q_j(\sigma, \sigma) - Q_j(\sigma, \tau) \big) } ~. \end{aligned}$$ It is easy to check that when $ \mu = M$, $ A_j(s / M^2, \eta)$ coincides with the amplitude $ \tilde{A}_j(s, M^2)$ obtained in Ref.   [ @egt ]. Eqs.   (\[aleptf\ ], \[solutionaj\ ]) identify all changeless amplitudes for $ e^+e^-$ -annihilation into a quark or a lepton couple in the collinear kinematics   (\[tmu\ ], \[umu\ ]). Scattering amplitude at large value of $ t$ and $ u$ { # FB } = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = In this section we calculate the disperse amplitudes $ A$ when the restriction of Eqs.   (\[tmu\ ], \[umu\ ]) for the kinematical shape (\[tkin\ ], \[ukin\ ]) are replaced by $ $ \label{tm } s \gg M^2 \geq -t \gg \mu^2$$ and $ $ \label{um } s \gg M^2 \geq -u \gg \mu^2 ~.$$ In this kinematical regions it is more commodious to study the scattering amplitudes $ A$ directly, rather than using the invariant amplitude $ A_j$. In order to unify the discussion for both kinematics   (\[tm\ ], \[um\ ]), let us precede $ $ \label{kappat } \kappa = -t ~,$$ when (\[tm\ ]) is considered and $ $ \label{kappau } \kappa = -u$$ for the other case (\[um\ ]). Using this note, the same parameterization $ A = A(s, \mu^2, M^2, \kappa)$ have for both kinematics   (\[tm\ ], \[um\ ]). Let us discuss now the development equations for $ A$. As in the previous case, it is convenient to consider separately the strictly QED region, $ A^{(QED)}$ and the mixed part, $ A'$: $ $ \label{separ } A(s, \kappa, \mu^2, M^2) = A^{(QED)}(s, \kappa, \mu^2) + A'(s, \kappa, \mu^2, M^2) .$$ Generalizing Eq.   (\[param\ ]), we can parameterize them as follows: $ $ A^{(QED
(x + y)) \big( Q_j(\sigma, \sigma) - Q_j(\rigma, \tau) \big)} ~. \gne{alignxd}$$ It ia easy tu check that when $\mu = M$, $A_j(s/M^2, \era)$ councides with the ampligude $\tildv{A}_j(s, M^2)$ obraintd in Ref. [@egt]. Eqs. (\[alxltf\], \[solmcionan\]) desermbe all invariakt amplitudas for $e^+e^-$ -anniviuacion into a quark or a lepton pair ig the cplpinear kinematycs (\[tkt\], \[umh\]). Scattering amplitudes at large vamues of $t$ and $u$ {#FB} ===================================================== Im this section we calculatf thf scattering amplihudes $A$ wheb thq restriction of Eqs. (\[tmu\], \[umu\]) for the kinematical configurations (\[tkiv\], \[ukiu\]) are replaxee bj $$\label{tm} s \gj M^2 \gez -t \gg \mu^2$$ and $$\label{uk} s \gg M^2 \geq -u \gg \mu^2 ~.$$ Ik thiv kunematical regions it is more convenienj to study tke scattering amplitueew $A$ dhrecdly, fqthdr uhai uaing tje mnvariant ajplitudes $A_h$. In order to unify trv discussion ror boeh kinematics (\[tm\], \[um\]), let us introduce $$\label{naplat} \kappa = -t ~,$$ when (\[tm\]) is considered and $$\label{nappau} \kap[a = -u$$ for the other case (\[um\]). Using this notation, tve sajd pcvwowtfrization $A = A(s, \mu^2, M^2, \kappa)$ holds for both kinqjauicx (\[tm\], \[um\]). Let us biscuss now the rvllijion equations for $A$. As in the previous cwse, it ys cobvenient uo comsider separately the purelt QED part, $A^{(ZWD)}$ and the mixed pcrt, $A'$: $$\label{szpar} A(s, \kapps, \mu^2, M^2) = A^{(QED)}(s, \kappa, \mu^2) + A'(s, \iappa, \mu^2, M^2 ).$$ Heneralizjvg Eq. (\[param\]), we cav psrdmeterizt them as follows: $$W^{(QED
(x + y)) \big( Q_j(\sigma, \sigma) - \big)} \end{aligned}$$ It easy to check $A_j(s/M^2, coincides with the $\tilde{A}_j(s, M^2)$ obtained Ref. [@egt]. Eqs. (\[aleptf\], \[solutionaj\]) describe invariant amplitudes for $e^+e^-$ -annihilation into a quark or a lepton pair in collinear kinematics (\[tmu\], \[umu\]). Scattering amplitudes at large values of $t$ and $u$ ===================================================== this we the scattering amplitudes $A$ when the restriction of Eqs. (\[tmu\], \[umu\]) for the kinematical configurations (\[tkin\], are replaced by $$\label{tm} s \gg M^2 \geq \gg \mu^2$$ and $$\label{um} \gg M^2 \geq -u \gg ~.$$ this kinematical it more to study the amplitudes $A$ directly, rather than using the invariant amplitudes $A_j$. In order to unify the discussion for kinematics (\[tm\], us introduce \kappa -t when (\[tm\]) is $$\label{kappau} \kappa = -u$$ for the Using this notation, the same parameterization $A = \mu^2, M^2, holds for both kinematics (\[tm\], \[um\]). us discuss now the evolution equations for $A$. in the previous case, it is convenient to consider separately the purely QED part, $A^{(QED)}$ mixed part, $A'$: $$\label{separ} \kappa, \mu^2, M^2) A^{(QED)}(s, \mu^2) A'(s, \mu^2, M^2 Generalizing Eq. (\[param\]), we can parameterize them as follows: $$A^{(QED
(x + y)) \big( Q_j(\sigma, \sigma) - Q_j(\sigma, \tAu) \big)} ~. \end{alIgned}$$ it iS eaSy To chEck tHat when $\mu = M$, $A_j(s/M^2, \ETa)$ coIncides with the amplitudE $\tildE{A}_J(S, M^2)$ obTAiNed in ref. [@egt]. EQS. (\[aLEPtf\], \[SoLuTioNaJ\]) DeScribE alL invariAnt amplituDes FoR $e^+e^-$ -annihilatIOn Into a quark Or a Lepton pair in The CollinEaR kiNEmatiCs (\[tMu\], \[umu\]). scatteRIng ampLitudes at LaRGe valuES of $t$ and $U$ {#fb} ===================================================== IN thiS section we calculaTE tHE scattering ampLitudeS $A$ WHeN THe rEstRiction of EQs. (\[Tmu\], \[umU\]) For the kINeMATIcaL ConfigurationS (\[tkin\], \[ukin\]) arE RepLaced bY $$\lAbeL{Tm} s \gg M^2 \Geq -t \gG \mU^2$$ And $$\Label{um} s \gg M^2 \Geq -u \Gg \mu^2 ~.$$ In thiS kinemATical reGIons it iS more cOnvEniEnt tO StUdY thE sCAttERiNg aMPliTudes $A$ diReCtLy, ratHer tHAN USing The InvaRiant Amplitudes $A_j$. IN orDer tO UniFy the DiscuSsioN fOr botH kinemAtics (\[Tm\], \[Um\]), let us introducE $$\labEl{kappat} \kAppA = -t ~,$$ WheN (\[tM\]) is coNSidereD anD $$\laBel{kappAu} \kappa = -U$$ For ThE OTHeR case (\[um\]). Using this noTaTIOn, The same pArametERiZaTIon $A = A(s, \mu^2, m^2, \kAppA)$ holDS For boTh kiNEmAtics (\[tm\], \[uM\]). Let us DIsCuSs now thE eVolutiOn EquAtiOns foR $a$. As iN the prEvious caSe, it iS Convenient to coNSider separateLY tHE PuREly QeD pArt, $A^{(QED)}$ and tHe miXEd paRt, $A'$: $$\lABeL{sePAr} A(s, \kAppa, \mU^2, M^2) = a^{(qEd)}(S, \kappa, \mu^2) + A'(s, \kappa, \mu^2, M^2 ).$$ GEnEralizIng Eq. (\[Param\]), we can parAmeterize tHEM As followS: $$A^{(QEd
(x + y)) \big( Q_j(\sigm a, \sigma) - Q_ j(\ sig ma , \t au)\big)} ~. \end { alig ned}$$ It is easy to c heckth a t wh e n$\mu= M$, $ A _j ( s /M^ 2, \ eta )$ co incid eswith th e amplitud e $ \t ilde{A}_j(s, M^ 2)$ obtain edin Ref. [@eg t]. Eqs.  ( \[a l eptf\ ],\[sol utiona j \]) de scribe al li nvaria n t ampli t u de s fo r $e^+e^-$ -annih i la t ion into a qua rk oral ep t o n p air in the co ll inear kinemat i cs ( \ [tm u \], \[umu\]). Scatterin g am plitud es at largevalue so f $ t$ and $u$ {#F B} ====== ====== = ======= = ======= ====== === === ==== = == == === =Int hi s s e cti on we ca lc ul ate t he s c a t t erin g a mpli tudes $A$ when the re stri c tio n ofEqs.(\[t mu \], \ [umu\] ) for t he kinematicalconf iguration s ( \[ tki n\ ], \[ u kin\]) ar e r eplaced by $$\ l abe l{ t m } s \gg M^2 \geq -t \ gg \ mu ^2$$ and $$\la b el {u m } s \ggM^ 2 \ geq- u \gg\mu^ 2 ~ .$$ Inthis k i ne ma tical r eg ions i tismor e con v enie nt tostudy th e sca t tering amplitu d es $A$ direct l y, r at h er t han using theinva r iant amp l it ude s $A_j $. In o r de r to unify the discu ss ion fo r bot h kinematics(\[tm\], \ [ u m \]), let usi nt r oduce $$\label {kapp at} \kappa = -t ~,$ $ whe n (\[tm\ ]) is con s i dered an d $ $\l abe l{k a p pa u} \kappa = - u $ $ fo rthe oth ercase (\ [um \]) . U sin gthis nota tion, th esa me p ara meter i zation $ A= A (s , \ mu^2, M^2, \ kappa )$ h ol ds for both k i ne m a tics  ( \[ tm\] , \ [u m\]). Let usdiscuss now theevo l utio neq uations for $A$. Asin the previ ou s c ase, i t is conve nient to consider separ a tely th e p urely QED part, $A ^{( QED)}$ an d the m ixed p art,$A '$: $ $\lab e l {s epa r} A(s, \kap p a , \ mu^2, M ^2)= A^{(Q ED)}(s, \kappa, \m u ^2) + A'(s, \ka ppa , \m u ^ 2, M^ 2 ) . $$ G e ner a l izing Eq. (\[pa ram\]), we c a nparameteri z e t he m as fo llows:$$A^{ ( QED
(x_+ y))_ \big( Q_j(\sigma, \sigma) -_Q_j(\sigma, \tau)_\big)}_~. \end{aligned}$$_It_is easy to_check that when_$\mu = M$, $A_j(s/M^2,_\eta)$ coincides with_the_amplitude $\tilde{A}_j(s, M^2)$ obtained in Ref. [@egt]. Eqs. (\[aleptf\], \[solutionaj\]) describe all invariant amplitudes for $e^+e^-$ -annihilation_into_a quark_or_a_lepton pair in the collinear_kinematics (\[tmu\], \[umu\]). Scattering amplitudes at large_values of_$t$ and $u$ {#FB} ===================================================== In this section we_calculate_the scattering amplitudes_$A$ when the restriction of Eqs. (\[tmu\], \[umu\]) for the_kinematical configurations (\[tkin\], \[ukin\]) are replaced_by $$\label{tm} s \gg_M^2_\geq_-t \gg \mu^2$$ and_$$\label{um} s \gg M^2 \geq -u \gg_\mu^2 ~.$$ In this kinematical regions it_is more convenient to study the scattering_amplitudes $A$ directly, rather than using_the invariant amplitudes $A_j$. In_order to_unify the discussion for both_kinematics (\[tm\], \[um\]), let_us introduce_$$\label{kappat} \kappa = -t_~,$$ when (\[tm\]) is considered and_$$\label{kappau} \kappa = -u$$_for the other case (\[um\]). Using_this_notation, the same_parameterization_$A_= A(s,_\mu^2, M^2, \kappa)$_holds_for both_kinematics (\[tm\],_\[um\]). Let us discuss now the_evolution_equations for $A$. As in the previous_case, it is convenient_to_consider separately the purely_QED part, $A^{(QED)}$ and the_mixed part, $A'$: $$\label{separ} A(s, \kappa, \mu^2,_M^2) =_A^{(QED)}(s, \kappa,_\mu^2) + A'(s, \kappa, \mu^2, M^2 ).$$ Generalizing Eq. (\[param\]), we can parameterize_them as follows: $$A^{(QED
lambda}|<\rho \right\}, \quad {\mathbb{D}}_\rho = \left\{ {\lambda}\in{\mathbb{C}}\mid |{\lambda}|<\rho \right\}.$$ Any smooth map $f:{\mathbb{S}}^1\longrightarrow {\mathcal{M}}(2,{\mathbb{C}})$ can be decomposed into its Fourier series $$f({\lambda}) = \sum_{i\in{\mathbb{Z}}}f_i{\lambda}^i.$$ Let $|\cdot|$ denote a norm on ${\mathcal{M}}(2,{\mathbb{C}})$. Fix some $\rho>1$ and consider $${\left\Vertf\right\Vert}_\rho := \sum_{i\in{\mathbb{Z}}} |f_i|\rho^{|i|}.$$ Let $G$ be a Lie group or algebra of ${\mathcal{M}}(2,{\mathbb{C}})$. We define - $\Lambda G$ as the set of smooth functions $f:{\mathbb{S}}^1 \longrightarrow G$. - $\Lambda G_\rho\subset \Lambda G$ as the set of functions $f$ such that ${\left\Vertf\right\Vert}_{\rho}$ is finite. If $G$ is a group (or an algebra) then $(\Lambda G_\rho,{\left\Vert\cdot\right\Vert}_{\rho})$ is a Banach Lie group (or algebra). - $\Lambda G_\rho^{\geq 0}\subset\Lambda G_\rho$ as the set of functions $f$ such that $f_i=0$ for all $i<0$. - $\Lambda_+ G_\rho\subset \Lambda G_\rho^{\geq 0}$ as the set of functions such that $f_0$ is upper-triangular. - $\Lambda_+^{\mathbb{R}}{\mathrm{SL}}(2,{\mathbb{C}})_\rho\subset \Lambda_+ {\mathrm{SL}}(2,{\mathbb{C}})_\rho$ as the set of functions that have positive elements on the diagonal. We also define $\Lambda{\mathbb{C}}$ as the set of smooth maps from ${\mathbb{S}}^1$ to ${\mathbb{C}}$, and $\Lambda{\mathbb{C}}_\rho$ and $\Lambda {\mathbb{C}}_\rho^{\geq 0}$ as above. Note that every function of $\Lambda G_\rho$ holomorphically extends to ${\mathbb{A}}_\rho$ and that every function of $\
lambda}|<\rho \right\ }, \quad { \mathbb{D}}_\rho = \left\ { { \lambda}\in{\mathbb{C}}\mid |{\lambda}|<\rho \right\}.$$ Any smooth map $ f:{\mathbb{S}}^1\longrightarrow { \mathcal{M}}(2,{\mathbb{C}})$ can be decomposed into its Fourier series $ $ f({\lambda }) = \sum_{i\in{\mathbb{Z}}}f_i{\lambda}^i.$$ lease $ |\cdot|$ announce a norm on $ { \mathcal{M}}(2,{\mathbb{C}})$. Fix some $ \rho>1 $ and consider $ $ { \left\Vertf\right\Vert}_\rho: = \sum_{i\in{\mathbb{Z } } } |f_i|\rho^{|i|}.$$ lease $ G$ be a Lie group or algebra of $ { \mathcal{M}}(2,{\mathbb{C}})$. We specify - $ \Lambda G$ as the set of smooth function $ f:{\mathbb{S}}^1 \longrightarrow G$. - $ \Lambda G_\rho\subset \Lambda G$ as the bent of functions $ f$ such that $ { \left\Vertf\right\Vert}_{\rho}$ is finite. If $ G$ is a group (or an algebra) then $ (\Lambda G_\rho,{\left\Vert\cdot\right\Vert}_{\rho})$ is a Banach Lie group (or algebra). - $ \Lambda G_\rho^{\geq 0}\subset\Lambda G_\rho$ as the set of functions $ f$ such that $ f_i=0 $ for all $ i<0$. - $ \Lambda_+ G_\rho\subset \Lambda G_\rho^{\geq 0}$ as the set of function such that $ f_0 $ is upper - triangular. - $ \Lambda_+^{\mathbb{R}}{\mathrm{SL}}(2,{\mathbb{C}})_\rho\subset \Lambda_+ { \mathrm{SL}}(2,{\mathbb{C}})_\rho$ as the set of functions that get positive elements on the diagonal. We also define $ \Lambda{\mathbb{C}}$ as the set of politic maps from $ { \mathbb{S}}^1 $ to $ { \mathbb{C}}$, and $ \Lambda{\mathbb{C}}_\rho$ and $ \Lambda { \mathbb{C}}_\rho^{\geq 0}$ as above. Note that every routine of $ \Lambda G_\rho$ holomorphically extends to $ { \mathbb{A}}_\rho$ and that every function of $ \
lamhda}|<\rho \right\}, \quad {\mathbb{A}}_\rho = \left\{ {\lambbq}\in{\matibb{C}}\mid |{\lambda}|<\rfo \right\}.$$ Any smooth map $f:{\matibb{S}}^1\oongrughtarrow {\mathcal{M}}(2,{\mathcb{C}})$ can bv decompowed mnto its Fourier series $$n({\jambsw}) = \sbm_{m\in{\mathbb{Z}}}f_i{\lamnda}^i.$$ Let $|\cdmt|$ denote a nosm ou ${\mathcal{M}}(2,{\mathbb{C}})$. Fix some $\rho>1$ and cjnsider $${\lfft\Vertf\right\Vgrt}_\rhp := \suj_{p\ik{\mathbb{Z}}} |f_i|\rho^{|i|}.$$ Let $G$ be a Lie gdoup or algebra of ${\msthcal{M}}(2,{\mathbb{C}})$. We define - $\Lamhda G$ as the set ov smooth fubctijbs $f:{\mathbb{S}}^1 \uongrightagxow G$. - $\Lamgda G_\rho\subset \Lambda G$ as the ret oy functions $f$ sufv that ${\left\Tertf\rpght\Vert}_{\rho}$ is finite. If $G$ ix a group (or ak algxbra) then $(\Lambda G_\rho,{\left\Tert\cdot\right\Vert}_{\rho})$ is a Bandck Lie group (or algebrq). - $\Lamtda C_\rho^{\ewq 0}\rubaev\Lajbda G_\gho$ as the sef of functiins $f$ such that $f_i=0$ gow all $i<0$. - $\Lambsa_+ G_\rhj\stbset \Lambda G_\rho^{\geq 0}$ as the set of funbtiohs such that $f_0$ is upper-rriangular. - $\Lambda_+^{\majhbb{R}}{\mathri{SL}}(2,{\mathbb{C}})_\rho\subset \Lambda_+ {\mathrm{SL}}(2,{\mathbb{C}})_\rho$ as dhe sxt of fmnctkind that have positive elements on the diagonal. Rs slxo define $\Lambba{\mathbb{C}}$ as the sft jf smooth mapr from ${\mafhbb{S}}^1$ to ${\mathbb{C}}$, ajd $\Lambqa{\matybb{C}}_\rho$ agd $\Lsmbda {\mathbb{C}}_\rho^{\geq 0}$ as aboce. Note that wvery function of $\Pambda G_\rho$ holpmorpnically extends to ${\mathyb{A}}_\rho$ and that egery funcfkon of $\
lambda}|<\rho \right\}, \quad {\mathbb{D}}_\rho = \left\{ {\lambda}\in{\mathbb{C}}\mid Any map $f:{\mathbb{S}}^1\longrightarrow can be decomposed = Let $|\cdot|$ denote norm on ${\mathcal{M}}(2,{\mathbb{C}})$. some $\rho>1$ and consider $${\left\Vertf\right\Vert}_\rho := |f_i|\rho^{|i|}.$$ Let $G$ be a Lie group or algebra of ${\mathcal{M}}(2,{\mathbb{C}})$. We define $\Lambda G$ as the set of smooth functions $f:{\mathbb{S}}^1 \longrightarrow G$. - $\Lambda \Lambda as set functions $f$ such that ${\left\Vertf\right\Vert}_{\rho}$ is finite. If $G$ is a group (or an algebra) then G_\rho,{\left\Vert\cdot\right\Vert}_{\rho})$ is a Banach Lie group (or algebra). $\Lambda G_\rho^{\geq 0}\subset\Lambda G_\rho$ the set of functions $f$ that for all - G_\rho\subset G_\rho^{\geq 0}$ as set of functions such that $f_0$ is upper-triangular. - $\Lambda_+^{\mathbb{R}}{\mathrm{SL}}(2,{\mathbb{C}})_\rho\subset \Lambda_+ {\mathrm{SL}}(2,{\mathbb{C}})_\rho$ as the set of functions have positive the diagonal. also $\Lambda{\mathbb{C}}$ the set of from ${\mathbb{S}}^1$ to ${\mathbb{C}}$, and $\Lambda{\mathbb{C}}_\rho$ 0}$ as above. Note that every function of G_\rho$ holomorphically to ${\mathbb{A}}_\rho$ and that every function $\
lambda}|<\rho \right\}, \quad {\mathbb{D}}_\Rho = \left\{ {\lamBda}\in{\MatHbb{c}}\mId |{\laMbda}|<\Rho \right\}.$$ Any smoOTh maP $f:{\mathbb{S}}^1\longrightarroW {\mathCaL{m}}(2,{\matHBb{c}})$ can bE decompOSeD INto ItS FOurIeR SeRies $$f({\LamBda}) = \sum_{i\In{\mathbb{Z}}}f_I{\laMbDa}^i.$$ Let $|\cdot|$ deNOtE a norm on ${\maThcAl{M}}(2,{\mathbb{C}})$. FiX soMe $\rho>1$ aNd ConSIder $${\lEft\vertf\Right\VERt}_\rho := \sUm_{i\in{\mathBb{z}}} |F_i|\rho^{|i|}.$$ lEt $G$ be a LIE GrOup oR algebra of ${\mathcal{m}}(2,{\MaTHbb{C}})$. We define - $\LaMbda G$ aS tHE sET Of sMooTh functionS $f:{\MathbB{s}}^1 \longriGHtARROw G$. - $\lAmbda G_\rho\subsEt \Lambda G$ as THe sEt of fuNcTioNS $f$ such That ${\lEfT\verTf\right\Vert}_{\Rho}$ iS finite. If $g$ is a grOUp (or an aLGebra) thEn $(\LambDa G_\Rho,{\Left\vErT\cDot\RiGHt\VERt}_{\Rho})$ IS a BAnach Lie GrOuP (or alGebrA). - $\lAMBda G_\Rho^{\Geq 0}\sUbset\lambda G_\rho$ as tHe sEt of FUncTions $F$ such That $F_i=0$ For alL $i<0$. - $\LambDa_+ G_\rhO\sUbset \Lambda G_\rho^{\Geq 0}$ aS the set of FunCtIonS sUch thAT $f_0$ is upPer-TriAngular. - $\lambda_+^{\mAThbB{R}}{\MATHrM{SL}}(2,{\mathbb{C}})_\rho\subseT \LAMBdA_+ {\mathrm{Sl}}(2,{\mathbB{c}})_\rHo$ AS the set oF fUncTionS THat haVe poSItIve elemeNts on tHE dIaGonal. We AlSo defiNe $\lamBda{\MathbB{c}}$ as tHe set oF smooth mAps frOM ${\mathbb{S}}^1$ to ${\mathBB{C}}$, and $\Lambda{\maTHbB{c}}_\RhO$ And $\LAmbDa {\mathbb{C}}_\rhO^{\geq 0}$ AS aboVe. NoTE tHat EVery fUnctiOn OF $\LAMbda G_\rho$ holomorphicAlLy exteNds to ${\Mathbb{A}}_\rho$ and That every fUNCTion of $\
lambda}|<\rho \right\}, \q uad {\math bb{D} }_\ rho = \le ft\{ {\lambda}\in{ \ math bb{C}}\mid |{\lambda}| <\rho \ r ight \ }. $$ An y smoot h m a p $f :{ \m ath bb { S} }^1\l ong rightar row {\math cal {M }}(2,{\mathb b {C }})$ can b e d ecomposed in toits Fo ur ier serie s $ $f({\ lambda } ) = \s um_{i\in{ \m a thbb{Z } }}f_i{\ l a mb da}^ i.$$ Let $|\cdot| $ d e note a norm on ${\ma th c al { M }}( 2,{ \mathbb{C} }) $. Fi x some $ \ rh o > 1 $ a n d consider $$ {\left\Vert f \ri ght\Ve rt }_\ r ho :=\sum_ {i \ in{ \mathbb{Z}} } |f _i|\rho^{ |i|}.$ $ Let $G $ be a L ie gro uporalge b ra o f $ {\ m ath c al {M} } (2, {\mathbb {C }} )$. W e de f i n e - $ \Lam bda G $ as the setofsmoo t h f uncti ons $ f:{\ ma thbb{ S}}^1\long ri ghtarrow G$. - $ \Lambda G _\r ho \su bs et \L a mbda G $ a s t he setof func t ion s$ f $ s uch that ${\left\V er t f \r ight\Ver t}_{\r h o} $i s finite .If$G$i s a gr oup( or an alge bra) t h en $ (\Lambd aG_\rho ,{ \le ft\ Vert\ c dot\ right\ Vert}_{\ rho}) $ is a Banach L i e group (or a l ge b r a) . - $ \Lambda G_\ rho^ { \geq 0}\ s ub set \ Lambd a G_\ rh o $a s the set of functi on s $f$suchthat $f_i=0$for all $i < 0 $ . - $ \Lam b da _ + G_\rho\subse t \La mbda G_\rh o ^{\geq 0 }$ as the set of funct i o ns suchtha t $ f_0 $ i s up per-triangula r . - $\Lambd a_+ ^{\math bb{ R}} {\m ath rm {SL}}(2,{ \mathbb{ C} }) _\ rh o\s ubset \Lambda_ +{\m at hrm {SL}} ( 2,{\ma thbb{ C}}) _\ rh o $ a s the s e to f fun ct io ns t hat h ave p osit i veelement s on thedia g onal . W e alsodefine $\Lamb da {\mathbb{C }} $ a s thes e t of smo oth maps from ${\mathbb { S}}^1$to${\ma thbb {C}}$, an d $ \Lambd a{\ m athbb{ C}}_\r ho$ a nd $\ L a mbda{ \ ma thb b{ C}}_\rho^{ \ g eq0}$ a sabov e. Note that every functi o n o f $\Lambda G_ \rh o$ h o l om orp h ic a lly e x ten d s to ${\mathbb{A }}_\rho$ a nd th at every f u nct io n of $\
lambda}|<\rho \right\},_\quad {\mathbb{D}}_\rho_= \left\{ {\lambda}\in{\mathbb{C}}\mid |{\lambda}|<\rho_\right\}.$$ Any_smooth_map $f:{\mathbb{S}}^1\longrightarrow_{\mathcal{M}}(2,{\mathbb{C}})$_can be decomposed_into its Fourier_series $$f({\lambda}) = \sum_{i\in{\mathbb{Z}}}f_i{\lambda}^i.$$_Let $|\cdot|$ denote_a_norm on ${\mathcal{M}}(2,{\mathbb{C}})$. Fix some $\rho>1$ and consider $${\left\Vertf\right\Vert}_\rho := \sum_{i\in{\mathbb{Z}}} |f_i|\rho^{|i|}.$$ Let $G$_be_a Lie_group_or_algebra of ${\mathcal{M}}(2,{\mathbb{C}})$. We define -_ $\Lambda G$ as_the set_of smooth functions $f:{\mathbb{S}}^1 \longrightarrow G$. - _$\Lambda_G_\rho\subset \Lambda G$_as the set of functions $f$ such that ${\left\Vertf\right\Vert}_{\rho}$_is finite. If $G$ is a_group (or an_algebra)_then_$(\Lambda G_\rho,{\left\Vert\cdot\right\Vert}_{\rho})$ is a_Banach Lie group (or algebra). - _ $\Lambda G_\rho^{\geq 0}\subset\Lambda G_\rho$ as_the set of functions $f$ such that_$f_i=0$ for all $i<0$. - _$\Lambda_+ G_\rho\subset \Lambda G_\rho^{\geq 0}$_as the_set of functions such that_$f_0$ is upper-triangular. -_ _$\Lambda_+^{\mathbb{R}}{\mathrm{SL}}(2,{\mathbb{C}})_\rho\subset \Lambda_+ {\mathrm{SL}}(2,{\mathbb{C}})_\rho$_as the set of functions that_have positive elements_on the diagonal. We also define $\Lambda{\mathbb{C}}$_as_the set of_smooth_maps_from ${\mathbb{S}}^1$_to ${\mathbb{C}}$, and_$\Lambda{\mathbb{C}}_\rho$_and $\Lambda_{\mathbb{C}}_\rho^{\geq_0}$ as above. Note that every_function_of $\Lambda G_\rho$ holomorphically extends to ${\mathbb{A}}_\rho$_and that every function_of_$\
starting point is the following classification of homological ring epimorphism from [@BS]: *([@BS Theorem 5.23])*\[T:BSepi\] Let $R$ be a valuation domain. Then there is a bijection between: 1. non-dense admissible systems $\mathcal{X}$ in $\operatorname*{Spec}(R)$, and 2. epiclasses of homological ring epimorphisms $\lambda: R \rightarrow S$. The bijection consists of two mutually inverse assignments: $(i) \rightarrow (ii)$: [@BS Proposition 5.5(2)] assigns to $\lambda$ the set of all intervals obtained as follows: For each maximal ideal $\operatorname*{\mathfrak{n}}\in \operatorname*{mSpec}(S)$, the composition map $R \xrightarrow{\lambda} S \xrightarrow{\text{can}} S_{\operatorname*{\mathfrak{n}}}$ is equivalent to the natural map $R \rightarrow R_{\operatorname*{\mathfrak{q}}}/\operatorname*{\mathfrak{p}}$ for some interval $[\operatorname*{\mathfrak{p}},\operatorname*{\mathfrak{q}}]$ in $\operatorname*{Spec}(R)$ with $\operatorname*{\mathfrak{p}}$ idempotent. Then $\mathcal{X}$ is the collection of all intervals obtained by going through all maximal ideals of the commutative ring $S$. $(ii) \rightarrow (i)$: [@BS Construction 5.22] constructs a homological ring epimorphism $\lambda: R \rightarrow S$ from $\mathcal{X}$ by taking certain direct limits of finite products of copies of modules of the form $R_{\operatorname*{\mathfrak{q}}}/\operatorname*{\mathfrak{p}}$ running over intervals $[\operatorname*{\mathfrak{p}},\operatorname*{\mathfrak{q}}] \in \mathcal{X}$. Combining Theorem \[T:BSepi\] and Theorem \[T:epiclass\], we see that non-dense admissible systems in $\operatorname*{Spec}(R)$ correspond to extension-closed bireflective subcategories of $\operatorname*{Mod-R}$. The next step is to compute these subcategories. \[L:birefl\] Let $R$ be a valuation domain and $\mathcal{X}$ be a non-dense admissible system in $\operatorname*{Spec}(R)$. Then the extension-closed bireflective subcategory $\operatorname*{Mod-S}\simeq \mathcal{B
starting point is the following classification of homologic gang epimorphism from [ @BS ]: * ([ @BS Theorem 5.23])*\[T: BSepi\ ] Let $ R$ be a valuation knowledge domain. Then there be a bijection between: 1. non - dense admissible systems $ \mathcal{X}$ in $ \operatorname*{Spec}(R)$, and 2. epiclasses of homological band epimorphisms $ \lambda: R \rightarrow S$. The bijection consist of two mutually inverse assignments: $ (iodine) \rightarrow (ii)$: [ @BS Proposition 5.5(2) ] assigns to $ \lambda$ the set of all intervals prevail as follows: For each maximal ideal $ \operatorname*{\mathfrak{n}}\in \operatorname*{mSpec}(S)$, the composition function $ R \xrightarrow{\lambda } S \xrightarrow{\text{can } } S_{\operatorname*{\mathfrak{n}}}$ is equivalent to the natural function $ R \rightarrow R_{\operatorname*{\mathfrak{q}}}/\operatorname*{\mathfrak{p}}$ for some time interval $ [ \operatorname*{\mathfrak{p}},\operatorname*{\mathfrak{q}}]$ in $ \operatorname*{Spec}(R)$ with $ \operatorname*{\mathfrak{p}}$ idempotent. Then $ \mathcal{X}$ is the solicitation of all intervals obtained by going through all maximal ideals of the commutative band $ S$. $ (ii) \rightarrow (i)$: [ @BS Construction 5.22 ] constructs a homological ring epimorphism $ \lambda: roentgen \rightarrow S$ from $ \mathcal{X}$ by taking certain direct limits of finite products of copy of modules of the form $ R_{\operatorname*{\mathfrak{q}}}/\operatorname*{\mathfrak{p}}$ running over intervals $ [ \operatorname*{\mathfrak{p}},\operatorname*{\mathfrak{q } } ] \in \mathcal{X}$. Combining Theorem   \[T: BSepi\ ] and Theorem   \[T: epiclass\ ], we see that non - dense admissible systems in $ \operatorname*{Spec}(R)$ correspond to extension - closed bireflective subcategories of $ \operatorname*{Mod - R}$. The next step is to compute these subcategories. \[L: birefl\ ] Let $ R$ be a valuation domain and $ \mathcal{X}$ be a non - dense admissible system in $ \operatorname*{Spec}(R)$. Then the extension - close bireflective subcategory $ \operatorname*{Mod - S}\simeq \mathcal{B
stwrting point is the folluwing classificcrion oh homolkgical rkng epimorphism from [@BS]: *([@BS Thxoren 5.23])*\[T:BStii\] Let $R$ be a valuatiun domain. Then thwre ms a bijection bxfween: 1. kjn-dehde abmmssible systems $\mathcal{X}$ hn $\operatornama*{Soee}(R)$, and 2. epiclasses of homological rigg epimprohisms $\lambda: T \rignearrkw S$. The bijection consists of two jutuallj inverse assignmrnts: $(i) \rightarrow (ii)$: [@BS Prooosihion 5.5(2)] assigns to $\lwmbda$ the sgf os all intervaus obtained as follows: For each maximal ideal $\operatofname*{\kathfrak{n}}\ib \ipegdtorname*{mSpxc}(S)$, thv composition map $R \xsightartow{\lambda} S \xrlghtacrow{\rext{can}} S_{\operatorname*{\kathfrak{n}}}$ is equivwlent to dhz natural map $R \rightqreow R_{\mperdtorvqme*{\oatgfcak{s}}}/\operahoriame*{\mathfrai{p}}$ for some interval $[\operatornsmq*{\nathfrak{p}},\operztornaie*{\iathfrak{q}}]$ in $\operatorname*{Spec}(R)$ with $\opegatodname*{\mathfrak{p}}$ idempotebt. Then $\mathcal{X}$ is tje colleceion of all intervals obtained by going through anl maeioal iqdqld of the commutative ring $S$. $(ii) \rightarrow (i)$: [@BS Domsnruction 5.22] construgts a homological tijg gpimorphism $\laobda: R \rifhtarrow S$ from $\mahhcal{X}$ fy tajing certwin cirect limits of finite proeucts of coppes if modules of the yorm $R_{\operaturnake*{\matnfrak{q}}}/\operatorname*{\mathfxak{p}}$ rhnning over intervala $[\operatorname*{\matffrsk{[}},\operatorname*{\mathfrak{q}}] \in \iathcal{X}$. Cimbiuing Theurem \[J:BSepi\] wnd Theorel \[T:epignass\], we see that nln-dende admissiblf systems in $\operatorname*{Spec}(R)$ rprrespond to eftetsion-clofed blreflective subsategories of $\pperatoxname*{Mud-R}$. The neqt step iv to computq these subcadggories. \[L:birefn\] Let $R$ fe a valyation aumain and $\mathval{X}$ be a non-dense qdmissible system ln $\opgrztorname*{Spec}(R)$. Tktn uhw extension-cloxed biwevlxctivq subcategory $\opefatutname*{Ood-S}\simtz \nxthcsl{B
starting point is the following classification of epimorphism [@BS]: *([@BS 5.23])*\[T:BSepi\] Let $R$ there a bijection between: non-dense admissible systems in $\operatorname*{Spec}(R)$, and 2. epiclasses of ring epimorphisms $\lambda: R \rightarrow S$. The bijection consists of two mutually inverse $(i) \rightarrow (ii)$: [@BS Proposition 5.5(2)] assigns to $\lambda$ the set of all obtained follows: each ideal $\operatorname*{\mathfrak{n}}\in \operatorname*{mSpec}(S)$, the composition map $R \xrightarrow{\lambda} S \xrightarrow{\text{can}} S_{\operatorname*{\mathfrak{n}}}$ is equivalent to the natural $R \rightarrow R_{\operatorname*{\mathfrak{q}}}/\operatorname*{\mathfrak{p}}$ for some interval $[\operatorname*{\mathfrak{p}},\operatorname*{\mathfrak{q}}]$ in with $\operatorname*{\mathfrak{p}}$ idempotent. Then is the collection of all obtained going through maximal of commutative ring $S$. \rightarrow (i)$: [@BS Construction 5.22] constructs a homological ring epimorphism $\lambda: R \rightarrow S$ from $\mathcal{X}$ by certain direct finite products copies modules the form $R_{\operatorname*{\mathfrak{q}}}/\operatorname*{\mathfrak{p}}$ intervals $[\operatorname*{\mathfrak{p}},\operatorname*{\mathfrak{q}}] \in \mathcal{X}$. Combining Theorem \[T:epiclass\], we see that non-dense admissible systems in correspond to bireflective subcategories of $\operatorname*{Mod-R}$. The next is to compute these subcategories. \[L:birefl\] Let $R$ a valuation domain and $\mathcal{X}$ be a non-dense admissible system in $\operatorname*{Spec}(R)$. Then the extension-closed $\operatorname*{Mod-S}\simeq \mathcal{B
starting point is the followiNg classifiCatioN of HomOlOgicAl riNg epimorphism fROm [@BS]: *([@bS Theorem 5.23])*\[T:BSepi\] Let $R$ be a ValuaTiON domAIn. then tHere is a BIjECTioN bEtWeeN: 1. nON-dEnse aDmiSsible sYstems $\mathCal{x}$ iN $\operatornamE*{spEc}(R)$, and 2. epicLasSes of homologIcaL ring ePiMorPHisms $\LamBda: R \rIghtarROw S$. The Bijection CoNSists oF Two mutuALLy InveRse assignments: $(i) \riGHtARrow (ii)$: [@BS ProposItion 5.5(2)] aSsIGnS TO $\laMbdA$ the set of aLl InterVAls obtaINeD AS FolLOws: For each maxImal ideal $\opERatOrname*{\MaThfRAk{n}}\in \oPeratOrNAme*{MSpec}(S)$, the coMposItion map $R \XrightARrow{\lamBDa} S \xrigHtarroW{\teXt{cAn}} S_{\oPErAtOrnAmE*{\MatHFrAk{n}}}$ IS eqUivalent To ThE natuRal mAP $r \RIghtArrOw R_{\oPeratOrname*{\mathfraK{q}}}/\oPeraTOrnAme*{\maThfraK{p}}$ foR sOme inTerval $[\OperaToRname*{\mathfrak{p}},\oPeraTorname*{\maThfRaK{q}}]$ iN $\oPeratORname*{SPec}(r)$ wiTh $\operaTorname*{\MAthFrAK{P}}$ IdEmpotent. Then $\mathcaL{X}$ IS ThE collectIon of aLL iNtERvals obtAiNed By goING throUgh aLL mAximal idEals of THe CoMmutatiVe Ring $S$. $(iI) \rIghTarRow (i)$: [@Bs consTructiOn 5.22] constrUcts a HOmological ring EPimorphism $\lamBDa: r \RIgHTarrOw S$ From $\mathcal{x}$ by tAKing CertAIn DirECt limIts of FiNItE Products of copies of mOdUles of The foRm $R_{\operatornaMe*{\mathfrak{Q}}}/\OPEratornaMe*{\maTHfRAk{p}}$ running over InterVals $[\operatORname*{\matHfrak{P}},\operatoRname*{\mathFRAk{q}}] \in \matHcaL{X}$. COmbIniNG thEorem \[T:BSepi\] anD tHeorEm \[t:epiclaSs\], wE see thaT noN-deNse AdmIsSible systEms in $\opeRaToRnAmE*{SpEc}(R)$ coRRespond tO eXteNsIon-CloseD BireflEctivE subCaTeGOriEs of $\opeRAtORName*{moD-R}$. the nExt StEp is tO comPUte These suBcategoriEs. \[L:BIrefL\] LEt $r$ be a valUation domain aNd $\Mathcal{X}$ be A nOn-dEnse adMISsible syStem in $\operatorname*{Spec}(R)$. tHen the eXteNsion-ClosEd birefleCtiVe subcAteGOry $\opeRatornAme*{MoD-S}\SimEQ \MathcAL{b
starting point is the fol lowing cla ssifi cat ion o f ho molo gical ring epi m orph ism from [@BS]: *([@B S The or e m 5. 2 3] )*\[T :BSepi\ ] L e t $R $be ava l ua tiondom ain. Th en there i s a b ijection bet w ee n: 1. no n-d ense admissi ble syste ms $\ m athca l{X }$ in $\ope r atorna me*{Spec} (R ) $, and 2. ep i c la sses of homological r i ng epimorphisms $ \lambd a: R\ r igh tar row S$. T he bije c tion co n si s t s of two mutuallyinverse ass i gnm ents: $ (i) \right arrow ( i i)$ : [@BS Prop osit ion 5.5(2 )] ass i gns to$ \lambda $ theset of all in te rva ls obt a in eda s f ollows:Fo reachmaxi m a l idea l $ \ope rator name*{\mathfr ak{ n}}\ i n \ opera torna me*{ mS pec}( S)$, t he co mp osition map $R\xri ghtarrow{ \la mb da} S \xri g htarro w{\ tex t{can}} S_{\op e rat or n a m e* {\mathfrak{n}}}$ i se q ui valent t o then at ur a l map $R \ rig htar r o w R_{ \ope r at orname*{ \mathf r ak {q }}}/\op er atorna me *{\ mat hfrak { p}}$ for s ome inte rval$ [\operatorname * {\mathfrak{p} } ,\ o p er a torn ame *{\mathfrak {q}} ] $ in $\o p er ato r name* {Spec }( R )$ with $\operatorname *{ \mathf rak{p }}$ idempoten t. Then $\ m a t hcal{X}$ ist he collection ofall i ntervals o b tained b y goi ng throu gh all ma x i mal idea lsofthe co m m ut ative ring $S $ . $( ii ) \righ tar row (i) $:[@B S C ons tr uction 5. 22] cons tr uc ts a ho molog i cal ring e pim or phi sm $\ l ambda: R \r ight ar ro w S$ from $ \ ma t h cal{ X} $by t aki ng cert aind ire ct limi ts of fin ite prod uc ts of cop ies of module sof the for m$R_ {\oper a t orname*{ \mathfrak{q}}}/\operato r name*{\ mat hfrak {p}} $ running ov er int erv a ls $[\ operat ornam e* {\m a t hfrak { p }} ,\o pe ratorname* { \ mat hfrak {q }}]\in \ma thcal{X}$. Combin i ngTheorem \[T:B Sep i\]a n dThe o re m  \[ T: e pic l a ss\], we see th at non-den se ad missible s y ste ms in $\o perator name* { Spec}(R )$ corres pond to e xt ensi o n -cl osed biref lectivesubcatego r ies o f $ \oper ato rname* {M od- R}$.The ne x t s tep i s to c om pute t hesesu bcategor ies. \[L:birefl\] Let$R$ be a va lua tion doma ina nd$\mathcal {X}$ be a non- den seadmis sib l e sys temi n$\o p erato rnam e *{Spec}(R ) $. Th e n t he extensio n - c los ed bi ref l ective sub category $\operat o rname*{Mod-S}\ sime q \ma thc a l{B
starting_point is_the following classification of_homological ring_epimorphism_from [@BS]: *([@BS_Theorem_5.23])*\[T:BSepi\] Let $R$_be a valuation_domain. Then there is_a bijection between: 1.__non-dense admissible systems $\mathcal{X}$ in $\operatorname*{Spec}(R)$, and 2. epiclasses of homological ring epimorphisms $\lambda:_R_\rightarrow S$. The_bijection_consists_of two mutually inverse assignments: $(i)_\rightarrow (ii)$: [@BS Proposition 5.5(2)]_assigns to_$\lambda$ the set of all intervals obtained as_follows:_For each maximal_ideal $\operatorname*{\mathfrak{n}}\in \operatorname*{mSpec}(S)$, the composition map $R \xrightarrow{\lambda} S_\xrightarrow{\text{can}} S_{\operatorname*{\mathfrak{n}}}$ is equivalent to the_natural map $R_\rightarrow_R_{\operatorname*{\mathfrak{q}}}/\operatorname*{\mathfrak{p}}$_for some interval $[\operatorname*{\mathfrak{p}},\operatorname*{\mathfrak{q}}]$_in $\operatorname*{Spec}(R)$ with $\operatorname*{\mathfrak{p}}$ idempotent. Then_$\mathcal{X}$ is the collection of all_intervals obtained by going through all maximal_ideals of the commutative ring $S$. $(ii)_\rightarrow (i)$: [@BS Construction 5.22]_constructs a_homological ring epimorphism $\lambda: R_\rightarrow S$ from_$\mathcal{X}$ by_taking certain direct_limits of finite products of copies_of modules of_the form $R_{\operatorname*{\mathfrak{q}}}/\operatorname*{\mathfrak{p}}$ running over intervals_$[\operatorname*{\mathfrak{p}},\operatorname*{\mathfrak{q}}]_\in \mathcal{X}$. Combining Theorem \[T:BSepi\]_and_Theorem \[T:epiclass\],_we see_that non-dense admissible_systems_in $\operatorname*{Spec}(R)$_correspond_to extension-closed bireflective subcategories of $\operatorname*{Mod-R}$._The_next step is to compute these subcategories. \[L:birefl\]_Let $R$ be a_valuation_domain and $\mathcal{X}$ be_a non-dense admissible system in_$\operatorname*{Spec}(R)$. Then the extension-closed bireflective subcategory_$\operatorname*{Mod-S}\simeq \mathcal{B
^{16})-q^2f(q^{4},q^{26})\bigl)\bigr(f(q^{13},q^{17})-qf(q^7,q^{23})\bigl)\notag\\ &\;\;\;+q\bigr(f(q^8,q^{22})-q^2f(q^2,q^{28})\bigl)\bigr(f(q^{11},q^{19})-q^3f(q,q^{29})\bigl).\label{60pre3}\end{aligned}$$ Recall that the Rogers-Ramanujan functions are defined by $$\label{GH} G(q) :=\sum_{n=0}^{{\infty}}{\dfrac}{q^{n^2}}{(q;q)_n} \qquad \text{and} \qquad H(q) :=\sum_{n=0}^{{\infty}}{\dfrac}{q^{n(n+1)}}{(q;q)_n}.$$ These functions satisfy the famous Rogers–Ramanujan identities [@cp pp. 214–215] $$\label{rridents} G(q) = {\dfrac}{1}{(q;q^5)_{{\infty}}(q^4;q^5)_{{\infty}}} \qquad \text{and} \qquad H(q) = {\dfrac}{1}{(q^2;q^5)_{{\infty}}(q^3;q^5)_{{\infty}}}.$$ Our proof makes use of one of Ramanujan’s forty identities for the Rogers-Ramanujan functions, namely [@watson] $$G(q)G(q^4)-qH(q)H(q^4)={\dfrac}{{\varphi}(q^5)}{E(q^2)}.\label{40gh}$$ Next, we employ the quintuple product identity, with $q$ replaced by $q^{10}$ and $a=-q$ to find that $$f(-q^{13},-q^{17})+qf(-q^7,-q^{23}) = E(q^{10}){\dfrac}{f(-q^2,-q^8)}{f(-q,-q^9)}=E(q^2)G(q).\label{qp1}$$ Similarly, from we find $$\begin{aligned} E(q^2)H(q)&=f(-q^{11},-q^{19})+q^3f(-
^{16})-q^2f(q^{4},q^{26})\bigl)\bigr(f(q^{13},q^{17})-qf(q^7,q^{23})\bigl)\notag\\ & \;\;\;+q\bigr(f(q^8,q^{22})-q^2f(q^2,q^{28})\bigl)\bigr(f(q^{11},q^{19})-q^3f(q, q^{29})\bigl).\label{60pre3}\end{aligned}$$ Recall that the Rogers - Ramanujan functions are defined by $ $ \label{GH } G(q): = \sum_{n=0}^{{\infty}}{\dfrac}{q^{n^2}}{(q;q)_n } \qquad \text{and } \qquad H(q): = \sum_{n=0}^{{\infty}}{\dfrac}{q^{n(n+1)}}{(q;q)_n}.$$ These functions satisfy the celebrated Rogers – Ramanujan identities [ @cp pp.   214–215 ] $ $ \label{rridents } G(q) = { \dfrac}{1}{(q;q^5)_{{\infty}}(q^4;q^5)_{{\infty } } } \qquad \text{and } \qquad H(q) = { \dfrac}{1}{(q^2;q^5)_{{\infty}}(q^3;q^5)_{{\infty}}}.$$ Our validation make use of one of Ramanujan ’s forty identity for the Rogers - Ramanujan function, namely [ @watson ] $ $ G(q)G(q^4)-qH(q)H(q^4)={\dfrac}{{\varphi}(q^5)}{E(q^2)}.\label{40gh}$$ Next, we employ the quintuple product identity, with $ q$ substitute by $ q^{10}$ and $ a=-q$ to find oneself that $ $ f(-q^{13},-q^{17})+qf(-q^7,-q^{23 }) = E(q^{10}){\dfrac}{f(-q^2,-q^8)}{f(-q,-q^9)}=E(q^2)G(q).\label{qp1}$$ Similarly, from we find $ $ \begin{aligned } E(q^2)H(q)&=f(-q^{11},-q^{19})+q^3f(-
^{16})-q^2f(q^{4},e^{26})\bigl)\bigr(f(q^{13},q^{17})-qf(q^7,q^{23})\bigl)\notan\\ &\;\;\;+q\bigr(f(q^8,q^{22})-q^2f(q^2,q^{28})\bigl)\bigr(f(q^{11},x^{19})-q^3f(q,q^{29})\bjgl).\label{60ore3}\end{aligned}$$ Recall that thx Roters-Rqmanujan functions are defined hy $$\label{TH} G(q) :=\wum_{n=0}^{{\infty}}{\dhdac}{q^{n^2}}{(q;q)_k} \qquzf \tert{end} \qquad H(q) :=\sum_{k=0}^{{\infty}}{\dfrac}{x^{n(n+1)}}{(q;q)_n}.$$ These futcgilns satisfy the famous Rogers–Ramanutan idemtlties [@cp pp. 214–215] $$\lafel{rgiqenta} G(q) = {\dfrac}{1}{(q;q^5)_{{\infty}}(q^4;q^5)_{{\infty}}} \qquad \texf{and} \qqlad H(q) = {\dfrac}{1}{(q^2;q^5)_{{\infyy}}(q^3;q^5)_{{\infty}}}.$$ Our proof makes kse lf one of Ramanujaj’s forty idghtieues for the Fogers-Ramanujan functikns, namely [@watson] $$G(q)G(q^4)-qH(q)H(q^4)={\dfraz}{{\varpki}(q^5)}{E(q^2)}.\label{40gy}$$ Nwxt, fe employ tie quigtuple produgn identhty, witn $q$ replaced bn $q^{10}$ aid $a=-w$ to find that $$f(-q^{13},-q^{17})+qf(-q^7,-x^{23}) = E(q^{10}){\dfrac}{f(-q^2,-q^8)}{f(-q,-q^9)}=E(q^2)G(z).\label{qp1}$$ Vijilarly, from we fune $$\beghn{alhgnea} W(q^2)H(d)&=f(-q^{11},-s^{19})+q^3h(-
^{16})-q^2f(q^{4},q^{26})\bigl)\bigr(f(q^{13},q^{17})-qf(q^7,q^{23})\bigl)\notag\\ &\;\;\;+q\bigr(f(q^8,q^{22})-q^2f(q^2,q^{28})\bigl)\bigr(f(q^{11},q^{19})-q^3f(q,q^{29})\bigl).\label{60pre3}\end{aligned}$$ Recall that the Rogers-Ramanujan functions by G(q) :=\sum_{n=0}^{{\infty}}{\dfrac}{q^{n^2}}{(q;q)_n} \text{and} \qquad H(q) famous identities [@cp pp. $$\label{rridents} G(q) = \qquad \text{and} \qquad H(q) = {\dfrac}{1}{(q^2;q^5)_{{\infty}}(q^3;q^5)_{{\infty}}}.$$ proof makes use of one of Ramanujan’s forty identities for the Rogers-Ramanujan functions, [@watson] $$G(q)G(q^4)-qH(q)H(q^4)={\dfrac}{{\varphi}(q^5)}{E(q^2)}.\label{40gh}$$ Next, we employ the quintuple product identity, with $q$ replaced by and to that = E(q^{10}){\dfrac}{f(-q^2,-q^8)}{f(-q,-q^9)}=E(q^2)G(q).\label{qp1}$$ Similarly, from we find $$\begin{aligned} E(q^2)H(q)&=f(-q^{11},-q^{19})+q^3f(-
^{16})-q^2f(q^{4},q^{26})\bigl)\bigr(f(q^{13},q^{17})-qf(q^7,q^{23})\bigl)\noTag\\ &\;\;\;+q\bigr(f(q^8,Q^{22})-q^2f(q^2,q^{28})\BigL)\biGr(F(q^{11},q^{19})-q^3F(q,q^{29})\bIgl).\label{60pre3}\end{ALignEd}$$ Recall that the Rogers-RAmanuJaN FuncTIoNs are Defined BY $$\lABEl{Gh} G(Q) :=\sUm_{n=0}^{{\InFTy}}{\Dfrac}{Q^{n^2}}{(q;Q)_n} \qquad \Text{and} \qquAd H(Q) :=\sUm_{n=0}^{{\infty}}{\dfraC}{Q^{n(N+1)}}{(q;q)_n}.$$ These fUncTions satisfy The Famous roGerS–ramanUjaN idenTities [@CP pp. 214–215] $$\labEl{rridentS} G(Q) = {\Dfrac}{1}{(q;Q^5)_{{\Infty}}(q^4;q^5)_{{\INFtY}}} \qquAd \text{and} \qquad H(q) = {\dFRaC}{1}{(Q^2;q^5)_{{\infty}}(q^3;q^5)_{{\infty}}}.$$ our proOf MAkES Use Of oNe of RamanuJaN’s forTY identiTIeS FOR thE rogers-RamanujAn functions, NAmeLy [@watsOn] $$g(q)G(Q^4)-QH(q)H(q^4)={\dFrac}{{\vArPHi}(q^5)}{e(q^2)}.\label{40gh}$$ NeXt, we Employ the QuintuPLe produCT identiTy, with $Q$ rePlaCed bY $Q^{10}$ aNd $A=-q$ tO fINd tHAt $$F(-q^{13},-q^{17})+QF(-q^7,-q^{23}) = e(q^{10}){\dfrac}{f(-Q^2,-q^8)}{F(-q,-Q^9)}=E(q^2)G(q).\LabeL{QP1}$$ sImilArlY, froM we fiNd $$\begin{aligneD} E(q^2)h(q)&=f(-q^{11},-Q^{19})+Q^3f(-
^{16})-q^2f(q^{4},q^{26})\ bigl)\bigr (f(q^ {13 },q ^{ 17}) -qf( q^7,q^{23})\bi g l)\n otag\\ &\;\;\;+q\bigr( f(q^8 ,q ^ {22} ) -q ^2f(q ^2,q^{2 8 }) \ b igl )\ bi gr( f( q ^{ 11},q ^{1 9})-q^3 f(q,q^{29} )\b ig l).\label{60 p re 3}\end{ali gne d}$$ Recalltha t theRo ger s -Rama nuj an fu nction s are d efined by $ $ \label { GH} G(q ) := \sum _{n=0}^{{\infty}} { \d f rac}{q^{n^2}}{ (q;q)_ n} \q q u ad\te xt{and} \q qu ad H( q ) :=\su m _{ n = 0 }^{ { \infty}}{\dfr ac}{q^{n(n+ 1 )}} {(q;q) _n }.$ $ Thes e fun ct i ons satisfy th e fa mous Roge rs–Ram a nujan i d entitie s [@cp pp . 2 14–2 1 5] $ $\l ab e l{r r id ent s } G (q) = {\ df ra c}{1} {(q; q ^ 5 ) _{{\ inf ty}} (q^4; q^5)_{{\infty }}} \qq u ad\text {and} \qq ua d H(q ) = {\ dfrac }{ 1}{(q^2;q^5)_{{ \inf ty}}(q^3; q^5 )_ {{\ in fty}} } .$$ Ou r p roo f makes use of one o f R am anujan’s forty ide nt i t ie s for th e Roge r s- Ra m anujan f un cti ons, n amely [@w a ts on] $$G( q)G(q^ 4 )- qH (q)H(q^ 4) ={\dfr ac }{{ \va rphi} ( q^5) }{E(q^ 2)}.\lab el{40 g h}$$ Next, we employ the qu i nt u p le prod uct identity,with $q$repl a ce d b y $q^{ 10}$an d $ a =-q$ to find that $ $f (-q^{1 3},-q ^{17})+qf(-q^ 7,-q^{23}) = E(q^{10} ){\d f ra c }{f(-q^2,-q^8) }{f(- q,-q^9)}=E ( q^2)G(q) .\lab el{qp1}$ $ Similar l y , from w e f ind $$ \be g i n{ aligned} E(q^ 2 ) H(q) &= f(-q^{1 1}, -q^{19} )+q ^3f (-
^{16})-q^2f(q^{4},q^{26})\bigl)\bigr(f(q^{13},q^{17})-qf(q^7,q^{23})\bigl)\notag\\ &\;\;\;+q\bigr(f(q^8,q^{22})-q^2f(q^2,q^{28})\bigl)\bigr(f(q^{11},q^{19})-q^3f(q,q^{29})\bigl).\label{60pre3}\end{aligned}$$ Recall_that the_Rogers-Ramanujan functions are defined_by $$\label{GH} G(q)_:=\sum_{n=0}^{{\infty}}{\dfrac}{q^{n^2}}{(q;q)_n}_\qquad \text{and} \qquad_H(q)_:=\sum_{n=0}^{{\infty}}{\dfrac}{q^{n(n+1)}}{(q;q)_n}.$$ These functions satisfy_the famous Rogers–Ramanujan_identities [@cp pp. 214–215] $$\label{rridents} G(q)_= {\dfrac}{1}{(q;q^5)_{{\infty}}(q^4;q^5)_{{\infty}}} \qquad_\text{and}_\qquad H(q) = {\dfrac}{1}{(q^2;q^5)_{{\infty}}(q^3;q^5)_{{\infty}}}.$$ Our proof makes use of one of Ramanujan’s forty identities for_the_Rogers-Ramanujan functions,_namely_[@watson]_$$G(q)G(q^4)-qH(q)H(q^4)={\dfrac}{{\varphi}(q^5)}{E(q^2)}.\label{40gh}$$ Next, we employ the quintuple_product identity, with $q$ replaced_by $q^{10}$_and $a=-q$ to find that $$f(-q^{13},-q^{17})+qf(-q^7,-q^{23}) = E(q^{10}){\dfrac}{f(-q^2,-q^8)}{f(-q,-q^9)}=E(q^2)G(q).\label{qp1}$$ Similarly,_from_we find $$\begin{aligned} E(q^2)H(q)&=f(-q^{11},-q^{19})+q^3f(-
\quad 0\leq t \leq T,$$ and we denote the continuous version of obtained by linear interpolation by $$\label{pir} \widehat{W}^n_t := \frac{1}{\sqrt{n}} \widehat{Z}^n_{\lfloor nt/T \rfloor}, \quad 0\leq t \leq T.$$ By the central limit theorem; $(W^n, \widehat{W}^n) \Rightarrow (W,W) $ as $n \rightarrow \infty$ on $D([0,T];\mathbb{R}^2)$ ($\Rightarrow$ implies convergence in distribution). i.e., the law $(P_n) $ converges to the law $P_0$ on the Skorohod space $D([0,T];\mathbb{R}^2)$ @Patrick [Theorem $27.1$]. Let $g \in \mathcal{C}([0,T]\times \Omega, { } \sqrt{\mathbf{D}})$ such that $$P = P_0 \circ \left( \underbrace{ \int g(t, W) dW_t}_\text{$M$} \right)^{-1}.$$ Since $g$ is continuous and $\widehat{W}^n_t$ is the interpolated version of, $$\left( W^n, \left( g\left( \lfloor nt/T \rfloor T/n, \widehat{W}^n_t \right) \right)_{t \in [0,T]} \right) \Rightarrow \left(W, (g(t, W_t))_{t \in [0,T]} \right) \text{ as } n \rightarrow \infty \text{ on } D([0,T];\mathbb{R}^2).$$ We introduce martingales with discrete-time integrals, $$\label{dis-mart} M^{n}_k := \sum _{l=1}^{k} g \left( (l-1)T/n, \widehat{W}^n \right) \widehat{W}^n_{lT/n} - \widehat{W}^n_{(l-1)T/n}.$$ In order to construct $M^n$ which is “close” to $M$ and also is such that ${P_n \circ {(M^n)}^{-1} \in \mathcal{Q}_{\mathbf{D}^{\
\quad 0\leq t \leq T,$$ and we denote the continuous version of obtained by analogue interjection by $ $ \label{pir } \widehat{W}^n_t: = \frac{1}{\sqrt{n } } \widehat{Z}^n_{\lfloor nt / T \rfloor }, \quad 0\leq thyroxine \leq T.$$ By the central limit theorem; $ (W^n, \widehat{W}^n) \Rightarrow (W, W) $ as $ north \rightarrow \infty$ on $ D([0,T];\mathbb{R}^2)$ ($ \Rightarrow$ implies convergence in distribution). i.e., the police $ (P_n) $ converges to the police $ P_0 $ on the Skorohod space $ D([0,T];\mathbb{R}^2)$ @Patrick [ Theorem   $ 27.1 $ ]. permit $ g \in \mathcal{C}([0,T]\times \Omega, { } \sqrt{\mathbf{D}})$ such that $ $ P = P_0 \circ \left (\underbrace { \int g(t, W) dW_t}_\text{$M$ } \right)^{-1}.$$ Since $ g$ is continuous and $ \widehat{W}^n_t$ is the interpolate version of, $ $ \left (W^n, \left (g\left (\lfloor nt / T \rfloor T / n, \widehat{W}^n_t \right) \right)_{t \in [ 0,T ] } \right) \Rightarrow \left(W, (g(t, W_t))_{t \in [ 0,T ] } \right) \text { as } n \rightarrow \infty \text { on } D([0,T];\mathbb{R}^2).$$ We introduce dolphin striker with discrete - meter integrals, $ $ \label{dis - mart } M^{n}_k: = \sum _ { l=1}^{k } g \left (( l-1)T / n, \widehat{W}^n \right) \widehat{W}^n_{lT / n } - \widehat{W}^n_{(l-1)T / n}.$$ In order to construct $ M^n$ which is “ close ” to $ M$ and also is such that $ { P_n \circ { (M^n)}^{-1 } \in \mathcal{Q}_{\mathbf{D}^{\
\quwd 0\leq t \leq T,$$ and we denute the continuous vervion or obtaindd by linear interpolation bb $$\lavel{pie} \widehat{W}^n_t := \frac{1}{\sdrt{n}} \widejat{Z}^n_{\lflior it/T \rfloor}, \quad 0\leq t \leq T.$$ By bhe cznvral limit theotem; $(W^n, \widehdt{W}^n) \Rightarrof (D,W) $ as $n \rightarrow \infty$ on $D([0,T];\mathbb{W}^2)$ ($\Rightsrgow$ implies cogverbqnce pn distribution). i.e., the law $(P_n) $ donvergts to the law $P_0$ on the Skorohod space $D([0,T];\mathhb{R}^2)$ @Oatrick [Theorem $27.1$]. Leh $g \in \mathxal{C}([0,E]\rimes \Omega, { } \sqrt{\mathbf{D}})$ such thaj $$P = P_0 \circ \left( \underbrace{ \int e(t, W) bW_t}_\text{$M$} \rithr)^{-1}.$$ Sltce $g$ is coitinuols and $\widehab{E}^n_t$ is the inyerpolated verxioi of, $$\left( W^n, \left( g\left( \lhloor nt/T \rfloor T/n, \ridehat{W}^n_d \xight) \right)_{t \in [0,T]} \rigyt) \Righjarrof \lewr(W, (e(t, S_t))_{v \ih [0,T]} \rihht) \text{ as } h \rightarroq \infty \text{ on } D([0,T];\kaeybb{R}^2).$$ We introsuce mwreingales with discrete-time integrals, $$\latel{sis-mart} M^{n}_k := \sum _{l=1}^{k} t \left( (l-1)T/n, \widehat{W}^n \tight) \widerat{W}^n_{lT/n} - \widehat{W}^n_{(l-1)T/n}.$$ In order to construct $M^n$ wvich ms “clisc” to $N$ wnd also is such that ${P_n \circ {(M^n)}^{-1} \in \mathcal{Q}_{\iztnbn{D}^{\
\quad 0\leq t \leq T,$$ and we continuous of obtained linear interpolation by nt/T \quad 0\leq t T.$$ By the limit theorem; $(W^n, \widehat{W}^n) \Rightarrow (W,W) as $n \rightarrow \infty$ on $D([0,T];\mathbb{R}^2)$ ($\Rightarrow$ implies convergence in distribution). i.e., the $(P_n) $ converges to the law $P_0$ on the Skorohod space $D([0,T];\mathbb{R}^2)$ @Patrick $27.1$]. $g \mathcal{C}([0,T]\times { } \sqrt{\mathbf{D}})$ such that $$P = P_0 \circ \left( \underbrace{ \int g(t, W) dW_t}_\text{$M$} \right)^{-1}.$$ $g$ is continuous and $\widehat{W}^n_t$ is the interpolated of, $$\left( W^n, \left( \lfloor nt/T \rfloor T/n, \widehat{W}^n_t \right)_{t [0,T]} \right) \left(W, W_t))_{t [0,T]} \right) \text{ } n \rightarrow \infty \text{ on } D([0,T];\mathbb{R}^2).$$ We introduce martingales with discrete-time integrals, $$\label{dis-mart} M^{n}_k := _{l=1}^{k} g \widehat{W}^n \right) - In to construct $M^n$ “close” to $M$ and also is \circ {(M^n)}^{-1} \in \mathcal{Q}_{\mathbf{D}^{\
\quad 0\leq t \leq T,$$ and we denote thE continuouS versIon Of oBtAineD by lInear interpolaTIon bY $$\label{pir} \widehat{W}^n_t := \fraC{1}{\sqrt{N}} \wIDehaT{z}^n_{\LflooR nt/T \rflOOr}, \QUAd 0\lEq T \lEq T.$$ by THe CentrAl lImit theOrem; $(W^n, \wideHat{w}^n) \rightarrow (W,W) $ AS $n \Rightarrow \InfTy$ on $D([0,T];\mathbb{r}^2)$ ($\RiGhtarrOw$ ImpLIes coNveRgencE in disTRibutiOn). i.e., the laW $(P_N) $ ConverGEs to the LAW $P_0$ On thE Skorohod space $D([0,T];\mAThBB{R}^2)$ @Patrick [TheorEm $27.1$]. Let $g \In \MAtHCAl{C}([0,t]\tiMes \Omega, { } \sqRt{\MathbF{d}})$ such thAT $$P = p_0 \CIRc \lEFt( \underbrace{ \iNt g(t, W) dW_t}_\texT{$m$} \riGht)^{-1}.$$ SinCe $G$ is COntinuOus anD $\wIDehAt{W}^n_t$ is the iNterPolated veRsion oF, $$\Left( W^n, \lEFt( g\left( \Lfloor Nt/T \RflOor T/N, \WiDeHat{w}^n_T \RigHT) \rIghT)_{T \in [0,t]} \right) \RiGhTaRrow \lEft(W, (G(T, w_T))_{T \in [0,T]} \RigHt) \teXt{ as } n \Rightarrow \infTy \tExt{ oN } d([0,T];\mAthbb{r}^2).$$ We inTrodUcE martIngaleS with DiScrete-time integRals, $$\Label{dis-mArt} m^{n}_K := \suM _{l=1}^{K} g \lefT( (L-1)T/n, \widEhaT{W}^n \Right) \wiDehat{W}^n_{Lt/n} - \wIdEHAT{W}^N_{(l-1)T/n}.$$ In order to constRuCT $m^n$ Which is “cLose” to $m$ AnD aLSo is such ThAt ${P_N \cirC {(m^N)}^{-1} \in \maThcaL{q}_{\mAthbf{D}^{\
\quad 0\leq t \leq T,$$ and we den ote t hecon ti nuou s ve rsion of obtai n ed b y linear interpolation by $ $\ l abel { pi r} \wide h at { W }^n _t : = \ fr a c{ 1}{\s qrt {n}} \w idehat{Z}^ n_{ \l floor nt/T \ r fl oor}, \qua d 0 \leq t \leqT.$ $ By t he ce n trallim it th eorem; $(W^n, \widehat {W } ^n) \R i ghtarro w (W ,W)$ as $n \rightarr o w\ infty$ on $D([ 0,T];\ ma t hb b { R}^ 2)$ ($\Righta rr ow$ i m plies c o nv e r g enc e in distribut ion). i.e., the law $ (P _n) $ conv erges t o th e law $P_0$ onthe Skoro hod sp a ce $D([ 0 ,T];\ma thbb{R }^2 )$@Pat r ic k[Th eo r em$ 27 .1$ ] . L et $g \i n\m athca l{C} ( [ 0 , T]\t ime s \O mega, { } \sqrt{\m ath bf{D } })$ such that $$P = P_0\circ\left (\underbrace{ \i nt g (t, W) dW _t} _\ tex t{ $M$}\ right) ^{- 1}. $$ Sinc e $g$ i s co nt i n u ou s and $\widehat{W} ^n _ t $is the i nterpo l at ed versionof , $ $\le f t ( W^n , \l e ft ( g\left ( \lfl o or n t/T \rf lo or T/n ,\wi deh at{W} ^ n_t\right ) \right )_{t\ in [0,T]} \rig h t) \Rightarro w \ l e ft ( W, ( g(t , W_t))_{t\in[ 0,T] } \r i gh t)\ text{ as } n \r i ghtarrow \infty \te xt { on } D([0 ,T];\mathbb{R }^2).$$ We i n troducemart i ng a les with discr ete-t ime integr a ls, $$\l abel{ dis-mart } M^{ n } _k := \s um_{l =1} ^{k } g\left( (l-1)T / n , \w id ehat{W} ^n\right) \w ide hat {W} ^n _{lT/n} - \wideha t{ W} ^n _{ (l- 1)T/n } .$$ In o rd erto co nstru c t $M^n $ whi ch i s“c l ose ” to $M $ a n d als ois suc h t ha t ${P _n \ c irc {(M^n) }^{-1} \ in\ math ca l{ Q}_{\ma thbf{D}^{\
\quad_0\leq t_\leq T,$$ and we denote_the continuous_version_of obtained_by_linear interpolation by_$$\label{pir} _ \widehat{W}^n_t := \frac{1}{\sqrt{n}}_\widehat{Z}^n_{\lfloor nt/T \rfloor},_\quad_0\leq t \leq T.$$ By the central limit theorem; $(W^n, \widehat{W}^n) \Rightarrow (W,W) $_as_$n \rightarrow_\infty$_on_$D([0,T];\mathbb{R}^2)$ ($\Rightarrow$ implies convergence in_distribution). i.e., the law $(P_n)_$ converges_to the law $P_0$ on the Skorohod space_$D([0,T];\mathbb{R}^2)$_@Patrick [Theorem $27.1$]. Let_$g \in \mathcal{C}([0,T]\times \Omega, { } \sqrt{\mathbf{D}})$ such that_$$P = P_0 \circ \left( \underbrace{_\int g(t, W)_dW_t}_\text{$M$}_\right)^{-1}.$$_Since $g$ is continuous_and $\widehat{W}^n_t$ is the interpolated version_of, $$\left( W^n, \left( g\left( \lfloor_nt/T \rfloor T/n, \widehat{W}^n_t \right) \right)_{t \in_[0,T]} \right) \Rightarrow \left(W, (g(t, W_t))_{t_\in [0,T]} \right) \text{ as_} n_\rightarrow \infty \text{ on }_D([0,T];\mathbb{R}^2).$$ We introduce_martingales with_discrete-time integrals, $$\label{dis-mart} _ M^{n}_k := \sum__{l=1}^{k} g \left(_(l-1)T/n, \widehat{W}^n \right) \widehat{W}^n_{lT/n} - \widehat{W}^n_{(l-1)T/n}.$$_In_order to construct_$M^n$_which_is “close”_to $M$ and_also_is such_that_${P_n \circ {(M^n)}^{-1} \in \mathcal{Q}_{\mathbf{D}^{\
\subseteq \mathbb{Z}_{2}$ ------------------------------------------------------------------- ------- ------------------- ---------------------------- ----------------------------------------- ---------------------------------------- --------------------------------------------------- -------------------------------------------------- --------------------------------------------------------------------------------------------------------------- ------- ---------------------- ---------------------------------- ------------------------------------------- ---------------------------------------------- ------------------------------------------------------- ---------------------------------------------------------------- $d=0$ $d=1$ $d=1$ $d=2$ $d=2$ $d=2$ Symmetry Class $d_{\parallel}=0$ $d_{\parallel}=0$ $d_{\parallel}=1$ $d_{\parallel}=0$ $d_{\parallel}=1$ $d_{\parallel}=2$ $\text{\ensuremath{\hat{\mathcal{U}}_{0,+}^{+}}},\text{\ensuremath{\hat{\mathcal{U}}_{T/2,+}^{+}}}$ AI $\mathbb{Z}^{2}$ $0 \subseteq 0$ $\mathbb{Z} \subseteq \mathbb{Z}$ $0 \subseteq 0 \subseteq 0$ $0 \subseteq 0 \subseteq 0$ $\mathbb{Z} \subseteq \mathbb{Z} \subseteq \mathbb{Z}$ $\hat{\mathcal{U}}_{0,++}^{+},\hat{\mathcal{U}}_{T/2,+-}^{+}$ BDI $\mathbb{Z}_{2}^{2}$ $0 \subseteq $\mathbb{Z}_{2} \subseteq \mathbb{Z}_2$ $0 \subseteq 0 \subseteq 0$ $0 \subseteq \mathbb{Z} \subseteq \mathbb{Z}$ $0 \subseteq 0 \subseteq 0$ \mathbb{Z}^2$ $\ensuremath{\hat{\mathcal{U}}_{0,+}^{+}},\text{\ensuremath{\hat{\mathcal{U}}_{T/2,-}^{+}}}$ D $\mathbb{Z}_{2}^{2}$ $0 \subseteq \mathbb{Z}_{2}^{2}$ $\mathbb{Z}_{2} \subseteq \mathbb{Z}_{2}$ $0 \subseteq 0 \subseteq \mathbb{Z}^{2}$ $0 \subseteq \mathbb{Z}_{2} \subseteq \mathbb{Z}_{2}$ $0 \subseteq 0 \subseteq \mathbb{Z}$ $\hat{\mathcal{U}}_{0
\subseteq \mathbb{Z}_{2}$ ------------------------------------------------------------------- ------- ------------------- ---------------------------- ----------------------------------------- ---------------------------------------- --------------------------------------------------- -------------------------------------------------- --------------------------------------------------------------------------------------------------------------- ------- ---------------------- ---------------------------------- ------------------------------------------- ---------------------------------------------- ------------------------------------------------------- ---------------------------------------------------------------- $ d=0 $ $ d=1 $ $ d=1 $ $ d=2 $ $ d=2 $ $ d=2 $ Symmetry Class $ d_{\parallel}=0 $ $ d_{\parallel}=0 $ $ d_{\parallel}=1 $ $ d_{\parallel}=0 $ $ d_{\parallel}=1 $ $ d_{\parallel}=2 $ $ \text{\ensuremath{\hat{\mathcal{U}}_{0,+}^{+}}},\text{\ensuremath{\hat{\mathcal{U}}_{T/2,+}^{+}}}$ AI $ \mathbb{Z}^{2}$ $ 0 \subseteq 0 $ $ \mathbb{Z } \subseteq \mathbb{Z}$ $ 0 \subseteq 0 \subseteq 0 $ $ 0 \subseteq 0 \subseteq 0 $ $ \mathbb{Z } \subseteq \mathbb{Z } \subseteq \mathbb{Z}$ $ \hat{\mathcal{U}}_{0,++}^{+},\hat{\mathcal{U}}_{T/2,+-}^{+}$ BDI $ \mathbb{Z}_{2}^{2}$ $ 0 \subseteq $ \mathbb{Z}_{2 } \subseteq \mathbb{Z}_2 $ $ 0 \subseteq 0 \subseteq 0 $ $ 0 \subseteq \mathbb{Z } \subseteq \mathbb{Z}$ $ 0 \subseteq 0 \subseteq 0 $ \mathbb{Z}^2 $ $ \ensuremath{\hat{\mathcal{U}}_{0,+}^{+}},\text{\ensuremath{\hat{\mathcal{U}}_{T/2,-}^{+}}}$ D $ \mathbb{Z}_{2}^{2}$ $ 0 \subseteq \mathbb{Z}_{2}^{2}$ $ \mathbb{Z}_{2 } \subseteq \mathbb{Z}_{2}$ $ 0 \subseteq 0 \subseteq \mathbb{Z}^{2}$ $ 0 \subseteq \mathbb{Z}_{2 } \subseteq \mathbb{Z}_{2}$ $ 0 \subseteq 0 \subseteq \mathbb{Z}$ $ \hat{\mathcal{U}}_{0
\subseteq \larhbb{Z}_{2}$ ------------------------------------------------------------------- ------- ------------------- ---------------------------- ----------------------------------------- ---------------------------------------- --------------------------------------------------- -------------------------------------------------- --------------------------------------------------------------------------------------------------------------- ------- ---------------------- ---------------------------------- ------------------------------------------- ---------------------------------------------- ------------------------------------------------------- ---------------------------------------------------------------- $d=0$ $d=1$ $d=1$ $d=2$ $d=2$ $d=2$ Symmetrn Rlass $d_{\parallel}=0$ $d_{\'arallel}=0$ $d_{\[arauoel}=1$ $d_{\parallel}=0$ $s_{\paraljej}=1$ $d_{\parallel}=2$ $\text{\wnsuremath{\hat{\mathcal{U}}_{0,+}^{+}}},\jext{\ensureiath{\hat{\mathcal{U}}_{T/2,+}^{+}}}$ AI $\mathbb{Z}^{2}$ $0 \rubwebeq 0$ $\mathbb{Z} \subseteq \mathbb{Z}$ $0 \subseteq 0 \smbseteq 0$ $0 \sucseteq 0 \shbseteq 0$ $\marhbb{Z} \subfetea \mathbb{Z} \subseteq \mathbb{Z}$ $\hat{\machcal{U}}_{0,++}^{+},\hat{\matkcal{U}}_{T/2,+-}^{+}$ YDI $\mathbb{Z}_{2}^{2}$ $0 \subseteq $\mathbb{E}_{2} \subseteq \mathbb{Z}_2$ $0 \xubseted 0 \sobseteq 0$ $0 \subseteq \mathbb{Z} \subsgteq \mdthbb{Z}$ $0 \subseteq 0 \subseteq 0$ \mathbb{Z}^2$ $\enwuremath{\rat{\mathcal{U}}_{0,+}^{+}},\tect{\ensuremath{\hat{\matjcal{U}}_{V/2,-}^{+}}}$ D $\mathbb{Z}_{2}^{2}$ $0 \subseteq \matgbb{Z}_{2}^{2}$ $\mahhbn{Z}_{2} \subseteq \mwthbn{Z}_{2}$ $0 \subsetzq 0 \subseteq \mathbb{Z}^{2}$ $0 \subseteq \mathbu{Z}_{2} \subseteq \mathbb{Z}_{2}$ $0 \subsetgq 0 \subseteq \methbb{Z}$ $\har{\mathcal{U}}_{0
\subseteq \mathbb{Z}_{2}$ ------------------------------------------------------------------- ------- ------------------- ---------------------------- ----------------------------------------- -------------------------------------------------- ------- ---------------------- ------------------------------------------- ---------------------------------------------- ------------------------------------------------------- $d=2$ Symmetry Class $d_{\parallel}=0$ $d_{\parallel}=1$ $d_{\parallel}=0$ $d_{\parallel}=1$ $\text{\ensuremath{\hat{\mathcal{U}}_{0,+}^{+}}},\text{\ensuremath{\hat{\mathcal{U}}_{T/2,+}^{+}}}$ AI $\mathbb{Z}^{2}$ $0 \subseteq 0$ \subseteq \mathbb{Z}$ $0 \subseteq 0 \subseteq 0$ $0 \subseteq 0 \subseteq 0$ $\mathbb{Z} \mathbb{Z} \subseteq \mathbb{Z}$ $\hat{\mathcal{U}}_{0,++}^{+},\hat{\mathcal{U}}_{T/2,+-}^{+}$ BDI $\mathbb{Z}_{2}^{2}$ $0 \subseteq $\mathbb{Z}_{2} \subseteq \mathbb{Z}_2$ $0 \subseteq \subseteq $0 \mathbb{Z} \mathbb{Z}$ $0 \subseteq 0 \subseteq 0$ \mathbb{Z}^2$ $\ensuremath{\hat{\mathcal{U}}_{0,+}^{+}},\text{\ensuremath{\hat{\mathcal{U}}_{T/2,-}^{+}}}$ D $\mathbb{Z}_{2}^{2}$ $0 \subseteq \mathbb{Z}_{2}^{2}$ $\mathbb{Z}_{2} \subseteq \mathbb{Z}_{2}$ \subseteq 0 \subseteq \mathbb{Z}^{2}$ $0 \subseteq \mathbb{Z}_{2} \subseteq $0 \subseteq 0 \subseteq $\hat{\mathcal{U}}_{0
\subseteq \mathbb{Z}_{2}$ ------------------------------------------------------------------- ------- ------------------- ---------------------------- ----------------------------------------- ---------------------------------------- --------------------------------------------------- -------------------------------------------------- --------------------------------------------------------------------------------------------------------------- ------- ---------------------- ---------------------------------- ------------------------------------------- ---------------------------------------------- ------------------------------------------------------- ---------------------------------------------------------------- $d=0$ $d=1$ $d=1$ $d=2$ $d=2$ $d=2$ SymmEtry Class $d_{\ParalLel}=0$ $D_{\paRaLlel}=0$ $D_{\parAllel}=1$ $d_{\parallel}=0$ $D_{\ParaLlel}=1$ $d_{\parallel}=2$ $\text{\ensurEmath{\HaT{\MathCAl{u}}_{0,+}^{+}}},\text{\EnsuremATh{\HAT{\maThCaL{U}}_{T/2,+}^{+}}}$ aI $\MAtHbb{Z}^{2}$ $0 \sUbsEteq 0$ $\matHbb{Z} \subsetEq \mAtHbb{Z}$ $0 \subseteq 0 \SUbSeteq 0$ $0 \subseTeq 0 \Subseteq 0$ $\mathBb{Z} \SubsetEq \MatHBb{Z} \suBseTeq \maThbb{Z}$ $\hAT{\mathcAl{U}}_{0,++}^{+},\hat{\matHcAL{U}}_{T/2,+-}^{+}$ BDI $\MAthbb{Z}_{2}^{2}$ $0 \sUBSeTeq $\mAthbb{Z}_{2} \subseteq \matHBb{z}_2$ $0 \Subseteq 0 \subsetEq 0$ $0 \subsEtEQ \mATHbb{z} \suBseteq \mathBb{z}$ $0 \subsETeq 0 \subsETeQ 0$ \MAThbB{z}^2$ $\ensuremath{\haT{\mathcal{U}}_{0,+}^{+}},\teXT{\enSuremaTh{\Hat{\MAthcal{u}}_{T/2,-}^{+}}}$ D $\maThBB{Z}_{2}^{2}$ $0 \sUbseteq \mathBb{Z}_{2}^{2}$ $\mAthbb{Z}_{2} \subSeteq \mAThbb{Z}_{2}$ $0 \suBSeteq 0 \suBseteq \MatHbb{z}^{2}$ $0 \subSEtEq \MatHbB{z}_{2} \suBSeTeq \MAthBb{Z}_{2}$ $0 \subseTeQ 0 \sUbsetEq \maTHBB{z}$ $\hat{\MatHcal{u}}_{0
\subseteq \mathb b{Z}_ {2 } $ - -- ----- ------- - -- - - --- -- -- --- -- - -- ----- --- ------- ---------- --- -- -- ------- - - -- ---------- --- -- --------- --- ------ -- --- - ------- ----- ------ - ------ --------- -- - ------ - - ----- - - -- ---- ----------------- - -- - ------ ------- ------ -- - -- - - --- --- ---------- -- ----- - ------- -- - - - --- - ------------- ----------- - --- ------ -- --- - - - ----- -- - --- ----------- ---- --------- ------ - ------- - ------- ------ --- --- ---- - -- -- --- -- - --- - -- --- - --- -------- -- -- - --- ---- - - - ---- --- ---- ----- --- --------- --- ---- - --- ----- ----- ---- - ----- ------ ----- -- --------------- ---- ----- --- --- -- --- -- ----- - ------ --- --- ------- ------- - -- -- - - - -- ------------------ -- - - -- -------- ------ - -- -- - -- ----- -- --- ---- - - ----- ---- - -- -------- ------ - -- -- ------- -- ------ -- $d=0$ $d=1$ $d=1$ $d=2$ $d = 2 $ $d =2 $ Sy mmetry C l ass $ d_{\par allel}=0$ $d_{ \p ara llel}= 0 $ $d_{\par a llel}=1 $ $ d_{ \ parall el}=0$ $ d _ {\p arall el }=1$ $d _{\p a r al lel } =2 $ $ \text{\ensurema th{\hat{\m at h ca l{U}}_{0,+ } ^{+ }} },\text {\ensur emath { \hat{\m athcal{U} }_{T/2,+} ^{ +}}} $ AI $\ma thbb{Z}^{ 2 }$ $ 0 \sub se teq 0$ $ \mathb b{ Z} \su bsete q\mathbb{ Z}$ $0 \s ubsete q 0 \ sub seteq 0$ $0 \su bse teq 0 \s ubs e teq 0 $ $\mathbb{ Z }\su b s et eq \mathbb{ Z } \su bsete q \ m athbb{ Z}$ $\ha t{\m a t hca l{U } }_{0 ,+ +}^{+},\hat{\m ath ca l { U}}_{T/2 ,+ -}^{+}$ BDI $\ma t h bb { Z}_{2} ^{2} $ $ 0\ subsete q $\ mathbb {Z}_{2 } \su b s eteq \mathbb{Z}_ 2$ $0 \sub se teq 0 \ s ubse teq 0$ $0 \s ubse teq \ mathbb{ Z} \subs ete q\mathbb{Z} $ $0\su bseteq 0 \ s u bsete q 0$ \m athbb{Z }^2 $ $\ensuremath{\hat{\ math cal{U } }_{ 0,+}^ {+}}, \t ext{\ens u rema th{\hat{ \ mathcal{U}}_{ T/2,-}^{+}}}$ D $\mathbb {Z} _ {2}^{ 2}$ $ 0 \s ubs eteq \ math b b{Z} _ {2 }^{2 }$ $ \mat hb b{ Z}_{2} \s ub seteq \ m athb b {Z} _{2} $ $ 0\ su bseteq0 \subse teq \ ma thbb{Z}^{ 2 }$ $0 \ su bseteq \mathb b{Z}_{2} \su bs eteq \math b b{Z }_{ 2 } $ $0 \s ubs eteq 0 \su bs ete q \m athbb {Z}$ $ \hat{\ma thcal{ U}}_ { 0
_ _ _ __ __ _ _ _ __ __ ___ _ _ _ __ _ _ \subseteq \mathbb{Z}_{2}$ -------------------------------------------------------------------_------- ------------------- ----------------------------_-----------------------------------------_----------------------------------------_--------------------------------------------------- -------------------------------------------------- ---------------------------------------------------------------------------------------------------------------_------- ---------------------- ---------------------------------- ------------------------------------------- ---------------------------------------------- -------------------------------------------------------_---------------------------------------------------------------- _ _ _ _ _ _ _ _ _ _ _ __ ___ _ __ __ __ _ __ _ _ _ _ _ _ _ _ _$d=0$_ __ _ _ _ _ $d=1$ __ _ _ ___ _ _ $d=1$_ __ __ ___ _ _ $d=2$ _ _ ___ _ _ _ _ _ $d=2$ _ _ _ __ ____ _ _ _ _ _ $d=2$ _ __ _ _ __ _ _ _ Symmetry __ _ _ _ _ _ _ _ _ __ ____ _ _ Class_ _$d_{\parallel}=0$_ _ _ $d_{\parallel}=0$ _ _ $d_{\parallel}=1$ _ _ _ __ _ _ $d_{\parallel}=0$ __ ___ __ _____ $d_{\parallel}=1$ _ __ _ __ _ _ __ ___$d_{\parallel}=2$ _ _ $\text{\ensuremath{\hat{\mathcal{U}}_{0,+}^{+}}},\text{\ensuremath{\hat{\mathcal{U}}_{T/2,+}^{+}}}$ AI __ $\mathbb{Z}^{2}$ _ __ _$0 \subseteq 0$ _ _ _ $\mathbb{Z} \subseteq \mathbb{Z}$ $0 \subseteq 0_\subseteq 0$ _ _ __ _ _ $0 \subseteq 0_\subseteq 0$ _ $\mathbb{Z} \subseteq \mathbb{Z} \subseteq_\mathbb{Z}$ _ _ _ _ $\hat{\mathcal{U}}_{0,++}^{+},\hat{\mathcal{U}}_{T/2,+-}^{+}$ __ _ _ BDI_ $\mathbb{Z}_{2}^{2}$__ _ _ _ $0 \subseteq_ _ _ $\mathbb{Z}_{2} \subseteq \mathbb{Z}_2$ _ $0 \subseteq 0 \subseteq 0$ _ __$0 \subseteq \mathbb{Z} \subseteq \mathbb{Z}$ _ _ _ $0 \subseteq 0_\subseteq_0$ _ __ _ _ _ _ _ _ _ __ _ __ _ __ _ _ __ _ __ _ __ _ _ _ __ _ _ _ _ _ _ _ __ _ _ __ \mathbb{Z}^2$ _ _ _ _ _ _ ___ _ _ _ __ _ _ _ __ __ _ _ _ _ ___ _ __ _ _ _ ___ __ _ __ _$\ensuremath{\hat{\mathcal{U}}_{0,+}^{+}},\text{\ensuremath{\hat{\mathcal{U}}_{T/2,-}^{+}}}$ D $\mathbb{Z}_{2}^{2}$__ $0 \subseteq_\mathbb{Z}_{2}^{2}$__ $\mathbb{Z}_{2}_\subseteq \mathbb{Z}_{2}$ _ $0_\subseteq 0_\subseteq \mathbb{Z}^{2}$ _ $0 \subseteq_\mathbb{Z}_{2} \subseteq \mathbb{Z}_{2}$ $0 \subseteq_0 \subseteq \mathbb{Z}$ _ _ _ __ _ _ _$\hat{\mathcal{U}}_{0
consider the corresponding admissible graph $\Gamma$ of type $(n,k,l)$, then an edge $e$ of $\Gamma$ determines either a projection $\pi_e:\mathcal C_{n,k,l}^+\to\mathcal C_{2,0,0}^+$ or $\pi_e:\mathcal C_{n,k,l}^+\to\mathcal C_{1,0,0}^+$. We now consider the vector space $X=\mathbb K^d$ and two linear (or affine) subspaces $U_i$, $i=1,2$, for which we assume there is a direct sum decomposition $$\label{eq-orth-split} X=(U_1\cap U_2)\overset{\perp}\oplus (U_1^\perp\cap U_2)\overset{\perp}\oplus (U_1\cap U_2^\perp)\overset{\perp}\oplus (U_1+U_2)^\perp,$$ w.r.t. a chosen inner product over $X$. Clearly, we have $$U_1=(U_1\cap U_2)\overset{\perp}\oplus (U_1\cap U_2^\perp),\ U_2=(U_1\cap U_2)\overset{\perp}\oplus (U_1^\perp\cap U_2).$$ We choose linear coordinates $\{x_i\}$ on $X$ which are adapted to the orthogonal decomposition , [*i.e.*]{} there are two non-disjoint subsets $I_i$, $i=1,2$, of $[d]$, such that $$[d]=\left(I_1\cap I_2\right)\sqcup\left(I_1\cap I_2^c\right)\sqcup\left(I_1^c\cap I_2\right)\sqcup\left(I_1^c\cap I_2^c\right),$$ w.r.t. which $\{x_i\}$ is a set of linear coordinates on $U_1\cap U_2$, $U_1\cap U_2^\perp$, $U_1^\perp\cap U_2$ or $(U_1+U_2)^\perp$, if the index $i$ belongs to $I_1\cap I_2$, $I_1\cap I_2^c$, $I_
consider the corresponding admissible graph $ \Gamma$ of type $ (n, k, l)$, then an edge $ e$ of $ \Gamma$ determine either a expulsion $ \pi_e:\mathcal C_{n, k, l}^+\to\mathcal C_{2,0,0}^+$ or $ \pi_e:\mathcal C_{n, k, l}^+\to\mathcal C_{1,0,0}^+$. We now consider the vector space $ X=\mathbb K^d$ and two analogue (or affine) subspaces $ U_i$, $ i=1,2 $, for which we assume there is a direct kernel decomposition $ $ \label{eq - orth - split } X=(U_1\cap U_2)\overset{\perp}\oplus (U_1^\perp\cap U_2)\overset{\perp}\oplus (U_1\cap U_2^\perp)\overset{\perp}\oplus (U_1+U_2)^\perp,$$ w.r.t.   a choose inner product over $ X$. distinctly, we have $ $ U_1=(U_1\cap U_2)\overset{\perp}\oplus (U_1\cap U_2^\perp),\ U_2=(U_1\cap U_2)\overset{\perp}\oplus (U_1^\perp\cap U_2).$$ We choose linear coordinates $ \{x_i\}$ on $ X$ which are adjust to the extraneous decomposition  , [ * i.e. * ] { } there are two non - disjoint subset $ I_i$, $ i=1,2 $, of $ [ d]$, such that $ $ [ d]=\left(I_1\cap I_2\right)\sqcup\left(I_1\cap I_2^c\right)\sqcup\left(I_1^c\cap I_2\right)\sqcup\left(I_1^c\cap I_2^c\right),$$ w.r.t.   which $ \{x_i\}$ is a set of analogue coordinate on $ U_1\cap U_2 $, $ U_1\cap U_2^\perp$, $ U_1^\perp\cap U_2 $ or $ (U_1+U_2)^\perp$, if the index $ i$ belongs to $ I_1\cap I_2 $, $ I_1\cap I_2^c$, $ I _
cojsider the corresponding admissible grakh $\Gamma$ of tyle $(n,k,l)$, tfen an edge $e$ of $\Gamma$ deterlibes euther a projection $\pi_e:\oathcal C_{j,k,l}^+\to\matycal X_{2,0,0}^+$ or $\pi_e:\mavgcal C_{n,k,l}^+\to\mafmcal E_{1,0,0}^+$. Wx now consider jhe vector s[ace $X=\mathbb K^g$ xnb two linear (or affine) subspaces $U_i$, $y=1,2$, for wnifh we assume trere ys a direct sum decomposition $$\label{eq-odth-spliu} X=(U_1\cap U_2)\overset{\perl}\oplus (U_1^\perp\cap U_2)\overset{\pegp}\oppus (U_1\cap U_2^\perp)\overdet{\perp}\opluw (U_1+U_2)^\[wrp,$$ w.r.t. a choren inner krmduct over $X$. Clearly, we have $$U_1=(U_1\cap U_2)\oversdt{\per'}\oplus (U_1\cap U_2^\pero),\ U_2=(U_1\cap U_2)\ovecset{\pegp}\oplus (U_1^\perp\gsp U_2).$$ We choose linear coordikates $\{x_i\}$ on $X$ which are adaptxd to the orthogonal decomposhtnon , [*i.e.*]{} there are two bob-disjmint subrwts $I_i$, $i=1,2$, or $[d]$, sufh vhat $$[d]=\left(I_1\dap I_2\right)\swcup\left(I_1\cap I_2^c\righu)\sqsll\left(I_1^c\cap I_2\dight)\szctp\left(I_1^c\cap I_2^c\right),$$ w.r.t. which $\{x_i\}$ is a seu of minear coordinates on $U_1\xap U_2$, $U_1\cap U_2^\perp$, $U_1^\perk\cap U_2$ or $(T_1+U_2)^\perp$, if the index $i$ belongs to $I_1\cap I_2$, $I_1\cap I_2^c$, $I_
consider the corresponding admissible graph $\Gamma$ of then edge $e$ $\Gamma$ determines either or C_{n,k,l}^+\to\mathcal C_{1,0,0}^+$. We consider the vector $X=\mathbb K^d$ and two linear (or subspaces $U_i$, $i=1,2$, for which we assume there is a direct sum decomposition X=(U_1\cap U_2)\overset{\perp}\oplus (U_1^\perp\cap U_2)\overset{\perp}\oplus (U_1\cap U_2^\perp)\overset{\perp}\oplus (U_1+U_2)^\perp,$$ w.r.t. a chosen inner product over Clearly, have U_2)\overset{\perp}\oplus U_2^\perp),\ U_2=(U_1\cap U_2)\overset{\perp}\oplus (U_1^\perp\cap U_2).$$ We choose linear coordinates $\{x_i\}$ on $X$ which are adapted to orthogonal decomposition , [*i.e.*]{} there are two non-disjoint $I_i$, $i=1,2$, of $[d]$, that $$[d]=\left(I_1\cap I_2\right)\sqcup\left(I_1\cap I_2^c\right)\sqcup\left(I_1^c\cap I_2\right)\sqcup\left(I_1^c\cap w.r.t. $\{x_i\}$ is set linear on $U_1\cap U_2$, U_2^\perp$, $U_1^\perp\cap U_2$ or $(U_1+U_2)^\perp$, if the index $i$ belongs to $I_1\cap I_2$, $I_1\cap I_2^c$, $I_
consider the corresponding aDmissible gRaph $\GAmmA$ of TyPe $(n,k,L)$, theN an edge $e$ of $\GammA$ DeteRmines either a projectioN $\pi_e:\mAtHCal C_{N,K,l}^+\To\matHcal C_{2,0,0}^+$ or $\PI_e:\MAThcAl c_{n,K,l}^+\tO\mAThCal C_{1,0,0}^+$. WE noW considEr the vectoR spAcE $X=\mathbb K^d$ anD TwO linear (or aFfiNe) subspaces $U_I$, $i=1,2$, fOr whicH wE asSUme thEre Is a diRect suM DecompOsition $$\laBeL{Eq-orth-SPlit} X=(U_1\cAP u_2)\oVersEt{\perp}\oplus (U_1^\perp\cAP U_2)\OVerset{\perp}\opluS (U_1\cap U_2^\PeRP)\oVERseT{\peRp}\oplus (U_1+U_2)^\pErP,$$ w.r.t. a CHosen inNEr PRODucT Over $X$. Clearly, wE have $$U_1=(U_1\cap U_2)\OVerSet{\perP}\oPluS (u_1\cap U_2^\pErp),\ U_2=(U_1\CaP u_2)\ovErset{\perp}\opLus (U_1^\Perp\cap U_2).$$ WE choosE Linear cOOrdinatEs $\{x_i\}$ on $x$ whIch Are aDApTeD to ThE OrtHOgOnaL DecOmpositiOn , [*I.e.*]{} There Are tWO NON-disJoiNt suBsets $i_i$, $i=1,2$, of $[d]$, such thaT $$[d]=\lEft(I_1\CAp I_2\Right)\Sqcup\Left(i_1\cAp I_2^c\rIght)\sqCup\leFt(i_1^c\cap I_2\right)\sqcuP\lefT(I_1^c\cap I_2^c\rIghT),$$ w.R.t. wHiCh $\{x_i\}$ iS A set of LinEar CoordinAtes on $U_1\CAp U_2$, $u_1\cAP u_2^\PeRp$, $U_1^\perp\cap U_2$ or $(U_1+U_2)^\perP$, iF THe Index $i$ beLongs tO $i_1\cAp i_2$, $i_1\cap I_2^c$, $I_
consider the correspondin g admissib le gr aph $\ Ga mma$ oftype $(n,k,l)$ , the n an edge $e$ of $\Gam ma$ d et e rmin e seithe r a pro j ec t i on$\ pi _e: \m a th cal C _{n ,k,l}^+ \to\mathca l C _{ 2,0,0}^+$ or $\ pi_e:\math cal C_{n,k,l}^+ \to \mathc al C_ { 1,0,0 }^+ $. W e nowc onside r the vec to r space $X=\mat h b bK^d$ and two linear ( o ra ffine) subspac es $U_ i$ , $ i = 1,2 $,for whichwe assu m e there is a dir e ct sum decomp osition $$\ l abe l{eq-o rt h-s p lit} X =(U_1 \c a p U _2)\overset {\pe rp}\oplus (U_1^ \ perp\ca p U_2)\o verset {\p erp }\op l us ( U_1 \c a p U _ 2^ \pe r p)\ overset{ \p er p}\op lus( U _ 1 +U_2 )^\ perp ,$$ w .r.t. a chose n i nner pro ductover$X$. C learl y, wehave$$ U_1=(U_1\cap U_ 2)\o verset{\p erp }\ opl us (U_1 \ cap U_ 2^\ per p),\ U_ 2=(U_1\ c apU_ 2 ) \ ov erset{\perp}\oplus ( U _ 1^ \perp\ca p U_2) . $$ W e choose l ine ar c o o rdina tes$ \{ x_i\}$ o n $X$w hi ch are ad ap ted to t heort hogon a l de compos ition ,[*i.e . *]{} there are two non-disjo i nt s ub s ets$I_ i$, $i=1,2$ , of $[d] $, s u ch th a t $$[ d]=\l ef t (I _ 1\cap I_2\right)\sq cu p\left (I_1\ cap I_2^c\rig ht)\sqcup\ l e f t(I_1^c\ capI _2 \ right)\sqcup\l eft(I _1^c\cap I _ 2^c\righ t),$$ w.r.t.which $\{ x _ i\}$ isa s etoflin e a rcoordinates o n $U_1 \c ap U_2$ , $ U_1\cap U_ 2^\ per p$, $ U_1^\perp \cap U_2 $or $ (U _1+ U_2)^ \ perp$, i fthe i nde x $i$ belong s to$I_1 \c ap I_2 $, $I_1 \ ca p I_2^ c$ ,$I_
consider_the corresponding_admissible graph $\Gamma$ of_type $(n,k,l)$,_then_an edge_$e$_of $\Gamma$ determines_either a projection_$\pi_e:\mathcal C_{n,k,l}^+\to\mathcal C_{2,0,0}^+$ or_$\pi_e:\mathcal C_{n,k,l}^+\to\mathcal C_{1,0,0}^+$. We_now_consider the vector space $X=\mathbb K^d$ and two linear (or affine) subspaces $U_i$, $i=1,2$,_for_which we_assume_there_is a direct sum decomposition_$$\label{eq-orth-split} X=(U_1\cap U_2)\overset{\perp}\oplus (U_1^\perp\cap U_2)\overset{\perp}\oplus (U_1\cap_U_2^\perp)\overset{\perp}\oplus (U_1+U_2)^\perp,$$_w.r.t. a chosen inner product over $X$. Clearly, we_have_$$U_1=(U_1\cap U_2)\overset{\perp}\oplus (U_1\cap_U_2^\perp),\ U_2=(U_1\cap U_2)\overset{\perp}\oplus (U_1^\perp\cap U_2).$$ We choose linear coordinates $\{x_i\}$_on $X$ which are adapted to_the orthogonal decomposition ,_[*i.e.*]{}_there_are two non-disjoint subsets_$I_i$, $i=1,2$, of $[d]$, such that_$$[d]=\left(I_1\cap I_2\right)\sqcup\left(I_1\cap I_2^c\right)\sqcup\left(I_1^c\cap I_2\right)\sqcup\left(I_1^c\cap I_2^c\right),$$ w.r.t. which_$\{x_i\}$ is a set of linear coordinates_on $U_1\cap U_2$, $U_1\cap U_2^\perp$, $U_1^\perp\cap_U_2$ or $(U_1+U_2)^\perp$, if the_index $i$_belongs to $I_1\cap I_2$, $I_1\cap_I_2^c$, $I_
term on the right-hand side of Eq. [(\[eq:s\])]{} vanishes. It is furthermore reasonable to suppose that an environmental heat bath in thermal equilibrium leads to a vanishing expectation value of the stochastic force $\hat{\boldsymbol{\xi}}(t)$, Eq. [(\[eq:stochasticforce\])]{}. Consequently, presuming a vanishing shift $\mathbf{s}(t)$ seems appropriate for a typical measurement configuration. We will confirm this presumption further below for a specifically chosen initial state. Note that the linearity of Eqs. [(\[eq:inferredsystem\])]{} and [(\[eq:inferredobservables\])]{}, which is essential for the inference process, Eq. [(\[eq:inferredexpecation\])]{}, is a direct result of the bilinear structure of the Hamiltonian, Eq. [(\[eq:H\])]{}. We assume here the non-pathological case that the interaction Hamiltonian $\hat{H}_{\mathrm{int}}(t)$, Eq. [(\[eq:H:int\])]{}, and the measurement vector $\hat{\mathbf{w}}(t)$ are chosen in such a way that inferring system observables from the pointers is possible in the first place, which is defined by the existence of the coefficient matrix $\mathbf{A}(t)$, Eq. [(\[eq:inferredobservables\])]{}. In other words, we require a sufficient information transfer from the system to the pointers. For example, for the classic Arthurs and Kelly model [@arthurs1965] with the interaction Hamiltonian mentioned in Sec. \[sec:generalopenpointer-basedsimultaneousmeasurements:hamiltonian\] and no environmental heat bath, measuring both pointer positions leads to an existing matrix $\mathbf{A}(t)$ for $t>0$, but measuring both pointer momenta does not [@busshardt2010]. Seeing now the role played by the inferred observables, we can quantify the uncertainty of a simultaneous pointer-based measurement with the help of the so-called noise operators [@arthurs1988] $$\begin{aligned} \label{eq:noiseoperators} \begin{pmatrix} \hat{N}_{\mathcal{X}}(t) \\ \hat{N}_{\mathcal{P}}(t) \end{pmatrix} \equiv \begin{pmatrix} \hat{\mathcal{X}}(t) - \hat{X}_{\mathrm{S}}(0) \\ \hat{\mathcal{P}}(t) - \hat{P}_{\mathrm{
term on the right - hand side of Eq.   [ (\[eq: s\ ]) ] { } vanishes. It is furthermore fair to presuppose that an environmental heat bath in thermal chemical equilibrium leave to a vanishing expectation value of the stochastic effect $ \hat{\boldsymbol{\xi}}(t)$, Eq.   [ (\[eq: stochasticforce\ ]) ] { }. Consequently, presume a vanishing transformation $ \mathbf{s}(t)$ seems appropriate for a typical measurement shape. We will confirm this presumption further below for a specifically chosen initial department of state. Note that the linearity of Eqs.   [ (\[eq: inferredsystem\ ]) ] { } and [ (\[eq: inferredobservables\ ]) ] { }, which is essential for the inference process, Eq.   [ (\[eq: inferredexpecation\ ]) ] { }, is a direct consequence of the bilinear structure of the Hamiltonian, Eq.   [ (\[eq: H\ ]) ] { }. We assume here the non - pathological character that the interaction Hamiltonian $ \hat{H}_{\mathrm{int}}(t)$, Eq.   [ (\[eq: H: int\ ]) ] { }, and the measurement vector $ \hat{\mathbf{w}}(t)$ are chosen in such a way that inferring system observables from the arrow is possible in the first place, which is defined by the existence of the coefficient matrix $ \mathbf{A}(t)$, Eq.   [ (\[eq: inferredobservables\ ]) ] { }. In early words, we require a sufficient information transfer from the system to the pointers. For case, for the classic Arthurs and Kelly model [ @arthurs1965 ] with the interaction Hamiltonian mentioned in Sec.   \[sec: generalopenpointer - basedsimultaneousmeasurements: hamiltonian\ ] and no environmental heat bath, measuring both pointer positions leads to an existing matrix $ \mathbf{A}(t)$ for $ t>0 $, but measure both pointer momenta does not [ @busshardt2010 ]. interpret nowadays the role played by the inferred observables, we can quantify the doubt of a simultaneous arrow - based measurement with the help of the therefore - called noise operators [ @arthurs1988 ] $ $ \begin{aligned } \label{eq: noiseoperators } \begin{pmatrix } \hat{N}_{\mathcal{X}}(t) \\ \hat{N}_{\mathcal{P}}(t) \end{pmatrix } \equiv \begin{pmatrix } \hat{\mathcal{X}}(t) - \hat{X}_{\mathrm{S}}(0) \\ \hat{\mathcal{P}}(t) - \hat{P}_{\mathrm {
tegm on the right-hand side of Eq. [(\[eq:s\])]{} vaniskws. It ms furtgermore feasonable to suppose that ai encironnental heat bath in thdrmal equplibrium oeadw to a vanishing exizctatjln vclne of the stochsstic forca $\hat{\boldsymbon{\xk}}(t)$, Eq. [(\[eq:stochasticforce\])]{}. Consequently, pwesuminb w vanishing shyft $\kwthbr{s}(t)$ seems appropriate for a typicam measugement configuratoon. We will confirm this pgesulption further bellw for a spgdifyxally chosen initial suace. Note thaj the linearity of Eqs. [(\[eq:inferredrystek\])]{} and [(\[eq:indeerefmbservables\])]{}, whicr is essential for tve infetence process, Cq. [(\[eq:iifereedexpecation\])]{}, is a dicect result of the bylinear sdrbcture of the Hamiltobiqn, Eq. [(\[gq:H\])]{}. We assjne fert tie hon-patjolkgical cass that the unteraction Hamiltomiwb $\hat{H}_{\mathrm{iht}}(t)$, Eq. [(\[qq:R:int\])]{}, and the measurement vector $\hat{\mathtf{w}}(f)$ are chosen in such a qay that inferring sydtem obsewvables from the pointers is possible in the firsd plare, whngm is eevined by the existence of the coefficient matwjx $\msthbf{A}(t)$, Eq. [(\[eq:innerredobservables\])]{}. On oyrer words, we tequire a aufficient informahion trwnsfee from tht sysyem to the pointers. For exanple, for the xlassic Arthurs anb Kelly modeu [@arjhurs1965] eith the interaction Haoiltknian mentilned in Ssz. \[sec:generalopenpuinner-bdsedsimultaneousmeasuremenes:hamiltoiian\] cnd no evvirpnmentwl heat bahh, measuring both pointeg posntionv leads to an existing matrix $\mathbf{A}(t)$ foc $t>0$, but measutinc bmth poincer mokenta does noe [@busshardt2010]. Seenng now che roue played gy the mnferred obsqrvables, we cdj quantify tie uncertwintt of a simuuganeous pointet-based mecfurenent with the help of fhe so-called nonrt iperators [@arthirs1988] $$\bedij{akidted} \label{eq:nmiseupefstors} \begin{kiabrib} \hay{N}_{\mathcal{X}}(t) \\ \hat{N}_{\matvcal{L}}(t) \end{pmatrix} \equif \negin{pmattix} \hat{\maehcal{X}}(t) - \hat{X}_{\kathrm{S}}(0) \\ \hat{\mathcap{P}}(t) - \iat{P}_{\mavhrm{
term on the right-hand side of Eq. It furthermore reasonable suppose that an equilibrium to a vanishing value of the force $\hat{\boldsymbol{\xi}}(t)$, Eq. [(\[eq:stochasticforce\])]{}. Consequently, presuming vanishing shift $\mathbf{s}(t)$ seems appropriate for a typical measurement configuration. We will confirm presumption further below for a specifically chosen initial state. Note that the linearity Eqs. and which essential for the inference process, Eq. [(\[eq:inferredexpecation\])]{}, is a direct result of the bilinear structure of Hamiltonian, Eq. [(\[eq:H\])]{}. We assume here the non-pathological that the interaction Hamiltonian Eq. [(\[eq:H:int\])]{}, and the measurement $\hat{\mathbf{w}}(t)$ chosen in a that system observables from pointers is possible in the first place, which is defined by the existence of the coefficient matrix Eq. [(\[eq:inferredobservables\])]{}. words, we a information from the system pointers. For example, for the classic model [@arthurs1965] with the interaction Hamiltonian mentioned in \[sec:generalopenpointer-basedsimultaneousmeasurements:hamiltonian\] and environmental heat bath, measuring both pointer leads to an existing matrix $\mathbf{A}(t)$ for $t>0$, measuring both pointer momenta does not [@busshardt2010]. Seeing now the role played by the inferred can quantify the uncertainty a simultaneous pointer-based with help the noise operators $$\begin{aligned} \label{eq:noiseoperators} \begin{pmatrix} \hat{N}_{\mathcal{X}}(t) \\ \hat{N}_{\mathcal{P}}(t) \end{pmatrix} \equiv \begin{pmatrix} \hat{\mathcal{X}}(t) - \\ \hat{\mathcal{P}}(t) - \hat{P}_{\mathrm{
term on the right-hand side of EQ. [(\[eq:s\])]{} vanishEs. It iS fuRthErMore ReasOnable to supposE That An environmental heat batH in thErMAl eqUIlIbriuM leads tO A vANIshInG eXpeCtATiOn valUe oF the stoChastic forCe $\hAt{\Boldsymbol{\xi}}(T)$, eq. [(\[Eq:stochastIcfOrce\])]{}. ConsequeNtlY, presuMiNg a VAnishIng Shift $\Mathbf{S}(T)$ seems AppropriaTe FOr a typICal measUREmEnt cOnfiguration. We wilL CoNFirm this presumPtion fUrTHeR BEloW foR a specificAlLy choSEn initiAL sTATE. NoTE that the lineaRity of Eqs. [(\[eq:INfeRredsyStEm\])]{} aND [(\[eq:infErredObSErvAbles\])]{}, which iS essEntial for The infERence prOCess, Eq. [(\[eQ:inferRedExpEcatIOn\])]{}, Is A diReCT reSUlT of THe bIlinear sTrUcTure oF the hAMILtonIan, eq. [(\[eq:h\])]{}. We asSume here the noN-paTholOGicAl casE that The iNtEractIon HamIltonIaN $\hat{H}_{\mathrm{int}}(t)$, eq. [(\[eq:h:int\])]{}, and thE meAsUreMeNt vecTOr $\hat{\mAthBf{w}}(T)$ are choSen in suCH a wAy THAT iNferring system obseRvABLeS from the PointeRS iS pOSsible in ThE fiRst pLACe, whiCh is DEfIned by thE existENcE oF the coeFfIcient MaTriX $\maThbf{A}(T)$, eq. [(\[eq:InferrEdobservAbles\])]{}. iN other words, we rEQuire a sufficiENt INFoRMatiOn tRansfer from The sYStem To thE PoIntERs. For ExampLe, FOr THe classic Arthurs and keLly modEl [@artHurs1965] with the inTeraction HAMILtonian mEntiONeD In Sec. \[sec:generaLopenPointer-basEDsimultaNeousMeasuremEnts:hamilTONian\] and nO enVirOnmEntAL HeAt bath, measuriNG Both PoInter poSitIons leaDs tO an ExiStiNg Matrix $\matHbf{A}(t)$ for $T>0$, bUt MeAsUriNg botH Pointer mOmEntA dOes Not [@buSShardt2010]. seeinG now ThE rOLe pLayed by THe INFerrEd ObServAblEs, We can QuanTIfy The unceRtainty of A siMUltaNeOuS pointeR-based measureMeNt with the hElP of The so-cALLed noise Operators [@arthurs1988] $$\begin{alIGned} \labEl{eQ:noisEopeRators} \begIn{pMatrix} \Hat{n}_{\MathcaL{X}}(t) \\ \hat{n}_{\mathCaL{P}}(t) \END{pmatRIX} \eQuiV \bEgin{pmatriX} \HAt{\mAthcaL{X}}(T) - \hat{x}_{\mathrm{s}}(0) \\ \hat{\mathcal{P}}(t) - \hat{P}_{\mAThrM{
term on the right-hand si de of Eq.[(\[e q:s \]) ]{ } va nish es. It is furt h ermo re reasonable to suppo se th at an e n vi ronme ntal he a tb a thin t her ma l e quili bri um lead s to a van ish in g expectatio n v alue of th e s tochastic fo rce $\hat {\ bol d symbo l{\ xi}}( t)$, E q . [(\[ eq:stocha st i cforce \ ])]{}.C o ns eque ntly, presuming a va n ishing shift $ \mathb f{ s }( t ) $ s eem s appropri at e for a typic a lm e a sur e ment configur ation. We w i llconfir mthi s presu mptio nf urt her below f or a specific ally c h osen in i tial st ate. N ote th at t h eli nea ri t y o f E qs. [(\ [eq:infe rr ed syste m\]) ] { } and[(\ [eq: infer redobservable s\] )]{} , wh ich i s ess enti al forthe in feren ce process, Eq. [ (\[e q:inferre dex pe cat io n\])] { }, isa d ire ct resu lt of t h e b il i n e ar structure of theHa m i lt onian, E q. [(\ [ eq :H \ ])]{}. We as sume h ere t he n o n- patholog ical c a se t hat the i nterac ti onHam ilton i an $ \hat{H }_{\math rm{in t }}(t)$, Eq. [( \ [eq:H:int\])] { }, a nd themea surement ve ctor $\ha t{\m a th bf{ w }}(t) $ are c h os e n in such a way tha tinferr ing s ystem observa bles fromt h e pointer s is po s sible in the f irstplace, whi c h is def inedby the e xistenceo f the coe ffi cie ntmat r i x$\mathbf{A}(t ) $ , Eq .[(\[eq: inf erredob ser vab les \]) ]{ }. In oth er words ,we r eq uir e a s u fficient i nfo rm ati on tr a nsferfromthesy st e m t o the p o in t e rs.Fo rexam ple ,for t he c l ass ic Arth urs and K ell y mod el [ @arthur s1965] with t he interacti on Ha milton i a n mentio ned in Sec. \[sec:gener a lopenpo int er-ba seds imultaneo usm easure men t s:hami ltonia n\] a nd no e nviro n m en tal h eat bath,m e asu ringbo th p ointerpositions leads to anexisting matr ix$\ma t h bf {A} ( t) $ fo r$ t>0 $ , but measuringboth point er mo menta does not [ @bussha rdt2010 ]. S e eing no w the rol e playedby the i nfe rred obser vables,we can qu a ntify th e unc ert aintyof asimul taneou s po inter -based m easure mentwi th the h elp of the so-called no ise op erato rs[@arthurs 198 8 ] $ $\begin{a lign ed} \labe l{e q:n oiseo per a tors} \ b eg in{ p matri x} \ h at{N}_{\m a th cal { X }} (t) \\ \hat { N } _{\ mathc al{ P }}(t)\end {pmatrix} \equiv\ begin{pmatrix} \ha t { \ma thc a l{X} }( t) - \hat{X}_{ \ma th r m {S}}(0)\\ \hat{\math cal{P}}( t) - \ha t{P}_{ \mathr m{
term_on the_right-hand side of Eq. [(\[eq:s\])]{}_vanishes. It_is_furthermore reasonable_to_suppose that an_environmental heat bath_in thermal equilibrium leads_to a vanishing_expectation_value of the stochastic force $\hat{\boldsymbol{\xi}}(t)$, Eq. [(\[eq:stochasticforce\])]{}. Consequently, presuming a vanishing shift $\mathbf{s}(t)$ seems_appropriate_for a_typical_measurement_configuration. We will confirm this_presumption further below for a_specifically chosen_initial state. Note that the linearity of Eqs. [(\[eq:inferredsystem\])]{}_and_[(\[eq:inferredobservables\])]{}, which is_essential for the inference process, Eq. [(\[eq:inferredexpecation\])]{}, is a direct_result of the bilinear structure of_the Hamiltonian, Eq. [(\[eq:H\])]{}. We_assume_here_the non-pathological case that_the interaction Hamiltonian $\hat{H}_{\mathrm{int}}(t)$, Eq. [(\[eq:H:int\])]{}, and_the measurement vector $\hat{\mathbf{w}}(t)$ are chosen_in such a way that inferring system_observables from the pointers is possible_in the first place, which_is defined_by the existence of the_coefficient matrix $\mathbf{A}(t)$,_Eq. [(\[eq:inferredobservables\])]{}. In_other words, we_require a sufficient information transfer from_the system to_the pointers. For example, for the_classic_Arthurs and Kelly_model_[@arthurs1965]_with the_interaction Hamiltonian mentioned_in_Sec. \[sec:generalopenpointer-basedsimultaneousmeasurements:hamiltonian\] and_no_environmental heat bath, measuring both pointer_positions_leads to an existing matrix $\mathbf{A}(t)$ for_$t>0$, but measuring both_pointer_momenta does not [@busshardt2010]. Seeing_now the role played by_the inferred observables, we can quantify_the uncertainty_of a_simultaneous pointer-based measurement with the help of the so-called noise operators_[@arthurs1988] $$\begin{aligned} \label{eq:noiseoperators} \begin{pmatrix} \hat{N}_{\mathcal{X}}(t)_\\ \hat{N}_{\mathcal{P}}(t) \end{pmatrix} \equiv_\begin{pmatrix} \hat{\mathcal{X}}(t)_-_\hat{X}_{\mathrm{S}}(0) \\ \hat{\mathcal{P}}(t)_-_\hat{P}_{\mathrm{
frac{1}{\sigma^2} \sum_{mn} mn v_{m0} \sum_{j=1}^\infty u_{jm} \left(\frac{1}{-\lambda_j}\right) v_{nj}.$$ ![Autocorrelation time computed (a) numerically using eigenfunction expansion or (b) by simulation using method of batch means. For sufficient cutoff $N$ or trajectory duration $T$, respectively, both methods converge to same value (dashed line). Parameters: $\theta = h = 0$ and $n_c = 100$. Time is in units of $1/k_1^-$. In (b), $\tau_b = 1000$, and error bars are standard error from $50$ trajectories.[]{data-label="fig:tauc"}](fig5){width="\linewidth"} In matrix notation, $$\label{eq:tauc_mat} \tau_c = \sigma^{-2} \vec{n} {\bf V} {\bf F} {\bf U} \vec{w},$$ where $\vec{n}$ is a row vector, $\vec{w} = mv_{m0}$ is a column vector, and neither the eigenvector matrices ${\bf V}$ and ${\bf U}$ nor the diagonal matrix $F_{jj'} = -\delta_{jj'}/\lambda_j$ contain the $j=0$ term. Numerically, we compute $\tau_c$ via Eq. \[eq:tauc\_mat\] using a cutoff $N > n_c$ for the vectors and matrices. The second method is to calculate $\tau_c$ from stochastic simulations [@gillespie1977exact] and the method of batch means [@thompson2010comparison]. The idea is to divide a simulation trajectory of length $T$ into batches of length $\tau_b$. In the limit $T\gg\tau_b\gg\tau_c$, the correlation time can be estimated by [@thompson2010comparison] $$\label{eq:batch} \tau_c = \frac{\tau_b\sigma_b^2}{2\sigma^2},$$ where $\sigma_b^2$ is the variance of the means of the batches. In Fig. \[fig:tauc\] we verify that the two methods converge to the same limit for sufficiently large $N$ or $T$, respectively. We find that the
frac{1}{\sigma^2 } \sum_{mn } mn v_{m0 } \sum_{j=1}^\infty u_{jm } \left(\frac{1}{-\lambda_j}\right) v_{nj}.$$ ! [ Autocorrelation time computed (a) numerically using eigenfunction expansion or (boron) by model using method of batch means. For sufficient shortcut $ N$ or trajectory duration $ T$, respectively, both methods converge to same value (dashed pipeline). Parameters: $ \theta = h = 0 $ and $ n_c = 100$. clock time is in units of $ 1 / k_1 ^ -$. In (b), $ \tau_b = 1000 $, and erroneousness bars are standard error from $ 50 $ trajectories.[]{data - label="fig: tauc"}](fig5){width="\linewidth " } In matrix note, $ $ \label{eq: tauc_mat } \tau_c = \sigma^{-2 } \vec{n } { \bf V } { \bf F } { \bf U } \vec{w},$$ where $ \vec{n}$ is a row vector, $ \vec{w } = mv_{m0}$ is a column vector, and neither the eigenvector matrices $ { \bf V}$ and $ { \bf U}$ nor the diagonal matrix $ F_{jj' } = -\delta_{jj'}/\lambda_j$ contain the $ j=0 $ term. Numerically, we compute $ \tau_c$ via Eq.   \[eq: tauc\_mat\ ] using a cutoff $ N > n_c$ for the vector and matrices. The second method is to calculate $ \tau_c$ from stochastic pretense [ @gillespie1977exact ] and the method of batch means [ @thompson2010comparison ]. The idea is to divide a model trajectory of length $ T$ into batches of length $ \tau_b$. In the limit $ T\gg\tau_b\gg\tau_c$, the correlation meter can be estimated by [ @thompson2010comparison ] $ $ \label{eq: batch } \tau_c = \frac{\tau_b\sigma_b^2}{2\sigma^2},$$ where $ \sigma_b^2 $ is the variance of the means of the batches. In Fig.   \[fig: tauc\ ] we verify that the two methods converge to the same limit for sufficiently large $ N$ or $ T$, respectively. We find that the
fraf{1}{\sigma^2} \sum_{mn} mn v_{m0} \suo_{j=1}^\infty u_{jm} \lefj(\feac{1}{-\lamuda_j}\riggt) v_{nj}.$$ ![Augocorrelation time computed (e) nunericqlly using eigenfunctiun expanspon or (b) vy smmulation using method on batdm meaus. For sufficienj cutoff $N$ os trajectory dgrxtnon $T$, respectively, both methods convqrge to swme value (dashgd limq). Padameters: $\theta = h = 0$ and $n_c = 100$. Time js in uiits of $1/k_1^-$. In (b), $\yau_b = 1000$, and error bars are dtanfard error from $50$ tgajectories.[]{eata-jqbel="fig:tauc"}](fkg5){width="\lintwndth"} In matrjx notation, $$\label{eq:tauc_mat} \tau_c = \sigmc^{-2} \vec{n} {\bf V} {\bd F} {\tf U} \vec{w},$$ wiere $\vvc{n}$ is a row yvctor, $\vac{w} = mv_{k0}$ is a column yectoc, ane neither the eigenvertor matrices ${\bf V}$ agd ${\bf U}$ nmr the diagonal mateiz $F_{jj'} = -\denta_{jg'}/\oamcda_n$ ronfain tje $o=0$ term. Numedically, we xompute $\tau_c$ via Eq. \[tq:twlv\_mat\] using a cutofs $G > n_c$ for the vectors and matrices. The stcond method is to calculate $\tau_c$ from stochastic simulatijns [@gillespie1977exact] and the method of batch means [@dhompaun2010cimiaririn]. The idea is to divide a simulation trajectorr og kength $T$ into natches of length $\yak_b$. Yn the limit $J\gg\tau_b\yf\tzu_c$, the correlatioj time san bw estimattd by [@thompson2010comparison] $$\label{eq:vatch} \tau_c = \fgac{\tqu_b\sigma_b^2}{2\sigma^2},$$ whexe $\sigma_b^2$ is the varisnce of the means of thz batcges. In Fig. \[fih:tauc\] we bdrify that the tdo kedhods converge to the same limit foc sufyicientlh latge $N$ ow $T$, respechiveln. We find that the
frac{1}{\sigma^2} \sum_{mn} mn v_{m0} \sum_{j=1}^\infty u_{jm} \left(\frac{1}{-\lambda_j}\right) time (a) numerically eigenfunction expansion or of means. For sufficient $N$ or trajectory $T$, respectively, both methods converge to value (dashed line). Parameters: $\theta = h = 0$ and $n_c = 100$. is in units of $1/k_1^-$. In (b), $\tau_b = 1000$, and error bars standard from trajectories.[]{data-label="fig:tauc"}](fig5){width="\linewidth"} matrix notation, $$\label{eq:tauc_mat} \tau_c = \sigma^{-2} \vec{n} {\bf V} {\bf F} {\bf U} \vec{w},$$ where $\vec{n}$ a row vector, $\vec{w} = mv_{m0}$ is a vector, and neither the matrices ${\bf V}$ and ${\bf nor diagonal matrix = contain $j=0$ term. Numerically, compute $\tau_c$ via Eq. \[eq:tauc\_mat\] using a cutoff $N > n_c$ for the vectors and matrices. The method is $\tau_c$ from simulations and method of batch The idea is to divide a length $T$ into batches of length $\tau_b$. In limit $T\gg\tau_b\gg\tau_c$, correlation time can be estimated by $$\label{eq:batch} \tau_c = \frac{\tau_b\sigma_b^2}{2\sigma^2},$$ where $\sigma_b^2$ is the of the means of the batches. In Fig. \[fig:tauc\] we verify that the two methods the same limit for large $N$ or respectively. find the
frac{1}{\sigma^2} \sum_{mn} mn v_{m0} \sum_{j=1}^\infTy u_{jm} \left(\fRac{1}{-\laMbdA_j}\rIgHt) v_{nJ}.$$ ![AutOcorrelation tiME comPuted (a) numerically using EigenFuNCtioN ExPansiOn or (b) by SImULAtiOn UsIng MeTHoD of baTch Means. FoR sufficienT cuToFf $N$ or trajectORy Duration $T$, rEspEctively, both MetHods coNvErgE To samE vaLue (daShed liNE). ParamEters: $\thetA = h = 0$ ANd $n_c = 100$. TiME is in unITS oF $1/k_1^-$. In (B), $\tau_b = 1000$, and error bars ARe STandard error frOm $50$ trajEcTOrIES.[]{daTa-lAbel="fig:tauC"}](fIg5){widTH="\linewiDTh"} iN MAtrIX notation, $$\labeL{eq:tauc_mat} \tAU_c = \sIgma^{-2} \veC{n} {\Bf V} {\BF F} {\bf U} \vEc{w},$$ whErE $\Vec{N}$ is a row vectOr, $\veC{w} = mv_{m0}$ is a cOlumn vECtor, and NEither tHe eigeNveCtoR matRIcEs ${\Bf V}$ AnD ${\Bf U}$ NOr The DIagOnal matrIx $f_{jJ'} = -\deltA_{jj'}/\lAMBDA_j$ coNtaIn thE $j=0$ terM. Numerically, wE coMputE $\Tau_C$ via EQ. \[eq:taUc\_maT\] uSing a Cutoff $n > n_c$ foR tHe vectors and matRiceS. The seconD meThOd iS tO calcULate $\taU_c$ fRom StochasTic simuLAtiOnS [@GILlEspie1977exact] and the meThOD Of Batch meaNs [@thomPSoN2010cOMparison]. thE idEa is TO DividE a siMUlAtion traJectorY Of LeNgth $T$ inTo BatcheS oF leNgtH $\tau_b$. iN the Limit $T\Gg\tau_b\gg\Tau_c$, tHE correlation tiME can be estimatED bY [@THoMPson2010ComParison] $$\labeL{eq:bATch} \tAu_c = \fRAc{\Tau_B\Sigma_B^2}{2\sigmA^2},$$ wHErE $\Sigma_b^2$ is the variance Of The meaNs of tHe batches. In FiG. \[fig:tauc\] we VERIfy that tHe twO MeTHods converge to The saMe limit for SUfficienTly laRge $N$ or $T$, rEspectiveLY. we find thAt tHe
frac{1}{\sigma^2} \sum_{mn } mn v_{m0 } \s um_ {j =1}^ \inf ty u_{jm} \lef t (\fr ac{1}{-\lambda_j}\righ t) v_ {n j }.$$ ! [Auto correla t io n tim eco mpu te d ( a) nu mer icallyusing eige nfu nc tion expansi o nor (b) bysim ulation usin g m ethodof ba t ch me ans . For suffi c ient c utoff $N$ o r traje c tory du r a ti on $ T$, respectively, bo t h methods conv erge t os am e val ue(dashed li ne ). Pa r ameters : $ \ t h eta = h = 0$ and$n_c = 100$ . Ti me isin un i ts of$1/k_ 1^ - $.In (b), $\t au_b = 1000$, and e r ror bar s are st andard er ror fro m $ 50 $ t ra j ect o ri es. [ ]{d ata-labe l= "f ig:ta uc"} ] ( f i g5){ wid th=" \line width"} In m atr ix n o tat ion,$$\la bel{ eq :tauc _mat}\tau_ c= \sigma^{-2} \ vec{ n} {\bf V } { \b f F }{\bfU } \vec {w} ,$$ where$\vec{n } $ i sa r ow vector, $\vec{w}=m v _{ m0}$ isa colu m nve c tor, and n eit hert h e eig enve c to r matric es ${\ b fV} $ and $ {\ bf U}$ n orthe diag o nalmatrix $F_{jj' } = - \ delta_{jj'}/\l a mbda_j$ conta i nt h e$ j=0$ te rm. Numeric ally , wecomp u te $\ t au_c$ viaEq .  \ [ eq:tauc\_mat\] usin ga cuto ff $N > n_c$ for t he vectors a n d matric es.Th e second method is t o calculat e $\tau_c $ fro m stocha stic simu l a tions [@ gil les pie 197 7 e xa ct] and the m e t hodof batchmea ns [@th omp son 201 0co mp arison].The idea i sto d ivi de as imulatio ntra je cto ry of length $T$into b at c hes of len g th $ \tau _b $. Inthe l imit$T\g g \ta u_b\gg\ tau_c$, t hec orre la ti on time can be estim at ed by [@th om pso n2010c o m parison] $$\label{eq:batch} \ta u _c = \f rac {\tau _b\s igma_b^2} {2\ sigma^ 2}, $ $ wher e $\si gma_b ^2 $ i s the v a r ia nce o f the mean s ofthe b at ches . In F ig. \[fig:tauc\] w e ve rify that the tw o me t h od s c o nv e rge t o th e same limit forsufficient ly la rge $N$ or $T$ ,respect ively.We fi n d thatthe
frac{1}{\sigma^2} \sum_{mn}_mn v_{m0} _ \sum_{j=1}^\infty_u_{jm} \left(\frac{1}{-\lambda_j}\right)_v_{nj}.$$ ![Autocorrelation_time computed_(a)_numerically using eigenfunction_expansion or (b)_by simulation using method_of batch means._For_sufficient cutoff $N$ or trajectory duration $T$, respectively, both methods converge to same value_(dashed_line). Parameters:_$\theta_=_h = 0$ and $n_c_= 100$. Time is in_units of_$1/k_1^-$. In (b), $\tau_b = 1000$, and error_bars_are standard error_from $50$ trajectories.[]{data-label="fig:tauc"}](fig5){width="\linewidth"} In matrix notation, $$\label{eq:tauc_mat} \tau_c = \sigma^{-2} \vec{n}_{\bf V} {\bf F} {\bf U}_\vec{w},$$ where $\vec{n}$_is_a_row vector, $\vec{w} =_mv_{m0}$ is a column vector, and_neither the eigenvector matrices ${\bf V}$_and ${\bf U}$ nor the diagonal matrix_$F_{jj'} = -\delta_{jj'}/\lambda_j$ contain the $j=0$_term. Numerically, we compute $\tau_c$_via Eq. \[eq:tauc\_mat\]_using a cutoff $N >_n_c$ for the_vectors and_matrices. The second method_is to calculate $\tau_c$ from stochastic_simulations [@gillespie1977exact] and_the method of batch means [@thompson2010comparison]._The_idea is to_divide_a_simulation trajectory_of length $T$_into_batches of_length_$\tau_b$. In the limit $T\gg\tau_b\gg\tau_c$, the_correlation_time can be estimated by [@thompson2010comparison] $$\label{eq:batch} \tau_c_= \frac{\tau_b\sigma_b^2}{2\sigma^2},$$ where $\sigma_b^2$_is_the variance of the_means of the batches. In Fig. \[fig:tauc\]_we verify that the two methods_converge to_the same_limit for sufficiently large $N$ or $T$, respectively. We find that_the
d_2,\ldots,d_c)$ for $(d_1,d_2,\ldots,d_c)\ne (2,2)$. If $X$ is not a cubic surface or an elliptic curve, then the map $$\mathrm{Aut}(X)\rightarrow \mathrm{Aut}(H^m(X,\mathbb{Q}))$$ is injective where $m = n - c$. The proposition is well-known for smooth algebraic curves of genus at least two and K3 surfaces. If $X$ is a surface with nontrivial canonical bundle, or the dimension of $X$ is at least $3$, then we have $$\mathrm{Aut}(X)=\mathrm{Aut}_L(X).$$Suppose $g$ is a linear automorphism of $X$ acting trivially on $H^m(X,{\mathbb{Q}})$. If $X$ is a Fano or Calabi-Yau complete intersection, then we know that $g$ is extendable to any nearby deformation $Y$ of $X$ by (\[equ\]). It follows from Theorem \[THMCIAUTO000\] that $g$ is a specialization of the identity maps. Therefore, we conclude that $g$ is the identity. Note that the induced action of $g$ on the ring $$S_X = \bigoplus_{n\geq 0}\mathrm{Sym}^d(H^0(X,\omega_X))$$ is trivial. If $X$ is a complete intersection of general type, then it follows that $g$ induces the identity on the scheme $X (\subseteq \mathrm{Proj} (S_X))$, thanks to the fact that $X$ has very ample canonical bundle. By [@PANL Theorem 3.2 and Corollary 3.3], we have an alternative way to show this theorem. Automorphism and Cohomology =========================== Let $X$ be a smooth cubic surface over an algebraically closed field $k$. If $g$ is an automorphism of $X$ such that $g^*=\operatorname{Id}$ on $H_{{\textup{\'et}}}^2(X,\mathbb{Q}_l)$ for a prime $l\ne \operatorname{char}(k)$, then $g=\operatorname{Id}$. It is clear that $X$ is the blow-up of ${\mathbb{P}}^2$ at general six points $p_1,\ldots,p_6$. Denote by
d_2,\ldots, d_c)$ for $ (d_1,d_2,\ldots, d_c)\ne (2,2)$. If $ X$ is not a cubic surface or an elliptic curve, then the map $ $ \mathrm{Aut}(X)\rightarrow \mathrm{Aut}(H^m(X,\mathbb{Q}))$$ is injective where $ m = newton - c$. The suggestion is well - known for smooth algebraic curves of genus at least two and K3 surface. If $ X$ is a surface with nontrivial canonical pile, or the proportion of $ X$ is at least $ 3 $, then we have $ $ \mathrm{Aut}(X)=\mathrm{Aut}_L(X).$$Suppose $ g$ is a analogue automorphism of $ X$ act trivially on $ H^m(X,{\mathbb{Q}})$. If $ X$ is a Fano or Calabi - Yau arrant intersection, then we know that $ g$ is extendable to any nearby distortion $ Y$ of $ X$ by (\[equ\ ]). It follows from Theorem   \[THMCIAUTO000\ ] that $ g$ is a specialization of the identity maps. Therefore, we conclude that $ g$ is the identity. Note that the induce action of $ g$ on the ring $ $ S_X = \bigoplus_{n\geq 0}\mathrm{Sym}^d(H^0(X,\omega_X))$$ is trivial. If $ X$ is a arrant intersection of general type, then it follows that $ g$ induce the identity on the scheme $ X (\subseteq \mathrm{Proj } (S_X))$, thanks to the fact that $ X$ has very ample canonical bundle. By [ @PANL Theorem 3.2 and Corollary 3.3 ], we have an alternate way to show this theorem. Automorphism and Cohomology = = = = = = = = = = = = = = = = = = = = = = = = = = = Let $ X$ be a smooth cubic surface over an algebraically closed field $ k$. If $ g$ is an automorphism of $ X$ such that $ g^*=\operatorname{Id}$ on $ H_{{\textup{\'et}}}^2(X,\mathbb{Q}_l)$ for a premier $ l\ne \operatorname{char}(k)$, then $ g=\operatorname{Id}$. It is clear that $ X$ is the blow - up of $ { \mathbb{P}}^2 $ at general six points $ p_1,\ldots, p_6$. Denote by
d_2,\ldlts,d_c)$ for $(d_1,d_2,\ldots,d_c)\ne (2,2)$. In $X$ is not a cubnx surfece or zn ellipgic curve, then the map $$\mathrl{Ayt}(X)\rithtarrow \mathrm{Aut}(H^m(X,\mxthbb{Q}))$$ is injectice wiere $m = n - c$. The 'dopositljn ia weln-jnown for smoojh algebraic curves of gengs ac least two and K3 surfaces. If $X$ is a surfacr aith nontriviaj camjniczl bundle, or the dimension of $X$ is at leavt $3$, then we hsve $$\mathrm{Aut}(X)=\mathrm{Aut}_L(X).$$Skppode $g$ is a linear aktomorphism of $V$ acting trivkally on $H^m(X,{\mathbb{Q}})$. If $X$ is a Fano or Calabi-Yau compldte iutersection, tyen fe know thav $g$ is extendable bp any tearby ceformation $Y$ pf $E$ by (\[equ\]). It follows from Vheorem \[THMCIAUTO000\] thaj $g$ is a s[eeialization of the idwnrity kaps. Thefwfofe, se ckncludf tiat $g$ is ths identity. Nite that the inducec wbyion of $g$ on the rynd $$S_X = \bigoplus_{n\geq 0}\mathrm{Sym}^d(H^0(X,\omega_X))$$ iv tdivial. If $X$ is a complere intersection of gejeral typq, then it follows that $g$ induces the identity on dhe srhdme $X (\rybdeteq \mathrm{Proj} (S_X))$, thanks to the fact that $X$ gax nery ample canonigal bundle. By [@PANL Yhfotgm 3.2 and Corollxry 3.3], wz hzve an alternative way to show this thejrem. Sutomorphism and Cohomology =========================== Oet $X$ be a siioth cubic surface over an alyebraivally closed field $k$. If $g$ is an zutomorphisl of $X$ sudf that $g^*=\operatorvamv{Id}$ mn $H_{{\textup{\'et}}}^2(X,\mathbb{Q}_l)$ for w prime $l\ie \opzratornaoe{chsr}(k)$, thqn $g=\operatlrnamc{Hd}$. It is clear that $X$ is tve blow-up lf ${\mathbb{P}}^2$ at general six points $p_1,\ldots,p_6$. Denoje ty
d_2,\ldots,d_c)$ for $(d_1,d_2,\ldots,d_c)\ne (2,2)$. If $X$ is cubic or an curve, then the where = n - The proposition is for smooth algebraic curves of genus least two and K3 surfaces. If $X$ is a surface with nontrivial canonical or the dimension of $X$ is at least $3$, then we have $$\mathrm{Aut}(X)=\mathrm{Aut}_L(X).$$Suppose is linear of acting trivially on $H^m(X,{\mathbb{Q}})$. If $X$ is a Fano or Calabi-Yau complete intersection, then we know $g$ is extendable to any nearby deformation $Y$ $X$ by (\[equ\]). It from Theorem \[THMCIAUTO000\] that $g$ a of the maps. we that $g$ is identity. Note that the induced action of $g$ on the ring $$S_X = \bigoplus_{n\geq 0}\mathrm{Sym}^d(H^0(X,\omega_X))$$ is trivial. $X$ is intersection of type, it that $g$ induces on the scheme $X (\subseteq \mathrm{Proj} the fact that $X$ has very ample canonical By [@PANL 3.2 and Corollary 3.3], we have alternative way to show this theorem. Automorphism and =========================== Let $X$ be a smooth cubic surface over an algebraically closed field $k$. If an automorphism of $X$ that $g^*=\operatorname{Id}$ on for prime \operatorname{char}(k)$, $g=\operatorname{Id}$. It clear that $X$ is the blow-up of ${\mathbb{P}}^2$ at general six $p_1,\ldots,p_6$. Denote by
d_2,\ldots,d_c)$ for $(d_1,d_2,\ldots,d_c)\ne (2,2)$. If $X$ Is not a cubiC surfAce Or aN eLlipTic cUrve, then the map $$\MAthrM{Aut}(X)\rightarrow \mathrm{AUt}(H^m(X,\MaTHbb{Q}))$$ IS iNjectIve wherE $M = n - C$. tHe pRoPoSitIoN Is Well-kNowN for smoOth algebraIc cUrVes of genus at LEaSt two and K3 sUrfAces. If $X$ is a suRfaCe with NoNtrIVial cAnoNical Bundle, OR the diMension of $x$ iS At leasT $3$, Then we hAVE $$\mAthrM{Aut}(X)=\mathrm{Aut}_L(X).$$SUPpOSe $g$ is a linear auTomorpHiSM oF $x$ ActIng Trivially oN $H^M(X,{\matHBb{Q}})$. If $X$ iS A FANO Or CALabi-Yau compleTe intersectIOn, tHen we kNoW thAT $g$ is exTendaBlE To aNy nearby defOrmaTion $Y$ of $X$ bY (\[equ\]). It FOllows fROm TheorEm \[THMCiAUtO000\] tHat $g$ IS a SpEciAlIZatIOn Of tHE idEntity maPs. thErefoRe, we CONCLude ThaT $g$ is The idEntity. Note thaT thE indUCed ActioN of $g$ oN the RiNg $$S_X = \bIgopluS_{n\geq 0}\MaThrm{Sym}^d(H^0(X,\omega_x))$$ is tRivial. If $X$ Is a CoMplEtE inteRSectioN of GenEral typE, then it FOllOwS THAt $G$ induces the identitY oN THe Scheme $X (\sUbseteQ \MaThRM{Proj} (S_X))$, tHaNks To thE FAct thAt $X$ hAS vEry ample CanoniCAl BuNdle. By [@PaNl TheorEm 3.2 And corOllarY 3.3], We haVe an alTernativE way tO Show this theoreM. automorphism aND COHOmOLogy =========================== let $x$ be a smooth cUbic SUrfaCe ovER aN alGEbraiCally ClOSeD Field $k$. If $g$ is an automoRpHism of $x$ such That $g^*=\operatorName{Id}$ on $H_{{\tEXTUp{\'et}}}^2(X,\matHbb{Q}_L)$ FoR A prime $l\ne \operaTornaMe{char}(k)$, theN $G=\operatoRname{id}$. It is clEar that $X$ iS THe blow-up Of ${\mAthBb{P}}^2$ At gENErAl six points $p_1,\lDOTs,p_6$. DEnOte by
d_2,\ldots,d_c)$ for $(d_1 ,d_2,\ldot s,d_c )\n e ( 2, 2)$. If$X$ is not a c u bicsurface or an elliptic curv e, then th e map $$\mat h rm { A ut} (X )\ rig ht a rr ow \m ath rm{Aut} (H^m(X,\ma thb b{ Q}))$$ is in j ec tive where $m = n - c$. The propo si tio n is w ell -know n fors moothalgebraic c u rves o f genusa t l east two and K3 surfa c es . If $X$ is a s urface w i th n ont riv ial canoni ca l bun d le, ort he d i men s ion of $X$ is at least $ 3 $,then w ehav e $$\ma thrm{ Au t }(X )=\mathrm{A ut}_ L(X).$$Su ppose$ g$ is a linearautomo rph ism of$ X$ a cti ng tri v ia lly on$H^m(X,{ \m at hbb{Q }})$ . I f $X$ is a F ano o r Calabi-Yaucom plet e in terse ction , th en we k now th at $g $is extendable t o an y nearbydef or mat io n $Y$ of $X$ by (\ [equ\]) . It fo l low sf r o mTheorem \[THMCIAUT O0 0 0 \] that $g $ is a sp ec i alizatio nofthei d entit y ma p s. Therefo re, we co nc lude th at $g$ i sthe id entit y . N ote th at the i nduce d action of $g$ on the ring $ $ S_ X =\ bigo plu s_{n\geq 0} \mat h rm{S ym}^ d (H ^0( X ,\ome ga_X) )$ $ i s trivial. If $X$ is a compl ete i ntersection o f generalt y p e, thenit f o ll o ws that $g$ in duces the ident i ty on th e sch eme $X ( \subseteq \ mathrm{P roj } ( S_X ))$ , th anks to the f a c t th at $X$ ha s v ery amp lecan oni cal b undle. B y [@PANL T he or em 3. 2 and Corollar y3.3 ], we have an alt ernat ivewa yt o s how thi s t h e orem . A utom orp hi sm an d Co h omo logy == ========= === = ==== == == ==== L et $X$ be a s mo oth cubicsu rfa ce ove r an algeb raically closed field $ k $. If $ g$is an aut omorphism of $X$ s uch that $ g^*=\o perat or nam e { Id}$o n $ H_{ {\ textup{\'e t } }}^ 2(X,\ ma thbb {Q}_l)$ for a prime $l\ne \op eratorname{ch ar} (k)$ , th en$ g= \ ope ra t orn a m e{Id}$. It isclear that $ X $is the blo w -up o f ${\ma thbb{P} }^2$a t gener al six po ints $p_1 ,\ ldot s , p_6 $. Denoteby
d_2,\ldots,d_c)$ for_$(d_1,d_2,\ldots,d_c)\ne (2,2)$._If $X$ is not_a cubic_surface_or an_elliptic_curve, then the_map $$\mathrm{Aut}(X)\rightarrow \mathrm{Aut}(H^m(X,\mathbb{Q}))$$_is injective where $m_= n -_c$. The_proposition is well-known for smooth algebraic curves of genus at least two and K3_surfaces._If $X$_is_a_surface with nontrivial canonical bundle,_or the dimension of $X$_is at_least $3$, then we have $$\mathrm{Aut}(X)=\mathrm{Aut}_L(X).$$Suppose $g$ is_a_linear automorphism of_$X$ acting trivially on $H^m(X,{\mathbb{Q}})$. If $X$ is a_Fano or Calabi-Yau complete intersection, then_we know that_$g$_is_extendable to any nearby_deformation $Y$ of $X$ by (\[equ\])._It follows from Theorem \[THMCIAUTO000\] that $g$_is a specialization of the identity maps._Therefore, we conclude that $g$ is_the identity. Note that the induced_action of_$g$ on the ring $$S_X_= \bigoplus_{n\geq 0}\mathrm{Sym}^d(H^0(X,\omega_X))$$_is trivial._If $X$ is_a complete intersection of general type,_then it follows_that $g$ induces the identity on_the_scheme $X (\subseteq_\mathrm{Proj}_(S_X))$,_thanks to_the fact that_$X$_has very_ample_canonical bundle. By [@PANL Theorem 3.2 and_Corollary_3.3], we have an alternative way to_show this theorem. Automorphism and_Cohomology =========================== Let_$X$ be a smooth_cubic surface over an algebraically_closed field $k$. If $g$ is_an automorphism_of $X$_such that $g^*=\operatorname{Id}$ on $H_{{\textup{\'et}}}^2(X,\mathbb{Q}_l)$ for a prime $l\ne \operatorname{char}(k)$, then_$g=\operatorname{Id}$. It is clear that $X$ is_the blow-up of ${\mathbb{P}}^2$_at general_six_points $p_1,\ldots,p_6$. Denote_by
) node[circle,fill=black, minimum size=0, inner sep=1] {} node[midway,above,scale=0.5] {$2$};}}}\mathcal D_2$]{}; (0,-0.5) node [0]{}; (1.5,-0.5) node [1]{}; ; (left2) at (-3,-2) (0,0) node [$\mathcal D_1 {{\tikz[baseline=-4]{\draw (0,0) node[circle,fill=black,minimum size=0,inner sep=1]{} -- (0.3,0) node[circle,fill=black, minimum size=0, inner sep=1] {} node[midway,above,scale=0.5] {$2$};}}}\mathcal D_2$]{}; (0.75,0) node [;]{}; (1.5,0) node [${\mathcal D_2^{\Box}}{{\tikz[baseline=-4]{\draw (0,0) node[circle,fill=black,minimum size=0,inner sep=1]{} -- (0.3,0) node[circle,fill=black, minimum size=0, inner sep=1] {} node[midway,above,scale=0.5] {$2$};}}}\mathcal D_2$]{}; (2.25,0) node [;]{}; (3,0) node [${\mathcal D_3^{\Box}}{{\tikz[baseline=-4]{\draw (0,0) node[circle,fill=black,minimum size=0,inner sep=1]{} -- (0.3,0) node[circle,fill=black, minimum size=0, inner sep=1] {} node[midway,above,scale=0.5] {$2$};}}}\mathcal D_2$]{}; (0,-0.5) node [3]{}; (1.5,-0.5) node [1]{}; (3,-0.5) node [0]{}; ; (left3) at (-3,-3) (0,0) node [$\mathcal D_1 {{\tikz[baseline=-4]{\draw (0,0) node[circle,fill=black,minimum size=0,inner sep=1]{} -- (0.3,0) node[circle,fill=black, minimum size=0, inner sep=1] {} node[midway,
) node[circle, fill = black, minimum size=0, inner sep=1 ] { } node[midway, above, scale=0.5 ] { $ 2$};}}}\mathcal D_2 $ ] { }; (0,-0.5) node [ 0 ] { }; (1.5,-0.5) lymph node [ 1 ] { }; ; (left2) at (-3,-2) (0,0) lymph node [ $ \mathcal D_1 { { \tikz[baseline=-4]{\draw (0,0) node[circle, filling = black, minimum size=0,inner sep=1 ] { } -- (0.3,0) node[circle, fill = bootleg, minimum size=0, inner sep=1 ] { } node[midway, above, scale=0.5 ] { $ 2$};}}}\mathcal D_2 $ ] { }; (0.75,0) node [; ] { }; (1.5,0) node [ $ { \mathcal D_2^{\Box}}{{\tikz[baseline=-4]{\draw (0,0) node[circle, fill = bootleg, minimal size=0,inner sep=1 ] { } -- (0.3,0) node[circle, fill = black, minimum size=0, inner sep=1 ] { } node[midway, above, scale=0.5 ] { $ 2$};}}}\mathcal D_2 $ ] { }; (2.25,0) node [; ] { }; (3,0) node [ $ { \mathcal D_3^{\Box}}{{\tikz[baseline=-4]{\draw (0,0) node[circle, filling = black, minimum size=0,inner sep=1 ] { } -- (0.3,0) node[circle, fill = black, minimal size=0, inner sep=1 ] { } node[midway, above, scale=0.5 ] { $ 2$};}}}\mathcal D_2 $ ] { }; (0,-0.5) node [ 3 ] { }; (1.5,-0.5) node [ 1 ] { }; (3,-0.5) lymph node [ 0 ] { }; ; (left3) at (-3,-3) (0,0) node [ $ \mathcal D_1 { { \tikz[baseline=-4]{\draw (0,0) node[circle, filling = bootleg, minimum size=0,inner sep=1 ] { } -- (0.3,0) node[circle, fill = black, minimum size=0, inside sep=1 ] { } node[midway,
) nofe[circle,fill=black, minimuo size=0, inner sek=1] {} node[mmdway,abkve,scale=0.5] {$2$};}}}\mathcal D_2$]{}; (0,-0.5) node [0]{}; (1.5,-0.5) node [1]{}; ; (lefv2) at (-3,-2) (0,0) nodt [$\mathcal D_1 {{\tikz[basdline=-4]{\draw (0,0) node[ciecle,hill=black,minimum size=0,inncx sep=1]{} -- (0.3,0) nobe[rircle,fill=black, minimum shze=0, inner sep=1] {} nudz[midway,above,scale=0.5] {$2$};}}}\mathcal D_2$]{}; (0.75,0) node [;]{}; (1.5,0) gode [${\mayhfal D_2^{\Box}}{{\tikz[bafelimq=-4]{\dras (0,0) node[circle,fill=black,minimum size=0,jnner stp=1]{} -- (0.3,0) node[circle,filk=black, minimum size=0, inner dep=1] {} node[midway,above,scwle=0.5] {$2$};}}}\mathcal D_2$]{}; (2.25,0) gide [;]{}; (3,0) node [${\mxthcal D_3^{\Boq}}{{\cikz[baseling=-4]{\draw (0,0) node[circle,fill=black,minimuo sizz=0,inner sep=1]{} -- (0.3,0) bodf[wircle,fill=boack, iinimum size=0, inner sap=1] {} nodr[midway,above,scsle=0.5] {$2$};}}}\marhcal D_2$]{}; (0,-0.5) node [3]{}; (1.5,-0.5) node [1]{}; (3,-0.5) node [0]{}; ; (left3) at (-3,-3) (0,0) njde [$\mathcdl D_1 {{\tikz[baseline=-4]{\drqw (0,0) nodg[circne,fiuo=blxck,jiiimhm sizf=0,inier sep=1]{} -- (0.3,0) nkde[circle,fiol=black, minimum sizt=0, igber sep=1] {} node[jidway,
) node[circle,fill=black, minimum size=0, inner sep=1] {} D_2$]{}; node [0]{}; node [1]{}; ; [$\mathcal {{\tikz[baseline=-4]{\draw (0,0) node[circle,fill=black,minimum sep=1]{} -- (0.3,0) minimum size=0, inner sep=1] {} node[midway,above,scale=0.5] D_2$]{}; (0.75,0) node [;]{}; (1.5,0) node [${\mathcal D_2^{\Box}}{{\tikz[baseline=-4]{\draw (0,0) node[circle,fill=black,minimum size=0,inner sep=1]{} -- node[circle,fill=black, minimum size=0, inner sep=1] {} node[midway,above,scale=0.5] {$2$};}}}\mathcal D_2$]{}; (2.25,0) node [;]{}; (3,0) [${\mathcal (0,0) size=0,inner -- (0.3,0) node[circle,fill=black, minimum size=0, inner sep=1] {} node[midway,above,scale=0.5] {$2$};}}}\mathcal D_2$]{}; (0,-0.5) node [3]{}; (1.5,-0.5) node (3,-0.5) node [0]{}; ; (left3) at (-3,-3) (0,0) [$\mathcal D_1 {{\tikz[baseline=-4]{\draw (0,0) size=0,inner sep=1]{} -- (0.3,0) node[circle,fill=black, size=0, sep=1] {}
) node[circle,fill=black, minimuM size=0, inner Sep=1] {} noDe[mIdwAy,AbovE,scaLe=0.5] {$2$};}}}\mathcal D_2$]{}; (0,-0.5) node [0]{}; (1.5,-0.5) NOde [1]{}; ; (lEft2) at (-3,-2) (0,0) node [$\mathcal D_1 {{\tikz[bAseliNe=-4]{\DRaw (0,0) nODe[CirclE,fill=blACk,MINimUm SiZe=0,iNnER sEp=1]{} -- (0.3,0) nodE[ciRcle,filL=black, miniMum SiZe=0, inner sep=1] {} noDE[mIdway,above,ScaLe=0.5] {$2$};}}}\mathcal D_2$]{}; (0.75,0) noDe [;]{}; (1.5,0) nOde [${\matHcAl D_2^{\bOx}}{{\tikZ[baSelinE=-4]{\draw (0,0) nODe[circLe,fill=blaCk,MInimum SIze=0,inneR SEp=1]{} -- (0.3,0) Node[Circle,fill=black, miNImUM size=0, inner sep=1] {} nOde[midWaY,AbOVE,scAle=0.5] {$2$};}}}\Mathcal D_2$]{}; (2.25,0) noDe [;]{}; (3,0) Node [${\mAThcal D_3^{\BOX}}{{\tIKZ[BasELine=-4]{\draw (0,0) node[cIrcle,fill=blACk,mInimum SiZe=0,iNNer sep=1]{} -- (0.3,0) Node[cIrCLe,fIll=black, minImum Size=0, inner Sep=1] {} nodE[Midway,aBOve,scalE=0.5] {$2$};}}}\mathcAl D_2$]{}; (0,-0.5) NodE [3]{}; (1.5,-0.5) nodE [1]{}; (3,-0.5) NoDe [0]{}; ; (LefT3) aT (-3,-3) (0,0) NodE [$\MaThcAL D_1 {{\tIkz[baselInE=-4]{\dRaw (0,0) noDe[ciRCLE,Fill=BlaCk,miNimum Size=0,inner sep=1]{} -- (0.3,0) nOde[CircLE,fiLl=blaCk, minImum SiZe=0, innEr sep=1] {} nOde[miDwAy,
) node[circle,fill=black,minimum si ze=0, in ner s ep=1 ] {} node[midway,a b ove, scale=0.5] {$2$};}}}\m athca lD _2$] { }; (0,- 0.5) no d e[ 0 ]{} ;(1 .5, -0 . 5) node [1 ]{}; ; (left2) a t ( -3 ,-2) (0,0)n od e [$\mathc alD_1 {{\tikz[ bas eline= -4 ]{\ d raw ( 0,0 ) nod e[circ l e,fill =black,mi ni m um siz e =0,inne r se p=1] {} -- (0.3,0) nod e [c i rcle,fill=blac k, min im u ms i ze= 0,inner sep= 1] {} n o de[midw a y, a b o ve, s cale=0.5] {$2 $};}}}\math c alD_2$]{ }; (0 . 75,0)node[; ] {}; (1.5,0) no de [ ${\mathca l D_2^ { \Box}}{ { \tikz[b aselin e=- 4]{ \dra w ( 0, 0)no d e[c i rc le, f ill =black,m in im um si ze=0 , i n n er s ep= 1]{} -- ( 0.3,0) node[c irc le,f i ll= black , min imum s ize=0 , inne r sep =1 ] {} node[midwa y,ab ove,scale =0. 5] {$ 2$ };}}} \ mathca l D _2$ ]{}; (2 .25,0)n ode [ ; ] { }; (3,0) node [${\ma th c a lD_3^{\Bo x}}{{\ t ik z[ b aseline= -4 ]{\ draw ( 0,0)node [ ci rcle,fil l=blac k ,m in imum si ze =0,inn er se p=1 ]{} - - (0. 3,0) n ode[circ le,fi l l=black, minim u m size=0, inn e rs e p= 1 ] {} no de[midway,a bove , scal e=0. 5 ]{$2 $ };}}} \math ca l D _ 2$]{}; (0,-0.5) nod e[3]{}; (1.5 ,-0.5) node [ 1]{}; (3,- 0 . 5 ) node [ 0]{} ; ; (left3) at (- 3,-3) (0,0) no d e [$\mat hcalD_1 {{\t ikz[basel i n e=-4]{\d raw (0 ,0) no d e [c ircle,fill=bl a c k,mi ni mum siz e=0 ,innersep =1] {}--(0 .3,0) nod e[circle ,f il l= bl ack , min i mum size =0 , i nn ersep=1 ] {} no de[mi dway ,
) node[circle,fill=black,_minimum size=0,_inner sep=1] {} node[midway,above,scale=0.5]_{$2$};}}}\mathcal D_2$]{};_(0,-0.5)_node [0]{};_(1.5,-0.5)_node [1]{}; ; (left2)_at (-3,-2) (0,0) node_[$\mathcal D_1 {{\tikz[baseline=-4]{\draw (0,0)_node[circle,fill=black,minimum size=0,inner sep=1]{}_--_(0.3,0) node[circle,fill=black, minimum size=0, inner sep=1] {} node[midway,above,scale=0.5] {$2$};}}}\mathcal D_2$]{}; (0.75,0) node [;]{}; (1.5,0)_node_[${\mathcal D_2^{\Box}}{{\tikz[baseline=-4]{\draw_(0,0)_node[circle,fill=black,minimum_size=0,inner sep=1]{} -- (0.3,0) node[circle,fill=black,_minimum size=0, inner sep=1] {}_node[midway,above,scale=0.5] {$2$};}}}\mathcal_D_2$]{}; (2.25,0) node [;]{}; (3,0) node [${\mathcal D_3^{\Box}}{{\tikz[baseline=-4]{\draw_(0,0)_node[circle,fill=black,minimum size=0,inner sep=1]{}_-- (0.3,0) node[circle,fill=black, minimum size=0, inner sep=1] {} node[midway,above,scale=0.5]_{$2$};}}}\mathcal D_2$]{}; (0,-0.5) node [3]{}; (1.5,-0.5)_node [1]{}; (3,-0.5)_node_[0]{}; ;_(left3) at (-3,-3) (0,0) node_[$\mathcal D_1 {{\tikz[baseline=-4]{\draw (0,0) node[circle,fill=black,minimum size=0,inner_sep=1]{} -- (0.3,0) node[circle,fill=black, minimum size=0,_inner sep=1] {} node[midway,
] considered the correlation functions for the Izergin-Korepin model for massive regime $-1<{q}<0$, within the framework of representation theory of $U_q(A_2^{(2)})$. They gave free field realizations of the vertex operators, and realized integral representation of the correlation functions, as the trace of the vertex operators. In the limiting case, our integral representation reproduce the correlation function for the Izergin-Korepin model at critical point $q=-1$, which was derived by Hou et al.[@HYZ; @M]. This gives a supporting argument that our integral representation (\[Main\]) gives a conjectural formula of the correlation function of the Izergin-Korepin model at massless regime $|q|=1$. In the special case where $\lambda=3\pi$ and $\xi \to \infty$, the auxiliary functions $g_{k}(\{\alpha_j\}_{j=-1}^k), (k=\pm1,0)$ tend to $\widetilde{g}_{k}(\{\alpha_j\}_{j=-1}^k), (k=\pm1,0) $ given by $$\begin{aligned} &&\widetilde{g}_{-1}(\{\alpha_{-1}\})=1,\\ &&\widetilde{g}_{0}(\{\alpha_{-1}, \alpha_0\})= \frac{2\pi i}{\alpha_0-\alpha_{-1}+\pi i},\\ &&\widetilde{g}_{1}(\{\alpha_{-1}, \alpha_0, \alpha_1\})= \frac{2\pi(2\alpha_{-1}-\alpha_0-\alpha_1)} {(\alpha_0-\alpha_{-1}+\pi i) (\alpha_1-\alpha_{-1}+\pi i)}.\end{aligned}$$ The intergal representation $G_{2N}$, (\[Main\]) becomes as follows. $$\begin{aligned} &&\prod_{1\leq s <t \leq 2N} \frac{1}{\Gamma\left( \frac{i(\beta_s-\beta_t)}{6\pi}+1 \right) \Gamma\left(\frac{i(\beta_s-\beta_t)}{6\pi}+\frac{5}{6} \right) \Gamma\left( \frac{i(-\beta_s+\beta_t)}{6\pi}+\frac{1}{2} \right)
] considered the correlation functions for the Izergin - Korepin model for massive government $ -1<{q}<0 $, within the model of representation theory of $ U_q(A_2^{(2)})$. They give detached field realizations of the vertex hustler, and gain integral theatrical performance of the correlation coefficient functions, as the trace of the vertex hustler. In the limiting case, our integral theatrical performance reproduce the correlation affair for the Izergin - Korepin model at critical point $ q=-1 $, which was deduce by Hou et al.[@HYZ; @M ]. This gives a supporting argument that our built-in representation (\[Main\ ]) gives a conjectural recipe of the correlation affair of the Izergin - Korepin model at massless regime $ |q|=1$. In the special case where $ \lambda=3\pi$ and $ \xi \to \infty$, the auxiliary routine $ g_{k}(\{\alpha_j\}_{j=-1}^k), (k=\pm1,0)$ tend to $ \widetilde{g}_{k}(\{\alpha_j\}_{j=-1}^k), (k=\pm1,0) $ given by $ $ \begin{aligned } & & \widetilde{g}_{-1}(\{\alpha_{-1}\})=1,\\ & & \widetilde{g}_{0}(\{\alpha_{-1 }, \alpha_0\})= \frac{2\pi i}{\alpha_0-\alpha_{-1}+\pi i},\\ & & \widetilde{g}_{1}(\{\alpha_{-1 }, \alpha_0, \alpha_1\})= \frac{2\pi(2\alpha_{-1}-\alpha_0-\alpha_1) } { (\alpha_0-\alpha_{-1}+\pi i) (\alpha_1-\alpha_{-1}+\pi i)}.\end{aligned}$$ The intergal representation $ G_{2N}$, (\[Main\ ]) becomes as follows. $ $ \begin{aligned } & & \prod_{1\leq s < t \leq 2N } \frac{1}{\Gamma\left ( \frac{i(\beta_s-\beta_t)}{6\pi}+1 \right) \Gamma\left(\frac{i(\beta_s-\beta_t)}{6\pi}+\frac{5}{6 } \right) \Gamma\left ( \frac{i(-\beta_s+\beta_t)}{6\pi}+\frac{1}{2 } \right )
] cojsidered the correlation functions for jhw Izerjin-Korelin modeu for massive regime $-1<{q}<0$, withii thw franework of representatiun theory of $U_q(A_2^{(2)})$. Rhey tave free hjeld realizatjlns mh the vertex opgrators, and sealized integsau xepresentation of the correlation fugctions, ad the trace of the dertsq iperators. In the limiting cass, our iitegral represemtation reproduce the corrflatlon function for tje Izergin-Kirepyb model at cfitical poput $q=-1$, which sas derived by Hou et al.[@HYZ; @M]. Tfis gnves a suppirrinh argument tiat oug integral reigesentadion (\[Maon\]) gives a conmectucal dormula of the correletion function of thg Izergin-Kmrzpin model at masslesw eegimg $|q|=1$. In the wpeziam rass wherf $\lembda=3\pi$ and $\xi \to \inftt$, the auxiliary funvtyins $g_{k}(\{\alpha_j\}_{j=-1}^i), (k=\pm1,0)$ eegd to $\widetilde{g}_{k}(\{\alpha_j\}_{j=-1}^k), (k=\pm1,0) $ given by $$\tegjn{aligned} &&\widetilde{g}_{-1}(\{\alphq_{-1}\})=1,\\ &&\widetilde{g}_{0}(\{\alpha_{-1}, \alphw_0\})= \frac{2\pi i}{\wlpha_0-\alpha_{-1}+\pi i},\\ &&\widetilde{g}_{1}(\{\alpha_{-1}, \alpha_0, \alpha_1\})= \frac{2\pi(2\alpva_{-1}-\alpia_0-\xlpkq_1)} {(\alphx_0-\qloha_{-1}+\pi i) (\alpha_1-\alpha_{-1}+\pi i)}.\end{aligned}$$ The intergal wsptexentation $G_{2N}$, (\[Mcin\]) becomes as fplpoef. $$\begin{alignea} &&\prod_{1\lzs a <t \leq 2N} \frac{1}{\Gamma\peft( \fras{i(\betq_s-\beta_t)}{6\pi}+1 \wighy) \Gamma\left(\frac{i(\beta_s-\beta_t)}{6\pi}+\drac{5}{6} \right) \Gamiq\left( \frac{i(-\beta_s+\betc_t)}{6\pi}+\frac{1}{2} \righc)
] considered the correlation functions for the for regime $-1<{q}<0$, the framework of gave field realizations of vertex operators, and integral representation of the correlation functions, the trace of the vertex operators. In the limiting case, our integral representation the correlation function for the Izergin-Korepin model at critical point $q=-1$, which was by et @M]. gives a supporting argument that our integral representation (\[Main\]) gives a conjectural formula of the correlation of the Izergin-Korepin model at massless regime $|q|=1$. the special case where and $\xi \to \infty$, the functions (k=\pm1,0)$ tend $\widetilde{g}_{k}(\{\alpha_j\}_{j=-1}^k), $ by $$\begin{aligned} &&\widetilde{g}_{-1}(\{\alpha_{-1}\})=1,\\ \alpha_0\})= \frac{2\pi i}{\alpha_0-\alpha_{-1}+\pi i},\\ &&\widetilde{g}_{1}(\{\alpha_{-1}, \alpha_0, \alpha_1\})= \frac{2\pi(2\alpha_{-1}-\alpha_0-\alpha_1)} {(\alpha_0-\alpha_{-1}+\pi i) (\alpha_1-\alpha_{-1}+\pi i)}.\end{aligned}$$ The intergal representation $G_{2N}$, (\[Main\]) as follows. s <t 2N} \frac{i(\beta_s-\beta_t)}{6\pi}+1 \Gamma\left(\frac{i(\beta_s-\beta_t)}{6\pi}+\frac{5}{6} \right) \Gamma\left(
] considered the correlation fUnctions foR the IZerGin-koRepiN modEl for massive reGIme $-1<{q}<0$, Within the framework of rePreseNtATion THeOry of $u_q(A_2^{(2)})$. They GAvE FRee FiElD reAlIZaTions Of tHe verteX operators, And ReAlized integrAL rEpresentatIon Of the correlaTioN functIoNs, aS The trAce Of the Vertex OPeratoRs. In the liMiTIng casE, Our inteGRAl ReprEsentation reproduCE tHE correlation fuNction FoR ThE iZerGin-korepin modEl At criTIcal poiNT $q=-1$, WHICh wAS derived by Hou Et al.[@HYZ; @M]. ThiS GivEs a supPoRtiNG argumEnt thAt OUr iNtegral reprEsenTation (\[MaiN\]) gives A ConjectURal formUla of tHe cOrrElatIOn FuNctIoN Of tHE IZerGIn-KOrepin moDeL aT massLess REGIMe $|q|=1$. IN thE speCial cAse where $\lambdA=3\pi$ And $\xI \To \iNfty$, tHe auxIliaRy FunctIons $g_{k}(\{\Alpha_J\}_{j=-1}^K), (k=\pm1,0)$ tend to $\widetIlde{G}_{k}(\{\alpha_j\}_{j=-1}^K), (k=\pM1,0) $ gIveN bY $$\begiN{AligneD} &&\wiDetIlde{g}_{-1}(\{\alPha_{-1}\})=1,\\ &&\wideTIldE{g}_{0}(\{\ALPHa_{-1}, \Alpha_0\})= \frac{2\pi i}{\alpha_0-\aLpHA_{-1}+\Pi I},\\ &&\widetilDe{g}_{1}(\{\alpHA_{-1}, \aLpHA_0, \alpha_1\})= \frAc{2\Pi(2\aLpha_{-1}-\ALPha_0-\alPha_1)} {(\aLPhA_0-\alpha_{-1}+\pi I) (\alpha_1-\ALpHa_{-1}+\Pi i)}.\end{aLiGned}$$ ThE iNteRgaL reprESentAtion $G_{2n}$, (\[Main\]) becOmes aS Follows. $$\begin{alIGned} &&\prod_{1\leq s <t \LEq 2n} \FRaC{1}{\gammA\leFt( \frac{i(\beta_S-\betA_T)}{6\pi}+1 \rIght) \gAmMa\lEFt(\fraC{i(\betA_s-\BEtA_T)}{6\pi}+\frac{5}{6} \right) \Gamma\leFt( \Frac{i(-\bEta_s+\bEta_t)}{6\pi}+\frac{1}{2} \rigHt)
] considered the correlati on functio ns fo r t heIz ergi n-Ko repin model fo r mas sive regime $-1<{q}<0$ , wit hi n the fr amewo rk of r e pr e s ent at io n t he o ry of $ U_q (A_2^{( 2)})$. The y g av e free field re alizations of the vertexope rators ,and reali zed inte gral r e presen tation of t h e corr e lationf u nc tion s, as the trace o f t h e vertex opera tors.In th e lim iti ng case, o ur inte g ral rep r es e n t ati o n reproduce t he correlat i onfuncti on fo r the I zergi n- K ore pin model a t cr itical po int $q = -1$, wh i ch wasderive d b y H ou e t a l. [@H YZ ; @M ] .Thi s gi ves a su pp or tingargu m e n t tha t o ur i ntegr al representa tio n (\ [ Mai n\])gives a c on jectu ral fo rmula o f the correlati on f unction o f t he Iz er gin-K o repinmod elat mass less re g ime $ | q | =1 $. In the special c a s ewhere $\ lambda = 3\ pi $ and $\x i\to \in f t y$, t he a u xi liary fu nction s $ g_ {k}(\{\ al pha_j\ }_ {j= -1} ^k),( k=\p m1,0)$ tend to $\wi d etilde{g}_{k}( \ {\alpha_j\}_{ j =- 1 } ^k ) , (k =\p m1,0) $ giv en b y $$\ begi n {a lig n ed} & &\wid et i ld e {g}_{-1}(\{\alpha_{ -1 }\})=1 ,\\ & &\widetilde{g }_{0}(\{\a l p h a_{-1},\alp h a_ 0 \})= \frac{2\p i i}{ \alpha_0-\ a lpha_{-1 }+\pi i},\\ & &\widetil d e {g}_{1}( \{\ alp ha_ {-1 } , \ alpha_0, \alp h a _1\} )= \frac{ 2\p i(2\alp ha_ {-1 }-\ alp ha _0-\alpha _1)} {(\ al ph a_ 0- \al pha_{ - 1}+\pi i )(\a lp ha_ 1-\al p ha_{-1 }+\pi i)} .\ en d {al igned}$ $ T h e int er ga l re pre se ntati on $ G _{2 N}$, (\ [Main\])bec o mesas f ollows. $$\begin{ali gn ed} &&\pro d_ {1\ leq s< t \leq 2N } \frac{1}{\Gamma\left( \frac{i (\b eta_s -\be ta_t)}{6\ pi} +1 \ri ght ) \Gamm a\left (\fra c{ i(\ b e ta_s- \ b et a_t )} {6\pi}+\fr a c {5} {6} \ ri ght) \Gamma \left( \frac{i(-\b e ta_ s+\beta_t)}{6 \pi }+\f r a c{ 1}{ 2 }\ rig ht )
] considered_the correlation_functions for the Izergin-Korepin_model for_massive_regime $-1<{q}<0$,_within_the framework of_representation theory of_$U_q(A_2^{(2)})$. They gave free_field realizations of_the_vertex operators, and realized integral representation of the correlation functions, as the trace of_the_vertex operators._In_the_limiting case, our integral representation_reproduce the correlation function for_the Izergin-Korepin_model at critical point $q=-1$, which was derived_by_Hou et al.[@HYZ;_@M]. This gives a supporting argument that our integral_representation (\[Main\]) gives a conjectural formula_of the correlation_function_of_the Izergin-Korepin model at_massless regime $|q|=1$. In the special case_where $\lambda=3\pi$ and $\xi \to \infty$,_the auxiliary functions $g_{k}(\{\alpha_j\}_{j=-1}^k), (k=\pm1,0)$ tend to_$\widetilde{g}_{k}(\{\alpha_j\}_{j=-1}^k), (k=\pm1,0) $ given by $$\begin{aligned} &&\widetilde{g}_{-1}(\{\alpha_{-1}\})=1,\\ &&\widetilde{g}_{0}(\{\alpha_{-1}, \alpha_0\})= \frac{2\pi_i}{\alpha_0-\alpha_{-1}+\pi i},\\ &&\widetilde{g}_{1}(\{\alpha_{-1}, \alpha_0, \alpha_1\})= \frac{2\pi(2\alpha_{-1}-\alpha_0-\alpha_1)} {(\alpha_0-\alpha_{-1}+\pi i) (\alpha_1-\alpha_{-1}+\pi i)}.\end{aligned}$$_The intergal_representation $G_{2N}$, (\[Main\]) becomes as_follows. $$\begin{aligned} &&\prod_{1\leq s_<t \leq_2N} \frac{1}{\Gamma\left( \frac{i(\beta_s-\beta_t)}{6\pi}+1 \right) \Gamma\left(\frac{i(\beta_s-\beta_t)}{6\pi}+\frac{5}{6} \right) \Gamma\left( \frac{i(-\beta_s+\beta_t)}{6\pi}+\frac{1}{2} \right)
cases, we initialized with a simple interpolation of each spectral channel by bicubic splines. Quality Assessment Indices -------------------------- For simulated data, the spatial and spectral consistency of the fused products with respect to the ground-truth images are numerically evaluated by means of several quality assessment indices. Let $\u^R=\left(u_1^R, \ldots, u_C^R\right)$ be the high-resolution reference multispectral image and let us use the same notations than in Section \[sec:numerics\]. We employ the following quality indices for the evaluation of the pansharpening techniques: - The [*Root Mean Squared Error*]{} (RMSE) is one of the most popular measures that accounts for spatial distortion. It is computed as $$\text{RMSE}\left(u^R_k, u_k\right) = \sqrt{\dfrac{1}{|I|}\sum_{\p\in I} \left(u_k^R(\p) - u_k(\p)\right)}, \quad \forall \,k\in\{1,\ldots, C\},$$ and its optimal value is zero. We shall calculate this measure for each band and average the results over all bands to obtain a global value. - The [*Erreur Relative Globale Adimensionelle de Synth[è]{}se*]{} (ERGAS) proposed by Ranchin and Wald [@RanchinWald2000] is an index that gives a global quality assessment of the fused product. It is defined as $$\text{ERGAS} = \dfrac{100}{s} \sqrt{\dfrac{1}{C} \sum_{k=1}^C \left( \dfrac{\text{RMSE}\left(u^R_k, u_k\right)}{\mu_{u_k^R}}\right)^2},$$ where $s$ is the sampling factor and $\mu_{u_k^R}$ is the mean value of the $k$th spectral component of the reference image. Since the ERGAS is composed by a sum of RMSE, its optimal value is zero. - The [*Spectral Angle Mapper*]{} (SAM) introduced by Alparone [*et al.*]{} [@AlparoneBaronti2004] is a measure of spectral quality computed in the space defined by considering each channel as a coordinate axis. Mathematically, it is written as the absolute
cases, we initialized with a simple interpolation of each apparitional distribution channel by bicubic splines. Quality Assessment Indices -------------------------- For simulated datum, the spatial and spectral consistency of the fuse products with respect to the background - truth images are numerically measure by means of several timbre assessment indices. Let $ \u^R=\left(u_1^R, \ldots, u_C^R\right)$ be the high - settlement reference multispectral image and let us use the same notation than in Section \[sec: numerics\ ]. We employ the following quality index for the evaluation of the pansharpening techniques: - The [ * Root Mean Squared Error * ] { } (RMSE) is one of the most popular measures that account for spatial distortion. It is computed as $ $ \text{RMSE}\left(u^R_k, u_k\right) = \sqrt{\dfrac{1}{|I|}\sum_{\p\in I } \left(u_k^R(\p) - u_k(\p)\right) }, \quad \forall \,k\in\{1,\ldots, C\},$$ and its optimal value is zero. We shall calculate this measurement for each band and average the results over all bands to obtain a global value. - The [ * Erreur Relative Globale Adimensionelle de Synth[è]{}se * ] { } (ERGAS) nominate by Ranchin and Wald [ @RanchinWald2000 ] is an exponent that gives a global quality assessment of the fused product. It is defined as $ $ \text{ERGAS } = \dfrac{100}{s } \sqrt{\dfrac{1}{C } \sum_{k=1}^C \left (\dfrac{\text{RMSE}\left(u^R_k, u_k\right)}{\mu_{u_k^R}}\right)^2},$$ where $ s$ is the sample factor and $ \mu_{u_k^R}$ is the average value of the $ k$th spectral part of the reference point prototype. Since the ERGAS is composed by a sum of RMSE, its optimal value is zero. - The [ * Spectral Angle Mapper * ] { } (SAM) introduced by Alparone [ * et al. * ] { } [ @AlparoneBaronti2004 ] is a measure of spectral quality calculate in the space defined by considering each channel as a coordinate axis. Mathematically, it is written as the absolute
cades, we initialized with x simple interpolation of eadh spectfal channel by bicubic splinxs. Quqlity Assessment Indices -------------------------- For simulatef data, tye skatial and spectrem consistency lf tke fused productx with res[ect to the grmuvd-cruth images are numerically evaluatqd by mrajs of several zualptr asavswment indices. Let $\u^R=\left(u_1^R, \ldkts, u_C^R\gight)$ be the high-tesolution reference multidpechral image and let us use the samq notations tfan in Secuimn \[sec:numetics\]. We employ the following quauity nndices for tye fealuation oh the iansharpening techniqges: - Tne [*Root Mean Sauaced Wrror*]{} (RMSE) is one of vhe most popular meafures thad cccounts for spatial eiwtorthon. Ht ir coopuued aa $$\text{GMSX}\left(u^R_k, u_k\dight) = \sqrt{\efrac{1}{|I|}\sum_{\p\in I} \left(i_k^W(\i) - u_k(\p)\right)}, \qhad \fowajl \,k\in\{1,\ldots, C\},$$ and its optimal value is eero. Se shall calculate this measure for each banf and avewage the results over all bands to obtain a globan valne. - Uhc [*Derfur Relative Globale Adimensionelle de Synth[è]{}sq*]{} (ETGSS) proposed by Ranchin anc Aakq [@RanchinWald2000] is an insex that gives a gpobal qoality assessmegt og the fused product. It is dwfined as $$\teqt{ERTAS} = \dfrac{100}{s} \sqrt{\dfxac{1}{C} \sum_{k=1}^C \lzft( \dftac{\texy{RMSE}\left(u^R_k, u_k\right)}{\mu_{u_y^R}}\rifht)^2},$$ where $s$ is the szopling factor ana $\ml_{u_k^R}$ is the mean value of the $h$th spectcal cpmponeng of the rqference ilage. Since the ERGAS is fompodeg by a sum of RMSE, its optimal value is zxco. - The [*Specjran Atgle Map'er*]{} (SAK) introduced fy Alparone [*et al.*]{} [@AlpcroneBxronti2004] is z measuce of spectrwl quality cokkuted in the vpace desinee by considdfing each chanmel as a bojrdibate axis. Mathematlcalli, jt is written aw tye absolute
cases, we initialized with a simple interpolation spectral by bicubic Quality Assessment Indices spatial spectral consistency of fused products with to the ground-truth images are numerically by means of several quality assessment indices. Let $\u^R=\left(u_1^R, \ldots, u_C^R\right)$ be the reference multispectral image and let us use the same notations than in Section We the quality for the evaluation of the pansharpening techniques: - The [*Root Mean Squared Error*]{} (RMSE) is one the most popular measures that accounts for spatial It is computed as u_k\right) = \sqrt{\dfrac{1}{|I|}\sum_{\p\in I} \left(u_k^R(\p) u_k(\p)\right)}, \forall \,k\in\{1,\ldots, and optimal is zero. We calculate this measure for each band and average the results over all bands to obtain a global - The Globale Adimensionelle Synth[è]{}se*]{} proposed Ranchin and Wald an index that gives a global the fused product. It is defined as $$\text{ERGAS} \dfrac{100}{s} \sqrt{\dfrac{1}{C} \left( \dfrac{\text{RMSE}\left(u^R_k, u_k\right)}{\mu_{u_k^R}}\right)^2},$$ where $s$ is sampling factor and $\mu_{u_k^R}$ is the mean value the $k$th spectral component of the reference image. Since the ERGAS is composed by a RMSE, its optimal value zero. - The Angle (SAM) by [*et al.*]{} is a measure of spectral quality computed in the space defined considering each channel as a coordinate axis. Mathematically, it is the
cases, we initialized with a siMple interpOlatiOn oF eaCh SpecTral Channel by bicubIC splInes. Quality Assessment INdiceS -------------------------- FOR simULaTed daTa, the spATiAL And SpEcTraL cONsIstenCy oF the fusEd products WitH rEspect to the gROuNd-truth imaGes Are numericalLy eValuatEd By mEAns of SevEral qUality ASsessmEnt indiceS. LET $\u^R=\lefT(U_1^R, \ldots, U_c^r\rIght)$ Be the high-resolutiON rEFerence multispEctral ImAGe AND leT us Use the same NoTatioNS than in sEcTION \[seC:Numerics\]. We empLoy the folloWIng QualitY iNdiCEs for tHe evaLuATioN of the panshArpeNing technIques: - THE [*Root MeAN SquareD Error*]{} (rMSe) is One oF ThE mOst PoPUlaR MeAsuREs tHat accouNtS fOr spaTial DISTOrtiOn. IT is cOmputEd as $$\text{RMSE}\lEft(U^R_k, u_K\RigHt) = \sqrT{\dfraC{1}{|I|}\suM_{\p\In I} \leFt(u_k^R(\p) - U_k(\p)\riGhT)}, \quad \forall \,k\in\{1,\lDots, c\},$$ and its opTimAl ValUe Is zerO. we shalL caLcuLate thiS measurE For EaCH BAnD and average the resuLtS OVeR all bandS to obtAIn A gLObal valuE. - THe [*ERreuR rElatiVe GlOBaLe AdimenSionelLE dE SYnth[è]{}se*]{} (eRgAS) proPoSed By RAnchiN And WAld [@RanChinWald2000] Is an iNDex that gives a gLObal quality asSEsSMEnT Of thE fuSed product. IT is dEFineD as $$\tEXt{eRGas} = \dfraC{100}{s} \sqrT{\dFRaC{1}{c} \sum_{k=1}^C \left( \dfrac{\text{rMsE}\left(U^R_k, u_k\Right)}{\mu_{u_k^R}}\rigHt)^2},$$ where $s$ is THE Sampling FactOR aND $\mu_{u_k^R}$ is the meaN valuE of the $k$th sPEctral coMponeNt of the rEference iMAGe. Since tHe ErGAs is ComPOSeD by a sum of RMSE, ITS optImAl value Is zEro. - The [*SPecTraL AnGle maPper*]{} (SAM) inTroduced By alPaRoNe [*eT al.*]{} [@AlPAroneBarOnTi2004] iS a MeaSure oF SpectrAl quaLity CoMpUTed In the spACe DEFineD bY cOnsiDerInG each ChanNEl aS a coordInate axis. matHEmatIcAlLy, it is wRitten as the abSoLute
cases, we initialized wit h a simple inte rpo lat io n of eac h spectral cha n nelby bicubic splines. Q ualit yA sses s me nt In dices - - -- - - --- -- -- --- -- - -- ----- F or simu lated data , t he spatial and sp ectral con sis tency of the fu sed pr od uct s with re spect to th e groun d-truth i ma g es are numeric a l ly eva luated by means o f s e veral qualityassess me n ti n dic es. Let $\u^R =\ left( u _1^R, \ l do t s , u_ C ^R\right)$ be the high-r e sol utionre fer e nce mu ltisp ec t ral image andletus use th e same notatio n s thanin Sec tio n \ [sec : nu me ric s\ ] . W e e mpl o y t he follo wi ng qual ityi n d i cesfor the eval uation of the pa nsha r pen ing t echni ques : - The [* RootMe an Squared Erro r*]{ } (RMSE)ison e o fthe m o st pop ula r m easures that a c cou nt s f or spatial distortio n. I tis compu ted as $$ \t e xt{RMSE} \l eft (u^R _ k , u_k \rig h t) = \sqrt {\dfra c {1 }{ |I|}\su m_ {\p\in I } \ lef t(u_k ^ R(\p ) - u_ k(\p)\ri ght)} , \quad \forall \,k\in\{1,\ld o ts , C\ } ,$$and its optima l va l ue i s ze r o. We shall calc ul a te this measure for ea ch bandand a verage the re sults over a l l bandsto o b ta i n a global val ue. - The [* E rreur Re lativ e Global e Adimens i o nelle de Sy nth [è] {}s e * ]{ } (ERGAS) pro p o sedby Ranchi n a nd Wald [@ Ran chi nWa ld 2000] isan index t ha tgi ves a gl o bal qual it y a ss ess mento f thefused pro du ct . It is def i ne d as $ $\ te xt{E RGA S} = \d frac { 100 }{s} \s qrt{\dfra c{1 } {C}\s um _{k=1}^ C \left( \dfr ac {\text{RMS E} \le ft(u^R _ k , u_k\ri ght)}{\mu_{u_k^R}}\righ t )^2},$$ wh ere $ s$ i s the sam pli ng fac tor and $\ mu_{u_ k^R}$ i s t h e mean v al ueof the $k$th s pec tralco mpon ent ofthe reference imag e . S ince the ERGA S i s co m p os edb ya su mo f R M S E, its optimalvalue is z er o . - The [ * Spe ct ral Ang le Mapp er*]{ } (SAM)introduce d by Alpa ro ne [ * e t a l.*]{} [@A lparoneB aronti200 4 ] isa m easur e o f spec tr alquali ty com p ute d inthe sp ac e defi ned b yconsider ing each channel as a c oordin ate a xis . Mathema tic a lly , it is w ritt en as theabs olu te
cases,_we initialized_with a simple interpolation_of each_spectral_channel by_bicubic_splines. Quality Assessment Indices -------------------------- For_simulated data, the_spatial and spectral consistency_of the fused_products_with respect to the ground-truth images are numerically evaluated by means of several quality_assessment_indices. Let_$\u^R=\left(u_1^R,_\ldots,_u_C^R\right)$ be the high-resolution reference_multispectral image and let us_use the_same notations than in Section \[sec:numerics\]. We employ_the_following quality indices_for the evaluation of the pansharpening techniques: - _The [*Root Mean Squared Error*]{} (RMSE)_is one of_the_most_popular measures that accounts_for spatial distortion. It is computed_as $$\text{RMSE}\left(u^R_k, u_k\right) = \sqrt{\dfrac{1}{|I|}\sum_{\p\in I}_\left(u_k^R(\p) - u_k(\p)\right)}, \quad \forall \,k\in\{1,\ldots, C\},$$_and its optimal value is zero._We shall calculate this measure_for each_band and average the results_over all bands_to obtain_a global value. -_ The [*Erreur Relative Globale_Adimensionelle de Synth[è]{}se*]{}_(ERGAS) proposed by Ranchin and Wald_[@RanchinWald2000]_is an index_that_gives_a global_quality assessment of_the_fused product._It_is defined as $$\text{ERGAS} = \dfrac{100}{s}_\sqrt{\dfrac{1}{C}_\sum_{k=1}^C \left( \dfrac{\text{RMSE}\left(u^R_k, u_k\right)}{\mu_{u_k^R}}\right)^2},$$ where $s$ is_the sampling factor and_$\mu_{u_k^R}$_is the mean value_of the $k$th spectral component_of the reference image. Since the_ERGAS is_composed by_a sum of RMSE, its optimal value is zero. - _The [*Spectral Angle Mapper*]{} (SAM) introduced_by Alparone [*et al.*]{}_[@AlparoneBaronti2004] is_a_measure of spectral_quality_computed in_the space defined by considering each channel_as a_coordinate axis. Mathematically, it is written_as the absolute
D_2)}} \right \} \\ &\qquad=\frac{1}{2} \log \frac{(1-\rho)^2}{D_1 + D_2 - (1+\rho^2) + 2\rho \sqrt{(1-D_1)(1-D_2)}} \label{general1}\end{aligned}$$ where the last equation follows from $$\begin{aligned} {(2-D_1-D_2)}^2 - 4\rho^2 (1-D_1)(1-D_2) \label{2d1d2} & = (1-\rho^2)(2-D_1-D_2)^2 + \rho^2(D_1-D_2)^2\\ &\geq 0 \label{3d1d2}\end{aligned}$$ and therefore, $$D_1 + D_2 - (1+\rho^2)+ 2\rho \sqrt{(1-D_1)(1-D_2)}\leq 1-\rho^2.$$ Then, by comparing with, we find that, Corollary \[cor6\] provides necessary conditions at least as tight as the LW bound if $$\rho \in \left\{\rho: \tau - \sqrt{D_2-1+\tau^2} \leq \rho \leq \tau + \sqrt{D_2-1+\tau^2}, \quad D_2+\tau^2\geq 1\right\}$$ where $$\tau = \frac{D_1}{2} + \sqrt{(1-D_1)(1-D_2)}.$$ By symmetry, it then follows for region $(D_1, D_2)\in \mathcal{F}$ that, Corollary \[cor6\] is at least as tight as the LW bound if $$\rho \in \left\{\rho: \lambda - \sqrt{D_1-1+\lambda^2} \leq \rho \leq \lambda + \sqrt{D_1-1+\lambda^2}, \quad D_1+\tau^2\geq 1 \right\},$$ where $$\lambda = \frac{D_2}{2} + \sqrt{(1-D_1)(1-D_2)}.$$ For $(D_1, D_2)\in \mathcal{G}$, we observe from and that, $$\begin{aligned}
D_2) } } \right \ } \\ & \qquad=\frac{1}{2 } \log \frac{(1-\rho)^2}{D_1 + D_2 - (1+\rho^2) + 2\rho \sqrt{(1 - D_1)(1 - D_2) } } \label{general1}\end{aligned}$$ where the last equation follows from $ $ \begin{aligned } { (2 - D_1 - D_2)}^2 - 4\rho^2 (1 - D_1)(1 - D_2) \label{2d1d2 } & = (1-\rho^2)(2 - D_1 - D_2)^2 + \rho^2(D_1 - D_2)^2\\ & \geq 0 \label{3d1d2}\end{aligned}$$ and therefore, $ $ D_1 + D_2 - (1+\rho^2)+ 2\rho \sqrt{(1 - D_1)(1 - D_2)}\leq 1-\rho^2.$$ Then, by compare with, we receive that, Corollary   \[cor6\ ] provides necessary condition at least equally tight as the LW bound if $ $ \rho \in \left\{\rho: \tau - \sqrt{D_2 - 1+\tau^2 } \leq \rho \leq \tau + \sqrt{D_2 - 1+\tau^2 }, \quad D_2+\tau^2\geq 1\right\}$$ where $ $ \tau = \frac{D_1}{2 } + \sqrt{(1 - D_1)(1 - D_2)}.$$ By symmetry, it then surveil for region $ (D_1, D_2)\in \mathcal{F}$ that, Corollary   \[cor6\ ] is at least as fast as the LW bound if $ $ \rho \in \left\{\rho: \lambda - \sqrt{D_1 - 1+\lambda^2 } \leq \rho \leq \lambda + \sqrt{D_1 - 1+\lambda^2 }, \quad D_1+\tau^2\geq 1 \right\},$$ where $ $ \lambda = \frac{D_2}{2 } + \sqrt{(1 - D_1)(1 - D_2)}.$$ For $ (D_1, D_2)\in \mathcal{G}$, we observe from and that, $ $ \begin{aligned }
D_2)}} \rlght \} \\ &\qquad=\frac{1}{2} \log \frxc{(1-\rho)^2}{D_1 + D_2 - (1+\rho^2) + 2\rho \sxrt{(1-D_1)(1-D_2)}} \mabel{gendral1}\end{aligned}$$ where the lasv eqyatiob follows from $$\begin{alkgned} {(2-D_1-D_2)}^2 - 4\gho^2 (1-D_1)(1-D_2) \lavel{2d1v2} & = (1-\rho^2)(2-D_1-D_2)^2 + \rho^2(D_1-D_2)^2\\ &\jsq 0 \labcj{3d1d2}\ehf{aliynxd}$$ and thereforg, $$D_1 + D_2 - (1+\rho^2)+ 2\sho \sqrt{(1-D_1)(1-D_2)}\leq 1-\shu^2.$$ Chen, by comparing with, we find that, Sorollaty \[for6\] provides ngcessswy cknditions at least as tight as the LW bouid if $$\rho \in \legt\{\rho: \tau - \sqrt{D_2-1+\tau^2} \leq \rhl \lee \tau + \sqrt{D_2-1+\tau^2}, \quwd D_2+\tau^2\geq 1\tjghe\}$$ where $$\tau = \wrac{D_1}{2} + \sqru{(1-D_1)(1-B_2)}.$$ By symmetty, it then follows for region $(D_1, A_2)\in \mcthcal{F}$ thaj, Dogmllary \[cor6\] iw at jeast as tigmn as tha LW boind if $$\rho \in \kefv\{\rho: \lambda - \sqrt{D_1-1+\lambda^2} \neq \rho \leq \lambda + \sqrt{D_1-1+\lakbba^2}, \quad D_1+\tau^2\geq 1 \rigyt\},$$ wherg $$\lamtda = \draz{D_2}{2} + \sarf{(1-D_1)(1-D_2)}.$$ For $(D_1, V_2)\in \mathcal{F}$, we observw from and that, $$\begon{woigned}
D_2)}} \right \} \\ &\qquad=\frac{1}{2} \log \frac{(1-\rho)^2}{D_1 - + 2\rho \label{general1}\end{aligned}$$ where the {(2-D_1-D_2)}^2 4\rho^2 (1-D_1)(1-D_2) \label{2d1d2} = (1-\rho^2)(2-D_1-D_2)^2 + &\geq 0 \label{3d1d2}\end{aligned}$$ and therefore, $$D_1 D_2 - (1+\rho^2)+ 2\rho \sqrt{(1-D_1)(1-D_2)}\leq 1-\rho^2.$$ Then, by comparing with, we find that, \[cor6\] provides necessary conditions at least as tight as the LW bound if \in \tau \sqrt{D_2-1+\tau^2} \rho \leq \tau + \sqrt{D_2-1+\tau^2}, \quad D_2+\tau^2\geq 1\right\}$$ where $$\tau = \frac{D_1}{2} + \sqrt{(1-D_1)(1-D_2)}.$$ By symmetry, then follows for region $(D_1, D_2)\in \mathcal{F}$ that, \[cor6\] is at least tight as the LW bound $$\rho \left\{\rho: \lambda \sqrt{D_1-1+\lambda^2} \rho \lambda + \sqrt{D_1-1+\lambda^2}, D_1+\tau^2\geq 1 \right\},$$ where $$\lambda = \frac{D_2}{2} + \sqrt{(1-D_1)(1-D_2)}.$$ For $(D_1, D_2)\in \mathcal{G}$, we observe from and $$\begin{aligned}
D_2)}} \right \} \\ &\qquad=\frac{1}{2} \log \frac{(1-\rho)^2}{d_1 + D_2 - (1+\rho^2) + 2\rho \sqRt{(1-D_1)(1-D_2)}} \lAbeL{geNeRal1}\eNd{alIgned}$$ where the lASt eqUation follows from $$\begin{AlignEd} {(2-d_1-d_2)}^2 - 4\rho^2 (1-d_1)(1-d_2) \lAbel{2d1D2} & = (1-\rho^2)(2-D_1-D_2)^2 + \rHO^2(D_1-d_2)^2\\ &\GEq 0 \lAbEl{3D1d2}\eNd{ALiGned}$$ aNd tHereforE, $$D_1 + D_2 - (1+\rho^2)+ 2\rho \sQrt{(1-d_1)(1-D_2)}\Leq 1-\rho^2.$$ Then, by COmParing with, We fInd that, CorolLarY \[cor6\] prOvIdeS NecesSarY condItions AT least As tight as ThE lW bounD If $$\rho \in \LEFt\{\Rho: \tAu - \sqrt{D_2-1+\tau^2} \leq \rho \lEQ \tAU + \sqrt{D_2-1+\tau^2}, \quad D_2+\Tau^2\geq 1\RiGHt\}$$ WHEre $$\Tau = \Frac{D_1}{2} + \sqrt{(1-D_1)(1-d_2)}.$$ BY symmETry, it thEN fOLLOws FOr region $(D_1, D_2)\in \mAthcal{F}$ that, cOroLlary \[cOr6\] Is aT Least aS tighT aS The lW bound if $$\rhO \in \lEft\{\rho: \lamBda - \sqrT{d_1-1+\lambda^2} \LEq \rho \leQ \lambdA + \sqRt{D_1-1+\LambDA^2}, \qUaD D_1+\tAu^2\GEq 1 \rIGhT\},$$ whERe $$\lAmbda = \fraC{D_2}{2} + \SqRt{(1-D_1)(1-D_2)}.$$ FOr $(D_1, D_2)\IN \MAThcaL{G}$, wE obsErve fRom and that, $$\begIn{aLignED}
D_2)}} \right \} \\ &\qqua d=\frac{1} {2} \lo g \f rac{ (1-\ rho)^2}{D_1 +D _2 - (1+\rho^2) + 2\rho \ sqrt{ (1 - D_1) ( 1- D_2)} } \lab e l{ g e ner al 1} \en d{ a li gned} $$where t he last eq uat io n follows fr o m$$\begin{a lig ned} {(2-D_1 -D_ 2)}^2-4\r h o^2 ( 1-D _1)(1 -D_2)\ label{ 2d1d2} & = (1-\rh o ^2)(2-D _ 1 -D _2)^ 2 + \rho^2(D_1-D_ 2 )^ 2 \\ &\geq 0 \la bel{3d 1d 2 }\ e n d{a lig ned}$$ and t heref o re, $$D _ 1+ D _2- (1+\rho^2)+2\rho \sqrt { (1- D_1)(1 -D _2) } \leq 1 -\rho ^2 . $$Then, by co mpar ing with, we fi n d that, Corolla ry \[c or6 \]prov i de snec es s ary co ndi t ion s at lea st a s tig ht a s t h e LW bo undif $$ \rho \in \lef t\{ \rho : \t au -\sqrt {D_2 -1 +\tau ^2} \l eq \r ho \leq \tau + \s qrt{ D_2-1+\ta u^2 }, \q ua d D_2 + \tau^2 \ge q 1 \right\ }$$ whe r e $ $\ t a u = \frac{D_1}{2} + \ sq r t {( 1-D_1)(1 -D_2)} . $$ B y symmetr y, it the n follo ws f o rregion $ (D_1,D _2 )\ in \mat hc al{F}$ t hat , C oroll a ry \ [cor6\ ] is atleast as tight as th e LW bound if$ $\ r h o\ in \ lef t\{\rho: \l ambd a - \ sqrt { D_ 1-1 + \lamb da^2} \ l eq \rho \leq \lambda + \ sqrt{D _1-1+ \lambda^2}, \ quad D_1+\ t a u ^2\geq 1 \r i gh t \},$$ where $$ \lamb da = \frac { D_2}{2}+ \sq rt{(1-D_ 1)(1-D_2) } . $$ For$(D _1, D_ 2)\ i n \ mathcal{G}$,w e obs er ve from an d that, $$ \be gin {al ig ned}
D_2)}} \right_\} \\ &\qquad=\frac{1}{2}_ \log \frac{(1-\rho)^2}{D_1_+ D_2_-_(1+\rho^2) +__2\rho \sqrt{(1-D_1)(1-D_2)}} \label{general1}\end{aligned}$$_where the last_equation follows from $$\begin{aligned} {(2-D_1-D_2)}^2_- 4\rho^2 (1-D_1)(1-D_2)_\label{2d1d2}_ & = (1-\rho^2)(2-D_1-D_2)^2 + \rho^2(D_1-D_2)^2\\ &\geq 0 \label{3d1d2}\end{aligned}$$ and therefore, $$D_1 + D_2 - (1+\rho^2)+_2\rho_\sqrt{(1-D_1)(1-D_2)}\leq 1-\rho^2.$$_Then,_by_comparing with, we find that,_Corollary \[cor6\] provides necessary conditions at_least as_tight as the LW bound if $$\rho \in_\left\{\rho:_\tau - \sqrt{D_2-1+\tau^2}_\leq \rho \leq \tau + \sqrt{D_2-1+\tau^2}, \quad D_2+\tau^2\geq 1\right\}$$_where $$\tau = \frac{D_1}{2} + \sqrt{(1-D_1)(1-D_2)}.$$_By symmetry, it_then_follows_for region $(D_1, D_2)\in_\mathcal{F}$ that, Corollary \[cor6\] is at least_as tight as the LW bound_if $$\rho \in \left\{\rho: \lambda - \sqrt{D_1-1+\lambda^2}_\leq \rho \leq \lambda + \sqrt{D_1-1+\lambda^2},_\quad D_1+\tau^2\geq 1 \right\},$$_where $$\lambda_= \frac{D_2}{2} + \sqrt{(1-D_1)(1-D_2)}.$$ For $(D_1,_D_2)\in \mathcal{G}$, we_observe from_and that, $$\begin{aligned}
0 \[Esigma\] where $\RR_{\mu\nu}$ and $\Box$ are the Ricci tensor and the D’Alembert operator corresponding to $\gg_{\mu\nu}$. Vacuum $D$-dimensional equations are thus reduced to scalar-vacuum ones in 4 dimensions. Although such SAS configurations were repeatedly considered [@Rad77; @RSh], it makes sense to return to them to reveal some new features, in particular, those connected with higher dimensions. Field equations for axial symmetry ================================== The SAS 4-metric in the Einstein gauge (\[GE\]) may be written in the Weyl canonical form [@Synge] = \^[2]{}dt\^2-\^[-2]{} \[\^[2]{}(d\^2 + dz\^2) + \^2 d\^2\] \[DsA\] The field equations then can be written as 0, \[Esig\]\ 0, \[Enu\]\ \_z (2\_\_z + \_0\_\_z) \[Bz\]\ \_\[Brho\] where the indices $z$ and $\rho$ denote the partial derivatives $\partial_\rho$ and $\partial_z$, respectively, and $\Delta$ is the “flat” Laplace operator in the cylindrical coordinates: $$\Delta = \rho^{-1}\partial_\rho(\rho\partial_\rho)+\partial_z\partial_z.$$ The integrability condition for (\[Bz\]) and (\[Brho\]) is satisfied automatically. Following the example of [@Rad77], let us seek solutions in the new coordinates $(x,y)$, connected with $\rho$ and $z$ by \^2 = L\^2(x\^2 +)(1-y\^2), z = Lxy \[XY\] where $L$ is a fixed positive constant and $\varepsilon= 0,\ \pm 1$, so that $x$ and $y$ are spherical ($\varepsilon=0$), prolate spheroidal ($\varepsilon=-1$), or oblate spheroidal ($\varepsilon=+1$) coordinates, respectively. The Laplace operator $\Delta$ acquires the form = \_x(x\^2+)\_x +\_y(1-y\^2)\_y. \[Delta\] Separating the variables in Eq.(\[Enu\]), i.e., putting $\nu (x,y) = \chi (x)\psi (y)$, one obtains \_x + 0, \[Ex\]
0 \[Esigma\ ] where $ \RR_{\mu\nu}$ and $ \Box$ are the Ricci tensor and the D’Alembert operator corresponding to $ \gg_{\mu\nu}$. Vacuum $ D$-dimensional equations are thus dilute to scalar - void ones in 4 dimensions. Although such SAS configuration were repeatedly study [ @Rad77; @RSh ], it makes sense to come back to them to reveal some new feature, in particular, those connected with high dimensions. Field equality for axile symmetry = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = The SAS 4 - metric in the Einstein bore (\[GE\ ]) may be written in the Weyl basic kind [ @Synge ] = \^[2]{}dt\^2-\^[-2 ] { } \[\^[2]{}(d\^2 + dz\^2) + \^2 d\^2\ ] \[DsA\ ] The field equations then can be written as 0, \[Esig\]\ 0, \[Enu\]\ \_z (2\_\_z + \_0\_\_z) \[Bz\]\ \_\[Brho\ ] where the index $ z$ and $ \rho$ denote the partial derivatives $ \partial_\rho$ and $ \partial_z$, respectively, and $ \Delta$ is the “ flat ” Laplace hustler in the cylindrical coordinates: $ $ \Delta = \rho^{-1}\partial_\rho(\rho\partial_\rho)+\partial_z\partial_z.$$ The integrability circumstance for (\[Bz\ ]) and (\[Brho\ ]) is satisfied automatically. Following the example of [ @Rad77 ], get us seek solutions in the new coordinates $ (x, y)$, connected with $ \rho$ and $ z$ by \^2 = L\^2(x\^2 +) (1 - y\^2), omega = Lxy \[XY\ ] where $ L$ is a sterilize positive constant and $ \varepsilon= 0,\ \pm 1 $, so that $ x$ and $ y$ are spherical ($ \varepsilon=0 $), prolate spheroidal ($ \varepsilon=-1 $), or oblate spheroidal ($ \varepsilon=+1 $) coordinates, respectively. The Laplace operator $ \Delta$ acquires the form = \_x(x\^2+)\_x + \_y(1 - y\^2)\_y. \[Delta\ ] Separating the variables in Eq.(\[Enu\ ]), i.e., putting $ \nu (x, y) = \chi (x)\psi (y)$, one obtain \_x + 0, \[Ex\ ]
0 \[Eslgma\] where $\RR_{\mu\nu}$ and $\Bow$ are the Ricci jebsor aid the S’Alemberg operator corresponding to $\jg_{\mu\bu}$. Vacyum $D$-dimensional equatkons are nhus reduxed uo scalar-vacuum oiss in 4 dimensjlns. Clvhough such SAS configuradions were repaagebly considered [@Rad77; @RSh], it makes sensq to reyugn to them to teveak soms new features, in particular, those connecued with higher dikensions. Field equations fog axlal symmetry ================================== The SAD 4-metric in the Winstein gauee (\[GE\]) may be written ih the Weyl canonical form [@Synge] = \^[2]{}dt\^2-\^[-2]{} \[\^[2]{}(b\^2 + dz\^2) + \^2 d\^2\] \[DwA\] Thf field equavions nhen can be wvptten av 0, \[Esig\]\ 0, \[Enu\]\ \_z (2\_\_z + \_0\_\_z) \[Bz\]\ \_\[Bvho\] wiere the indices $z$ and $\rhm$ denote the partiwl derivadires $\partial_\rho$ and $\paetual_z$, tespewtivdoy, xnd $\Dxltz$ is tje “hlat” Laplacs operator un the cylindrical vojgcinates: $$\Deltz = \rro^{-1}\partial_\rho(\rho\partial_\rho)+\partial_z\partian_z.$$ Fhe integrability condirion for (\[Bz\]) and (\[Brho\]) ls satisfyed automatically. Following the example of [@Rad77], let us sxey silmtiovw ln the new coordinates $(x,y)$, connected with $\rho$ whd $z$ by \^2 = L\^2(x\^2 +)(1-y\^2), z = Jxy \[XY\] wherr $P$ of a fixed posktive eknatant and $\varepsilln= 0,\ \pm 1$, so tyat $x$ and $y$ ate spherical ($\varepsilon=0$), prooate spheroibal ($\varepsilon=-1$), or oblcte spheroidcl ($\vargpsilom=+1$) coordinates, respectivzly. Ths Laplace ooerator $\Dsuta$ acquires the fogm = \_f(x\^2+)\_x +\_y(1-y\^2)\_y. \[Delta\] Separating tre variaboes nn Eq.(\[Enu\]), i.e., kutting $\nu (x,y) = \chi (x)\psi (y)$, one obtains \_x + 0, \[Gx\]
0 \[Esigma\] where $\RR_{\mu\nu}$ and $\Box$ are tensor the D’Alembert corresponding to $\gg_{\mu\nu}$. reduced scalar-vacuum ones in dimensions. Although such configurations were repeatedly considered [@Rad77; @RSh], makes sense to return to them to reveal some new features, in particular, connected with higher dimensions. Field equations for axial symmetry ================================== The SAS 4-metric the gauge may written in the Weyl canonical form [@Synge] = \^[2]{}dt\^2-\^[-2]{} \[\^[2]{}(d\^2 + dz\^2) + \^2 d\^2\] \[DsA\] field equations then can be written as 0, 0, \[Enu\]\ \_z (2\_\_z \_0\_\_z) \[Bz\]\ \_\[Brho\] where the $z$ $\rho$ denote partial $\partial_\rho$ $\partial_z$, respectively, and is the “flat” Laplace operator in the cylindrical coordinates: $$\Delta = \rho^{-1}\partial_\rho(\rho\partial_\rho)+\partial_z\partial_z.$$ The integrability condition for (\[Bz\]) (\[Brho\]) is Following the of let seek solutions in coordinates $(x,y)$, connected with $\rho$ and = L\^2(x\^2 +)(1-y\^2), z = Lxy \[XY\] where is a positive constant and $\varepsilon= 0,\ \pm so that $x$ and $y$ are spherical ($\varepsilon=0$), spheroidal ($\varepsilon=-1$), or oblate spheroidal ($\varepsilon=+1$) coordinates, respectively. The Laplace operator $\Delta$ acquires the form +\_y(1-y\^2)\_y. \[Delta\] Separating the in Eq.(\[Enu\]), i.e., $\nu = (x)\psi one obtains + 0, \[Ex\]
0 \[Esigma\] where $\RR_{\mu\nu}$ and $\Box$ arE the Ricci tEnsor And The d’ALembErt oPerator correspONdinG to $\gg_{\mu\nu}$. Vacuum $D$-dimensIonal EqUAtioNS aRe thuS reduceD To SCAlaR-vAcUum OnES iN 4 dimeNsiOns. AlthOugh such SAs coNfIgurations weRE rEpeatedly cOnsIdered [@Rad77; @RSh], It mAkes seNsE to REturn To tHem to Reveal SOme new Features, iN pARticulAR, those cONNeCted With higher dimensiONs. fIeld equations fOr axiaL sYMmETRy ================================== THe SaS 4-metric in ThE EinsTEin gaugE (\[gE\]) MAY Be wRItten in the WeyL canonical fORm [@SYnge] = \^[2]{}dt\^2-\^[-2]{} \[\^[2]{}(D\^2 + dZ\^2) + \^2 d\^2\] \[DSa\] The fiEld eqUaTIonS then can be wRittEn as 0, \[Esig\]\ 0, \[ENu\]\ \_z (2\_\_z + \_0\_\_z) \[BZ\]\ \_\[brho\] wheRE the indIces $z$ aNd $\rHo$ dEnotE ThE pArtIaL DerIVaTivES $\paRtial_\rho$ AnD $\pArtiaL_z$, reSPECTiveLy, aNd $\DeLta$ is The “flat” LaplacE opEratOR in The cyLindrIcal CoOrdinAtes: $$\DeLta = \rhO^{-1}\pArtial_\rho(\rho\parTial_\Rho)+\partiaL_z\pArTiaL_z.$$ the inTEgrabiLitY coNdition For (\[Bz\]) anD (\[brhO\]) iS SATiSfied automatically. foLLOwIng the exAmple oF [@raD77], lET us seek sOlUtiOns iN THe new CoorDInAtes $(x,y)$, coNnecteD WiTh $\Rho$ and $z$ By \^2 = l\^2(x\^2 +)(1-y\^2), z = LxY \[Xy\] whEre $l$ is a fIXed pOsitivE constanT and $\vARepsilon= 0,\ \pm 1$, so thAT $x$ and $y$ are spheRIcAL ($\VaREpsiLon=0$), Prolate spheRoidAL ($\varEpsiLOn=-1$), Or oBLate sPheroIdAL ($\vARepsilon=+1$) coordinates, ReSpectiVely. THe Laplace operAtor $\Delta$ aCQUIres the fOrm = \_x(X\^2+)\_X +\_y(1-Y\^2)\_Y. \[Delta\] SeparatiNg the Variables iN eq.(\[Enu\]), i.e., pUttinG $\nu (x,y) = \chi (X)\psi (y)$, one oBTAins \_x + 0, \[Ex\]
0 \[Esigma\] where $\RR_{\ mu\nu}$ an d $\B ox$ ar etheRicc i tensor and t h e D’ Alembert operator corr espon di n g to $\ gg_{\ mu\nu}$ . V a cuu m$D $-d im e ns ional eq uations are thusred uc ed to scalar - va cuum onesin4 dimensions . A lthoug hsuc h SAScon figur ations were r epeatedly c o nsider e d [@Rad 7 7 ;@RSh ], it makes sense to return to them to re ve a ls o menew features, i n par t icular, th o s e co n nected with h igher dimen s ion s. Fi el d e q uation s for a x ial symmetry = ==== ========= ====== = ======= = ===== The SA S 4 -me tric in t heEi n ste i ngau g e ( \[GE\])ma ybe wr itte n i n the We yl c anoni cal form [@Sy nge ] =\ ^[2 ]{}dt \^2-\ ^[-2 ]{ } \[\ ^[2]{} (d\^2 + dz\^2) + \^2 d \^2\ ] \[DsA\] Th efie ld equa t ions t hen ca n be wr itten a s 0, \ [ E s ig \]\ 0, \[Enu\]\ \_ z( 2 \_ \_z + \_ 0\_\_z ) \ [B z \]\ \_\[ Br ho\ ] wh e r e the ind i ce s $z$ an d $\rh o $de note th epartia lder iva tives $\pa rtial_ \rho$ an d $\p a rtial_z$, resp e ctively, and$ \D e l ta $ isthe “flat” Lap lace oper ator in th e cyli ndric al co o rdinates: $$\Delta= \ rho^{ -1}\partial_\ rho(\rho\p a r t ial_\rho )+\p a rt i al_z\partial_z .$$ T he integra b ility co nditi on for ( \[Bz\]) a n d (\[Brho \]) is sa tis f i ed automaticall y . Fo ll owing t heexample of [@ Rad 77] ,let us se ek solut io ns i nthe newc oordinat es $( x, y)$ , con n ectedwith$\rh o$ a n d $ z$ by \ ^ 2= L\^2 (x \^ 2 +) (1- y\ ^2),z =L xy\[XY\]where $L$ is a fi xe dpositiv e constant an d$\varepsil on = 0 ,\ \pm 1 $, so th at $x$ and $y$ are sphe r ical ($ \va repsi lon= 0$), prol ate spher oid a l ($\v arepsi lon=- 1$ ),o r obla t e s phe ro idal ($\va r e psi lon=+ 1$ ) co ordinat es, respectively.T heLaplace opera tor $\D e l ta $ a c qu i res t h e f o r m = \_x(x\^2+)\ _x +\_y(1- y\ ^ 2) \_y. \[Del t a\] S eparati ng thevaria b les inEq.(\[Enu \]), i.e. ,putt i n g $ \nu ( x,y) = \ chi (x)\p s i (y) $ ,one o bta ins \_ x+ 0 , \[E x\]
0 \[Esigma\]_where $\RR_{\mu\nu}$_and $\Box$ are the_Ricci tensor_and_the D’Alembert_operator_corresponding to $\gg_{\mu\nu}$. Vacuum_$D$-dimensional equations are_thus reduced to scalar-vacuum_ones in 4_dimensions._Although such SAS configurations were repeatedly considered [@Rad77; @RSh], it makes sense to return_to_them to_reveal_some_new features, in particular, those_connected with higher dimensions. Field equations_for axial_symmetry ================================== The SAS 4-metric in the Einstein gauge (\[GE\])_may_be written in_the Weyl canonical form [@Synge] = \^[2]{}dt\^2-\^[-2]{} \[\^[2]{}(d\^2 +_dz\^2) + \^2 d\^2\] \[DsA\] The_field equations then_can_be_written as 0, \[Esig\]\ 0,_\[Enu\]\ \_z (2\_\_z + \_0\_\_z) \[Bz\]\ \_\[Brho\] where_the indices $z$ and $\rho$ denote_the partial derivatives $\partial_\rho$ and $\partial_z$, respectively,_and $\Delta$ is the “flat” Laplace_operator in the cylindrical coordinates:_$$\Delta = _ \rho^{-1}\partial_\rho(\rho\partial_\rho)+\partial_z\partial_z.$$_The integrability condition_for (\[Bz\])_and (\[Brho\]) is_satisfied automatically. Following the example of [@Rad77],_let us seek_solutions in the new coordinates $(x,y)$,_connected_with $\rho$ and_$z$_by_\^2 =_L\^2(x\^2 +)(1-y\^2), z_=_Lxy \[XY\]_where_$L$ is a fixed positive constant_and_$\varepsilon= 0,\ \pm 1$, so that $x$_and $y$ are spherical_($\varepsilon=0$),_prolate spheroidal ($\varepsilon=-1$), or_oblate spheroidal ($\varepsilon=+1$) coordinates, respectively._The Laplace operator $\Delta$ acquires the_form =_\_x(x\^2+)\_x +\_y(1-y\^2)\_y._\[Delta\] Separating the variables in Eq.(\[Enu\]), i.e., putting $\nu _ (x,y) = \chi (x)\psi_(y)$, one obtains \_x_+ 0,_\[Ex\]
X\xrightarrow{1_{X}}X\xrightarrow{}0$ is a left triangle. And for every morphism $z:Y\rightarrow Z$ in $\mathcal{C}$, there exists a left triangle $\Omega Z\xrightarrow{x}X\xrightarrow{y}Y\xrightarrow{z}Z$. (LTR2)If $\Omega Z\xrightarrow{x}X\xrightarrow{y}Y\xrightarrow{z}Z$ is a left triangle, then so is $\Omega Y\xrightarrow{-\Omega z}\Omega Z\xrightarrow{x}X\xrightarrow{y}Y$. (LTR3)For any two left triangles $\Omega Z\xrightarrow{x}X\xrightarrow{y}Y\xrightarrow{z}Z$ and $\Omega Z'\xrightarrow{x'}X'\xrightarrow{y'}Y'\xrightarrow{z'}Z'$, and any two morphisms $g:Y\rightarrow Y'$, $h:Z\rightarrow Z'$ such that $hz=z'g$, there exists $f:X\rightarrow X'$ making the following diagram commutative $$\xymatrix{ \Omega Z\ar[r]^{x}\ar[d]^{\Omega h} & X\ar[r]^{y}\ar@{-->}[d]^{f} & Y\ar[r]^{z}\ar[d]^{g} & Z\ar[d]^{h} \\ \Omega Z'\ar[r]^{x'} & X'\ar[r]^{y'} & Y'\ar[r]^{z'} & Z' }$$ (LTR4)For any two left triangles $\Omega Z\xrightarrow{x}X\xrightarrow{y}Y\xrightarrow{z}Z$ and $\Omega Z'\xrightarrow{x'}X'\xrightarrow{y'}Z\xrightarrow{z'}Z'$, there exists a commutative diagram $$\xymatrix{ & \Omega Y'\ar[r]^{\Omega y'}\ar[d]^{x\cdot\Omega y'} & \Omega Z\ar[d]^{x} & \\ & X\ar@{=}[r]\ar[d]^{g} & X\ar[d]^{y} & \\ \Omega Z'\ar[r]^{u}\ar@{=}[d] & X'\ar[r]^{v}\ar[d]^{h} & Y\ar[r]^{z'\
X\xrightarrow{1_{X}}X\xrightarrow{}0 $ is a left triangle. And for every morphism $ z: Y\rightarrow Z$ in $ \mathcal{C}$, there exists a leftover triangulum $ \Omega Z\xrightarrow{x}X\xrightarrow{y}Y\xrightarrow{z}Z$. (LTR2)If $ \Omega Z\xrightarrow{x}X\xrightarrow{y}Y\xrightarrow{z}Z$ is a leftover triangle, then so is $ \Omega Y\xrightarrow{-\Omega z}\Omega Z\xrightarrow{x}X\xrightarrow{y}Y$. (LTR3)For any two left triangles $ \Omega Z\xrightarrow{x}X\xrightarrow{y}Y\xrightarrow{z}Z$ and $ \Omega Z'\xrightarrow{x'}X'\xrightarrow{y'}Y'\xrightarrow{z'}Z'$, and any two morphisms $ guanine: Y\rightarrow Y'$, $ h: Z\rightarrow Z'$ such that $ hz = z'g$, there exist $ f: X\rightarrow X'$ form the following diagram commutative $ $ \xymatrix { \Omega Z\ar[r]^{x}\ar[d]^{\Omega h } & X\ar[r]^{y}\ar@{-->}[d]^{f } & Y\ar[r]^{z}\ar[d]^{g } & Z\ar[d]^{h } \\ \Omega Z'\ar[r]^{x' } & X'\ar[r]^{y' } & Y'\ar[r]^{z' } & Z' } $ $ (LTR4)For any two entrust triangles $ \Omega Z\xrightarrow{x}X\xrightarrow{y}Y\xrightarrow{z}Z$ and $ \Omega Z'\xrightarrow{x'}X'\xrightarrow{y'}Z\xrightarrow{z'}Z'$, there exists a commutative diagram $ $ \xymatrix { & \Omega Y'\ar[r]^{\Omega y'}\ar[d]^{x\cdot\Omega y' } & \Omega Z\ar[d]^{x } & \\ & X\ar@{=}[r]\ar[d]^{g } & X\ar[d]^{y } & \\ \Omega Z'\ar[r]^{u}\ar@{=}[d ] & X'\ar[r]^{v}\ar[d]^{h } & Y\ar[r]^{z'\
X\xrlghtarrow{1_{X}}X\xrightarrow{}0$ ir a left triangle. And hor evedy morphksm $z:Y\rightarrow Z$ in $\mathcap{C}$, thert exists a left trixngle $\Omeha Z\xrighrarriq{x}X\xrightacdow{y}Y\xrlyhtardlw{z}Z$. (NVR2)If $\Omega Z\xrigmtarrow{x}X\xrhghtarrow{y}Y\xrichgaxrow{z}Z$ is a left triangle, then so is $\Omega U\xgightarrow{-\Omegw z}\Okqga Z\sginhtarrow{x}X\xrightarrow{y}Y$. (LTR3)For ang two ltft triangles $\Omegs Z\xrightarrow{x}X\xrightarrow{j}Y\xrlghtarrow{z}Z$ and $\Omfga Z'\xrightatdow{v'}Z'\xrightarrow{h'}Y'\xrightargmw{z'}Z'$, and ahy two morphisms $g:Y\rightarrow Y'$, $h:Z\riyhtarrow Z'$ wuxh hvat $hz=z'g$, thxre expsts $f:X\rightavgow X'$ mdking tne following dlagrak cimmutative $$\xymatrix{ \Omxga Z\ar[r]^{x}\ar[d]^{\Omega h} & X\ar[r]^{y}\ar@{-->}[d]^{x} & V\ar[r]^{z}\ar[d]^{g} & Z\ar[d]^{h} \\ \Omegq Z'\ar[r]^{x'} & X'\as[r]^{y'} & T'\ar[f]^{z'} & Z' }$$ (LFR4)For wny two left friangles $\Onega Z\xrightarrow{x}X\xtidytarrow{y}Y\xriggtarror{z}S$ and $\Omega Z'\xrightarrow{x'}X'\xrightarrow{y'}Z\xgighfarrow{z'}Z'$, there exists a commutative diagram $$\dymatrix{ & \Omega Y'\ar[r]^{\Omega y'}\ar[g]^{x\cdov\Ooegc y'} & \Oowgw Z\ar[d]^{x} & \\ & X\ar@{=}[r]\wd[d]^{b} & X\ar[d]^{y} & \\ \Omega Z'\wr[t]^{o}\ar@{=}[d] & X'\ar[r]^{v}\ar[d]^{f} & Y\ar[x]^{a'\
X\xrightarrow{1_{X}}X\xrightarrow{}0$ is a left triangle. And for $z:Y\rightarrow in $\mathcal{C}$, exists a left Z\xrightarrow{x}X\xrightarrow{y}Y\xrightarrow{z}Z$ a left triangle, so is $\Omega z}\Omega Z\xrightarrow{x}X\xrightarrow{y}Y$. (LTR3)For any two left $\Omega Z\xrightarrow{x}X\xrightarrow{y}Y\xrightarrow{z}Z$ and $\Omega Z'\xrightarrow{x'}X'\xrightarrow{y'}Y'\xrightarrow{z'}Z'$, and any two morphisms $g:Y\rightarrow Y'$, $h:Z\rightarrow Z'$ that $hz=z'g$, there exists $f:X\rightarrow X'$ making the following diagram commutative $$\xymatrix{ \Omega h} X\ar[r]^{y}\ar@{-->}[d]^{f} Y\ar[r]^{z}\ar[d]^{g} Z\ar[d]^{h} \\ \Omega Z'\ar[r]^{x'} & X'\ar[r]^{y'} & Y'\ar[r]^{z'} & Z' }$$ (LTR4)For any two left triangles Z\xrightarrow{x}X\xrightarrow{y}Y\xrightarrow{z}Z$ and $\Omega Z'\xrightarrow{x'}X'\xrightarrow{y'}Z\xrightarrow{z'}Z'$, there exists a commutative $$\xymatrix{ & \Omega Y'\ar[r]^{\Omega y'} & \Omega Z\ar[d]^{x} & & & X\ar[d]^{y} \\ Z'\ar[r]^{u}\ar@{=}[d] X'\ar[r]^{v}\ar[d]^{h} & Y\ar[r]^{z'\
X\xrightarrow{1_{X}}X\xrightarrow{}0$ Is a left triAngle. and For EvEry mOrphIsm $z:Y\rightarroW z$ in $\mAthcal{C}$, there exists a lefT triaNgLE $\OmeGA Z\XrighTarrow{x}x\XrIGHtaRrOw{Y}Y\xRiGHtArrow{Z}Z$. (LtR2)If $\OmeGa Z\xrightaRroW{x}x\xrightarrow{Y}y\xRightarrow{Z}Z$ iS a left triangLe, tHen so iS $\OMegA y\xrigHtaRrow{-\OMega z}\OMEga Z\xrIghtarrow{X}X\XRightaRRow{y}Y$. (LTr3)fOr Any tWo left triangles $\OmEGa z\Xrightarrow{x}X\xRightaRrOW{y}y\XRigHtaRrow{z}Z$ and $\OMeGa Z'\xrIGhtarroW{X'}X'\XRIGhtARrow{y'}Y'\xrightaRrow{z'}Z'$, and anY Two MorphiSmS $g:Y\RIghtarRow Y'$, $h:z\rIGhtArrow Z'$ such tHat $hZ=z'g$, there eXists $f:x\RightarROw X'$ makiNg the fOllOwiNg diAGrAm ComMuTAtiVE $$\xYmaTRix{ \omega Z\ar[R]^{x}\Ar[D]^{\OmegA h} & X\aR[R]^{Y}\AR@{-->}[d]^{f} & Y\Ar[r]^{Z}\ar[d]^{G} & Z\ar[d]^{H} \\ \Omega Z'\ar[r]^{x'} & X'\aR[r]^{y'} & y'\ar[r]^{Z'} & z' }$$ (LTr4)For aNy two Left TrIanglEs $\OmegA Z\xriGhTarrow{x}X\xrightaRrow{Y}Y\xrightaRroW{z}z$ anD $\OMega Z'\XRightaRroW{x'}X'\XrightaRrow{y'}Z\xRIghTaRROW{z'}z'$, there exists a commuTaTIVe Diagram $$\xYmatriX{ & \omEgA y'\ar[r]^{\OmegA y'}\Ar[d]^{X\cdoT\oMega y'} & \omegA z\aR[d]^{x} & \\ & X\ar@{=}[r]\aR[d]^{g} & X\ar[D]^{Y} & \\ \OMeGa Z'\ar[r]^{u}\Ar@{=}[D] & X'\ar[r]^{v}\Ar[D]^{h} & Y\Ar[r]^{Z'\
X\xrightarrow{1_{X}}X\xrig htarrow{}0 $ isa l eft t rian gle. And for every morp hism $z:Y\rightarrow Z $ in$\ m athc a l{ C}$,there e x is t s ale ft tr ia n gl e $\O meg a Z\xri ghtarrow{x }X\ xr ightarrow{y} Y \x rightarrow {z} Z$. (LTR2)I f $ \Omega Z \xr i ghtar row {x}X\ xright a rrow{y }Y\xright ar r ow{z}Z $ is a l e f ttria ngle, then so is$ \O m ega Y\xrightar row{-\ Om e ga z }\O meg a Z\xright ar row{x } X\xrigh t ar r o w {y} Y $. (LTR3)For any two le f t t riangl es $\ O mega Z \xrig ht a rro w{x}X\xrigh tarr ow{y}Y\xr ightar r ow{z}Z$ and $\O mega Z '\x rig htar r ow {x '}X '\ x rig h ta rro w {y' }Y'\xrig ht ar row{z '}Z' $ , a nd a nytwomorph isms $g:Y\rig hta rrow Y'$ , $h: Z\rig htar ro w Z'$ suchthat$h z=z'g$, there e xist s $f:X\ri ght ar row X '$ ma k ing th e f oll owing d iagramc omm ut a t i ve $$\xymatrix{ \Ome ga Z \a r[r]^{x} \ar[d] ^ {\ Om e ga h} &X\ ar[ r]^{ y } \ar@{ -->} [ d] ^{f} & Y \ar[r] ^ {z }\ ar[d]^{ g} & Z\a r[ d]^ {h} \\ \ O mega Z'\ar [r]^{x'} & X' \ ar[r]^{y'} & Y ' \ar[r]^{z'} & Z' } $$ (LT R4) For any two lef t tri angl e s$\O m ega Z \xrig ht a rr o w{x}X\xrightarrow{y }Y \xrigh tarro w{z}Z$ and $\ Omega Z'\x r i g htarrow{ x'}X ' \x r ightarrow{y'}Z \xrig htarrow{z' } Z'$, the re ex ists a c ommutativ e diagram$$\ xym atr ix{ & \O mega Y' \ar [r] ^{\ Ome ga y'}\ar[d ]^{x\cdo t\ Om eg ay'} & \O m ega Z\ar [d ]^{ x} & \\ &X \a r @ {=}[ r] \a r[d] ^{g } & X\ ar[d ] ^{y } & \ \ \ O mega Z '\ ar[r]^{ u}\ar@{=}[d]&X'\ar[r]^{ v} \ar [d]^{h } & Y\ar[r ]^{z'\
X\xrightarrow{1_{X}}X\xrightarrow{}0$ is_a left_triangle. And for every_morphism $z:Y\rightarrow_Z$_in $\mathcal{C}$,_there_exists a left_triangle $\Omega Z\xrightarrow{x}X\xrightarrow{y}Y\xrightarrow{z}Z$. (LTR2)If $\Omega Z\xrightarrow{x}X\xrightarrow{y}Y\xrightarrow{z}Z$_is a left triangle,_then so is_$\Omega_Y\xrightarrow{-\Omega z}\Omega Z\xrightarrow{x}X\xrightarrow{y}Y$. (LTR3)For any two left triangles $\Omega Z\xrightarrow{x}X\xrightarrow{y}Y\xrightarrow{z}Z$ and $\Omega Z'\xrightarrow{x'}X'\xrightarrow{y'}Y'\xrightarrow{z'}Z'$, and any two morphisms $g:Y\rightarrow_Y'$,_$h:Z\rightarrow Z'$_such_that_$hz=z'g$, there exists $f:X\rightarrow X'$_making the following diagram commutative_$$\xymatrix{ \Omega Z\ar[r]^{x}\ar[d]^{\Omega_h} & X\ar[r]^{y}\ar@{-->}[d]^{f} & Y\ar[r]^{z}\ar[d]^{g} & Z\ar[d]^{h} \\ \Omega Z'\ar[r]^{x'}_&_X'\ar[r]^{y'} & Y'\ar[r]^{z'}_& Z' }$$ (LTR4)For any two left triangles $\Omega Z\xrightarrow{x}X\xrightarrow{y}Y\xrightarrow{z}Z$ and_$\Omega Z'\xrightarrow{x'}X'\xrightarrow{y'}Z\xrightarrow{z'}Z'$, there exists a commutative diagram_$$\xymatrix{ ___ _ _ _ _ & \Omega Y'\ar[r]^{\Omega y'}\ar[d]^{x\cdot\Omega_y'} & \Omega Z\ar[d]^{x} &_ \\ _ _ _ _ _ _ _ __& X\ar@{=}[r]\ar[d]^{g} _&_X\ar[d]^{y}_ &_ __ __ \\ \Omega Z'\ar[r]^{u}\ar@{=}[d] &_X'\ar[r]^{v}\ar[d]^{h}_& Y\ar[r]^{z'\
Technically, this approach, with $N$ modes being taken into account, means following the dynamics of $7N+3$ real variables. According to our experience, for most of our results, the frequency interval $[0, 30\nu]$ with $N=3000$ together with the realistic assumption of $\Omega_n/\omega_0=0.001\sqrt{\tilde{\omega}_n/\omega_0}$ satisfy the requirement mentioned at the end of the previous section: even an exciting pulse as long as few hundred optical cycles is much shorter than the time scale on which the quantized modes observably modify the atom’s dynamics. This means that the number of modes that we take into account has negligible influence on the time evolution of the atom, it is mainly the resolution of the spectra that is determined by the value of $N.$ By requiring 100 frequency values (points on the graphs) in an interval of “length” $\nu,$ the value of $N=3000$ means covering the interval of $[0, 30\nu]$ sufficiently densely. For most of the cases we considered, this is appropriate, since no harmonics of higher order than 30 appear. However, for the strongly detuned excitation shown in Fig. \[resnonresfig\](d), there are roughly ten times more observable harmonics and the value of $N$ had to be increased correspondingly. As an indication of the correctness of the factorization described above, let us mention that without excitation ($E_0=0$), the initial conditions $\langle N_i\rangle=n\delta_{i0}$ (where $\tilde{\omega}_0=\omega_0$) lead to Rabi flopping between mode $0$ and the atomic system with a frequency very close to $\Omega_0\sqrt{n+1}.$ The time evolution of the photon number expectation values is shown in Figs. \[nexpfig1\] and \[nexpfig2\]. As we can see, the longer the exciting pulse is, the more pronounced peaks can be observed. As one can expect, these peaks appear around integer multiples of the carrier frequency of the exciting pulse. For long enough pulses (see Fig. \[nexpfig2\]) and higher harmonic orders, a significant, qualitative difference can be perceived between the consecutive peaks: the ones at odd multiples of $\nu$ are considerably broader than those corresponding to even harmonics. We will return to this point in the next section.
Technically, this approach, with $ N$ modes being taken into account, mean take after the dynamics of $ 7N+3 $ real variables. accord to our experience, for most of our results, the frequency interval $ [ 0, 30\nu]$ with $ N=3000 $ in concert with the realistic assumption of $ \Omega_n/\omega_0=0.001\sqrt{\tilde{\omega}_n/\omega_0}$ satisfy the requirement mention at the end of the previous incision: even an exciting pulse as long as few hundred ocular cycles is much shorter than the time scale on which the quantize modes observably modify the atom ’s moral force. This means that the number of mode that we take into history has negligible influence on the time evolution of the atom, it is mainly the resolution of the spectrum that is determined by the value of $ N.$ By requiring 100 frequency values (points on the graphs) in an interval of “ length ” $ \nu,$ the value of $ N=3000 $ means covering the interval of $ [ 0, 30\nu]$ sufficiently densely. For most of the cases we considered, this is appropriate, since no harmonics of higher order than 30 appear. However, for the powerfully detuned excitement shown in Fig.   \[resnonresfig\](d), there are roughly ten time more observable harmonics and the value of $ N$ had to be increase correspondingly. As an indication of the correctness of the factorization described above, let us mention that without excitement ($ E_0=0 $), the initial conditions $ \langle N_i\rangle = n\delta_{i0}$ (where $ \tilde{\omega}_0=\omega_0 $) lead to Rabi flopping between mode $ 0 $ and the atomic system with a frequency very close to $ \Omega_0\sqrt{n+1}.$ The time evolution of the photon number expectation values is show in Figs.   \[nexpfig1\ ] and \[nexpfig2\ ]. As we can interpret, the longer the exciting pulse is, the more pronounced peaks can be observed. As one can expect, these bill appear around integer multiples of the carrier frequency of the exciting pulsation. For long adequate pulses (see Fig.   \[nexpfig2\ ]) and higher harmonic orders, a meaning, qualitative difference can be perceived between the consecutive peaks: the ones at curious multiples of $ \nu$ are well broader than those corresponding to even harmonic. We will return to this point in the future part.
Tecjnically, this approach, wlth $N$ modes beiny taken into zccount, oeans following the dynamics od $7N+3$ rtcl variables. Accordivg to our experiebce, hor most of our cssults, bke frseueney interval $[0, 30\nu]$ eith $N=3000$ togather with the rdapistic assumption of $\Omega_n/\omega_0=0.001\sqre{\tilde{\okeha}_n/\omega_0}$ satissy tnq reslivement mentioned at the end of tge prevpous section: even an exciting pulse as long as vew hundred opticap cycles is mucr shorter thav the time scale on whjch the quantized modes observacly mpdify the qtim’s gynamics. Thms meags that the klmber ox modes that we take lnto eccoynt has negligible inhluence on the time gvolution mf the atom, it is mqibly tve rasoljriov or vhe spectga vhat is detsrmined by rhe value of $N.$ By rtquygong 100 frequendy valtef (points on the graphs) in an interval ox “lsngth” $\nu,$ the value of $N=3000$ means covering the ijterval os $[0, 30\nu]$ sufficiently densely. For most of the cases fe coiskdexcq, rhls is appropriate, since no harmonics of highew otdvr than 30 appear. Hjwever, for yhf xjrongly detunea excicztjon shown in Fig. \[rednonressig\](d), rhere are roubhly ten times more observavle harmonicf and the value of $U$ had to be nncreaxed cprrespondingly. As an indncatioh of the cogrectness uf the factorizagiok dascribed above, let us mentyon that xithobt excitxtiom ($E_0=0$), thq initial fondibhons $\langle N_i\rangpe=n\deptd_{i0}$ (where $\tllde{\omega}_0=\omega_0$) lead to Rabi flo''ing between koge $0$ and the atomlc system with w frequency vety close co $\Omeea_0\sqrt{n+1}.$ The time etolution of ehe photon nukher expectatmon valuef is shoqn in Fkes. \[nexpfig1\] and \[mexpfig2\]. As we can swe, the longer the cxcitkhg pulse is, the motw pronounced prakr cwn bx obsqsved. As one wan dxpdvt, thdse peaks aipexr atound integer multipnes kf the carrier freaucncy of tye excityng pulse. For long enough pulsed (see Fij. \[nexpfog2\]) wnd higher harmonic orders, a sjgnificanh, qmalitative disfercnce can be pexceived between the consecutive peaks: thx ones at odd multiples of $\nu$ are considercbky broader vhan trose corrasponding to even haemonics. We will rtturn to this point in fhe neft seftion.
Technically, this approach, with $N$ modes being account, following the of $7N+3$ real for of our results, frequency interval $[0, with $N=3000$ together with the realistic of $\Omega_n/\omega_0=0.001\sqrt{\tilde{\omega}_n/\omega_0}$ satisfy the requirement mentioned at the end of the previous section: an exciting pulse as long as few hundred optical cycles is much shorter the scale which quantized modes observably modify the atom’s dynamics. This means that the number of modes that we into account has negligible influence on the time of the atom, it mainly the resolution of the that determined by value $N.$ requiring 100 frequency (points on the graphs) in an interval of “length” $\nu,$ the value of $N=3000$ means covering the of $[0, densely. For of cases considered, this is no harmonics of higher order than for the strongly detuned excitation shown in Fig. there are ten times more observable harmonics and value of $N$ had to be increased correspondingly. an indication of the correctness of the factorization described above, let us mention that without the initial conditions $\langle (where $\tilde{\omega}_0=\omega_0$) lead Rabi between $0$ the atomic with a frequency very close to $\Omega_0\sqrt{n+1}.$ The time evolution of photon number expectation values is shown in Figs. \[nexpfig1\] and we see, the longer exciting pulse is, the pronounced can be observed. As expect, peaks multiples the frequency of the exciting For long enough pulses (see \[nexpfig2\]) and higher harmonic can be perceived between the consecutive peaks: the at odd multiples of $\nu$ are considerably than those corresponding to even harmonics. We will return to this point the next
Technically, this approach, wiTh $N$ modes beIng taKen IntO aCcouNt, meAns following thE DynaMics of $7N+3$ real variables. AcCordiNg TO our EXpErienCe, for moST oF OUr rEsUlTs, tHe FReQuencY inTerval $[0, 30\nU]$ with $N=3000$ togeTheR wIth the realisTIc Assumption Of $\OMega_n/\omega_0=0.001\sqRt{\tIlde{\omEgA}_n/\oMEga_0}$ saTisFy the RequirEMent meNtioned at ThE End of tHE previoUS SeCtioN: even an exciting puLSe AS long as few hundRed optIcAL cYCLes Is mUch shorter ThAn the TIme scalE On WHICh tHE quantized modEs observablY ModIfy the AtOm’s DYnamicS. This MeANs tHat the numbeR of mOdes that wE take iNTo accouNT has negLigiblE inFluEnce ON tHe TimE eVOluTIoN of THe aTom, it is mAiNlY the rEsolUTION of tHe sPectRa thaT is determined By tHe vaLUe oF $N.$ By rEquirIng 100 fReQuencY valueS (poinTs On the graphs) in an InteRval of “lenGth” $\Nu,$ The VaLue of $n=3000$ Means cOveRinG the intErval of $[0, 30\NU]$ suFfICIEnTly densely. For most oF tHE CaSes we conSidereD, ThIs IS approprIaTe, sInce NO HarmoNics OF hIgher ordEr than 30 APpEaR. HoweveR, fOr the sTrOngLy dEtuneD ExciTation Shown in FIg. \[resNOnresfig\](d), there ARe roughly ten tIMeS MOrE ObseRvaBle harmonicS and THe vaLue oF $n$ hAd tO Be incReaseD cORrESpondingly. As an indicAtIon of tHe corRectness of the FactorizatION DescribeD aboVE, lET us mention that WithoUt excitatiON ($E_0=0$), the iniTial cOnditionS $\langle N_i\RANgle=n\delTa_{i0}$ (WheRe $\tIldE{\OMeGa}_0=\omega_0$) lead to rABi flOpPing betWeeN mode $0$ anD thE atOmiC syStEm with a frEquency vErY cLoSe To $\OMega_0\sQRt{n+1}.$ The tiMe EvoLuTioN of thE Photon NumbeR expEcTaTIon Values iS ShOWN in FIgS. \[nExpfIg1\] aNd \[NexpfIg2\]. As WE caN see, the Longer the ExcITing PuLsE is, the mOre pronounced PeAks can be obSeRveD. As one CAN expect, tHese peaks appear around inTEger mulTipLes of The cArrier freQueNcy of tHe eXCiting Pulse. FOr lonG eNouGH PulseS (SEe fig. \[NeXpfig2\]) and hiGHEr hArmonIc OrdeRs, a signIficant, qualitative DIffErence can be peRceIved BETwEen THe COnsEcUTivE PEaks: the ones at odD multiples Of $\NU$ aRe considerABly BrOader thAn those CorreSPonding To even harMonics. We wIlL retURN to This point iN the next Section.
Technically, this approach , with $N$ mode s b ein gtake n in to account, me a ns f ollowing the dynamicsof $7 N+ 3 $ re a lvaria bles. A c co r d ing t oour e x pe rienc e,for mos t of our r esu lt s, the frequ e nc y interval $[ 0, 30\nu]$ w ith $N=30 00 $ t o gethe r w ith t he rea l isticassumptio no f $\Om e ga_n/\o m e ga _0=0 .001\sqrt{\tilde{ \ om e ga}_n/\omega_0 }$ sat is f yt h e r equ irement me nt ioned at thee nd o f th e previous sec tion: evena n e xcitin gpul s e as l ong a sf ewhundred opt ical cycles i s much shorter than th e time sc ale onw hi ch th eq uan t iz edm ode s observ ab ly modi fy t h e a tom’ s d ynam ics.This means th atthen umb er of mode s th at we t ake in to ac co unt has negligi bleinfluence on t heti me ev o lution of th e atom, it ism ain ly t h eresolution of thesp e c tr a that i s dete r mi ne d by theva lue of$ N .$ By req u ir ing 100freque n cy v alues ( po ints o nthe gr aphs) in a n inte rval of“leng t h” $\nu,$ thev alue of $N=30 0 0$ m ea n s co ver ing the int erva l of$[0, 30 \nu ] $ suf ficie nt l yd ensely. For most of t he cas es we considered,this is ap p r o priate,sinc e n o harmonics ofhighe r order th a n 30 app ear.However, for thes t rongly d etu ned ex cit a t io n shown in Fi g .  \[r es nonresf ig\ ](d), t her e a rerou gh ly ten ti mes more o bs er va ble harm o nics and t heva lue of $ N $ hadto be inc re as e d c orrespo n di n g ly. A san i ndi ca tionof t h e c orrectn ess of th e f a ctor iz at ion des cribed above, l et us ment io n t hat wi t h out exci tation ($E_0=0$), the i n itial c ond ition s $\ langle N_ i\r angle= n\d e lta_{i 0}$ (w here$\ til d e {\ome g a }_ 0=\ om ega_0$) le a d to Rabi f lopp ing bet ween mode $0$ andt heatomic system wi th a f re que n cy ver yc los e to $\Omega_0\sq rt{n+1}.$ T h etime evolu t ion o f the p hoton n umber expecta tion valu es is sho wn inF i gs.  \[nexpfig 1\] and\[nexpfig 2 \]. A s w e can se e, the l ong er th e exci t ing puls e is,th e more pron ou nced pea ks can be observed. Asone ca n exp ect , these p eak s ap pear arou nd i nteger mul tip les of t hec arrie r fr e qu enc y of t he e x citing pu l se . F o r l ong enoughp u l ses (see Fi g . \[ne xpfi g2\]) and higherh armonic orders , as i gni fic a nt,qu alitative diff ere nc e can be p er ceived betw een theco n secut ive pe aks: t he ones a to dd mul tipl esof $\nu$are c o nsidera bl yb roader tha nthosecorres p ondi n g to even harmoni cs. W e willr etu rn to t his poi n t in the nextsection.
Technically, this_approach, with_$N$ modes being taken_into account,_means_following the_dynamics_of $7N+3$ real_variables. According to_our experience, for most_of our results,_the_frequency interval $[0, 30\nu]$ with $N=3000$ together with the realistic assumption of $\Omega_n/\omega_0=0.001\sqrt{\tilde{\omega}_n/\omega_0}$ satisfy_the_requirement mentioned_at_the_end of the previous section:_even an exciting pulse as_long as_few hundred optical cycles is much shorter than_the_time scale on_which the quantized modes observably modify the atom’s dynamics._This means that the number of_modes that we_take_into_account has negligible influence_on the time evolution of the_atom, it is mainly the resolution_of the spectra that is determined by_the value of $N.$ By requiring_100 frequency values (points on_the graphs)_in an interval of “length”_$\nu,$ the value_of $N=3000$_means covering the_interval of $[0, 30\nu]$ sufficiently densely._For most of_the cases we considered, this is_appropriate,_since no harmonics_of_higher_order than_30 appear. However,_for_the strongly_detuned_excitation shown in Fig. \[resnonresfig\](d), there are_roughly_ten times more observable harmonics and the_value of $N$ had_to_be increased correspondingly. As an_indication of the correctness of_the factorization described above, let us_mention that_without excitation_($E_0=0$), the initial conditions $\langle N_i\rangle=n\delta_{i0}$ (where $\tilde{\omega}_0=\omega_0$) lead to Rabi_flopping between mode $0$ and the_atomic system with a_frequency very_close_to $\Omega_0\sqrt{n+1}.$ The time_evolution_of the_photon number expectation values is shown in_Figs. \[nexpfig1\] and_\[nexpfig2\]. As we can see, the_longer the exciting pulse_is,_the more pronounced peaks can be_observed. As one can expect, these_peaks appear around integer multiples_of_the_carrier frequency of the exciting_pulse. For long enough pulses (see_Fig. \[nexpfig2\]) and higher_harmonic orders, a significant, qualitative difference can_be_perceived between the consecutive peaks: the_ones_at odd multiples of $\nu$ are_considerably_broader_than those corresponding to even_harmonics. We will return to this_point in the next section.
}}^{(n)}_\text{occ}$ is constructed as before and the transformed density matrix $\tilde{\mathbf{P}}^{(n)}$ is calculated from the transformed eigenvector matrix $\tilde{\mathbf{C}}^{(n)}$, $$\begin{aligned} \tilde{\mathbf{P}}^{(n)} = 2 \hspace{0.1cm} \tilde{\mathbf{C}}^{(n)}_\text{occ} \left(\tilde{\mathbf{C}}^{(n)}_\text{occ}\right)^\mathsf{T} \label{eq:P_T_construction}\end{aligned}$$ The density matrix $\mathbf{P}^{(n)}$ in the original basis $B$ can be recovered by a back-transformation of $\tilde{\mathbf{P}}^{(n)}$, $$\begin{aligned} \mathbf{P}^{(n)} = \mathbf{W}^\mathsf{T} \tilde{\mathbf{P}}^{(n)} \mathbf{W}. \label{eq:p_backtransform}\end{aligned}$$ This back-transformation is necessary because it would be inefficient to evaluate a new transformed Fock matrix $\tilde{\mathbf{F}}^{(n+1)}$ from the transformed density matrix $\tilde{\mathbf{P}}^{(n)}$ directly because of the 4-index transformation required for Eqs.  and. Hence, the calculation of the new transformed Fock matrix $\tilde{\mathbf{F}}^{(n+1)}$ may be more efficiently achieved by a sequence of backward and forward transformations, $$\begin{aligned} \tilde{\mathbf{F}}^{(n+1)} = \mathbf{W} \mathbf{F}{{\left[{\scriptstyle \mathbf{W}^\mathsf{T} \tilde{\mathbf{P}}^{(n)} \mathbf{W}}\right]}} \mathbf{W}^\mathsf{T}.\end{aligned}$$ Partitioning of the system {#sec:part} -------------------------- In most of the following, we consider a subdivision of a molecular system into two parts, denoted *subsystem* ($\mathcal{S}$) and *environment* ($\mathcal{E}$). However, an extension to an arbitrary number of subsystems (including hierarchical subsystem nesting) is also discussed. In this work, subsystem and environment are chosen according to a partitioning of the atom-centered basis set $B$ into the subsets $B_\mathcal{S}$ and $B_\mathcal{E}$, such that $$\begin{aligned} B_\mathcal{S} \cup B_\mathcal{E} &=
} } ^{(n)}_\text{occ}$ is constructed as before and the transformed density matrix $ \tilde{\mathbf{P}}^{(n)}$ is account from the transformed eigenvector matrix $ \tilde{\mathbf{C}}^{(n)}$, $ $ \begin{aligned } \tilde{\mathbf{P}}^{(n) } = 2 \hspace{0.1 cm } \tilde{\mathbf{C}}^{(n)}_\text{occ } \left(\tilde{\mathbf{C}}^{(n)}_\text{occ}\right)^\mathsf{T } \label{eq: P_T_construction}\end{aligned}$$ The concentration matrix $ \mathbf{P}^{(n)}$ in the original basis $ B$ can be recovered by a binding - transformation of $ \tilde{\mathbf{P}}^{(n)}$, $ $ \begin{aligned } \mathbf{P}^{(n) } = \mathbf{W}^\mathsf{T } \tilde{\mathbf{P}}^{(n) } \mathbf{W }. \label{eq: p_backtransform}\end{aligned}$$ This spinal column - transformation is necessary because it would be inefficient to evaluate a modern translate Fock matrix $ \tilde{\mathbf{F}}^{(n+1)}$ from the transformed concentration matrix $ \tilde{\mathbf{P}}^{(n)}$ immediately because of the 4 - index transformation required for Eqs.   and. therefore, the calculation of the new transformed Fock matrix $ \tilde{\mathbf{F}}^{(n+1)}$ may be more efficiently achieve by a sequence of backward and forward transformations, $ $ \begin{aligned } \tilde{\mathbf{F}}^{(n+1) } = \mathbf{W } \mathbf{F}{{\left[{\scriptstyle \mathbf{W}^\mathsf{T } \tilde{\mathbf{P}}^{(n) } \mathbf{W}}\right ] } } \mathbf{W}^\mathsf{T}.\end{aligned}$$ Partitioning of the system { # sec: part } -------------------------- In most of the following, we consider a branch of a molecular system into two parts, denoted * subsystem * ($ \mathcal{S}$) and * environment * ($ \mathcal{E}$). However, an annex to an arbitrary number of subsystems (including hierarchical subsystem nesting) is also discourse. In this work, subsystem and environment are chosen according to a partitioning of the atom - centered footing set $ B$ into the subsets $ B_\mathcal{S}$ and $ B_\mathcal{E}$, such that $ $ \begin{aligned } B_\mathcal{S } \cup B_\mathcal{E } & =
}}^{(n)}_\tedt{occ}$ is constructed as nefore and the ttabsformxd densjty matrkx $\tilde{\mathbf{P}}^{(n)}$ is calculatxd feom tye transformed eigenveztor matrpx $\tilde{\mqthbh{C}}^{(n)}$, $$\begin{aligned} \vjlde{\matmyf{P}}^{(n)} = 2 \hspccx{0.1cm} \tilde{\mathbf{G}}^{(n)}_\text{occ} \laft(\tilde{\mathbf{W}}^{(n)}_\gert{occ}\right)^\mathsf{T} \label{eq:P_T_constructyon}\end{akihned}$$ The densijy mauriv $\mafhbf{P}^{(n)}$ in the original basis $B$ can be recmvered by a bsck-transformation of $\tilde{\lathhf{P}}^{(n)}$, $$\begin{aligned} \mwthbf{P}^{(n)} = \majgbf{R}^\nathsf{T} \tildd{\mathbf{P}}^{(n)} \mathbf{W}. \labem{eq:p_backtransform}\end{aligned}$$ Thir bacl-transformqtuon hs necessarb becalse it would nv ineffhcient yo evaluate a kew tcansdormed Fock matrix $\tinde{\mathbf{F}}^{(n+1)}$ from tre transfmrjed density matriz $\rilde{\kathtf{P}}^{(n)}$ eirdctmy bscause of the 4-index transformarion required for Eas.  qnd. Hence, the calcujaeion of the new transformed Fock matrix $\timde{\mathbf{F}}^{(n+1)}$ may be more efficiently achieved by a seqtence of backward and forward transformations, $$\beghn{alijndd} \tnode{\magybv{F}}^{(n+1)} = \mathbf{W} \mathbf{F}{{\left[{\scriptstyle \mathbf{W}^\maegsg{T} \tilde{\mathbf{P}}^{(n)} \mathbf{W}}\rignt]}} \msjhbf{W}^\mathsf{T}.\ena{alignzs}$$ Pzrtitioning of the system {#sec:pqrt} -------------------------- In mosu of yhe following, we consider a subdivision if a molecular syscem into two parjs, denpted *subsystem* ($\mathcal{S}$) and *environmenh* ($\mathcal{S}$). However, an extevsipn to an arbitrary number of subsystens (iucluding hietarchicwl subsystfm nesting) is also discudsed. Iu thiv work, subdystem and environment are chosxi according tp d pdrtitionnng of the atom-centqred basis set $B$ into the rubsets $B_\mzthcal{S}$ and $B_\mathcwl{E}$, such that $$\hegin{aligned} U_\mathcal{S} \cup B_\marhcal{E} &=
}}^{(n)}_\text{occ}$ is constructed as before and the matrix is calculated the transformed eigenvector 2 \tilde{\mathbf{C}}^{(n)}_\text{occ} \left(\tilde{\mathbf{C}}^{(n)}_\text{occ}\right)^\mathsf{T} \label{eq:P_T_construction}\end{aligned}$$ density matrix $\mathbf{P}^{(n)}$ the original basis $B$ can be by a back-transformation of $\tilde{\mathbf{P}}^{(n)}$, $$\begin{aligned} \mathbf{P}^{(n)} = \mathbf{W}^\mathsf{T} \tilde{\mathbf{P}}^{(n)} \mathbf{W}. \label{eq:p_backtransform}\end{aligned}$$ This is necessary because it would be inefficient to evaluate a new transformed Fock $\tilde{\mathbf{F}}^{(n+1)}$ the density $\tilde{\mathbf{P}}^{(n)}$ directly because of the 4-index transformation required for Eqs. and. Hence, the calculation of the transformed Fock matrix $\tilde{\mathbf{F}}^{(n+1)}$ may be more efficiently by a sequence of and forward transformations, $$\begin{aligned} \tilde{\mathbf{F}}^{(n+1)} \mathbf{W} \mathbf{W}^\mathsf{T} \tilde{\mathbf{P}}^{(n)} \mathbf{W}^\mathsf{T}.\end{aligned}$$ of system {#sec:part} -------------------------- most of the following, we consider a subdivision of a molecular system into two parts, denoted *subsystem* and *environment* an extension an number subsystems (including hierarchical is also discussed. In this work, are chosen according to a partitioning of the basis set into the subsets $B_\mathcal{S}$ and $B_\mathcal{E}$, that $$\begin{aligned} B_\mathcal{S} \cup B_\mathcal{E} &=
}}^{(n)}_\text{occ}$ is constructed as beFore and the TransForMed DeNsitY matRix $\tilde{\mathbf{p}}^{(N)}$ is cAlculated from the transfOrmed EiGEnveCToR matrIx $\tilde{\MAtHBF{C}}^{(n)}$, $$\BeGiN{alIgNEd} \Tilde{\MatHbf{P}}^{(n)} = 2 \hsPace{0.1cm} \tildE{\maThBf{C}}^{(n)}_\text{occ} \lEFt(\Tilde{\mathbF{C}}^{(n)}_\Text{occ}\right)^\MatHsf{T} \laBeL{eq:p_t_consTruCtion}\End{aliGNed}$$ The Density maTrIX $\mathbF{p}^{(n)}$ in the ORIgInal Basis $B$ can be recoveREd BY a back-transforMation Of $\TIlDE{\MatHbf{p}}^{(n)}$, $$\begin{aliGnEd} \matHBf{P}^{(n)} = \matHBf{w}^\MAThsF{t} \tilde{\mathbf{P}}^{(N)} \mathbf{W}. \labEL{eq:P_backtRaNsfORm}\end{aLigneD}$$ THIs bAck-transforMatiOn is necesSary beCAuse it wOUld be inEfficiEnt To eValuATe A nEw tRaNSfoRMeD FoCK maTrix $\tildE{\mAtHbf{F}}^{(n+1)}$ From THE TRansForMed dEnsitY matrix $\tilde{\mAthBf{P}}^{(n)}$ DIreCtly bEcausE of tHe 4-Index TransfOrmatIoN required for Eqs.  And. HEnce, the caLcuLaTioN oF the nEW transForMed fock matRix $\tildE{\MatHbF{f}}^{(N+1)}$ MaY be more efficiently AcHIEvEd by a seqUence oF BaCkWArd and foRwArd TranSFOrmatIons, $$\BEgIn{aligneD} \tilde{\MAtHbF{F}}^{(n+1)} = \mathBf{w} \mathbF{F}{{\LefT[{\scRiptsTYle \mAthbf{W}^\Mathsf{T} \tIlde{\mAThbf{P}}^{(n)} \mathbf{W}}\rIGht]}} \mathbf{W}^\matHSf{t}.\ENd{ALignEd}$$ PArtitioning Of thE SystEm {#seC:PaRt} -------------------------- IN Most oF the fOlLOwINg, we consider a subdivIsIon of a MolecUlar system intO two parts, dENOTed *subsyStem* ($\MAtHCal{S}$) and *environMent* ($\mAthcal{E}$). HowEVer, an extEnsioN to an arbItrary numBER of subsySteMs (iNclUdiNG HiErarchical subSYStem NeSting) is AlsO discusSed. in tHis WorK, sUbsystem aNd enviroNmEnT aRe ChoSen acCOrding to A pArtItIonIng of THe atom-CenteRed bAsIs SEt $B$ Into the SUbSETs $B_\mAtHcAl{S}$ aNd $B_\MaThcal{e}$, sucH ThaT $$\begin{aLigned} B_\maThcAL{S} \cuP B_\MaThcal{E} &=
}}^{(n)}_\text{occ}$ is co nstructedas be for e a nd the tra nsformed densi t y ma trix $\tilde{\mathbf{P }}^{( n) } $ is ca lcula ted fro m t h e tr an sf orm ed ei genve cto r matri x $\tilde{ \ma th bf{C}}^{(n)} $ ,$$\begin{a lig ned} \tilde{ \ma thbf{P }} ^{( n )} =2 \ hspac e{0.1c m } \til de{\mathb f{ C }}^{(n ) }_\text { o cc } \l eft(\tilde{\mathb f {C } }^{(n)}_\text{ occ}\r ig h t) ^ \ mat hsf {T} \label {e q:P_T _ constru c ti o n } \en d {aligned}$$ T he densitym atr ix $\m at hbf { P}^{(n )}$ i nt heoriginal ba sis$B$ can b e reco v ered by a back- transf orm ati on o f $ \t ild e{ \ mat h bf {P} } ^{( n)}$, $$ \b eg in{al igne d } \ math bf{ P}^{ (n)}= \mathbf{W}^ \ma thsf { T}\tild e{\ma thbf {P }}^{( n)} \m athbf {W }. \label{eq:p_ back transform }\e nd {al ig ned}$ $ Thisbac k-t ransfor mationi s n ec e s s ar y because it would b e in efficien t to e v al ua t e a newtr ans form e d Fock mat r ix $\tilde {\math b f{ F} }^{(n+1 )} $ from t hetra nsfor m ed d ensity matrix$\til d e{\mathbf{P}}^ { (n)}$ directl y b e c au s e of th e 4-index t rans f orma tion re qui r ed fo r Eqs .an d . Hence, the calcul at ion of thenew transform ed Fock ma t r i x $\tild e{\m a th b f{F}}^{(n+1)}$ maybe more ef f iciently achi eved bya sequenc e of backw ard an d f orw a r dtransformatio n s , $$ \b egin{al ign ed} \ti lde {\m ath bf{ F} }^{(n+1)} = \math bf {W }\m ath bf{F} { {\left[{ \s cri pt sty le \m a thbf{W }^\ma thsf {T }\ til de{\mat h bf { P }}^{ (n )} \ma thb f{ W}}\r ight ] }}\mathbf {W}^\math sf{ T }.\e nd {a ligned} $$ Partition in g of the s ys tem {#sec : p art} --- ----------------------- In mos t o f the fol lowing, w e c onside r a subdiv isionof amo lec u l ar sy s t em in to two parts , den oted*s ubsy stem* ( $\mathcal{S}$) and *en vironment* ($ \ma thca l { E} $). Ho w eve r, ane x tension to an a rbitrary n um b er of subsys t ems ( includi ng hier archi c al subs ystem nes ting) isal so d i s cus sed. In t his work , subsyst e m and en viron men t arech ose n acc ording toa par tition in g of t he at om -centere d basis set $B$ into th e subs ets $ B_\ mathcal{S }$a nd$B_\mathc al{E }$, such t hat $$ \begi n{a l igned } B_ \ ma thc a l{S}\cup B_\mathca l {E } & =
}}^{(n)}_\text{occ}$ is_constructed as_before and the transformed_density matrix_$\tilde{\mathbf{P}}^{(n)}$_is calculated_from_the transformed eigenvector_matrix $\tilde{\mathbf{C}}^{(n)}$, $$\begin{aligned} \tilde{\mathbf{P}}^{(n)}_= 2 \hspace{0.1cm} \tilde{\mathbf{C}}^{(n)}_\text{occ}_\left(\tilde{\mathbf{C}}^{(n)}_\text{occ}\right)^\mathsf{T} \label{eq:P_T_construction}\end{aligned}$$ The_density_matrix $\mathbf{P}^{(n)}$ in the original basis $B$ can be recovered by a back-transformation of_$\tilde{\mathbf{P}}^{(n)}$,_$$\begin{aligned} \mathbf{P}^{(n)} =_\mathbf{W}^\mathsf{T}_\tilde{\mathbf{P}}^{(n)}_\mathbf{W}. \label{eq:p_backtransform}\end{aligned}$$ This back-transformation is_necessary because it would be_inefficient to_evaluate a new transformed Fock matrix $\tilde{\mathbf{F}}^{(n+1)}$ from_the_transformed density matrix_$\tilde{\mathbf{P}}^{(n)}$ directly because of the 4-index transformation required for_Eqs.  and. Hence, the calculation of_the new transformed_Fock_matrix_$\tilde{\mathbf{F}}^{(n+1)}$ may be more_efficiently achieved by a sequence of_backward and forward transformations, $$\begin{aligned} \tilde{\mathbf{F}}^{(n+1)} =_\mathbf{W} \mathbf{F}{{\left[{\scriptstyle \mathbf{W}^\mathsf{T} \tilde{\mathbf{P}}^{(n)} \mathbf{W}}\right]}} \mathbf{W}^\mathsf{T}.\end{aligned}$$ Partitioning of_the system {#sec:part} -------------------------- In most of the_following, we consider a subdivision_of a_molecular system into two parts,_denoted *subsystem* ($\mathcal{S}$)_and *environment*_($\mathcal{E}$). However, an_extension to an arbitrary number of_subsystems (including hierarchical_subsystem nesting) is also discussed. In this_work,_subsystem and environment_are_chosen_according to_a partitioning of_the_atom-centered basis_set_$B$ into the subsets $B_\mathcal{S}$ and_$B_\mathcal{E}$,_such that $$\begin{aligned} B_\mathcal{S} \cup B_\mathcal{E} &=
bm \psi, \bm \tau, \bm \beta)$ is defined as the corresponding optimization objective. Denote the estimates of the complex gain, the AoA, and the path delay at the $n$th iteration as $\bm \beta^{(n)} $, $\bm \psi^{(n)} $, and $\bm \tau^{(n)} $, respectively. Utilizing the majorization-minorization (MM) iterative approach [@log-sum-2; @log-sum-3] and similar to [@SFW-BS-GC], the optimization in can be transformed into minimizing the surrogate function as $$\begin{aligned} \min_{\bm \psi, \bm \tau } \ \underbrace{ - \mathbf y_k^H \mathbf W^H \mathbf P_k (\bm \psi, \bm \tau) \Big( \mathbf P_k^H (\bm \psi, \bm \tau) \mathbf W \, \mathbf W^H\mathbf P_k (\bm \psi, \bm \tau) + \lambda ^{-1} \mathbf D^{(n)} \Big)^{-1} \mathbf P_k^H (\bm \psi, \bm \tau) \mathbf W\mathbf y_k + C(\bm \beta^{(n)}) }_{S_1 (\bm \psi, \bm \tau)} , \label{Sl2}\end{aligned}$$ where ${S_1 (\bm \psi, \bm \tau)}$ is as defined, $$\begin{aligned} \mathbf D^{(n)} & \triangleq {\textup{diag}}\Bigg\{ \frac{ 1 } { \big| [\bm \beta^{(n)}]_1 \big| ^2 + \epsilon } ,\dots, \frac{ 1 } { \big| [\bm \beta^{(n)}]_{L_k} \big| ^2 + \epsilon } \Bigg\} , \label{Dn}\end{aligned}$$ and $C(\bm \beta^{(n)})$ is the constant independent of $\bm \psi$, $\bm \tau$, and $\bm \beta$. For given $\bm \psi $ and $\bm \tau$, the optimal value of $\bm \beta$ can be obtained as $$\begin{aligned} \bm \beta^* (\bm \psi, \bm \tau) & =
bm \psi, \bm \tau, \bm \beta)$ is defined as the corresponding optimization objective. Denote the estimates of the complex amplification, the AoA, and the way delay at the $ n$th iteration as $ \bm \beta^{(n) } $, $ \bm \psi^{(n) } $, and $ \bm \tau^{(n) } $, respectively. Utilizing the majorization - minorization (MM) iterative approach [ @log - sum-2; @log - sum-3 ] and exchangeable to [ @SFW - BS - GC ], the optimization in can be transform into minimizing the surrogate routine as $ $ \begin{aligned } \min_{\bm \psi, \bm \tau } \ \underbrace { - \mathbf y_k^H \mathbf W^H \mathbf P_k (\bm \psi, \bm \tau) \Big ( \mathbf P_k^H (\bm \psi, \bm \tau) \mathbf W \, \mathbf W^H\mathbf P_k (\bm \psi, \bm \tau) + \lambda ^{-1 } \mathbf D^{(n) } \Big)^{-1 } \mathbf P_k^H (\bm \psi, \bm \tau) \mathbf W\mathbf y_k + C(\bm \beta^{(n) }) } _ { S_1 (\bm \psi, \bm \tau) } , \label{Sl2}\end{aligned}$$ where $ { S_1 (\bm \psi, \bm \tau)}$ is as define, $ $ \begin{aligned } \mathbf D^{(n) } & \triangleq { \textup{diag}}\Bigg\ { \frac { 1 } { \big| [ \bm \beta^{(n)}]_1 \big| ^2 + \epsilon } , \dots, \frac { 1 } { \big| [ \bm \beta^{(n)}]_{L_k } \big| ^2 + \epsilon } \Bigg\ } , \label{Dn}\end{aligned}$$ and $ C(\bm \beta^{(n)})$ is the constant mugwump of $ \bm \psi$, $ \bm \tau$, and $ \bm \beta$. For given $ \bm \psi $ and $ \bm \tau$, the optimal value of $ \bm \beta$ can be prevail as $ $ \begin{aligned } \bm \beta^ * (\bm \psi, \bm \tau) & =
bm \osi, \bm \tau, \bm \beta)$ is denined as the cortewpondiig optijization objective. Denote the estimatxs od the complex gain, the AoA, xnd the pwth delat at rhe $n$th itxdation as $\bm \gcta^{(n)} $, $\um \psi^{(n)} $, and $\bm \tau^{(n)} $, res[ectively. Utilhzkny the majorization-minorization (MM) itqrative aoproach [@log-sum-2; @log-xtm-3] ahd similar to [@SFW-BS-GC], the optimizafion in can be transgormed into minimizing the surgogate function as $$\begin{aligngs} \mig_{\vm \psi, \bm \taj } \ \underbrace{ - \mathbf y_i^H \mathbf W^H \mathbf P_k (\bm \psi, \bo \tau) \Big( \mathbf P_j^H (\hk \psi, \bm \tan) \mathff W \, \mathbf W^H\mathbx P_k (\bm \psi, \bm \tau) + \lakbde ^{-1} \mqthbf D^{(n)} \Big)^{-1} \mathbf P_k^I (\bm \psi, \bm \tau) \mathff W\mathbx v_k + C(\bm \beta^{(n)}) }_{S_1 (\bm \psi, \bn \rau)} , \latel{Sn2}\end{xoigved}$$ wiers ${S_1 (\bm \psm, \bm \tau)}$ is as defined, $$\begin{aligned} \mathbf D^{(g)} & \triangleq {\testup{diwg}}\Figg\{ \frac{ 1 } { \big| [\bm \beta^{(n)}]_1 \big| ^2 + \epsilon } ,\gota, \frac{ 1 } { \big| [\bm \beta^{(n)}]_{L_k} \big| ^2 + \epsilon } \Bigg\} , \lahel{Dn}\end{ajigned}$$ and $C(\bm \beta^{(n)})$ is the constant independent mf $\bm \osi$, $\bi \rak$, and $\bm \beta$. For given $\bm \psi $ and $\bm \tau$, thq oktikal value of $\bi \beta$ can ne onjained as $$\begiv{alignzs} \bj \beta^* (\bm \psi, \bm \twu) & =
bm \psi, \bm \tau, \bm \beta)$ is the optimization objective. the estimates of and path delay at $n$th iteration as \beta^{(n)} $, $\bm \psi^{(n)} $, and \tau^{(n)} $, respectively. Utilizing the majorization-minorization (MM) iterative approach [@log-sum-2; @log-sum-3] and similar [@SFW-BS-GC], the optimization in can be transformed into minimizing the surrogate function as \min_{\bm \bm } \underbrace{ - \mathbf y_k^H \mathbf W^H \mathbf P_k (\bm \psi, \bm \tau) \Big( \mathbf P_k^H (\bm \bm \tau) \mathbf W \, \mathbf W^H\mathbf P_k \psi, \bm \tau) + ^{-1} \mathbf D^{(n)} \Big)^{-1} \mathbf (\bm \bm \tau) W\mathbf + \beta^{(n)}) }_{S_1 (\bm \bm \tau)} , \label{Sl2}\end{aligned}$$ where ${S_1 (\bm \psi, \bm \tau)}$ is as defined, $$\begin{aligned} \mathbf D^{(n)} & {\textup{diag}}\Bigg\{ \frac{ { \big| \beta^{(n)}]_1 ^2 \epsilon } ,\dots, } { \big| [\bm \beta^{(n)}]_{L_k} \big| } \Bigg\} , \label{Dn}\end{aligned}$$ and $C(\bm \beta^{(n)})$ is constant independent $\bm \psi$, $\bm \tau$, and $\bm For given $\bm \psi $ and $\bm \tau$, optimal value of $\bm \beta$ can be obtained as $$\begin{aligned} \bm \beta^* (\bm \psi, \bm =
bm \psi, \bm \tau, \bm \beta)$ is defined As the correSpondIng OptImIzatIon oBjective. Denote THe esTimates of the complex gaiN, the AOA, ANd thE PaTh delAy at the $N$Th ITEraTiOn As $\bM \bETa^{(N)} $, $\bm \psI^{(n)} $, aNd $\bm \tau^{(N)} $, respectivEly. utIlizing the maJOrIzation-minOriZation (MM) iterAtiVe apprOaCh [@lOG-sum-2; @lOg-sUm-3] and SimilaR To [@SFW-Bs-GC], the optImIZation IN can be tRANsFormEd into minimizing tHE sURrogate functioN as $$\begIn{ALiGNEd} \mIn_{\bM \psi, \bm \tau } \ \uNdErbraCE{ - \mathbf Y_K^H \MATHbf w^h \mathbf P_k (\bm \psI, \bm \tau) \Big( \maTHbf p_k^H (\bm \pSi, \Bm \tAU) \mathbF W \, \matHbF w^H\mAthbf P_k (\bm \psI, \bm \tAu) + \lambda ^{-1} \mAthbf D^{(N)} \big)^{-1} \mathBF P_k^H (\bm \pSi, \bm \taU) \maThbF W\maTHbF y_K + C(\bM \bETa^{(n)}) }_{s_1 (\Bm \Psi, \BM \taU)} , \label{Sl2}\EnD{aLigneD}$$ wheRE ${s_1 (\BM \psi, \Bm \tAu)}$ is As defIned, $$\begin{aligNed} \MathBF D^{(n)} & \TrianGleq {\tExtuP{dIag}}\BiGg\{ \frac{ 1 } { \Big| [\bm \BeTa^{(n)}]_1 \big| ^2 + \epsilon } ,\doTs, \frAc{ 1 } { \big| [\bm \beTa^{(n)}]_{l_k} \Big| ^2 + \EpSilon } \bIgg\} , \labEl{DN}\enD{aligneD}$$ and $C(\bm \BEta^{(N)})$ iS THE cOnstant independent Of $\BM \PsI$, $\bm \tau$, anD $\bm \betA$. foR gIVen $\bm \psi $ AnD $\bm \Tau$, tHE OptimAl vaLUe Of $\bm \beta$ Can be oBTaInEd as $$\begIn{AligneD} \bM \beTa^* (\bM \psi, \bM \Tau) & =
bm \psi, \bm \tau, \bm \be ta)$ is de fined as th ecorr espo nding optimiza t ionobjective. Denote the esti ma t es o f t he co mplex g a in , the A oA , a nd th e pat h d elay at the $n$th it er ation as $\b m \ beta^{(n)} $, $\bm \psi^{ (n) } $, a nd $\ b m \ta u^{ (n)}$, res p ective ly. Utili zi n g them ajoriza t i on -min orization (MM) it e ra t ive approach [ @log-s um - 2; @ log -su m-3] and s im ilart o [@SFW - BS - G C ],t he optimizati on in can b e tr ansfor me d i n to min imizi ng the surrogatefunc tion as $ $\begi n {aligne d } \min_ {\bm \ psi , \ bm \ t au } \\u n der b ra ce{ - \ mathbf y _k ^H \mat hbfW ^ H \mat hbf P_k (\bm \psi, \bm \t au) \Bi g ( \ mathb f P_k ^H ( \b m \ps i, \bm \tau )\mathbf W \, \m athb f W^H\mat hbf P _k(\ bm \p s i, \bm \t au) + \lam bda ^{- 1 } \ ma t h b fD^{(n)} \Big)^{-1} \ m a th bf P_k^H (\bm\ ps i, \bm \tau )\ma thbf W \math bf y _ k+ C(\bm\beta^ { (n )} ) }_{S_ 1(\bm \ ps i,\bm \tau ) } ,\label {Sl2}\en d{ali g ned}$$ where $ { S_1 (\bm \psi , \ b m \ t au)} $ i s as define d, $ $ \beg in{a l ig ned } \mat hbf D ^{ ( n) } & \triangleq {\tex tu p{diag }}\Bi gg\{ \frac{ 1 } { \big| [ \ bm \beta ^{(n ) }] _ 1 \big| ^2 + \ epsil on } ,\dot s , \frac{ 1 }{ \big|[\bm \bet a ^ {(n)}]_{ L_k } \ big | ^ 2 +\epsilon } \B i g g\},\label{ Dn} \end{al ign ed} $$and $ C(\bm \be ta^{(n)} )$ i sth e c onsta n t indepe nd ent o f $ \bm \ p si$, $ \bm \ tau$ ,an d $\ bm \bet a $. F or g iv en $\b m \ ps i $ a nd $ \ bm\tau$,the optim alv alue o f$\bm \b eta$ can be o bt ained as $ $\ beg in{ali g n ed} \bm\beta^* (\bm \psi, \bm\ tau) &=
bm \psi,_\bm \tau,_\bm \beta)$ is defined_as the_corresponding_optimization objective. Denote_the_estimates of the_complex gain, the_AoA, and the path_delay at the_$n$th_iteration as $\bm \beta^{(n)} $, $\bm \psi^{(n)} $, and $\bm \tau^{(n)} $, respectively. Utilizing_the_majorization-minorization (MM)_iterative_approach_[@log-sum-2; @log-sum-3] and similar to_[@SFW-BS-GC], the optimization in can_be transformed_into minimizing the surrogate function as $$\begin{aligned} \min_{\bm \psi,_\bm_\tau } \ \underbrace{ -_\mathbf y_k^H \mathbf W^H \mathbf P_k (\bm \psi, \bm_\tau) \Big( \mathbf P_k^H (\bm \psi, \bm_\tau) \mathbf W_\,_\mathbf_W^H\mathbf P_k (\bm \psi,_\bm \tau) + \lambda ^{-1} \mathbf D^{(n)} \Big)^{-1}_\mathbf P_k^H (\bm \psi, \bm \tau)_\mathbf W\mathbf y_k + C(\bm \beta^{(n)}) }_{S_1 (\bm \psi, \bm_\tau)} , \label{Sl2}\end{aligned}$$ where ${S_1 (\bm \psi, \bm_\tau)}$ is as defined, $$\begin{aligned} \mathbf_D^{(n)} &_\triangleq {\textup{diag}}\Bigg\{ \frac{ 1 } { \big|_[\bm \beta^{(n)}]_1 \big|_^2 +_\epsilon } ,\dots, \frac{ 1_} { \big| [\bm \beta^{(n)}]_{L_k} \big|_^2 + \epsilon_} \Bigg\} , \label{Dn}\end{aligned}$$ and $C(\bm \beta^{(n)})$ is the_constant_independent of $\bm_\psi$,_$\bm_\tau$, and_$\bm \beta$. For_given_$\bm \psi_$_and $\bm \tau$, the optimal value_of_$\bm \beta$ can be obtained as $$\begin{aligned} \bm_\beta^* (\bm \psi, \bm_\tau)_& =
)$, the commutation relation equal to the corresponding commutator of position and momentum, that is, $[A,C]= i\hbar $. Physically, we are looking for an observable that could act as a *generator of correlations*. One can prove that such an operator $A(X,P)$ can not be expanded in a power series $\sum a_{kr} X^{k}P^{r}$ (notice that any power series can be brought to this “normal order” with all powers of $X$ at the left of all powers of $P$). In order to prove this we use the commutation relations in Eq.(\[ComRel3\]) and we can see that there exists no choice of the coefficients $a_{kr}$ that satisfy the commutation relation $[A,C]=i\hbar$. The eigenvectors of $A(X,P)$ (if they exist) and $C(X,P)$ would build two mutually unbiased bases and these two observables could be chosen as a pair of canonical conjugate coordinates for the description of the system. This choice is related to the canonical transformation of classical mechanics where the coordinates $(x,p)$ are transformed to $a(x,p)$ and $c(x,p)$ in a way to preserve the Poisson brackets, that is $\{x,p\}=\{a,c\}=1$. If we take $c(x,p)=xp$, then the conjugate coordinate is $a(x,p)=\frac{1}{2} \ln(\frac{x}{p})$. Following this suggestion we can see that the operator $$\label{AcanconjtoC} A=\frac{1}{2}(\ln X - \ln P)\,$$ at least *formally*, has the wanted commutation relation. To prove this we use Eqs.(\[commutPderivX\]) and (\[commutXderivP\]) in order to obtain $[\ln X,P]=i\hbar/X$ and $[\ln P,X]=-i\hbar/P$. With more mathematical rigour it is not clear that such an operator exists. Furthermore, the physical meaning of an observable such as $\ln X$, undefined for negative values of position, is unclear, leaving alone what would be the mysterious physical procedure to measure $A$. The question of the existence of a generator of correlations is open. We come now to the question of the existence of the eigenvectors of the correlation operator $C$, that is, to determine the
) $, the commutation relation equal to the corresponding commutator of placement and momentum, that is, $ [ deoxyadenosine monophosphate, C]= i\hbar $. Physically, we are looking for an observable that could act as a * generator of correlation coefficient *. One can prove that such an operator $ A(X, P)$ cannot be boom in a might series $ \sum a_{kr } X^{k}P^{r}$ (detect that any power series can be bring to this “ normal order ” with all powers of $ X$ at the left of all powers of $ P$). In order to prove this we practice the commutation relations in Eq.(\[ComRel3\ ]) and we can see that there exist no choice of the coefficients $ a_{kr}$ that meet the commutation relative $ [ A, C]=i\hbar$. The eigenvectors of $ A(X, P)$ (if they exist) and $ C(X, P)$ would build up two mutually unbiased bases and these two observables could be chosen as a couple of canonical conjugate coordinates for the description of the system. This choice is related to the canonical transformation of classical mechanics where the coordinates $ (x, p)$ are transformed to $ a(x, p)$ and $ c(x, p)$ in a way to preserve the Poisson brackets, that is $ \{x, p\}=\{a, c\}=1$. If we take $ c(x, p)=xp$, then the conjugate coordinate is $ a(x, p)=\frac{1}{2 } \ln(\frac{x}{p})$. Following this trace we can see that the operator $ $ \label{AcanconjtoC } A=\frac{1}{2}(\ln X - \ln P)\,$$ at least * formally *, have the want commutation relation. To prove this we use Eqs.(\[commutPderivX\ ]) and (\[commutXderivP\ ]) in order to obtain $ [ \ln X, P]=i\hbar / X$ and $ [ \ln P, X]=-i\hbar / P$. With more mathematical severity it is not clear that such an operator exists. Furthermore, the physical meaning of an observable such as $ \ln X$, undefined for negative value of position, is unclear, leaving alone what would be the cryptic physical routine to quantify $ A$. The doubt of the existence of a generator of correlations is open. We derive now to the question of the existence of the eigenvectors of the correlation coefficient operator $ C$, that is, to decide the
)$, thf commutation relation edual to the cortewpondiig commhtator ow position and momentum, that iw, $[A,C]= u\hbar $. Physically, we afe lookinh for an obstrvable that coulv act as a *gensvator if correlationx*. One can [rove that sucv xn operator $A(X,P)$ can not be expanded ig a powrr series $\sum a_{kt} X^{k}P^{g}$ (gotidv uhat any power series can be broufht to uhis “normal order” eith all powers of $X$ at thf levt of all powers ov $P$). In ordet to prove this wd use the bmmmutation relations in Eq.(\[ComRel3\]) and we cxn sez that therg sxlvts no choire of nhe coefficiekns $a_{kr}$ dhat sayisfy the commmtatimn eelation $[A,C]=i\hbar$. The emgenvectors of $A(X,P)$ (is they exhsc) and $C(X,P)$ would build tqo mujuallf uncuasdd gaxea and hheae two obssrvables coyld be chosen as a kaiw of canonical conjudaee coordinates for the description of tve aystem. This choice is rwlated to the canonicwl transfjrmation of classical mechanics where the coordindtes $(e,p)$ art bvanswirled to $a(x,p)$ and $c(x,p)$ in a way to preserve the Pjjsxok brackets, that ls $\{x,p\}=\{a,c\}=1$. If we take $c(d,p)=ck$, then the congugate cokrdinate is $a(x,p)=\fraf{1}{2} \ln(\frac{v}{p})$. Foolowing tris xuggestion we can see that rhe operator $$\oabel{AcanconjtoC} A=\yrac{1}{2}(\ln X - \ln P)\,$$ aj leasy *formally*, has the wantzd comjutation repation. To orove this we usd Eas.(\[wommutPdtfivX\]) and (\[commutXdqrivP\]) in irdex to obtxin $[\kn X,P]=i\rbar/X$ and $[\pn P,X]=-l\vbar/P$. With more mahhemajical sigour it ls not clear that such an operavpr exists. Futtharmmre, the 'hysicsl meaning of an observable such af $\ln B$, undefines for nxgative valuqs of positiot, is unclear, neaving wlonw whqt woula be the mysteroous physpccl procedyre to measure $A$. Tme qugsfion of the exiwtebce of a generayor of clrcelatymns is open. Wa cooe vpw to the qutftlon of yhe existence of the eigsnvectors of the cprvelation iperator $C$, that is, to determine the
)$, the commutation relation equal to the of and momentum, is, $[A,C]= i\hbar for observable that could as a *generator correlations*. One can prove that such operator $A(X,P)$ can not be expanded in a power series $\sum a_{kr} X^{k}P^{r}$ that any power series can be brought to this “normal order” with all of at left all powers of $P$). In order to prove this we use the commutation relations in Eq.(\[ComRel3\]) we can see that there exists no choice the coefficients $a_{kr}$ that the commutation relation $[A,C]=i\hbar$. The of (if they and would two mutually unbiased and these two observables could be chosen as a pair of canonical conjugate coordinates for the description the system. is related the transformation classical mechanics where $(x,p)$ are transformed to $a(x,p)$ and way to preserve the Poisson brackets, that is If we $c(x,p)=xp$, then the conjugate coordinate is \ln(\frac{x}{p})$. Following this suggestion we can see that operator $$\label{AcanconjtoC} A=\frac{1}{2}(\ln X - \ln P)\,$$ at least *formally*, has the wanted commutation relation. this we use Eqs.(\[commutPderivX\]) (\[commutXderivP\]) in order obtain X,P]=i\hbar/X$ $[\ln With more rigour it is not clear that such an operator exists. Furthermore, physical meaning of an observable such as $\ln X$, undefined values position, is unclear, alone what would be mysterious procedure to measure $A$. of existence of is We come now to question of the existence of eigenvectors of the correlation determine the
)$, the commutation relation equAl to the corRespoNdiNg cOmMutaTor oF position and moMEntuM, that is, $[A,C]= i\hbar $. PhysicalLy, we aRe LOokiNG fOr an oBservabLE tHAT coUlD aCt aS a *GEnEratoR of CorrelaTions*. One caN prOvE that such an oPErAtor $A(X,P)$ can Not Be expanded in A poWer serIeS $\suM A_{kr} X^{k}p^{r}$ (nOtice That anY Power sEries can bE bROught tO This “norMAL oRder” With all powers of $X$ aT ThE Left of all powerS of $P$). In OrDEr TO ProVe tHis we use thE cOmmutATion relATiONS In EQ.(\[comRel3\]) and we caN see that theRE exIsts no ChOicE Of the cOeffiCiENts $A_{kr}$ that satiSfy tHe commutaTion reLAtion $[A,C]=I\Hbar$. The EigenvEctOrs Of $A(X,p)$ (If ThEy eXiST) anD $c(X,p)$ woULd bUild two mUtUaLly unBiasED BASes aNd tHese Two obServables coulD be ChosEN as A pair Of canOnicAl ConjuGate coOrdinAtEs for the descripTion Of the systEm. THiS chOiCe is rELated tO thE caNonical TransfoRMatIoN OF ClAssical mechanics whErE THe CoordinaTes $(x,p)$ aRE tRaNSformed tO $a(X,p)$ aNd $c(x,P)$ IN a way To prESeRve the PoIsson bRAcKeTs, that iS $\{x,P\}=\{a,c\}=1$. If wE tAke $C(x,p)=Xp$, theN The cOnjugaTe coordiNate iS $A(x,p)=\frac{1}{2} \ln(\frac{x}{P})$. following this SUgGEStIOn we Can See that the oPeraTOr $$\laBel{ACAnConJToC} A=\fRac{1}{2}(\ln x - \lN p)\,$$ aT Least *formally*, has the WaNted coMmutaTion relation. TO prove this WE USe Eqs.(\[comMutPDErIVX\]) and (\[commutXdeRivP\]) iN order to obTAin $[\ln X,P]=i\Hbar/X$ And $[\ln P,X]=-i\Hbar/P$. With MORe mathemAtiCal RigOur IT Is Not clear that sUCH an oPeRator exIstS. FurtheRmoRe, tHe pHysIcAl meaning Of an obseRvAbLe SuCh aS $\ln X$, uNDefined fOr NegAtIve ValueS Of posiTion, iS uncLeAr, LEavIng alonE WhAT WoulD bE tHe mySteRiOus phYsicAL prOcedure To measure $a$. ThE QuesTiOn Of the exIstence of a genErAtor of corrElAtiOns is oPEN. We come nOw to the question of the exiSTence of The EigenVectOrs of the cOrrElatioN opERator $C$, That is, To detErMinE THe
)$, the commutation relati on equal t o the co rre sp ondi ng c ommutator of p o siti on and momentum, thatis, $ [A , C]=i \h bar $ . Physi c al l y , w ear e l oo k in g for an observ able thatcou ld act as a *g e ne rator of c orr elations*. O necan pr ov e t h at su chan op erator $A(X,P )$ can no tb e expa n ded ina po werseries $\sum a_{k r }X ^{k}P^{r}$ (no tice t ha t a n y po wer series ca nbe br o ught to th i s “no r mal order” wi th all powe r s o f $X$at th e leftof al lp owe rs of $P$). Inorder toprovet his weu se thecommut ati onrela t io ns in E q .(\ [ Co mRe l 3\] ) and we c an seethat t h e re e xis ts n o cho ice of the co eff icie n ts$a_{k r}$ t hatsa tisfy the c ommut at ion relation $[ A,C] =i\hbar$. T he ei ge nvect o rs of$A( X,P )$ (ifthey ex i st) a n d $C (X,P)$ would build t w o m utuallyunbias e dba s es and t he setwoo b serva bles co uld be c hosena sapair of c anonic al co nju gatec oord inates for the desc r iption of thes ystem. This c h oi c e i s rel ate d to the ca noni c al t rans f or mat i on of clas si c al mechanics where the c oordin ates$(x,p)$ are t ransformed t o $a(x,p) $ an d $ c (x,p)$ in a wa y topreserve t h e Poisso n bra ckets, t hat is $\ { x ,p\}=\{a ,c\ }=1 $.Ifw e t ake $c(x,p)=x p $ , th en the co nju gate co ord ina teis$a (x,p)=\fr ac{1}{2} \ ln (\ fr ac{ x}{p} ) $. Follo wi ngth issugge s tion w e can see t ha t th e opera t or $ $\la be l{ Acan con jt oC} A=\f r ac{ 1}{2}(\ ln X - \l n P ) \,$$ a tleast * formally*, ha sthe wanted c omm utatio n relation . To prove this we useE qs.(\[c omm utPde rivX \]) and ( \[c ommutX der i vP\])in ord er to o bta i n $[\l n X, P]= i\ hbar/X$ an d $[\ ln P, X] =-i\ hbar/P$ . With more mathem a tic al rigour itisnotc l ea r t h at suc ha n o p e rator exists. F urthermore ,t he physicalm ean in g of an observ ables uch as$\ln X$,undefined f or n e g ati ve valuesof posit ion, is u n clear , l eavin g a lone w ha t w ouldbe the mys terio us phy si cal pr ocedu re to meas ure $A$. The question o f theexist enc e of a ge ner a tor of corre lati ons is ope n. We come no w to t he q u es tio n of t he e x istence o f t hee i ge nvectors of t h e c orrel ati o n oper ator $C$, that is, to determine the
)$, the_commutation relation_equal to the corresponding_commutator of_position_and momentum,_that_is, $[A,C]= i\hbar_$. Physically, we_are looking for an_observable that could_act_as a *generator of correlations*. One can prove that such an operator $A(X,P)$ can_not_be expanded_in_a_power series $\sum a_{kr} X^{k}P^{r}$_(notice that any power series_can be_brought to this “normal order” with all powers_of_$X$ at the_left of all powers of $P$). In order to_prove this we use the commutation_relations in Eq.(\[ComRel3\])_and_we_can see that there_exists no choice of the coefficients_$a_{kr}$ that satisfy the commutation relation_$[A,C]=i\hbar$. The eigenvectors of $A(X,P)$ (if they exist)_and $C(X,P)$ would build two mutually_unbiased bases and these two_observables could_be chosen as a pair_of canonical conjugate_coordinates for_the description of_the system. This choice is related_to the canonical_transformation of classical mechanics where the_coordinates_$(x,p)$ are transformed_to_$a(x,p)$_and $c(x,p)$_in a way_to_preserve the_Poisson_brackets, that is $\{x,p\}=\{a,c\}=1$. If we_take_$c(x,p)=xp$, then the conjugate coordinate is $a(x,p)=\frac{1}{2} \ln(\frac{x}{p})$._Following this suggestion we_can_see that the operator_$$\label{AcanconjtoC} A=\frac{1}{2}(\ln X - \ln_P)\,$$ at least *formally*, has the_wanted commutation_relation. To_prove this we use Eqs.(\[commutPderivX\]) and (\[commutXderivP\]) in order to obtain_$[\ln X,P]=i\hbar/X$ and $[\ln P,X]=-i\hbar/P$. With_more mathematical rigour it_is not_clear_that such an_operator_exists. Furthermore,_the physical meaning of an observable such_as $\ln_X$, undefined for negative values of_position, is unclear, leaving_alone_what would be the mysterious physical_procedure to measure $A$. The question_of the existence of a_generator_of_correlations is open. We come now_to the question of the existence_of the eigenvectors_of the correlation operator $C$, that is,_to_determine the
(possible PRG’s; shown as filled circles in Figure 7c). None of these implied star-formation rates are high, compared to normal disk galaxies. For those galaxies with known  masses (see Table 2), we use these star-formation rates to compute gas-consumption timescales, assuming a Salpeter initial mass function truncated at $M$$<$0.4[M${_\odot}$]{}(Figure 7d). Only one of the galaxies in our sample, PRC C-51, could use up all of its gas in $\sim$10$^8$ years, and thus may fall into the category of “starburst” galaxies. Although the majority of the radio emission in these candidate PRGs likely arises from recent star-formation activity, the rates implied are not high compared to the total available gas in the galaxies. We conclude that polar-ring galaxies have radio continuum fluxes typical of normal (non-polar-ring) elliptical or S0 galaxies. The morphology of the extended sources, the spectral indices, and FIR fluxes of the PRG’s in our sample indicate that the radio emission is predominantly due to star-forming regions rather than nuclear activity. However, the rate of starbirth is modest — all but one of our sample galaxies have enough cool gas to fuel their star formation for at least another gigayear. ALC and LSS acknowledge support from the National Science Foundation through grants AST-9320403 and AST-9803114. The work reported here forms part of the PhD thesis of Andrea Cox, who was an NRAO Predoctoral Fellow while much of it was carried out. ALC would also like to acknowledge Beth Blount (Beloit College ’99), who contributed to this paper. The idea for this survey was a suggestion by Peter Biermann at the Max-Planck-Institut f[ü]{}r Radioastronomie in Bonn, Germany; we are grateful to Barry Clark for arranging “filler time” at the VLA. 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. Optical images in Figures 1-6 were taken from the Digitized Sky Survey (DSS), produced at the Space Telescope Science Institute under U.S. Government grant NAG W-216
(possible PRG ’s; shown as filled circles in Figure   7c). None of these entail asterisk - formation rates are high, compare to normal disk galaxies. For those galaxy with known   masses (attend Table   2), we use these star - geological formation rates to compute flatulence - pulmonary tuberculosis timescales, assuming a Salpeter initial mass affair truncated at $ M$$<$0.4[M${_\odot}$]{}(Figure   7d). Only one of the galaxy in our sample, PRC   C-51, could use up all of its gas in $ \sim$10$^8 $   year, and thus may fall into the category of “ starburst ” galaxies. Although the majority of the radio receiver emission in these campaigner PRGs likely arises from recent star - geological formation activity, the rates implied are not high compared to the entire available gas in the galaxies. We conclude that polar - ring galaxies have radio continuum fluxes typical of normal (non - diametric - ring) elliptical or S0 galaxy. The morphology of the extended sources, the spectral indices, and FIR flux density of the PRG ’s in our sample indicate that the radio emission is predominantly ascribable to star - forming regions rather than nuclear activity. However, the rate of starbirth is modest — all but one of our sample galaxies have enough cool gas to fuel their asterisk formation for at least another gigayear. ALC and LSS admit support from the National Science Foundation through grants AST-9320403 and AST-9803114. The work reported here forms contribution of the PhD thesis of Andrea Cox, who was an NRAO Predoctoral Fellow while much of it was carried out. ALC would also like to acknowledge Beth Blount (Beloit College ’ 99), who contributed to this paper. The estimate for this survey was a suggestion by Peter Biermann at the Max - Planck - Institut f[ü]{}r Radioastronomie in Bonn, Germany; we are grateful to Barry Clark for arrange “ filler clock time ” at the VLA. This research has reach use of the NASA / IPAC Extragalactic Database (NED) which is operate by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration. Optical images in Figures 1 - 6 were take from the Digitized Sky Survey (DSS), produced at the Space Telescope Science Institute under U.S. Government grant NAG   W-216
(podsible PRG’s; shown as filued circles in Yugure 7c). None kf these implied star-formation rates aee hith, compared to normal aisk galaqies. For rhost galaxies with kikwn  masses (ses Tabnx 2), we use these xtar-formathon rates to cmmouce gas-consumption timescales, assumind a Salleher initial mafs flnstioh truncated at $M$$<$0.4[M${_\odot}$]{}(Figure 7d). Only kne of uhe galaxies in out sample, PRC C-51, could use up all of its gas in $\sim$10$^8$ jears, and tyus iqy fall into the category of “starborst” galaxies. Although the majorkty oy the radio enisdhon in thesx candpdate PRGs likely arives frok recent star-fprmetiob activity, the rates mmplied are not high compared tk the total availqboe gav in the talxxits. Wx ckncludf tiat polar-rihg galaxies have radio continuim dluxes typicam of njrial (non-polar-ring) elliptical or S0 galaxits. Ths morphology of the extwnded sources, the speftral indyces, and FIR fluxes of the PRG’s in our sample indhcate ghau bme rxeil emission is predominantly due to star-formind rtgipns rather thak nuclear activity. Hlwrder, the rate uf staxgidth is modest — all but ong of oyr sample galsxies have enough cool gas ro fuel theig stqr formation for ac least anotker gibayeat. ALC and LSS acknowledgz suppkrt from thf Nationam Science Foundatkon tvrough grants AST-9320403 and AST-9803114. Ehe work ceporced here forks pare of the PjD thcvis of Andrea Cox, aho wcs an NRAO Predlctoral Fellow while much of it was carried oot. DLC would ajso llke to acknowleqge Beth Blounj (Beloit Eollegd ’99), who connributed vo this papew. The idea fos this survey was a stggewtiob by Pegdr Biermann at the Max-Planck-Instirut f[ü]{}r Radioastronpmig jn Bonn, Germany; we are grateful tp Bxrrr Black fow arranging “fhllef tkke” at the VLA. Thls fesesrch has made use of the NASA/IPAC Extragalscbic Databqse (NED) rhich is opersted by the Jet Prlpulsmon Lauoratoty, Salifornia Institute of Technomogy, undeg cjntract with the National Aexonautics and Space Administration. Opticel images in Figures 1-6 wwre taken from the Binitized Sky Xurver (DSS), proguced at the Space Twlescope Science Lnstitute under U.S. Govsrnmend grajt NAG W-216
(possible PRG’s; shown as filled circles in None these implied rates are high, For galaxies with known (see Table 2), use these star-formation rates to compute timescales, assuming a Salpeter initial mass function truncated at $M$$<$0.4[M${_\odot}$]{}(Figure 7d). Only one the galaxies in our sample, PRC C-51, could use up all of its in years, thus fall into the category of “starburst” galaxies. Although the majority of the radio emission in these PRGs likely arises from recent star-formation activity, the implied are not high to the total available gas the We conclude polar-ring have continuum fluxes typical normal (non-polar-ring) elliptical or S0 galaxies. The morphology of the extended sources, the spectral indices, and FIR of the our sample that radio is predominantly due regions rather than nuclear activity. However, starbirth is modest — all but one of sample galaxies enough cool gas to fuel their formation for at least another gigayear. ALC and acknowledge support from the National Science Foundation through grants AST-9320403 and AST-9803114. The work reported part of the PhD of Andrea Cox, was NRAO Fellow much of was carried out. ALC would also like to acknowledge Beth Blount College ’99), who contributed to this paper. The idea for was suggestion by Peter at the Max-Planck-Institut f[ü]{}r in Germany; we are grateful Clark arranging the This has made use of NASA/IPAC Extragalactic Database (NED) which operated by the Jet Technology, under contract with the National Aeronautics and Administration. Optical images in Figures 1-6 were from the Digitized Sky Survey (DSS), produced at the Space Telescope Science under U.S. NAG W-216
(possible PRG’s; shown as filled Circles in FIgure 7C). NoNe oF tHese ImplIed star-formatiON ratEs are high, compared to norMal diSk GAlaxIEs. for thOse galaXIeS WIth KnOwN  maSsES (sEe TabLe 2), wE use theSe star-formAtiOn Rates to compuTE gAs-consumptIon Timescales, asSumIng a SaLpEteR InitiAl mAss fuNction TRuncatEd at $M$$<$0.4[M${_\odoT}$]{}(FIGure 7d). ONLy one of THE gAlaxIes in our sample, PRC c-51, CoULd use up all of itS gas in $\SiM$10$^8$ YeARS, anD thUs may fall iNtO the cATegory oF “StARBUrsT” Galaxies. AlthoUgh the majorITy oF the raDiO emISsion iN thesE cANdiDate PRGs likEly aRises from Recent STar-formATion actIvity, tHe rAteS impLIeD aRe nOt HIgh COmParED to The total AvAiLable Gas iN THE GalaXieS. We cOncluDe that polar-riNg gAlaxIEs hAve raDio coNtinUuM fluxEs typiCal of NoRmal (non-polar-rinG) ellIptical or s0 gaLaXieS. THe morPHology Of tHe eXtended Sources, THe sPeCTRAl Indices, and FIR fluxeS oF THe pRG’s in ouR samplE InDiCAte that tHe RadIo emISSion iS preDOmInantly dUe to stAR-fOrMing regIoNs rathEr ThaN nuClear ACtivIty. HowEver, the rAte of STarbirth is modeST — all but one of oUR sAMPlE GalaXieS have enough Cool GAs to Fuel THeIr sTAr forMatioN fOR aT Least another gigayeaR. AlC and LsS ackNowledge suppoRt from the NATIOnal ScieNce FOUnDAtion through grAnts AsT-9320403 and AST-9803114. ThE Work repoRted hEre forms Part of the pHd thesis oF AnDreA CoX, whO WAs An NRAO PredoctORAl FeLlOw while MucH of it waS caRriEd oUt. AlC Would also Like to acKnOwLeDgE BeTh BloUNt (Beloit coLleGe ’99), Who ContrIButed tO this PapeR. THe IDea For this SUrVEY was A sUgGestIon By peter bierMAnn At the MaX-Planck-InStiTUt f[ü]{}R RAdIoastroNomie in Bonn, GeRmAny; we are grAtEfuL to BarRY clark for Arranging “filler time” at thE vLA. 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. OptiCaL ImAges in FiguREs 1-6 wErE taken fRom the DIgitiZEd Sky SuRvey (DSS), prOduced at tHe spacE tEleScope ScienCe InstitUte under U.s. goverNMeNt graNt NaG W-216
(possible PRG’s; shown as filled ci rcles in Fi gu re 7 c).None of thesei mpli ed star-formation rate s are h i gh,c om pared to nor m al d isk g al axi es . F or th ose galaxi es with kn own masses (seeT ab le 2), weuse these star- for mation r ate s to c omp ute g as-con s umptio n timesca le s , assu m ing a S a l pe terinitial mass func t io n truncated at$M$$<$ 0. 4 [M $ { _\o dot }$]{}(Figu re  7d). Only on e o f t heg alaxies in ou r sample, P R C C -51, c ou ldu se upall o fi tsgas in $\si m$10 $^8$ year s, and thus ma y fall i nto th e c ate gory of “ sta rb u rst ” g ala x ies . Althou gh t he ma jori t y o f th e r adio emis sion in these ca ndid a tePRGslikel y ar is es fr om rec ent s ta r-formation act ivit y, the ra tes i mpl ie d are not hi ghcom pared t o the t o tal a v a i la ble gas in the gal ax i e s. We con cludet ha tp olar-rin ggal axie s haveradi o c ontinuum fluxe s t yp ical of n ormal(n on- pol ar-ri n g) e llipti cal or S 0 gal a xies. The morp h ology of thee xt e n de d sou rce s, the spec tral indi ces, an d F I R flu xes o ft he PRG’s in our sample i ndicat e tha t the radio e mission is p r edominan tlyd ue to star-formin g reg ions rathe r than nu clear activit y. Howeve r , the rat e o f s tar bir t h i s modest — al l buton e of ou r s ample g ala xie s h ave e nough coo l gas to f ue lth eir star formatio nfor a t l easta nother giga year . A L C a nd LSSa ck n o wled ge s uppo rtfr om th e Na t ion al Scie nce Found ati o n th ro ug h grant s AST-9320403 a nd AST-980 31 14. The w o r k report ed here forms part of t h e PhD t hes is of And rea Cox,who was a n N R AO Pre doctor al Fe ll oww h ile m u c hofit was carri e d ou t. AL Cwoul d alsolike to acknowledg e Be th Blount (Be loi t Co l l eg e ’ 9 9) , wh oc ont r i buted to this p aper. Theid e afor this s u rve ywas a s uggesti on by Peter B iermann a t the Max -P lanc k - Ins titut f[ü] {}r Radi oastronom i e inB on n, Ge rma ny; we a regrate ful to Bar ry Cl ark fo rarrang ing “ fi ller tim e” at the VLA. This res earchhas m ade use of t heN ASA /IPAC Ext raga lactic Dat aba se(NED) wh i ch is ope r at edb y the Jet Propulsio n L abo r a to ry, Califor n i a In stitu teo f Tech nolo gy, under contrac t with the Nati onal A ero nau t icsan d Space Admini str at i o n. Optic al images inFigures1- 6 were taken fromthe Dig i t iz e d SkySurv ey(DSS), pr odu ce d at the S pa c e Tele scop eScienc e Inst i tute u nder U.S. Govern mentg r ant N A G W -216
(possible_PRG’s; shown_as filled circles in_Figure 7c). None_of_these implied_star-formation_rates are high,_compared to normal_disk galaxies. For those_galaxies with known_ masses_(see Table 2), we use these star-formation rates to compute gas-consumption timescales, assuming a Salpeter_initial_mass function_truncated_at_$M$$<$0.4[M${_\odot}$]{}(Figure 7d). Only one of the_galaxies in our sample, PRC C-51,_could use_up all of its gas in $\sim$10$^8$ years, and_thus_may fall into_the category of “starburst” galaxies. Although the majority of_the radio emission in these candidate_PRGs likely arises_from_recent_star-formation activity, the rates_implied are not high compared to_the total available gas in the_galaxies. We conclude that polar-ring galaxies have radio_continuum fluxes typical of normal (non-polar-ring)_elliptical or S0 galaxies. The_morphology of_the extended sources, the spectral_indices, and FIR_fluxes of_the PRG’s in_our sample indicate that the radio_emission is predominantly_due to star-forming regions rather than_nuclear_activity. However, the_rate_of_starbirth is_modest — all_but_one of_our_sample galaxies have enough cool gas_to_fuel their star formation for at least_another gigayear. ALC and LSS_acknowledge_support from the National_Science Foundation through grants AST-9320403_and AST-9803114. The work reported here_forms part_of the_PhD thesis of Andrea Cox, who was an NRAO Predoctoral Fellow_while much of it was carried_out. ALC would also_like to_acknowledge_Beth Blount (Beloit_College_’99), who_contributed to this paper. The idea for_this survey_was a suggestion by Peter Biermann_at the Max-Planck-Institut f[ü]{}r_Radioastronomie_in Bonn, Germany; we are grateful_to Barry Clark for arranging “filler_time” at the VLA. 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._Optical images in Figures 1-6 were_taken_from_the Digitized Sky Survey (DSS),_produced at the Space Telescope Science_Institute under U.S. Government grant NAG W-216
} ================== ![FER comparison of proposed model and the model of [@Chatzigeorgiou2008] for case 0: general MIMO channels with $N_T=1$, $N=N_R=1,2,4$, uncoded.[]{data-label="fig.case.0"}](FER_MIMO_Comparison_Result_Without_Outaget_Uncoded){width="80.00000%"} ![Average FER of the HRS scheme for case 1: $\Omega_0=\Omega_{1i}=\Omega_{2i}=1$, uncoded.[]{data-label="fig.case.1"}](UnCoded_L100_case1){width="80.00000%"} ![Theoretical and simulated FER of the HRS scheme for case 4 and 5: $\Omega_0=\Omega_{1i}=\Omega_{2i}=1$, coded.[]{data-label="fig.case.4.5"}](Coded_1n2_case_4_5){width="80.00000%"} ![Theoretical and simulated FER of the HRS scheme for case 6 and 7: $\Omega_0=\Omega_{1i}=\Omega_{2i}=1$, coded.[]{data-label="fig.case.6.7"}](Coded_2n3_case_6_7){width="80.00000%"} ![Average FER of the HRS, AF-RS and PDF-RS schemes for case 4 with $n=4$: $\Omega_0=\Omega_{1i}=\Omega_{2i}=1$, coded.[]{data-label="fig.case.4.AF.PDF.N4"}](Coded_1n2_case_4_AF_PDF_HRS_N4){width="80.00000%"} In this section, we compare the proposed approximate FER expressions with the exact FER obtained by Monte-Carlo simulations. Unless specifically mentioned, the simulations are performed for a BPSK modulation and a frame size of $100$ or $200$ symbols over block Rayleigh fading channels. We consider two basic cases: (1) the *uncoded* case, where none channel code is used; and (2) the *coded* case, where a systematic convolutional code with a code rate of $1/2$ and the generator matrix of $(5,7)_8$ or a code rate of $2/
} = = = = = = = = = = = = = = = = = = ! [ FER comparison of proposed model and the model of [ @Chatzigeorgiou2008 ] for case 0: cosmopolitan MIMO groove with $ N_T=1 $, $ N = N_R=1,2,4 $, uncoded.[]{data - label="fig.case.0"}](FER_MIMO_Comparison_Result_Without_Outaget_Uncoded){width="80.00000% " } ! [ Average FER of the HRS scheme for subject 1: $ \Omega_0=\Omega_{1i}=\Omega_{2i}=1 $, uncoded.[]{data - label="fig.case.1"}](UnCoded_L100_case1){width="80.00000% " } ! [ Theoretical and fake FER of the HRS scheme for case 4 and 5: $ \Omega_0=\Omega_{1i}=\Omega_{2i}=1 $, coded.[]{data - label="fig.case.4.5"}](Coded_1n2_case_4_5){width="80.00000% " } ! [ Theoretical and simulated FER of the HRS dodge for case 6 and 7: $ \Omega_0=\Omega_{1i}=\Omega_{2i}=1 $, coded.[]{data - label="fig.case.6.7"}](Coded_2n3_case_6_7){width="80.00000% " } ! [ Average FER of the HRS, AF - RS and PDF - RS scheme for case 4 with $ n=4 $: $ \Omega_0=\Omega_{1i}=\Omega_{2i}=1 $, coded.[]{data - label="fig.case.4.AF.PDF.N4"}](Coded_1n2_case_4_AF_PDF_HRS_N4){width="80.00000% " } In this section, we compare the nominate approximate FER expressions with the exact FER receive by Monte - Carlo model. Unless specifically mentioned, the simulations are performed for a BPSK modulation and a frame size of $ 100 $ or $ 200 $ symbol over block Rayleigh fading channels. We regard two basic cases: (1) the * uncoded * case, where none channel code is used; and (2) the * code * case, where a systematic convolutional code with a code rate of $ 1/2 $ and the generator matrix of $ (5,7)_8 $ or a code rate of $ 2/
} ================== ![FER comparison of proposed oodel and the model of [@Chatzjgeorgioj2008] for case 0: general MIMO chainelw wity $N_T=1$, $N=N_R=1,2,4$, uncoded.[]{data-lacel="fig.casv.0"}](FER_MIMO_Cimpacison_Result_Withonf_Outageb_Bncodsf){widch="80.00000%"} ![Everage FER of jhe HRS scheke for case 1: $\Okeea_0=\Lmega_{1i}=\Omega_{2i}=1$, uncoded.[]{data-label="fig.casq.1"}](UnCodec_L100_fase1){width="80.00000%"} ![Theorgticak and simulated FER of the HRS scheme fkr case 4 and 5: $\Omega_0=\Okega_{1i}=\Omega_{2i}=1$, coded.[]{data-label="vig.cwse.4.5"}](Coded_1n2_case_4_5){width="80.00000%"} ![Hheoretical and wimulated FEF of the HGV scheme fkr case 6 and 7: $\Omega_0=\Omega_{1i}=\Omega_{2i}=1$, codeb.[]{data-label="fug.xasf.6.7"}](Woded_2n3_case_6_7){wmdth="80.00000%"} ![Avvrage FER of bne HRS, AF-RS amd PDF-RS schemcs foc cawe 4 with $n=4$: $\Omega_0=\Omega_{1m}=\Omega_{2i}=1$, coded.[]{data-labgl="fig.case.4.AX.PBF.N4"}](Coded_1n2_case_4_AF_PDF_HRS_B4){wudth="80.00000%"} It thhs sdxtiun, se ckmpare thx proposed zpproximate FER expressions wiuh eye exact FER kbtainqd by Monte-Carlo simulations. Unless specixiczlly mentioned, the simuoations are performed for a BPFK modulation and a frame size of $100$ or $200$ symbols ovar blkzk Xqyleiey vading channels. We consider two basic cases: (1) ege *ukcoded* case, wherc none channel codr ls osed; and (2) the *zoded* ezss, where a systematlc convjlutiinal code witn a code rate of $1/2$ and the gwnerator matgix if $(5,7)_8$ or a code rate of $2/
} ================== ![FER comparison of proposed model model [@Chatzigeorgiou2008] for 0: general MIMO ![Average of the HRS for case 1: uncoded.[]{data-label="fig.case.1"}](UnCoded_L100_case1){width="80.00000%"} ![Theoretical and simulated FER of HRS scheme for case 4 and 5: $\Omega_0=\Omega_{1i}=\Omega_{2i}=1$, coded.[]{data-label="fig.case.4.5"}](Coded_1n2_case_4_5){width="80.00000%"} ![Theoretical and simulated FER the HRS scheme for case 6 and 7: $\Omega_0=\Omega_{1i}=\Omega_{2i}=1$, coded.[]{data-label="fig.case.6.7"}](Coded_2n3_case_6_7){width="80.00000%"} ![Average FER of HRS, and schemes case 4 with $n=4$: $\Omega_0=\Omega_{1i}=\Omega_{2i}=1$, coded.[]{data-label="fig.case.4.AF.PDF.N4"}](Coded_1n2_case_4_AF_PDF_HRS_N4){width="80.00000%"} In this section, we compare the proposed approximate FER expressions the exact FER obtained by Monte-Carlo simulations. Unless mentioned, the simulations are for a BPSK modulation and frame of $100$ $200$ over Rayleigh fading channels. consider two basic cases: (1) the *uncoded* case, where none channel code is used; and (2) the case, where convolutional code a rate $1/2$ and the of $(5,7)_8$ or a code rate
} ================== ![FER comparison of proposed moDel and the mOdel oF [@ChAtzIgEorgIou2008] fOr case 0: general MimO chAnnels with $N_T=1$, $N=N_R=1,2,4$, uncoded.[]{Data-lAbEL="fig.CAsE.0"}](FER_MiMO_CompARiSON_ReSuLt_witHoUT_OUtageT_UnCoded){wiDth="80.00000%"} ![Average fER Of The HRS scheme FOr Case 1: $\Omega_0=\OMegA_{1i}=\Omega_{2i}=1$, uncoDed.[]{Data-laBeL="fiG.Case.1"}](UNCoDed_L100_cAse1){widTH="80.00000%"} ![TheorEtical and SiMUlated feR of the hrs sChemE for case 4 and 5: $\Omega_0=\OMEgA_{1I}=\Omega_{2i}=1$, coded.[]{daTa-labeL="fIG.cASE.4.5"}](CoDed_1N2_case_4_5){width="80.00000%"} ![thEoretICal and sIMuLATEd Fer of the HRS scheMe for case 6 anD 7: $\omeGa_0=\OmegA_{1i}=\omeGA_{2i}=1$, codeD.[]{data-LaBEl="fIg.case.6.7"}](Coded_2N3_casE_6_7){width="80.00000%"} ![AveRage FEr Of the HRs, aF-RS and pDF-RS sCheMes For cASe 4 WiTh $n=4$: $\omEGa_0=\OMEgA_{1i}=\OMEga_{2I}=1$, coded.[]{daTa-LaBel="fiG.casE.4.af.pdF.N4"}](COdeD_1n2_caSe_4_AF_PdF_HRS_N4){width="80.00000%"} In ThiS secTIon, We comPare tHe prOpOsed aPproxiMate FeR Expressions with The eXact FER obTaiNeD by moNte-CaRLo simuLatIonS. Unless SpecifiCAllY mENTIoNed, the simulations aRe PERfOrmed for A BPSK mODuLaTIon and a fRaMe sIze oF $100$ OR $200$ symbOls oVEr Block RayLeigh fADiNg ChannelS. WE consiDeR twO baSic caSEs: (1) thE *uncodEd* case, whEre noNE channel code is USed; and (2) the *codeD* CaSE, WhERe a sYstEmatic convoLutiONal cOde wITh A coDE rate Of $1/2$ and ThE GeNErator matrix of $(5,7)_8$ or a coDe Rate of $2/
} ================== ![FE R comparis on of pr opo se d mo deland the modelo f [@ Chatzigeorgiou2008] fo r cas e0 : ge n er al MI MO chan n el s wit h$N _T= 1$ , $ N=N_R =1, 2,4$, u ncoded.[]{ dat a- label="fig.c a se .0"}](FER_ MIM O_Comparison _Re sult_W it hou t _Outa get _Unco ded){w i dth="8 0.00000%" }![Aver a ge FERo f t he H RS scheme for cas e 1 : $\Omega_0=\Om ega_{1 i} = \O m e ga_ {2i }=1$, unco de d.[]{ d ata-lab e l= " f i g.c a se.1"}](UnCod ed_L100_cas e 1){ width= "8 0.0 0 000%"} ![T he o ret ical and si mula ted FER o f theH RS sche m e for c ase 4and 5: $\O m eg a_ 0=\ Om e ga_ { 1i }=\ O meg a_{2i}=1 $, c oded. []{d a t a - labe l=" fig. case. 4.5"}](Coded_ 1n2 _cas e _4_ 5){wi dth=" 80.0 00 00%"} ![Th eoret ic al and simulate d FE R of theHRS s che me forc ase 6and 7: $\Omeg a_0=\Om e ga_ {1 i } = \O mega_{2i}=1$, code d. [ ] {d ata-labe l="fig . ca se . 6.7"}](C od ed_ 2n3_ c a se_6_ 7){w i dt h="80.00 000%"} ! [A verageFE R of t he HR S,AF-RS andPDF-RS schemes forc ase 4 with $n= 4 $: $\Omega_0= \ Om e g a_ { 1i}= \Om ega_{2i}=1$ , co d ed.[ ]{da t a- lab e l="fi g.cas e. 4 .A F .PDF.N4"}](Coded_1n 2_ case_4 _AF_P DF_HRS_N4){wi dth="80.00 0 0 0 %"} Inthis se c tion, we compa re th e proposed approxim ate F ER expre ssions wi t h the exa ctFER ob tai n e dby Monte-Carl o simu la tions.Unl ess spe cif ica lly me nt ioned, th e simula ti on sar e p erfor m ed for a B PSK m odu latio n and a fram e si ze o f $1 00$ or$ 20 0 $ sym bo ls ove r b lo ck Ra ylei g h f ading c hannels.Wec onsi de rtwo bas ic cases: (1) t he *uncode d* ca se, wh e r e none c hannel code is used; an d (2) th e * coded * ca se, where asystem ati c convo lution al co de wi t h a co d e r ate o f $1/2$ an d the gene ra tormatrixof $(5,7)_8$ or ac ode rate of $2/
} ================== ![FER comparison_of proposed_model and the model_of [@Chatzigeorgiou2008]_for_case 0:_general_MIMO channels with_$N_T=1$, $N=N_R=1,2,4$, uncoded.[]{data-label="fig.case.0"}](FER_MIMO_Comparison_Result_Without_Outaget_Uncoded){width="80.00000%"} ![Average_FER of the HRS_scheme for case_1:_$\Omega_0=\Omega_{1i}=\Omega_{2i}=1$, uncoded.[]{data-label="fig.case.1"}](UnCoded_L100_case1){width="80.00000%"} ![Theoretical and simulated FER of the HRS scheme for case 4 and 5:_$\Omega_0=\Omega_{1i}=\Omega_{2i}=1$,_coded.[]{data-label="fig.case.4.5"}](Coded_1n2_case_4_5){width="80.00000%"} ![Theoretical and_simulated_FER_of the HRS scheme for_case 6 and 7: $\Omega_0=\Omega_{1i}=\Omega_{2i}=1$,_coded.[]{data-label="fig.case.6.7"}](Coded_2n3_case_6_7){width="80.00000%"} ![Average FER_of the HRS, AF-RS and PDF-RS schemes for_case_4 with $n=4$:_$\Omega_0=\Omega_{1i}=\Omega_{2i}=1$, coded.[]{data-label="fig.case.4.AF.PDF.N4"}](Coded_1n2_case_4_AF_PDF_HRS_N4){width="80.00000%"} In this section, we compare the proposed approximate_FER expressions with the exact FER_obtained by Monte-Carlo_simulations._Unless_specifically mentioned, the simulations_are performed for a BPSK modulation_and a frame size of $100$_or $200$ symbols over block Rayleigh fading_channels. We consider two basic cases:_(1) the *uncoded* case, where_none channel_code is used; and (2)_the *coded* case,_where a_systematic convolutional code_with a code rate of $1/2$_and the generator_matrix of $(5,7)_8$ or a code_rate_of $2/
and $\ell$ and any subset $S \subset\{1,\dots,N\}$: $$\label{eq:e-commute-uv} e_k(\mathbf{u}_S)\,e_\ell(\mathbf{v}_S)=e_\ell(\mathbf{v}_S)\,e_k(\mathbf{u}_S);$$ - the commutation relation holds for $|S|\le 3$ and any $k, \ell$; - the commutation relation holds for $|S|\le 3$ and $k\ell\le 3$. Explicitly, Theorem \[t es commute uv 2\] asserts that the commutation relations  hold for all $k$ and $\ell$ and all subsets $S \subset \{1,\dots,N\}$ if and only if the following relations hold: $$\begin{aligned} &e_1(\mathbf{u}_S)e_1(\mathbf{v}_S) = e_1(\mathbf{v}_S)e_1(\mathbf{u}_S) \qquad \text{for }\, 1 \le |S| \le 3, \label{e uv 12vars AB}\\ &e_2(\mathbf{u}_S)e_1(\mathbf{v}_S) = e_1(\mathbf{v}_S)e_2(\mathbf{u}_S) \qquad \text{for }\, 2 \le |S| \le 3, \label{e uuv 23vars AB}\\ &e_1(\mathbf{u}_S)e_2(\mathbf{v}_S) = e_2(\mathbf{v}_S)e_1(\mathbf{u}_S) \qquad \text{for }\, 2 \le |S| \le 3, \label{e uvv 23vars AB}\\ &e_3(\mathbf{u}_S)e_1(\mathbf{v}_S) = e_1(\mathbf{v}_S)e_3(\mathbf{u}_S) \qquad \text{for }\, |S| = 3, \label{e uuuv 23vars AB}\\ &e_1(\mathbf{u}_S)e_3(\mathbf{v}_S) = e_3(\mathbf{
and   $ \ell$ and any subset $ S \subset\{1,\dots, N\}$: $ $ \label{eq: e - commute - uv } e_k(\mathbf{u}_S)\,e_\ell(\mathbf{v}_S)=e_\ell(\mathbf{v}_S)\,e_k(\mathbf{u}_S);$$ - the substitution relation back holds for $ |S|\le 3 $ and any $ k, \ell$; - the commutation relation back hold for $ |S|\le 3 $ and $ k\ell\le 3$. Explicitly, Theorem   \[t es commute uv 2\ ] asserts that the substitution relations   hold for all $ k$ and   $ \ell$ and all subsets $ S \subset \{1,\dots, N\}$ if and merely if the following relations detention: $ $ \begin{aligned } & e_1(\mathbf{u}_S)e_1(\mathbf{v}_S) = e_1(\mathbf{v}_S)e_1(\mathbf{u}_S) \qquad \text{for } \, 1 \le |S| \le 3, \label{e uv 12vars AB}\\ & e_2(\mathbf{u}_S)e_1(\mathbf{v}_S) = e_1(\mathbf{v}_S)e_2(\mathbf{u}_S) \qquad \text{for } \, 2 \le |S| \le 3, \label{e uuv 23vars AB}\\ & e_1(\mathbf{u}_S)e_2(\mathbf{v}_S) = e_2(\mathbf{v}_S)e_1(\mathbf{u}_S) \qquad \text{for } \, 2 \le |S| \le 3, \label{e uvv 23vars AB}\\ & e_3(\mathbf{u}_S)e_1(\mathbf{v}_S) = e_1(\mathbf{v}_S)e_3(\mathbf{u}_S) \qquad \text{for } \, |S| = 3, \label{e uuuv 23vars AB}\\ & e_1(\mathbf{u}_S)e_3(\mathbf{v}_S) = e_3(\mathbf {
anf $\ell$ and any subset $S \sunset\{1,\dots,N\}$: $$\label{eq:e-commuve-uv} e_k(\mathbw{u}_S)\,e_\ell(\mathbf{v}_S)=e_\ell(\mathbf{v}_S)\,e_n(\mqthbf{y}_S);$$ - the commutation rdlation hllds for $|S|\le 3$ and any $k, \ell$; - tmz comjmtatimi relation holdx for $|S|\le 3$ and $k\ell\le 3$. Ex[lkcntly, Theorem \[t es commute uv 2\] asserts that tne commutation rglatipgs  hkld for all $k$ and $\ell$ and all subsefs $S \suuset \{1,\dots,N\}$ if amd only if the following rflatlons hold: $$\begin{alihned} &e_1(\mathbf{o}_A)e_1(\mwrhbf{v}_S) = e_1(\maghbf{v}_S)e_1(\mathbf{u}_S) \qquad \jext{for }\, 1 \le |S| \le 3, \label{e uv 12vafs AB}\\ &z_2(\mathbf{u}_S)e_1(\mqtybf{g}_V) = e_1(\mathbf{t}_S)e_2(\matrbf{u}_S) \qquad \bvxt{for }\, 2 \le |S| \le 3, \label{e uuy 23varv AV}\\ &e_1(\mathbf{u}_S)e_2(\mathbf{v}_S) = e_2(\mathbf{v}_S)e_1(\mathbf{u}_F) \qquad \taxc{for }\, 2 \le |S| \le 3, \labeo{e uvv 23ears AB}\\ &e_3(\oqthcf{u}_A)e_1(\kafhbf{v}_S) = x_1(\mathbf{v}_S)e_3(\mzthbf{u}_S) \qquqd \text{for }\, |S| = 3, \lanej{v uuuv 23vars AG}\\ &e_1(\mathff{t}_S)e_3(\mathbf{v}_S) = e_3(\mathbf{
and $\ell$ and any subset $S \subset\{1,\dots,N\}$: - commutation relation for $|S|\le 3$ the relation holds for 3$ and $k\ell\le Explicitly, Theorem \[t es commute uv asserts that the commutation relations hold for all $k$ and $\ell$ and all $S \subset \{1,\dots,N\}$ if and only if the following relations hold: $$\begin{aligned} &e_1(\mathbf{u}_S)e_1(\mathbf{v}_S) e_1(\mathbf{v}_S)e_1(\mathbf{u}_S) \text{for 1 |S| \le 3, \label{e uv 12vars AB}\\ &e_2(\mathbf{u}_S)e_1(\mathbf{v}_S) = e_1(\mathbf{v}_S)e_2(\mathbf{u}_S) \qquad \text{for }\, 2 \le |S| 3, \label{e uuv 23vars AB}\\ &e_1(\mathbf{u}_S)e_2(\mathbf{v}_S) = e_2(\mathbf{v}_S)e_1(\mathbf{u}_S) \text{for }\, 2 \le \le 3, \label{e uvv 23vars &e_3(\mathbf{u}_S)e_1(\mathbf{v}_S) e_1(\mathbf{v}_S)e_3(\mathbf{u}_S) \qquad }\, = \label{e uuuv 23vars &e_1(\mathbf{u}_S)e_3(\mathbf{v}_S) = e_3(\mathbf{
and $\ell$ and any subset $S \subset\{1,\Dots,N\}$: $$\label{Eq:e-coMmuTe-uV} e_K(\matHbf{u}_s)\,e_\ell(\mathbf{v}_S)=e_\ELl(\maThbf{v}_S)\,e_k(\mathbf{u}_S);$$ - the comMutatIoN RelaTIoN holdS for $|S|\le 3$ ANd ANY $k, \eLl$; - ThE coMmUTaTion rElaTion holDs for $|S|\le 3$ anD $k\eLl\Le 3$. Explicitly, tHeOrem \[t es comMutE uv 2\] asserts thAt tHe commUtAtiON relaTioNs  holD for alL $K$ and $\elL$ and all suBsETs $S \subSEt \{1,\dots,N\}$ IF AnD onlY if the following reLAtIOns hold: $$\begin{alIgned} &e_1(\MaTHbF{U}_s)e_1(\mAthBf{v}_S) = e_1(\mathbF{v}_s)e_1(\matHBf{u}_S) \qquAD \tEXT{For }\, 1 \LE |S| \le 3, \label{e uv 12vArs AB}\\ &e_2(\mathbF{U}_S)e_1(\Mathbf{V}_S) = E_1(\maTHbf{v}_S)e_2(\MathbF{u}_s) \QquAd \text{for }\, 2 \le |s| \le 3, \lAbel{e uuv 23vArs AB}\\ &e_1(\MAthbf{u}_S)E_2(\Mathbf{v}_s) = e_2(\mathBf{v}_s)e_1(\mAthbF{U}_S) \QqUad \TeXT{foR }\, 2 \Le |s| \le 3, \LAbeL{e uvv 23varS Ab}\\ &e_3(\MathbF{u}_S)e_1(\MATHBf{v}_S) = E_1(\maThbf{V}_S)e_3(\maThbf{u}_S) \qquad \teXt{fOr }\, |S| = 3, \lABel{E uuuv 23Vars Ab}\\ &e_1(\maThBf{u}_S)e_3(\Mathbf{V}_S) = e_3(\maThBf{
and $\ell$ and any subset $S \subse t\{1, \do ts, N\ }$:$$\l abel{eq:e-comm u te-u v} e_k(\mathbf{u}_ S)\,e _\ e ll(\ m at hbf{v }_S)=e_ \ el l ( \ma th bf {v} _S ) \, e_k(\ mat hbf{u}_ S);$$ - th ecommutationr el ation hold s f or $|S|\le 3 $ a nd any $ k,\ ell$; - th e comm u tation relation h o lds fo r $|S|\l e 3$ and $k\ell\le 3$. E x pl i citly, Theorem  \[t e sc om m u teuv2\] assert sthatt he comm u ta t i o n r e lations  hold for all $k $ an d $\el l$ an d all s ubset s$ S \ subset \{1, \dot s,N\}$ if and o n ly if t h e follo wing r ela tio ns h o ld :$$\ be g in{ a li gne d } & e_1(\mat hb f{ u}_S) e_1( \ m a t hbf{ v}_ S) = e_ 1(\mathbf{v}_ S)e _1(\ m ath bf{u} _S) \ qqua d\text {for } \, 1\l e |S| \le 3,\lab el{e uv 1 2va rs AB }\ \ &e_ 2 (\math bf{ u}_ S)e_1(\ mathbf{ v }_S )= e _1(\mathbf{v}_S)e_ 2( \ m at hbf{u}_S ) \qqu a d\t e xt{for } \, 2\le| S | \ le 3 , \ label{euuv 23 v ar sAB}\\ & e_ 1(\mat hb f{u }_S )e_2( \ math bf{v}_ S) = e _2(\m a thbf{v}_S)e_1( \ mathbf{u}_S)\ qq u a d\ text {fo r }\, 2 \le |S| \l e 3, \l abe l {e uv v 23v ar s A B }\\ &e_3(\mathbf{u} _S )e_1(\ mathb f{v}_S) = e _1(\mathbf { v } _S)e_3(\ math b f{ u }_S) \qquad \t ext{f or }\, |S| = 3, \ label {e uuuv23vars AB } \ \ &e_1(\ mat hbf {u} _S) e _ 3( \mathbf{v}_S) = e _3 (\mathb f{
and $\ell$_and any_subset $S \subset\{1,\dots,N\}$: $$\label{eq:e-commute-uv} _ __e_k(\mathbf{u}_S)\,e_\ell(\mathbf{v}_S)=e_\ell(\mathbf{v}_S)\,e_k(\mathbf{u}_S);$$ - __the commutation relation_holds for $|S|\le_3$ and any $k,_\ell$; - _the_commutation relation holds for $|S|\le 3$ and $k\ell\le 3$. Explicitly, Theorem \[t es commute uv 2\]_asserts_that the_commutation_relations _hold for all $k$ and $\ell$_and all subsets $S \subset_\{1,\dots,N\}$ if_and only if the following relations hold: $$\begin{aligned} &e_1(\mathbf{u}_S)e_1(\mathbf{v}_S)_= _ e_1(\mathbf{v}_S)e_1(\mathbf{u}_S) \qquad_\text{for }\, 1 \le |S| \le 3, \label{e_uv 12vars AB}\\ &e_2(\mathbf{u}_S)e_1(\mathbf{v}_S) = e_1(\mathbf{v}_S)e_2(\mathbf{u}_S)_\qquad \text{for }\,_2_\le_|S| \le 3,_\label{e uuv 23vars AB}\\ &e_1(\mathbf{u}_S)e_2(\mathbf{v}_S) = _e_2(\mathbf{v}_S)e_1(\mathbf{u}_S) \qquad \text{for }\, 2 \le_|S| \le 3, \label{e uvv 23vars_AB}\\ &e_3(\mathbf{u}_S)e_1(\mathbf{v}_S) = e_1(\mathbf{v}_S)e_3(\mathbf{u}_S) \qquad \text{for_}\, |S| = 3,_\label{e uuuv_23vars AB}\\ &e_1(\mathbf{u}_S)e_3(\mathbf{v}_S) = e_3(\mathbf{
cumulant and cubic response. The latter could be due to either nonlinear dissipation mechanisms or to dependence on actual current temperature maintained by Joulean heating. But, in any case, the left-hand side can not grow in a more fast way than $\propto t$. On the contrary, in presence of $f^{-\gamma }$ excess noise the first term on right-hand side should grow as $\propto t^{1+\gamma }$. Hence, at sufficiently long time intervals it must be compensated by the second term, and we come to relations between low-field excess nonequilibrium noise and four-point cumulant of equilibrium noise. The result can be written also as $K_m(\tau )=\frac 1{T^2}K_D(\tau )$ \[4,6,9\], where $K_D(\tau )$ is the above defined correlator of diffusivity and $K_m(\tau )$ is the mobility correlation function as introducedd by the expansion $\left\langle Q(t),Q(t)\right\rangle =\left\langle Q(t),Q(t)\right\rangle _0+x^2\int_0^t\int_0^tK_m(t^{\prime }-t^{\prime \prime })dt^{\prime }dt^{\prime \prime }+O(x^4)$. This is the extension of Einstein relation and similar relations to 1/f-fluctuations. UNIFICATION OF EQUILIBRIUM AND\ NONEQUILIBRIUM BROWNIAN MOTION ------------------------------- The producing FDR can be reformulated in terms of characteristic function (CF): $$\left\langle \exp [(ik-\frac xT)Q(t)]\right\rangle =\left\langle \exp [-ikQ(t)]\right\rangle$$ This is functional equation with respect to dependencies on $ik$ and $x$. It helps to find the connections between the whole statistics of nonequilibrium random walk and that of equilibrium one, in presence of 1/f-noise, if we know something about statistics under neglecting 1/f-noise \[4,6,8,38\]. One of possible solutions on the functional equation is $\left\langle \exp [ikQ(t)]\right\rangle =\Xi (ikx/T-k^2,t)$, with only two independent arguments. It is easy to see that
cumulant and cubic response. The latter could be due to either nonlinear dissipation mechanisms or to dependence on actual current temperature sustain by Joulean heating system. But, in any case, the left - bridge player english cannot grow in a more fast room than $ \propto t$. On the reverse, in presence of $ f^{-\gamma } $ excess noise the first term on right - bridge player side should grow as $ \propto t^{1+\gamma } $. therefore, at sufficiently long time interval it must be compensated by the second term, and we arrive to relations between low - field excess nonequilibrium randomness and four - point cumulant of equilibrium randomness. The result can be write also as $ K_m(\tau) = \frac 1{T^2}K_D(\tau) $ \[4,6,9\ ], where $ K_D(\tau) $ is the above defined correlator of diffusivity and $ K_m(\tau) $ is the mobility correlation function as introducedd by the expansion $ \left\langle Q(t),Q(t)\right\rangle = \left\langle Q(t),Q(t)\right\rangle _ 0+x^2\int_0^t\int_0^tK_m(t^{\prime } -t^{\prime \prime }) dt^{\prime } dt^{\prime \prime } + O(x^4)$. This is the extension of Einstein relation back and similar relations to 1 / f - fluctuations. UNIFICATION OF EQUILIBRIUM AND\ NONEQUILIBRIUM BROWNIAN MOTION ------------------------------- The producing FDR can be reformulated in terms of characteristic function (CF ): $ $ \left\langle \exp [ (ik-\frac xT)Q(t)]\right\rangle = \left\langle \exp [ -ikQ(t)]\right\rangle$$ This is functional equation with respect to dependencies on $ ik$ and $ x$. It helps to find the connections between the whole statistics of nonequilibrium random base on balls and that of equilibrium one, in presence of 1 / f - randomness, if we acknowledge something about statistic under neglecting 1 / f - noise \[4,6,8,38\ ]. One of possible solutions on the functional equation is $ \left\langle \exp [ ikQ(t)]\right\rangle = \Xi (ikx / T - k^2,t)$, with only two independent argumentation. It is easy to see that
cululant and cubic responst. The latter coulb be dux to eifher nonuinear dissipation mechanismd ir to dependence on actual zurrent tvmperaturw mamntained by Joulxzn heatlug. Buf, in cnb case, the left-mand side cdn not grow in a mlre fast way than $\propto t$. On the cjntrary, ij presence of $s^{-\gamkw }$ esbews noise the first term on rifht-hand side should brow as $\propto t^{1+\gamma }$. Henfe, ah sufficiently lonh time intetbalf it must be zompensated by the secknd term, and we come to relatiovs becween low-figls ffcess noneqnilibrpum noise and four-poitt cumukant of equilinrium nouse. The result can be written also as $K_i(\tau )=\frac 1{T^2}I_D(\tau )$ \[4,6,9\], where $K_D(\tqu )$ is jhe atove eefknes rordelatog oh diffusivify and $K_m(\tay )$ is the mobility vowgrlation funcfion af yntroducedd by the expansion $\left\langle Q(t),S(t)\right\rangle =\left\langlw Q(t),Q(t)\right\rangle _0+x^2\int_0^j\int_0^tK_m(t^{\pryme }-t^{\prime \prime })dt^{\prime }dt^{\prime \prime }+O(x^4)$. This is dhe eetdnsnin of Wijstein relation and similar relations to 1/f-flusfustpons. UNIFICATION ON EQUILIBRIUM AND\ NPNFQIYLIBRIUM BROWVIAN MOTIKN ------------------------------- The producing FDG can bg refoemulated yn trrms of characteristic funcrion (CF): $$\left\jqngle \exp [(ik-\frac xC)Q(t)]\right\rangue =\lgft\lanble \exp [-ikQ(t)]\right\rangle$$ Chis ia functionap equatioh with respect to deiendancies on $ik$ and $x$. It helpf to find the connecgionx betwqen the whlle sbdtistics of nonequllibrnum rdndom walk and that of equilibrium one, in presence of 1/f-mohse, if we kuow sokething about statistics unber neglzcting 1/f-noise \[4,6,8,38\]. Onv of possmble solutiogs on the funwjional equatimn is $\lest\labgle \exp [ikQ(g)]\fight\rangle =\Xi (ikx/T-k^2,t)$, wptk only twi independent argukenjs. It is easy to wee that
cumulant and cubic response. The latter could to nonlinear dissipation or to dependence by heating. But, in case, the left-hand can not grow in a more way than $\propto t$. On the contrary, in presence of $f^{-\gamma }$ excess the first term on right-hand side should grow as $\propto t^{1+\gamma }$. Hence, sufficiently time it be compensated by the second term, and we come to relations between low-field excess nonequilibrium noise four-point cumulant of equilibrium noise. The result can written also as $K_m(\tau 1{T^2}K_D(\tau )$ \[4,6,9\], where $K_D(\tau is above defined of and )$ is the correlation function as introducedd by the expansion $\left\langle Q(t),Q(t)\right\rangle =\left\langle Q(t),Q(t)\right\rangle _0+x^2\int_0^t\int_0^tK_m(t^{\prime }-t^{\prime \prime })dt^{\prime }dt^{\prime \prime This is of Einstein and relations 1/f-fluctuations. UNIFICATION OF NONEQUILIBRIUM BROWNIAN MOTION ------------------------------- The producing reformulated in terms of characteristic function (CF): $$\left\langle [(ik-\frac xT)Q(t)]\right\rangle \exp [-ikQ(t)]\right\rangle$$ This is functional equation respect to dependencies on $ik$ and $x$. It to find the connections between the whole statistics of nonequilibrium random walk and that of in presence of 1/f-noise, we know something statistics neglecting \[4,6,8,38\]. of possible on the functional equation is $\left\langle \exp [ikQ(t)]\right\rangle =\Xi (ikx/T-k^2,t)$, with two independent arguments. It is easy to see that
cumulant and cubic response. THe latter coUld be Due To eItHer nOnliNear dissipatioN MechAnisms or to dependence on ActuaL cURrenT TeMperaTure maiNTaINEd bY JOuLeaN hEAtIng. BuT, in Any case, The left-hanD siDe Can not grow in A MoRe fast way tHan $\Propto t$. On the ConTrary, iN pResENce of $F^{-\gaMma }$ exCess noISe the fIrst term oN rIGht-hanD Side shoULD gRow aS $\propto t^{1+\gamma }$. HencE, At SUfficiently lonG time iNtERvALS it MusT be compensAtEd by tHE second TErM, AND we COme to relationS between low-FIelD excesS nOneQUilibrIum noIsE And Four-point cuMulaNt of equilIbrium NOise. The REsult caN be wriTteN alSo as $k_M(\tAu )=\FraC 1{T^2}k_d(\taU )$ \[4,6,9\], WhEre $k_d(\taU )$ is the abOvE dEfineD corRELATor oF diFfusIvity And $K_m(\tau )$ is the MobIlitY CorRelatIon fuNctiOn As intRoduceDd by tHe Expansion $\left\laNgle q(t),Q(t)\right\RanGlE =\leFt\LanglE q(t),Q(t)\riGht\RanGle _0+x^2\int_0^T\int_0^tK_m(T^{\PriMe }-T^{\PRImE \prime })dt^{\prime }dt^{\priMe \PRImE }+O(x^4)$. This iS the exTEnSiON of EinstEiN reLatiON And siMilaR ReLations tO 1/f-flucTUaTiOns. UNIFiCaTION Of EqUIlIBrIUM And\ NONeQUILIbRIUM BROwNIAN moTION ------------------------------- The producINg FDR can be refORmULAtED in tErmS of characteRistIC funCtioN (cF): $$\LefT\LanglE \exp [(iK-\fRAc Xt)Q(t)]\right\rangle =\left\lAnGle \exp [-IkQ(t)]\rIght\rangle$$ ThiS is functioNAL Equation With REsPEct to dependencIes on $Ik$ and $x$. It heLPs to find The coNnectionS between tHE Whole staTisTicS of NonEQUiLibrium random WALk anD tHat of eqUilIbrium oNe, iN prEseNce Of 1/F-noise, if wE know somEtHiNg AbOut StatiSTics undeR nEglEcTinG 1/f-noiSE \[4,6,8,38\]. One of PossiBle sOlUtIOns On the fuNCtIONal eQuAtIon iS $\leFt\LanglE \exp [IKQ(t)]\Right\raNgle =\Xi (ikx/t-k^2,t)$, WIth oNlY tWo indepEndent argumenTs. it is easy to SeE thAt
cumulant and cubic respon se. The la ttercou ldbe due toeither nonline a r di ssipation mechanisms o r tode p ende n ce on a ctual c u rr e n t t em pe rat ur e m ainta ine d by Jo ulean heat ing .But, in anyc as e, the lef t-h and side can no t grow i n a morefas t way than$ \propt o t$. Onth e contr a ry, inp r es ence of $f^{-\gamma } $ e x cess noise the first t e rm o n r igh t-hand sid eshoul d grow a s $ \ p r opt o t^{1+\gamma}$. Hence,a t s uffici en tly long t ime i nt e rva ls it mustbe c ompensate d by t h e secon d term,and we co meto r e la ti ons b e twe e nlow - fie ld exces sno nequi libr i u m nois e a nd f our-p oint cumulant of equ i lib riumnoise . Th eresul t canbe wr it ten also as $K_ m(\t au )=\fra c 1 {T ^2} K_ D(\ta u )$ \[ 4,6 ,9\ ], wher e $K_D( \ tau ) $ i sthe above definedco r r el ator ofdiffus i vi ty and $K_m (\ tau )$i s themobi l it y correl ationf un ct ion asin troduc ed d b y t he ex p ansi on $\l eft\lang le Q( t ),Q(t)\right\r a ngle =\left\l a ng l e Q ( t),Q (t) \right\rang le _ 0 +x^2 \int _ 0^ t\i n t_0^t K_m(t ^{ \ pr i me }-t^{\prime \pri me })dt^ {\pri me }dt^{\prim e \prime } + O ( x^4)$. T hisi st he extension o f Ein stein rela t ion andsimil ar relat ions to 1 / f -fluctua tio ns. U NIF I C AT ION OF EQUILI B R IUMAN D\ NONE QUI LIBRIUM BR OWN IAN MO TI ON ------ -------- -- -- -- -- --- ----- - The pr od uci ng FD R can be ref ormul ated i nt erm s of ch a ra c t eris ti cfunc tio n(CF): $$\ l eft \langle \exp [(i k-\ f racxT )Q (t)]\ri ght\rangle =\ le ft\langle\e xp[-ikQ( t ) ]\right\ rangle$$ This is functi o nal equ ati on wi th r espect to de penden cie s on $i k$ and $x$. I t h e l ps to f in d t he connectio n s be tween t he w hole st atistics of nonequ i lib rium random w alk and t ha t o f e q uil ib r ium o ne, in presence of 1/f-no is e ,if we know som et hing ab out sta tisti c s under neglecti ng 1/f-no is e \[ 4 , 6,8 ,38\]. On e of pos sible sol u tions on thefun ctiona lequ ation is $\ l eft \lang le \ex p[ikQ(t )]\ri gh t\rangle =\Xi (ikx/T-k^2,t)$, w ith on ly tw o i ndependen t a r gum ents. Itis e asy to see th at
cumulant_and cubic_response. The latter could_be due_to_either nonlinear_dissipation_mechanisms or to_dependence on actual_current temperature maintained by_Joulean heating. But,_in_any case, the left-hand side can not grow in a more fast way than_$\propto_t$. On_the_contrary,_in presence of $f^{-\gamma }$_excess noise the first term_on right-hand_side should grow as $\propto t^{1+\gamma }$. Hence,_at_sufficiently long time_intervals it must be compensated by the second term,_and we come to relations between_low-field excess nonequilibrium_noise_and_four-point cumulant of equilibrium_noise. The result can be written_also as $K_m(\tau )=\frac 1{T^2}K_D(\tau )$_\[4,6,9\], where $K_D(\tau )$ is the above_defined correlator of diffusivity and $K_m(\tau_)$ is the mobility correlation_function as_introducedd by the expansion $\left\langle_Q(t),Q(t)\right\rangle =\left\langle Q(t),Q(t)\right\rangle _0+x^2\int_0^t\int_0^tK_m(t^{\prime_}-t^{\prime \prime })dt^{\prime_}dt^{\prime \prime }+O(x^4)$._This is the extension of Einstein_relation and similar_relations to 1/f-fluctuations. UNIFICATION OF EQUILIBRIUM AND\ NONEQUILIBRIUM_BROWNIAN_MOTION ------------------------------- The producing FDR_can_be_reformulated in_terms of characteristic_function_(CF): $$\left\langle_\exp_[(ik-\frac xT)Q(t)]\right\rangle =\left\langle \exp [-ikQ(t)]\right\rangle$$ This is_functional_equation with respect to dependencies on $ik$_and $x$. It helps_to_find the connections between_the whole statistics of nonequilibrium_random walk and that of equilibrium_one, in_presence of_1/f-noise, if we know something about statistics under neglecting 1/f-noise \[4,6,8,38\]. One_of possible solutions on the functional_equation is $\left\langle \exp [ikQ(t)]\right\rangle_=\Xi (ikx/T-k^2,t)$,_with_only two independent_arguments._It is_easy to see that
Mitas L, Drobný G, Wagner L K and Schmidt K E 2006 *Phys. Rev. Lett.* **96** 130201 Bajdich M, Mitas L, Wagner L K and Schmidt K E 2008 *Phys. Rev.* B **77** 115112 Lüchow A, Petz R and Scott T C 2007 *J. Chem.Phys.* **126** 144110 F A Reboredo, R Q Hood and P R C Kent 2009 *Phys. Rev.* B **79** 195117 Conroy H 1964 *J. Chem. Phys.* **41** 1331 Umrigar C J, Wilson K G and Wilkins J W 1988 *Phys. Rev. Lett.* **60** 1719 Dewing M and Ceperley D M 2002 *Methods for Coupled Electronic-Ionic Monte Carlo* in *Recent Advances in Quantum Monte Carlo Methods, Part II*, ed by Lester W A, Rothstein S M and Tanaka S (World Scientific, Singapore) Drummond N D and Needs R J 2005 *Phys.Rev.* B **72** 085124 Moroni S, Fantoni S and Senatore G 1995 *Phys. Rev. B* **52** 13547 Riley K E and Anderson J B 2003 *Mol. Phys.* **101** 3129 Nightingale M P and Melik-Alaverdian V 2001 *Phys. Rev. Lett.* **87** 043401 Umrigar C J, Toulouse J, Filippi C, Sorella S and Hennig R G 2007 *Phys. Rev. Lett.* **98** 110201 Toulouse J and Umrigar C J 2007 *J. Chem.Phys.* **126** 084102 Rajagopal G, Needs R J, Kenny S, Foulkes W M C and James A 1994 *Phys. Rev. Lett.* **73** 1959 Rajagopal G, Needs R J, James A, Kenny S D and Foulkes W M C 1995 *Phys. Rev.* B **51** 10591 Lin C, Zong F H and C
Mitas L, Drobný G, Wagner L K and Schmidt K E 2006 * Phys.   Rev.   Lett. *   * * 96 * * 130201 Bajdich M, Mitas L, Wagner L K and Schmidt K E 2008 * Phys.   Rev. *   B * * 77 * * 115112 Lüchow A, Petz R and Scott T C 2007 * J.   Chem. Phys. *   * * 126 * * 144110 F A Reboredo, R Q Hood and P R C Kent 2009 * Phys.   Rev. *   B * * 79 * * 195117 Conroy H 1964 * J.   Chem.   Phys. *   * * 41 * * 1331 Umrigar C J, Wilson K G and Wilkins J W 1988 * Phys.   Rev.   Lett. *   * * 60 * * 1719 Dewing M and Ceperley D M 2002 * Methods for Coupled Electronic - Ionic Monte Carlo * in * Recent Advances in Quantum Monte Carlo Methods, Part II *, erectile dysfunction by Lester W A, Rothstein S M and Tanaka S (World Scientific, Singapore) Drummond N D and need R J 2005 * Phys. Rev. *   B * * 72 * * 085124 Moroni S, Fantoni S and Senatore G 1995 * Phys. Rev.   B * * * 52 * * 13547 Riley K E and Anderson J B 2003 * Mol.   Phys. * * * 101 * * 3129 Nightingale M P and Melik - Alaverdian five 2001 * Phys.   Rev.   Lett. *   * * 87 * * 043401 Umrigar C joule, Toulouse J, Filippi C, Sorella S and Hennig R G 2007 * Phys.   Rev.   Lett. *   * * 98 * * 110201 Toulouse J and Umrigar C J 2007 * J.   Chem. Phys. *   * * 126 * * 084102 Rajagopal G, Needs R J, Kenny S, Foulkes W M C and James A 1994 * Phys.   Rev.   Lett. *   * * 73 * * 1959 Rajagopal G, Needs R J, James A, Kenny S D and Foulkes W M C 1995 * Phys.   Rev. *   B * * 51 * * 10591 Lin C, Zong F H and cytosine
Mihas L, Drobný G, Wagner L K and Schmidt K G 2006 *Phys. Rxv. Lett.* **96** 130201 Gajdich O, Mitas L, Wagner L K and Schliet K T 2008 *Phys. Rev.* B **77** 115112 Lüchow X, Petz R wnd Scotr T R 2007 *J. Chem.Phys.* **126** 144110 F A Reboredo, R Q Glod cnv P R C Kent 2009 *Pmys. Rev.* B **79** 195117 Cotroy H 1964 *J. Chem. Pvyr.* **41** 1331 Bmrigar C J, Wilson K G and Wilkins J W 1988 *Phyx. Rfv. Lett.* **60** 1719 Dewing I anc Cepsglty D M 2002 *Methods for Coupled Electdonic-Ioiic Monte Carlo* in *Recent Advances in Quajtum Monte Carlo Methofs, Part II*, gs br Lester W A, Fothstein S M and Tanaia S (World Scientific, Singapore) Arummpnd N D ane Beefv R J 2005 *Phys.Cev.* B **72** 085124 Ioroni S, Fanbpni S dnd Senstore G 1995 *Phys. Vev. B* **52** 13547 Rioey K E and Anderson O B 2003 *Mol. Phys.* **101** 3129 Nightiggale M P aud Melik-Alaverdian V 2001 *Pyys. Ree. Letd.* **87** 043401 Umfugaf C J, Tkulousf J, Filippi C, Sorella S qnd Hennig R G 2007 *Phyx. Rqn. Kett.* **98** 110201 Toulouss J anq Tmrigar C J 2007 *J. Chem.Phys.* **126** 084102 Rajagopal G, Needv R J, Kenny S, Foulkes W M X and James A 1994 *Phys. Reg. Lett.* **73** 1959 Rajwgopal G, Needs R J, James A, Kenny S D and Foulkes F M C 1995 *Phvw. Rev.* B **51** 10591 Lln C, Zong F H and C
Mitas L, Drobný G, Wagner L K K 2006 *Phys. Lett.* **96** 130201 L and Schmidt K 2008 *Phys. Rev.* **77** 115112 Lüchow A, Petz R Scott T C 2007 *J. Chem.Phys.* **126** 144110 F A Reboredo, R Q and P R C Kent 2009 *Phys. Rev.* B **79** 195117 Conroy H *J. Phys.* 1331 C J, Wilson K G and Wilkins J W 1988 *Phys. Rev. Lett.* **60** 1719 Dewing and Ceperley D M 2002 *Methods for Coupled Monte Carlo* in *Recent in Quantum Monte Carlo Methods, II*, by Lester A, S and Tanaka S Scientific, Singapore) Drummond N D and Needs R J 2005 *Phys.Rev.* B **72** 085124 Moroni S, Fantoni and Senatore *Phys. Rev. **52** Riley E and Anderson 2003 *Mol. Phys.* **101** 3129 Nightingale Melik-Alaverdian V 2001 *Phys. Rev. Lett.* **87** 043401 C J, J, Filippi C, Sorella S and R G 2007 *Phys. Rev. Lett.* **98** 110201 J and Umrigar C J 2007 *J. Chem.Phys.* **126** 084102 Rajagopal G, Needs R J, Foulkes W M C James A 1994 Rev. **73** Rajagopal Needs R James A, Kenny S D and Foulkes W M C 1995 Rev.* B **51** 10591 Lin C, Zong F H and
Mitas L, Drobný G, Wagner L K and SChmidt K E 2006 *PhYs. Rev. letT.* **96** 130201 BaJdIch M, mitaS L, Wagner L K and SCHmidT K E 2008 *Phys. Rev.* B **77** 115112 Lüchow A, Petz r and SCoTT T C 2007 *J. cHeM.Phys.* **126** 144110 f A ReborEDo, r q hooD aNd p R C keNT 2009 *PHys. ReV.* B **79** 195117 COnroy H 1964 *J. chem. Phys.* **41** 1331 UmRigAr c J, Wilson K G anD wiLkins J W 1988 *PhyS. ReV. Lett.* **60** 1719 Dewing M And ceperlEy d M 2002 *METhods For couplEd ElecTRonic-IOnic Monte caRLo* in *ReCEnt AdvaNCEs In QuAntum Monte Carlo MeTHoDS, Part II*, ed by LesTer W A, ROtHStEIN S M And tanaka S (WorLd scienTIfic, SinGApORE) druMMond N D and NeedS R J 2005 *Phys.Rev.* B **72** 085124 mOroNi S, FanToNi S ANd SenaTore G 1995 *phYS. ReV. B* **52** 13547 Riley K E anD AndErson J B 2003 *MoL. Phys.* **101** 3129 NIGhtingaLE M P and MElik-AlAveRdiAn V 2001 *PHYs. reV. LeTt.* **87** 043401 uMriGAr c J, TOUloUse J, FiliPpI C, sorelLa S aND hENnig r G 2007 *PHys. REv. LetT.* **98** 110201 Toulouse J and umrIgar c j 2007 *J. CHem.PhYs.* **126** 084102 RajAgopAl g, NeedS R J, KenNy S, FoUlKes W M C and James A 1994 *phys. rev. Lett.* **73** 1959 RaJagOpAl G, neEds R J, jAmes A, KEnnY S D And FoulKes W M C 1995 *PHYs. REv.* b **51** 10591 lIN C, zong F H and C
Mitas L, Drobný G, Wagner L K and S chmid t K E20 06 * Phys . Rev. Lett.** *96* * 130201 Bajdich M, M itasL, Wagn e rL K a nd Schm i dt K E20 08 *P hy s .Rev.*  B**77**115112 Lü cho wA, Petz R an d S cott T C 2 007 *J. Chem.Ph ys. * **12 6* * 1 4 4110 FA Reb oredo, R Q Ho od and PRC Kent2 009 *Ph y s .Rev. * B **79** 195117 C o nroy H 1964 *J . Chem .P hy s . * * *41 ** 1331 U mr igarC J, Wil s on K G a n d Wilkins J W 1988 *Phys .  Re v. Let t. * * * 60** 1 719 De w ing M and Cepe rley D M 2002 *Meth o ds forC oupledElectr oni c-I onic Mo nt e C ar l o*i n*Re c ent Advance sin Quan tumM o n t e Ca rlo Met hods, Part II*, ed by Les t erW A,Roths tein S M an d Tana ka S(W orld Scientific , Si ngapore) Dr um mon dN D a n d Need s R J2005 *P hys.Rev . * B * * 7 2 ** 085124 Moroni S, F a n to ni S and Senat o re G 1995 *Ph ys . R ev.B * **52 ** 1 3 54 7 Riley K E a n dAn dersonJB 2003 * Mol . P hys.* **10 1** 31 29 Nigh tinga l e M P and Meli k -Alaverdian V 20 0 1 * P hys.  Re v. Lett.* * *87* * 043 401Um rig a r C J , Tou lo u se J, Filippi C, Sorel la S and Henn ig R G 2007 * Phys. Rev. L e tt.* **9 8**1 10 2 01 Toulouse J andUmrigar CJ 2007 *J . Che m.Phys.*  **126**0 8 4102 Ra jag opa l G , N e e ds R J, Kenny S , Foul ke s W M C an d James A199 4 * Phy s.  Rev. Let t.* **73 ** 1 95 9 Ra jagop a l G, Nee ds RJ, Ja mes A , Kenny S DandFo ul k esW M C 1 9 95 * Phys .Re v.*B * *5 1** 1 0591 Li n C, Zo ng F H an d C
Mitas_L, Drobný_G, Wagner L K_and Schmidt_K_E 2006_*Phys. Rev. Lett.* **96**_130201 Bajdich M, Mitas_L, Wagner L_K and Schmidt K_E 2008 *Phys. Rev.* B_**77**_115112 Lüchow A, Petz R and Scott T C 2007 *J. Chem.Phys.* **126** 144110 F A Reboredo, R_Q_Hood and_P_R_C Kent 2009 *Phys. Rev.* B **79**_195117 Conroy H 1964 *J. Chem. Phys.* **41** 1331 Umrigar_C J,_Wilson K G and Wilkins J W 1988_*Phys. Rev. Lett.* **60**_1719 Dewing M and_Ceperley D M 2002 *Methods for Coupled Electronic-Ionic Monte_Carlo* in *Recent Advances in Quantum_Monte Carlo Methods,_Part_II*,_ed by Lester W_A, Rothstein S M and Tanaka_S (World Scientific, Singapore) Drummond N D_and Needs R J 2005 *Phys.Rev.* B **72**_085124 Moroni S, Fantoni S and Senatore_G 1995 *Phys. Rev. B* **52**_13547 Riley K_E and Anderson J B_2003 *Mol. Phys.* **101**_3129 Nightingale M_P and Melik-Alaverdian_V 2001 *Phys. Rev. Lett.* **87** 043401 Umrigar C J,_Toulouse J, Filippi_C, Sorella S and Hennig R_G_2007 *Phys. Rev. Lett.* **98** 110201 Toulouse_J_and_Umrigar C_J 2007 *J. Chem.Phys.* **126**_084102 Rajagopal_G, Needs_R_J, Kenny S, Foulkes W M_C_and James A 1994 *Phys. Rev. Lett.* **73** 1959 Rajagopal G,_Needs R J, James_A,_Kenny S D and_Foulkes W M C 1995_*Phys. Rev.* B **51** 10591 Lin C, Zong F_H and_C
M$ will generally have different kinematic properties to the structure involving $T^{(2)}M$. This difference is difficult to see if we only consider the configuration space itself but not how it is constructed. For example, when $M=\mathbb{R}^n$, both $(T\oplus T)M$ and $T^{(2)}M$ can be considered as (or more properly, are diffeomorphic to) $\mathbb{R}^{3n}$, but there are nontheless significant differences, namely that for quadratic energy, one model supports torque while the other does not! (In fact, we can construct an isomorphism of $(T\oplus T)M$ and $T^{(2)}M$ as fibre bundles using a metric – see appendix **\[subapp:vectTNM\]** for more details. However, the isomorphism depends on the metric, and the complete lifts of vector fields to $(T\oplus T)M$ and $T^{(2)}M$ respectively do not coincide under the isomorphism.) Dissipation bracket for $3$-bead chains {#subsec:diss3bead} --------------------------------------- We continue to follow the same strategy as the bead-spring pairs to construct the dissipation bracket for the $3$-bead chain suspension. Let $M = \mathbb{R}^n$, so we can identify $T^{(2)}M$ with $\mathbb{R}^{3n}$ with coordinates $(x^i,y^i,z^i)$. Consider the usual Riemannian metric $g$ on $\mathbb{R}^{3n}$, which can be written as $$\begin{aligned} \mathrm{d}s^2 = \delta_{ij}\mathrm{d}x^i\mathrm{d}x^j + \delta_{ij}\mathrm{d}y^i\mathrm{d}y^j + \delta_{ij}\mathrm{d}z^i\mathrm{d}z^j.\end{aligned}$$ The construction of Riemannian metrics on higher order tangent bundles of general Riemannian manifolds is considered in appendix **\[app:vect\]**. We can define the dissipation bracket $$\begin{aligned} (F,G) = \int \mathrm{d}^nx\mathrm{d}^ny\mathrm{d}^nz \ \psi \ \widetilde{g}\left( \mathrm{d}\left(\frac{\delta F}{\delta \psi}\right), \mathrm
M$ will generally have different kinematic properties to the social organization imply $ T^{(2)}M$. This difference is difficult to visualize if we entirely consider the configuration outer space itself but not how it is constructed. For example, when $ M=\mathbb{R}^n$, both $ (T\oplus T)M$ and $ T^{(2)}M$ can be consider as (or more properly, are diffeomorphic to) $ \mathbb{R}^{3n}$, but there are nontheless meaning differences, namely that for quadratic energy, one model supports torsion while the other does not! (In fact, we can manufacture an isomorphism of $ (T\oplus T)M$ and $ T^{(2)}M$ as fibre bundles use a metric – see appendix * * \[subapp: vectTNM\ ] * * for more details. However, the isomorphism count on the metric, and the complete lifts of vector fields to $ (T\oplus T)M$ and $ T^{(2)}M$ respectively do not concur under the isomorphism .) Dissipation bracket for $ 3$-bead chains { # subsec: diss3bead } --------------------------------------- We cover to follow the same strategy as the bead - give pairs to construct the dissipation bracket for the $ 3$-bead chain suspension. Let $ M = \mathbb{R}^n$, so we can identify $ T^{(2)}M$ with $ \mathbb{R}^{3n}$ with coordinates $ (x^i, y^i, z^i)$. Consider the usual Riemannian metric $ g$ on $ \mathbb{R}^{3n}$, which can be written as $ $ \begin{aligned } \mathrm{d}s^2 = \delta_{ij}\mathrm{d}x^i\mathrm{d}x^j + \delta_{ij}\mathrm{d}y^i\mathrm{d}y^j + \delta_{ij}\mathrm{d}z^i\mathrm{d}z^j.\end{aligned}$$ The construction of Riemannian metrics on higher order tangent package of general Riemannian manifolds is view in appendix * * \[app: vect\ ] * *. We can specify the dissipation bracket $ $ \begin{aligned } (F, G) = \int \mathrm{d}^nx\mathrm{d}^ny\mathrm{d}^nz \ \psi \ \widetilde{g}\left ( \mathrm{d}\left(\frac{\delta F}{\delta \psi}\right), \mathrm
M$ wlll generally have diffevent kinematic ptopertiev to tge strucgure involving $T^{(2)}M$. This differxnce is dufficult to see if we unly conspder the xonfmguration space mfself bmc not mow ic ms constructed. Nor example, when $M=\mathbb{R}^t$, coch $(T\oplus T)M$ and $T^{(2)}M$ can be considereq as (or mlre properly, ate digseomkgpmic to) $\mathbb{R}^{3n}$, but there are nohthelesv significant differences, namely that flr qkadratic energy, onf model supkkrtf torque whild the otheg does not! (Jn fact, we can construct an isooorphnsm of $(T\oplos T)L$ and $T^{(2)}M$ as hibre fundles usinn a metsic – ser appendix **\[subspp:tectRNM\]** for more details. Iowever, the isomorphysm depengs on the metric, ane rhe cmmplate uuftr or tecfor fiflda to $(T\oplua T)M$ and $T^{(2)}M$ respectively do nou cjpmcide under fhe isjmjrphism.) Dissipation bracket for $3$-bead chapns {#aubsec:diss3bead} --------------------------------------- We continye to follow the same strategy as the bead-spring pairs to construct the dissipadion urxcktt for gye $3$-bead chain suspension. Let $M = \mathbb{R}^n$, so we cwh odvntify $T^{(2)}M$ with $\mabhbb{R}^{3n}$ with coordimahex $(x^i,y^i,z^i)$. Consiaer thz uaual Riemannian mehric $g$ jn $\marhbb{R}^{3n}$, whych van be written as $$\begin{aligbed} \mathrm{d}s^2 = \eelta_{ij}\mathrm{d}x^i\machrm{d}x^j + \delca_{ij}\majhrm{d}y^o\mathrm{d}y^j + \delta_{ij}\mathxm{d}z^i\mzthrm{d}z^j.\end{wligned}$$ Tgd construction ow Rpematnian meufics on higher orqer tangeit buudles of gengral Riqmannian mwnifolds is considered ij appgndix **\[dpp:vect\]**. We fan define the dissipation bracket $$\begin{aligngd} (F,C) = \pnt \mathri{d}^nx\msthrm{d}^ny\mathri{d}^nz \ \psi \ \widgtilde{g}\leyt( \maghrm{d}\left(\fgac{\delta H}{\delta \psi}\ridht), \mathrm
M$ will generally have different kinematic properties structure $T^{(2)}M$. This is difficult to the space itself but how it is For example, when $M=\mathbb{R}^n$, both $(T\oplus and $T^{(2)}M$ can be considered as (or more properly, are diffeomorphic to) $\mathbb{R}^{3n}$, there are nontheless significant differences, namely that for quadratic energy, one model supports while other not! fact, we can construct an isomorphism of $(T\oplus T)M$ and $T^{(2)}M$ as fibre bundles using a – see appendix **\[subapp:vectTNM\]** for more details. However, isomorphism depends on the and the complete lifts of fields $(T\oplus T)M$ $T^{(2)}M$ do coincide under the Dissipation bracket for $3$-bead chains {#subsec:diss3bead} --------------------------------------- We continue to follow the same strategy as the bead-spring to construct bracket for $3$-bead suspension. $M = \mathbb{R}^n$, can identify $T^{(2)}M$ with $\mathbb{R}^{3n}$ with the usual Riemannian metric $g$ on $\mathbb{R}^{3n}$, which be written $$\begin{aligned} \mathrm{d}s^2 = \delta_{ij}\mathrm{d}x^i\mathrm{d}x^j + \delta_{ij}\mathrm{d}y^i\mathrm{d}y^j \delta_{ij}\mathrm{d}z^i\mathrm{d}z^j.\end{aligned}$$ The construction of Riemannian metrics on higher tangent bundles of general Riemannian manifolds is considered in appendix **\[app:vect\]**. We can define the $$\begin{aligned} (F,G) = \int \ \psi \ \mathrm{d}\left(\frac{\delta \psi}\right),
M$ will generally have differeNt kinematiC propErtIes To The sTrucTure involving $T^{(2)}m$. this Difference is difficult tO see iF wE Only COnSider The confIGuRATioN sPaCe iTsELf But noT hoW it is coNstructed. FOr eXaMple, when $M=\matHBb{r}^n$, both $(T\oplUs T)m$ and $T^{(2)}M$ can be cOnsIdered As (Or mORe proPerLy, are DiffeoMOrphic To) $\mathbb{R}^{3N}$, bUT there ARe nonthELEsS sigNificant differencES, nAMely that for quaDratic EnERgY, ONe mOdeL supports tOrQue whILe the otHEr DOES noT! (in fact, we can coNstruct an isOMorPhism oF $(T\OplUS T)M$ and $t^{(2)}M$ as fIbRE buNdles using a MetrIc – see appeNdix **\[suBApp:vecttnM\]** for moRe detaIls. howEver, THe IsOmoRpHIsm DEpEndS On tHe metric, AnD tHe comPletE LIFTs of VecTor fIelds To $(T\oplus T)M$ and $t^{(2)}M$ rEspeCTivEly do Not coInciDe Under The isoMorphIsM.) Dissipation braCket For $3$-bead chAinS {#sUbsEc:Diss3bEAd} --------------------------------------- We coNtiNue To folloW the samE StrAtEGY As The bead-spring pairs To CONsTruct the DissipATiOn BRacket foR tHe $3$-bEad cHAIn susPensIOn. let $M = \mathBb{R}^n$, so WE cAn IdentifY $T^{(2)}m$ with $\mAtHbb{r}^{3n}$ wIth coORdinAtes $(x^i,Y^i,z^i)$. ConsIder tHE usual RiemanniAN metric $g$ on $\matHBb{r}^{3N}$, WhICh caN be Written as $$\beGin{aLIgneD} \matHRm{D}s^2 = \dELta_{ij}\MathrM{d}X^I\mAThrm{d}x^j + \delta_{ij}\mathrM{d}Y^i\mathRm{d}y^j + \Delta_{ij}\mathrm{D}z^i\mathrm{d}Z^J.\ENd{aligneD}$$ The COnSTruction of RiemAnniaN metrics on HIgher ordEr tanGent bundLes of geneRAL RiemannIan ManIfoLds IS CoNsidered in appENDix **\[aPp:Vect\]**. We cAn dEfine thE diSsiPatIon BrAcket $$\begiN{aligned} (f,G) = \InT \mAtHrm{D}^nx\maTHrm{d}^ny\maThRm{d}^Nz \ \Psi \ \WidetILde{g}\leFt( \matHrm{d}\LeFt(\FRac{\Delta F}{\dELtA \PSi}\riGhT), \mAthrM
M$ will generally have dif ferent kin emati c p rop er ties tothe structurei nvol ving $T^{(2)}M$. This diff er e ncei sdiffi cult to se e ifwe o nly c o ns iderthe config uration sp ace i tself but no t h ow it is c ons tructed. For ex ample, w hen $M=\m ath bb{R} ^n$, b o th $(T \oplus T) M$ and $T ^ {(2)}M$ c an beconsidered as (or mo r e properly, ar e diff eo m or p h icto) $\mathbb{ R} ^{3n} $ , but t h er e a ren ontheless sig nificant di f fer ences, n ame l y that forqu a dra tic energy, one model su pports torquew hile th e othe r d oes not ! ( In fa ct , we ca n c o nst ruct anis om orphi sm o f $ ( T\op lus T)M $ and $T^{(2)}M$ a s f ibre bun dlesusing a m et ric – see a ppend ix **\[subapp:vec tTNM \]** formor edet ai ls. H o wever, th e i somorph ism dep e nds o n t he metric, and the c om p l et e liftsof vec t or f i elds to$( T\o plus T )M$ a nd $ T ^{ (2)}M$ r espect i ve ly do not c oincid eund erthe i s omor phism. ) Dissi patio n bracket for $ 3 $-bead chains {# s u bs e c:di ss3 bead} ----- ---- - ---- ---- - -- --- - ----- ----- -- - -We continue to foll ow the s ame s trategy as th e bead-spr i n g pairs t o co n st r uct the dissip ation bracket f o r the $3 $-bea d chainsuspensio n . Let $M =\ma thb b{R } ^ n$ , so we can i d e ntif y$T^{(2) }M$ with $ \ma thb b{R }^{ 3n }$ with c oordinat es $ (x ^i ,y^ i,z^i ) $. Consi de r t he us ual R i emanni an me tric $ g$ on$\mathb b {R } ^ {3n} $, w hich ca nbe wr itte n as $$\beg in{aligne d}\ math rm {d }s^2 =\delta_{ij}\m at hrm{d}x^i\ ma thr m{d}x^ j + \delta _{ij}\mathrm{d}y^i\math r m{d}y^j +\delt a_{i j}\mathrm {d} z^i\ma thr m {d}z^j .\end{ align ed }$$ T he co n s tr uct io n of Riema n n ian metr ic s on higher order tangent bun d les of general R iem anni a n m ani f ol d s i sc ons i d ered in appendi x **\[app: ve c t\ ]**. We c a n d ef ine the dissip ation bracket $$\begin {aligned} ( F,G) = \i nt \mathrm {d}^nx\m athrm{d}^ n y\mat h rm {d}^n z \ \psi\\wi detil de{g}\ l eft ( \m athrm{ d} \left( \frac {\ delta F} {\delta \psi}\right), \ mathrm
M$ will_generally have_different kinematic properties to_the structure_involving_$T^{(2)}M$. This difference_is_difficult to see_if we only_consider the configuration space_itself but not_how_it is constructed. For example, when $M=\mathbb{R}^n$, both $(T\oplus T)M$ and $T^{(2)}M$ can be_considered_as (or_more_properly,_are diffeomorphic to) $\mathbb{R}^{3n}$, but_there are nontheless significant differences,_namely that_for quadratic energy, one model supports torque while_the_other does not!_(In fact, we can construct an isomorphism of $(T\oplus_T)M$ and $T^{(2)}M$ as fibre bundles_using a metric_–_see_appendix **\[subapp:vectTNM\]** for more_details. However, the isomorphism depends on_the metric, and the complete lifts_of vector fields to $(T\oplus T)M$ and_$T^{(2)}M$ respectively do not coincide under_the isomorphism.) Dissipation bracket for $3$-bead_chains {#subsec:diss3bead} --------------------------------------- We_continue to follow the same_strategy as the_bead-spring pairs_to construct the_dissipation bracket for the $3$-bead chain_suspension. Let $M =_\mathbb{R}^n$, so we can identify $T^{(2)}M$_with_$\mathbb{R}^{3n}$ with coordinates_$(x^i,y^i,z^i)$._Consider_the usual_Riemannian metric $g$_on_$\mathbb{R}^{3n}$, which_can_be written as $$\begin{aligned} \mathrm{d}s^2 = \delta_{ij}\mathrm{d}x^i\mathrm{d}x^j_+_\delta_{ij}\mathrm{d}y^i\mathrm{d}y^j + \delta_{ij}\mathrm{d}z^i\mathrm{d}z^j.\end{aligned}$$ The construction of Riemannian_metrics on higher order_tangent_bundles of general Riemannian_manifolds is considered in appendix_**\[app:vect\]**. We can define the dissipation bracket_$$\begin{aligned} (F,G) =_\int \mathrm{d}^nx\mathrm{d}^ny\mathrm{d}^nz_\ \psi \ \widetilde{g}\left( \mathrm{d}\left(\frac{\delta F}{\delta \psi}\right), \mathrm
, T. Vidick, arXiv:1210.1810 \[quant-ph\]. J. Silman et al., Phys. Rev. Lett. **106**, 220501 (2011). C.-E. Bardyn et al., Phys. Rev. A **80**, 062327 (2009). M. McKague, T.H. Yang, and V. Scarani, arXiv:1203.2976 [\[]{}quant-ph[\]]{}. J.-D. Bancal et al., Phys. Rev. Lett. 106, 250404 (2011). F. Magniez et al., in Proceedings of the 33rd International Colloquium on Automata, Languages and Programming (Springer, 2006), p. 72. M. McKague and M. Mosca, in Proceedings of the 5th Conference on Theory of Quantum Computation, Communication, and Cryptography (Springer, 2011), p. 113. R. Colbeck, PhD dissertation, Univ. Cambridge (2007), arXiv:0911.3814 [\[]{}quant-ph[\]]{}; R. Colbeck and A. Kent, J. Phys. A 44, 095305 (2011). S. Pironio et al., Nature 464, 1021 (2010). S. Pironio and S. Massar, arXiv:1111.6056 [\[]{}quant-ph[\]]{}. S. Fehr, R. Gelles, and C. Schaffner, arXiv:1111.6052 [\[]{}quant-ph[\]]{}. U. Vazirani and T. Vidick, Phil. Trans. R. Soc. A **370**, 3432 (2012). Y. Zhao et al., Phys. Rev. A **78**, 042333 (2008). F. Xu, B. Qi, and H.-K. Lo, New J. Phys. **12**, 113026 (2010). L. Lydersen et al., Nature Photonics **4**, 686 (2010). M. Ansmann et al., Nature **461**, 504 (2009). M.A. Rowe et al., Nature 409, 791 (2001). T. Monz et al., Phys. Rev. Lett. **106**, 130506 (2011). J.
, T. Vidick, arXiv:1210.1810 \[quant - ph\ ]. J. Silman et al. , Phys. Rev. Lett. * * 106 * *, 220501 (2011). C.-E. Bardyn et al. , Phys. Rev. A * * 80 * *, 062327 (2009). M. McKague, T.H. Yang, and V. Scarani, arXiv:1203.2976 [ \[]{}quant - ph[\ ] ] { }. J.-D. Bancal et al. , Phys. Rev. Lett. 106, 250404 (2011). F. Magniez et al. , in Proceedings of the 33rd International Colloquium on Automata, Languages and Programming (Springer, 2006), p. 72. M. McKague and M. Mosca, in Proceedings of the 5th Conference on Theory of Quantum Computation, Communication, and Cryptography (Springer, 2011), p. 113. R. Colbeck, PhD dissertation, Univ. Cambridge (2007), arXiv:0911.3814 [ \[]{}quant - ph[\ ] ] { }; R. Colbeck and A. Kent, J. Phys. A 44, 095305 (2011). S. Pironio et al. , Nature 464, 1021 (2010). S. Pironio and S. Massar, arXiv:1111.6056 [ \[]{}quant - ph[\ ] ] { }. S. Fehr, R. Gelles, and C. Schaffner, arXiv:1111.6052 [ \[]{}quant - ph[\ ] ] { }. U. Vazirani and T. Vidick, Phil. Trans. R. Soc. A * * 370 * *, 3432 (2012). Y. Zhao et al. , Phys. Rev. A * * 78 * *, 042333 (2008). F. Xu, B. Qi, and H.-K. Lo, New J. Phys. * * 12 * *, 113026 (2010). L. Lydersen et al. , Nature Photonics * * 4 * *, 686 (2010). M. Ansmann et al. , Nature * * 461 * *, 504 (2009). M.A. Rowe et al. , Nature 409, 791 (2001). T. Monz et al. , Phys. Rev. Lett. * * 106 * *, 130506 (2011). J.
, T. Gidick, arXiv:1210.1810 \[quant-ph\]. J. Siuman et al., Phys. Rev. Levt. **106**, 220501 (2011). C.-E. Bardyn dt al., Phys. Rev. A **80**, 062327 (2009). M. McKague, T.Y. Yant, and V. Scarani, arXiv:1203.2976 [\[]{}duant-ph[\]]{}. J.-D. Bancal wt ao., Phys. Rev. Lett. 106, 250404 (2011). F. Magnisd et cl., in Proceedingx of the 33rg Internationan Zoploquium on Automata, Languages and [rogramkijg (Springer, 2006), p. 72. M. MbKwgue and M. Mosca, in Proceedings of the 5th Conherence on Theoty of Quantum Computation, Fommknication, and Cryphography (Sptjngqe, 2011), p. 113. R. Colbezk, PhD dissertation, Unjv. Cambridge (2007), arXiv:0911.3814 [\[]{}quant-ph[\]]{}; R. Culbecl and A. Kebt, J. Ovys. A 44, 095305 (2011). S. Pmronio et al., Naturc 464, 1021 (2010). S. Phronio snd S. Massar, avXiv:1111.6056 [\[]{}xuabt-ph[\]]{}. S. Fehr, R. Gelles, aid C. Schaffner, arXiv:1111.6052 [\[]{}quant-ph[\]]{}. U. Vczirani and T. Vidick, Phul. Trdns. S. Soz. A **370**, 3432 (2012). Y. Ziao et al., Phbs. Rev. A **78**, 042333 (2008). R. Xu, B. Qi, abd H.-K. Lo, New J. Phys. **12**, 113026 (2010). O. Lydersen et al., Naeuwe Photonics **4**, 686 (2010). M. Ansmann et al., Nature **461**, 504 (2009). M.A. Rowe et al., Nature 409, 791 (2001). T. Nonz et al., Phys. Rev. Lgtt. **106**, 130506 (2011). J.
, T. Vidick, arXiv:1210.1810 \[quant-ph\]. J. Silman Phys. Lett. **106**, (2011). C.-E. Bardyn **80**, (2009). M. McKague, Yang, and V. arXiv:1203.2976 [\[]{}quant-ph[\]]{}. J.-D. Bancal et al., Rev. Lett. 106, 250404 (2011). F. Magniez et al., in Proceedings of the International Colloquium on Automata, Languages and Programming (Springer, 2006), p. 72. M. McKague M. in of 5th Conference on Theory of Quantum Computation, Communication, and Cryptography (Springer, 2011), p. 113. R. Colbeck, dissertation, Univ. Cambridge (2007), arXiv:0911.3814 [\[]{}quant-ph[\]]{}; R. Colbeck A. Kent, J. Phys. 44, 095305 (2011). S. Pironio al., 464, 1021 S. and Massar, arXiv:1111.6056 [\[]{}quant-ph[\]]{}. Fehr, R. Gelles, and C. Schaffner, arXiv:1111.6052 [\[]{}quant-ph[\]]{}. U. Vazirani and T. Vidick, Phil. Trans. R. Soc. **370**, 3432 Zhao et Phys. A 042333 (2008). F. Qi, and H.-K. Lo, New J. (2010). L. Lydersen et al., Nature Photonics **4**, (2010). M. et al., Nature **461**, 504 (2009). Rowe et al., Nature 409, 791 (2001). T. et al., Phys. Rev. Lett. **106**, 130506 (2011). J.
, T. Vidick, arXiv:1210.1810 \[quant-ph\]. J. SilmaN et al., Phys. REv. LetT. **106**, 220501 (2011). C.-E. barDyN et aL., PhyS. Rev. A **80**, 062327 (2009). M. McKague, T.h. yang, And V. Scarani, arXiv:1203.2976 [\[]{}quant-pH[\]]{}. J.-D. BaNcAL et aL., phYs. Rev. lett. 106, 250404 (2011). F. MaGNiEZ Et aL., iN PRocEeDInGs of tHe 33rD InternAtional ColLoqUiUm on Automata, lAnGuages and PRogRamming (SprinGer, 2006), P. 72. M. McKaGuE anD m. MoscA, in proceEdings OF the 5th conferencE oN theory OF QuantuM cOmPutaTion, Communication, ANd cRyptography (SprInger, 2011), p. 113. r. COLbECK, Phd diSsertation, unIv. CamBRidge (2007), arxIv:0911.3814 [\[]{}QUANt-pH[\]]{}; r. Colbeck and A. KEnt, J. Phys. A 44, 095305 (2011). S. PIRonIo et al., naTurE 464, 1021 (2010). s. PironIo and s. MASsaR, arXiv:1111.6056 [\[]{}quant-Ph[\]]{}. S. FEhr, R. GelleS, and C. SCHaffner, ARXiv:1111.6052 [\[]{}quaNt-ph[\]]{}. U. VAziRanI and t. viDiCk, PHiL. traNS. R. soc. a **370**, 3432 (2012). y. ZhAo et al., PhYs. reV. A **78**, 042333 (2008). F. Xu, b. Qi, aND h.-k. lo, NeW J. PHys. **12**, 113026 (2010). L. lyderSen et al., Nature phoToniCS **4**, 686 (2010). M. ANsmanN et al., natuRe **461**, 504 (2009). m.A. RowE et al., NAture 409, 791 (2001). t. MOnz et al., Phys. Rev. LEtt. **106**, 130506 (2011). J.
, T. Vidick, arXiv:1210.18 10 \[quant -ph\] . J.Si lman etal., Phys. Rev . Let t. **106**, 220501 (20 11). C . -E.B ar dyn e t al.,P hy s . Re v. A ** 80 * *, 0623 27(2009). M. McKag ue, T .H. Yang, an d V . Scarani, ar Xiv:1203.297 6 [ \[]{}q ua nt- p h[\]] {}. J.- D. Ban c al etal., Phys .R ev. Le t t. 106, 2 50 404(2011). F. Magni e ze t al., in Proc eeding so ft h e 3 3rd Internati on al Co l loquium on A u tom a ta, Languages and Progra m min g (Spr in ger , 2006) , p.72 . M . McKague a nd M . Mosca,in Pro c eedings of the5th Co nfe ren ce o n T he ory o f Qu a nt umC omp utation, C om munic atio n , a nd C ryp togr aphy(Springer, 20 11) , p. 113 . R. Colb eck, P hD di sserta tion, U niv. Cambridge(200 7), arXiv :09 11 .38 14 [\[] { }quant -ph [\] ]{}; R. Colbec k an dA . Ke nt, J. Phys. A 44, 0 9 5 30 5 (2011) . S.P ir on i o et al. ,Nat ure4 6 4, 10 21 ( 2 01 0). S.Pironi o a nd S. Mas sa r, arX iv :11 11. 6056[ \[]{ }quant -ph[\]]{ }. S . Fehr, R. Gell e s, and C. Sch a ff n e r, arXi v:1 111.6052 [\ []{} q uant -ph[ \ ]] {}. U. V azira ni an d T. Vidick, Phil. T ra ns. R. Soc. A **370**, 3 432 (2012) . Y. Zhaoet a l ., Phys. Rev. A * *78** , 042333 ( 2 008). F . Xu, B. Qi,and H.-K. L o, New J . P hys . * *12 * * ,113026 (2010) . L.Ly dersenetal., Na tur e P hot oni cs **4**, 6 86 (2010 ). M. A nsm ann e t al., Na tu re** 461 **, 5 0 4 (200 9). M.A. R ow e et al., N a tu r e 409 ,79 1 (2 001 ). T.Monz etal., Ph ys. Rev.Let t . ** 10 6* *, 1305 06 (2011). J .
, T._Vidick, arXiv:1210.1810_\[quant-ph\]. J. Silman et al.,_Phys. Rev._Lett._**106**, 220501_(2011). C.-E._Bardyn et al.,_Phys. Rev. A_**80**, 062327 (2009). M. McKague,_T.H. Yang, and_V._Scarani, arXiv:1203.2976 [\[]{}quant-ph[\]]{}. J.-D. Bancal et al., Phys. Rev. Lett. 106, 250404 (2011). F. Magniez et_al.,_in Proceedings_of_the_33rd International Colloquium on Automata,_Languages and Programming (Springer, 2006),_p. 72. M._McKague and M. Mosca, in Proceedings of the_5th_Conference on Theory_of Quantum Computation, Communication, and Cryptography (Springer, 2011), p._113. R. Colbeck, PhD dissertation, Univ. Cambridge_(2007), arXiv:0911.3814 [\[]{}quant-ph[\]]{};_R._Colbeck_and A. Kent, J._Phys. A 44, 095305 (2011). S. Pironio_et al., Nature 464, 1021 (2010). S._Pironio and S. Massar, arXiv:1111.6056 [\[]{}quant-ph[\]]{}. S. Fehr,_R. Gelles, and C. Schaffner, arXiv:1111.6052_[\[]{}quant-ph[\]]{}. U. Vazirani and T. Vidick,_Phil. Trans._R. Soc. A **370**, 3432_(2012). Y. Zhao et_al., Phys._Rev. A **78**,_042333 (2008). F. Xu, B. Qi, and_H.-K. Lo, New_J. Phys. **12**, 113026 (2010). L. Lydersen_et_al., Nature Photonics_**4**,_686_(2010). M. Ansmann_et al., Nature_**461**,_504 (2009). M.A._Rowe_et al., Nature 409, 791 (2001). T._Monz_et al., Phys. Rev. Lett. **106**, 130506_(2011). J.
d_2$ least significant bits of the binary representation of $x$. We thus decompose the raw keys as $X=({\hat{X}},{\check{X}})$, where ${\hat{X}}$ and ${\check{X}}$ denote the sequence of the $d_2$ most and the $d_1$ least significant bits of each key symbol, respectively. The reconciliation module performs the following steps: (iii-a) Based on the variance of her binned raw key and the samples $X_A^\text{pe}$ and $X_B^\text{pe}$, Alice determines $d_1$, $d_2$, and the code rate $R$ such that the expected leakage is minimized w.r.t. the entropy in Bob’s symbols, and transmits these parameters to Bob. (iii-b) Then Alice communicates ${\check{X}}_A$ to Bob who reconciles ${\check{X}}_B$ simply by setting $ {\check{X}}_B:={\check{X}}_A $. Hence, the errors which are left in Bob’s key $X_B$ are reduced to the errors in ${\hat{X}}_B$. Non-binary LDPC reconciliation is used to correct ${\hat{X}}_B$ as described in the next step. (iii-c) Both Alice and Bob split their ${\hat{X}}_A$ and ${\hat{X}}_B$ into blocks ${\hat{X}}_A^{(\ell)}$ and ${\hat{X}}_B^{(\ell)}$, $\ell=1,\dots, \frac{N-k}{n}$, each with $n=10^5$ elements of ${\hat{\chi}}$. For this step we identify $ {\hat{\chi}}$ with GF($2^{d_2}$), the Galois field with $2^{d_2}$ elements. For each block ${\hat{X}}_A^{(\ell)}$, Alice uses the parity check matrix $H$ of an LDPC code over GF($2^{d_2}$) and rate $R$ to calculate the syndrome $s^{(\ell)}:=H\cdot {\hat{X}}_A^{(\ell)}$. Alice sends the syndrome $s^{(\ell)}$ to Bob. For all elements $j \in GF(2^{d_2})$ and for all indices $i\in\{1,\dots,n\}$ in the block Bob calculates the conditional probability that $({\hat{X}}_A^{(\ell)})_i=j
d_2 $ least significant bits of the binary representation of $ x$. We thus disintegrate the crude keys as $ X=({\hat{X}},{\check{X}})$, where $ { \hat{X}}$ and $ { \check{X}}$ denote the succession of the $ d_2 $ most and the $ d_1 $ least significant snatch of each key symbol, respectively. The reconciliation module perform the following steps: (iii - a) Based on the discrepancy of her binned raw keystone and the samples $ X_A^\text{pe}$ and $ X_B^\text{pe}$, Alice determines $ d_1 $, $ d_2 $, and the code pace $ R$ such that the expected leakage is minimized w.r.t.   the information in Bob ’s symbols, and transmits these parameter to Bob. (iii - b) Then Alice communicates $ { \check{X}}_A$ to Bob who reconciles $ { \check{X}}_B$ plainly by setting $ { \check{X}}_B:={\check{X}}_A $. Hence, the errors which are leave in Bob ’s key $ X_B$ are reduced to the errors in $ { \hat{X}}_B$. Non - binary LDPC reconciliation is used to correct $ { \hat{X}}_B$ as trace in the next step. (iii - c) Both Alice and Bob split their $ { \hat{X}}_A$ and $ { \hat{X}}_B$ into blocks $ { \hat{X}}_A^{(\ell)}$ and $ { \hat{X}}_B^{(\ell)}$, $ \ell=1,\dots, \frac{N - k}{n}$, each with $ n=10 ^ 5 $ elements of $ { \hat{\chi}}$. For this step we identify $ { \hat{\chi}}$ with GF($2^{d_2}$), the Galois field with $ 2^{d_2}$ elements. For each block $ { \hat{X}}_A^{(\ell)}$, Alice uses the parity check matrix $ H$ of an LDPC code over GF($2^{d_2}$) and rate $ R$ to calculate the syndrome $ s^{(\ell)}:=H\cdot { \hat{X}}_A^{(\ell)}$. Alice sends the syndrome $ s^{(\ell)}$ to Bob. For all elements $ joule \in GF(2^{d_2})$ and for all index $ i\in\{1,\dots, n\}$ in the block Bob account the conditional probability that $ ({ \hat{X}}_A^{(\ell)})_i = j
d_2$ lfast significant bits of the binary reptewentatmon of $s$. We thur decompose the raw keys as $E=({\hat{Z}},{\checj{X}})$, where ${\hat{X}}$ and ${\checy{X}}$ denote the seqyenct of the $d_2$ most ais the $d_1$ least dignnfmcant bits of esch key sykbol, respectivalh. Che reconciliation module performs tre follpwlng steps: (iii-a) Fasec on fhe variance of her binned raw key and tht samples $X_A^\text{pe}$ and $X_B^\text{pe}$, Alice determlnes $d_1$, $d_2$, and the code gate $R$ such thae the expectea leakage pv minimizes w.r.t. the entropy in Bob’s symbolr, and transmits tyesf parameters to Bjb. (iii-b) Then Alice cokmunicayes ${\check{X}}_A$ to Bou whi reconciles ${\check{X}}_B$ vimply by setting $ {\check{X}}_B:={\cveek{X}}_A $. Hence, the errorw qhich are lefg in Bog’s ksy $X_B$ wre reduced tk the errorw in ${\hat{X}}_B$. Non-binaru JEPC reconciliztion ys used to correct ${\hat{X}}_B$ as described in uhe nsxt step. (iii-c) Both Alice and Bob split their ${\jat{X}}_A$ and ${\hat{X}}_B$ into blocks ${\hat{X}}_A^{(\ell)}$ and ${\hat{X}}_B^{(\ell)}$, $\ell=1,\dots, \xrac{N-i}{v}$, ecgm wigy $j=10^5$ elements of ${\hat{\chi}}$. For this step we identifr $ {\nan{\chi}}$ with GF($2^{d_2}$), the Galois fiekd wojh $2^{d_2}$ elements. Wor eaeg glock ${\hat{X}}_A^{(\ell)}$, Alife uses the parity chtck mstrix $H$ of an LDPC code ovee GF($2^{d_2}$) and rane $R$ to calculate the dyndrome $s^{(\eul)}:=H\ccot {\hst{X}}_A^{(\ell)}$. Alice sends the synsrome $s^{(\ell)}$ ho Bob. Fod all elements $j \kn NF(2^{d_2})$ and for all indices $i\in\{1,\does,n\}$ in thx bloek Bob cxlcukates ehe conditlonal probability that $({\hwt{X}}_A^{(\epl)})_h=j
d_2$ least significant bits of the binary $x$. thus decompose raw keys as denote sequence of the most and the least significant bits of each key respectively. The reconciliation module performs the following steps: (iii-a) Based on the variance her binned raw key and the samples $X_A^\text{pe}$ and $X_B^\text{pe}$, Alice determines $d_1$, and code $R$ that the expected leakage is minimized w.r.t. the entropy in Bob’s symbols, and transmits these parameters Bob. (iii-b) Then Alice communicates ${\check{X}}_A$ to Bob reconciles ${\check{X}}_B$ simply by $ {\check{X}}_B:={\check{X}}_A $. Hence, the which left in key are to the errors ${\hat{X}}_B$. Non-binary LDPC reconciliation is used to correct ${\hat{X}}_B$ as described in the next step. (iii-c) Both and Bob ${\hat{X}}_A$ and into ${\hat{X}}_A^{(\ell)}$ ${\hat{X}}_B^{(\ell)}$, $\ell=1,\dots, \frac{N-k}{n}$, $n=10^5$ elements of ${\hat{\chi}}$. For this $ {\hat{\chi}}$ with GF($2^{d_2}$), the Galois field with elements. For block ${\hat{X}}_A^{(\ell)}$, Alice uses the parity matrix $H$ of an LDPC code over GF($2^{d_2}$) rate $R$ to calculate the syndrome $s^{(\ell)}:=H\cdot {\hat{X}}_A^{(\ell)}$. Alice sends the syndrome $s^{(\ell)}$ to Bob. elements $j \in GF(2^{d_2})$ for all indices in block calculates conditional probability $({\hat{X}}_A^{(\ell)})_i=j
d_2$ least significant bits of thE binary repResenTatIon Of $X$. We tHus dEcompose the raw KEys aS $X=({\hat{X}},{\check{X}})$, where ${\hat{X}}$ aNd ${\cheCk{x}}$ DenoTE tHe seqUence of THe $D_2$ MOst AnD tHe $d_1$ LeASt SigniFicAnt bits Of each key sYmbOl, Respectively. tHe ReconciliaTioN module perfoRms The folLoWinG Steps: (Iii-A) BaseD on the VAriancE of her binNeD Raw key ANd the saMPLeS $X_A^\tExt{pe}$ and $X_B^\text{pe}$, ALIcE Determines $d_1$, $d_2$, anD the coDe RAtE $r$ SucH thAt the expecTeD leakAGe is minIMiZED W.r.t. THe entropy in BoB’s symbols, anD TraNsmits ThEse PArametErs to boB. (Iii-B) Then Alice cOmmuNicates ${\chEck{X}}_A$ tO bob who rEConcileS ${\check{x}}_B$ sImpLy by SEtTiNg $ {\cHeCK{X}}_B:={\CHeCk{X}}_a $. henCe, the errOrS wHich aRe leFT IN bob’s Key $x_B$ arE reduCed to the errorS in ${\Hat{X}}_b$. non-BinarY LDPC RecoNcIliatIon is uSed to CoRrect ${\hat{X}}_B$ as desCribEd in the neXt sTeP. (iiI-c) both ALIce and bob SplIt their ${\Hat{X}}_A$ anD ${\Hat{x}}_B$ INTO bLocks ${\hat{X}}_A^{(\ell)}$ and ${\haT{X}}_b^{(\ELl)}$, $\Ell=1,\dots, \fRac{N-k}{n}$, EAcH wITh $n=10^5$ elemeNtS of ${\Hat{\cHI}}$. for thIs stEP wE identifY $ {\hat{\chI}}$ WiTh gF($2^{d_2}$), the GAlOis fieLd WitH $2^{d_2}$ eLemenTS. For Each blOck ${\hat{X}}_A^{(\Ell)}$, AlICe uses the paritY Check matrix $H$ oF An ldpC COde oVer gF($2^{d_2}$) and rate $R$ To caLCulaTe thE SyNdrOMe $s^{(\elL)}:=H\cdoT {\hAT{X}}_a^{(\Ell)}$. Alice sends the synDrOme $s^{(\elL)}$ to BoB. For all elemenTs $j \in GF(2^{d_2})$ anD FOR all indiCes $i\IN\{1,\dOTs,n\}$ in the block BOb calCulates the COnditionAl proBability That $({\hat{X}}_A^{(\ELL)})_i=j
d_2$ least significant bit s of the b inary re pre se ntat ionof $x$. We thu s dec ompose the raw keys as $X=( {\ h at{X } }, {\che ck{X}}) $ ,w h ere $ {\ hat {X } }$ and${\ check{X }}$ denote th esequence oft he $d_2$ mos t a nd the $d_1$ le ast si gn ifi c ant b its of e ach ke y symbo l, respec ti v ely. T h e recon c i li atio n module performs th e following ste ps: ( ii i -a ) Bas edon the var ia nce o f her bi n ne d r awk ey and the sa mples $X_A^ \ tex t{pe}$ a nd$ X_B^\t ext{p e} $ , A lice determ ines $d_1$, $ d_2$,a nd thec ode rat e $R$suc h t hatt he e xpe ct e d l e ak age isminimize dw. r.t.thee n t r opyinBob’ s sym bols, and tra nsm itst hes e par amete rs t oBob. (iii- b) Th en Alice communic ates ${\check {X} }_ A$to Bobw ho rec onc ile s ${\ch eck{X}} _ B$si m p l yby setting $ {\che ck { X }} _B:={\ch eck{X} } _A $ . Hence,th e e rror s which are le ft in Bo b’s ke y $ X_ B$ arere ducedto th e e rrors in $ {\hat{ X}}_B$.Non-b i nary LDPC reco n ciliation isu se d to corr ect ${\hat{X}} _B$a s de scri b ed in the n ext s te p .(iii-c) Both Alicean d Bobsplit their ${\hat {X}}_A$ an d $ {\hat{X} }_B$ in t o blocks ${\ha t{X}} _A^{(\ell) } $ and ${ \hat{ X}}_B^{( \ell)}$,$ \ ell=1,\d ots , \ fra c{N - k }{ n}$, each wit h $n=1 0^ 5$ elem ent s of ${ \ha t{\ chi }}$ .For thisstep weid en ti fy ${\hat { \chi}}$wi thGF ($2 ^{d_2 } $), th e Gal oisfi el d wi th $2^{ d _2 } $ ele me nt s. F orea ch bl ock$ {\h at{X}}_ A^{(\ell) }$, Alic eus es theparity checkma trix $H$ o fanLDPC c o d e over G F($2^{d_2}$) and rate $ R $ to ca lcu latethesyndrome$s^ {(\ell )}: = H\cdot {\hat {X}}_ A^ {(\ e l l)}$. A li cese nds the sy n d rom e $s^ {( \ell )}$ toBob. For all eleme n ts$j \in GF(2^{ d_2 })$a n dfor al l in di c es$ i \in\{1,\dots,n\ }$ in thebl o ck Bob calcu l ate sthe con ditiona l pro b ability that $({ \hat{X}}_ A^ {(\e l l )}) _i=j
d_2$ least_significant bits_of the binary representation_of $x$._We_thus decompose_the_raw keys as_$X=({\hat{X}},{\check{X}})$, where ${\hat{X}}$_and ${\check{X}}$ denote the_sequence of the_$d_2$_most and the $d_1$ least significant bits of each key symbol, respectively. The reconciliation_module_performs the_following_steps: (iii-a)_Based on the variance of_her binned raw key and_the samples_$X_A^\text{pe}$ and $X_B^\text{pe}$, Alice determines $d_1$, $d_2$, and_the_code rate $R$_such that the expected leakage is minimized w.r.t. the entropy_in Bob’s symbols, and transmits these_parameters to Bob. (iii-b)_Then_Alice_communicates ${\check{X}}_A$ to Bob_who reconciles ${\check{X}}_B$ simply by setting_$ {\check{X}}_B:={\check{X}}_A $. Hence, the errors_which are left in Bob’s key $X_B$_are reduced to the errors in_${\hat{X}}_B$. Non-binary LDPC reconciliation is_used to_correct ${\hat{X}}_B$ as described in_the next step. (iii-c)_Both Alice_and Bob split_their ${\hat{X}}_A$ and ${\hat{X}}_B$ into blocks_${\hat{X}}_A^{(\ell)}$ and ${\hat{X}}_B^{(\ell)}$,_$\ell=1,\dots, \frac{N-k}{n}$, each with $n=10^5$ elements_of_${\hat{\chi}}$. For this_step_we_identify $_{\hat{\chi}}$ with GF($2^{d_2}$),_the_Galois field_with_$2^{d_2}$ elements. For each block ${\hat{X}}_A^{(\ell)}$,_Alice_uses the parity check matrix $H$ of_an LDPC code over_GF($2^{d_2}$)_and rate $R$ to_calculate the syndrome $s^{(\ell)}:=H\cdot {\hat{X}}_A^{(\ell)}$._Alice sends the syndrome $s^{(\ell)}$ to_Bob. For_all elements_$j \in GF(2^{d_2})$ and for all indices $i\in\{1,\dots,n\}$ in the block_Bob calculates the conditional probability that_$({\hat{X}}_A^{(\ell)})_i=j
(u, \omega) = A(u, -\omega)^*$, where $A^*$ is the complex conjugate of $A$. Formally, we consider $\mathbb{R}^N$–valued time series of length $T$, $\left\{ \mb X_t : t =1, \dots, T \right\}$, of the form $$\label{eq:X} \mb X_t = \int_{-1/2}^{1/2} A(t/T, \omega) \exp{(2 \pi i \omega t)} d \mb Z(\omega),$$ for a time-varying transfer function $A(u, \omega)$ and an $N$–dimensional mean-zero orthogonal process $\mb Z(\omega)$ that is Hermitian and where $E\left\{ d \mb Z(\omega) d \mb Z^*(\zeta) \right \}$ is the identity matrix if $\omega = \zeta$ and zero otherwise. The focus of this article is on estimating the time-varying spectrum $$f(u, \omega) = A(u, \omega)A(u, \omega)^*, \quad u \in [0,1], \omega \in \mathbb{R}.$$ The time-varying spectrum $f(u, \omega)$ is a positive definite Hermitian $N \times N$ matrix. We assume it satisfies some regularity conditions both as a function of $\omega$ and as a function of $u$. For every $u$, we assume that each component of $f(u, \cdot)$ possesses a square-integrable first derivative as a function of frequency. For every $\omega$, we assume that each component of $f(\cdot, \omega)$ is continuous as a function of scaled time at all but a possible finite number of points. We refer to this class of time series as locally stationary and note that it differs somewhat from the locally stationary time series models considered by [@dahlhaus2000] and @guo2006. First, [@dahlhaus2000] assumed a series of transfer functions $A^0_{t,T}(\omega)$ that converge to a large sample transfer function $A(u, \omega)$ in order to allow for the fitting of parametric models. Since we are considering nonparametric estimation, in a manner similar to @guo2006, we define our model directly using $A(u, \omega)$. Second, the models of [@dahl
( u, \omega) = A(u, -\omega)^*$, where $ A^*$ is the complex conjugate of $ A$. Formally, we consider $ \mathbb{R}^N$–valued time serial of duration $ T$, $ \left\ { \mb X_t: t = 1, \dots, T \right\}$, of the form $ $ \label{eq :X } \mb X_t = \int_{-1/2}^{1/2 } A(t / T, \omega) \exp{(2 \pi i \omega t) } d \mb Z(\omega),$$ for a time - vary transportation function $ A(u, \omega)$ and an $ N$–dimensional mean - zero orthogonal process $ \mb Z(\omega)$ that is Hermitian and where $ E\left\ { d \mb Z(\omega) d \mb Z^*(\zeta) \right \}$ is the identity matrix if $ \omega = \zeta$ and zero differently. The stress of this article is on estimating the time - varying spectrum $ $ f(u, \omega) = A(u, \omega)A(u, \omega)^ *, \quad u \in [ 0,1 ], \omega \in \mathbb{R}.$$ The time - vary spectrum $ f(u, \omega)$ is a positive definite Hermitian $ N \times N$ matrix. We assume it satisfy some regularity condition both as a function of $ \omega$ and as a function of $ u$. For every $ u$, we assume that each component of $ f(u, \cdot)$ possesses a straight - integrable first derivative as a function of frequency. For every $ \omega$, we wear that each component of $ f(\cdot, \omega)$ is continuous as a function of scaled clock time at all but a possible finite number of points. We refer to this course of time series as locally stationary and note that it differs somewhat from the locally stationary time serial model considered by [ @dahlhaus2000 ] and @guo2006. First, [ @dahlhaus2000 ] assumed a series of transfer functions $ A^0_{t, T}(\omega)$ that converge to a large sample transfer affair $ A(u, \omega)$ in order to allow for the appointment of parametric model. Since we are considering nonparametric estimation, in a manner similar to @guo2006, we specify our model directly using $ A(u, \omega)$. Second, the models of [ @dahl
(u, \olega) = A(u, -\omega)^*$, where $A^*$ ir the complex conjugatx of $A$. Rormally, we consider $\mathbb{R}^N$–valued vime seritf of length $T$, $\left\{ \mb X_t : t =1, \dots, T \rigit\}$, of the form $$\legel{eq:X} \mn X_t = \lnt_{-1/2}^{1/2} A(c/T, \omega) \exp{(2 \pi l \omega t)} g \mb Z(\omega),$$ fos x cime-varying transfer function $A(u, \omeda)$ and sn $N$–dimensional iean-eerj orfhogonal process $\mb Z(\omega)$ that is Hermitpan and where $E\legt\{ d \mb Z(\omega) d \mb Z^*(\zeta) \rigjt \}$ is the identitj matrix if $\omedq = \zeta$ and xero otherwise. The focua of this article is on estimatkng tke time-varyunt sogctrum $$f(u, \omxga) = A(l, \omega)A(u, \omens)^*, \quad u \in [0,1], \pmega \in \mathbn{R}.$$ Thx tine-varying spectrum $f(u, \omega)$ is a positide definide Hermitian $N \timew B$ mattix. Wa asryme it setiafies domx regularitg conditionw both as a functiom jd $\omega$ and aa a fugceion of $u$. For every $u$, we assume that eabh ckmponent of $f(u, \cdot)$ poswesses a square-integrwble firse derivative as a function of frequency. For every $\omege$, de cwsume rhwt each component of $f(\cdot, \omega)$ is continuouf ax s function of fcaled time ah sjl but a posskble fnhife number of pointd. We refgr to rhis clasf of time series as locally starionary and uotw that it differs domewhat frum tne lovally stationary time szries jodels consldered by [@aahlhaus2000] and @guo2006. Figst, [@gahlhaus2000] assumed a series jf transfxr fuuctions $X^0_{t,T}(\okega)$ trat converhe to a large sample trajsfer fgnction $A(u, \omega)$ in order to allow for thx fitting of kardmenric modejs. Sikce we are consydering nonparcmetric zstimagion, in a janner vimilar to @duo2006, we define lur model dicectly usyng $Q(u, \onega)$. Sezund, the models of [@dahl
(u, \omega) = A(u, -\omega)^*$, where $A^*$ complex of $A$. we consider $\mathbb{R}^N$–valued $\left\{ X_t : t \dots, T \right\}$, the form $$\label{eq:X} \mb X_t = A(t/T, \omega) \exp{(2 \pi i \omega t)} d \mb Z(\omega),$$ for a time-varying function $A(u, \omega)$ and an $N$–dimensional mean-zero orthogonal process $\mb Z(\omega)$ that is and $E\left\{ \mb d \mb Z^*(\zeta) \right \}$ is the identity matrix if $\omega = \zeta$ and zero otherwise. focus of this article is on estimating the spectrum $$f(u, \omega) = \omega)A(u, \omega)^*, \quad u \in \omega \mathbb{R}.$$ The spectrum \omega)$ a positive definite $N \times N$ matrix. We assume it satisfies some regularity conditions both as a function of $\omega$ as a $u$. For $u$, assume each component of possesses a square-integrable first derivative as frequency. For every $\omega$, we assume that each of $f(\cdot, is continuous as a function of time at all but a possible finite number points. We refer to this class of time series as locally stationary and note that somewhat from the locally time series models by and First, assumed a of transfer functions $A^0_{t,T}(\omega)$ that converge to a large sample transfer $A(u, \omega)$ in order to allow for the fitting of Since are considering nonparametric in a manner similar @guo2006, define our model directly \omega)$. the
(u, \omega) = A(u, -\omega)^*$, where $A^*$ is the cOmplex conjUgate Of $A$. forMaLly, wE conSider $\mathbb{R}^N$–vALued Time series of length $T$, $\lefT\{ \mb X_t : T =1, \dOTs, T \rIGhT\}$, of thE form $$\laBEl{EQ:x} \mb x_t = \InT_{-1/2}^{1/2} A(t/t, \oMEgA) \exp{(2 \pI i \oMega t)} d \mB Z(\omega),$$ for A tiMe-Varying transFEr Function $A(u, \OmeGa)$ and an $N$–dimeNsiOnal meAn-ZerO OrthoGonAl proCess $\mb z(\Omega)$ tHat is HermItIAn and wHEre $E\lefT\{ D \Mb z(\omeGa) d \mb Z^*(\zeta) \right \}$ is THe IDentity matrix iF $\omega = \ZeTA$ aND ZerO otHerwise. The FoCus of THis artiCLe IS ON esTImating the timE-varying speCTruM $$f(u, \omeGa) = a(u, \oMEga)A(u, \oMega)^*, \qUaD U \in [0,1], \Omega \in \mathBb{R}.$$ THe time-varYing spECtrum $f(u, \OMega)$ is a PositiVe dEfiNite hErMiTiaN $N \TImeS n$ mAtrIX. We Assume it SaTiSfies Some REGULariTy cOndiTions Both as a functiOn oF $\omeGA$ anD as a fUnctiOn of $U$. FOr eveRy $u$, we aSsume ThAt each component Of $f(u, \Cdot)$ posseSseS a SquArE-inteGRable fIrsT deRivativE as a funCTioN oF FREqUency. For every $\omega$, We ASSuMe that eaCh compONeNt OF $f(\cdot, \omEgA)$ is ContINUous aS a fuNCtIon of scaLed timE At AlL but a poSsIble fiNiTe nUmbEr of pOInts. we refeR to this cLass oF Time series as loCAlly stationarY AnD NOtE That It dIffers somewHat fROm thE locALlY stATionaRy timE sERiES models considered by [@DaHlhaus2000] And @guO2006. First, [@dahlhauS2000] assumed a sERIEs of tranSfer FUnCTions $A^0_{t,T}(\omega)$ tHat coNverge to a lARge samplE tranSfer funcTion $A(u, \omeGA)$ In order tO alLow For The FITtIng of parametrIC ModeLs. since we Are ConsideRinG noNpaRamEtRic estimaTion, in a mAnNeR sImIlaR to @guO2006, We define OuR moDeL diRectlY Using $A(U, \omegA)$. SecOnD, tHE moDels of [@dAHl
(u, \omega) = A(u, -\omega )^*$, wher e $A^ *$isth e co mple x conjugate of $A$. Formally, we consider $\ma th b b{R} ^ N$ –valu ed time se r i esof l eng th $T $, $\ lef t\{ \mb X_t : t = 1,\d ots, T \righ t \} $, of thefor m $$\label{e q:X } \mbX_ t = \int_ {-1 /2}^{ 1/2} A ( t/T, \ omega) \e xp { (2 \pi i \omeg a t) } d \mb Z(\omega),$$ fo r a time-varyin g tran sf e rf u nct ion $A(u, \om eg a)$ a n d an $N $ –d i m e nsi o nal mean-zero orthogonal pro cess $ \m b Z ( \omega )$ th at isHermitian a nd w here $E\l eft\{d \mb Z ( \omega) d \mb Z^ *(\ zeta ) \ ri ght \ } $ i s t hei den tity mat ri xif $\ omeg a = \zet a$andzerootherwise. T hefocu s of this arti cleis on e stimat ing t he time-varying s pect rum $$f(u , \ om ega )= A(u , \omeg a)A (u, \omega )^*, \q u adu\ i n [ 0,1], \omega \in \ ma t h bb {R}.$$ T he tim e -v ar y ing spec tr um$f(u , \omeg a)$i sa positi ve def i ni te Hermit ia n $N \ ti mes N$ matr i x. W e assu me it sa tisfi e s some regular i ty conditions bo t h a s a f unc tion of $\o mega $ and asa f unc t ion o f $u$ .F or every $u$, we assum ethat e ach c omponent of $ f(u, \cdot ) $ possesse s as qu a re-integrablefirst derivativ e as a fu nctio n of fre quency. F o r every $ \om ega $,wea s su me that eachc o mpon en t of $f (\c dot, \o meg a)$ is co nt inuous as a funct io nof s cal ed ti m e at all b utapos sible finite numb er o fpo i nts . We r e fe r to t hi sclas s o ftimeseri e s a s local ly statio nar y and n ot e thatit differs so me what fromth e l ocally s tationar y time series models co n sidered by [@da hlha us2000] a nd@guo20 06. First, [@dah lhaus 20 00] a ssume d aser ie s of trans f e r f uncti on s $A ^0_{t,T }(\omega)$ that co n ver ge to a large sa mple t ra nsf e rf unc ti o n $ A ( u, \omega)$ inorder to a ll o wfor the fi t tin gof para metricmodel s . Since we are c onsiderin gnonp a r ame tric estim ation, i n a manne r simi l ar to @ guo 2006,we de fineour mo d eldirec tly us in g $A(u , \om eg a)$. Sec ond, the models of [@da hl
(u, \omega)_= A(u,_-\omega)^*$, where $A^*$ is_the complex_conjugate_of $A$._Formally,_we consider $\mathbb{R}^N$–valued_time series of_length $T$, $\left\{ \mb_X_t : t_=1,_\dots, T \right\}$, of the form $$\label{eq:X} \mb X_t = \int_{-1/2}^{1/2} A(t/T, \omega) \exp{(2 \pi_i_\omega t)}__d_\mb Z(\omega),$$ for a time-varying_transfer function $A(u, \omega)$ and_an $N$–dimensional_mean-zero orthogonal process $\mb Z(\omega)$ that is Hermitian_and_where $E\left\{ _d \mb Z(\omega) d \mb Z^*(\zeta) \right \}$ is_the identity matrix if $\omega =_\zeta$ and zero_otherwise. The_focus_of this article is_on estimating the time-varying spectrum $$f(u,_\omega) = A(u, \omega)A(u, \omega)^*, \quad_u \in [0,1], \omega \in \mathbb{R}.$$ The_time-varying spectrum $f(u, \omega)$ is a_positive definite Hermitian $N \times_N$ matrix._We assume it satisfies some_regularity conditions both_as a_function of $\omega$_and as a function of $u$._For every $u$,_we assume that each component of_$f(u,_\cdot)$ possesses a_square-integrable_first_derivative as_a function of_frequency._For every_$\omega$,_we assume that each component of_$f(\cdot,_\omega)$ is continuous as a function of_scaled time at all_but_a possible finite number_of points. We refer to this_class of time series as locally_stationary and_note that_it differs somewhat from the locally stationary time series models considered_by [@dahlhaus2000] and @guo2006. First, [@dahlhaus2000]_assumed a series of_transfer functions_$A^0_{t,T}(\omega)$_that converge to_a_large sample_transfer function $A(u, \omega)$ in order to_allow for_the fitting of parametric models. Since_we are considering nonparametric_estimation,_in a manner similar to @guo2006,_we define our model directly using_$A(u, \omega)$. Second, the models_of_[@dahl
-varying covariate (e.g., presence of emphysema) that is a predictor of both future exposure and of failure. In 1982, the standard analytic approach was to model the conditional probability (i.e., the hazard) of failure time $t$ as a function of past exposure history using a time-dependent Cox proportional hazards model. Robins formally showed that, even when confounding by unmeasured factors and model specification are absent, this approach may result in estimates of effect that may fail to have a causal interpretation, regardless of whether or not one also adjusts for the measured time-dependent confounders in the analysis. In fact, if previous exposure also predicts the subsequent evolution of the time-dependent confounders (e.g., since smoking is a cause of emphysema, it predicts this disease) then the standard approach can find an artifactual exposure effect even under the sharp null hypothesis of no net, direct or indirect effect of exposure on the failure time of any subject. Prior to @robins:1986 ([-@robins:1986]), although informal discussions of net, direct and indirect (i.e., mediated) effects of time varying exposures were to be found in the discussion sections of most epidemiologic papers, no formal mathematical definitions existed. To address this, @robins:1986 ([-@robins:1986]) introduced a new counterfactual model, the *finest fully randomized causally interpreted structured tree graph* (FFRCISTG)[^4] model that extended the point treatment counterfactual model of @neyman:sur:1923 ([-@neyman:sur:1923]) and @rubin:estimating:1974 ([-@rubin:estimating:1974; -@Rubi:baye:1978])[^5] to longitudinal studies with time-varying treatments, direct and indirect effects and feedback of one cause on another. Due to his lack of formal statistical training, the notation and formalisms in @robins:1986 ([-@robins:1986]) differ from those found in the mainstream literature; as a consequence the paper can be a difficult read.[^6] @richardson:robins:2013 ([-@richardson:robins:2013], Appendix C) present the FFRCISTG model using a more familiar notation.[^7] ![Causal tree graph depicting a simple scenario with treatments at two times $A_1$, $
-varying covariate (e.g.,   presence of emphysema) that is a predictor of both future exposure and of bankruptcy. In 1982, the standard analytic overture was to model the conditional probability (i.e., the hazard) of failure meter $ t$ as a routine of past exposure history using a prison term - subject Cox proportional hazards exemplar. Robins formally showed that, even when confounding by immeasurable factors and model specification are absent, this overture may result in estimates of effect that may fail to have a causal interpretation, regardless of whether or not one besides adjusts for the measured time - subject confounders in the analysis. In fact, if previous vulnerability also predict the subsequent evolution of the time - dependent confounders (e.g.,   since smoking is a lawsuit of emphysema, it predicts this disease) then the standard approach can find an artifactual exposure effect even under the sharp null hypothesis of no net, direct or indirect effect of exposure on the failure time of any subject. Prior to @robins:1986 ([ -@robins:1986 ]), although informal discussions of internet, direct and indirect (i.e.,   mediate) effect of time varying exposures were to be determine in the discussion sections of most epidemiologic papers, no formal mathematical definition existed. To address this, @robins:1986 ([ -@robins:1986 ]) introduced a new counterfactual model, the * finest in full randomized causally interpreted structured tree graph * (FFRCISTG)[^4 ] model that extend the point discussion counterfactual exemplar of @neyman: sur:1923 ([ -@neyman: sur:1923 ]) and @rubin: estimating:1974 ([ -@rubin: estimating:1974; -@Rubi: baye:1978])[^5 ] to longitudinal studies with time - varying treatments, lineal and indirect effects and feedback of one cause on another. Due to his lack of formal statistical training, the notation and formalisms in @robins:1986 ([ -@robins:1986 ]) differ from those found in the mainstream literature; as a consequence the newspaper can be a difficult read.[^6 ] @richardson: robins:2013 ([ -@richardson: robins:2013 ], Appendix C) portray the FFRCISTG model using a more conversant notation.[^7 ] ! [ Causal tree graph depicting a simple scenario with treatments at two times $ A_1 $, $
-varjing covariate (e.g., presenct of emphysema) thcr is a predidtor of coth future exposure and of hailyre. Ib 1982, the standard analytkc approabh was to modtl the conditional probabljity (l.e., thz iazard) of failute time $t$ as a function of pxsc exposure history using a time-depenqent Coc oroportional hwzarcf mosvl. Robins formally showed that, sven whtn confounding by inmeasured factors and modfl soecification are ahsent, this qpprjqch may resuut in estimates of effgct that may fail to have a causxl incerpretatiob, eegwtdless of whxther jr not one also adjuvts for the measured bime-dxpeneent confounders in tie analysis. In fact, yf previogs exposure also prwducts jhe sgbsedyeng ebokufion ov tie time-depehdent confoynders (e.g., since smokond is a cause or emphrsqma, it predicts this disease) then the suandadd approach can find an artifactual exposure effect eden under the sharp null hypothesis of no net, diract oc kndnvcct ddffct of exposure on the failure time of any sufnevt. Irior to @robins:1986 ([-@rjbins:1986]), althoigj ogformal discursions of net, direct and infirect (y.e., meduated) efftcts pf time varying exposures wwre to be folnd un the discussion dections of mosj epidrmiologic papers, no foroal jathematicap definitjuns existed. To aadrvss dhis, @robins:1986 ([-@robins:1986]) introdused a new couuterfactjal kodel, ehe *finest fulln randomized causalpy injerpreded structkred tree graph* (FFRCISTG)[^4] model viat extended jhe popnt treatient gounterfactual iodel of @neymau:sur:1923 ([-@neyian:suf:1923]) and @rubih:estimaving:1974 ([-@rubin:eseimating:1974; -@Rubi:twye:1978])[^5] to longivudinal seudiws wuth timd-xarying treatmrnts, direbt and indirect effects end wsedback of one ecuwe on another. Die go rid kasn of formal vtatkstkval tfaining, the votayion and formalisms hn @rkbins:1986 ([-@robins:1986]) diffet nrom thosg found ig the mainstrram literature; as w conveqnence yhe paper can be a difficult read.[^6] @richardsln:rjbins:2013 ([-@richarqson:vobigs:2013], Appendir C) present the FFRCISTG model using a mire familiar notajion.[^7] ![Causal tree graph cepicting a simpje scenarho with treatments ar two times $A_1$, $
-varying covariate (e.g., presence of emphysema) that predictor both future and of failure. approach to model the probability (i.e., the of failure time $t$ as a of past exposure history using a time-dependent Cox proportional hazards model. Robins formally that, even when confounding by unmeasured factors and model specification are absent, this may in of that may fail to have a causal interpretation, regardless of whether or not one also adjusts the measured time-dependent confounders in the analysis. In if previous exposure also the subsequent evolution of the confounders since smoking a of it predicts this then the standard approach can find an artifactual exposure effect even under the sharp null hypothesis of net, direct effect of on failure of any subject. @robins:1986 ([-@robins:1986]), although informal discussions of indirect (i.e., mediated) effects of time varying exposures to be in the discussion sections of most papers, no formal mathematical definitions existed. To address @robins:1986 ([-@robins:1986]) introduced a new counterfactual model, the *finest fully randomized causally interpreted structured tree model that extended the treatment counterfactual model @neyman:sur:1923 and ([-@rubin:estimating:1974; to longitudinal with time-varying treatments, direct and indirect effects and feedback of one on another. Due to his lack of formal statistical training, and in @robins:1986 ([-@robins:1986]) from those found in mainstream as a consequence the be difficult Appendix present FFRCISTG model using a familiar notation.[^7] ![Causal tree graph a simple scenario with $
-varying covariate (e.g., presencE of emphyseMa) thaT is A prEdIctoR of bOth future exposURe anD of failure. In 1982, the standarD analYtIC appROaCh was To model THe CONdiTiOnAl pRoBAbIlity (I.e., tHe hazarD) of failure TimE $t$ As a function oF PaSt exposure HisTory using a tiMe-dEpendeNt cox PRoporTioNal haZards mODel. RobIns formalLy SHowed tHAt, even wHEN cOnfoUnding by unmeasureD FaCTors and model spEcificAtIOn ARE abSenT, this approAcH may rESult in eSTiMATEs oF Effect that may Fail to have a CAusAl inteRpRetATion, reGardlEsS Of wHether or not One aLso adjustS for thE MeasureD Time-depEndent ConFouNderS In ThE anAlYSis. iN fAct, IF prEvious exPoSuRe alsO preDICTS the SubSequEnt evOlution of the tIme-DepeNDenT confOundeRs (e.g., SiNce smOking iS a cauSe Of emphysema, it prEdicTs this disEasE) tHen ThE stanDArd appRoaCh cAn find aN artifaCTuaL eXPOSuRe effect even under tHe SHArP null hypOthesiS Of No NEt, direct Or IndIrecT EFfect Of exPOsUre on the FailurE TiMe Of any suBjEct. PriOr To @rObiNs:1986 ([-@robINs:1986]), alThough Informal DiscuSSions of net, direCT and indirect (i.E., MeDIAtED) effEctS of time varyIng eXPosuRes wERe To bE Found In the DiSCuSSion sections of most ePiDemiolOgic pApers, no formal MathematicAL DEfinitioNs exIStED. To address this, @RobinS:1986 ([-@robins:1986]) intROduced a nEw couNterfactUal model, tHE *Finest fuLly RanDomIzeD CAuSally interpreTED strUcTured trEe gRaph* (FFRcIStG)[^4] mOdeL thAt Extended tHe point tReAtMeNt CouNterfACtual modEl Of @nEyMan:Sur:1923 ([-@neYMan:sur:1923]) And @ruBin:eStImATinG:1974 ([-@rubin:eSTiMATing:1974; -@ruBi:Baye:1978])[^5] To lOnGitudInal STudIes with Time-varyiNg tREatmEnTs, Direct aNd indirect effEcTs and feedbAcK of One cauSE On anotheR. Due to his lack of formal stATisticaL trAininG, the Notation aNd fOrmaliSms IN @robinS:1986 ([-@robinS:1986]) diffEr FroM THose fOUNd In tHe Mainstream LITerAture; As A conSequencE the paper can be a difFIcuLt read.[^6] @richardSon:RobiNS:2013 ([-@RiChaRDsON:roBiNS:2013], ApPENdix C) present the fFRCISTG moDeL UsIng a more faMIliAr NotatioN.[^7] ![Causal Tree gRAph depiCting a simPle scenarIo With TREatMents at two Times $A_1$, $
-varying covariate (e.g.,presence o f emp hys ema )that isa predictor of both future exposure and o f fai lu r e. I n 1 982,the sta n da r d an al yt icap p ro ach w asto mode l the cond iti on al probabili t y(i.e., the ha zard) of fai lur e time $ t$a s a f unc tionof pas t expos ure histo ry usinga time-d e p en dent Cox proportional ha z ards model. Ro bins f or m al l y sh owe d that, ev en when confoun d in g b y u n measured fact ors and mod e l s pecifi ca tio n are a bsent ,t his approach m ay r esult inestima t es of e f fect th at may fa ilto h a ve a ca us a l i n te rpr e tat ion, reg ar dl ess o f wh e t h e r or no t on e als o adjusts for th e me a sur ed ti me-de pend en t con founde rs in t he analysis. In fac t, if pre vio us ex po surea lso pr edi cts the su bsequen t ev ol u t i on of the time-depen de n t c onfounde rs (e. g .,  s i nce smok in g i s ac a use o f em p hy sema, it predi c ts t his dis ea se) th en th e s tanda r d ap proach can fin d ana rtifactual exp o sure effect e v en u nd e r th e s harp null h ypot h esis ofn onet , dire ct or i n di r ect effect of expos ur e on t he fa ilure time of any subje c t . Priorto @ r ob i ns:1986 ([-@ro bins: 1986]), al t hough in forma l discus sions ofn e t, direc t a ndind ire c t ( i.e., mediate d ) eff ec ts of t ime varyin g e xpo sur eswe re to befound in t he d is cus sions ectionsof mo st ep idemi o logicpaper s, n ofo r mal mathem a ti c a l de fi ni tion s e xi sted. Toa ddr ess thi s, @robin s:1 9 86 ( [- @r obins:1 986]) introdu ce d a new co un ter factua l model, t he *finest fully random i zed cau sal ly in terp reted str uct ured t ree graph* (FFRC ISTG) [^ 4]m o del t h a text en ded the po i n t t reatm en t co unterfa ctual model of @ne y man :sur:1923 ([- @ne yman : s ur :19 2 3] ) an d@ rub i n :estimating:197 4 ([-@rubi n: e st imating:19 7 4;-@ Rubi:ba ye:1978 ])[^5 ] to lon gitudinal studieswi th t i m e-v arying tre atments, direct a n d ind i re ct ef fec ts and f eed backof one cau se on anoth er . Dueto hi slack offormal statistical trai ning,the n ota tion andfor m ali sms in @r obin s:1986 ([- @ro bin s:198 6]) diffe r fr o mtho s e fou nd i n the main s tr eam l it erature; as a con seque nce the pa percan be a difficul t read.[^6] @ri char d s on: rob i ns:2 01 3 ([-@richards on: ro b i ns:2013] ,Appendix C) present t h e FFR CISTGmodelusing a m or e famil iarnot ation.[^7 ] ![ C ausal t re eg raph d epic ti ng a s imples cena r i o with treatment s att w o tim e s $ A_1$, $
-varying covariate_(e.g., presence of_emphysema) that is a_predictor of_both_future exposure_and_of failure. In_1982, the standard_analytic approach was to_model the conditional_probability_(i.e., the hazard) of failure time $t$ as a function of past exposure history_using_a time-dependent_Cox_proportional_hazards model. Robins formally showed_that, even when confounding by_unmeasured factors_and model specification are absent, this approach may_result_in estimates of_effect that may fail to have a causal interpretation,_regardless of whether or not one_also adjusts for_the_measured_time-dependent confounders in the_analysis. In fact, if previous exposure_also predicts the subsequent evolution of_the time-dependent confounders (e.g., since smoking is a_cause of emphysema, it predicts this_disease) then the standard approach_can find_an artifactual exposure effect even_under the sharp_null hypothesis_of no net,_direct or indirect effect of exposure_on the failure_time of any subject. Prior to @robins:1986_([-@robins:1986]),_although informal discussions_of_net,_direct and_indirect (i.e., mediated) effects_of_time varying_exposures_were to be found in the_discussion_sections of most epidemiologic papers, no formal_mathematical definitions existed. To_address_this, @robins:1986 ([-@robins:1986]) introduced_a new counterfactual model, the_*finest fully randomized causally interpreted structured_tree graph*_(FFRCISTG)[^4] model_that extended the point treatment counterfactual model of @neyman:sur:1923 ([-@neyman:sur:1923]) and_@rubin:estimating:1974 ([-@rubin:estimating:1974; -@Rubi:baye:1978])[^5] to longitudinal studies_with time-varying treatments, direct_and indirect_effects_and feedback of_one_cause on_another. Due to his lack of formal_statistical training,_the notation and formalisms in @robins:1986_([-@robins:1986]) differ from those_found_in the mainstream literature; as a_consequence the paper can be a_difficult read.[^6] @richardson:robins:2013 ([-@richardson:robins:2013], Appendix_C)_present_the FFRCISTG model using a_more familiar notation.[^7] ![Causal tree graph depicting_a simple scenario_with treatments at two times $A_1$, $
is non-zero. This completes the proof of the theorem. 1. The proof of Theorem\[cuplength\] actually shows that if some product $x=x_1\cdots x_t\neq 0$ with $1\leq \mathrm{deg}(x_i)\leq \ell$, then for any vector bundle $\xi$ over $X$ with $\mathrm{charrank}_X(\xi)\geq\ell$ some product of the Stiefel-Whitney classes of $\xi$ of length greater than or equal to $t$ is non-zero. 2. The conclusion of Theorem\[cuplength\] is not true if $\mathrm{ucharrank}(X)<\mathrm{dim}(X)$. If $X=S^k$, $k\neq 1,2,4,8$, then $\mathrm{ucharrank}(X)=k-1<k$, $\mathrm{cup}(X)=1$ however $w(\xi)=1$ for any vector bundle $\xi$ over $X$. [**Proof of Theorem\[maintheorem\].**]{} First note that the assumption $\mathrm{ucharrank}(X)\geq 1$ is odd clearly implies that the function $$f:\mathrm{Vect}_{\mathbb R}(X)\longrightarrow \mathbb Z_2$$ defined by $$f(\xi)=\mathrm{charrank}_X(\xi) ~~~~~(\mathrm{mod}~~~~~ 2)$$ is surjective. We shall now check that $f$ is a semi-group homomorphism. To see this, let $\xi$ and $\eta$ be two bundles over $X$. We have the following cases. If $\xi$ and $\eta$ are both orientable, then so is $\xi\oplus\eta$. Hence $w_1(\xi\oplus\eta)=0$. As $r_X=1$, it follows that $\mathrm{charrank}_X(\xi\oplus\eta)=0$. The same argument shows that $\mathrm{charrank}_X(\xi)=0=\mathrm{charrank}_X(\eta)$. Thus in this case we have $f(\xi\oplus\eta)=f(\xi)+f(\eta)$. Next suppose that both $\xi$ and $\eta$ are non-orientable. Then, on the one hand, $\xi\oplus\eta$ is orientable and hence $f(\xi\oplus\eta
is non - zero. This completes the proof of the theorem. 1. The proof of Theorem\[cuplength\ ] actually shows that if some merchandise $ x = x_1\cdots x_t\neq 0 $ with $ 1\leq \mathrm{deg}(x_i)\leq \ell$, then for any vector package $ \xi$ over $ X$ with $ \mathrm{charrank}_X(\xi)\geq\ell$ some product of the Stiefel - Whitney classes of $ \xi$ of length great than or equal to $ t$ is non - zero. 2. The conclusion of Theorem\[cuplength\ ] is not on-key if $ \mathrm{ucharrank}(X)<\mathrm{dim}(X)$. If $ adam = S^k$, $ k\neq 1,2,4,8 $, then $ \mathrm{ucharrank}(X)=k-1 < k$, $ \mathrm{cup}(X)=1 $ however $ w(\xi)=1 $ for any vector bundle $ \xi$ over $ X$. [ * * validation of Theorem\[maintheorem\ ]. * * ] { } inaugural note that the presumption $ \mathrm{ucharrank}(X)\geq 1 $ is odd clearly entail that the function $ $ f:\mathrm{Vect}_{\mathbb R}(X)\longrightarrow \mathbb Z_2$$ defined by $ $ f(\xi)=\mathrm{charrank}_X(\xi) ~~~~~(\mathrm{mod}~~~~~ 2)$$ is surjective. We shall now check that $ f$ is a semi - group homomorphism. To see this, let $ \xi$ and $ \eta$ be two package over $ X$. We have the following cases. If $ \xi$ and $ \eta$ are both orientable, then so is $ \xi\oplus\eta$. therefore $ w_1(\xi\oplus\eta)=0$. As $ r_X=1 $, it follows that $ \mathrm{charrank}_X(\xi\oplus\eta)=0$. The same controversy shows that $ \mathrm{charrank}_X(\xi)=0=\mathrm{charrank}_X(\eta)$. therefore in this case we consume $ f(\xi\oplus\eta)=f(\xi)+f(\eta)$. Next suppose that both $ \xi$ and $ \eta$ are non - orientable. Then, on the one hand, $ \xi\oplus\eta$ is orientable and therefore $ f(\xi\oplus\eta
is non-zero. This completes uhe proof of the jhworem. 1. The pdoof of Gheorem\[cuplength\] actually shlww thau if some product $x=b_1\cdots x_t\jeq 0$ wity $1\lew \mathrm{deg}(e_j)\leq \ell$, then nor auy vector bundle $\xi$ over $X$ with $\mathrm{chdrfauk}_X(\xi)\geq\ell$ some product of the Stiesel-Whitmej classes of $\xy$ of jengfh greater than or equal to $t$ is nkn-zero. 2. The conclusipn of Theorem\[cuplength\] is jot hrue if $\mathrm{uchagrank}(X)<\mathrn{dim}(V)$. If $X=S^k$, $k\neq 1,2,4,8$, then $\mathrm{ucharrank}(S)=k-1<k$, $\mathrm{cup}(X)=1$ however $w(\xi)=1$ for xny vzctor bundlg $\si$ mver $X$. [**Proof of Treorem\[mainthcprem\].**]{} Fhrst noye that the asxum'tiob $\mathrm{ucharrank}(X)\geq 1$ is odd clearly iiplies thdt the function $$f:\marhem{Vecj}_{\mathtb R}(B)\oonerifhvardow \mahhbu Z_2$$ defined by $$f(\xi)=\mathem{charrank}_X(\xi) ~~~~~(\mathrk{mje}~~~~~ 2)$$ is surjectjve. We srall now check that $f$ is a semi-group hokomkrphism. To see this, let $\xi$ and $\eta$ be two bujdles ovew $X$. We have the following cases. If $\xi$ and $\eta$ are toth kfieubwcoe, then so is $\xi\oplus\eta$. Hence $w_1(\xi\oplus\eta)=0$. As $r_V=1$, iu fpllows that $\mabhrm{charrank}_X(\xi\oplis\fts)=0$. The same argoment skkwa that $\mathrm{charrwnk}_X(\xi)=0=\mwthrm{xharrank}_X(\tta)$. Tnus in this case we have $f(\xu\oplus\eta)=f(\xi)+y(\etq)$. Next suppose that both $\xi$ anb $\eta$ sre npn-orientable. Then, on thz one gand, $\xi\oplud\eta$ is odkentable and henze $n(\xi\mplus\eta
is non-zero. This completes the proof of 1. proof of actually shows that 0$ $1\leq \mathrm{deg}(x_i)\leq \ell$, for any vector $\xi$ over $X$ with $\mathrm{charrank}_X(\xi)\geq\ell$ some of the Stiefel-Whitney classes of $\xi$ of length greater than or equal to is non-zero. 2. The conclusion of Theorem\[cuplength\] is not true if $\mathrm{ucharrank}(X)<\mathrm{dim}(X)$. If $k\neq then $\mathrm{cup}(X)=1$ $w(\xi)=1$ for any vector bundle $\xi$ over $X$. [**Proof of Theorem\[maintheorem\].**]{} First note that the assumption 1$ is odd clearly implies that the function R}(X)\longrightarrow \mathbb Z_2$$ defined $$f(\xi)=\mathrm{charrank}_X(\xi) ~~~~~(\mathrm{mod}~~~~~ 2)$$ is surjective. shall check that is semi-group To see this, $\xi$ and $\eta$ be two bundles over $X$. We have the following cases. If $\xi$ and $\eta$ both orientable, is $\xi\oplus\eta$. $w_1(\xi\oplus\eta)=0$. $r_X=1$, follows that $\mathrm{charrank}_X(\xi\oplus\eta)=0$. argument shows that $\mathrm{charrank}_X(\xi)=0=\mathrm{charrank}_X(\eta)$. Thus in have $f(\xi\oplus\eta)=f(\xi)+f(\eta)$. Next suppose that both $\xi$ and are non-orientable. on the one hand, $\xi\oplus\eta$ is and hence $f(\xi\oplus\eta
is non-zero. This completes the Proof of the TheorEm. 1. THe pRoOf of theoRem\[cuplength\] acTUallY shows that if some producT $x=x_1\cdOtS X_t\neQ 0$ WiTh $1\leq \Mathrm{dEG}(x_I)\LEq \eLl$, ThEn fOr ANy VectoR buNdle $\xi$ oVer $X$ with $\maThrM{cHarrank}_X(\xi)\geQ\ElL$ some produCt oF the Stiefel-WHitNey claSsEs oF $\Xi$ of lEngTh greAter thAN or equAl to $t$ is noN-zERo. 2. The cONclusioN OF THeorEm\[cuplength\] is not tRUe IF $\mathrm{ucharraNk}(X)<\matHrM{DiM}(x)$. if $X=s^k$, $k\Neq 1,2,4,8$, then $\matHrM{uchaRRank}(X)=k-1<k$, $\MAtHRM{Cup}(x)=1$ However $w(\xi)=1$ for Any vector buNDle $\Xi$ over $x$. [**PRooF Of TheoRem\[maInTHeoRem\].**]{} First notE thaT the assumPtion $\mAThrm{uchARrank}(X)\gEq 1$ is odD clEarLy imPLiEs ThaT tHE fuNCtIon $$F:\MatHrm{Vect}_{\mAtHbB R}(X)\loNgriGHTARrow \MatHbb Z_2$$ DefinEd by $$f(\xi)=\mathrm{ChaRranK}_x(\xi) ~~~~~(\MathrM{mod}~~~~~ 2)$$ iS surJeCtive. we shalL now cHeCk that $f$ is a semi-gRoup HomomorphIsm. to See ThIs, let $\XI$ and $\etA$ be Two Bundles Over $X$. We HAve ThE FOLlOwing cases. If $\xi$ and $\eTa$ ARE bOth orienTable, tHEn So IS $\xi\oplus\EtA$. HeNce $w_1(\XI\Oplus\Eta)=0$. AS $R_X=1$, It followS that $\mAThRm{CharranK}_X(\Xi\opluS\eTa)=0$. THe sAme arGUmenT shows That $\mathRm{chaRRank}_X(\xi)=0=\mathrm{cHArrank}_X(\eta)$. ThuS In THIs CAse wE haVe $f(\xi\oplus\eTa)=f(\xI)+F(\eta)$. next SUpPosE That bOth $\xi$ AnD $\EtA$ Are non-orientable. TheN, oN the onE hand, $\Xi\oplus\eta$ is oRientable aND HEnce $f(\xi\oPlus\ETa
is non-zero. This complet es the pro of of th e t he orem . 1 . The proof o f The orem\[cuplength\] actu allysh o ws t h at if s ome pro d uc t $x= x_ 1\ cdo ts x_ t\neq 0$ with $ 1\leq \mat hrm {d eg}(x_i)\leq \e ll$, thenfor any vectorbun dle $\ xi $ o v er $X $ w ith $ \mathr m {charr ank}_X(\x i) \ geq\el l $ somep r od uctof the Stiefel-Wh i tn e y classes of $ \xi$ o fl en g t h g rea ter than o requal to $t$i sn o n -ze r o. 2. The c onclusion o f Th eorem\ [c upl e ngth\] is n ot tru e if $\math rm{u charrank} (X)<\m a thrm{di m }(X)$.If $X= S^k $,$k\n e q1, 2,4 ,8 $ , t h en $\ m ath rm{uchar ra nk }(X)= k-1< k $ , $\ma thr m{cu p}(X) =1$ however $ w(\ xi)= 1 $ f or an y vec torbu ndle$\xi$over$X $. [**Proof of The orem\[mai nth eo rem \] .**]{ } First no tethat th e assum p tio n$ \ m at hrm{ucharrank}(X)\ ge q 1$ is oddclearl y i mp l ies that t hefunc t i on $$ f:\m a th rm{Vect} _{\mat h bb R }(X)\lo ng righta rr ow\ma thbbZ _2$$ defin ed by $$ f(\xi ) =\mathrm{charr a nk}_X(\xi) ~~ ~ ~~ ( \ ma t hrm{ mod }~~~~~ 2)$$ iss urje ctiv e .Wes hallnow c he c kt hat $f$ is a semi-g ro up hom omorp hism. To seethis, let$ \ x i$ and $ \eta $ b e two bundles o ver $ X$. We hav e the fol lowin g cases. If $\xi $ and $\et a$are bo tho r ie ntable, thens o is$\ xi\oplu s\e ta$. He nce $w _1( \xi \o plus\eta) =0$. As$r _X =1 $, it foll o ws that$\ mat hr m{c harra n k}_X(\ xi\op lus\ et a) = 0$. The sa m ea r gume nt s hows th at $\ma thrm { cha rrank}_ X(\xi)=0= \ma t hrm{ ch ar rank}_X (\eta)$. Thus i n this cas ewehave $ f ( \xi\oplu s\eta)=f(\xi)+f(\eta)$. Next s upp ose t hatboth $\xi $ a nd $\e ta$ are no n-orie ntabl e. Th e n , ont h eone h and, $\xi\ o p lus \eta$ i s or ientabl e and hence $f(\xi \ opl us\eta
is_non-zero. This_completes the proof of_the theorem. 1.__The proof_of_Theorem\[cuplength\] actually shows_that if some_product $x=x_1\cdots x_t\neq 0$_with $1\leq \mathrm{deg}(x_i)\leq_\ell$,_then for any vector bundle $\xi$ over $X$ with $\mathrm{charrank}_X(\xi)\geq\ell$ some product of the_Stiefel-Whitney_classes of_$\xi$_of_length greater than or equal_to $t$ is non-zero. 2. _The conclusion_of Theorem\[cuplength\] is not true if $\mathrm{ucharrank}(X)<\mathrm{dim}(X)$. If_$X=S^k$,_$k\neq 1,2,4,8$, then_$\mathrm{ucharrank}(X)=k-1<k$, $\mathrm{cup}(X)=1$ however $w(\xi)=1$ for any vector bundle $\xi$_over $X$. [**Proof of Theorem\[maintheorem\].**]{} First note_that the assumption_$\mathrm{ucharrank}(X)\geq_1$_is odd clearly implies_that the function $$f:\mathrm{Vect}_{\mathbb R}(X)\longrightarrow \mathbb_Z_2$$ defined by $$f(\xi)=\mathrm{charrank}_X(\xi) ~~~~~(\mathrm{mod}~~~~~ 2)$$_is surjective. We shall now check that_$f$ is a semi-group homomorphism. To_see this, let $\xi$ and_$\eta$ be_two bundles over $X$. We_have the following_cases. If $\xi$_and $\eta$ are_both orientable, then so is $\xi\oplus\eta$._Hence $w_1(\xi\oplus\eta)=0$. As_$r_X=1$, it follows that $\mathrm{charrank}_X(\xi\oplus\eta)=0$. The_same_argument shows that_$\mathrm{charrank}_X(\xi)=0=\mathrm{charrank}_X(\eta)$._Thus_in this_case we have_$f(\xi\oplus\eta)=f(\xi)+f(\eta)$. Next_suppose that_both_$\xi$ and $\eta$ are non-orientable. Then,_on_the one hand, $\xi\oplus\eta$ is orientable and_hence $f(\xi\oplus\eta
spaced subsets of $S^1_{eucl}$ is somewhat unnatural; we leave open the following question, which would define the ‘upper and lower Rips magnitude’ of the circle intrinsically: Does this asymptotic behaviour extend to arbitrary finite subsets of $S^1_{eucl}$? For instance, given any $\epsilon>0$, is there a $\delta>0$ such that for all finite $A\subseteq S^1_{eucl}$ with $d_H(A,S^1_{eucl})<\delta$ we have $$e^{-2t}+2\pi t-\epsilon<|tA|_\mathrm{Rips}<e^{-2t}+2\pi t\sum_{r\text{ odd}}\frac1r e^{-2t\cos\left(\frac\pi{2r}\right)}\sin\left(\frac\pi{2r}\right)+\epsilon$$ for all $t$ in a given interval? Geodesic circle --------------- Finally, we note that the case of the geodesic circle $S^1_{geo}$ of total arclength $2\pi$ can be treated using the same methods as we used for $S^1_{eucl}$. Namely, we restrict attention to equally spaced subsets $C_n^{geo}$ described in Section \[section-cycles\]. (Note that $C_n^{geo}$ is just the $n$-cycle graph $C_n$ rescaled by $\frac{2\pi}{n}$.) We could calculate the limits along the same subsequences we examined in the case $S^1_{eucl}$ and find that they again exist, but instead we just state the final result regarding the lower and upper limit. In this case it turns out that the lower limit is still finite and can be expressed explicitly, whereas the upper limit becomes infinite. Thus the Rips magnitude of $S^1_{eucl}$ and $S^1_{geo}$ behave quite differently. \[theorem-GC\] The Rips magnitudes of geodesic cycles $C^{\mathrm{geo}}_n$ satisfy $$\liminf_{n\to\infty}|t C^{\mathrm{geo}}_n|_{\mathrm{Rips}}=e^{-\pi t}+2\pi t$$ and $$\limsup_{n\to\infty}|t C^{\mathrm{geo}}_n|_{\mathrm{Rips}}=\infty.$$ The behaviour of $|t C^{\mathrm
spaced subsets of $ S^1_{eucl}$ is somewhat unnatural; we bequeath loose the following question, which would define the ‘ upper and lower Rips order of magnitude ’ of the circle intrinsically: Does this asymptotic behaviour extend to arbitrary finite subset of $ S^1_{eucl}$? For example, given any $ \epsilon>0 $, is there a $ \delta>0 $ such that for all finite $ A\subseteq S^1_{eucl}$ with $ d_H(A, S^1_{eucl})<\delta$ we suffer $ $ e^{-2t}+2\pi t-\epsilon<|tA|_\mathrm{Rips}<e^{-2t}+2\pi t\sum_{r\text { odd}}\frac1r e^{-2t\cos\left(\frac\pi{2r}\right)}\sin\left(\frac\pi{2r}\right)+\epsilon$$ for all $ t$ in a given interval? Geodesic r-2 --------------- Finally, we note that the sheath of the geodesic traffic circle $ S^1_{geo}$ of total arclength $ 2\pi$ can be treated using the same methods as we use for $ S^1_{eucl}$. Namely, we restrict attention to equally space subsets $ C_n^{geo}$ described in Section \[section - cycles\ ]. (notice that $ C_n^{geo}$ is just the $ n$-cycle graph $ C_n$ rescale by $ \frac{2\pi}{n}$.) We could calculate the limits along the same subsequences we examined in the case $ S^1_{eucl}$ and determine that they again exist, but instead we just state the final result involve the lower and upper limit. In this case it turns out that the lower limit is still finite and can be expressed explicitly, whereas the upper limit becomes countless. therefore the Rips magnitude of $ S^1_{eucl}$ and $ S^1_{geo}$ behave quite differently. \[theorem - GC\ ] The Rips magnitude of geodesic cycles $ C^{\mathrm{geo}}_n$ satisfy $ $ \liminf_{n\to\infty}|t C^{\mathrm{geo}}_n|_{\mathrm{Rips}}=e^{-\pi t}+2\pi t$$ and $ $ \limsup_{n\to\infty}|t C^{\mathrm{geo}}_n|_{\mathrm{Rips}}=\infty.$$ The behavior of $ |t C^{\mathrm
spwced subsets of $S^1_{eucl}$ is somewhat unnatorql; we neave kpen the following question, which wonld eefint the ‘upper and lowdr Rips mwgnitude’ of uhe circle intrinsically: Does thjd asvm'totic behaviout extend to drbitrary finide sbbsets of $S^1_{eucl}$? For instance, given agy $\epsikoj>0$, is there a $\dglta>0$ xtch fhat for all finite $A\subseteq S^1_{eucm}$ with $v_H(A,S^1_{eucl})<\delta$ wr have $$e^{-2t}+2\pi t-\epsilon<|tA|_\mathgm{Rios}<e^{-2t}+2\pi t\sum_{r\text{ ofd}}\frac1r e^{-2t\cis\lesr(\frac\pi{2r}\righg)}\sin\left(\frac\pi{2r}\right)+\epailon$$ for all $t$ in a given intefval? Gzodesic cirxlw --------------- Fijdlly, we notx that the case of the geogesic corcle $S^1_{geo}$ of botal arxlength $2\pi$ can be treeted using the same iethods av ce used for $S^1_{eucl}$. Namwlt, we testrhct xrtevtikn tk equaply spaced sugsets $C_n^{geo}$ described in Sectipn \[wection-cycles\]. (Note ehwt $C_n^{geo}$ is just the $n$-cycle graph $C_n$ revcamed by $\frac{2\pi}{n}$.) We could xalculate the limits wlong the same subsequences we examined in the case $S^1_{eucl}$ dnd fmna tkqt thdt wgain exist, but instead we just state the finwm texult regarding the lower snf ikper limit. In jhis case jt turns out that hhe lowgr limut is stijl fonite and can be expressed wxplicitly, wkerwas the upper limic becomes inyinite. Thus the Rips magnitude of $R^1_{eucm}$ and $S^1_{geo}$ hehave qujge differently. \[thdorvm-GC\] The Rips magnitudes of gejdesic cyrles $E^{\mathrm{gdo}}_n$ xatisfr $$\liminf_{n\tl\inftn}|d C^{\mathrm{geo}}_n|_{\mathrl{Rips}}=g^{-\pi t}+2\ph t$$ and $$\lilsup_{n\to\infty}|t C^{\mathrm{geo}}_n|_{\mathrm{Cmps}}=\infty.$$ The bghaeiolr of $|t C^{\iathrk
spaced subsets of $S^1_{eucl}$ is somewhat unnatural; open following question, would define the of circle intrinsically: Does asymptotic behaviour extend arbitrary finite subsets of $S^1_{eucl}$? For given any $\epsilon>0$, is there a $\delta>0$ such that for all finite $A\subseteq with $d_H(A,S^1_{eucl})<\delta$ we have $$e^{-2t}+2\pi t-\epsilon<|tA|_\mathrm{Rips}<e^{-2t}+2\pi t\sum_{r\text{ odd}}\frac1r e^{-2t\cos\left(\frac\pi{2r}\right)}\sin\left(\frac\pi{2r}\right)+\epsilon$$ for all $t$ in given Geodesic --------------- we note that the case of the geodesic circle $S^1_{geo}$ of total arclength $2\pi$ can be using the same methods as we used for Namely, we restrict attention equally spaced subsets $C_n^{geo}$ described Section (Note that is the graph $C_n$ rescaled $\frac{2\pi}{n}$.) We could calculate the limits along the same subsequences we examined in the case $S^1_{eucl}$ and that they but instead just the result regarding the upper limit. In this case it the lower limit is still finite and can expressed explicitly, the upper limit becomes infinite. Thus Rips magnitude of $S^1_{eucl}$ and $S^1_{geo}$ behave quite \[theorem-GC\] The Rips magnitudes of geodesic cycles $C^{\mathrm{geo}}_n$ satisfy $$\liminf_{n\to\infty}|t C^{\mathrm{geo}}_n|_{\mathrm{Rips}}=e^{-\pi t}+2\pi t$$ and $$\limsup_{n\to\infty}|t behaviour of $|t C^{\mathrm
spaced subsets of $S^1_{eucl}$ is somEwhat unnatUral; wE leAve OpEn thE folLowing question, WHich Would define the ‘upper and Lower riPS magNItUde’ of The circLE iNTRinSiCaLly: doES tHis asYmpTotic beHaviour extEnd To Arbitrary finITe Subsets of $S^1_{EucL}$? For instance, GivEn any $\ePsIloN>0$, Is theRe a $\Delta>0$ Such thAT for alL finite $A\sUbSEteq S^1_{eUCl}$ with $d_h(a,s^1_{eUcl})<\dElta$ we have $$e^{-2t}+2\pi t-\epSIlON<|tA|_\mathrm{Rips}<e^{-2T}+2\pi t\suM_{r\TExT{ ODd}}\fRac1R e^{-2t\cos\left(\FrAc\pi{2r}\RIght)}\sin\LEfT(\FRAc\pI{2R}\right)+\epsilon$$ For all $t$ in a gIVen IntervAl? geoDEsic ciRcle --------------- FInALly, We note that tHe caSe of the geOdesic CIrcle $S^1_{gEO}$ of totaL arcleNgtH $2\pi$ Can bE TrEaTed UsINg tHE sAme MEthOds as we uSeD fOr $S^1_{euCl}$. NaMELY, We reStrIct aTtentIon to equally sPacEd suBSetS $C_n^{geO}$ descRibeD iN SectIon \[secTion-cYcLes\]. (Note that $C_n^{geO}$ is jUst the $n$-cyCle GrAph $c_n$ RescaLEd by $\frAc{2\pI}{n}$.) WE could cAlculatE The LiMITS aLong the same subsequEnCES wE examineD in the CAsE $S^1_{EUcl}$ and fiNd ThaT theY AGain eXist, BUt Instead wE just sTAtE tHe final ReSult reGaRdiNg tHe lowER and Upper lImit. In thIs casE It turns out that THe lower limit iS StILL fINite And Can be expresSed eXPlicItly, WHeReaS The upPer liMiT BeCOmes infinite. Thus the riPs magnItude Of $S^1_{eucl}$ and $S^1_{geO}$ behave quiTE DIfferentLy. \[thEOrEM-GC\] The Rips magnItudeS of geodesiC Cycles $C^{\mAthrm{Geo}}_n$ satiSfy $$\liminf_{N\TO\infty}|t C^{\MatHrm{Geo}}_N|_{\maTHRm{rips}}=e^{-\pi t}+2\pi t$$ anD $$\LImsuP_{n\To\infty}|T C^{\mAthrm{geO}}_n|_{\mAthRm{RIps}}=\InFty.$$ The behAviour of $|T C^{\MaThRm
spaced subsets of $S^1_{e ucl}$ is s omewh atunn at ural ; we leave open th e fol lowing question, which woul dd efin e t he ‘u pper an d l o w erRi ps ma gn i tu de’ o f t he circ le intrins ica ll y: Does thi s a symptoticbeh aviour exten d t o arbi tr ary finit e s ubset s of $ S ^1_{eu cl}$? For i n stance , givena n y$\ep silon>0$, is ther e a $\delta>0$ suc h that f o ra l l f ini te $A\subs et eq S^ 1 _{eucl} $ w i t h $d _ H(A,S^1_{eucl })<\delta$w e h ave $$ e^ {-2 t }+2\pi t-\e ps i lon <|tA|_\math rm{R ips}<e^{- 2t}+2\ p i t\sum _ {r\text { odd} }\f rac 1r e ^ {- 2t \co s\ l eft ( \f rac \ pi{ 2r}\righ t) }\ sin\l eft( \ f r a c\pi {2r }\ri ght)+ \epsilon$$ fo r a ll $ t $ i n a g iveninte rv al? Geodes ic ci rc le ------------ --- Finally, we n ote t hat t h e case of th e geode sic cir c le$S ^ 1 _ {g eo}$ of total arcl en g t h$2\pi$ c an bet re at e d usingth e s amem e thods asw eused for $S^1_ { eu cl }$. Nam el y, were str ict atte n tion to eq ually sp aceds ubsets $C_n^{g e o}$ described in S ec t ion\[s ection-cycl es\] . (No te t h at $C _ n^{ge o}$ i sj us t the $n$-cycle grap h$C_n$resca led by $\frac {2\pi}{n}$ . ) We coul d ca l cu l ate the limits alon g the same subseque nceswe exami ned in th e case $S^ 1_{ euc l}$ an d fi nd that theya g ainex ist, bu t i nsteadwejus t s tat ethe final resultre ga rd in g t he lo w er and u pp erli mit . Int his ca se it tur ns o u t t hat the lo w e r li mi tis s til lfinit e an d ca n be ex pressed e xpl i citl y, w hereasthe upper lim it becomes i nf ini te. Th u s the Rip s magnitude of $S^1_{eu c l}$ and $S ^1_{g eo}$ behave q uit e diff ere n tly. \[theo rem-G C\ ] T h e Rips m ag nit ud es of geod e s iccycle s$C^{ \mathrm {geo}}_n$ satisfy$ $\l iminf_{n\to\i nft y}|t C ^{ \ma t hr m {ge o} } _n| _ { \mathrm{Rips}}= e^{-\pi t} +2 \ pi t$$ and $ $ \li ms up_{n\t o\infty }|t C ^ {\mathr m{geo}}_n |_{\mathr m{ Rips } } =\i nfty.$$ T he behav iour of $ | t C^{ \ ma thrm
spaced_subsets of_$S^1_{eucl}$ is somewhat unnatural;_we leave_open_the following_question,_which would define_the ‘upper and_lower Rips magnitude’ of_the circle intrinsically: Does_this_asymptotic behaviour extend to arbitrary finite subsets of $S^1_{eucl}$? For instance, given any $\epsilon>0$,_is_there a_$\delta>0$_such_that for all finite $A\subseteq_S^1_{eucl}$ with $d_H(A,S^1_{eucl})<\delta$ we have_$$e^{-2t}+2\pi t-\epsilon<|tA|_\mathrm{Rips}<e^{-2t}+2\pi_t\sum_{r\text{ odd}}\frac1r e^{-2t\cos\left(\frac\pi{2r}\right)}\sin\left(\frac\pi{2r}\right)+\epsilon$$ for all $t$ in a_given_interval? Geodesic circle --------------- Finally, we_note that the case of the geodesic circle $S^1_{geo}$_of total arclength $2\pi$ can be_treated using the_same_methods_as we used for_$S^1_{eucl}$. Namely, we restrict attention to_equally spaced subsets $C_n^{geo}$ described in_Section \[section-cycles\]. (Note that $C_n^{geo}$ is just_the $n$-cycle graph $C_n$ rescaled by_$\frac{2\pi}{n}$.) We could calculate the limits_along the_same subsequences we examined in_the case $S^1_{eucl}$_and find_that they again_exist, but instead we just state_the final result_regarding the lower and upper limit._In_this case it_turns_out_that the_lower limit is_still_finite and_can_be expressed explicitly, whereas the upper_limit_becomes infinite. Thus the Rips magnitude of_$S^1_{eucl}$ and $S^1_{geo}$ behave_quite_differently. \[theorem-GC\] The Rips magnitudes_of geodesic cycles $C^{\mathrm{geo}}_n$ satisfy_$$\liminf_{n\to\infty}|t C^{\mathrm{geo}}_n|_{\mathrm{Rips}}=e^{-\pi t}+2\pi t$$ and $$\limsup_{n\to\infty}|t_C^{\mathrm{geo}}_n|_{\mathrm{Rips}}=\infty.$$ The behaviour_of $|t_C^{\mathrm
zebruch class, with $\theta(fh)=[fh]$ the relative fundamental class. Moreover, these characteristic classes commute with the corresponding orientations $\theta$ of a smooth morphism (as already explained before). So we only have to show that - the corresponding Grothendieck transformation $$\ga_{c\ell}=:\La_y^{mot}: \bM(\m V/X \xrightarrow{f} Y) \to \bK(X \xrightarrow{f} Y)$$ from Corollary \[twisting\] vanishes on the subgroup $\mathbb {BL}(\m V/ X \xrightarrow{f} Y)$, and - the relation $\ga_{T^*_y}=\tau\circ \La_y^{mot}$ up to the renormalization by the multiplication with $(1+y)^i$ on $\bH^i(-)\otimes \bQ[y]$. \(i) : Let us identify the vector bundle $T^*_{fh}$ for the smooth morphism $fh: V\to Y$ with the corresponding locally free sheaf $\Omega^1_{fh}$ of sections given by the relative one-forms, so that $$\Lambda_y^{mot}([V \xrightarrow{h} X]) := \sum_{p \geq 0} h_*([\Omega^p_{fh}] \bullet \m O_{fh}) \cdot y^p\:.$$ Note that by the definition of relative perfectness, $D^b_{id-perf}(V)=D^b_{fh-perf}(V)$ for the smooth morphism $fh$, so that $$\bullet \m O_{fh}: \bK(V \xrightarrow{id_V} V)=K_0(D^b_{id-perf}(V)) \xrightarrow{\sim} K_0(D^b_{fh-perf}(V))=\bK(V \xrightarrow{fh} X)\:,$$ with $h_*(\;-\; \bullet \m O_{fh})$ induced by the total direct image $$Rh_*: D^b_{id-perf}(V)= D^b_{fh-perf}(V)\to D^b_{f-perf}(X)\:.$$ Consider now a blow-up diagram $$\begin{CD} E @> i'>> Bl_{S}X' \\ @V q' VV @VV q
zebruch class, with $ \theta(fh)=[fh]$ the relative fundamental class. furthermore, these characteristic class commute with the corresponding orientation $ \theta$ of a fluent morphism (as already explained ahead). indeed we only suffer to show that - the corresponding Grothendieck transformation $ $ \ga_{c\ell}=:\La_y^{mot }: \bM(\m V / X \xrightarrow{f } Y) \to \bK(X \xrightarrow{f } Y)$$ from Corollary \[twisting\ ] vanishes on the subgroup $ \mathbb { BL}(\m V/ X \xrightarrow{f } Y)$, and - the relation back $ \ga_{T^*_y}=\tau\circ \La_y^{mot}$ up to the renormalization by the multiplication with $ (1+y)^i$ on $ \bH^i(-)\otimes \bQ[y]$. \(i): Let us name the vector bundle $ T^*_{fh}$ for the smooth morphism $ fh: V\to Y$ with the comparable locally free bundle $ \Omega^1_{fh}$ of section given by the relative one - forms, so that $ $ \Lambda_y^{mot}([V \xrightarrow{h } X ]): = \sum_{p \geq 0 } h_*([\Omega^p_{fh } ] \bullet \m O_{fh }) \cdot y^p\:.$$ notice that by the definition of relative perfectness, $ D^b_{id - perf}(V)=D^b_{fh - perf}(V)$ for the smooth morphism $ fh$, so that $ $ \bullet \m O_{fh }: \bK(V \xrightarrow{id_V } V)=K_0(D^b_{id - perf}(V) ) \xrightarrow{\sim } K_0(D^b_{fh - perf}(V))=\bK(V \xrightarrow{fh } X)\:,$$ with $ h_*(\;-\; \bullet \m O_{fh})$ induced by the total direct persona $ $ Rh _ *: D^b_{id - perf}(V)= D^b_{fh - perf}(V)\to D^b_{f - perf}(X)\:.$$ Consider now a blow - up diagram $ $ \begin{CD } E @ > i' > > Bl_{S}X' \\ @V q' VV @VV q
zebguch class, with $\theta(fh)=[fm]$ the relative foneamentel clasa. Moreovef, these characteristic classxs cimmutt with the correspovding orivntations $\theua$ of a smooth moclhism (as alrezfy erpoained before). Xo we only have to show dhxt - the corresponding Grothendieck twansforkahion $$\ga_{c\ell}=:\La_y^{iot}: \nI(\m V/S \xrightarrow{f} Y) \to \bK(X \xrightadrow{f} J)$$ from Corollary \[ywisting\] vanishes on the skbgrlup $\mathbb {BL}(\m V/ D \xrightarriw{f} T)$, and - the felation $\ga_{T^*_y}=\tau\circ \Lz_y^{mot}$ up to the renormalization by tke multiplixarioj with $(1+y)^i$ on $\bH^i(-)\onimes \bQ[y]$. \(i) : Lcn us idantify yhe vector buncle $T^*_{fy}$ for the smooth morpiism $fh: V\to Y$ with tre corres[ouding locally free shwad $\Omeca^1_{fh}$ of rwctkona jivsn by hhe relative kne-forms, so that $$\Lambda_y^{mot}([V \xtidytarrow{h} X]) := \aum_{p \gqq 0} h_*([\Omega^p_{fh}] \bullet \m O_{fh}) \cdot y^p\:.$$ Note thdt gy the definition of reoative perfectness, $D^b_{ld-perf}(V)=D^b_{sh-perf}(V)$ for the smooth morphism $fh$, so that $$\bullet \m O_{fi}: \cK(V \xvighgqrgow{id_V} V)=K_0(D^b_{id-perf}(V)) \xrightarrow{\sim} K_0(D^b_{fh-perf}(V))=\bK(D \xtinhtarrow{fh} X)\:,$$ wibh $h_*(\;-\; \bullet \m O_{fh})$ onfuvgd by the totau direef jmage $$Rh_*: D^b_{id-perf}(V)= D^b_{fh-petf}(V)\to E^b_{f-perf}(X)\:.$$ Cjnsicer now a blow-up diagram $$\betin{CD} E @> i'>> Bl_{F}Z' \\ @V q' VV @VV q
zebruch class, with $\theta(fh)=[fh]$ the relative fundamental these classes commute the corresponding orientations (as explained before). So only have to that - the corresponding Grothendieck transformation \bM(\m V/X \xrightarrow{f} Y) \to \bK(X \xrightarrow{f} Y)$$ from Corollary \[twisting\] vanishes on subgroup $\mathbb {BL}(\m V/ X \xrightarrow{f} Y)$, and - the relation $\ga_{T^*_y}=\tau\circ \La_y^{mot}$ to renormalization the with $(1+y)^i$ on $\bH^i(-)\otimes \bQ[y]$. \(i) : Let us identify the vector bundle $T^*_{fh}$ for the morphism $fh: V\to Y$ with the corresponding locally sheaf $\Omega^1_{fh}$ of sections by the relative one-forms, so $$\Lambda_y^{mot}([V X]) := \geq h_*([\Omega^p_{fh}] \m O_{fh}) \cdot Note that by the definition of relative perfectness, $D^b_{id-perf}(V)=D^b_{fh-perf}(V)$ for the smooth morphism $fh$, so that $$\bullet O_{fh}: \bK(V \xrightarrow{\sim} K_0(D^b_{fh-perf}(V))=\bK(V X)\:,$$ $h_*(\;-\; \m O_{fh})$ induced total direct image $$Rh_*: D^b_{id-perf}(V)= D^b_{fh-perf}(V)\to a blow-up diagram $$\begin{CD} E @> i'>> Bl_{S}X' @V q' @VV q
zebruch class, with $\theta(fh)=[fh]$ The relativE fundAmeNtaL cLass. moreOver, these charaCTeriStic classes commute with The coRrESponDInG orieNtationS $\ThETA$ of A sMoOth MoRPhIsm (as AlrEady expLained befoRe). SO wE only have to sHOw That - the corResPonding GrothEndIeck trAnSfoRMatioN $$\ga_{C\ell}=:\LA_y^{mot}: \bm(\M V/X \xriGhtarrow{f} y) \tO \BK(X \xriGHtarrow{F} y)$$ FrOm CoRollary \[twisting\] vaNIsHEs on the subgrouP $\mathbB {Bl}(\M V/ x \XRigHtaRrow{f} Y)$, and - tHe RelatIOn $\ga_{T^*_y}=\tAU\cIRC \la_y^{MOt}$ up to the renoRmalization BY thE multiPlIcaTIon witH $(1+y)^i$ on $\BH^I(-)\OtiMes \bQ[y]$. \(i) : Let uS ideNtify the vEctor bUNdle $T^*_{fh}$ FOr the smOoth moRphIsm $Fh: V\tO y$ wItH thE cORreSPoNdiNG loCally freE sHeAf $\OmeGa^1_{fh}$ OF SECtioNs gIven By the Relative one-foRms, So thAT $$\LaMbda_y^{Mot}([V \xRighTaRrow{h} x]) := \sum_{p \gEq 0} h_*([\OmEgA^p_{fh}] \bullet \m O_{fh}) \cDot y^P\:.$$ Note that By tHe DefInItion OF relatIve PerFectnesS, $D^b_{id-peRF}(V)=D^B_{fH-PERf}(v)$ for the smooth morphIsM $FH$, sO that $$\bulLet \m O_{fH}: \BK(v \xRIghtarroW{iD_V} V)=k_0(D^b_{iD-PErf}(V)) \xRighTArRow{\sim} K_0(D^B_{fh-perF}(v))=\bk(V \XrightaRrOw{fh} X)\:,$$ wItH $h_*(\;-\; \bUllEt \m O_{fH})$ InduCed by tHe total dIrect IMage $$Rh_*: D^b_{id-perf}(v)= d^b_{fh-perf}(V)\to D^b_{F-PeRF}(x)\:.$$ CONsidEr nOw a blow-up diAgraM $$\BegiN{CD} E @> I'>> bl_{s}X' \\ @V Q' vV @VV q
zebruch class, with $\thet a(fh)=[fh] $ the re lat iv e fu ndam ental class. M oreo ver, these characteris tic c la s sesc om mutewith th e c o r res po nd ing o r ie ntati ons $\thet a$ of a sm oot hmorphism (as al ready expl ain ed before).Sowe onl yhav e to s how that - t he cor respondin gG rothen d ieck tr a n sf orma tion $$\ga_{c\ell } =: \ La_y^{mot}: \b M(\m V /X \ x r igh tar row{f} Y) \ to \b K (X \xr i gh t a r row { f} Y)$$ from Corollary\ [tw isting \] va n isheson th es ubg roup $\math bb { BL}(\m V/ X \x r ightarr o w{f} Y )$, an d - the re la tio n$ \ga _ {T ^*_ y }=\ tau\circ \ La _y^{m ot}$ u p to t hereno rmali zation by the mu ltip l ica tionwith$(1+ y) ^i$ o n $\bH ^i(-) \o times \bQ[y]$. \(i ) : Let u s i de nti fy thev ectorbun dle $T^*_{ fh}$ fo r th es m o ot h morphism $fh: V\ to Y $with the corre s po nd i ng local ly fr ee s h e af $\ Omeg a ^1 _{fh}$ o f sect i on sgiven b ythe re la tiv e o ne-fo r ms,so tha t $$\Lam bda_y ^ {mot}([V \xrig h tarrow{h} X] ) : = \s u m_{p \g eq 0} h_*([ \Ome g a^p_ {fh} ] \ bul l et \m O_{f h} ) \ c dot y^p\:.$$ Note t ha t by t he de finition of r elative pe r f e ctness,$D^b _ {i d -perf}(V)=D^b_ {fh-p erf}(V)$ f o r the sm oothmorphism $fh$, so t hat $$\b ull et\mO_{ f h }: \bK(V \xrigh t a rrow {i d_V} V )=K _0(D^b_ {id -pe rf} (V) )\xrightar row{\sim }K_ 0( D^ b_{ fh-pe r f}(V))=\ bK (V\x rig htarr o w{fh} X)\: ,$$wi th $h_ *(\;-\; \b u l let\m O _{fh })$ i nduce d by the totaldirect im age $$Rh _* :D^b_{id -perf}(V)= D^ b_ {fh-perf}( V) \to D^b_{ f - perf}(X) \:.$$ Consider now a b l ow-up d iag ram $ $\be gin{CD} E @> i'>>Bl_ { S}X' \ \ @V q ' VV @ VVq
zebruch class,_with $\theta(fh)=[fh]$_the relative fundamental class. Moreover,_these characteristic_classes_commute with_the_corresponding orientations $\theta$_of a smooth_morphism (as already explained_before). So we_only_have to show that - the corresponding Grothendieck transformation $$\ga_{c\ell}=:\La_y^{mot}: \bM(\m V/X _\xrightarrow{f}_ Y)_\to_\bK(X_ \xrightarrow{f} Y)$$ from_Corollary \[twisting\] vanishes on the_subgroup $\mathbb_{BL}(\m V/ X \xrightarrow{f} Y)$, and -__ the relation_$\ga_{T^*_y}=\tau\circ \La_y^{mot}$ up to the renormalization by the multiplication_with $(1+y)^i$ on $\bH^i(-)\otimes \bQ[y]$. \(i) :_Let us identify_the_vector_bundle $T^*_{fh}$ for the_smooth morphism $fh: V\to Y$ with_the corresponding locally free sheaf $\Omega^1_{fh}$_of sections given by the relative one-forms,_so that $$\Lambda_y^{mot}([V \xrightarrow{h} X])_:= \sum_{p \geq 0} h_*([\Omega^p_{fh}] \bullet_\m O_{fh})_\cdot y^p\:.$$ Note that by_the definition of_relative perfectness,_$D^b_{id-perf}(V)=D^b_{fh-perf}(V)$ for the_smooth morphism $fh$, so that $$\bullet_\m O_{fh}: \bK(V_\xrightarrow{id_V} V)=K_0(D^b_{id-perf}(V)) \xrightarrow{\sim} K_0(D^b_{fh-perf}(V))=\bK(V \xrightarrow{fh} X)\:,$$_with_$h_*(\;-\; \bullet \m_O_{fh})$_induced_by the_total direct image_$$Rh_*:_D^b_{id-perf}(V)= D^b_{fh-perf}(V)\to_D^b_{f-perf}(X)\:.$$ Consider_now a blow-up diagram $$\begin{CD} E @>_i'>>_Bl_{S}X' \\ @V q' VV @VV q
LFPs at each electrode were recorded for 18s while the fly was awake and 18s more after the fly was anaesthetised (isoflurane, 0.6% by volume, through an evaporator). Flies’ unresponsiveness during anaesthesia was confirmed by the absence of behavioural responses to a series of air puffs, and recovery was also confirmed after isoflurane gas was turned off [@CohenEneuro2016]. We used data sampled at 1kHz for the analysis [@CohenEneuro2016], and to obtain an estimate of local neural activity, the 16 electrodes were re-referenced by subtracting adjacent signals giving 15 channels which we parametrise as $c \in [1,15]$. Line noise was removed from the recordings, followed by linear de-trending and removing the mean. The resulting data is a fluctuating voltage signal, which is time-binned (1ms bins) and binarised by splitting over the median, leading to a time series, see Fig. \[fig:workflow\](b). For each of the 13 flies in our data set, we have 30 time series of length $N = 18,000$. They correspond to the 15 channels, labelled numerically from the central to peripheral region as depicted in Fig. \[fig:workflow\](a), and the two states of conscious arousal. Using the CSSR algorithm [@CSSR2], we construct [$\epsilon\text{-machines}$]{} for each of these time series as a function of maximum memory length within the range $\lambda \in [2,11]$. This is below the memory length $L(N) \sim 14$ beyond which we would be unable to reliably determine transition probabilities for a sequence of length $N$ (see Sec. \[sec:Bkg-cssr\]) [^2]. We record the resulting $3,900$ [$\epsilon\text{-machine}$]{} structures and their corresponding statistical complexities, and group them according to their respective level of conscious arousal, $\psi$, channel location, $c$, and maximum memory length, $\lambda$. Thus, statistical complexity $C_{\mu}$ is a function of the set of parameters $\{\psi,c,\lambda\}$ for each fly, $f$. We are principally interested in the difference in statistical complexity over states of conscious arousal $$\begin{gathered} \Delta C_{\mu} = C_{\mu}^{\text{wake}} - C_{\mu}^{\text{ana
LFPs at each electrode were recorded for 18s while the fly was awake and eighteen more after the tent-fly was anaesthetised (isoflurane, 0.6% by book, through an evaporator). Flies ’ unresponsiveness during anesthesia was confirmed by the absence of behavioral responses to a series of air travel quilt, and recovery was besides confirm after isoflurane gas was turn off   [ @CohenEneuro2016 ]. We used datum sampled at 1kHz for the analysis   [ @CohenEneuro2016 ], and to obtain an estimate of local neural activity, the 16 electrodes were re - reference by subtracting adjacent signals giving 15 channel which we parametrise as $ c \in [ 1,15]$. Line noise was remove from the recordings, come by linear de - trending and removing the mean. The resulting data is a fluctuating voltage signal, which is meter - binned (1ms bins) and binarised by splitting over the median, leading to a time series, see Fig.   \[fig: workflow\](b). For each of the 13 flies in our data set, we have 30 time series of length $ N = 18,000$. They correspond to the 15 channels, labelled numerically from the cardinal to peripheral area as depicted in Fig.   \[fig: workflow\](a), and the two states of conscious arousal. use the CSSR algorithm   [ @CSSR2 ], we construct [ $ \epsilon\text{-machines}$ ] { } for each of these time series as a affair of maximum memory length within the range $ \lambda \in [ 2,11]$. This is below the memory length $ L(N) \sim 14 $ beyond which we would be ineffective to reliably determine transition probabilities for a sequence of length $ N$ (see Sec.   \[sec: Bkg - cssr\ ])   [ ^2 ]. We record the resulting $ 3,900 $ [ $ \epsilon\text{-machine}$ ] { } structures and their corresponding statistical complexities, and group them harmonize to their respective level of conscious foreplay, $ \psi$, distribution channel location, $ c$, and maximum memory length, $ \lambda$. Thus, statistical complexity $ C_{\mu}$ is a affair of the set of parameters $ \{\psi, c,\lambda\}$ for each fly, $ f$. We are chiefly interested in the dispute in statistical complexity over states of conscious arousal $ $ \begin{gathered } \Delta C_{\mu } = C_{\mu}^{\text{wake } } - C_{\mu}^{\text{ana
LFOs at each electrode wert recorded for 18s cyile tie fly sas awakd and 18s more after the fly wes abaestyetised (isoflurane, 0.6% by volume, tjrough ab eveporator). Flies’ uidesponslrenesa durnnj anaesthesia wss confirmad by the absetcd lf behavioural responses to a serief of ait ouffs, and recodery ras zlso confirmed after isoflurane gaa was tlrned off [@CohenEneiro2016]. We used data sampled at 1kHz for the analysis [@ClhenEneuro2016], qnd ei obtain an dstimate of local neurzl activity, the 16 electrodes werd re-rzferenced bi auhjracting adjecent fignals givikb 15 chatnels wnich we paramebrise as $c \in [1,15]$. Line noise was removed from the tecordings, fkllowed by linear dw-trenging and eemuvihg tge meaj. Tie resultinf data is a fluctuating voltagt sytnal, which is time-bynged (1ms bins) and binarised by splitting mved the median, leading to a time series, see Fih. \[fig:workfjow\](b). For each of the 13 flies in our data set, we hava 30 tijd strlcs ow lfngth $N = 18,000$. They correspond to the 15 channels, lafslkec numerically nrom the central tp oetypheral regiov as dzlidted in Fig. \[fig:workvlow\](a), agd thw two staues og conscious arousal. Using tye CSSR algogithn [@CSSR2], we construct [$\epsilon\texc{-machimes}$]{} fpr each of these time szries zs a functiln of maxjoum memory lengtf wpthit the range $\lambda \in [2,11]$. Thif is belox the memory lenbth $L(N) \sim 14$ beyojd whlwh we would be unahle tl seliably dftermine transition probabilitixx for a sequgnca ox length $N$ (sec Sec. \[sec:Bkg-cssr\]) [^2]. We record the resultnng $3,900$ [$\eosilon\text{-jachine}$]{} structures and their costesponding stetistical complexuties, ava group them avcording no theur respective levek ow conscious arouwal, $\psi$, channel lovatkon, $c$, aid mavhmum memory nenggh, $\usmbda$. Thus, suctiwgicak complexity $C_{\mu}$ is d fuhction of the set pf parametgrs $\{\psi,c,\lwmbda\}$ for eacn fly, $f$. We are prijcipanly intetesjed in the difference in statisfical comolewity over staees if conscious arousal $$\begin{gathered} \Delta C_{\mu} = C_{\mu}^{\texv{wake}} - C_{\mu}^{\text{ana
LFPs at each electrode were recorded for the was awake 18s more after 0.6% volume, through an Flies’ unresponsiveness during was confirmed by the absence of responses to a series of air puffs, and recovery was also confirmed after gas was turned off [@CohenEneuro2016]. We used data sampled at 1kHz for the [@CohenEneuro2016], to an of local neural activity, the 16 electrodes were re-referenced by subtracting adjacent signals giving 15 channels we parametrise as $c \in [1,15]$. Line noise removed from the recordings, by linear de-trending and removing mean. resulting data a voltage which is time-binned bins) and binarised by splitting over the median, leading to a time series, see Fig. \[fig:workflow\](b). For of the in our set, have time series of = 18,000$. They correspond to the numerically from the central to peripheral region as in Fig. and the two states of conscious Using the CSSR algorithm [@CSSR2], we construct [$\epsilon\text{-machines}$]{} each of these time series as a function of maximum memory length within the range [2,11]$. This is below memory length $L(N) 14$ which would unable to determine transition probabilities for a sequence of length $N$ (see Sec. [^2]. We record the resulting $3,900$ [$\epsilon\text{-machine}$]{} structures and their complexities, group them according their respective level of arousal, channel location, $c$, and length, Thus, is function the set of parameters for each fly, $f$. We principally interested in the states of conscious arousal $$\begin{gathered} \Delta C_{\mu} = - C_{\mu}^{\text{ana
LFPs at each electrode were reCorded for 18s While The Fly WaS awaKe anD 18s more after the FLy waS anaesthetised (isofluraNe, 0.6% by vOlUMe, thROuGh an eVaporatOR). FLIEs’ uNrEsPonSiVEnEss duRinG anaestHesia was coNfiRmEd by the absenCE oF behaviourAl rEsponses to a sEriEs of aiR pUffS, And reCovEry waS also cONfirmeD after isoFlURane gaS Was turnED OfF [@CohEnEneuro2016]. We used datA SaMPled at 1kHz for thE analySiS [@coHENEnEurO2016], and to obtaIn An estIMate of lOCaL NEUraL Activity, the 16 elEctrodes werE Re-rEferenCeD by SUbtracTing aDjACenT signals givIng 15 cHannels whIch we pARametriSE as $c \in [1,15]$. LIne noiSe wAs rEmovED fRoM thE rECorDInGs, fOLloWed by linEaR dE-trenDing AND REmovIng The mEan. ThE resulting datA is A fluCTuaTing vOltagE sigNaL, whicH is timE-binnEd (1Ms bins) and binariSed bY splittinG ovEr The MeDian, lEAding tO a tIme Series, sEe Fig. \[fiG:WorKfLOW\](B). FOr each of the 13 flies in OuR DAtA set, we haVe 30 time SErIeS Of length $n = 18,000$. THey CorrESPond tO the 15 CHaNnels, labElled nUMeRiCally frOm The cenTrAl tO peRipheRAl reGion as Depicted In Fig. \[FIg:workflow\](a), and THe two states of COnSCIoUS aroUsaL. Using the CSsR alGOritHm [@CSsr2], wE coNStrucT [$\epsiLoN\TeXT{-machines}$]{} for each of tHeSe time SerieS as a function oF maximum meMORY length wIthiN ThE Range $\lambda \in [2,11]$. THis is Below the meMOry lengtH $L(N) \siM 14$ beyond wHich we wouLD Be unable To rEliAblY deTERmIne transition PRObabIlIties foR a sEquence Of lEngTh $N$ (See seC. \[sec:Bkg-csSr\]) [^2]. We recoRd ThE rEsUltIng $3,900$ [$\epSIlon\text{-MaChiNe}$]{} StrUcturES and thEir coRresPoNdINg sTatistiCAl COMpleXiTiEs, anD grOuP them AccoRDinG to theiR respectiVe lEVel oF cOnScious aRousal, $\psi$, chanNeL location, $c$, AnD maXimum mEMOry lengtH, $\lambda$. Thus, statistical cOMplexitY $C_{\mU}$ is a fUnctIon of the sEt oF paramEteRS $\{\psi,c,\lAmbda\}$ fOr eacH fLy, $f$. wE Are prINCiPalLy Interested IN The DiffeReNce iN statisTical complexity oveR StaTes of consciouS arOusaL $$\BEgIn{gAThERed} \deLTa C_{\MU} = c_{\mu}^{\text{wake}} - C_{\mu}^{\tExt{ana
LFPs at each electrode we re recorde d for 18 s w hi le t he f ly was awake a n d 18 s more after the fly w as an ae s thet i se d (is ofluran e ,0 . 6%by v olu me , t hroug h a n evapo rator). Fl ies ’unresponsive n es s during a nae sthesia wascon firmed b y t h e abs enc e ofbehavi o ural r esponsesto a seri e s of ai r pu ffs, and recovery was al s o confirmed af ter is of l ur a n e g aswas turned o ff [@ C ohenEne u ro 2 0 1 6]. We used data sampled at 1kH z forth e a n alysis  [@Co he n Ene uro2016], a nd t o obtainan est i mate of local n euralact ivi ty,t he 1 6 e le c tro d es we r e r e-refere nc ed by s ubtr a c t i ng a dja cent sign als giving 15 ch anne l s w hichwe pa rame tr ise a s $c \ in [1 ,1 5]$. Line noise was removedfro mthe r ecord i ngs, f oll owe d by li near de - tre nd i n g a nd removing the me an . Th e result ing da t ais a fluctu at ing vol t a ge si gnal , w hich istime-b i nn ed (1ms b in s) and b ina ris ed by spli ttingover the medi a n, leading toa time series, se e Fi g . \[ fig :workflow\] (b). For eac h o f t h e 13flies i n o u r data set, we have 3 0 time seri es of length$N = 18,00 0 $ . They co rres p on d to the 15 cha nnels , labelled numerica lly f rom thecentral t o peripher alreg ion as d ep icted in Fig. \ [fig :w orkflow \]( a), and th e t wosta te s of cons cious ar ou sa l. U sin g the CSSR alg or ith m[@C SSR2] , we co nstru ct [ $\ ep s ilo n\text{ - ma c h ines }$ ]{ } fo r e ac h ofthes e ti me seri es as a f unc t ionof m aximummemory length w ithin thera nge $\lam b d a \in [2 ,11]$. This is below th e memory le ngth$L(N ) \sim 14 $ b eyondwhi c h we w ould b e una bl e t o relia b l ydet er mine trans i t ion prob ab ilit ies for a sequence of len g th$N$ (see Sec.  \[ sec: B k g- css r \] )  [^ 2] . We r ecord the resul ting $3,90 0$ [$ \epsilon\t e xt{ -m achine} $]{} st ructu r es andtheir cor respondin gstat i s tic al complex ities, a nd groupt hem a c co rding to their r esp ectiv e leve l of cons ciousar ousal, $\ps i$ , channe l location, $c$, and ma ximummemor y l ength, $\ lam b da$ . Thus, s tati stical com ple xit y $C_ {\m u }$ is a f u nc tio n of t he s e t of para m et ers $ \{ \psi,c,\lam b d a \}$ foreac h fly,$f$. We are principal l y interested i n th e dif fer e ncein statistical c omp le x i ty overst ates of con scious a ro u sal $ $\begi n{gath ered} \ D e lt a C_{\m u} = C_ {\mu}^{\t ext {w a ke}} -C_ {\ m u}^{\t ext{ an a
LFPs_at each_electrode were recorded for_18s while_the_fly was_awake_and 18s more_after the fly_was anaesthetised (isoflurane, 0.6%_by volume, through_an_evaporator). Flies’ unresponsiveness during anaesthesia was confirmed by the absence of behavioural responses to_a_series of_air_puffs,_and recovery was also confirmed_after isoflurane gas was turned_off [@CohenEneuro2016]. We used_data sampled at 1kHz for the analysis [@CohenEneuro2016], and_to_obtain an estimate_of local neural activity, the 16 electrodes were re-referenced_by subtracting adjacent signals giving 15_channels which we_parametrise_as_$c \in [1,15]$. Line_noise was removed from the recordings,_followed by linear de-trending and removing_the mean. The resulting data is a_fluctuating voltage signal, which is time-binned_(1ms bins) and binarised by_splitting over_the median, leading to a_time series, see_Fig. \[fig:workflow\](b). For each_of the 13_flies in our data set, we_have 30 time_series of length $N = 18,000$._They_correspond to the_15_channels,_labelled numerically_from the central_to_peripheral region_as_depicted in Fig. \[fig:workflow\](a), and the two_states_of conscious arousal. Using the CSSR algorithm [@CSSR2],_we construct [$\epsilon\text{-machines}$]{} for_each_of these time series_as a function of maximum_memory length within the range $\lambda_\in [2,11]$._This is_below the memory length $L(N) \sim 14$ beyond which we would_be unable to reliably determine transition_probabilities for a sequence_of length_$N$_(see Sec. \[sec:Bkg-cssr\]) [^2]. We_record_the resulting_$3,900$ [$\epsilon\text{-machine}$]{} structures and their corresponding statistical_complexities, and_group them according to their respective_level of conscious arousal,_$\psi$,_channel location, $c$, and maximum memory_length, $\lambda$. Thus, statistical complexity $C_{\mu}$_is a function of the_set_of_parameters $\{\psi,c,\lambda\}$ for each fly,_$f$. We are principally interested in_the difference in_statistical complexity over states of conscious arousal_$$\begin{gathered} \Delta_C_{\mu} = C_{\mu}^{\text{wake}} - C_{\mu}^{\text{ana
\end{array}\right)V^{\mathrm{T}}.$$ Moreover, amplitudes (\[eq:explicit\_e(p)\]) fulfil the following conditions: \[seq:amplitudes\] $$\begin{gathered} p_\mu e^{\mu}_{\phantom{\mu}\sigma}(p) = 0, \label{eq:amplitudes_eq1} \\ e^{*\mu}_{\phantom{*\mu}\sigma}(p) e_{\mu\sigma^\prime}(p) = - \delta_{\sigma\sigma^\prime}, \label{eq:amplitudes_eq2} \\ e^{\mu}_{\phantom{\mu}\sigma}(p) e_{\mu\sigma^\prime}(p) = -(VV^{\mathrm{T}})_{\sigma\sigma^\prime}, \label{eq:amplitudes_eq3} \\ e^{*\mu}_{\phantom{*\mu}\sigma}(p) e^{\nu}_{\phantom{\nu}\sigma}(p) = - \eta^{\mu\nu}+\tfrac{p^{\mu}p^{\nu}}{m^2}, \label{eq:amplitudes_eq4} \end{gathered}$$ where $e(p)VV^{\mathrm{T}}=e^*(p)$, and $VV^{\mathrm{T}}=\left( \begin{array}{ccc} 0 & 0 & -1 \\ 0 & 1 & 0 \\ -1 & 0 & 0 \\ \end{array} \right)$. The covariant states (\[eq:covariant\_state\]) are normalized as follows \[c.f. Eq. (\[eq:spin\_basis\_normalization\])\] $$\label{eq:covariant_states_normaliztion} {\langle(\mu, p^\prime)|(\nu,p)\rangle} = 2p^0 \delta^3({\textbf}{p}^\prime-{\textbf}{p}) e^{*\mu}_{\phantom{*\mu}\sigma}(p^\prime) e^{\nu}_{\phantom{\nu}\sigma}(p).$$ Using helicity basis we have $${|(\mu,p)\rangle} = e^{\mu}_{\phantom{\mu}\sigma}(p) {\mathcal D}_{\lambda\sigma}(R_{p}^{-1})
\end{array}\right)V^{\mathrm{T}}.$$ Moreover, amplitudes (\[eq: explicit\_e(p)\ ]) fulfil the following condition: \[seq: amplitudes\ ] $ $ \begin{gathered } p_\mu e^{\mu}_{\phantom{\mu}\sigma}(p) = 0, \label{eq: amplitudes_eq1 } \\ e^{*\mu}_{\phantom{*\mu}\sigma}(p) e_{\mu\sigma^\prime}(p) = - \delta_{\sigma\sigma^\prime }, \label{eq: amplitudes_eq2 } \\ e^{\mu}_{\phantom{\mu}\sigma}(p) e_{\mu\sigma^\prime}(p) = -(VV^{\mathrm{T}})_{\sigma\sigma^\prime }, \label{eq: amplitudes_eq3 } \\ e^{*\mu}_{\phantom{*\mu}\sigma}(p) e^{\nu}_{\phantom{\nu}\sigma}(p) = - \eta^{\mu\nu}+\tfrac{p^{\mu}p^{\nu}}{m^2 }, \label{eq: amplitudes_eq4 } \end{gathered}$$ where $ e(p)VV^{\mathrm{T}}=e^*(p)$, and $ VV^{\mathrm{T}}=\left ( \begin{array}{ccc } 0 & 0 & -1 \\ 0 & 1 & 0 \\ -1 & 0 & 0 \\ \end{array } \right)$. The covariant state (\[eq: covariant\_state\ ]) are normalized as follows \[c.f.   Eq.   (\[eq: spin\_basis\_normalization\])\ ] $ $ \label{eq: covariant_states_normaliztion } { \langle(\mu, p^\prime)|(\nu, p)\rangle } = 2p^0 \delta^3({\textbf}{p}^\prime-{\textbf}{p }) e^{*\mu}_{\phantom{*\mu}\sigma}(p^\prime) e^{\nu}_{\phantom{\nu}\sigma}(p).$$ Using helicity footing we have $ $ { |(\mu, p)\rangle } = e^{\mu}_{\phantom{\mu}\sigma}(p) { \mathcal D}_{\lambda\sigma}(R_{p}^{-1 } )
\fnd{array}\right)V^{\mathrm{T}}.$$ Moveover, amplitudes (\[eq:expnicit\_e(l)\]) fulfil the following conditions: \[seq:emplutudew\] $$\begin{gathered} p_\mu e^{\mu}_{\phantlm{\mu}\sigmq}(p) = 0, \label{eq:amplitudes_eq1} \\ e^{*\mu}_{\'hentom{*\mu}\sigma}(p) e_{\ku\sigma^\prike}(p) = - \delta_{\viemc\sigma^\prime}, \label{eq:amplitudes_eq2} \\ e^{\mu}_{\phsnhom{\mu}\sigma}(p) e_{\mo\sigms^\[rims}(i) = -(VV^{\mathrm{T}})_{\sigma\sigma^\prime}, \laben{eq:amplitudes_rq3} \\ e^{*\mu}_{\phantom{*\mu}\sigma}(p) e^{\ju}_{\phantom{\nu}\sigma}(p) = - \eta^{\mu\no}+\ffrwx{p^{\mu}p^{\nu}}{m^2}, \lxbel{eq:amplpcudes_eq4} \ehd{gathered}$$ where $e(p)VV^{\mathrm{T}}=e^*(p)$, avd $VV^{\kathrm{T}}=\lefj( \begin{drray}{ccv} 0 & 0 & -1 \\ 0 & 1 & 0 \\ -1 & 0 & 0 \\ \tnd{wgtay} \right)$. The covariant statts (\[eq:dovariant\_state\]) are normqlized as follows \[c.f. Ee. (\[eq:spin\_bafis\_normalization\])\] $$\label{eq:covariant_states_normaliztimn} {\uanyoe(\mu, o^\prlme)|(\nu,p)\rangle} = 2p^0 \delta^3({\textbf}{p}^\prime-{\textbf}{p}) e^{*\ku}_{\ihantom{*\mu}\sigma}(p^\prlme) e^{\nu}_{\phantom{\ni}\slgkw}(p).$$ Using helizity bcaia we have $${|(\mu,p)\ranglf} = e^{\mu}_{\pranton{\mu}\sigma}(p) {\kathcal D}_{\lambda\sigma}(R_{p}^{-1})
\end{array}\right)V^{\mathrm{T}}.$$ Moreover, amplitudes (\[eq:explicit\_e(p)\]) fulfil the following $$\begin{gathered} e^{\mu}_{\phantom{\mu}\sigma}(p) = \label{eq:amplitudes_eq1} \\ e^{*\mu}_{\phantom{*\mu}\sigma}(p) \\ e_{\mu\sigma^\prime}(p) = -(VV^{\mathrm{T}})_{\sigma\sigma^\prime}, \\ e^{*\mu}_{\phantom{*\mu}\sigma}(p) e^{\nu}_{\phantom{\nu}\sigma}(p) - \eta^{\mu\nu}+\tfrac{p^{\mu}p^{\nu}}{m^2}, \label{eq:amplitudes_eq4} \end{gathered}$$ where $e(p)VV^{\mathrm{T}}=e^*(p)$, $VV^{\mathrm{T}}=\left( \begin{array}{ccc} 0 & 0 & -1 \\ 0 & 1 & 0 -1 & 0 & 0 \\ \end{array} \right)$. The covariant states (\[eq:covariant\_state\]) are as \[c.f. (\[eq:spin\_basis\_normalization\])\] {\langle(\mu, p^\prime)|(\nu,p)\rangle} = 2p^0 \delta^3({\textbf}{p}^\prime-{\textbf}{p}) e^{*\mu}_{\phantom{*\mu}\sigma}(p^\prime) e^{\nu}_{\phantom{\nu}\sigma}(p).$$ Using helicity basis we have $${|(\mu,p)\rangle} = e^{\mu}_{\phantom{\mu}\sigma}(p) {\mathcal
\end{array}\right)V^{\mathrm{T}}.$$ MoreOver, amplitUdes (\[eQ:exPliCiT\_e(p)\]) fUlfiL the following cONditIons: \[seq:amplitudes\] $$\begin{GatheReD} P_\mu e^{\MU}_{\pHantoM{\mu}\sigmA}(P) = 0, \lABEl{eQ:aMpLitUdES_eQ1} \\ e^{*\mu}_{\pHanTom{*\mu}\siGma}(p) e_{\mu\sigMa^\pRiMe}(p) = - \delta_{\sigmA\SiGma^\prime}, \laBel{Eq:amplitudes_Eq2} \\ e^{\Mu}_{\phanToM{\mu}\SIgma}(p) E_{\mu\Sigma^\Prime}(p) = -(vv^{\mathrM{T}})_{\sigma\siGmA^\Prime}, \lABel{eq:amPLItUdes_Eq3} \\ e^{*\mu}_{\phantom{*\mu}\sigMA}(p) E^{\Nu}_{\phantom{\nu}\sigMa}(p) = - \eta^{\Mu\NU}+\tFRAc{p^{\Mu}p^{\Nu}}{m^2}, \label{eq:AmPlituDEs_eq4} \end{GAtHEREd}$$ wHEre $e(p)VV^{\mathrm{t}}=e^*(p)$, and $VV^{\matHRm{T}}=\Left( \beGiN{arRAy}{ccc} 0 & 0 & -1 \\ 0 & 1 & 0 \\ -1 & 0 & 0 \\ \eNd{arrAy} \RIghT)$. The covariaNt stAtes (\[eq:covAriant\_STate\]) are NOrmalizEd as foLloWs \[c.F. Eq. (\[eQ:SpIn\_BasIs\_NOrmALiZatIOn\])\] $$\lAbel{eq:coVaRiAnt_stAtes_NORMAlizTioN} {\lanGle(\mu, P^\prime)|(\nu,p)\rangLe} = 2p^0 \DeltA^3({\TexTbf}{p}^\pRime-{\tExtbF}{p}) E^{*\mu}_{\phAntom{*\mU}\sigmA}(p^\Prime) e^{\nu}_{\phantom{\Nu}\siGma}(p).$$ Using HelIcIty BaSis we HAve $${|(\mu,p)\RanGle} = E^{\mu}_{\phanTom{\mu}\siGMa}(p) {\MaTHCAl d}_{\lambda\sigma}(R_{p}^{-1})
\end{array}\right)V^{\ mathrm{T}} .$$ M ore ove r, amp litu des (\[eq:expl i cit\ _e(p)\]) fulfil the fo llowi ng cond i ti ons: \[seq: a mp l i tud es \] $$ \b e gi n{gat her ed} p_\mu e^{ \mu }_ {\phantom{\m u }\ sigma}(p)= 0 , \label{ eq: amplit ud es_ e q1} \ \ e^ {*\mu} _ {\phan tom{*\mu} \s i gma}(p ) e_{\mu \ s ig ma^\ prime}(p) = - \d e lta_{\sigma\si gma^\p ri m e} , \la bel{eq:amp li tudes _ eq2} \\ e ^{\ m u}_{\phantom{ \mu}\sigma} ( p)e_{\mu \s igm a ^\prim e}(p) = -(VV^{\ma thrm {T}})_{\s igma\s i gma^\pr i me}, \labe l{e q:a mpli t ud es _eq 3} \\ e^ { *\m u}_{\pha nt om {*\mu }\si g m a } (p) e^{ \nu}_ {\phantom{\nu }\s igma } (p) = - \ eta^ {\ mu\nu }+\tfr ac{p^ {\ mu}p^{\nu}}{m^2 }, \label{ eq: am pli tu des_e q 4} \en d{g athered }$$ wh e re$e ( p ) VV ^{\mathrm{T}}=e^*( p) $ , a nd $VV^{ \mathr m {T }} = \left( \be g in {a rray}{c cc } 0 &0 & -1 \\ 0 & 1 &0 \\ -1 & 0 & 0 \\ \ e n d{array} \ri g ht)$. T he co variantstates (\ [ e q:covari ant \_s tat e\] ) ar e normalizeda s fol lo ws \[c. f.Eq. (\[ eq: spi n\_ bas is \_normali zation\] )\ ]$$ \l abe l{eq: c ovariant _s tat es _no rmali z tion} { \lan gl e( \ mu, p^\pri m e) | ( \nu, p) \r angl e}=2p^0 \ del ta^3({\ textbf}{p }^\ p rime -{ \t extbf}{ p}) e^{*\ mu }_{\phanto m{ *\m u}\sig m a }(p^\pri me) e^{\nu}_{\phant o m{\nu}\ sig ma}(p ).$$ Using he lic ity ba sis we hav e $${| (\mu, p) \ra n g le} = e ^{ \mu }_ {\phantom{ \ m u}\ sigma }( p) {\ma thcal D}_{\lambda\ s igm a}(R_{p}^{-1} )
_ \end{array}\right)V^{\mathrm{T}}.$$_Moreover, amplitudes (\[eq:explicit\_e(p)\]) fulfil_the following_conditions: \[seq:amplitudes\]_$$\begin{gathered} __ p_\mu e^{\mu}_{\phantom{\mu}\sigma}(p)_= 0, _ \label{eq:amplitudes_eq1} \\ _ e^{*\mu}_{\phantom{*\mu}\sigma}(p)_e_{\mu\sigma^\prime}(p)_= - \delta_{\sigma\sigma^\prime}, \label{eq:amplitudes_eq2} \\ e^{\mu}_{\phantom{\mu}\sigma}(p)_e_{\mu\sigma^\prime}(p)_= ___ -(VV^{\mathrm{T}})_{\sigma\sigma^\prime}, \label{eq:amplitudes_eq3}_\\ e^{*\mu}_{\phantom{*\mu}\sigma}(p) _ _ e^{\nu}_{\phantom{\nu}\sigma}(p) = - \eta^{\mu\nu}+\tfrac{p^{\mu}p^{\nu}}{m^2}, __ \label{eq:amplitudes_eq4} _ \end{gathered}$$ where $e(p)VV^{\mathrm{T}}=e^*(p)$, and $VV^{\mathrm{T}}=\left( _ _ ___ _ _ _ \begin{array}{ccc} _ _ _ _ _ _ _ _ _ 0 &_0 & -1 \\ __ ___ _ __ __ __ _ __0 & 1 &_0 \\ _ _ _ _ _ _ _ _-1_& 0 &_0_\\ _ _ _ _ __ _ _ \end{array} ___ _ _ _ __ \right)$. The_covariant_states (\[eq:covariant\_state\]) are normalized as follows_\[c.f. Eq. (\[eq:spin\_basis\_normalization\])\]_$$\label{eq:covariant_states_normaliztion} _ {\langle(\mu, p^\prime)|(\nu,p)\rangle}_= 2p^0 \delta^3({\textbf}{p}^\prime-{\textbf}{p}) _ e^{*\mu}_{\phantom{*\mu}\sigma}(p^\prime) e^{\nu}_{\phantom{\nu}\sigma}(p).$$ Using_helicity basis we_have $${|(\mu,p)\rangle} = e^{\mu}_{\phantom{\mu}\sigma}(p) ___{\mathcal D}_{\lambda\sigma}(R_{p}^{-1})
{aligned}$$ By inverting relation (\[z\_of\_f\]) one obtains an explicit equation for the amplitude of the order parameter as a function of position along the wire (see Fig. \[u\_of\_x\]): $$\begin{aligned} f^2(z) &= u_0 + u_1 \sin^2 \big[\text{JacobiAmplitude}\big[z \sqrt{\frac{u_2}{2}},\frac{u_1}{u_2}\big]\big]\\ &=u_0+u_1 \text{JacobiSn}^2\big[z \sqrt{\frac{u_2}{2}},\frac{u_1}{u_2}\big] \label{f_of_z}\end{aligned}$$ ![Squared amplitude $u$ of the order parameter as a function of position along the wire for the two types of solution: metastable and saddle point. []{data-label="u_of_x"}](\figdir/u_of_x2c.eps){width="7cm"} The end-to-end phase difference along the wire may be found by using the current conservation law. Thus one obtains $$\theta=\int_{-b/2}^{b/2} \frac{J}{f^2(z)} dz = 2 J \int_0^{b/2} \frac{dz}{ u_0+u_1 \text{JacobiSn}^2\big[z \sqrt{\frac{u_2}{2}},\frac{u_1}{u_2}\big]}. \label{etoe}$$ The Helmholtz free energy can be found by substituting the expressions for $f(z)$ and $\phi'(z)$ into the expression for the free energy. One then obtains $$F=4 \Ecore \int_0^{b/2} dz (\frac{1}{2}-2 f^2+f^4+J^2/u_0+u_0-u_0^2/2), \label{fexact}$$ where $E$ was expressed in terms of $u_0$. provide expressions for $\theta$ and $F$ which are true regardless of the length of the wire, and therefore may be used as a starting point for computing the Gibbs free-energy of the various metast
{ aligned}$$ By inverting relation (\[z\_of\_f\ ]) one obtains an explicit equality for the amplitude of the decree parameter as a function of placement along the telegram (see Fig.   \[u\_of\_x\ ] ): $ $ \begin{aligned } f^2(z) & = u_0 + u_1 \sin^2 \big[\text{JacobiAmplitude}\big[z \sqrt{\frac{u_2}{2}},\frac{u_1}{u_2}\big]\big]\\ & = u_0+u_1 \text{JacobiSn}^2\big[z \sqrt{\frac{u_2}{2}},\frac{u_1}{u_2}\big ] \label{f_of_z}\end{aligned}$$ ! [ Squared amplitude $ u$ of the order parameter as a affair of position along the wire for the two type of solution: metastable and saddle point. [ ] { data - label="u_of_x"}](\figdir / u_of_x2c.eps){width="7 cm " } The conclusion - to - end phase difference along the telegram may be found by using the current conservation law. Thus one obtains $ $ \theta=\int_{-b/2}^{b/2 } \frac{J}{f^2(z) } dz = 2 J \int_0^{b/2 } \frac{dz } { u_0+u_1 \text{JacobiSn}^2\big[z \sqrt{\frac{u_2}{2}},\frac{u_1}{u_2}\big ] }. \label{etoe}$$ The Helmholtz free energy can be rule by substituting the expressions for $ f(z)$ and $ \phi'(z)$ into the construction for the free energy. One then obtains $ $ F=4 \Ecore \int_0^{b/2 } dz (\frac{1}{2}-2 f^2+f^4+J^2 / u_0+u_0 - u_0 ^ 2/2), \label{fexact}$$ where $ E$ was expressed in footing of $ u_0$. provide expressions for $ \theta$ and $ F$ which are true regardless of the length of the wire, and therefore may be use as a starting point for computing the Gibbs free - energy of the diverse metast
{alihned}$$ By inverting relatiok (\[z\_of\_f\]) one obtaiuw an eeplicit equatiov for the amplitude of the ocder paraneter as a function of position along tye wmre (see Fig. \[u\_of\_x\]): $$\bxfin{aligkzd} f^2(z) &= m_0 + u_1 \vmn^2 \big[\text{JacoblAmplitude}\bhg[z \sqrt{\frac{u_2}{2}},\frdc{j_1}{u_2}\yig]\big]\\ &=u_0+u_1 \text{JacobiSn}^2\big[z \sqrt{\frac{u_2}{2}},\frwc{u_1}{u_2}\big] \lwbel{f_of_z}\end{alidned}$$ ![Xzuarsd amplitude $u$ of the order paramefer as e function of ppsition along the wire for the two types of soluhion: metastqble qnd saddle puint. []{data-label="u_of_x"}](\figdjr/u_of_x2c.eps){width="7cm"} The end-to-end pfase bifference qling jhe wire may be fjund by usinn the cgrrent vonservation lsw. Vhus one obtains $$\theta=\int_{-u/2}^{b/2} \frac{J}{f^2(z)} dz = 2 J \inj_0^{b/2} \frac{dz}{ u_0+g_1 \cext{JacobiSn}^2\big[z \sqrt{\feax{u_2}{2}},\fraw{u_1}{u_2}\bhg]}. \lacwl{egoe}$$ Uhe Hslmholhz hree energy can be foubd by substituting uhe vcpressions fkr $f(z)$ wnq $\phi'(z)$ into the expression for the free ensrgy. One then obtains $$F=4 \Ecore \int_0^{b/2} dz (\frac{1}{2}-2 f^2+f^4+U^2/u_0+u_0-u_0^2/2), \label{sexact}$$ where $E$ was expressed in terms of $u_0$. provida expcersiinf dog $\theta$ and $F$ which are true regardless of thq ltngnh of the wire, anb therefore may ne uxgd as a startivg poiuf ror computing the Hibbs ftee-eneegy of tht varoous metast
{aligned}$$ By inverting relation (\[z\_of\_f\]) one obtains equation the amplitude the order parameter along wire (see Fig. $$\begin{aligned} f^2(z) &= + u_1 \sin^2 \big[\text{JacobiAmplitude}\big[z \sqrt{\frac{u_2}{2}},\frac{u_1}{u_2}\big]\big]\\ &=u_0+u_1 \sqrt{\frac{u_2}{2}},\frac{u_1}{u_2}\big] \label{f_of_z}\end{aligned}$$ ![Squared amplitude $u$ of the order parameter as a function of along the wire for the two types of solution: metastable and saddle point. The phase along wire may be found by using the current conservation law. Thus one obtains $$\theta=\int_{-b/2}^{b/2} \frac{J}{f^2(z)} dz 2 J \int_0^{b/2} \frac{dz}{ u_0+u_1 \text{JacobiSn}^2\big[z \sqrt{\frac{u_2}{2}},\frac{u_1}{u_2}\big]}. \label{etoe}$$ Helmholtz free energy can found by substituting the expressions $f(z)$ $\phi'(z)$ into expression the energy. One then $$F=4 \Ecore \int_0^{b/2} dz (\frac{1}{2}-2 f^2+f^4+J^2/u_0+u_0-u_0^2/2), \label{fexact}$$ where $E$ was expressed in terms of $u_0$. provide expressions $\theta$ and are true of length the wire, and be used as a starting point Gibbs free-energy of the various metast
{aligned}$$ By inverting relatioN (\[z\_of\_f\]) one obTains An eXplIcIt eqUatiOn for the amplitUDe of The order parameter as a fuNctioN oF PosiTIoN alonG the wirE (SeE fIg. \[u\_Of\_X\]): $$\bEgiN{aLIgNed} f^2(z) &= U_0 + u_1 \sIn^2 \big[\teXt{JacobiAmPliTuDe}\big[z \sqrt{\frAC{u_2}{2}},\Frac{u_1}{u_2}\big]\bIg]\\ &=u_0+U_1 \text{JacobiSN}^2\biG[z \sqrt{\FrAc{u_2}{2}},\FRac{u_1}{u_2}\Big] \Label{F_of_z}\enD{AligneD}$$ ![Squared aMpLItude $u$ OF the ordER PaRameTer as a function of pOSiTIon along the wirE for thE tWO tYPEs oF soLution: metaStAble aND saddle POiNT. []{DAta-LAbel="u_of_x"}](\figdiR/u_of_x2c.eps){wiDTh="7cM"} The enD-tO-enD Phase dIfferEnCE alOng the wire mAy be Found by usIng the CUrrent cONservatIon law. thuS onE obtAInS $$\tHetA=\iNT_{-b/2}^{b/2} \FRaC{J}{f^2(Z)} Dz = 2 J \Int_0^{b/2} \frac{Dz}{ U_0+u_1 \Text{JAcobIsN}^2\BIg[z \sQrt{\Frac{U_2}{2}},\frac{U_1}{u_2}\big]}. \label{etoE}$$ ThE HelMHolTz freE enerGy caN bE founD by subStituTiNg the expressionS for $F(z)$ and $\phi'(z)$ IntO tHe eXpRessiON for thE frEe eNergy. OnE then obTAinS $$F=4 \eCORe \Int_0^{b/2} dz (\frac{1}{2}-2 f^2+f^4+J^2/u_0+u_0-u_0^2/2), \laBeL{FExAct}$$ where $e$ was exPReSsED in terms Of $U_0$. prOvidE EXpresSionS FoR $\theta$ anD $F$ whicH ArE tRue regaRdLess of ThE leNgtH of thE Wire, And theRefore maY be usED as a starting poINt for computinG ThE gIbBS freE-enErgy of the vaRiouS MetaSt
{aligned}$$ By invertingrelation ( \[z\_ of\ _f\ ]) one obt ains an explic i t eq uation for the amplitu de of t h e or d er para meter a s a f unc ti on of p o si tionalo ng thewire (seeFig .\[u\_of\_x\] ) : $$\begin{ ali gned} f^2(z) &= u_0 + u _1\ sin^2 \b ig[\t ext{Ja c obiAmp litude}\b ig [ z \sqr t {\frac{ u _ 2} {2}} ,\frac{u_1}{u_2}\ b ig ] \big]\\ &=u_0+ u_1 \t ex t {J a c obi Sn} ^2\big[z \ sq rt{\f r ac{u_2} { 2} } , \ fra c {u_1}{u_2}\bi g] \label{f _ of_ z}\end {a lig n ed}$$ ![Sq ua r edamplitude $ u$ o f the ord er par a meter a s a func tion o f p osi tion al on g t he wir e f ort hetwo type sof solu tion : m e tast abl e an d sad dle point. [] {da ta-l a bel ="u_o f_x"} ](\f ig dir/u _of_x2 c.eps ){ width="7cm"} T he e nd-to-end ph as e d if feren c e alon g t hewire ma y be fo u ndby u s in g the current cons er v a ti on law.Thus o n eob t ains $$\ th eta =\in t _ {-b/2 }^{b / 2} \frac{J }{f^2( z )} d z = 2 J \ int_0^ {b /2} \f rac{d z }{ u _0+u_1 \text{J acobi S n}^2\big[z \sq r t{\frac{u_2}{ 2 }} , \ fr a c{u_ 1}{ u_2}\big]}. \la b el{e toe} $ $ Th e Helm holtz f r ee energy can be found b y subs titut ing the expre ssions for $ f (z)$ and $\p h i' ( z)$ into the e xpres sion for t h e free e nergy . One th en obtain s $$F=4 \E cor e \ int _0^ { b /2 } dz (\frac{1 } { 2}-2 f ^2+f^4+ J^2 /u_0+u_ 0-u _0^ 2/2 ),\l abel{fexa ct}$$ wh er e$E $was expr e ssed inte rms o f $ u_0$. provid e exp ress io ns for $\thet a $a n d $F $wh ichare t rue r egar d les s of th e lengthoft he w ir e, and th erefore may b eused as ast art ing po i n t for co mputing the Gibbs free- e nergy o f t he va riou s metast
{aligned}$$ By inverting_relation (\[z\_of\_f\])_one obtains an explicit_equation for_the_amplitude of_the_order parameter as_a function of_position along the wire_(see Fig. \[u\_of\_x\]): $$\begin{aligned} f^2(z) &=_u_0_+ u_1 \sin^2 \big[\text{JacobiAmplitude}\big[z \sqrt{\frac{u_2}{2}},\frac{u_1}{u_2}\big]\big]\\ &=u_0+u_1 \text{JacobiSn}^2\big[z \sqrt{\frac{u_2}{2}},\frac{u_1}{u_2}\big] \label{f_of_z}\end{aligned}$$ ![Squared amplitude $u$ of the order parameter as a_function_of position_along_the_wire for the two types_of solution: metastable and saddle_point. []{data-label="u_of_x"}](\figdir/u_of_x2c.eps){width="7cm"} The_end-to-end phase difference along the wire may be_found_by using the_current conservation law. Thus one obtains $$\theta=\int_{-b/2}^{b/2} \frac{J}{f^2(z)} dz_= 2 J \int_0^{b/2} \frac{dz}{ u_0+u_1 \text{JacobiSn}^2\big[z \sqrt{\frac{u_2}{2}},\frac{u_1}{u_2}\big]}. \label{etoe}$$ The_Helmholtz free energy_can_be_found by substituting the_expressions for $f(z)$ and $\phi'(z)$ into_the expression for the free energy._One then obtains $$F=4 \Ecore \int_0^{b/2} dz (\frac{1}{2}-2_f^2+f^4+J^2/u_0+u_0-u_0^2/2), \label{fexact}$$ where $E$ was expressed in_terms of $u_0$. provide expressions_for $\theta$_and $F$ which are true_regardless of the_length of_the wire, and_therefore may be used as a_starting point for_computing the Gibbs free-energy of the_various_metast
we arrive at the bound $$\label{eqn:gm-am-temp2} \begin{aligned} {\operatorname{tr}}\big[ {{\bm{H}}}{{\bm{W}}}^q {{\bm{H}}} {{\bm{Y}}}^{2r-q} \big] + {\operatorname{tr}}\big[ {{\bm{H}}}{{\bm{W}}}^{2r-q} {{\bm{H}}} {{\bm{Y}}}^{q} \big] &\leq \sum_{i,j=1}^d \big( {\left\vert {\lambda_i} \right\vert}^q {\left\vert {\smash{\mu_j}} \right\vert}^{2r-q} + {\left\vert {\lambda_i} \right\vert}^{2r-q} {\left\vert {\smash{\mu_j}} \right\vert}^{q}\big) \cdot {{{\left\vert { {{\bm{u}}}_i^{*}{{\bm{H}}} {{\bm{v}}}_j } \right\vert}}^2} \\ &\leq \sum_{i,j=1}^d \big( \lambda_i^{2r} + \mu_j^{2r} \big) \cdot {{{\left\vert { {{\bm{u}}}_i^{*}{{\bm{H}}} {{\bm{v}}}_j } \right\vert}}^2}. \end{aligned}$$ The second inequality is , with $\theta = q/(2r)$ and $\lambda = \lambda_i^{2r}$ and $\mu = \mu_j^{2r}$. It remains to rewrite the right-hand side of  a more recognizable form. To that end, observe that $$\begin{aligned} {\operatorname{tr}}\big[ {{\bm{H}}}{{\bm{W}}}^q {{\bm{H}}} {{\bm{Y}}}^{2r-q} \big] &+ {\operatorname{tr}}\big[ {{\bm{H}}}{{\bm{W}}}^{2r-q} {{\bm{H}}} {{\bm{Y}}}^{q} \big] \\ &\leq \sum_{i,j=1}^d \big( \lambda_i^{2r} + \mu_j^{2r} \big) \cdot {\operatorname{tr}}\big[ {{\bm{H}}} {{\bm{u
we arrive at the bound $ $ \label{eqn: gm - am - temp2 } \begin{aligned } { \operatorname{tr}}\big [ { { \bm{H}}}{{\bm{W}}}^q { { \bm{H } } } { { \bm{Y}}}^{2r - q } \big ] + { \operatorname{tr}}\big [ { { \bm{H}}}{{\bm{W}}}^{2r - q } { { \bm{H } } } { { \bm{Y}}}^{q } \big ] & \leq \sum_{i, j=1}^d \big ({ \left\vert { \lambda_i } \right\vert}^q { \left\vert { \smash{\mu_j } } \right\vert}^{2r - q } + { \left\vert { \lambda_i } \right\vert}^{2r - q } { \left\vert { \smash{\mu_j } } \right\vert}^{q}\big) \cdot { { { \left\vert { { { \bm{u}}}_i^{*}{{\bm{H } } } { { \bm{v}}}_j } \right\vert}}^2 } \\ & \leq \sum_{i, j=1}^d \big (\lambda_i^{2r } + \mu_j^{2r } \big) \cdot { { { \left\vert { { { \bm{u}}}_i^{*}{{\bm{H } } } { { \bm{v}}}_j } \right\vert}}^2 }. \end{aligned}$$ The second inequality is  , with $ \theta = q/(2r)$ and $ \lambda = \lambda_i^{2r}$ and $ \mu = \mu_j^{2r}$. It remains to rewrite the right - bridge player slope of   a more recognizable form. To that end, observe that $ $ \begin{aligned } { \operatorname{tr}}\big [ { { \bm{H}}}{{\bm{W}}}^q { { \bm{H } } } { { \bm{Y}}}^{2r - q } \big ] & + { \operatorname{tr}}\big [ { { \bm{H}}}{{\bm{W}}}^{2r - q } { { \bm{H } } } { { \bm{Y}}}^{q } \big ] \\ & \leq \sum_{i, j=1}^d \big (\lambda_i^{2r } + \mu_j^{2r } \big) \cdot { \operatorname{tr}}\big [ { { \bm{H } } } { { \bm{u
we arrive at the bound $$\labtl{eqn:gm-am-temp2} \begiu{qlignev} {\operatkrname{tr}}\cig[ {{\bm{H}}}{{\bm{W}}}^q {{\bm{H}}} {{\bm{Y}}}^{2r-q} \big] + {\opxratirnamt{nr}}\big[ {{\bm{H}}}{{\bm{W}}}^{2r-q} {{\bm{H}}} {{\bm{H}}}^{q} \big] &\leq \sum_{u,j=1}^d \uig( {\left\vert {\lamusa_i} \rigmc\vert}^s {\lefc\vxrt {\smash{\mu_j}} \rinht\vert}^{2r-q} + {\left\vert {\lakbaa_n} \right\vert}^{2r-q} {\left\vert {\smash{\mu_j}} \righe\vert}^{q}\bog) \cdot {{{\left\vgrt { {{\ni{u}}}_i^{*}{{\bj{H}}} {{\bm{v}}}_j } \right\vert}}^2} \\ &\leq \sum_{i,j=1}^d \gig( \lamuda_i^{2r} + \mu_j^{2r} \big) \cdot {{{\left\vert { {{\bm{u}}}_i^{*}{{\bm{H}}} {{\bm{v}}}_u } \right\vert}}^2}. \end{alihned}$$ The sexond unequality ir , with $\theua = q/(2r)$ and $\lzmbda = \lambda_i^{2r}$ and $\mu = \mu_j^{2r}$. It femaius to rewrijz thf right-hand wide jf  a more regpgnizatle fork. To that end, pbsxrve that $$\begin{aligned} {\opecatorname{tr}}\big[ {{\bm{H}}}{{\bm{R}}}^q {{\bm{H}}} {{\bm{F}}}^{2r-s} \big] &+ {\operatornamw{te}}\big[ {{\tm{H}}}{{\bk{W}}}^{2r-q} {{\vm{H}}} {{\bm{G}}}^{q} \bjg] \\ &\les \sum_{i,j=1}^d \bjg( \lambda_i^{2r} + \mu_j^{2r} \big) \cdot {\ppqgstorname{tr}}\bif[ {{\bm{H}}} {{\fm{t
we arrive at the bound $$\label{eqn:gm-am-temp2} \begin{aligned} {{\bm{H}}} \big] + {{\bm{H}}}{{\bm{W}}}^{2r-q} {{\bm{H}}} {{\bm{Y}}}^{q} {\lambda_i} {\left\vert {\smash{\mu_j}} \right\vert}^{2r-q} {\left\vert {\lambda_i} \right\vert}^{2r-q} {\smash{\mu_j}} \right\vert}^{q}\big) \cdot {{{\left\vert { {{\bm{u}}}_i^{*}{{\bm{H}}} } \right\vert}}^2} \\ &\leq \sum_{i,j=1}^d \big( \lambda_i^{2r} + \mu_j^{2r} \big) \cdot {{{\left\vert { {{\bm{v}}}_j } \right\vert}}^2}. \end{aligned}$$ The second inequality is , with $\theta = q/(2r)$ $\lambda \lambda_i^{2r}$ $\mu \mu_j^{2r}$. It remains to rewrite the right-hand side of a more recognizable form. To that end, that $$\begin{aligned} {\operatorname{tr}}\big[ {{\bm{H}}}{{\bm{W}}}^q {{\bm{H}}} {{\bm{Y}}}^{2r-q} \big] &+ {{\bm{H}}}{{\bm{W}}}^{2r-q} {{\bm{H}}} {{\bm{Y}}}^{q} \big] &\leq \sum_{i,j=1}^d \big( \lambda_i^{2r} + \big) {\operatorname{tr}}\big[ {{\bm{H}}}
we arrive at the bound $$\label{eqN:gm-am-temp2} \bEgin{aLigNed} {\OpEratOrnaMe{tr}}\big[ {{\bm{H}}}{{\bm{W}}}^q {{\BM{H}}} {{\bm{y}}}^{2r-q} \big] + {\operatorname{tr}}\biG[ {{\bm{H}}}{{\bM{W}}}^{2R-Q} {{\bm{H}}} {{\BM{Y}}}^{Q} \big] &\lEq \sum_{i,j=1}^D \BiG( {\LEft\VeRt {\LamBdA_I} \rIght\vErt}^Q {\left\veRt {\smash{\mu_j}} \RigHt\Vert}^{2r-q} + {\left\veRT {\lAmbda_i} \righT\veRt}^{2r-q} {\left\vert {\SmaSh{\mu_j}} \rIgHt\vERt}^{q}\biG) \cdOt {{{\lefT\vert { {{\bM{U}}}_i^{*}{{\bm{H}}} {{\bM{v}}}_j } \right\vErT}}^2} \\ &\Leq \sum_{I,J=1}^d \big( \laMBDa_I^{2r} + \mu_J^{2r} \big) \cdot {{{\left\vert { {{\BM{u}}}_I^{*}{{\Bm{H}}} {{\bm{v}}}_j } \right\veRt}}^2}. \end{aLiGNeD}$$ tHe sEcoNd inequaliTy Is , witH $\Theta = q/(2r)$ ANd $\LAMBda = \LAmbda_i^{2r}$ and $\mu = \mU_j^{2r}$. It remainS To rEwrite ThE riGHt-hand Side oF  a MOre RecognizablE forM. To that enD, obserVE that $$\beGIn{alignEd} {\operAtoRnaMe{tr}}\BIg[ {{\Bm{h}}}{{\bm{w}}}^q {{\BM{H}}} {{\bM{y}}}^{2r-Q} \biG] &+ {\OpeRatornamE{tR}}\bIg[ {{\bm{H}}}{{\Bm{W}}}^{2r-Q} {{\BM{h}}} {{\Bm{Y}}}^{q} \Big] \\ &\Leq \sUm_{i,j=1}^d \Big( \lambda_i^{2r} + \mu_J^{2r} \bIg) \cdOT {\opEratoRname{Tr}}\biG[ {{\bM{H}}} {{\bm{u
we arrive at the bound $$ \label{eqn :gm-a m-t emp 2} \be gin{ aligned} {\ope r ator name{tr}}\big[ {{\bm{H }}}{{ \b m {W}} } ^q {{\b m{H}}}{ {\ b m {Y} }} ^{ 2r- q} \b ig] + {\ operato rname{tr}} \bi g[ {{\bm{H}}}{ { \b m{W}}}^{2r -q} {{\bm{H}}}{{\ bm{Y}} }^ {q} \big] &\l eq \su m _{i,j= 1}^d \big ({ \left\ v ert {\l a m bd a_i} \right\vert}^q { \ le f t\vert {\smash {\mu_j }} \r i g ht\ ver t}^{2r-q} + { \ left\ve r t{ \ l amb d a_i} \right\v ert}^{2r-q} {\l eft\ve rt {\ s mash{\ mu_j} }\ rig ht\vert}^{q }\bi g) \c dot {{ { \left\v e rt { {{ \bm{u} }}_ i^{ *}{{ \ bm {H }}} { { \bm { v} }}_ j }\right\v er t} }^2}\\ & \leq \s um_{ i,j=1 }^d \big( \la mbd a_i^ { 2r} + \m u_j^{ 2r}\b ig) \cd ot {{ {\ left\vert { {{\ bm{u }}}_i^{*} {{\ bm {H} }} {{\b m {v}}}_ j } \r ight\ve rt}}^2} . \e nd { a l ig ned}$$ The secondin e q ua lity is, with $\ th e ta = q/( 2r )$and$ \ lambd a =\ la mbda_i^{ 2r}$ a n d$\ mu = \m u_ j^{2r} $. I t r emain s torewrit e the ri ght-h a nd side of  am ore recogniza b le f or m . To th at end, obs erve that $$\ b eg in{ a ligne d} {\ op e ra t orname{tr}}\big[ {{ \b m{H}}} {{\bm {W}}}^q {{\bm {H}}} {{\b m { Y }}}^{2r- q} \ b ig ] &+ {\operator name{ tr}}\big[{ {\bm{H}} }{{\b m{W}}}^{ 2r-q} {{\ b m {H}}} {{ \bm {Y} }}^ {q} \ bi g] \\ &\l e q \su m_ {i,j=1} ^d\big( \ lam bda _i^ {2r }+ \mu_j^{ 2r} \big ) \c dot {\op e ratornam e{ tr} }\ big [ {{\ b m{H}}} {{\b m{u
we_arrive at_the bound $$\label{eqn:gm-am-temp2} \begin{aligned} {\operatorname{tr}}\big[ {{\bm{H}}}{{\bm{W}}}^q_{{\bm{H}}} {{\bm{Y}}}^{2r-q}_\big] +_{\operatorname{tr}}\big[ {{\bm{H}}}{{\bm{W}}}^{2r-q}_{{\bm{H}}}_{{\bm{Y}}}^{q} \big] _ &\leq_\sum_{i,j=1}^d \big( {\left\vert {\lambda_i}_\right\vert}^q {\left\vert {\smash{\mu_j}}_\right\vert}^{2r-q} _ + {\left\vert {\lambda_i} \right\vert}^{2r-q} {\left\vert {\smash{\mu_j}} \right\vert}^{q}\big) \cdot_{{{\left\vert_{ {{\bm{u}}}_i^{*}{{\bm{H}}}_{{\bm{v}}}_j_}_\right\vert}}^2} \\ _&\leq \sum_{i,j=1}^d \big( \lambda_i^{2r} +_\mu_j^{2r} \big) _ \cdot {{{\left\vert { {{\bm{u}}}_i^{*}{{\bm{H}}} {{\bm{v}}}_j_}_\right\vert}}^2}. \end{aligned}$$ The second_inequality is , with $\theta = q/(2r)$ and $\lambda =_\lambda_i^{2r}$ and $\mu = \mu_j^{2r}$. It remains_to rewrite the_right-hand_side_of  a more recognizable_form. To that end, observe that_$$\begin{aligned} {\operatorname{tr}}\big[ {{\bm{H}}}{{\bm{W}}}^q {{\bm{H}}} {{\bm{Y}}}^{2r-q} \big] &+ {\operatorname{tr}}\big[_{{\bm{H}}}{{\bm{W}}}^{2r-q} {{\bm{H}}} {{\bm{Y}}}^{q} \big] \\ _ &\leq \sum_{i,j=1}^d \big( \lambda_i^{2r} +_\mu_j^{2r} \big) _\cdot {\operatorname{tr}}\big[_{{\bm{H}}} {{\bm{u
bf q} = V/N$ where $V$ is the strength of the impurity potential. In what follows we use standard finite temperature Green’s function formalism [@DS88; @BF04]. Because of the existence of two sub-lattices, the Green’s function can be written as a $2\times 2$ matrix: $$\begin{aligned} \bm G_{\sigma}(\bm k,{\bf p},\tau) = \left(\begin{array}{cc} G_{AA,\sigma} (\bm k,{\bf p}, \tau) \hspace{0.5cm} & G_{AB,\sigma}(\bm k,{\bf p},\tau) \\ G_{BA,\sigma} (\bm k,{\bf p},\tau) \hspace{0.5cm} & G_{BB,\sigma}(\bm k,{\bf p},\tau) \end{array}\right) \,,\end{aligned}$$ where $$\begin{aligned} G_{AA,\sigma}(\bm k,{\bf p}, \tau) &=& - \langle {\cal T} a_{\bm k,\sigma}(\tau) a_{\bm p,\sigma}^{\dag}(0) \rangle \,, \nonumber \\ G_{AB,\sigma}(\bm k,{\bf p}, \tau) &=& - \langle {\cal T} a_{\bm k,\sigma}(\tau) b_{\bm p,\sigma}^{\dag}(0) \rangle \,, \nonumber \\ G_{BA,\sigma}(\bm k,{\bf p},\tau) &=& - \langle {\cal T} b_{\bm k,\sigma}(\tau) a_{\bm p,\sigma}^{\dag}(0) \rangle \,, \nonumber \\ G_{BB,\sigma}(\bm k,{\bf p},\tau) &=& - \langle {\cal T} b_{\bm k,\sigma}(\tau) b_{\bm p,\sigma}^{\dag}(0) \rangle \,, \label{defg}\end{aligned}$$ where $\tau$ is the “imaginary” time, and ${\cal T}$ is the time ordering operator. For a single impurity the Green’s function can be written as $\bm G(\bm k,{\bf p},\tau) = \delta_{\bm k,\bm p} \bm G(\bm k,\tau)$, where [@DS88]: $$\bm G(\bm k,\omega_n)=\bm G^0(\bm
bf q } = V / N$ where $ V$ is the strength of the impurity potential. In what follows we practice standard finite temperature Green ’s affair formalism [ @DS88; @BF04 ]. Because of the existence of two sub - lattice, the Green ’s routine can be written as a $ 2\times 2 $ matrix: $ $ \begin{aligned } \bm G_{\sigma}(\bm k,{\bf p},\tau) = \left(\begin{array}{cc } G_{AA,\sigma } (\bm k,{\bf p }, \tau) \hspace{0.5 cm } & G_{AB,\sigma}(\bm k,{\bf p},\tau) \\ G_{BA,\sigma } (\bm k,{\bf p},\tau) \hspace{0.5 curium } & G_{BB,\sigma}(\bm k,{\bf p},\tau) \end{array}\right) \,,\end{aligned}$$ where $ $ \begin{aligned } G_{AA,\sigma}(\bm k,{\bf p }, \tau) & = & - \langle { \cal T } a_{\bm k,\sigma}(\tau) a_{\bm p,\sigma}^{\dag}(0) \rangle \, , \nonumber \\ G_{AB,\sigma}(\bm k,{\bf p }, \tau) & = & - \langle { \cal T } a_{\bm k,\sigma}(\tau) b_{\bm p,\sigma}^{\dag}(0) \rangle \, , \nonumber \\ G_{BA,\sigma}(\bm k,{\bf p},\tau) & = & - \langle { \cal T } b_{\bm k,\sigma}(\tau) a_{\bm p,\sigma}^{\dag}(0) \rangle \, , \nonumber \\ G_{BB,\sigma}(\bm k,{\bf p},\tau) & = & - \langle { \cal T } b_{\bm k,\sigma}(\tau) b_{\bm p,\sigma}^{\dag}(0) \rangle \, , \label{defg}\end{aligned}$$ where $ \tau$ is the “ fanciful ” time, and $ { \cal T}$ is the time ordering operator. For a single impurity the Green ’s routine can be written as $ \bm G(\bm k,{\bf p},\tau) = \delta_{\bm k,\bm p } \bm G(\bm k,\tau)$, where [ @DS88 ]: $ $ \bm G(\bm k,\omega_n)=\bm G^0(\bm
bf e} = V/N$ where $V$ is the strtngth of the imputiry potxntial. Jn what wollows we use standard finive twmperqture Green’s function wormalism [@DS88; @BF04]. Bwcauww of the eejstence of twk sub-nettices, the Gregn’s function can be writtet xs a $2\times 2$ matrix: $$\begin{aligned} \bm G_{\sidma}(\bm k,{\nf p},\tau) = \left(\begyn{argar}{cc} G_{ZA,\sigma} (\bm k,{\bf p}, \tau) \hspace{0.5cm} & G_{AB,\aigma}(\bm k,{\bf p},\tau) \\ G_{BA,\xigma} (\bm k,{\bf p},\tau) \hspace{0.5cm} & G_{BH,\sigma}(\bm k,{\bf p},\tau) \end{array}\ritht) \,,\qbd{aligned}$$ whdre $$\begin{aligned} G_{AA,\sigja}(\bm k,{\bf p}, \tau) &=& - \langle {\cal T} a_{\cm k,\sngma}(\tau) a_{\bm p,\wigld}^{\dag}(0) \rangle \,, \ionumbvr \\ G_{AB,\sigma}(\bm k,{\bf p}, \tag) &=& - \lanble {\cal T} a_{\bm l,\sijma}(\tqu) b_{\bm p,\sigma}^{\dag}(0) \ranglx \,, \nonumber \\ G_{BA,\sigma}(\bm k,{\bf p},\tau) &=& - \langle {\cal T} b_{\bm k,\wigma}(\jau) a_{\tm p,\rugmx}^{\daf}(0) \rengme \,, \nonkmbxr \\ G_{BB,\sigma}(\bj k,{\bf p},\tau) &=& - \langle {\cal T} b_{\bm l,\sytma}(\tau) b_{\bm p,\sjgma}^{\dad}(0) \rwngle \,, \label{defg}\end{aligned}$$ where $\tau$ is uhe “ijaginary” time, and ${\cal T}$ is the time ordering operator. Sor a single impurity the Green’s function can be frittxn as $\bi T(\bl k,{\bf p},\tau) = \delta_{\bm k,\bm p} \bm G(\bm k,\tau)$, where [@DF88]: $$\bk N(\bm k,\omega_n)=\bm G^0(\bi
bf q} = V/N$ where $V$ is of impurity potential. what follows we function [@DS88; @BF04]. Because the existence of sub-lattices, the Green’s function can be as a $2\times 2$ matrix: $$\begin{aligned} \bm G_{\sigma}(\bm k,{\bf p},\tau) = \left(\begin{array}{cc} G_{AA,\sigma} k,{\bf p}, \tau) \hspace{0.5cm} & G_{AB,\sigma}(\bm k,{\bf p},\tau) \\ G_{BA,\sigma} (\bm k,{\bf p},\tau) & k,{\bf \end{array}\right) where $$\begin{aligned} G_{AA,\sigma}(\bm k,{\bf p}, \tau) &=& - \langle {\cal T} a_{\bm k,\sigma}(\tau) a_{\bm p,\sigma}^{\dag}(0) \rangle \nonumber \\ G_{AB,\sigma}(\bm k,{\bf p}, \tau) &=& - {\cal T} a_{\bm k,\sigma}(\tau) p,\sigma}^{\dag}(0) \rangle \,, \nonumber \\ k,{\bf &=& - {\cal b_{\bm a_{\bm p,\sigma}^{\dag}(0) \rangle \nonumber \\ G_{BB,\sigma}(\bm k,{\bf p},\tau) &=& - \langle {\cal T} b_{\bm k,\sigma}(\tau) b_{\bm p,\sigma}^{\dag}(0) \rangle \,, \label{defg}\end{aligned}$$ $\tau$ is time, and T}$ the ordering operator. For impurity the Green’s function can be G(\bm k,{\bf p},\tau) = \delta_{\bm k,\bm p} \bm k,\tau)$, where $$\bm G(\bm k,\omega_n)=\bm G^0(\bm
bf q} = V/N$ where $V$ is the strength oF the impuriTy potEntIal. in What FollOws we use standaRD finIte temperature Green’s fuNctioN fORmalISm [@dS88; @BF04]. BEcause oF ThE EXisTeNcE of TwO SuB-lattIceS, the GreEn’s functioN caN bE written as a $2\tIMeS 2$ matrix: $$\begIn{aLigned} \bm G_{\sigMa}(\bM k,{\bf p},\tAu) = \LefT(\Begin{ArrAy}{cc} G_{aA,\sigmA} (\Bm k,{\bf p}, \Tau) \hspace{0.5Cm} & g_{aB,\sigmA}(\Bm k,{\bf p},\tAU) \\ g_{Ba,\sigMa} (\bm k,{\bf p},\tau) \hspace{0.5CM} & G_{bb,\sigma}(\bm k,{\bf p},\taU) \end{arRaY}\RiGHT) \,,\enD{alIgned}$$ where $$\BeGin{alIGned} G_{AA,\SIgMA}(\BM k,{\bF P}, \tau) &=& - \langle {\cal t} a_{\bm k,\sigma}(\tAU) a_{\bM p,\sigmA}^{\dAg}(0) \rANgle \,, \noNumbeR \\ G_{ab,\siGma}(\bm k,{\bf p}, \taU) &=& - \lanGle {\cal T} a_{\bM k,\sigmA}(\Tau) b_{\bm p,\SIgma}^{\dag}(0) \Rangle \,, \NonUmbEr \\ G_{Ba,\SiGmA}(\bm K,{\bF P},\taU) &=& - \LaNglE {\Cal t} b_{\bm k,\sigMa}(\TaU) a_{\bm p,\SigmA}^{\DAG}(0) \RangLe \,, \nOnumBer \\ G_{Bb,\sigma}(\bm k,{\bf p},\tAu) &=& - \lAnglE {\Cal t} b_{\bm k,\Sigma}(\Tau) b_{\Bm P,\sigmA}^{\dag}(0) \raNgle \,, \lAbEl{defg}\end{aligneD}$$ wheRe $\tau$ is thE “imAgInaRy” Time, aND ${\cal T}$ iS thE tiMe orderIng operATor. foR A SInGle impurity the GreeN’s FUNcTion can bE writtEN aS $\bM g(\bm k,{\bf p},\tAu) = \DelTa_{\bm K,\BM p} \bm G(\Bm k,\tAU)$, wHere [@DS88]: $$\bm g(\bm k,\omEGa_N)=\bM G^0(\bm
bf q} = V/N$ where $V$ isthe streng th of th e i mp urit y po tential. In wh a t fo llows we use standardfinit et empe r at ure G reen’sf un c t ion f or mal is m [ @DS88 ; @ BF04].Because of th eexistence of tw o sub-latt ice s, the Green ’sfuncti on ca n be w rit ten a s a $2 \ times2$ matrix :$ $\begi n {aligne d } \ bm G _{\sigma}(\bm k,{ \ bf p},\tau) = \le ft(\be gi n {a r r ay} {cc } G_{AA,\s ig ma} ( \ bm k,{\ b fp } , \t a u) \hspace{0. 5cm} & G_{A B ,\s igma}( \b m k , {\bf p },\ta u) \\G_{BA,\sigm a} ( \bm k,{\b f p},\ t au) \hs p ace{0.5 cm} &G_{ BB, \sig m a} (\ bmk, { \bf p} ,\t a u) \end{a rr ay }\rig ht)\ , , \ end{ ali gned }$$ w here $$\begin {al igne d } G _{AA, \sigm a}(\ bm k,{\ bf p}, \tau )&=& - \langle { \cal T} a_{\b m k ,\ sig ma }(\ta u ) a_{\ bmp,\ sigma}^ {\dag}( 0 ) \ ra n g l e\,, \nonumber \\ G _{ A B ,\ sigma}(\ bm k,{ \ bf p } , \tau)&= & - \la n g le {\ calT }a_{\bm k ,\sigm a }( \t au) b_{ \b m p,\s ig ma} ^{\ dag}( 0 ) \r angle\,, \non umber \\ G_{BA,\sigm a }(\bm k,{\bfp }, \ t au ) &=& -\langle {\c al T } b_{ \bmk ,\ sig m a}(\t au) a _{ \ bm p,\sigma}^{\dag}(0) \ rangle \,,\nonumber \\G_{BB,\sig m a } (\bm k,{ \bfp }, \ tau) &=& - \la ngle{\cal T} b _ {\bm k,\ sigma }(\tau)b_{\bm p, \ s igma}^{\ dag }(0 ) \ ran g l e\,, \label{de f g }\en d{ aligned }$$ where$\t au$ is th e“imaginar y” time, a nd $ {\ cal T}$i s the ti me or de rin g ope r ator. Fora si ng le imp urity t h eG r een’ sfu ncti onca n bewrit t enas $\bm G(\bm k, {\b f p}, \t au ) = \de lta_{\bm k,\b mp} \bm G(\ bm k, \tau)$ , where [@ DS88]: $$\bm G(\bm k,\o m ega_n)= \bm G^0( \bm
bf q}_= V/N$_where $V$ is the_strength of_the_impurity potential._In_what follows we_use standard finite_temperature Green’s function formalism_[@DS88; @BF04]. Because_of_the existence of two sub-lattices, the Green’s function can be written as a $2\times_2$_matrix: $$\begin{aligned} \bm_G_{\sigma}(\bm_k,{\bf_p},\tau) = \left(\begin{array}{cc} G_{AA,\sigma} (\bm k,{\bf_p}, \tau) \hspace{0.5cm} & G_{AB,\sigma}(\bm_k,{\bf p},\tau)_\\ G_{BA,\sigma} (\bm k,{\bf p},\tau) \hspace{0.5cm} & G_{BB,\sigma}(\bm k,{\bf_p},\tau)_ \end{array}\right) \,,\end{aligned}$$_where $$\begin{aligned} G_{AA,\sigma}(\bm k,{\bf p}, \tau) &=& - \langle {\cal_T} a_{\bm k,\sigma}(\tau) a_{\bm p,\sigma}^{\dag}(0) \rangle \,, \nonumber \\ G_{AB,\sigma}(\bm_k,{\bf p}, \tau)_&=&_-_\langle {\cal T} a_{\bm_k,\sigma}(\tau) b_{\bm p,\sigma}^{\dag}(0) \rangle \,, \nonumber \\ G_{BA,\sigma}(\bm k,{\bf p},\tau)_&=& - \langle {\cal T} b_{\bm_k,\sigma}(\tau) a_{\bm p,\sigma}^{\dag}(0) \rangle \,, \nonumber \\ G_{BB,\sigma}(\bm k,{\bf p},\tau) &=&_- \langle {\cal T} b_{\bm k,\sigma}(\tau)_b_{\bm p,\sigma}^{\dag}(0) \rangle \,, \label{defg}\end{aligned}$$ where $\tau$_is the_“imaginary” time, and ${\cal T}$_is the time_ordering operator. For_a single impurity_the Green’s function can be written_as $\bm G(\bm_k,{\bf p},\tau) = \delta_{\bm k,\bm p}_\bm_G(\bm k,\tau)$, where_[@DS88]:_$$\bm_G(\bm k,\omega_n)=\bm_G^0(\bm
the data. A large body of literature has been devoted to studying codes that are minimax optimal with respect to (\[Redundancy\]), exactly [@Topsoe79; @Shtarkov87en2; @Haussler97] or asymptotically [@BarronRissanenYu98; @Grunwald07]. Let us notice that if the minimax expected Shannon redundancy $$\begin{aligned} \label{MinimaxRed} \min_{C} \sup_{P\in\mathcal{M}} {\textbf{E}\, _}{x\sim P} {\left[ |C(x)|+\log P(x) \right]}\end{aligned}$$ or the minimax regret $$\begin{aligned} \label{MinimaxReg} \min_{C} \sup_{P\in\mathcal{M}} \max_{x} {\left[ |C(x)|+\log P(x) \right]}\end{aligned}$$ are finite, plausibly bounded in terms of the data length, and achieved by a unique code $C$ then the corresponding minimax properties appear a plausible rationale to argue for code $C$’s optimality against data typical of a class of distributions $\mathcal{M}$. Things change when (\[MinimaxRed\]) or (\[MinimaxReg\]) are infinite since then every code is a minimizer. Infinite or unbounded minimax values appear in fact in many statistical models: (i) There are no universal redundancy rates for stationary ergodic processes [@Shields93]. (ii) Even in the parametric iid case, like Poisson or geometric, one often has to restrict the parameter range to a compact subset to have a reasonable minimax code [@Grunwald07 Theorem 7.1 and Sections 11.1.1–2]. In a surprising contrast, the redundancy for computable parameters can be very small, which is known as superefficient estimation/compression [@LeCam53; @Vovk91en2; @BarronHengartner98]. The minimax values (\[MinimaxRed\]) or (\[MinimaxReg\]) may be infinite because there is no worst case of data rather than no intuitively good code. Often there exists an intuitively good code but to single it out with the minimax criterion, we have to modify the score (\[Redundancy\])
the data. A   large body of literature has been devoted to study code that are minimax optimal with respect to (\[Redundancy\ ]), exactly [ @Topsoe79; @Shtarkov87en2; @Haussler97 ] or asymptotically [ @BarronRissanenYu98; @Grunwald07 ]. get us notice that if the minimax expected Shannon redundancy $ $ \begin{aligned } \label{MinimaxRed } \min_{C } \sup_{P\in\mathcal{M } } { \textbf{E}\, _ } { x\sim P } { \left [ |C(x)|+\log P(x) \right]}\end{aligned}$$ or the minimax sorrow $ $ \begin{aligned } \label{MinimaxReg } \min_{C } \sup_{P\in\mathcal{M } } \max_{x } { \left [ |C(x)|+\log P(x) \right]}\end{aligned}$$ are finite, plausibly bounded in terms of the datum duration, and achieved by a   singular code $ C$ then the corresponding minimax properties look a   plausible rationale to argue for code $ C$ ’s optimality against data typical of a   class of distributions $ \mathcal{M}$. Things change when (\[MinimaxRed\ ]) or (\[MinimaxReg\ ]) are infinite since then every code is a   minimizer. countless or unbounded minimax values appear in fact in many statistical model: (i) There are no cosmopolitan redundancy rates for stationary ergodic process [ @Shields93 ]. (ii) Even in the parametric iid case, like Poisson or geometric, one much has to restrict the parameter compass to a   compact subset to have a   reasonable minimax code [ @Grunwald07 Theorem 7.1 and Sections 11.1.1–2 ]. In a   surprising contrast, the redundancy for computable parameters can be very small, which is known as superefficient estimation / compression [ @LeCam53; @Vovk91en2; @BarronHengartner98 ]. The minimax values (\[MinimaxRed\ ]) or (\[MinimaxReg\ ]) may be infinite because there is no worst case of data rather than no intuitively good code. frequently there exists an intuitively good code but to single it out with the minimax standard, we take to change the score (\[Redundancy\ ] )
thf data. A large body of littrature has been bwvoted to sthdying cudes that are minimax optimap qith eespect to (\[Redundancy\]), dxactly [@Tlpsoe79; @Shrarkic87en2; @Hausslxd97] or asniptoflcallv [@UarronRissanenYo98; @Grunwald07]. Lat us notice tvag nf the minimax expected Shannon redugdancy $$\nehin{aligned} \lafel{MpnymaxDvd} \min_{C} \sup_{P\in\mathcal{M}} {\textbf{E}\, _}{x\sim P} {\left[ |C(x)|+\log P(c) \right]}\end{aligned}$$ or the mlnimwx regret $$\begin{alihned} \label{NiniiqxReg} \min_{C} \rup_{P\in\mathbcl{M}} \max_{x} {\lert[ |C(x)|+\log P(x) \right]}\end{aligned}$$ are finice, plausibli goktded in terns of the data lekbth, ang achiefed by a unique cove $C$ then the correspondiig minimax propertief appear d pmausible rationalw ro arcue xor zide $C$’s o'tijality ageinst data fypical of q class of distributoogw $\mathcal{M}$. Thihgs chwnde when (\[MinimaxRed\]) or (\[MinimaxReg\]) are inxinjte since then every coee is a minimizer. Infijite or ugbounded minimax values appear in fact in many stdtistmcxl noqdos: (i) There are no universal redundancy rates fow suatponary ergodic prjcesses [@Shirlfs93]. (yi) Even in thg paramzfrjc iid case, like Plisson jr geimetric, oge ogten has to restrict the paeameter rangv to a compact subset tl have a rearonanle monimax code [@Grunwald07 Thzorem 7.1 and Sectiojs 11.1.1–2]. In a shfprising contrasg, tme sedundancy for computable [arameterw cau be verh smsll, whych is knoan as superefficient estlmatiln/wompressioj [@LeCam53; @Vovk91en2; @BarronHengartner98]. Vie minimax vakuas (\[KinimaxRzd\]) or (\[KinimaxReg\]) mar be infinite yecause chere ks no worsn case of data rathew than no intgltively good code. Ofeen rherw existr an intuitivelu good cobt but to sungle it out with bhe mkhimax criterion, we have to modify thd ssoge (\[Cedunqdncy\])
the data. A large body of literature devoted studying codes are minimax optimal [@Topsoe79; @Haussler97] or asymptotically @Grunwald07]. Let us that if the minimax expected Shannon $$\begin{aligned} \label{MinimaxRed} \min_{C} \sup_{P\in\mathcal{M}} {\textbf{E}\, _}{x\sim P} {\left[ |C(x)|+\log P(x) \right]}\end{aligned}$$ or the regret $$\begin{aligned} \label{MinimaxReg} \min_{C} \sup_{P\in\mathcal{M}} \max_{x} {\left[ |C(x)|+\log P(x) \right]}\end{aligned}$$ are finite, plausibly in of data and achieved by a unique code $C$ then the corresponding minimax properties appear a plausible rationale argue for code $C$’s optimality against data typical a class of distributions Things change when (\[MinimaxRed\]) or are since then code a Infinite or unbounded values appear in fact in many statistical models: (i) There are no universal redundancy rates for stationary processes [@Shields93]. in the iid like or geometric, one to restrict the parameter range to to have a reasonable minimax code [@Grunwald07 Theorem and Sections In a surprising contrast, the redundancy computable parameters can be very small, which is as superefficient estimation/compression [@LeCam53; @Vovk91en2; @BarronHengartner98]. The minimax values (\[MinimaxRed\]) or (\[MinimaxReg\]) may be infinite is no worst case data rather than intuitively code. there an intuitively code but to single it out with the minimax criterion, we to modify the score (\[Redundancy\])
the data. A large body of literaTure has beeN devoTed To sTuDyinG codEs that are minimAX optImal with respect to (\[RedunDancy\]), ExACtly [@tOpSoe79; @ShTarkov87eN2; @haUSSleR97] oR aSymPtOTiCally [@barRonRissAnenYu98; @GrunWalD07]. LEt us notice thAT iF the minimaX exPected ShannoN reDundanCy $$\BegIN{aligNed} \Label{minimaXred} \min_{c} \sup_{P\in\maThCAl{M}} {\texTBf{E}\, _}{x\sim p} {\LEfT[ |C(x)|+\lOg P(x) \right]}\end{alignED}$$ oR The minimax regrEt $$\begiN{aLIgNED} \laBel{minimaxReg} \MiN_{C} \sup_{p\In\mathcAL{M}} \MAX_{X} {\leFT[ |C(x)|+\log P(x) \right]}\End{aligned}$$ aRE fiNite, plAuSibLY boundEd in tErMS of The data lengTh, anD achieved By a uniQUe code $C$ THen the cOrrespOndIng MiniMAx PrOpeRtIEs aPPeAr a PLauSible ratIoNaLe to aRgue FOR COde $C$’S opTimaLity aGainst data typIcaL of a CLasS of diStribUtioNs $\MathcAl{M}$. ThiNgs chAnGe when (\[MinimaxReD\]) or (\[MInimaxReg\]) Are InFinItE sincE Then evEry CodE is a minImizer. INFinItE OR UnBounded minimax valuEs APPeAr in fact In many STaTiSTical modElS: (i) THere ARE no unIverSAl RedundanCy rateS FoR sTationaRy ErgodiC pRocEssEs [@ShiELds93]. (iI) Even iN the paraMetriC Iid case, like PoiSSon or geometriC, OnE OFtEN has To rEstrict the pAramETer rAnge TO a ComPAct suBset tO hAVe A Reasonable minimax coDe [@grunwaLd07 TheOrem 7.1 and SectioNs 11.1.1–2]. In a surprISINg contraSt, thE ReDUndancy for compUtablE parameterS Can be verY smalL, which is Known as suPERefficieNt eStiMatIon/COMpRession [@LeCam53; @VOVK91en2; @BArRonHengArtNer98]. The mIniMax ValUes (\[miNimaxRed\]) oR (\[MinimaxreG\]) mAy Be InfInite BEcause thErE is No WorSt casE Of data RatheR thaN nO iNTuiTively gOOd CODe. OfTeN tHere ExiStS an inTuitIVelY good coDe but to siNglE It ouT wItH the minImax criterion, We Have to modiFy The Score (\[REDUndancy\])
the data. A large body o f literatu re ha s b een d evot ed t o studying cod e s th at are minimax optimal with r e spec t t o (\[ Redunda n cy \ ] ),ex ac tly [ @ To psoe7 9;@Shtark ov87en2; @ Hau ss ler97] or as y mp totically[@B arronRissane nYu 98; @G ru nwa l d07]. Le t usnotice that i f the min im a x expe c ted Sha n n on red undancy $$\begin{ a li g ned} \label{ Minima xR e d} \m in_ {C} \sup_{ P\ in\ma t hcal{M} } { \ t e xtb f {E}\, _}{x\si m P} {\left [ |C (x)|+\ lo g P ( x) \ri ght]} \e n d{a ligned}$$ o r th e minimax regre t $$\beg i n{align ed} \la bel {Min i ma xR eg} \mi n _{ C}\ sup _{P\in\m at hc al{M} } \m a x _ { x} { \le ft[|C(x) |+\log P(x) \ rig ht]} \ end {alig ned}$ $ ar efinit e, pla usibl ybounded in term s of the data le ng th, a nd ac h ievedbya u nique c ode $C$ the nt h e c orresponding minim ax p ro pertiesappear apl a usible r at ion alet o argu e fo r c ode $C$’ s opti m al it y again st dataty pic alof ac lass of di stributi ons $ \ mathcal{M}$. T hings changew he n (\ [ Mini max Red\]) or ( \[Mi n imax Reg\ ] )are infin ite s in c et hen every code is a  m inimiz er. I nfinite or un bounded mi n i m ax value s ap p ea r in fact in ma ny st atisticalm odels: ( i) Th ere areno univer s a l redund anc y r ate s f o r s tationary erg o d ic p ro cesses[@S hields9 3]. (i i)Eve nin the pa rametric i id c as e,likeP oisson o rgeo me tri c, on e often hasto r es tr i ctthe par a me t e r ra ng eto a  co mp act s ubse t to have a  reasonab lem inim ax c ode [@G runwald07 The or em 7.1 and S ect ions 1 1 . 1.1–2].In a surprising contras t , the r edu ndanc y fo r computa ble param ete r s canbe ver y sma ll , w h i ch is k no wnas supereffi c i ent esti ma tion /compre ssion [@LeCam53; @ V ovk 91en2; @Barro nHe ngar t n er 98] . T hemi n ima x values (\[Minim axRed\]) o r( \[ MinimaxReg \ ])ma y be in finitebecau s e there is no wo rst caseof dat a rat her than n o intuit ively goo d code . O ftenthe re exi st s a n int uitive l y g ood c ode bu tto sin gle i tout with the minimax criterion, we ha ve to mo dify thesco r e ( \[Redunda ncy\ ])
the_data. A large body_of literature has been_devoted to_studying_codes that_are_minimax optimal with_respect to (\[Redundancy\]),_exactly [@Topsoe79; @Shtarkov87en2; @Haussler97]_or asymptotically [@BarronRissanenYu98;_@Grunwald07]._Let us notice that if the minimax expected Shannon redundancy $$\begin{aligned} \label{MinimaxRed} _\min_{C}_\sup_{P\in\mathcal{M}} {\textbf{E}\,__}{x\sim_P}_{\left[ |C(x)|+\log P(x) \right]}\end{aligned}$$ or_the minimax regret $$\begin{aligned} _\label{MinimaxReg} _\min_{C} \sup_{P\in\mathcal{M}} \max_{x} {\left[ |C(x)|+\log P(x) \right]}\end{aligned}$$ are_finite,_plausibly bounded in_terms of the data length, and achieved by a unique_code $C$ then the corresponding minimax_properties appear a plausible_rationale_to_argue for code $C$’s_optimality against data typical of a class_of distributions $\mathcal{M}$. Things change when (\[MinimaxRed\])_or (\[MinimaxReg\]) are infinite since then every_code is a minimizer. Infinite or unbounded_minimax values appear in fact_in many_statistical models: (i) There are_no universal redundancy_rates for_stationary ergodic processes_[@Shields93]. (ii) Even in the parametric_iid case, like_Poisson or geometric, one often has_to_restrict the parameter_range_to_a compact subset_to have a reasonable_minimax_code [@Grunwald07_Theorem_7.1 and Sections 11.1.1–2]. In a surprising_contrast,_the redundancy for computable parameters can be_very small, which is_known_as superefficient estimation/compression [@LeCam53;_@Vovk91en2; @BarronHengartner98]. The minimax values (\[MinimaxRed\])_or (\[MinimaxReg\]) may be infinite because_there is_no worst_case of data rather than no intuitively good code. Often there_exists an intuitively good code but_to single it out_with the_minimax_criterion, we have_to_modify the_score (\[Redundancy\])
mathtt{i})$. If ${{\sc\mbox{C}\hspace{1.0pt}}}$ has more than one object, one can fix an ordering of its objects and define the rank of a finitary $2$-representation of ${{\sc\mbox{C}\hspace{1.0pt}}}$ as a suitable tuple of positive integers. However, in this document we will only consider $2$-categories with a single object. We say that ${{\sc\mbox{C}\hspace{1.0pt}}}$ is [*weakly fiat*]{} if it is finitary and has a weak antiautomorphism $(-)^{\ast}$ of finite order and adjunction morphisms, see [@MM6 Subsection 2.5]. If $(-)^{\ast}$ is involutive, we say that ${{\sc\mbox{C}\hspace{1.0pt}}}$ is [*fiat*]{}. The existence of left and right adjoints suffices to conclude weak fiatness: taking right (alternatively left) adjoints is functorial and gives the desired weak $2$-equivalence; see [@EGNO]. A $2$-category of the form ${{\sc\mbox{C}\hspace{1.0pt}}}_{\!A}$ is weakly fiat if and only if $A$ is self-injective. This is an immediate consequence of [@MM1 Lemma 45]: \[AdjSelf\] Let $f,e$ be primitive mutually orthogonal idempotents of $A$. Then $$\big( \left(Ae \otimes_{\Bbbk} fA\right) \otimes_{A} -, \left( (fA)^{*} \otimes_{\Bbbk} eA \right) \otimes_{A} - \big)$$ is an adjoint pair of endofunctors of $A\!\on{-mod}$. Abelianization -------------- Given a finitary $2$-representation $\mathbf{M}$, we may consider its [*abelianization*]{} $\overline{\mathbf{M}}$, an abelian $2$-representation of ${{\sc\mbox{C}\hspace{1.0pt}}}$, defined in [@MMMT Section 3]. The abelianization defined therein is a significant improvement of that given in [@MM2 Subsection 4.2], however the main features we will use are shared by both constructions: we may recover $\mathbf{M}$ via a canonical embedding and the action of $1$-morphisms in $\overline{\mathbf{M}}$ is exact if ${{\sc\mbox
mathtt{i})$. If $ { { \sc\mbox{C}\hspace{1.0pt}}}$ has more than one object, one can fix an ordering of its object and specify the rank of a finitary $ 2$-representation of $ { { \sc\mbox{C}\hspace{1.0pt}}}$ as a suitable tuple of positive integer. However, in this document we will only consider $ 2$-categories with a individual object. We say that $ { { \sc\mbox{C}\hspace{1.0pt}}}$ is [ * decrepit fiat * ] { } if it is finitary and have a decrepit antiautomorphism $ (-)^{\ast}$ of finite order and adjunction morphisms, see [ @MM6 Subsection 2.5 ]. If $ (-)^{\ast}$ is involutive, we read that $ { { \sc\mbox{C}\hspace{1.0pt}}}$ is [ * fiat * ] { }. The being of left and right adjoints suffice to conclude weak fiatness: taking mighty (alternatively left) adjoints is functorial and gives the desire weak $ 2$-equivalence; understand [ @EGNO ]. A $ 2$-category of the form $ { { \sc\mbox{C}\hspace{1.0pt}}}_{\!A}$ is weakly fiat if and only if $ A$ is self - injective. This is an immediate consequence of [ @MM1 Lemma 45 ]: \[AdjSelf\ ] lease $ f, e$ be primitive mutually orthogonal idempotents of $ A$. Then $ $ \big (\left(Ae \otimes_{\Bbbk } fA\right) \otimes_{A } -, \left (( fA)^ { * } \otimes_{\Bbbk } eA \right) \otimes_{A } - \big)$$ is an adjoint pair of endofunctors of $ A\!\on{-mod}$. Abelianization -------------- Given a finitary $ 2$-representation $ \mathbf{M}$, we may consider its [ * abelianization * ] { } $ \overline{\mathbf{M}}$, an abelian $ 2$-representation of $ { { \sc\mbox{C}\hspace{1.0pt}}}$, defined in [ @MMMT Section 3 ]. The abelianization defined therein is a significant improvement of that given in [ @MM2 Subsection 4.2 ], however the main features we will use are shared by both constructions: we may recover $ \mathbf{M}$ via a canonical embedding and the military action of $ 1$-morphisms in $ \overline{\mathbf{M}}$ is exact if $ { { \sc\mbox
matjtt{i})$. If ${{\sc\mbox{C}\hspace{1.0pt}}}$ mas more than ong ibject, one czn fix av ordering of its objects anv dedine uke rank of a finitarh $2$-represejtation if ${{\sr\mbox{C}\hspace{1.0pt}}}$ as a suitable thile oy 'ositive integets. However, it this documend de will only consider $2$-categories with a singke object. We say jhat ${{\xs\mbos{B}\hwpace{1.0pt}}}$ is [*weakly fiat*]{} if it js finiuary and has a weal antiautomorphism $(-)^{\ast}$ of vinihe order and adjunftion morphusms, wee [@MM6 Subseztion 2.5]. If $(-)^{\ast}$ is involotive, we say that ${{\sc\mbox{C}\hspace{1.0pg}}}$ is [*yiat*]{}. The exusrenfg of left anv righn adjoints sungices do conckude weak fiatkess: vakibg right (alternativelb left) adjoints is fonctorial dnb gives the desired wwaj $2$-equhvalance; wee [@EGHO]. E $2$-cztegorj oh the form ${{\ac\mbox{C}\hspaxe{1.0pt}}}_{\!A}$ is weakly fiau is and only if $Z$ is sqls-injective. This is an immediate consequtnce kf [@MM1 Lemma 45]: \[AdjSelf\] Let $f,e$ be primitive mutuwlly orthjgonal idempotents of $A$. Then $$\big( \left(Ae \otimes_{\Bbbn} fA\rmgft) \itlmes_{X} -, \peft( (fA)^{*} \otimes_{\Bbbk} eA \right) \otimes_{A} - \big)$$ is ag acjpint pair of ekdofunctors of $A\!\on{-kof}$. Anglianization -------------- Gixen a yjnjtary $2$-representatiln $\mathff{M}$, ww may confidet its [*abelianization*]{} $\overlibe{\mathbf{M}}$, an qbelian $2$-representacion of ${{\sc\mbux{C}\hxpace{1.0lt}}}$, defined in [@MMMT Sectnon 3]. Tge abelianixation derkned therein is x spgnixicant improvement of that given in [@MM2 Xubsectkon 4.2], howevqr the maij feabgres we will use age shcred ty both cojstructions: we may recover $\mathuh{M}$ via a canomiwal embeddiug and the action os $1$-morphisms in $\overliue{\mathcf{M}}$ is exabt if ${{\sc\muox
mathtt{i})$. If ${{\sc\mbox{C}\hspace{1.0pt}}}$ has more than one can an ordering its objects and finitary of ${{\sc\mbox{C}\hspace{1.0pt}}}$ as suitable tuple of integers. However, in this document we only consider $2$-categories with a single object. We say that ${{\sc\mbox{C}\hspace{1.0pt}}}$ is [*weakly if it is finitary and has a weak antiautomorphism $(-)^{\ast}$ of finite order adjunction see Subsection If $(-)^{\ast}$ is involutive, we say that ${{\sc\mbox{C}\hspace{1.0pt}}}$ is [*fiat*]{}. The existence of left and right suffices to conclude weak fiatness: taking right (alternatively adjoints is functorial and the desired weak $2$-equivalence; see A of the ${{\sc\mbox{C}\hspace{1.0pt}}}_{\!A}$ weakly if and only $A$ is self-injective. This is an immediate consequence of [@MM1 Lemma 45]: \[AdjSelf\] Let $f,e$ be primitive orthogonal idempotents Then $$\big( \otimes_{\Bbbk} \otimes_{A} \left( (fA)^{*} \otimes_{\Bbbk} \otimes_{A} - \big)$$ is an adjoint of $A\!\on{-mod}$. Abelianization -------------- Given a finitary $2$-representation we may its [*abelianization*]{} $\overline{\mathbf{M}}$, an abelian $2$-representation ${{\sc\mbox{C}\hspace{1.0pt}}}$, defined in [@MMMT Section 3]. The abelianization therein is a significant improvement of that given in [@MM2 Subsection 4.2], however the main will use are shared both constructions: we recover via canonical and the of $1$-morphisms in $\overline{\mathbf{M}}$ is exact if ${{\sc\mbox
mathtt{i})$. If ${{\sc\mbox{C}\hspace{1.0pt}}}$ hAs more than One obJecT, onE cAn fiX an oRdering of its obJEcts And define the rank of a finItary $2$-RePReseNTaTion oF ${{\sc\mbox{c}\HsPACe{1.0pT}}}$ aS a SuiTaBLe Tuple Of pOsitive Integers. HoWevEr, In this documeNT wE will only cOnsIder $2$-categoriEs wIth a siNgLe oBJect. WE saY that ${{\Sc\mbox{c}\Hspace{1.0Pt}}}$ is [*weaklY fIAt*]{} if it IS finitaRY AnD has A weak antiautomorpHIsM $(-)^{\Ast}$ of finite ordEr and aDjUNcTIOn mOrpHisms, see [@MM6 suBsectIOn 2.5]. If $(-)^{\ast}$ IS iNVOLutIVe, we say that ${{\sc\Mbox{C}\hspace{1.0PT}}}$ is [*Fiat*]{}. ThE eXisTEnce of Left aNd RIghT adjoints suFficEs to conclUde weaK FiatnesS: Taking rIght (alTerNatIvelY LeFt) AdjOiNTs iS FuNctORiaL and giveS tHe DesirEd weAK $2$-EQUivaLenCe; seE [@EGNO]. a $2$-category of thE foRm ${{\sc\MBox{c}\hspaCe{1.0pt}}}_{\!A}$ Is weAkLy fiaT if and Only iF $A$ Is self-injective. this Is an immedIatE cOnsEqUence OF [@MM1 LemMa 45]: \[ADjSElf\] Let $f,E$ be primITivE mUTUAlLy orthogonal idempoTeNTS oF $A$. Then $$\biG( \left(AE \OtImES_{\Bbbk} fA\rIgHt) \oTimeS_{a} -, \Left( (fa)^{*} \otiMEs_{\bbbk} eA \riGht) \otiMEs_{a} - \bIg)$$ is an aDjOint paIr Of eNdoFunctORs of $a\!\on{-mod}$. abelianiZatioN -------------- given a finitary $2$-REpresentation $\MAtHBF{M}$, WE may ConSider its [*abeLianIZatiOn*]{} $\ovERlIne{\MAthbf{m}}$, an abElIAn $2$-REpresentation of ${{\sc\mbOx{c}\hspacE{1.0pt}}}$, deFined in [@MMMT SeCtion 3]. The abELIAnizatioN defINeD Therein is a signIficaNt improvemENt of that Given In [@MM2 SubsEction 4.2], howEVEr the maiN feAtuRes We wILL uSe are shared by BOTh coNsTructioNs: wE may recOveR $\maThbF{M}$ vIa A canonicaL embeddiNg AnD tHe ActIon of $1$-MOrphisms In $\OveRlIne{\MathbF{m}}$ is exaCt if ${{\sC\mboX
mathtt{i})$. If ${{\sc\mbo x{C}\hspac e{1.0 pt} }}$ h as m orethan one objec t , on e can fix an orderingof it so bjec t sand d efine t h er a nkof a fi ni t ar y $2$ -re present ation of $ {{\ sc \mbox{C}\hsp a ce {1.0pt}}}$ as a suitabletup le ofpo sit i ve in teg ers.Howeve r , in t his docum en t we wi l l onlyc o ns ider $2$-categories w i th a single objec t. We s a yt h at${{ \sc\mbox{C }\ hspac e {1.0pt} } }$ i s [* w eakly fiat*]{ } if it isf ini tary a nd ha s a wea k ant ia u tom orphism $(- )^{\ ast}$ offinite order a n d adjun ctionmor phi sms, se e[@M M6 Sub s ec tio n 2. 5]. If $ (- )^ {\ast }$ i s i n volu tiv e, w e say that ${{\sc\ mbo x{C} \ hsp ace{1 .0pt} }}$is [*fi at*]{} . The e xistence of lef t an d right a djo in tssu ffice s to co ncl ude weak f iatness : ta ki n g ri ght (alternatively l e f t) adjoint s is f u nc to r ial andgi ves the d esire d we a k$2$-equi valenc e ;se e [@EGN O] . A $ 2$ -ca teg ory o f the form${{\sc\m box{C } \hspace{1.0pt} } }_{\!A}$ is w e ak l y f i at i f a nd only if$A$i s se lf-i n je cti v e. Th is is a n i m mediate consequence o f [@MM 1 Lem ma 45]: \[Ad jSelf\] Le t $ f,e$ beprim i ti v e mutually ort hogon al idempot e nts of $ A$. T hen $$\b ig( \left ( A e \otime s_{ \Bb bk} fA \ r ig ht) \otimes_{ A } -,\l eft( (f A)^ {*} \ot ime s_{ \Bb bk} e A \right) \otimes _{ A} - \ big )$$ i s an adjo in t p ai r o f end o functo rs of $A\ !\ on { -mo d}$. A b el i a niza ti on --- --- -- ----- - G i ven a fini tary $2$- rep r esen ta ti on $\ma thbf{M}$, wema y consider i ts[*abel i a nization *]{} $\overline{\mathbf { M}}$, a n a belia n $2 $-represe nta tion o f $ { {\sc\m box{C} \hspa ce {1. 0 p t}}}$ , de fin ed in [@MMMT S ect ion 3 ]. The abelia nization defined t h ere in is a signi fic anti m pr ove m en t of t h atg i ven in [@MM2 Su bsection 4 .2 ] ,however th e ma in featur es we w ill u s e are s hared byboth cons tr ucti o n s:we may rec over $\m athbf{M}$ via a ca nonic alembedd in g a nd th e acti o n o f $1$ -morph is ms in$\ove rl ine{\mat hbf{M}}$ is exact if ${ {\sc\m box
mathtt{i})$. If_${{\sc\mbox{C}\hspace{1.0pt}}}$ has_more than one object,_one can_fix_an ordering_of_its objects and_define the rank_of a finitary $2$-representation_of ${{\sc\mbox{C}\hspace{1.0pt}}}$ as_a_suitable tuple of positive integers. However, in this document we will only consider $2$-categories_with_a single_object. We_say_that ${{\sc\mbox{C}\hspace{1.0pt}}}$ is [*weakly fiat*]{}_if it is finitary and_has a_weak antiautomorphism $(-)^{\ast}$ of finite order and adjunction_morphisms,_see [@MM6 Subsection_2.5]. If $(-)^{\ast}$ is involutive, we say that ${{\sc\mbox{C}\hspace{1.0pt}}}$_is [*fiat*]{}. The existence of left_and right adjoints_suffices_to_conclude weak fiatness: taking_right (alternatively left) adjoints is functorial_and gives the desired weak $2$-equivalence;_see [@EGNO]. A $2$-category of the form ${{\sc\mbox{C}\hspace{1.0pt}}}_{\!A}$_is weakly fiat if and only_if $A$ is self-injective. This_is an_immediate consequence of [@MM1 Lemma_45]: \[AdjSelf\] Let $f,e$_be primitive_mutually orthogonal idempotents_of $A$. Then $$\big( \left(Ae \otimes_{\Bbbk}_fA\right) \otimes_{A} -,_\left( (fA)^{*} \otimes_{\Bbbk} eA \right) \otimes_{A}_-_\big)$$ is an_adjoint_pair_of endofunctors_of $A\!\on{-mod}$. Abelianization -------------- Given a_finitary_$2$-representation $\mathbf{M}$,_we_may consider its [*abelianization*]{} $\overline{\mathbf{M}}$, an_abelian_$2$-representation of ${{\sc\mbox{C}\hspace{1.0pt}}}$, defined in [@MMMT Section_3]. The abelianization defined_therein_is a significant improvement_of that given in [@MM2_Subsection 4.2], however the main features_we will_use are_shared by both constructions: we may recover $\mathbf{M}$ via a canonical_embedding and the action of $1$-morphisms_in $\overline{\mathbf{M}}$ is exact_if ${{\sc\mbox
of the state $s_{i}$. The transition probability matrix is approximated by converting the elements of P by approximating the transition probabilities using $P_{ij}=\nicefrac{N_{ij}}{N_{i}}$. The resulting matrix is often described as a first order Markov Matrix [@Ross2013]. State changes are based on only the observation-to-observation amplitude changes; the matrix is a representation of the linearly interpolated sequence [@Ge2000]. Furthermore, the matrix is unpacked similar to image analysis methods into a feature space vector, with dimensions depend on the resolution and bounds of the states. The algorithm to process the time-domain conditioned data is given in Algorithm \[Algorithm2\]. $markovMatrix = []$ $markovMatrixPrime \gets []$ $currentSlotSet \gets markovMatrixPrime[j]$ $idxIn \gets $ find state containing $currentSlotSet[k-1]$ $idxOut \gets $ find state containing $currentSlotSet[k]$ $markovMatrixPrime[idxIn, idxOut]++$ $markovMatrix \gets markovMatrix + markovMatrixPrime$ $N_i =$ sum along row of $markovMatrix$ $markovMatrix[:,j] \gets \frac{N_{ij}}{N_i}$ The resulting Markov Matrix is unpacked into a feature vector given by: $$\mathbf{P_{\mathit{i}}}=\left[\begin{array}{cccc} p_{11} & p_{12} &... & p_{1r}\\ p_{21} & p_{22} & \cdots & \cdots\\ \vdots & \vdots & \ddots & \vdots\\ p_{r1} & p_{21} &... & p_{rr} \end{array}\right]\Rightarrow x_{i}=\left[\begin{array}{cccccc} p_{11} & p_{12} &... & p_{21} &... & p_{rr}\end{array}\right]$$ Where $\mathbf{P}_{i}$ is the Markov Chain of the $i^{th}$ input training set, and $x_{i}$ is the $i^{th}$ input unfolded training pattern. Feature Space Reduction (ECVA) ------------------------------ The resolution of the state set needs to be small to avoid loss of information resulting from over generalization. However, if the state resolution is too small the sparsity
of the state $ s_{i}$. The transition probability matrix is approximated by converting the chemical element of phosphorus by approximating the transition probability use $ P_{ij}=\nicefrac{N_{ij}}{N_{i}}$. The resulting matrix is often identify as a beginning order Markov Matrix [ @Ross2013 ]. state of matter changes are based on entirely the observation - to - observation amplitude change; the matrix is a representation of the linearly interpolated sequence [ @Ge2000 ]. Furthermore, the matrix is unpacked similar to image analysis method into a feature space vector, with dimensions count on the resolution and bounds of the state of matter. The algorithm to process the clock time - domain conditioned data is given in Algorithm \[Algorithm2\ ]. $ markovMatrix = [ ] $ $ markovMatrixPrime \gets [ ] $ $ currentSlotSet \gets markovMatrixPrime[j]$ $ idxIn \gets $ find department of state containing $ currentSlotSet[k-1]$ $ idxOut \gets $ find state containing $ currentSlotSet[k]$ $ markovMatrixPrime[idxIn, idxOut]++$ $ markovMatrix \gets markovMatrix + markovMatrixPrime$ $ N_i = $ sum along row of $ markovMatrix$ $ markovMatrix[:,j ] \gets \frac{N_{ij}}{N_i}$ The resulting Markov Matrix is unpacked into a feature vector given by: $ $ \mathbf{P_{\mathit{i}}}=\left[\begin{array}{cccc } p_{11 } & p_{12 } &... & p_{1r}\\ p_{21 } & p_{22 } & \cdots & \cdots\\ \vdots & \vdots & \ddots & \vdots\\ p_{r1 } & p_{21 } &... & p_{rr } \end{array}\right]\Rightarrow x_{i}=\left[\begin{array}{cccccc } p_{11 } & p_{12 } &... & p_{21 } &... & p_{rr}\end{array}\right]$$ Where $ \mathbf{P}_{i}$ is the Markov Chain of the $ i^{th}$ input training set, and $ x_{i}$ is the $ i^{th}$ input unfolded training pattern. Feature Space Reduction (ECVA) ------------------------------ The resolution of the state of matter set needs to be belittled to avoid loss of information resulting from over abstraction. However, if the state settlement is too small the sparseness
of the state $s_{i}$. The transiuion probability matrix ms apprkximated by converting the elements lf P by approximating the travsition pgobabilitues nsing $P_{ij}=\nicefrar{H_{ij}}{N_{i}}$. Tmz reshptiny natrix is oftek described as a first orgef Larkov Matrix [@Ross2013]. State changes arq based oj only the obsgrvatpog-to-ogservation amplitude changes; the mztrix iv a representstion of the linearly integpolwted sequence [@Ge2000]. Fkrthermore, jge iqtrix is unpxcked similar to image analysis methods into a featurd spaee vector, wuty dlkensions de'end og the resolubpon and bounds of the states. Thx altorithm to process thx time-domain conditijned data ia given in Algorirhn \[Algmritvm2\]. $mafjovOatdie = []$ $markogMavrixPrime \gsts []$ $currenrSlotSet \gets markofMwntixPrime[j]$ $idsIn \gqtf $ find state containing $currentSlotSet[n-1]$ $isxOut \gets $ find state xontaining $currentSlojSet[k]$ $markjvMatrixPrime[idxIn, idxOut]++$ $markovMatrix \gets markoeMatrmx + mcvhucMwtrixPrime$ $N_i =$ sum along row of $markovMatrix$ $izrlonMatrix[:,j] \gets \frag{N_{ij}}{N_i}$ The resultinb Lathov Matrix is unpacked into a feature veftor giden bt: $$\mathbf{P_{\mwthiy{i}}}=\left[\begin{array}{cccc} p_{11} & p_{12} &... & p_{1r}\\ p_{21} & p_{22} & \cdotf & \cdots\\ \vdots & \vdotd & \ddots & \vbots\\ p_{r1} & p_{21} &... & p_{rr} \end{array}\right]\Rightaxrow x_{j}=\left[\begin{agray}{cccccd} o_{11} & p_{12} &... & p_{21} &... & p_{rr}\ena{argay}\rhght]$$ Whert $\mathbf{P}_{i}$ is the Iarkov Chein oy the $i^{tf}$ inkut trayning set, wnd $x_{l}$ is the $i^{th}$ input knfolbed tsaining pahtern. Feature Space Reduction (ECTE) ------------------------------ The resolutipn of the stace set needs to be fmall to avoid loss oy infofmation reaulting from over deneralizatiot. However, if vhe state resilutuon is guo small the slarsity
of the state $s_{i}$. The transition probability approximated converting the of P by $P_{ij}=\nicefrac{N_{ij}}{N_{i}}$. resulting matrix is described as a order Markov Matrix [@Ross2013]. State changes based on only the observation-to-observation amplitude changes; the matrix is a representation of linearly interpolated sequence [@Ge2000]. Furthermore, the matrix is unpacked similar to image analysis into feature vector, dimensions depend on the resolution and bounds of the states. The algorithm to process the time-domain data is given in Algorithm \[Algorithm2\]. $markovMatrix = $markovMatrixPrime \gets []$ $currentSlotSet markovMatrixPrime[j]$ $idxIn \gets $ find containing $idxOut \gets find containing $markovMatrixPrime[idxIn, idxOut]++$ $markovMatrix markovMatrix + markovMatrixPrime$ $N_i =$ sum along row of $markovMatrix$ $markovMatrix[:,j] \gets \frac{N_{ij}}{N_i}$ The resulting Markov Matrix unpacked into vector given $$\mathbf{P_{\mathit{i}}}=\left[\begin{array}{cccc} & &... & p_{1r}\\ p_{22} & \cdots & \cdots\\ \vdots \ddots & \vdots\\ p_{r1} & p_{21} &... & \end{array}\right]\Rightarrow x_{i}=\left[\begin{array}{cccccc} & p_{12} &... & p_{21} &... p_{rr}\end{array}\right]$$ Where $\mathbf{P}_{i}$ is the Markov Chain of $i^{th}$ input training set, and $x_{i}$ is the $i^{th}$ input unfolded training pattern. Feature Space ------------------------------ The resolution of state set needs be to loss information resulting over generalization. However, if the state resolution is too small the
of the state $s_{i}$. The transition ProbabilitY matrIx iS apPrOximAted By converting thE ElemEnts of P by approximating The trAnSItioN PrObabiLities uSInG $p_{Ij}=\nIcEfRac{n_{iJ}}{n_{i}}$. the reSulTing matRix is often DesCrIbed as a first ORdEr Markov MaTriX [@Ross2013]. State chAngEs are bAsEd oN Only tHe oBservAtion-tO-ObservAtion amplItUDe chanGEs; the maTRIx Is a rEpresentation of thE LiNEarly interpolaTed seqUeNCe [@gE2000]. furTheRmore, the maTrIx is uNPacked sIMiLAR To iMAge analysis meThods into a fEAtuRe spacE vEctOR, with dImensIoNS dePend on the reSoluTion and boUnds of THe stateS. the algoRithm tO prOceSs thE TiMe-DomAiN ConDItIonED daTa is giveN iN ALgoriThm \[ALGORIthm2\]. $MarKovMAtrix = []$ $MarkovMatrixPRimE \getS []$ $CurRentSLotSeT \getS mArkovmatrixprime[J]$ $iDxIn \gets $ find staTe coNtaining $cUrrEnTSlOtset[k-1]$ $iDXOut \geTs $ fInd State coNtaininG $CurReNTsLoTSet[k]$ $markovMatrixPRiME[IdXIn, idxOuT]++$ $markoVmaTrIX \gets marKoVMaTrix + MARkovMAtriXprIme$ $N_i =$ sum Along rOW oF $mArkovMaTrIx$ $markOvmatRix[:,J] \gets \FRac{N_{Ij}}{N_i}$ ThE resultiNg MarKOv Matrix is unpaCKed into a featuRE vECToR GiveN by: $$\Mathbf{P_{\mathIt{i}}}=\lEFt[\beGin{aRRaY}{ccCC} p_{11} & p_{12} &... & p_{1r}\\ P_{21} & p_{22} & \cdoTs & \CDoTS\\ \vdots & \vdots & \ddots & \vdoTs\\ P_{r1} & p_{21} &... & p_{rr} \End{arRay}\right]\RightArrow x_{i}=\lefT[\BEGin{array}{CcccCC} p_{11} & P_{12} &... & P_{21} &... & p_{rr}\end{array}\riGht]$$ WhEre $\mathbf{P}_{I}$ Is the MarKov ChAin of the $I^{th}$ input tRAIning set, And $X_{i}$ iS thE $i^{tH}$ INpUt unfolded traINIng pAtTern. FeaTurE Space REduCtiOn (EcVA) ------------------------------ thE resolutiOn of the sTaTe SeT nEedS to be SMall to avOiD loSs Of iNformATion reSultiNg frOm OvER geNeralizATiON. howeVeR, iF the StaTe ResolUtioN Is tOo small The sparsiTy
of the state $s_{i}$. The transitio n pro bab ili ty mat rixis approximate d byconverting the element s ofPb y ap p ro ximat ing the tr a n sit io npro ba b il ities us ing $P_ {ij}=\nice fra c{ N_{ij}}{N_{i } }$ . The resu lti ng matrix is of ten de sc rib e d asa f irstorderM arkovMatrix [@ Ro s s2013] . Statec h an gesare based on only th e observation-t o-obse rv a ti o n am pli tude chang es ; the matrixi sa r epr e sentation ofthe linearl y in terpol at eds equenc e [@G e2 0 00] . Furthermo re,the matri x is u n packeds imilarto ima geana lysi s m et hod si nto afea t ure space v ec to r, wi th d i m e n sion s d epen d onthe resolutio n a nd b o und s ofthe s tate s. Thealgori thm t oprocess the tim e-do main cond iti on edda ta is giveninAlg orithm\[Algor i thm 2\ ] . $ markovMatrix = []$ $ m a rk ovMatrix Prime\ ge ts []$ $cur re ntS lotS e t \get s ma r ko vMatrixP rime[j ] $$i dxIn \ ge ts $ f in d s tat e con t aini ng $cu rrentSlo tSet[ k -1]$ $idxOut \ g ets $ find st a te c on t aini ng$currentSlo tSet [ k]$$mar k ov Mat r ixPri me[id xI n ,i dxOut]++$ $markovMa tr ix \ge ts ma rkovMatrix +markovMatr i x P rime$ $N _i = $ s u m along row of $mar kovMatrix$ $markovM atrix [:,j] \g ets \frac { N _{ij}}{N _i} $ The re s u lt ing Markov Ma t r ix i sunpacke d i nto a f eat ure ve cto rgiven by: $$\mat hb f{ P_ {\ mat hit{i } }}=\left [\ beg in {ar ray}{ c ccc} p _{11} & p _{ 12 } &. .. & p_ { 1r } \ \ p_ {2 1} & p _{2 2} & \c dots & \ cdots\\ \vdots & \v d ots&\d dots &\vdots\\ p_{r 1} & p_{21}&. ..& p_{r r } \end{ar ray}\right]\Rightarrowx _{i}=\l eft [\beg in{a rray}{ccc ccc } p_{1 1}& p_{12 } &... & p_ {2 1}& . .. &p _ {r r}\ en d{array}\r i g ht] $$ W he re $ \mathbf {P}_{i}$ is the Ma r kov Chain of the $i ^{th } $ i npu t t r ain in g se t , and $x_{i}$ is the $i^{t h} $ i nput unfol d edtr ainingpattern . Fe a ture Sp ace Reduc tion (ECV A) --- - - --- ---------- -------- ---- The resol u ti on of th e stat eset need s to b e sm all t o avoi dloss o f inf or mation r esulting from over gene raliza tion. Ho wever, if th e st ate resol utio n is too s mal l t he sp ars i ty
of_the state_$s_{i}$. The transition probability_matrix is_approximated_by converting_the_elements of P_by approximating the_transition probabilities using $P_{ij}=\nicefrac{N_{ij}}{N_{i}}$._The resulting matrix_is_often described as a first order Markov Matrix [@Ross2013]. State changes are based on_only_the observation-to-observation_amplitude_changes;_the matrix is a representation_of the linearly interpolated sequence_[@Ge2000]. Furthermore,_the matrix is unpacked similar to image analysis_methods_into a feature_space vector, with dimensions depend on the resolution and_bounds of the states. The algorithm_to process the_time-domain_conditioned_data is given in_Algorithm \[Algorithm2\]. $markovMatrix = []$ $markovMatrixPrime \gets_[]$ $currentSlotSet \gets markovMatrixPrime[j]$ $idxIn _\gets $ find state containing $currentSlotSet[k-1]$ $idxOut_\gets $ find state containing $currentSlotSet[k]$_$markovMatrixPrime[idxIn, idxOut]++$ $markovMatrix \gets markovMatrix_+ markovMatrixPrime$_$N_i =$ sum along row_of $markovMatrix$ $markovMatrix[:,j]_\gets \frac{N_{ij}}{N_i}$ The_resulting Markov Matrix_is unpacked into a feature vector_given by: $$\mathbf{P_{\mathit{i}}}=\left[\begin{array}{cccc} p_{11} &_p_{12} &... & p_{1r}\\ p_{21} & p_{22}_&_\cdots & \cdots\\ \vdots_&_\vdots_& \ddots_& \vdots\\ p_{r1} &_p_{21}_&... &_p_{rr} \end{array}\right]\Rightarrow_x_{i}=\left[\begin{array}{cccccc} p_{11} & p_{12} &... & p_{21}_&..._& p_{rr}\end{array}\right]$$ Where $\mathbf{P}_{i}$ is the Markov Chain_of the $i^{th}$ input_training_set, and $x_{i}$ is_the $i^{th}$ input unfolded training_pattern. Feature Space Reduction (ECVA) ------------------------------ The resolution of_the state_set needs_to be small to avoid loss of information resulting from over_generalization. However, if the state resolution_is too small the_sparsity
the notification (see Supplementary Note 4 for how these load profiles were generated). Subsequently, power flows within the distribution network were calculated. Finally, we make the reasonable assumption that the capacity of each line in the distribution network is limited to 10% over the peak power flow in that line under regular circumstances, i.e., when no resident receives the notification from the attacker. Finally, we analyze how the distribution network is affected by the attack. In this analysis, among the many variables that could be considered such as voltage and reactive power flows, we focused on line capacity limits since they are the most dominantly affected by energy consumption patterns, and are the most critical to power system stability [@quiros2018electric; @coignard2019will]. Once a line overloads it goes offline, leading all lines below it in the power distribution tree to go offline as well. By averaging over 100 such simulations, we obtained the fraction of residences suffering from a blackout that were depicted in Fig. \[LondonMaps\]a. The illustrations in Fig. \[LondonMaps\]b and \[LondonMaps\]c represent a single simulation each. Ethics statement {#ethics-statement.unnumbered} ================ The research was approved by the Institutional Review Boards of New York University-Abu Dhabi and the National University of Singapore. Data availability {#data-availability.unnumbered} ================= All power systems data that were used in this study are available to download from [@website_data]. The codes for generating the specific plots are available from the corresponding authors upon request. --- abstract: 'We develop a general deformation theory of objects in homotopy and derived categories of DG categories. The main result is a general pro-representability theorem for the corresponding deformation functor.' address: - 'Department of Mathematics, Indiana University, Bloomington, IN 47405, USA' - 'Steklov Mathematical Institute, 8 Gubkina St. Moscow, Russia' author: - 'Valery A. Lunts' - Dmitri Orlov title: DG deformation theory of objects in homotopy and derived categories I --- Introduction ============ It is well known (\[De\],\[Dr2\]) that for many mathematical objects $X$ (defined over a field of characteristic zero) the formal deformation theory of $X$ is controlled by a DG Lie algebra $\frak{g}=\frak{g}(X)$ of
the notification (see Supplementary Note   4 for how these load profiles were generate). Subsequently, exponent flows within the distribution network were calculated. last, we make the reasonable presumption that the capacitance of each line in the distribution net is specify to 10% over the peak might flow in that line under unconstipated circumstance, i.e., when no resident pick up the notification from the attacker. Finally, we analyze how the distribution net is affected by the attack. In this analysis, among the many variable that could be considered such as voltage and reactive world power flows, we concentrate on line capacity limits since they are the most dominantly affected by department of energy consumption patterns, and are the most critical to power system stability   [ @quiros2018electric; @coignard2019will ]. Once a lineage overloads it goes offline, leading all lines below it in the power distribution tree to go offline as well. By averaging over 100 such simulations, we obtained the fraction of residence suffering from a blackout that were portray in Fig.   \[LondonMaps\]a. The illustrations in Fig.   \[LondonMaps\]b and \[LondonMaps\]c represent a single simulation each. Ethics instruction { # ethics-statement.unnumbered } = = = = = = = = = = = = = = = = The research was approved by the Institutional Review Boards of New York University - Abu Dhabi and the National University of Singapore. Data availability { # data-availability.unnumbered } = = = = = = = = = = = = = = = = = All power system data that were used in this study are available to download from   [ @website_data ]. The codes for generating the specific plots are available from the equate authors upon request. --- abstract:' We develop a general deformation hypothesis of objects in homotopy and derive class of DG categories. The main result is a general pro - representability theorem for the comparable deformation functor.' address: -' Department of Mathematics, Indiana University, Bloomington, IN 47405, USA' -' Steklov Mathematical Institute, 8 Gubkina St. Moscow, Russia' author: -' Valery A.   Lunts' - Dmitri Orlov title: DG deformation theory of objects in homotopy and derive categories I --- insertion = = = = = = = = = = = = It is well known (\[De\],\[Dr2\ ]) that for many mathematical object $ X$ (defined over a battlefield of characteristic zero) the formal distortion theory of $ X$ is master by a DG Lie algebra $ \frak{g}=\frak{g}(X)$ of
thf notification (see Suppltmentary Note 4 for how thxse loas profilds were generated). Subsequentpy, powee flows within the disgribution network wert calculated. Finally, we make ths reavinable assumptlon that tha capacity of aazh line in the distribution network if limitrd to 10% over the keak kowqr fmow in that line under regular cirdumstanbes, i.e., when no rexident receives the notififatiln from the attackfr. Finally, wg anwoyze how the distributpmn network is affected by the attack. In tfis aualysis, amobg thf many variaules trat could be considesed sucn as voltage akd reectice power flows, we focnsed on line capaciti limits shnee they are the most eoninanjly axfecgwd cy tnecgy consulptmon patterna, and are tye most critical to pjqer system stzbilitr [@qtiros2018electric; @coignard2019will]. Once a line mvedloads it goes offline, oeading all lines bellw it in ehe power distribution tree to go offline as well. By atefagnkn ovde 100 such simulations, we obtained the fraction of dexicences sufferikg from a blackout tjay were depictea in Fnf. \[LkndonMaps\]a. The illkstratijns ib Fig. \[LondjnMals\]b and \[LondonMaps\]c represebt a single fumulation each. Ethies statement {#ethocs-ststement.unnumbered} ================ The rereardh was apprlved by tgd Institutional Fevpew Toards of New York Universyty-Abu Dhebi aud the Nxtiomal Unyversity ov Sinndpore. Data availabipity {#bata-aeailabilitj.unnumbered} ================= All power systems dave that were uxeg it this scudy ave available to download from [@cebsite_dcta]. Thd codes fog generatmng the specyfic plots ara available fcom the cjrrewponeing augfors upon requrst. --- abstrcet: 'We devwlop a general defprmxfion theory of ibjtcrs in homotopy ana dqrpvev catqcories of DG catdgofoes. Tfe main resmlt is s general pro-represettabjlity theorem for yhc correspinding dqformation fumctor.' address: - 'Deparument mf Oathekatycs, Indiana University, Bloominfton, IN 47405, KSA' - 'Steklov Matremabicaj Institutz, 8 Gubkina St. Moscow, Russia' author: - 'Valery A. Lunts' - Dmitri Orliv title: DG deformatnok theory of pbjeces in hommtopy and derived caregories I --- Introdugtion ============ It is well known (\[Se\],\[Dr2\]) tvat flr many mathematical objects $X$ (defined over a field of characteristic zeri) the howmal deformztiom themrv oy $X$ is sontcouled by a DG Lie algebra $\frak{g}=\frak{g}(X)$ of
the notification (see Supplementary Note 4 for load were generated). power flows within Finally, make the reasonable that the capacity each line in the distribution network limited to 10% over the peak power flow in that line under regular i.e., when no resident receives the notification from the attacker. Finally, we analyze the network affected the attack. In this analysis, among the many variables that could be considered such as voltage reactive power flows, we focused on line capacity since they are the dominantly affected by energy consumption and the most to system [@quiros2018electric; @coignard2019will]. Once line overloads it goes offline, leading all lines below it in the power distribution tree to go as well. over 100 simulations, obtained fraction of residences a blackout that were depicted in illustrations in Fig. \[LondonMaps\]b and \[LondonMaps\]c represent a simulation each. statement {#ethics-statement.unnumbered} ================ The research was by the Institutional Review Boards of New York Dhabi and the National University of Singapore. Data availability {#data-availability.unnumbered} ================= All power systems data used in this study available to download [@website_data]. codes generating specific plots available from the corresponding authors upon request. --- abstract: 'We develop general deformation theory of objects in homotopy and derived categories categories. main result is general pro-representability theorem for corresponding functor.' address: - 'Department Indiana Bloomington, - Mathematical 8 Gubkina St. Moscow, author: - 'Valery A. Lunts' Dmitri Orlov title: DG homotopy and derived categories I --- Introduction ============ is well known (\[De\],\[Dr2\]) that for many objects $X$ (defined over a field of characteristic zero) the formal deformation of $X$ by a DG Lie algebra $\frak{g}=\frak{g}(X)$ of
the notification (see SupplemEntary Note 4 For hoW thEse LoAd prOfilEs were generateD). subsEquently, power flows withIn the DiSTribUTiOn netWork werE CaLCUlaTeD. FInaLlY, We Make tHe rEasonabLe assumptiOn tHaT the capacity OF eAch line in tHe dIstribution nEtwOrk is lImIteD To 10% oveR thE peak Power fLOw in thAt line undEr REgular CIrcumstANCeS, i.e., wHen no resident receIVeS The notificatioN from tHe ATtACKer. finAlly, we analYzE how tHE distriBUtION NetWOrk is affected By the attack. iN thIs analYsIs, aMOng the Many vArIAblEs that could Be coNsidered sUch as vOLtage anD ReactivE power FloWs, wE focUSeD oN liNe CApaCItY liMIts Since theY aRe The moSt doMINANtly AffEcteD by enErgy consumptiOn pAtteRNs, aNd are The moSt crItIcal tO power SysteM sTability [@quiros2018eLectRic; @coignaRd2019wIlL]. OnCe A line OVerloaDs iT goEs offliNe, leadiNG alL lINES bElow it in the power diStRIBuTion tree To go ofFLiNe AS well. By aVeRagIng oVER 100 such SimuLAtIons, we obTained THe FrAction oF rEsidenCeS suFfeRing fROm a bLackouT that werE depiCTed in Fig. \[LondonmAps\]a. The illustRAtIONs IN Fig. \[lonDonMaps\]b and \[londONMapS\]c rePReSenT A singLe simUlATiON each. Ethics statemenT {#eThics-sTatemEnt.unnumbered} ================ the researcH WAS approveD by tHE INStitutional RevIew BoArds of New YORk UniverSity-ABu Dhabi aNd the NatiONAl UniverSitY of sinGapORE. DAta availabiliTY {#Data-AvAilabilIty.UnnumbeRed} ================= all PowEr sYsTems data tHat were uSeD iN tHiS stUdy arE AvailablE tO doWnLoaD from [@WEbsite_Data]. THe coDeS fOR geNeratinG ThE SPeciFiC pLots Are AvAilabLe frOM thE corresPonding auThoRS upoN rEqUest. --- absTract: 'We develoP a General defOrMatIon theORY of objecTs in homotopy and derived cATegorieS of dG catEgorIes. The maiN reSult is A geNEral prO-repreSentaBiLitY THeoreM FOr The CoRrespondinG DEfoRmatiOn FuncTor.' addrEss: - 'Department of MatHEmaTics, Indiana UnIveRsitY, bLoOmiNGtON, IN 47405, uSa' - 'steKLOv Mathematical INstitute, 8 GuBkINa st. Moscow, RuSSia' AuThor: - 'ValEry A. LunTs' - DmiTRi Orlov Title: DG deFormation ThEory OF ObjEcts in homoTopy and dErived catEGorieS i --- INtrodUctIon ============ It iS wEll Known (\[de\],\[Dr2\]) thAT foR many MathemAtIcal obJects $x$ (dEfined ovEr a field of characteristiC zero) tHe forMal DeformatiOn tHEorY of $X$ is conTrolLed by a DG LiE alGebRa $\fraK{g}=\fRAk{g}(X)$ oF
the notification (see Sup plementary Note  4for h ow t hese load profiles were generated). Subsequen tly,po w er f l ow s wit hin the di s t rib ut io n n et w or k wer e c alculat ed. Finall y,we make the re a so nable assu mpt ion that the ca pacity o f e a ch li nein th e dist r ibutio n network i s limit e d to 10 % ov er t he peak power flo w i n that line und er reg ul a rc i rcu mst ances, i.e ., when no resi d en t r ece i ves the notif ication fro m th e atta ck er. Final ly, w ea nal yze how the dis tribution netwo r k is af f ected b y theatt ack . In th is an al y sis , a mon g th e many v ar ia blesthat c o u ld b e c onsi dered such as volt age and rea ctive powe r fl ow s, we focus ed on l ine capacity li mits since th eyar e t he most domina ntl y a ffected by ene r gyco n s u mp tion patterns, and a r e t he mostcritic a lto power sy st emstab i l ity [ @qui r os 2018elec tric;@ co ig nard201 9w ill].On cea l ine o v erlo ads it goes of fline , leading all l i nes below iti nt h ep ower di stributiontree to g o of f li nea s wel l. By a v er a ging over 100 suchsi mulati ons,we obtained t he fractio n o f reside nces su f fering from ablack out that w e re depic ted i n Fig. \ [LondonMa p s \]a. The il lus tra tio n s i n Fig. \[Lond o n Maps \] b and \ [Lo ndonMap s\] c r epr ese nt a single simulat io nea ch . Ethic s stateme nt {# et hic s-sta t ement. unnum bere d} = = === ======= = == = The r es earc h w as appr oved bythe Ins titutiona l R e view B oa rds ofNew York Univ er sity-Abu D ha biand th e National University of Singapor e . Data av ailab ilit y {#data- ava ilabil ity . unnumb ered}===== == === = = ===== Al l p ow er systems d ata that w ereused in this study are av a ila ble to downlo adfrom [ @w ebs i te _ dat a] . Th e codes for gener ating thesp e ci fic plotsa reav ailable from t he co r respond ing autho rs upon r eq uest . -- - abstract : 'We de velop a g e neral de forma tio n theo ry of obje cts in hom otopy and d er ived c atego ri es of DG categories. The main r esultis agen eral pro- rep r ese ntability the orem for t hecor respo ndi n g def orma t io n f u nctor .' a d dress: -' De par t m en t of Mathem a t i cs, Indi ana Univer sity , Bloomington, IN 47405, USA' -'Ste k l ovMat h emat ic al Institute,8 G ub k i na St. M os cow, Russia ' author :- 'Val ery A.  Lunts ' - Dmi t r iO rlov t itle : D G deforma tio nt heory o fob j ects i n ho mo topy a nd der i vedc a tegories I --- Intro d u ction === ===== == == Iti s we ll known ( \[De\],\[Dr 2\]) t hatfor m any mat he matica l o bj ects $X$ ( d efined ov er afield o fchar act eristi c ze r o ) the for ma l d eformatio n th e or yo f $ X$ i s con tr olle d by a DG Lie alge bra $\frak{ g} =\f r a k{g}(X ) $o f
the_notification (see_Supplementary Note 4 for how_these load_profiles_were generated)._Subsequently,_power flows within_the distribution network_were calculated. Finally, we_make the reasonable_assumption_that the capacity of each line in the distribution network is limited to 10%_over_the peak_power_flow_in that line under regular_circumstances, i.e., when no resident_receives the_notification from the attacker. Finally, we analyze how the_distribution_network is affected_by the attack. In this analysis, among the many_variables that could be considered such_as voltage and_reactive_power_flows, we focused on_line capacity limits since they are_the most dominantly affected by energy_consumption patterns, and are the most critical_to power system stability [@quiros2018electric; @coignard2019will]. Once_a line overloads it goes_offline, leading_all lines below it in_the power distribution_tree to_go offline as_well. By averaging over 100 such_simulations, we obtained_the fraction of residences suffering from_a_blackout that were_depicted_in_Fig. \[LondonMaps\]a. The_illustrations in Fig. \[LondonMaps\]b_and_\[LondonMaps\]c represent_a_single simulation each. Ethics statement {#ethics-statement.unnumbered} ================ The research_was_approved by the Institutional Review Boards of_New York University-Abu Dhabi_and_the National University of_Singapore. Data availability {#data-availability.unnumbered} ================= All power systems_data that were used in this_study are_available to_download from [@website_data]. The codes for generating the specific plots are available_from the corresponding authors upon request. _--- abstract: 'We develop a_general deformation_theory_of objects in_homotopy_and derived_categories of DG categories. The main result_is a_general pro-representability theorem for the corresponding_deformation functor.' address: - 'Department of_Mathematics,_Indiana University, Bloomington, IN 47405, USA' -_'Steklov Mathematical Institute, 8 Gubkina St._Moscow, Russia' author: - 'Valery A. Lunts' - Dmitri_Orlov title:_DG_deformation theory of objects in_homotopy and derived categories I --- Introduction ============ It is_well known (\[De\],\[Dr2\])_that for many mathematical objects $X$ (defined_over_a field of characteristic zero) the_formal_deformation theory of $X$ is controlled_by_a_DG Lie algebra $\frak{g}=\frak{g}(X)$ of
interesting to note that even though $\delta V$ is added only in region C, the scattering states (WF1, WF2) are affected in all regions L, C, and R reflecting the scattering processes. To correlate the robustness of transmission $T(E,0)$ against the strength of the disorder potential $\delta V$, we have calculated configuration C8 (see Table-\[tab1\]) where $\delta V$ is randomly drawn from ranges $[0,U]$, results are shown in the inset of Fig.\[fig2\]c versus $U$. At Fermi energy $E=0$ (blue dotted line), $T(E,0) = 2$ until $U$ reaches $\sim 0.2$eV at which $T(E,0)$ substantially reduces (inset of Fig.\[fig2\]c). By investigating which atomic orbitals contribute to the scattering wave functions and also the real space WF1/WF2, the scattering states are found to be unconfined from the surfaces at this value of $U$, namely the helical states are destroyed. When $U$ increases further, $T(E,0)$ oscillates by the random disorder scattering configurations. At higher energies, we found that the helical edge states are much more robust. For instance, at $E=0.1$eV (solid curve in the inset of Fig.\[fig2\]c), $T(E,0)\approx 2$ until $U \sim 0.5$eV. This is reasonable since electrons with higher energy are less affected by potential perturbations. So far we focused on transmission in the x-direction, $T(E,k_y=0)$. For the film, a 2D transmission $T(E)$ is obtained by integrating $T(E,k_y)$ over $k_y$, which gives the 2D conductance $G(E) = T(E)\times G_o$ where $G_o\equiv e^2/h$. The inset of Fig.\[fig2\]b plots $G(E)$. For energies outside the bulk gap, e.g. $E < 0$ or $E \gtrsim 0.27$eV, bulk propagating modes participate transport and therefore $G$ decreases according to the increase of the disorder. Within the gap, $G(E)$ changes negligibly for weak disorder C4 and C6, but more significantly in strong disorder C5 and C7. In contrast to $
interesting to note that even though $ \delta V$ is added only in region C, the scatter state (WF1, WF2) are affected in all regions L, C, and R chew over the scattering processes. To correlate the robustness of infection $ T(E,0)$ against the persuasiveness of the disorder electric potential $ \delta V$, we have account configuration C8 (experience Table-\[tab1\ ]) where $ \delta V$ is randomly drawn from ranges $ [ 0,U]$, solution are indicate in the inset of Fig.\[fig2\]c versus $ U$. At Fermi energy $ E=0 $ (blue dotted line), $ T(E,0) = 2 $ until $ U$ strive $ \sim 0.2$eV at which $ T(E,0)$ substantially reduces (inset of Fig.\[fig2\]c). By investigate which atomic orbitals contribute to the scattering wave functions and besides the real distance WF1 / WF2, the scattering states are found to be unconfined from the surface at this value of $ U$, namely the helical states are destroyed. When $ U$ increases further, $ T(E,0)$ oscillates by the random disorderliness disperse configurations. At higher energies, we found that the helical edge states are much more robust. For case, at $ E=0.1$eV (solid curve in the gusset of Fig.\[fig2\]c), $ T(E,0)\approx 2 $ until $ U \sim 0.5$eV. This is fair since electron with higher energy are less affected by potential perturbations. So far we focused on transmission in the ten - direction, $ T(E, k_y=0)$. For the film, a 2D transmission $ T(E)$ is obtained by integrating $ T(E, k_y)$ over $ k_y$, which gives the 2D conductance $ G(E) = T(E)\times G_o$ where $ G_o\equiv e^2 / h$. The gusset of Fig.\[fig2\]b plots $ G(E)$. For energies outside the bulk gap, e.g. $ east < 0 $ or $ E \gtrsim 0.27$eV, bulk propagate modes participate transport and therefore $ G$ decreases harmonize to the increase of the disorder. Within the gap, $ G(E)$ change negligibly for unaccented disorder C4 and C6, but more significantly in solid disorder C5 and C7. In line to $
inheresting to note that eyen though $\delta V$ is avded onmy in reeion C, the scattering states (WD1, WF2) qre affected in all reeions L, C, and R rwflerting the scattecjng progzsses. Fl coxrxlate the robusjness of tratsmission $T(E,0)$ acakndt the strength of the disorder potqntial $\cepta V$, we have salcllwted bokfiguration C8 (see Table-\[tab1\]) where $\delta N$ is randomly draen from ranges $[0,U]$, results age sjown in the inset lf Fig.\[fig2\]c cerstw $U$. At Fermi energy $E=0$ (blue dotted mine), $T(E,0) = 2$ until $U$ reaches $\sim 0.2$eX at chich $T(E,0)$ suvsranhhally reducxs (insvt of Fig.\[fig2\]c). By invevtigatimg which atomig orbmtalw contribute to the srattering wave functyons and dlao the real space WD1/WF2, tve swattdeine suatxs zre foknd to be uncknfined fron the surfaces at tnif value of $U$, nzmely ehq helical states are destroyed. When $U$ itcrsases further, $T(E,0)$ oscillqtes by the random didorder scwttering configurations. At higher energies, we foutd thet tht mclicxo fdge states are much more robust. For instance, zt $E=0.1$vV (solid curve in the inset pf Fod.\[fig2\]c), $T(E,0)\approb 2$ untnm $H \sim 0.5$eV. This is rfasonabje sibce electwons with higher energy are lesw affected bj porential perturbatilns. So far wz focuxed om transmission in the x-birectjon, $T(E,k_y=0)$. Fog the filj, a 2D transmissiov $T(V)$ is obtained by integrating $T(Q,k_y)$ over $j_y$, wkich givds tne 2D cjnductance $G(E) = B(A)\times G_o$ where $G_o\fquiv e^2/v$. The inseh of Fig.\[fig2\]b plots $G(E)$. For energmxs outside thg bglk gap, e.g. $Z < 0$ or $E \gtrsim 0.27$eV, fulk propagatiug modes partkcipate trznsport and therefjre $G$ decreasad according vo the insreawe od the dkrorder. Within yhe gap, $G(V)$ ehanges nwgligibly for weak dirkrder C4 and C6, bbu mire significantky kn ftgonj disjsder C5 and C7. In zongtast go $
interesting to note that even though $\delta added in region the scattering states all L, C, and reflecting the scattering To correlate the robustness of transmission against the strength of the disorder potential $\delta V$, we have calculated configuration (see Table-\[tab1\]) where $\delta V$ is randomly drawn from ranges $[0,U]$, results are in inset Fig.\[fig2\]c $U$. At Fermi energy $E=0$ (blue dotted line), $T(E,0) = 2$ until $U$ reaches $\sim 0.2$eV which $T(E,0)$ substantially reduces (inset of Fig.\[fig2\]c). By which atomic orbitals contribute the scattering wave functions and the space WF1/WF2, scattering are to be unconfined the surfaces at this value of $U$, namely the helical states are destroyed. When $U$ increases further, oscillates by disorder scattering At energies, found that the states are much more robust. For (solid curve in the inset of Fig.\[fig2\]c), $T(E,0)\approx until $U 0.5$eV. This is reasonable since electrons higher energy are less affected by potential perturbations. far we focused on transmission in the x-direction, $T(E,k_y=0)$. For the film, a 2D transmission obtained by integrating $T(E,k_y)$ $k_y$, which gives 2D $G(E) T(E)\times where $G_o\equiv The inset of Fig.\[fig2\]b plots $G(E)$. For energies outside the bulk e.g. $E < 0$ or $E \gtrsim 0.27$eV, bulk propagating transport therefore $G$ decreases to the increase of disorder. the gap, $G(E)$ changes weak C4 more in disorder C5 and C7. contrast to $
interesting to note that even Though $\deltA V$ is aDdeD onLy In reGion c, the scattering STateS (WF1, WF2) are affected in all rEgionS L, c, And R REfLectiNg the scATtERIng PrOcEssEs. tO cOrrelAte The robuStness of trAnsMiSsion $T(E,0)$ againST tHe strength Of tHe disorder poTenTial $\deLtA V$, wE Have cAlcUlateD confiGUratioN C8 (see TablE-\[tAB1\]) where $\DElta V$ is RANdOmly Drawn from ranges $[0,U]$, rESuLTs are shown in thE inset Of fIg.\[FIG2\]c vErsUs $U$. At Fermi EnErgy $E=0$ (BLue dottED lINE), $t(E,0) = 2$ uNTil $U$ reaches $\siM 0.2$eV at which $T(e,0)$ SubStantiAlLy rEDuces (iNset oF FIG.\[fiG2\]c). By investiGatiNg which atOmic orBItals coNTribute To the sCatTerIng wAVe FuNctIoNS anD AlSo tHE reAl space Wf1/Wf2, tHe scaTterING STateS arE fouNd to bE unconfined frOm tHe suRFacEs at tHis vaLue oF $U$, NamelY the heLical StAtes are destroyeD. WheN $U$ increasEs fUrTheR, $T(e,0)$ osciLLates bY thE raNdom disOrder scATteRiNG COnFigurations. At higheR eNERgIes, we fouNd that THe HeLIcal edge StAteS are MUCh morE robUSt. for instaNce, at $E=0.1$Ev (sOlId curve In The insEt Of FIg.\[fIg2\]c), $T(E,0)\APproX 2$ until $u \sim 0.5$eV. ThIs is rEAsonable since eLEctrons with hiGHeR ENeRGy arE leSs affected bY potENtiaL perTUrBatIOns. So Far we FoCUsED on transmission in thE x-DirectIon, $T(E,K_y=0)$. For the film, a 2d transmissION $t(E)$ is obtaIned BY iNTegrating $T(E,k_y)$ oVer $k_y$, Which gives THe 2D conduCtancE $G(E) = T(E)\timEs G_o$ where $g_O\Equiv e^2/h$. THe iNseT of fig.\[FIG2\]b Plots $G(E)$. For eneRGIes oUtSide the BulK gap, e.g. $E < 0$ Or $E \GtrSim 0.27$EV, bUlK propagatIng modes PaRtIcIpAte TransPOrt and thErEfoRe $g$ deCreasES accorDing tO the InCrEAse Of the diSOrDER. WitHiN tHe gaP, $G(E)$ ChAnges NeglIGibLy for weAk disordeR C4 aND C6, buT mOrE signifIcantly in stroNg Disorder C5 aNd c7. In ContraST To $
interesting to note thateven thoug h $\d elt a V $is a dded only in regio n C,the scattering states(WF1, W F 2) a r eaffec ted ina ll r egi on sL,C, an d R r efl ectingthe scatte rin gprocesses. T ocorrelatethe robustnessoftransm is sio n $T(E ,0) $ aga inst t h e stre ngth of t he disord e r poten t i al $\d elta V$, we havec al c ulated configu ration C 8 ( s e e T abl e-\[tab1\] )where $\delta V$ i s ra n domly drawn f rom ranges$ [0, U]$, r es ult s are s hownin the inset of F ig.\ [fig2\]cversus $U$. At Fermi e nergy$E= 0$(blu e d ot ted l i ne) , $ T(E , 0)= 2$ unt il $ U$ re ache s $ \ sim0.2 $eVat wh ich $T(E,0)$sub stan t ial ly re duces (in se t ofFig.\[ fig2\ ]c ). By investiga ting which at omi corb it als c o ntribu tetothe sca ttering wav ef u n ct ions and also there a l s pace WF1 /WF2,t he s c attering s tat es a r e foun d to be unconfi ned fr o mth e surfa ce s at t hi s v alu e of$ U$,namely the hel icals tates are dest r oyed. When $U $ i n c re a sesfur ther, $T(E, 0)$o scil late s b y t h e ran dom d is o rd e r scattering config ur ations . Athigher energi es, we fou n d that the hel i ca l edge states a re mu ch more ro b ust. For inst ance, at $E=0.1$e V (solid c urv e i n t hei n se t of Fig.\[fi g 2 \]c) ,$T(E,0) \ap prox 2$ un til $U \s im 0.5$eV.This isre as on ab lesince electron swit hhig her e n ergy a re le ss a ff ec t edby pote n ti a l per tu rb atio ns. So fa r we foc used on transmis sio n inth ex-direc tion, $T(E,k_ y= 0)$. For t he fi lm, a2 D transmi ssion $T(E)$ is obtaine d by int egr ating $T( E,k_y)$ o ver $k_y$ , w h ich gi ves th e 2Dco ndu c t ance$ G (E ) = T (E)\timesG _ o$where $ G_o\ equiv e ^2/h$. The inset o f Fi g.\[fig2\]b p lot s $G ( E )$ . F o re ner gi e s o u t side the bulk g ap, e.g. $ E< 0 $ or $E \g t rsi m0.27$eV , bulkpropa g ating m odes part icipate t ra nspo r t an d therefor e $G$ de creases a c cordi n gto th e i ncreas eofthe d isorde r . W ithin the g ap , $G(E )$ ch an ges negl igibly for weak disorde r C4 a nd C6 , b ut more s ign i fic antly instro ng disorde r C 5 a nd C7 . I n cont rast to $
interesting_to note_that even though $\delta_V$ is_added_only in_region_C, the scattering_states (WF1, WF2)_are affected in all_regions L, C,_and_R reflecting the scattering processes. To correlate the robustness of transmission $T(E,0)$ against the strength_of_the disorder_potential_$\delta_V$, we have calculated configuration_C8 (see Table-\[tab1\]) where $\delta_V$ is_randomly drawn from ranges $[0,U]$, results are shown_in_the inset of_Fig.\[fig2\]c versus $U$. At Fermi energy $E=0$ (blue dotted_line), $T(E,0) = 2$ until $U$_reaches $\sim 0.2$eV_at_which_$T(E,0)$ substantially reduces (inset_of Fig.\[fig2\]c). By investigating which atomic_orbitals contribute to the scattering wave_functions and also the real space WF1/WF2,_the scattering states are found to_be unconfined from the surfaces_at this_value of $U$, namely the_helical states are_destroyed. When_$U$ increases further,_$T(E,0)$ oscillates by the random disorder_scattering configurations. At_higher energies, we found that the_helical_edge states are_much_more_robust. For_instance, at $E=0.1$eV_(solid_curve in_the_inset of Fig.\[fig2\]c), $T(E,0)\approx 2$ until_$U_\sim 0.5$eV. This is reasonable since electrons_with higher energy are_less_affected by potential perturbations. So_far we focused on transmission_in the x-direction, $T(E,k_y=0)$. For the_film, a_2D transmission_$T(E)$ is obtained by integrating $T(E,k_y)$ over $k_y$, which gives the_2D conductance $G(E) = T(E)\times G_o$_where $G_o\equiv e^2/h$. The_inset of_Fig.\[fig2\]b_plots $G(E)$. For_energies_outside the_bulk gap, e.g. $E < 0$ or_$E \gtrsim_0.27$eV, bulk propagating modes participate transport_and therefore $G$ decreases_according_to the increase of the disorder._Within the gap, $G(E)$ changes negligibly_for weak disorder C4 and_C6,_but_more significantly in strong disorder_C5 and C7. In contrast to_$
s}{d}$. Now recall that ${\mathcal F}^{-1} : L^1 \to L^{\infty}$ and ${\mathcal F}^{-1} : L^2 \to L^2$. Therefore, by real interpolation $$\begin{aligned} {\mathcal F}^{-1} : (L^1, L^2)_{\theta, 2} \to (L^{\infty}, L^2)_{\theta, 2}\end{aligned}$$ which, by is exactly the statement that $$\begin{aligned} {\mathcal F}^{-1} : L^{{\alpha}, 2}({\mathbb R}^d) \to L^{{\beta},2}({\mathbb R}^d)\end{aligned}$$ where $\frac{1}{{\alpha}} = 1 - \frac{\theta}{2}$ and $\frac{1}{{\beta}} = \frac{\theta}{2}$ and we notice that $\frac{1}{{\alpha}}+ \frac{1}{{\beta}} =1$. Hence, with $\frac{1}{{\gamma}} = \frac{1}{2} + \frac{s}{d}$ we have that ${\mathcal F}f \in L^{{\gamma}, 2}({\mathbb R}^d)$ which implies that $f \in L^{{\gamma}^{\prime}, 2}({\mathbb R}^d)$ where $\frac{1}{{\gamma}^{\prime}} =\frac{1}{2} - \frac{s}{d}$ which is exactly. The $L^p$ and Besov space versions of this statement are slightly more complicated to prove as they require additional facts from real interpolation theory and we refer the reader to [@Tar] for a detailed proof. We also need the following version of the Calderon-Zygmund theorem for Lorentz spaces. \[CZ\] Let $T$ be a Calderon-Zygmund operator. Then $T: L^{p,r} \to L^{p,r}$ for $1<p<\infty$ and $1\le r\le \infty$, $$\begin{aligned} \|T f\|_{L^{p,r}} \lesssim \|f\|_{L^{p,r}}\end{aligned}$$ where the constant above does not depend on $r$. This extension of the Calderon-Zygmund theorem is an easy consequence of the $L^p$ version given the following interpolation theorem of Calderon, see  [@Ber-Lof Theorem 5.3.4]. \[CZ I\]
s}{d}$. Now recall that $ { \mathcal F}^{-1 }: L^1 \to L^{\infty}$ and $ { \mathcal F}^{-1 }: L^2 \to L^2$. Therefore, by real interjection $ $ \begin{aligned } { \mathcal F}^{-1 }: (L^1, L^2)_{\theta, 2 } \to (L^{\infty }, L^2)_{\theta, 2}\end{aligned}$$ which, by is precisely the statement that $ $ \begin{aligned } { \mathcal F}^{-1 }: L^{{\alpha }, 2}({\mathbb R}^d) \to L^{{\beta},2}({\mathbb R}^d)\end{aligned}$$ where $ \frac{1}{{\alpha } } = 1 - \frac{\theta}{2}$ and $ \frac{1}{{\beta } } = \frac{\theta}{2}$ and we notice that $ \frac{1}{{\alpha}}+ \frac{1}{{\beta } } = 1$. therefore, with $ \frac{1}{{\gamma } } = \frac{1}{2 } + \frac{s}{d}$ we have that $ { \mathcal F}f \in L^{{\gamma }, 2}({\mathbb R}^d)$ which incriminate that $ f \in L^{{\gamma}^{\prime }, 2}({\mathbb R}^d)$ where $ \frac{1}{{\gamma}^{\prime } } = \frac{1}{2 } - \frac{s}{d}$ which is exactly. The $ L^p$ and Besov outer space versions of this statement are slenderly more complicated to rise as they require extra facts from real interpolation theory and we denote the reader to [ @Tar ] for a detailed proof. We also need the following adaptation of the Calderon - Zygmund theorem for Lorentz spaces. \[CZ\ ] Let $ T$ be a Calderon - Zygmund operator. Then $ triiodothyronine: L^{p, r } \to L^{p, r}$ for $ 1 < p<\infty$ and $ 1\le r\le \infty$, $ $ \begin{aligned } \|T f\|_{L^{p, r } } \lesssim \|f\|_{L^{p, r}}\end{aligned}$$ where the constant above does not depend on $ r$. This extension of the Calderon - Zygmund theorem is an comfortable consequence of the $ L^p$ version give the following interpolation theorem of Calderon, see   [ @Ber - Lof Theorem 5.3.4 ]. \[CZ I\ ]
s}{d}$. Jow recall that ${\mathcal N}^{-1} : L^1 \to L^{\infty}$ aue ${\mathral F}^{-1} : M^2 \to L^2$. Tferefore, by real interpolatiln $$\begib{aligned} {\mathcal F}^{-1} : (L^1, L^2)_{\gheta, 2} \to (L^{\infty}, O^2)_{\theua, 2}\end{aligned}$$ whirg, by is exactmn the wtatement that $$\begin{aligted} {\mathcal F}^{-1} : N^{{\aupka}, 2}({\mathbb R}^d) \to L^{{\beta},2}({\mathbb R}^d)\end{alidned}$$ whrrf $\frac{1}{{\alpha}} = 1 - \frab{\treta}{2}$ and $\frac{1}{{\beta}} = \frac{\theta}{2}$ and we nofice thet $\frac{1}{{\alpha}}+ \frsc{1}{{\beta}} =1$. Hence, with $\frac{1}{{\gamla}} = \vrac{1}{2} + \frac{s}{d}$ we hage that ${\matycal D}f \in L^{{\gamma}, 2}({\mathbb R}^d)$ which impligs that $f \in L^{{\gamma}^{\prime}, 2}({\mathbb F}^d)$ whzre $\frac{1}{{\gamna}^{\prilg}} =\frac{1}{2} - \frac{w}{d}$ whpch is exactln. Nhe $L^p$ dnd Bespv space versipns of this statement are snightly more complycated to pxove as they require qdeitiotal xactr frum deel jnterpllavion theory and we refwr the reader to [@Tat] sir a detailed proof. Re also need the following version of the Camderon-Zygmund theorem fir Lorentz spaces. \[CZ\] Lgt $T$ be a Salderon-Zygmund operator. Then $T: L^{p,r} \to L^{p,r}$ for $1<p<\itfty$ ena $1\lt v\jd \ijfty$, $$\begin{aligned} \|T f\|_{L^{p,r}} \lesssim \|f\|_{L^{p,r}}\end{aligneq}$$ wnege the constant anove does not depemd om $r$. This extenskon of ths Calderon-Zygmund hheorem is ab easy cogseqience of the $L^p$ version givwn the follocint interpolation thzorem of Calberon, xee  [@Brr-Lof Theorem 5.3.4]. \[CZ I\]
s}{d}$. Now recall that ${\mathcal F}^{-1} : L^{\infty}$ ${\mathcal F}^{-1} L^2 \to L^2$. {\mathcal : (L^1, L^2)_{\theta, \to (L^{\infty}, L^2)_{\theta, which, by is exactly the statement $$\begin{aligned} {\mathcal F}^{-1} : L^{{\alpha}, 2}({\mathbb R}^d) \to L^{{\beta},2}({\mathbb R}^d)\end{aligned}$$ where $\frac{1}{{\alpha}} = - \frac{\theta}{2}$ and $\frac{1}{{\beta}} = \frac{\theta}{2}$ and we notice that $\frac{1}{{\alpha}}+ \frac{1}{{\beta}} =1$. with = + we have that ${\mathcal F}f \in L^{{\gamma}, 2}({\mathbb R}^d)$ which implies that $f \in L^{{\gamma}^{\prime}, 2}({\mathbb where $\frac{1}{{\gamma}^{\prime}} =\frac{1}{2} - \frac{s}{d}$ which is exactly. $L^p$ and Besov space of this statement are slightly complicated prove as require facts real interpolation theory we refer the reader to [@Tar] for a detailed proof. We also need the following version of Calderon-Zygmund theorem spaces. \[CZ\] $T$ a operator. Then $T: L^{p,r}$ for $1<p<\infty$ and $1\le r\le f\|_{L^{p,r}} \lesssim \|f\|_{L^{p,r}}\end{aligned}$$ where the constant above does depend on This extension of the Calderon-Zygmund theorem an easy consequence of the $L^p$ version given following interpolation theorem of Calderon, see [@Ber-Lof Theorem 5.3.4]. \[CZ I\]
s}{d}$. Now recall that ${\mathcal F}^{-1} : L^1 \tO L^{\infty}$ and ${\MathcAl F}^{-1} : l^2 \to l^2$. THereFore, By real interpolATion $$\Begin{aligned} {\mathcal F}^{-1} : (L^1, L^2)_{\Theta, 2} \To (l^{\InftY}, l^2)_{\tHeta, 2}\eNd{alignED}$$ wHICh, bY iS eXacTlY ThE statEmeNt that $$\bEgin{aligneD} {\maThCal F}^{-1} : L^{{\alpha}, 2}({\maTHbB R}^d) \to L^{{\beta},2}({\MatHbb R}^d)\end{aligNed}$$ Where $\fRaC{1}{{\alPHa}} = 1 - \fraC{\thEta}{2}$ anD $\frac{1}{{\bETa}} = \frac{\Theta}{2}$ and wE nOTice thAT $\frac{1}{{\alPHA}}+ \fRac{1}{{\bEta}} =1$. Hence, with $\frac{1}{{\gAMmA}} = \Frac{1}{2} + \frac{s}{d}$ we haVe that ${\MaTHcAL f}f \iN L^{{\gAmma}, 2}({\mathbb r}^d)$ Which IMplies tHAt $F \IN l^{{\gaMMa}^{\prime}, 2}({\mathbb r}^d)$ where $\frac{1}{{\GAmmA}^{\prime}} =\FrAc{1}{2} - \fRAc{s}{d}$ whIch is ExACtlY. The $L^p$ and BeSov sPace versiOns of tHIs stateMEnt are sLightlY moRe cOmplICaTeD to PrOVe aS ThEy rEQuiRe additiOnAl Facts From REAL InteRpoLatiOn theOry and we refer The ReadER to [@tar] foR a detAileD pRoof. WE also nEed thE fOllowing version Of thE Calderon-zygMuNd tHeOrem fOR LorenTz sPacEs. \[CZ\] Let $t$ be a CalDEroN-ZYGMUnD operator. Then $T: L^{p,r} \tO L^{P,R}$ FoR $1<p<\infty$ aNd $1\le r\lE \InFtY$, $$\Begin{aliGnEd} \|T F\|_{L^{p,r}} \LESssim \|F\|_{L^{p,r}}\ENd{Aligned}$$ wHere thE CoNsTant aboVe Does noT dEpeNd oN $r$. ThiS ExteNsion oF the CaldEron-ZYGmund theorem is AN easy consequeNCe OF ThE $l^p$ veRsiOn given the fOlloWIng iNterPOlAtiON theoRem of caLDeROn, see  [@Ber-Lof Theorem 5.3.4]. \[Cz I\]
s}{d}$. Now recall that ${ \mathcal F }^{-1 } : L^ 1\toL^{\ infty}$ and ${ \ math cal F}^{-1} : L^2 \toL^2$. T h eref o re , byreal in t er p o lat io n$$\ be g in {alig ned } {\mat hcal F}^{- 1}:(L^1, L^2)_{ \ th eta, 2} \t o ( L^{\infty},L^2 )_{\th et a,2 }\end {al igned }$$ wh i ch, by is exact ly the st a tementt h at $$\ begin{aligned} {\ m at h cal F}^{-1} :L^{{\a lp h a} , 2}( {\m athbb R}^d )\to L ^ {{\beta } ,2 } ( { \ma t hbb R}^d)\end {aligned}$$ whe re $\f ra c{1 } {{\alp ha}}=1 -\frac{\thet a}{2 }$ and $\ frac{1 } {{\beta } } = \fr ac{\th eta }{2 }$ a n dwe no ti c e t h at $\ f rac {1}{{\al ph a} }+ \f rac{ 1 } { { \bet a}} =1$ . Hen ce, with $\fr ac{ 1}{{ \ gam ma}}= \fr ac{1 }{ 2} +\frac{ s}{d} $we have that ${ \mat hcal F}f\in L ^{{ \g amma} , 2}({\ mat hbb R}^d)$ whichi mpl ie s t ha t $f \in L^{{\gamm a} ^ { \p rime}, 2 }({\ma t hb bR }^d)$ wh er e $ \fra c { 1}{{\ gamm a }^ {\prime} } =\fr a c{ 1} {2} - \ fr ac{s}{ d} $ w hic h ise xact ly. T he $L^p$ andB esov space ver s ions of thiss ta t e me n t ar e s lightly mor e co m plic ated to pr o ve as they r e qu i re additional facts f rom re al in terpolation t heory andw e refer th e re a de r to [@Tar] for a de tailed pro o f. We a lso n eed thefollowing v ersion o f t heCal der o n -Z ygmund theore m forLo rentz s pac es. \[ CZ\ ] L et$T$ b e a Calde ron-Zygm un dop er ato r. Th e n $T: L^ {p ,r} \ toL^{p, r }$ for $1<p <\in ft y$ and $1\ler \l e \inf ty $, $$\ beg in {alig ned} \|T f\|_{L ^{p,r}} \ les s sim\| f\ |_{L^{p ,r}}\end{alig ne d}$$ where t heconsta n t above d oes not depend on $r$.This ex ten sionof t he Calder on- Zygmun d t h eoremis aneasyco nse q u enceo f t he$L ^p$ versio n giv en th efoll owing i nterpolation theor e m o f Calderon, s ee [@B e r -L ofT he o rem 5 . 3.4 ] . \[CZ I\]
s}{d}$. Now_recall that_${\mathcal F}^{-1} : L^1_\to L^{\infty}$_and_${\mathcal F}^{-1}_:_L^2 \to L^2$._Therefore, by real_interpolation $$\begin{aligned} {\mathcal F}^{-1} :_(L^1, L^2)_{\theta, 2}_\to_(L^{\infty}, L^2)_{\theta, 2}\end{aligned}$$ which, by is exactly the statement that $$\begin{aligned} {\mathcal F}^{-1} : L^{{\alpha},_2}({\mathbb_R}^d) \to_L^{{\beta},2}({\mathbb_R}^d)\end{aligned}$$_where $\frac{1}{{\alpha}} = 1 -_\frac{\theta}{2}$ and $\frac{1}{{\beta}} = \frac{\theta}{2}$_and we_notice that $\frac{1}{{\alpha}}+ \frac{1}{{\beta}} =1$. Hence, with $\frac{1}{{\gamma}}_=_\frac{1}{2} + \frac{s}{d}$_we have that ${\mathcal F}f \in L^{{\gamma}, 2}({\mathbb R}^d)$_which implies that $f \in L^{{\gamma}^{\prime},_2}({\mathbb R}^d)$ where_$\frac{1}{{\gamma}^{\prime}}_=\frac{1}{2}_- \frac{s}{d}$ which is_exactly. The $L^p$ and Besov space versions_of this statement are slightly more_complicated to prove as they require additional_facts from real interpolation theory and_we refer the reader to_[@Tar] for_a detailed proof. We also need_the following version_of the_Calderon-Zygmund theorem for_Lorentz spaces. \[CZ\] Let $T$ be a_Calderon-Zygmund operator. Then_$T: L^{p,r} \to L^{p,r}$ for $1<p<\infty$_and_$1\le r\le \infty$,_$$\begin{aligned} \|T_f\|_{L^{p,r}}_\lesssim \|f\|_{L^{p,r}}\end{aligned}$$_where the constant_above_does not_depend_on $r$. This extension of the Calderon-Zygmund_theorem_is an easy consequence of the $L^p$_version given the following_interpolation_theorem of Calderon, see_ [@Ber-Lof Theorem 5.3.4]. \[CZ I\]
$\mathscr{I\!O}\!_{\infty}(\mathbb{Z}^n_{\operatorname{lex}})$ implies that $\chi$ and its inverse $\chi^{-1}$ are elements of $\mathscr{I\!O}\!_{\infty}(\mathbb{Z}^n_{\operatorname{lex}})$. Simple verifications show that $\mathbb{I}=\chi\chi^{-1}=\chi\mathbb{I}\chi^{-1}$. Also, since $\mathfrak{C}$ is a congruence on the semigroup $\mathscr{I\!O}\!_{\infty}(\mathbb{Z}^n_{\operatorname{lex}})$ we conclude that $\mathbb{I}=\chi\mathbb{I}\chi^{-1}\mathfrak{C}\chi\kappa\chi^{-1}$. Now simple calculations imply that - if $m_i>0$ then $$(k)\chi_i\kappa_i\chi_i^{-1}= \left\{ \begin{array}{cl} k+m_i, & \hbox{if~}\, k\geqslant 1;\\ \hbox{undefined}, & \hbox{if~}\, -m_i<k\leqslant 0;\\ k+m_i, & \hbox{if~}\, k\leqslant-m_i, \end{array} \right.$$ and similarly - if $m_i<0$ then $$(k)\chi_i\kappa_i\chi_i^{-1}= \left\{ \begin{array}{cl} k+m_i, & \hbox{if~}\, k\geqslant-m_i;\\ \hbox{undefined}, & \hbox{if~}\, 0\leqslant k<-m_i;\\ k+m_i, & \hbox{if~}\, k\leqslant-1. \end{array} \right.$$ Next we put $\psi=\chi\kappa\chi^{-1}$, and hence we obtain that $\mathbb{I}\mathfrak{C}\psi$ but $\operatorname{dom}\psi\neq\mathbb{Z}$. This completes the proof of our lemma. \[remark-3\] The proof of Lemma \[lemma-2\] implies that for element $\psi=(\psi_1,\ldots,\psi_n)$ the following property holds: > $\psi_i$ is not
$ \mathscr{I\!O}\!_{\infty}(\mathbb{Z}^n_{\operatorname{lex}})$ implies that $ \chi$ and its inverse $ \chi^{-1}$ are elements of $ \mathscr{I\!O}\!_{\infty}(\mathbb{Z}^n_{\operatorname{lex}})$. Simple verifications picture that $ \mathbb{I}=\chi\chi^{-1}=\chi\mathbb{I}\chi^{-1}$. besides, since $ \mathfrak{C}$ is a congruence on the semigroup $ \mathscr{I\!O}\!_{\infty}(\mathbb{Z}^n_{\operatorname{lex}})$ we conclude that $ \mathbb{I}=\chi\mathbb{I}\chi^{-1}\mathfrak{C}\chi\kappa\chi^{-1}$. nowadays bare calculations imply that - if $ m_i>0 $ then $ $ (k)\chi_i\kappa_i\chi_i^{-1}= \left\ { \begin{array}{cl } k+m_i, & \hbox{if~}\, k\geqslant 1;\\ \hbox{undefined }, & \hbox{if~}\, -m_i < k\leqslant 0;\\ k+m_i, & \hbox{if~}\, k\leqslant - m_i, \end{array } \right.$$ and similarly - if $ m_i<0 $ then $ $ (k)\chi_i\kappa_i\chi_i^{-1}= \left\ { \begin{array}{cl } k+m_i, & \hbox{if~}\, k\geqslant - m_i;\\ \hbox{undefined }, & \hbox{if~}\, 0\leqslant k<-m_i;\\ k+m_i, & \hbox{if~}\, k\leqslant-1. \end{array } \right.$$ Next we arrange $ \psi=\chi\kappa\chi^{-1}$, and therefore we obtain that $ \mathbb{I}\mathfrak{C}\psi$ but $ \operatorname{dom}\psi\neq\mathbb{Z}$. This completes the proof of our lemma. \[remark-3\ ] The proof of Lemma   \[lemma-2\ ] implies that for chemical element $ \psi=(\psi_1,\ldots,\psi_n)$ the following property hold: > $ \psi_i$ is not
$\mahhscr{I\!O}\!_{\infty}(\mathbb{Z}^n_{\operxtorname{lex}})$ implies thet $\chi$ znd its knverse $\chi^{-1}$ are elements of $\larhscr{U\!O}\!_{\infty}(\mathbb{Z}^n_{\operatofname{lex}})$. Dimple vwrifmcations show thef $\mathbn{N}=\chi\cgl^{-1}=\chi\mctibb{I}\chi^{-1}$. Also, sikce $\mathfran{C}$ is a congruanze on the semigroup $\mathscr{I\!O}\!_{\infty}(\matrbb{Z}^n_{\oprrwtorname{lex}})$ we conbltde fhat $\mathbb{I}=\chi\mathbb{I}\chi^{-1}\mathfrak{C}\dhi\kappe\chi^{-1}$. Now simple valculations imply that - lf $m_l>0$ then $$(k)\chi_i\kappa_i\fhi_i^{-1}= \left\{ \fwgin{array}{cl} k+m_i, & \hbox{if~}\, k\geqslant 1;\\ \hbox{uvdefiued}, & \hbox{if~}\, -m_u<k\lfxslant 0;\\ k+i_i, & \hbox{ix~}\, k\leqskant-m_i, \enc{arcay} \right.$$ and similarly - if $m_i<0$ then $$(k)\chi_y\kappa_i\chh_i^{-1}= \left\{ \begin{aerqy}{cl} k+o_u, & \hblx{ih~}\, k\geqslant-j_i;\\ \hvox{undefined}, & \hbox{ig~}\, 0\jvaslant k<-m_i;\\ h+m_y, & \hbox{if~}\, k\leqslant-1. \end{drrzy} \right.$$ Next we put $\psi=\chi\kappa\chi^{-1}$, and hejce we obeain that $\mathbb{I}\mathfrak{C}\psi$ but $\operatorname{dom}\[si\nes\oatknn{Z}$. Tfus completes the proof of our lemma. \[remark-3\] The pwkog pf Lemma \[lemma-2\] lmplies that for ekelemj $\psi=(\psi_1,\ldots,\pri_n)$ thz fkllowing property jolds: > $\pfi_i$ iw not
$\mathscr{I\!O}\!_{\infty}(\mathbb{Z}^n_{\operatorname{lex}})$ implies that $\chi$ and its inverse elements $\mathscr{I\!O}\!_{\infty}(\mathbb{Z}^n_{\operatorname{lex}})$. Simple show that $\mathbb{I}=\chi\chi^{-1}=\chi\mathbb{I}\chi^{-1}$. congruence the semigroup $\mathscr{I\!O}\!_{\infty}(\mathbb{Z}^n_{\operatorname{lex}})$ conclude that $\mathbb{I}=\chi\mathbb{I}\chi^{-1}\mathfrak{C}\chi\kappa\chi^{-1}$. simple calculations imply that - if then $$(k)\chi_i\kappa_i\chi_i^{-1}= \left\{ \begin{array}{cl} k+m_i, & \hbox{if~}\, k\geqslant 1;\\ \hbox{undefined}, & \hbox{if~}\, -m_i<k\leqslant k+m_i, & \hbox{if~}\, k\leqslant-m_i, \end{array} \right.$$ and similarly - if $m_i<0$ then $$(k)\chi_i\kappa_i\chi_i^{-1}= \begin{array}{cl} & k\geqslant-m_i;\\ & \hbox{if~}\, 0\leqslant k<-m_i;\\ k+m_i, & \hbox{if~}\, k\leqslant-1. \end{array} \right.$$ Next we put $\psi=\chi\kappa\chi^{-1}$, and hence obtain that $\mathbb{I}\mathfrak{C}\psi$ but $\operatorname{dom}\psi\neq\mathbb{Z}$. This completes the of our lemma. \[remark-3\] proof of Lemma \[lemma-2\] implies for $\psi=(\psi_1,\ldots,\psi_n)$ the property > is not
$\mathscr{I\!O}\!_{\infty}(\mathbb{Z}^n_{\opeRatorname{lEx}})$ impLieS thAt $\Chi$ aNd itS inverse $\chi^{-1}$ are ELemeNts of $\mathscr{I\!O}\!_{\infty}(\matHbb{Z}^n_{\OpERatoRNaMe{lex}})$. simple vERiFICatIoNs ShoW tHAt $\MathbB{I}=\cHi\chi^{-1}=\chI\mathbb{I}\chI^{-1}$. AlSo, Since $\mathfraK{c}$ iS a congruenCe oN the semigrouP $\maThscr{I\!o}\!_{\iNftY}(\MathbB{Z}^n_{\OperaTornamE{Lex}})$ we cOnclude thAt $\MAthbb{I}=\CHi\mathbB{i}\ChI^{-1}\matHfrak{C}\chi\kappa\chi^{-1}$. nOw SImple calculatiOns impLy THaT - IF $m_i>0$ TheN $$(k)\chi_i\kappA_i\Chi_i^{-1}= \lEFt\{ \begin{ARrAY}{CL} k+m_I, & \Hbox{if~}\, k\geqslaNt 1;\\ \hbox{undefINed}, & \Hbox{if~}\, -M_i<K\leQSlant 0;\\ k+M_i, & \hboX{iF~}\, K\leQslant-m_i, \end{ArraY} \right.$$ and SimilaRLy - if $m_i<0$ tHEn $$(k)\chi_i\Kappa_i\Chi_I^{-1}= \leFt\{ \beGIn{ArRay}{Cl} K+M_i, & \hBOx{If~}\, k\GEqsLant-m_i;\\ \hbOx{UnDefinEd}, & \hbOX{IF~}\, 0\LeqsLanT k<-m_i;\\ K+m_i, & \hbOx{if~}\, k\leqslant-1. \End{ArraY} \RigHt.$$ NexT we puT $\psi=\ChI\kappA\chi^{-1}$, anD hencE wE obtain that $\mathBb{I}\mAthfrak{C}\pSi$ bUt $\OpeRaTornaME{dom}\psI\neQ\maThbb{Z}$. ThIs complETes ThE PROoF of our lemma. \[remark-3\] THe PROoF of Lemma \[Lemma-2\] iMPlIeS That for eLeMenT $\psi=(\PSI_1,\ldotS,\psi_N)$ ThE followiNg propERtY hOlds: > $\psi_I$ iS not
$\mathscr{I\!O}\!_{\infty }(\mathbb{ Z}^n_ {\o per at orna me{l ex}})$ implies that $\chi$ and its invers e $\c hi ^ {-1} $ a re el ementso f$ \ mat hs cr {I\ !O } \! _{\in fty }(\math bb{Z}^n_{\ ope ra torname{lex} } )$ . Simple v eri fications sh owthat $ \m ath b b{I}= \ch i\chi ^{-1}= \ chi\ma thbb{I}\c hi ^ {-1}$. Also, s i n ce $\m athfrak{C}$ is ac on g ruence on thesemigr ou p $ \ m ath scr {I\!O}\!_{ \i nfty} ( \mathbb { Z} ^ n _ {\o p eratorname{le x}})$ we co n clu de tha t$\m a thbb{I }=\ch i\ m ath bb{I}\chi^{ -1}\ mathfrak{ C}\chi \ kappa\c h i^{-1}$ . Now si mpl e ca l cu la tio ns imp l ytha t - if $m _i >0 $ the n $$ ( k ) \ chi_ i\k appa _i\ch i_i^{-1}= \le ft\ { \begi n{arr ay}{ cl } k +m_i, & \hb ox{i f~}\, k\g eqs la nt1; \\ \ hbo x{u ndefine d}, & \ h box {i f ~ } \, -m_i<k\leqslant 0 ;\ \ k+ m_i, & \h bo x{i f~}\ , k\leq slan t -m _i, \end { ar ra y} \r ight.$ $ an d s imila r ly - if $m_i<0$ then $$(k)\chi_i\ka p pa_i\chi_i^{- 1 }= \ le f t\{ \begin{a rray } {cl} k + m_i, & \hbox{if~}\, k\ge qs lant-m _i;\\ \hbo x{undefine d } , & \hbox {if~ } \, 0\leqslant k<- m_i;\ \ k +m_i, & \h box{if~}\ , k\leqsla nt- 1. \ e nd {array} \ r i ght. $$ Nextweput $\p si= \ch i\k app a\ chi^{-1}$ , and he nc ewe o bta in th a t $\math bb {I} \m ath frak{ C }\psi$ but$\op er at o rna me{dom} \ ps i \ neq\ ma th bb{Z }$. T his c ompl e tes the pr oof of ou r l e mma. \[ remark- 3\] The proof o f Lemma \[ le mma -2\] i m p lies tha t for element $\psi=(\p s i_1,\ld ots ,\psi _n)$ the foll owi ng pro per t y hold s: >$\psi _i $ i s not
$\mathscr{I\!O}\!_{\infty}(\mathbb{Z}^n_{\operatorname{lex}})$_implies that_$\chi$ and its inverse_$\chi^{-1}$ are_elements_of $\mathscr{I\!O}\!_{\infty}(\mathbb{Z}^n_{\operatorname{lex}})$._Simple_verifications show that_$\mathbb{I}=\chi\chi^{-1}=\chi\mathbb{I}\chi^{-1}$. Also, since_$\mathfrak{C}$ is a congruence_on the semigroup_$\mathscr{I\!O}\!_{\infty}(\mathbb{Z}^n_{\operatorname{lex}})$_we conclude that $\mathbb{I}=\chi\mathbb{I}\chi^{-1}\mathfrak{C}\chi\kappa\chi^{-1}$. Now simple calculations imply that - if $m_i>0$ then $$(k)\chi_i\kappa_i\chi_i^{-1}=_\left\{ _ ___ \begin{array}{cl} _ k+m_i,_ _ __& \hbox{if~}\, k\geqslant_1;\\ \hbox{undefined},_& \hbox{if~}\, -m_i<k\leqslant 0;\\ _ ___k+m_i, _ _ & \hbox{if~}\, k\leqslant-m_i, _ \end{array} _ \right.$$ and similarly - if_$m_i<0$ then $$(k)\chi_i\kappa_i\chi_i^{-1}= \left\{ _ _ \begin{array}{cl} _ _ _k+m_i, _ _ _& \hbox{if~}\, k\geqslant-m_i;\\ __ _\hbox{undefined},_&_\hbox{if~}\, 0\leqslant_k<-m_i;\\ __ __ k+m_i, __ &_\hbox{if~}\, k\leqslant-1. __ \end{array} _ \right.$$ Next we put_$\psi=\chi\kappa\chi^{-1}$, and hence we obtain that_$\mathbb{I}\mathfrak{C}\psi$ but_$\operatorname{dom}\psi\neq\mathbb{Z}$. This_completes the proof of our lemma. \[remark-3\] The proof of Lemma \[lemma-2\] implies_that for element $\psi=(\psi_1,\ldots,\psi_n)$ the following_property holds: > $\psi_i$ is_not
stray capacitances $C_s$ from the board, components and packaging. The bandwidth $BW$ of the photoreceiver can be estimated as $$BW=\sqrt{\frac{GBWP}{2\pi R_f\,C_T}}, \label{BW}$$ where $GBWP$ is the gain-bandwidth product of the operational amplifier. The total TIA input current noise $I_{\mathrm{noise}}(f)$ model can be expressed as $$I_{noise}(f)=\sqrt{i_T^2+i_{TIA}^2(f)}\,\cdot\|\overline{TF}(f)\|, \label{i_noise}$$ where $\|\overline{TF}(f)\|$ is the normalized TIA transfer function, $i_{TIA}(f)$ is a frequency dependent component of the input current noise (see Equation \[i\_tia\]), and $i_T$ is the quadrature sum of various contributors that can be approximated by neglecting their frequency dependency for modeling purposes. For example, the expected current noise $i_T$ in the photoreceiver shown in Figure \[tia\], can be computed as $$i_T=\sqrt{i_n^2+i_J^2+i_d^2}, \label{it}$$ where $i_n$ is the op-amp current noise, $i_J$ is the Johnson noise of the feedback resistor $R_f$ (see Equation \[i\_j\]), and $i_d$ is the shot-noise from the photodiode dark current. It can be seen from Equation \[Ct\] that the photodiode capacitance and the op-amp common-mode input capacitance are crucial factors in the total noise budget. The challenge for the photodiode manufacture lays in achieving a minimum capacitance per unit area while maintaining high responsivity and low leakage if reverse-biased. For the TIA electronics, it is necessary to identify an op-amp with minimal common-mode input capacitance, current and voltage noise, and a gain-bandwidth product large enough to maintain the required sensitivity over the required measurement bandwidth of 2–20MHz. Prototype photoreceivers ======================== Collaboration with industry {#DS_proto} --------------------------- Under a Small Business Innovation Research (SBIR) grant, the company Discovery Semiconductors has developed a large-area quadrant photodiode (QPD) of 1mm diameter
stray capacitances $ C_s$ from the board, components and promotion. The bandwidth $ BW$ of the photoreceiver can be estimate as $ $ BW=\sqrt{\frac{GBWP}{2\pi R_f\,C_T } }, \label{BW}$$ where $ GBWP$ is the gain - bandwidth product of the functional amplifier. The total TIA remark current noise $ I_{\mathrm{noise}}(f)$ exemplar can be expressed as $ $ I_{noise}(f)=\sqrt{i_T^2+i_{TIA}^2(f)}\,\cdot\|\overline{TF}(f)\|, \label{i_noise}$$ where $ \|\overline{TF}(f)\|$ is the normalized TIA transportation affair, $ i_{TIA}(f)$ is a frequency dependent part of the input current noise (see Equation   \[i\_tia\ ]), and $ i_T$ is the quadrature sum of diverse contributors that can be approximated by neglecting their frequency dependency for modeling purpose. For example, the expected current noise $ i_T$ in the photoreceiver show in Figure   \[tia\ ], can be computed as $ $ i_T=\sqrt{i_n^2+i_J^2+i_d^2 }, \label{it}$$ where $ i_n$ is the op - amp current noise, $ i_J$ is the Johnson noise of the feedback resistor $ R_f$ (see Equation   \[i\_j\ ]), and $ i_d$ is the shooting - noise from the photodiode dark current. It can be seen from Equation   \[Ct\ ] that the photodiode capacitor and the op - amp common - mode input capacitance are crucial factors in the entire noise budget. The challenge for the photodiode manufacture lays in achieving a minimum capacitance per unit area while maintaining high responsivity and abject leakage if reverse - bias. For the TIA electronics, it is necessary to name an op - amp with minimal coarse - mode input capacitor, current and voltage randomness, and a gain - bandwidth product large enough to keep the required sensitivity over the required measurement bandwidth of 2–20MHz. Prototype photoreceivers = = = = = = = = = = = = = = = = = = = = = = = = Collaboration with industry { # DS_proto } --------------------------- Under a Small Business Innovation Research (SBIR) grant, the company Discovery Semiconductors has developed a large - area quadrant photodiode (QPD) of 1 mm diameter
stgay capacitances $C_s$ from the board, components end paciaging. Thd bandwidth $BW$ of the photorxceicer cqn be estimated as $$BW=\sdrt{\frac{GBAP}{2\pi R_f\,C_R}}, \lauel{BW}$$ where $GBWP$ is the ncin-bahfwidch product of thg operationan amplifier. Tha gocal TIA input current noise $I_{\mathrm{njise}}(f)$ mpdfl can be exprgssed ws $$I_{hoise}(f)=\sqrt{i_T^2+i_{TIA}^2(f)}\,\cdot\|\overline{TF}(f)\|, \lzbel{i_nopse}$$ where $\|\overlinr{TF}(f)\|$ is the normalized TIA trajsfer function, $i_{TIW}(f)$ is a frewuenst dependent zomponent of the input current noise (see Equation \[i\_tia\]), and $n_T$ is the qocerahore sum of verious contributors that cdn be alproximated by nejlecring their frequency vependency for modelyng purpovea. For example, the ezpectgd cusreng nokse $i_V$ ih the ohovoreceiver ahown in Fiture \[tia\], can be compitqe as $$i_T=\sqrt{i_n^2+j_J^2+i_d^2}, \lwbql{it}$$ where $i_n$ is the op-amp current noist, $i_J$ js the Johnson noise of the feedback resistot $R_f$ (see Ezuation \[i\_j\]), and $i_d$ is the shot-noise from the photodhode vafk emvreng. Ih can be seen from Equation \[Ct\] that the photodykdt cspacitance and the op-amp volmpg-mode input cxpacitchcs are crucial factlrs in jhe toral noise budbet. The challenge for the pyotodiode mauufqcture lays in achneving a minnmum cspaciyance per unit area whiue mzintaining jigh respkvsivity and low uealace if reverse-biased. For thq TIA elertronncs, it ir nevessarr to identlfy ak op-amp with minimap comlot-mode inpuh capacitance, current and voltajx noise, and a gdin-tandwidtk prodmct large enougr to maintain jhe requixed sevsitivity kver thx required mqasurement batfwidth of 2–20MHv. Prototypq phitorwceiverr ======================== Zollaboration eith industry {#DS_proro} --------------------------- Under a Small Buxingsa Innovation Reweatxh (SBIR) grant, yhe coipwnb Dissmvery Semicotducgorr has aeveloptb a uargr-area quadrant photogiods (QPD) of 1mm diametrr
stray capacitances $C_s$ from the board, components The $BW$ of photoreceiver can be where is the gain-bandwidth of the operational The total TIA input current noise model can be expressed as $$I_{noise}(f)=\sqrt{i_T^2+i_{TIA}^2(f)}\,\cdot\|\overline{TF}(f)\|, \label{i_noise}$$ where $\|\overline{TF}(f)\|$ is the normalized TIA function, $i_{TIA}(f)$ is a frequency dependent component of the input current noise (see \[i\_tia\]), $i_T$ the sum of various contributors that can be approximated by neglecting their frequency dependency for modeling purposes. example, the expected current noise $i_T$ in the shown in Figure \[tia\], be computed as $$i_T=\sqrt{i_n^2+i_J^2+i_d^2}, \label{it}$$ $i_n$ the op-amp noise, is Johnson noise of feedback resistor $R_f$ (see Equation \[i\_j\]), and $i_d$ is the shot-noise from the photodiode dark current. It be seen \[Ct\] that photodiode and op-amp common-mode input crucial factors in the total noise for the photodiode manufacture lays in achieving a capacitance per area while maintaining high responsivity and leakage if reverse-biased. For the TIA electronics, it necessary to identify an op-amp with minimal common-mode input capacitance, current and voltage noise, and product large enough to the required sensitivity the measurement of Prototype photoreceivers Collaboration with industry {#DS_proto} --------------------------- Under a Small Business Innovation Research grant, the company Discovery Semiconductors has developed a large-area quadrant of diameter
stray capacitances $C_s$ from thE board, compOnentS anD paCkAginG. The Bandwidth $BW$ of tHE phoToreceiver can be estimatEd as $$Bw=\sQRt{\frAC{GbWP}{2\pi r_f\,C_T}}, \labEL{Bw}$$ WHerE $GbWp$ is ThE GaIn-banDwiDth prodUct of the opEraTiOnal amplifieR. thE total TIA iNpuT current noisE $I_{\mAthrm{nOiSe}}(f)$ MOdel cAn bE exprEssed aS $$i_{noise}(F)=\sqrt{i_T^2+i_{TiA}^2(F)}\,\Cdot\|\ovERline{TF}(F)\|, \LAbEl{i_nOise}$$ where $\|\overline{tf}(f)\|$ IS the normalized tIA traNsFEr FUNctIon, $I_{TIA}(f)$ is a frEqUency DEpendenT CoMPONenT Of the input curRent noise (seE equAtion \[i\_TiA\]), anD $I_T$ is thE quadRaTUre Sum of variouS conTributors That caN Be approXImated bY negleCtiNg tHeir FReQuEncY dEPenDEnCy fOR moDeling puRpOsEs. For ExamPLE, THe exPecTed cUrrenT noise $i_T$ in the PhoToreCEivEr shoWn in FIgurE \[tIa\], can Be compUted aS $$i_t=\sqrt{i_n^2+i_J^2+i_d^2}, \labeL{it}$$ wHere $i_n$ is tHe oP-aMp cUrRent nOIse, $i_J$ iS thE JoHnson noIse of thE FeeDbACK ReSistor $R_f$ (see EquatioN \[i\_J\]), ANd $I_d$ is the sHot-noiSE fRoM The photoDiOde Dark CURrent. it caN Be Seen from equatiON \[CT\] tHat the pHoTodiodE cApaCitAnce aND the Op-amp cOmmon-modE inpuT Capacitance are CRucial factors IN tHE ToTAl noIse Budget. The chAlleNGe foR the PHoTodIOde maNufacTuRE lAYs in achieving a minimUm CapaciTance Per unit area whIle maintaiNING high resPonsIViTY and low leakage If revErse-biased. fOr the TIA ElectRonics, it Is necessaRY To identiFy aN op-Amp WitH MInImal common-modE INput CaPacitanCe, cUrrent aNd vOltAge NoiSe, And a gain-bAndwidth PrOdUcT lArgE enouGH to maintAiN thE rEquIred sENsitivIty ovEr thE rEqUIreD measurEMeNT BandWiDtH of 2–20Mhz. PRoTotypE phoTOreCeivers ======================== collaboraTioN With InDuStry {#DS_pRoto} --------------------------- Under a SmaLl business InNoVatIon ResEARch (SBIR) gRant, the company Discovery sEmicondUctOrs haS devEloped a laRge-Area quAdrANt photOdiode (qPD) of 1Mm DiaMETer
stray capacitances $C_s$from the b oard, co mpo ne ntsandpackaging. Th e ban dwidth $BW$ of the pho torec ei v er c a nbe es timated as $ $BW =\ sq rt{ \f r ac {GBWP }{2 \pi R_f \,C_T}}, \la be l{BW}$$ wher e $ GBWP$ is t hegain-bandwid thproduc toft he op era tiona l ampl i fier.The total T I A inpu t curren t no ise$I_{\mathrm{noise } }( f )$ model can b e expr es s ed a s $ $I_ {noise}(f) =\ sqrt{ i _T^2+i_ { TI A } ^ 2(f ) }\,\cdot\|\ov erline{TF}( f )\| , \la be l{i _ noise} $$ wh er e $\ |\overline{ TF}( f)\|$ isthe no r malized TIA tra nsferfun cti on,$ i_ {T IA} (f ) $ i s a fr e que ncy depe nd en t com pone n t o f th e i nput curr ent noise (se e E quat i on\[i\_ tia\] ), a nd $i_T $ is t he qu ad rature sum of v ario us contri but or s t ha t can be app rox ima ted byneglect i ngth e i r f requency dependenc yf o rmodeling purpo s es .F or examp le , t he e x p ected cur r en t noise$i_T$i nth e photo re ceiver s how n i n Fig u re \ [tia\] , can be comp u ted as $$i_T=\ s qrt{i_n^2+i_J ^ 2+ i _ d^ 2 }, \la bel{it}$$ w here $i_n $ is th e o p -ampcurre nt no i se, $i_J$ is the Jo hn son no ise o f the feedbac k resistor $ R _f$ (see Equ a ti o n \[i\_j\]), a nd $i _d$ is the shot-noi se fr om the p hotodiode d ark curr ent . I t c anb e s een from Equa t i on \ [C t\] tha t t he phot odi ode ca pac it ance andthe op-a mp c om mo n-m ode i n put capa ci tan ce ar e cru c ial fa ctors inth et ota l noise bu d g et.Th echal len ge forthep hot odiodemanufactu rel aysin a chievin g a minimum c ap acitance p er un it are a while ma intaining high responsi v ity and lo w lea kage if rever se- biased . F o r theTIA el ectro ni cs, i t isn e ce ssa ry to identi f y an op-a mp wit h minim al common-mode inp u t c apacitance, c urr enta n dvol t ag e no is e , a n d a gain-bandwid th product l a rg e enough t o ma in tain th e requi red s e nsitivi ty over t he requir ed mea s u rem ent bandwi dth of 2 –20MHz. P rotot y pe phot ore ceiver s=== ===== ====== = === ===== = Col la borati on wi th industr y {#DS_proto} --------- ------ ----- --- ---- Und era Sm all Busin essInnovation Re sea rch ( SBI R ) gra nt,t he co m panyDisc o very Semi c on duc t o rs has develo p e d alarge -ar e a quad rant photodiode (QPD) of 1mm diamete r
stray_capacitances $C_s$_from the board, components_and packaging. The_bandwidth_$BW$ of_the_photoreceiver can be_estimated as $$BW=\sqrt{\frac{GBWP}{2\pi_R_f\,C_T}}, \label{BW}$$ where $GBWP$_is the gain-bandwidth_product_of the operational amplifier. The total TIA input current noise $I_{\mathrm{noise}}(f)$ model can be_expressed_as $$I_{noise}(f)=\sqrt{i_T^2+i_{TIA}^2(f)}\,\cdot\|\overline{TF}(f)\|, _\label{i_noise}$$_where_$\|\overline{TF}(f)\|$ is the normalized TIA_transfer function, $i_{TIA}(f)$ is a_frequency dependent_component of the input current noise (see Equation \[i\_tia\]),_and_$i_T$ is the_quadrature sum of various contributors that can be approximated_by neglecting their frequency dependency for_modeling purposes. For_example,_the_expected current noise $i_T$_in the photoreceiver shown in Figure \[tia\],_can be computed as $$i_T=\sqrt{i_n^2+i_J^2+i_d^2}, \label{it}$$_where $i_n$ is the op-amp current noise,_$i_J$ is the Johnson noise of_the feedback resistor $R_f$ (see_Equation \[i\_j\]), and_$i_d$ is the shot-noise from_the photodiode dark_current. It_can be seen_from Equation \[Ct\] that the photodiode capacitance_and the op-amp_common-mode input capacitance are crucial factors_in_the total noise_budget._The_challenge for_the photodiode manufacture_lays_in achieving_a_minimum capacitance per unit area while_maintaining_high responsivity and low leakage if reverse-biased._For the TIA electronics,_it_is necessary to identify_an op-amp with minimal common-mode_input capacitance, current and voltage noise,_and a_gain-bandwidth product_large enough to maintain the required sensitivity over the required measurement_bandwidth of 2–20MHz. Prototype photoreceivers ======================== Collaboration with industry_{#DS_proto} --------------------------- Under a Small Business_Innovation Research_(SBIR)_grant, the company_Discovery_Semiconductors has_developed a large-area quadrant photodiode (QPD) of_1mm diameter
{\partial^{2} \vartheta}{\partial x_{k} \partial x_{i}}} +\frac{\partial}{\partial x_{k}} \frac{\partial \rho L_0} {\partial \frac{\partial \vartheta}{\partial x_{k} }} -\frac{\partial \rho L_0}{\partial\vartheta} -\frac{\hbar \rho}{2} \left( G_1\cos \varphi - G_2\sin \varphi \right) =0 \label{eq:D6DSV3ARTH}\\ -\frac{\partial}{\partial x_{k}} \frac{\partial}{\partial x_{i}} \frac{\partial \rho L_0} {\partial\ \frac{\partial^{2} \varphi}{\partial x_{k} \partial x_{i}}} +\frac{\partial}{\partial x_{k}} \frac{\partial \rho L_0} {\partial \frac{\partial \varphi}{\partial x_{k} }} -\frac{\partial \rho L_0}{\partial\varphi} -\frac{\hbar }{2} \rho G_3 =0 \label{eq:D9ZSG4PHI} \mbox{.}\end{gathered}$$ The variable $S$ does not occur in (\[eq:DI2DS8RHO\])-(\[eq:D9ZSG4PHI\]) in agreement with our assumptions about the form of $L_0$. It is easy to see that a proper solution (with vanishing variational contributions from the second order derivatives) of (\[eq:DI2DS8RHO\])-(\[eq:D9ZSG4PHI\]) is given by $$\begin{gathered} L_0 = \frac{\hbar^{2}}{2m} \bigg[ \frac{1}{\sqrt{\rho}} \frac{\partial}{\partial \vec{x}} \frac{\partial}{\partial \vec{x}} \sqrt{\rho} -\frac{1}{4}\sin^{2} \vartheta \left( \frac{\partial \varphi}{\partial \vec{x}} \right)^{2} -\frac{1}{4} \left( \frac{\partial \vartheta}{\partial \vec{x}} \right)^{2} \bigg] \label{eq:DI3SOLL0HO}
{ \partial^{2 } \vartheta}{\partial x_{k } \partial x_{i } } } + \frac{\partial}{\partial x_{k } } \frac{\partial \rho L_0 } { \partial \frac{\partial \vartheta}{\partial x_{k } } } -\frac{\partial \rho L_0}{\partial\vartheta } -\frac{\hbar \rho}{2 } \left (G_1\cos \varphi - G_2\sin \varphi \right) = 0 \label{eq: D6DSV3ARTH}\\ -\frac{\partial}{\partial x_{k } } \frac{\partial}{\partial x_{i } } \frac{\partial \rho L_0 } { \partial\ \frac{\partial^{2 } \varphi}{\partial x_{k } \partial x_{i } } } + \frac{\partial}{\partial x_{k } } \frac{\partial \rho L_0 } { \partial \frac{\partial \varphi}{\partial x_{k } } } -\frac{\partial \rho L_0}{\partial\varphi } -\frac{\hbar } { 2 } \rho G_3 = 0 \label{eq: D9ZSG4PHI } \mbox{.}\end{gathered}$$ The variable $ S$ does not occur in   (\[eq: DI2DS8RHO\])-(\[eq: D9ZSG4PHI\ ]) in agreement with our assumptions about the form of $ L_0$. It is comfortable to examine that a proper solution (with vanishing variational contributions from the second decree derivative) of   (\[eq: DI2DS8RHO\])-(\[eq: D9ZSG4PHI\ ]) is given by $ $ \begin{gathered } L_0 = \frac{\hbar^{2}}{2 m } \bigg [ \frac{1}{\sqrt{\rho } } \frac{\partial}{\partial \vec{x } } \frac{\partial}{\partial \vec{x } } \sqrt{\rho } -\frac{1}{4}\sin^{2 } \vartheta \left (\frac{\partial \varphi}{\partial \vec{x } } \right)^{2 } -\frac{1}{4 } \left (\frac{\partial \vartheta}{\partial \vec{x } } \right)^{2 } \bigg ] \label{eq: DI3SOLL0HO }
{\parhial^{2} \vartheta}{\partial x_{k} \kartial x_{i}}} +\frac{\parjiql}{\partmal x_{k}} \fdac{\partixl \rho L_0} {\partial \frac{\partial \verthwta}{\paetial x_{k} }} -\frac{\partial \rfo L_0}{\partiwl\varthera} -\frec{\hbar \rho}{2} \left( G_1\rks \varpmn - G_2\sjk \var'hm \right) =0 \label{ea:D6DSV3ARTH}\\ -\fsac{\partial}{\parthau r_{k}} \frac{\partial}{\partial x_{i}} \frac{\partial \rho L_0} {\psrhial\ \frac{\partiaj^{2} \vagpri}{\padnicl x_{k} \partial x_{i}}} +\frac{\partial}{\partjal x_{k}} \fgac{\partial \rho L_0} {\psrtial \frac{\partial \varphi}{\pagtiap x_{k} }} -\frac{\partial \rjo L_0}{\partial\carpru} -\frac{\hbar }{2} \rfo G_3 =0 \label{tq:B9ZSG4PHI} \mbox{.}\gnd{gathered}$$ The variable $S$ does vot oecur in (\[eq:DI2ES8EHO\])-(\[fx:D9ZSG4PHI\]) in agrevment with ouv assum[tions sbout the form of $L_0$. Ut is easy to see thav a proper solution (rith vanivhnng variational contrubytionv frmm tfw sdcohd odder dfritatives) of (\[es:DI2DS8RHO\])-(\[eq:D9ZSG4PHI\]) is given by $$\nedpm{gathered} L_0 = \rrac{\hbwr^{2}}{2i} \bigg[ \frac{1}{\sqrt{\rho}} \frac{\partial}{\partial \vec{q}} \frzc{\partial}{\partial \vec{x}} \sqet{\rho} -\frac{1}{4}\sin^{2} \vartheta \peft( \frac{\[artial \varphi}{\partial \vec{x}} \right)^{2} -\frac{1}{4} \left( \frac{\parthal \vergheua}{\iwfriwl \vec{x}} \right)^{2} \bigg] \label{eq:DI3SOLL0HO}
{\partial^{2} \vartheta}{\partial x_{k} \partial x_{i}}} +\frac{\partial}{\partial x_{k}} L_0} \frac{\partial \vartheta}{\partial }} -\frac{\partial \rho \varphi G_2\sin \varphi \right) \label{eq:D6DSV3ARTH}\\ -\frac{\partial}{\partial x_{k}} x_{i}} \frac{\partial \rho L_0} {\partial\ \frac{\partial^{2} x_{k} \partial x_{i}}} +\frac{\partial}{\partial x_{k}} \frac{\partial \rho L_0} {\partial \frac{\partial \varphi}{\partial x_{k} }} \rho L_0}{\partial\varphi} -\frac{\hbar }{2} \rho G_3 =0 \label{eq:D9ZSG4PHI} \mbox{.}\end{gathered}$$ The variable $S$ does occur (\[eq:DI2DS8RHO\])-(\[eq:D9ZSG4PHI\]) agreement our assumptions about the form of $L_0$. It is easy to see that a proper solution vanishing variational contributions from the second order derivatives) (\[eq:DI2DS8RHO\])-(\[eq:D9ZSG4PHI\]) is given by L_0 = \frac{\hbar^{2}}{2m} \bigg[ \frac{1}{\sqrt{\rho}} \vec{x}} \vec{x}} \sqrt{\rho} \vartheta \frac{\partial \vec{x}} \right)^{2} -\frac{1}{4} \frac{\partial \vartheta}{\partial \vec{x}} \right)^{2} \bigg] \label{eq:DI3SOLL0HO}
{\partial^{2} \vartheta}{\partial x_{k} \pArtial x_{i}}} +\frAc{\parTiaL}{\paRtIal x_{K}} \fraC{\partial \rho L_0} {\paRTial \Frac{\partial \vartheta}{\parTial x_{K} }} -\fRAc{\paRTiAl \rho l_0}{\partiaL\VaRTHetA} -\fRaC{\hbAr \RHo}{2} \Left( G_1\Cos \Varphi - G_2\Sin \varphi \rIghT) =0 \lAbel{eq:D6DSV3ARth}\\ -\fRac{\partial}{\ParTial x_{k}} \frac{\paRtiAl}{\partIaL x_{i}} \FRac{\paRtiAl \rho l_0} {\partiAL\ \frac{\pArtial^{2} \varPhI}{\PartiaL X_{k} \partiAL X_{i}}} +\Frac{\Partial}{\partial x_{k}} \fRAc{\PArtial \rho L_0} {\partIal \fraC{\pARtIAL \vaRphI}{\partial x_{k} }} -\FrAc{\parTIal \rho L_0}{\PArTIAL\vaRPhi} -\frac{\hbar }{2} \rhO G_3 =0 \label{eq:D9Zsg4PHi} \mbox{.}\eNd{GatHEred}$$ ThE variAbLE $S$ dOes not occur In (\[eq:dI2DS8RHO\])-(\[eq:d9ZSG4PHi\]) In agreeMEnt with Our assUmpTioNs abOUt ThE foRm OF $L_0$. IT Is EasY To sEe that a pRoPeR soluTion (WITH VaniShiNg vaRiatiOnal contributIonS froM The SeconD ordeR derIvAtiveS) of (\[eq:Di2DS8RHo\])-(\[eQ:D9ZSG4PHI\]) is given By $$\beGin{gatherEd} L_0 = \FrAc{\hBaR^{2}}{2m} \bigG[ \Frac{1}{\sqRt{\rHo}} \fRac{\partIal}{\partIAl \vEc{X}} \FRAc{\Partial}{\partial \vec{x}} \SqRT{\RhO} -\frac{1}{4}\sin^{2} \VartheTA \lEfT( \Frac{\partIaL \vaRphi}{\PARtial \Vec{x}} \RIgHt)^{2} -\frac{1}{4} \leFt( \frac{\PArTiAl \varthEtA}{\partiAl \Vec{X}} \riGht)^{2} \biGG] \labEl{eq:DI3sOLL0HO}
{\partial^{2} \vartheta}{\ partial x_ {k} \ par tia lx_{i }}}+\frac{\partia l }{\p artial x_{k}} \frac{\p artia l\ rhoL _0 } {\p artial\ fr a c {\p ar ti al\v a rt heta} {\p artialx_{k} }} - \fr ac {\partial \r h oL_0}{\part ial \vartheta} - \fr ac{\hb ar \r h o}{2} \l eft(G_1\co s \varp hi - G_2\ si n \varp h i \rig h t )=0 \label{eq:D6DSV3A R TH } \\ -\frac{\par tial}{ \p a rt i a l x _{k }} \frac{ \p artia l }{\part i al x _ {i} } \frac{\parti al \rho L_0 } {\ partia l\ \f r ac{\pa rtial ^{ 2 } \ varphi}{\pa rtia l x_{k} \ partia l x_{i}} } +\frac {\part ial }{\ part i al x _{k }} \fr a c{ \pa r tia l \rho L _0 }{\par tial \ f r ac{\ par tial \var phi}{\partial x_ {k}} } - \frac {\par tial \ rho L _0}{\p artia l\ varphi} -\frac{ \hba r }{2} \r hoG_ 3 = 0\labe l {eq:D9 ZSG 4PH I} \mbo x{.}\en d {ga th e r e d} $$ The variable $S $d o es not occ ur in( \[ eq : DI2DS8RH O\ ])- (\[e q : D9ZSG 4PHI \ ]) in agre ementw it hour ass um ptions a bou t t he fo r m of $L_0$ . It iseasyt o see that a p r oper solution (w i t hv anis hin g variation al c o ntri buti o ns fr o m the seco nd or d er derivatives) of(\ [eq:DI 2DS8R HO\])-(\[eq:D 9ZSG4PHI\] ) i s givenby $ $ \b e gin{gathered}L_0 = \frac{\hb a r^{2}}{2 m} \b igg[ \fr ac{1}{\sq r t {\rho}}\fr ac{ \pa rti a l }{ \partial \vec { x }} \ fr ac{\par tia l}{\par tia l \ vec {x} }\sqrt{\rh o} -\fra c{ 1} {4 }\ sin ^{2}\ vartheta \ lef t( \f rac{\ p artial \var phi} {\ pa r tia l \vec{ x }} \ righ t) ^{ 2} - \fr ac {1}{4 } \l e ft( \frac{ \partial\va r thet a} {\ partial \vec{x}} \ri gh t)^{2} \bi gg ] \label { e q:DI3SOL L0HO}
{\partial^{2} \vartheta}{\partial_x_{k} \partial_x_{i}}} +\frac{\partial}{\partial x_{k}} \frac{\partial \rho L_0} {\partial \frac{\partial_\vartheta}{\partial x_{k}_}} -\frac{\partial_\rho L_0}{\partial\vartheta} -\frac{\hbar_\rho}{2} \left(_G_1\cos \varphi -_G_2\sin \varphi _\right) =0 \label{eq:D6DSV3ARTH}\\ -\frac{\partial}{\partial x_{k}} _\frac{\partial}{\partial x_{i}} \frac{\partial \rho_L_0} {\partial\ \frac{\partial^{2}_\varphi}{\partial x_{k} \partial x_{i}}} +\frac{\partial}{\partial x_{k}} \frac{\partial \rho L_0} {\partial \frac{\partial \varphi}{\partial x_{k} }} -\frac{\partial \rho L_0}{\partial\varphi} -\frac{\hbar }{2} \rho_G_3_=0 \label{eq:D9ZSG4PHI} \mbox{.}\end{gathered}$$ The_variable_$S$_does not occur in (\[eq:DI2DS8RHO\])-(\[eq:D9ZSG4PHI\]) in_agreement with our assumptions about_the form_of $L_0$. It is easy to see that_a_proper solution (with_vanishing variational contributions from the second order derivatives) of (\[eq:DI2DS8RHO\])-(\[eq:D9ZSG4PHI\])_is given by $$\begin{gathered} L_0 = \frac{\hbar^{2}}{2m}_\bigg[ \frac{1}{\sqrt{\rho}} \frac{\partial}{\partial \vec{x}} \frac{\partial}{\partial_\vec{x}} \sqrt{\rho} -\frac{1}{4}\sin^{2}_\vartheta \left(_\frac{\partial \varphi}{\partial \vec{x}} \right)^{2} -\frac{1}{4} \left(_\frac{\partial \vartheta}{\partial \vec{x}} \right)^{2} \bigg] \label{eq:DI3SOLL0HO}
u_1,u_2)=\Phi(u_1)+\Phi(u_2)$ and set $\Omega=\Omega_1\times\Omega_2$. Then $$\nabla_G\text{-}\deg(\nabla\Phi, \Omega)=\nabla_G\text{-}\deg(\nabla\Phi_1, \Omega_1)\star\nabla_G\text{-}\deg(\nabla\Phi_2, \Omega_2).$$ \[GLOB\] Fix $\Phi\in C^2_G({\mathcal{H}}\times\Lambda,{\mathbb{R}})$ such that $\nabla_u\Phi(u,\lambda)=Lu-\nabla_u\eta(u,\lambda)$, where the mapping $\nabla_u\eta\colon\Omega\times\Lambda\to{\mathcal{H}}$ is $G$-equivariant and compact. Suppose that $\nabla_u\Phi(0,\lambda)=0$ for every $\lambda\in\Lambda$. If there exist $\gamma_1,\ \gamma_2>0$ such that $$\nabla_G\text{-}\deg(\nabla_u\Phi(\cdot,\lambda_1), B_{\gamma_1}({\mathcal{H}}))\neq\nabla_G\text{-}\deg(\nabla_u\Phi(\cdot,\lambda_2), B_{\gamma}({\mathcal{H}})),$$ then at every path joining $(0,\lambda_1)$ and $(0,\lambda_2)$ exists a global bifurcation point of solutions of the equation $\nabla_u\Phi(u,\lambda)=0$. See [@degree] for properties of the degree and [@Geba; @Rybicki] for the definition of the degree for gradient $G$-equivariant maps. For the general theory of the equivariant degree we refer the reader to [@BKS], [@BKR]. [99]{} Z. Balanov, W. Krawcewicz and H. Steinlein, *Applied Equivariant Degree*, AIMS Series on Differential Equations and Dynamical Systems, American Institute of Mathematical Sciences (2006). Z. Balanov, W. Krawcewicz, H. Ruan *Applied equivariant degree. I. An axiomatic approach to primary degree.* Discrete Contin. Dyn. Syst. **15(3)** (2006), 983-1016. Ch. Budd *Symmetry breaking and semilinear elliptic equations* J.
u_1,u_2)=\Phi(u_1)+\Phi(u_2)$ and set $ \Omega=\Omega_1\times\Omega_2$. Then $ $ \nabla_G\text{-}\deg(\nabla\Phi, \Omega)=\nabla_G\text{-}\deg(\nabla\Phi_1, \Omega_1)\star\nabla_G\text{-}\deg(\nabla\Phi_2, \Omega_2).$$ \[GLOB\ ] Fix $ \Phi\in C^2_G({\mathcal{H}}\times\Lambda,{\mathbb{R}})$ such that $ \nabla_u\Phi(u,\lambda)=Lu-\nabla_u\eta(u,\lambda)$, where the mapping $ \nabla_u\eta\colon\Omega\times\Lambda\to{\mathcal{H}}$ is $ G$-equivariant and compact. presuppose that $ \nabla_u\Phi(0,\lambda)=0 $ for every $ \lambda\in\Lambda$. If there exist $ \gamma_1,\ \gamma_2>0 $ such that $ $ \nabla_G\text{-}\deg(\nabla_u\Phi(\cdot,\lambda_1), B_{\gamma_1}({\mathcal{H}}))\neq\nabla_G\text{-}\deg(\nabla_u\Phi(\cdot,\lambda_2), B_{\gamma}({\mathcal{H}})),$$ then at every way joining $ (0,\lambda_1)$ and $ (0,\lambda_2)$ exists a ball-shaped bifurcation point of solution of the equation $ \nabla_u\Phi(u,\lambda)=0$. See [ @degree ] for property of the academic degree and [ @Geba; @Rybicki ] for the definition of the degree for gradient $ G$-equivariant maps. For the general theory of the equivariant degree we denote the reader to [ @BKS ], [ @BKR ]. [ 99 ] { } Z. Balanov, W. Krawcewicz and H. Steinlein, * Applied Equivariant Degree *, AIMS Series on Differential Equations and Dynamical Systems, American Institute of Mathematical Sciences (2006). Z. Balanov, W. Krawcewicz, H. Ruan * Applied equivariant degree. I. An axiomatic overture to primary degree. * Discrete Contin. Dyn. Syst. * * 15(3) * * (2006), 983 - 1016. Ch. Budd * Symmetry breaking and semilinear elliptic equations * J.
u_1,u_2)=\Pji(u_1)+\Phi(u_2)$ and set $\Omega=\Omena_1\times\Omega_2$. Theu $$\nabla_J\text{-}\def(\nabla\Phk, \Omega)=\nabla_G\text{-}\deg(\nabla\Phi_1, \Onega_1)\sucr\nabla_G\text{-}\deg(\nabla\Ohi_2, \Omega_2).$$ \[HLOB\] Fix $\Phi\mn C^2_G({\mathcal{H}}\timxa\Lambda,{\mathbb{D}})$ suck vhat $\nabla_u\Phi(u,\kambda)=Lu-\natla_u\eta(u,\lambda)$, wfexe the mapping $\nabla_u\eta\colon\Omega\tiies\Lambca\ho{\mathcal{H}}$ is $D$-equpvwriahn cnd compact. Suppose that $\nabla_u\Lhi(0,\lambva)=0$ for every $\lakbda\in\Lambda$. If there exish $\galma_1,\ \gamma_2>0$ such thah $$\nabla_G\texj{-}\seg(\gqbla_u\Phi(\cdot,\uambda_1), B_{\gamma_1}({\mathcal{H}}))\ngq\nabla_G\text{-}\deg(\nabla_u\Phi(\cdot,\lambaa_2), B_{\gcmma}({\mathcal{Y}})),$$ rhej at every peth jopning $(0,\lambda_1)$ and $(0,\lambga_2)$ exisys a global binurcavion point of solutions oh the equation $\nabla_o\Phi(u,\lambdd)=0$. Sze [@degree] for propertuew of jhe dagred ana [@Gtba; @Rgbicki] foc the definjtion of thw degree for gradiemt $T$-equivariant japs. Fjr the general theory of the equivariant gegdee we refer the reader to [@BKS], [@BKR]. [99]{} Z. Balanov, A. Krawcewycz and H. Steinlein, *Applied Equivariant Degree*, AIKS Secids in Aufverential Equations and Dynamical Systems, Amewjcsn Institute of Iathematicak Dcognces (2006). Z. Balanuv, W. Kxzwdewicz, H. Ruan *Applled equyvariqnt degret. I. Am axiomatic approach to prinary degree.* Bisxrete Contin. Dyn. Svst. **15(3)** (2006), 983-1016. Ch. Bubd *Symketry breaking and semilineax elliltic equatilns* J.
u_1,u_2)=\Phi(u_1)+\Phi(u_2)$ and set $\Omega=\Omega_1\times\Omega_2$. Then $$\nabla_G\text{-}\deg(\nabla\Phi, \Omega)=\nabla_G\text{-}\deg(\nabla\Phi_1, \[GLOB\] $\Phi\in C^2_G({\mathcal{H}}\times\Lambda,{\mathbb{R}})$ that $\nabla_u\Phi(u,\lambda)=Lu-\nabla_u\eta(u,\lambda)$, where and Suppose that $\nabla_u\Phi(0,\lambda)=0$ every $\lambda\in\Lambda$. If exist $\gamma_1,\ \gamma_2>0$ such that $$\nabla_G\text{-}\deg(\nabla_u\Phi(\cdot,\lambda_1), B_{\gamma}({\mathcal{H}})),$$ then at every path joining $(0,\lambda_1)$ and $(0,\lambda_2)$ exists a global bifurcation of solutions of the equation $\nabla_u\Phi(u,\lambda)=0$. See [@degree] for properties of the degree [@Geba; for definition the degree for gradient $G$-equivariant maps. For the general theory of the equivariant degree we refer reader to [@BKS], [@BKR]. [99]{} Z. Balanov, W. and H. Steinlein, *Applied Degree*, AIMS Series on Differential and Systems, American of Sciences Z. Balanov, W. H. Ruan *Applied equivariant degree. I. An axiomatic approach to primary degree.* Discrete Contin. Dyn. Syst. **15(3)** 983-1016. Ch. breaking and elliptic J.
u_1,u_2)=\Phi(u_1)+\Phi(u_2)$ and set $\Omega=\OmegA_1\times\OmegA_2$. Then $$\NabLa_G\TeXt{-}\deG(\nabLa\Phi, \Omega)=\nablA_g\texT{-}\deg(\nabla\Phi_1, \Omega_1)\star\nAbla_G\TeXT{-}\deg(\NAbLa\Phi_2, \omega_2).$$ \[GLob\] FIX $\phi\In c^2_G({\MatHcAL{H}}\Times\lamBda,{\mathBb{R}})$ such thaT $\naBlA_u\Phi(u,\lambda)=lU-\nAbla_u\eta(u,\lAmbDa)$, where the maPpiNg $\nablA_u\Eta\COlon\OMegA\timeS\LambdA\To{\mathCal{H}}$ is $G$-eqUiVAriant ANd compaCT. suPposE that $\nabla_u\Phi(0,\lamBDa)=0$ FOr every $\lambda\iN\LambdA$. IF ThERE exIst $\Gamma_1,\ \gamma_2>0$ SuCh thaT $$\Nabla_G\tEXt{-}\DEG(\NabLA_u\Phi(\cdot,\lambDa_1), B_{\gamma_1}({\matHCal{h}}))\neq\naBlA_G\tEXt{-}\deg(\nAbla_u\phI(\CdoT,\lambda_2), B_{\gamMa}({\maThcal{H}})),$$ theN at eveRY path joINing $(0,\lamBda_1)$ and $(0,\LamBda_2)$ ExisTS a GlObaL bIFurCAtIon POinT of solutIoNs Of the EquaTION $\NablA_u\PHi(u,\lAmbda)=0$. see [@degree] for pRopErtiES of The deGree aNd [@GeBa; @rybicKi] for tHe defInItion of the degreE for Gradient $G$-EquIvAriAnT maps. fOr the gEneRal Theory oF the equIVarIaNT DEgRee we refer the readeR tO [@bkS], [@bKR]. [99]{} Z. BalaNov, W. KrAWcEwICz and H. StEiNleIn, *ApPLIed EqUivaRIaNt Degree*, aIMS SeRIeS oN DifferEnTial EqUaTioNs aNd DynAMicaL SysteMs, AmericAn InsTItute of MathemaTIcal Sciences (2006). Z. bAlANOv, w. krawCewIcz, H. Ruan *AppLied EQuivAriaNT dEgrEE. I. An aXiomaTiC ApPRoach to primary degreE.* DIscretE ContIn. Dyn. Syst. **15(3)** (2006), 983-1016. Ch. BuDd *Symmetry BREAking and SemiLInEAr elliptic equaTions* j.
u_1,u_2)=\Phi(u_1)+\Phi(u_ 2)$ and se t $\O meg a=\ Om ega_ 1\ti mes\Omega_2$.T hen$$\nabla_G\text{-}\deg (\nab la \ Phi, \O mega) =\nabla _ G\ t e xt{ -} \d eg( \n a bl a\Phi _1, \Omega _1)\star\n abl a_ G\text{-}\de g (\ nabla\Phi_ 2,\Omega_2).$$ \ [GLOB\ ]Fix $\Phi \in C^2_ G({\ma t hcal{H }}\times\ La m bda,{\ m athbb{R } } )$ suc h that $\nabla_u\ P hi ( u,\lambda)=Lu- \nabla _u \ et a ( u,\ lam bda)$, whe re them apping$ \n a b l a_u \ eta\colon\Ome ga\times\La m bda \to{\m at hca l {H}}$is $G $- e qui variant and com pact. Sup pose t h at $\na b la_u\Ph i(0,\l amb da) =0$f or e ver y$ \la m bd a\i n \La mbda$. I fth ere e xist $ \ g amma _1, \ \g amma_ 2>0$ such tha t $ $\na b la_ G\tex t{-}\ deg( \n abla_ u\Phi( \cdot ,\ lambda_1), B_{\ gamm a_1}({\ma thc al {H} }) )\neq \ nabla_ G\t ext {-}\deg (\nabla _ u\P hi ( \ c do t,\lambda_2), B_{\ ga m m a} ({\mathc al{H}} ) ), $$ then atev ery pat h joini ng $ ( 0, \lambda_ 1)$ an d $ (0 ,\lambd a_ 2)$ ex is tsa g lobal bifu rcatio n pointof so l utions of thee quation $\nab l a_ u \ Ph i (u,\ lam bda)=0$. S ee [ @ degr ee]f or pr o perti es of t h ed egree and [@Geba; @ Ry bicki] forthe definitio n of the d e g r ee for g radi e nt $G$-equivarian t map s. For the generaltheor y of the equivari a n t degree we re fer th e re ader to [@BKS ] , [@B KR ]. [99 ]{} Z. Ba lan ov, W. Kr aw cewicz an d H. Ste in le in ,*Ap plied Equivari an t D eg ree *, AI M S Seri es on Dif fe re n tia l Equat i on s andDy na mica l S ys tems, Ame r ica n Insti tute of M ath e mati ca lScience s (2006). Z.Ba lanov, W.Kr awc ewicz, H . Ruan * Applied equivariant deg r ee. I.Anaxiom atic approach to prima ryd egree. * Disc reteCo nti n . Dyn. S ys t.** 15(3)** (2 0 0 6), 983- 10 16.Ch. Bud d *Symmetry breaki n g a nd semilinear el lipt i c e qua t io n s*J.
u_1,u_2)=\Phi(u_1)+\Phi(u_2)$ and_set $\Omega=\Omega_1\times\Omega_2$._Then $$\nabla_G\text{-}\deg(\nabla\Phi, \Omega)=\nabla_G\text{-}\deg(\nabla\Phi_1, \Omega_1)\star\nabla_G\text{-}\deg(\nabla\Phi_2,_\Omega_2).$$ \[GLOB\] Fix_$\Phi\in_C^2_G({\mathcal{H}}\times\Lambda,{\mathbb{R}})$ such_that_$\nabla_u\Phi(u,\lambda)=Lu-\nabla_u\eta(u,\lambda)$, where the_mapping $\nabla_u\eta\colon\Omega\times\Lambda\to{\mathcal{H}}$ is_$G$-equivariant and compact. Suppose_that $\nabla_u\Phi(0,\lambda)=0$ for_every_$\lambda\in\Lambda$. If there exist $\gamma_1,\ \gamma_2>0$ such that $$\nabla_G\text{-}\deg(\nabla_u\Phi(\cdot,\lambda_1), B_{\gamma_1}({\mathcal{H}}))\neq\nabla_G\text{-}\deg(\nabla_u\Phi(\cdot,\lambda_2), B_{\gamma}({\mathcal{H}})),$$ then at every_path_joining $(0,\lambda_1)$_and_$(0,\lambda_2)$_exists a global bifurcation point_of solutions of the equation_$\nabla_u\Phi(u,\lambda)=0$. See [@degree]_for properties of the degree and [@Geba; @Rybicki]_for_the definition of_the degree for gradient $G$-equivariant maps. For the general_theory of the equivariant degree we_refer the reader_to_[@BKS],_[@BKR]. [99]{} Z. Balanov, W. Krawcewicz_and H. Steinlein, *Applied Equivariant Degree*,_AIMS Series on Differential Equations and_Dynamical Systems, American Institute of Mathematical Sciences_(2006). Z. Balanov, W. Krawcewicz, H._Ruan *Applied equivariant degree. I._An axiomatic_approach to primary degree.* Discrete_Contin. Dyn. Syst._**15(3)** (2006),_983-1016. Ch. Budd_*Symmetry breaking and semilinear elliptic equations*_J.
in the opposite direction. We define the 2-category $2^\prime \text{-}{\operatorname{dgalg}}$ of DG algebras as follows. The objects are DG algebras. For DG algebras ${{\mathcal B}}, {{\mathcal C}}$ the collection of 1-morphisms $1\text{-}\Hom({{\mathcal B}},{{\mathcal C}})$ consists of pairs $(M,\theta)$, where - $M\in D({{\mathcal B}}^0 \otimes {{\mathcal C}})$ and there exists an isomorphism (in $D({{\mathcal C}})$) $\theta :{{\mathcal C}}\to \nu _*M$ (where $\nu _*:D({{\mathcal B}}^0 \otimes {{\mathcal C}})\to D({{\mathcal C}})$ is the functor of restriction of scalars corresponding to the natural homomorphism $\nu :{{\mathcal C}}^0 \to {{\mathcal B}}\otimes {{\mathcal C}}^0$); - and $\theta :M{\stackrel{\bL}{\otimes }}_{{{\mathcal C}}}k\to k$ is an isomorphism in $D({{\mathcal B}}^0)$. The composition of 1-morphisms $$1\text{-}\Hom({{\mathcal B}},{{\mathcal C}})\times 1\text{-}\Hom({{\mathcal C}},{{\mathcal D}})\to 1\text{-}\Hom({{\mathcal B}},{{\mathcal D}})$$ is defined by the tensor product $\cdot \stackrel{\bL}{\otimes}_{{{\mathcal C}}}\cdot $. Given 1-morphisms $(M_1,\theta _1), (M_2,\theta _2)\in 1\text{-}\Hom({{\mathcal B}},{{\mathcal C}})$ a 2-morphism $f: (M_1,\theta _1)\to (M_2,\theta _2)$ is an isomorphism (in $D({{\mathcal B}}^0 \otimes {{\mathcal C}})$) $f:M_1\to M_2$ such that $\theta _1=\theta _2\cdot (f){\stackrel{\bL}{\otimes}}_{{{\mathcal C}}}k$. So in particular the category $1\text{-}\Hom({{\mathcal B}},{{\mathcal C}})$ is a groupoid. Denote by $2^\prime\text{-}{\operatorname{dgart}}$ the full 2-subcategory of $2^\prime\text{-}{\operatorname{dgalg}}$ consisting of artinian DG algebras. Similarly we define the full 2-sub
in the opposite direction. We define the 2 - category $ 2^\prime \text{-}{\operatorname{dgalg}}$ of DG algebra as take after. The objects are DG algebras. For DG algebra $ { { \mathcal B } }, { { \mathcal C}}$ the collection of 1 - morphisms $ 1\text{-}\Hom({{\mathcal B}},{{\mathcal C}})$ consist of pairs $ (M,\theta)$, where - $ M\in D({{\mathcal B}}^0 \otimes { { \mathcal C}})$ and there exists an isomorphism (in $ D({{\mathcal C}})$) $ \theta: { { \mathcal C}}\to \nu _ * M$ (where $ \nu _ *: D({{\mathcal B}}^0 \otimes { { \mathcal C}})\to D({{\mathcal C}})$ is the functor of limitation of scalar corresponding to the natural homomorphism $ \nu: { { \mathcal C}}^0 \to { { \mathcal B}}\otimes { { \mathcal C}}^0 $); - and $ \theta: M{\stackrel{\bL}{\otimes } } _ { { { \mathcal C}}}k\to k$ is an isomorphism in $ D({{\mathcal B}}^0)$. The composition of 1 - morphisms $ $ 1\text{-}\Hom({{\mathcal B}},{{\mathcal C}})\times 1\text{-}\Hom({{\mathcal C}},{{\mathcal D}})\to 1\text{-}\Hom({{\mathcal B}},{{\mathcal D}})$$ is defined by the tensor product $ \cdot \stackrel{\bL}{\otimes}_{{{\mathcal C}}}\cdot $. give 1 - morphisms $ (M_1,\theta _ 1), (M_2,\theta _ 2)\in 1\text{-}\Hom({{\mathcal B}},{{\mathcal C}})$ a 2 - morphism $ f: (M_1,\theta _ 1)\to (M_2,\theta _ 2)$ is an isomorphism (in $ D({{\mathcal B}}^0 \otimes { { \mathcal C}})$) $ f: M_1\to M_2 $ such that $ \theta _ 1=\theta _ 2\cdot (f){\stackrel{\bL}{\otimes}}_{{{\mathcal C}}}k$. So in finical the class $ 1\text{-}\Hom({{\mathcal B}},{{\mathcal C}})$ is a groupoid. Denote by $ 2^\prime\text{-}{\operatorname{dgart}}$ the full 2 - subcategory of $ 2^\prime\text{-}{\operatorname{dgalg}}$ consist of artinian DG algebras. Similarly we define the entire 2 - sub
in the opposite direction. Wt define the 2-cateyiry $2^\prmme \texf{-}{\operatofname{dgalg}}$ of DG algebras as filloww. The objects are DG augebras. Flr DG altebres ${{\mathcal B}}, {{\matidal C}}$ tmz colmcctiou if 1-morphisms $1\tgxt{-}\Hom({{\mathcan B}},{{\mathcal C}})$ cmnridts of pairs $(M,\theta)$, where - $M\in D({{\maehcal B}}^0 \ohimes {{\mathcal S}})$ anc thedv txists an isomorphism (in $D({{\mathcal C}})$) $\thete :{{\mathcal C}}\to \ni _*M$ (where $\nu _*:D({{\mathcal B}}^0 \otlmes {{\mathcal C}})\to D({{\mwthcal C}})$ is the dunctor of rdstriction of scalars dorresponding to the natural hooomor'hism $\nu :{{\majkxal W}}^0 \to {{\mathcel B}}\otpmes {{\mathcal G}}^0$); - and $\theta :K{\stackrel{\bL}{\otikes }}_{{{\marhcal C}}}k\to k$ is an ismmorphism in $D({{\mathsal B}}^0)$. The wojposition of 1-morpyiwms $$1\tgxt{-}\Hok({{\matfxal B}},{{\mzticam C}})\timfs 1\vext{-}\Hom({{\mathdal C}},{{\mathcao D}})\to 1\text{-}\Hom({{\mathcal B}},{{\iqthcal D}})$$ is dsfined br the tensor product $\cdot \stackrel{\bL}{\otikes}_{{{\jathcal C}}}\cdot $. Given 1-moephisms $(M_1,\theta _1), (M_2,\thetw _2)\in 1\text{-}\Hjm({{\mathcal B}},{{\mathcal C}})$ a 2-morphism $f: (M_1,\theta _1)\to (M_2,\theta _2)$ is en isimjfphlsm (in $D({{\mathcal B}}^0 \otimes {{\mathcal C}})$) $f:M_1\to M_2$ sucr tnan $\theta _1=\theta _2\cdob (f){\stackrel{\bL}{\otimes}}_{{{\kahhvwl C}}}k$. So in pxrticular the category $1\text{-}\Jom({{\mathsal B}},{{\nathcal C}})$ is s groupoid. Denote by $2^\prime\twxt{-}{\operatorncme{egart}}$ the full 2-subeategory of $2^\'rime\tgxt{-}{\opetatorname{dgalg}}$ consistiug of zrtinian DG algebras. Rimilarly we defkne tve full 2-sub
in the opposite direction. We define the \text{-}{\operatorname{dgalg}}$ DG algebras follows. The objects algebras B}}, {{\mathcal C}}$ collection of 1-morphisms B}},{{\mathcal C}})$ consists of pairs $(M,\theta)$, - $M\in D({{\mathcal B}}^0 \otimes {{\mathcal C}})$ and there exists an isomorphism (in C}})$) $\theta :{{\mathcal C}}\to \nu _*M$ (where $\nu _*:D({{\mathcal B}}^0 \otimes {{\mathcal C}})\to C}})$ the of of scalars corresponding to the natural homomorphism $\nu :{{\mathcal C}}^0 \to {{\mathcal B}}\otimes {{\mathcal C}}^0$); - $\theta :M{\stackrel{\bL}{\otimes }}_{{{\mathcal C}}}k\to k$ is an isomorphism $D({{\mathcal B}}^0)$. The composition 1-morphisms $$1\text{-}\Hom({{\mathcal B}},{{\mathcal C}})\times 1\text{-}\Hom({{\mathcal D}})\to B}},{{\mathcal D}})$$ defined the product $\cdot \stackrel{\bL}{\otimes}_{{{\mathcal $. Given 1-morphisms $(M_1,\theta _1), (M_2,\theta _2)\in 1\text{-}\Hom({{\mathcal B}},{{\mathcal C}})$ a 2-morphism $f: (M_1,\theta _1)\to (M_2,\theta _2)$ an isomorphism B}}^0 \otimes C}})$) M_2$ that $\theta _1=\theta C}}}k$. So in particular the category is a groupoid. Denote by $2^\prime\text{-}{\operatorname{dgart}}$ the full of $2^\prime\text{-}{\operatorname{dgalg}}$ of artinian DG algebras. Similarly we the full 2-sub
in the opposite direction. We dEfine the 2-caTegorY $2^\prIme \TeXt{-}{\opEratOrname{dgalg}}$ of Dg AlgeBras as follows. The objectS are Dg aLGebrAS. FOr DG aLgebras ${{\MAtHCAl B}}, {{\MaThCal c}}$ tHE cOllecTioN of 1-morpHisms $1\text{-}\HOm({{\mAtHcal B}},{{\mathcal c}})$ CoNsists of paIrs $(m,\theta)$, where - $M\In D({{\MathcaL B}}^0 \OtiMEs {{\matHcaL C}})$ and There eXIsts an IsomorphiSm (IN $D({{\mathCAl C}})$) $\thetA :{{\MAtHcal c}}\to \nu _*M$ (where $\nu _*:D({{\matHCaL b}}^0 \otimes {{\mathcal c}})\to D({{\maThCAl c}})$ IS thE fuNctor of resTrIctioN Of scalaRS cORREspONding to the natUral homomorPHisM $\nu :{{\matHcAl C}}^0 \TO {{\mathcAl B}}\otImES {{\maThcal C}}^0$); - and $\thEta :M{\Stackrel{\bl}{\otimeS }}_{{{\Mathcal c}}}K\to k$ is aN isomoRphIsm In $D({{\mAThCaL B}}^0)$. THe COmpOSiTioN Of 1-mOrphisms $$1\TeXt{-}\hom({{\maThcaL b}},{{\MAThcaL C}})\tImes 1\Text{-}\HOm({{\mathcal C}},{{\matHcaL D}})\to 1\TExt{-}\hom({{\maThcal b}},{{\matHcAl D}})$$ is DefineD by thE tEnsor product $\cdoT \staCkrel{\bL}{\otImeS}_{{{\mAthCaL C}}}\cdoT $. given 1-mOrpHisMs $(M_1,\thetA _1), (M_2,\theta _2)\IN 1\teXt{-}\hOM({{\MaThcal B}},{{\mathcal C}})$ a 2-morPhISM $f: (m_1,\theta _1)\to (m_2,\theta _2)$ IS aN iSOmorphisM (iN $D({{\mAthcAL b}}^0 \otimEs {{\maTHcAl C}})$) $f:M_1\to M_2$ Such thAT $\tHeTa _1=\theta _2\CdOt (f){\staCkRel{\BL}{\oTimes}}_{{{\MAthcAl C}}}k$. So In particUlar tHE category $1\text{-}\HOM({{\mathcal B}},{{\mathCAl c}})$ IS a GRoupOid. denote by $2^\priMe\teXT{-}{\opeRatoRNaMe{dGArt}}$ thE full 2-SuBCaTEgory of $2^\prime\text{-}{\opeRaTornamE{dgalG}}$ consisting of Artinian DG ALGEbras. SimIlarLY wE Define the full 2-sUb
in the opposite direction . We defi ne th e 2 -ca te gory $2^ \prime \text{- } {\op eratorname{dgalg}}$ of DG a lg e bras as foll ows. Th e o b j ect sar e D Ga lg ebras . F or DG a lgebras ${ {\m at hcal B}}, {{ \ ma thcal C}}$ th e collection of 1-mor ph ism s $1\t ext {-}\H om({{\ m athcal B}},{{\m at h cal C} } )$ cons i s ts ofpairs $(M,\theta) $ ,w here - $M\i n D({{ \m a th c a l B }}^ 0 \otimes{{ \math c al C}}) $ a n d the r e exists an i somorphism( in$D({{\ ma thc a l C}}) $) $\ th e ta:{{\mathcal C}} \to \nu _ *M$ (w h ere $\n u _*:D({ {\math cal B} }^0\ ot im es {{ \ ma thc a l C }})\to D ({ {\ mathc al C } } ) $ isthe fun ctorof restrictio n o f sc a lar s cor respo ndin gto th e natu ral h om omorphism $\nu:{{\ mathcal C }}^ 0 \t o{{\ma t hcal B }}\ oti mes {{\ mathcal C}} ^0 $ ) ; - and $\theta :M {\ s t ac krel{\bL }{\oti m es } } _{{{\mat hc alC}}} k \ to k$ isa nisomorph ism in $D ({ {\mathc al B}}^0 )$ . The comp o siti on of1-morphi sms $ $ 1\text{-}\Hom( { {\mathcal B}} , {{ \ m at h calC}} )\times 1\t ext{ - }\Ho m({{ \ ma thc a l C}} ,{{\m at h ca l D}})\to 1\text{-}\ Ho m({{\m athca l B}},{{\math cal D}})$$ i s defined byt he tensor product $\cd ot \stack r el{\bL}{ \otim es}_{{{\ mathcal C } } }\cdot $ . G ive n 1 -mo r p hi sms $(M_1,\th e t a _1 ), (M_2,\ the ta _2)\ in1\t ext {-} \H om({{\mat hcal B}} ,{ {\ ma th cal C}}) $ a 2-mor ph ism $ f:(M_1, \ theta_1)\t o (M _2 ,\ t het a _2)$i sa n iso mo rp hism (i n$D({{ \mat h cal B}}^0\otimes { {\m a thca lC} })$) $f :M_1\to M_2$su ch that $\ th eta _1=\t h e ta _2\cd ot (f){\stackrel{\bL}{\ o times}} _{{ {\mat hcal C}}}k$.Soin par tic u lar th e cate gory$1 \te x t {-}\H o m ({ {\m at hcal B}},{ { \ mat hcalC} })$is a gr oupoid. Denote by$ 2^\ prime\text{-} {\o pera t o rn ame { dg a rt} }$ the f ull 2-subcatego ry of $2^\ pr i me \text{-}{\ o per at orname{ dgalg}} $ con s istingof artini an DG alg eb ras. S imi larly we d efine th e full 2- s ub
in_the opposite_direction. We define the 2-category_$2^\prime \text{-}{\operatorname{dgalg}}$_of_DG algebras_as_follows. The objects_are DG algebras._For DG algebras ${{\mathcal_B}}, {{\mathcal C}}$_the_collection of 1-morphisms $1\text{-}\Hom({{\mathcal B}},{{\mathcal C}})$ consists of pairs $(M,\theta)$, where - $M\in_D({{\mathcal_B}}^0 \otimes_{{\mathcal_C}})$_and there exists an isomorphism_(in $D({{\mathcal C}})$) $\theta :{{\mathcal_C}}\to \nu__*M$ (where $\nu _*:D({{\mathcal B}}^0 \otimes __{{\mathcal C}})\to D({{\mathcal_C}})$ is the functor of restriction of scalars corresponding_to the natural homomorphism $\nu :{{\mathcal_C}}^0 \to_{{\mathcal_B}}\otimes_{{\mathcal C}}^0$); - _and $\theta :M{\stackrel{\bL}{\otimes }}_{{{\mathcal C}}}k\to k$_is an isomorphism in $D({{\mathcal B}}^0)$. The_composition of 1-morphisms $$1\text{-}\Hom({{\mathcal B}},{{\mathcal C}})\times 1\text{-}\Hom({{\mathcal_C}},{{\mathcal D}})\to 1\text{-}\Hom({{\mathcal B}},{{\mathcal D}})$$ is defined_by the tensor product $\cdot _\stackrel{\bL}{\otimes}_{{{\mathcal C}}}\cdot_$. Given 1-morphisms $(M_1,\theta _1),_(M_2,\theta _2)\in 1\text{-}\Hom({{\mathcal B}},{{\mathcal_C}})$ a_2-morphism $f: (M_1,\theta__1)\to (M_2,\theta _2)$ is an isomorphism (in_$D({{\mathcal B}}^0 \otimes_{{\mathcal C}})$) $f:M_1\to M_2$ such that_$\theta__1=\theta _2\cdot (f){\stackrel{\bL}{\otimes}}_{{{\mathcal C}}}k$._So_in_particular the_category $1\text{-}\Hom({{\mathcal B}},{{\mathcal_C}})$_is a_groupoid._Denote by $2^\prime\text{-}{\operatorname{dgart}}$ the full 2-subcategory_of_$2^\prime\text{-}{\operatorname{dgalg}}$ consisting of artinian DG algebras. Similarly_we define the full_2-sub
further out in the stellar atmosphere (see below). We restricted $\log \tau_{\rm c} < -2$ since for larger values of $\tau_{\rm c}$ the outer boundary is placed in a region where the temperature gradient starts to become significant and the periods of all modes become dependent of $\tau_{\rm c}$. It is clear that the periods of the normal modes are independent of $\log \tau_{\rm c}$ i.e. they are independent of the placement of the stellar surface. For the strange modes, the periods vary markedly with $\log \tau_{\rm c}$. This is consistent with the fact that the strange modes are predominantly surface modes largely confined to the region between the hydrogen ionization zone and the stellar surface [@buc97; @sai98]. The reason that the periods of the strange modes increase as the outer boundary is placed at larger radii is that the ratio $z_0$ of the radius of the hydrogen ionization zone to the stellar surface decreases. As shown in the toy models of @buc97 (see their Figure 13), decreasing $z_0$ causes the strange mode period to increase relative to the normal mode periods. Note also that it is the decrease in $z_0$ with luminosity in the sequence of 1.6M$_{\odot}$ models shown in Figure \[nonad\_p+gr-logl\] that causes the strange mode periods to increase faster with luminosity than the normal mode periods. The strange modes in Figure \[nonad\_p+gr-tau\] all have lower growth rates (larger damping rates) than the normal modes. They also cross the periods of the normal modes and in each case it can be seen that mode interaction influences the growth rate as the modes come in and out of resonance. The top panel of Figure \[eigfn\] shows the outer structure of a typical luminous red giant from the centre to the surface for two placements of the outer boundary, corresponding to $\log \tau_{\rm c} = -3$ and $\log \tau_{\rm c} = -8$. The model with $\log \tau_{\rm c} = -8$ has a considerably larger surface radius than the model with $\log \tau_{\rm c} = -3$. The physical structure of the two models is essentially indistinguishable at common radii. The eigenfunctions of the radius perturbation for the 4 lowest order modes are shown in the bottom panel of Figure \[eigfn\]
further out in the stellar atmosphere (visualize downstairs). We restricted $ \log \tau_{\rm c } < -2 $ since for larger values of $ \tau_{\rm c}$ the extinct boundary is placed in a area where the temperature gradient starts to become significant and the periods of all manner become dependent of $ \tau_{\rm c}$. It is clear that the period of the normal modes are independent of $ \log \tau_{\rm c}$ i they are independent of the placement of the stellar surface. For the strange modes, the periods vary markedly with $ \log \tau_{\rm c}$. This is reproducible with the fact that the strange modes are predominantly surface mood largely confined to the region between the hydrogen ionization zone and the stellar surface [ @buc97; @sai98 ]. The rationality that the periods of the strange modes increase as the outer limit is placed at larger radii is that the ratio $ z_0 $ of the radius of the hydrogen ionization zone to the stellar open decreases. As shown in the toy models of @buc97 (see their Figure 13), decreasing $ z_0 $ causes the strange mode time period to increase relative to the normal mode period. Note also that it is the decrease in $ z_0 $ with luminosity in the sequence of 1.6M$_{\odot}$ models indicate in Figure   \[nonad\_p+gr - logl\ ] that causes the strange mode period to increase faster with luminosity than the normal mode periods. The strange modes in Figure   \[nonad\_p+gr - tau\ ] all get lower growth rates (larger damping rates) than the normal mode. They also cross the periods of the normal modes and in each case it can be watch that mode interaction influences the growth pace as the modes issue forth in and out of resonance. The top panel of Figure   \[eigfn\ ] show the outer social organization of a typical luminous red giant from the centre to the open for two placements of the out boundary, correspond to $ \log \tau_{\rm c } = -3 $ and $ \log \tau_{\rm c } = -8$. The model with $ \log \tau_{\rm c } = -8 $ has a well larger surface radius than the model with $ \log \tau_{\rm c } = -3$. The physical structure of the two models is essentially indistinguishable at common radii. The eigenfunctions of the radius disturbance for the 4 lowest order manner are shown in the bottom panel of Figure   \[eigfn\ ]
fugther out in the stellar atmosphere (see below). Xe restdicted $\lug \tau_{\rm c} < -2$ since for largec vaoues if $\tau_{\rm c}$ the outer buundary id placed in e region where tis tempevcture nradiznv starts to becpme signifhcant and the [efilds of all modes become dependent os $\tau_{\rm c}$. Lt is clear thwt tnq pedpobs of the normal modes are indelendent of $\log \tau_{\rm c}$ i.e. they are independent of hhe placement of tje stellar wurfwxe. For the sgrange modts, the perioss vary markedly with $\log \tau_{\rm c}$. Thns is consiwtwnt fith the fart than the strange modes ase predpminantly surfsce moees largely confined vo the region betweeg the hydsoyen ionization zone abd the vtelnar ryrfxce [@bnc97; @aai98]. Thf rxason that fhe periods of the strange modts ybcrease as ths outew foundary is placed at larger radii is tvat the ratio $z_0$ of the raduus of the hydrogen ilnization zone to the stellar surface decreases. As shown it the goy mjawld of @buc97 (see their Figure 13), decreasing $z_0$ causef tne strange mode ieriod to increase rflsjive to the notmal mobs leriods. Note also hhat it is tye decreafe im $z_0$ with luminosity in the wequence of 1.6I$_{\idot}$ models shown nn Figure \[noncd\_p+gr-lpgl\] tnat causes the strange oode periods to increase waster with lumivospty dhan the normal mode perioqs. The strenge kodes iv Fibure \[nogad\_p+gr-tau\] wll have lower growth rahes (lcrger damping rwtes) than the normal modes. They also cross thg parimds of tke norkal modes and in each case nt can bz seen that mode interartion influegces the growdj rate as thx modes cjme un abd out uw resonance. The top panel of Figurw \[eigfn\] shows the omter rfructure of a tvkicql luminous red gixnt fgom tra centre to dhe rurwsce fur two kjagemdnts of the outer boundasy, ckrresponding to $\lob \bau_{\rm c} = -3$ and $\log \tau_{\rm c} = -8$. Thr model with $\log \twu_{\rm r} = -8$ har a cpnsyderably larger surface radius than the moqel with $\log \eau_{\rn c} = -3$. The phvsical structure of the two models is eswentially indistintuishable at common vadii. The eigxnfunceions of dhe radius perturbatuon for the 4 lowext order modes are shkwn in the hottom panel of Figure \[eigfn\]
further out in the stellar atmosphere (see restricted \tau_{\rm c} -2$ since for the boundary is placed a region where temperature gradient starts to become significant the periods of all modes become dependent of $\tau_{\rm c}$. It is clear the periods of the normal modes are independent of $\log \tau_{\rm c}$ i.e. are of placement the stellar surface. For the strange modes, the periods vary markedly with $\log \tau_{\rm c}$. This consistent with the fact that the strange modes predominantly surface modes largely to the region between the ionization and the surface @sai98]. reason that the of the strange modes increase as the outer boundary is placed at larger radii is that the $z_0$ of of the ionization to stellar surface decreases. in the toy models of @buc97 13), decreasing $z_0$ causes the strange mode period increase relative the normal mode periods. Note also it is the decrease in $z_0$ with luminosity the sequence of 1.6M$_{\odot}$ models shown in Figure \[nonad\_p+gr-logl\] that causes the strange mode periods faster with luminosity than normal mode periods. strange in \[nonad\_p+gr-tau\] have lower rates (larger damping rates) than the normal modes. They also cross periods of the normal modes and in each case it seen mode interaction influences growth rate as the come and out of resonance. panel Figure outer of typical luminous red giant the centre to the surface two placements of the \tau_{\rm c} = -3$ and $\log \tau_{\rm c} -8$. The model with $\log \tau_{\rm c} -8$ has a considerably larger surface radius than the model with $\log c} = physical structure of the two models is essentially at common radii. The of the radius perturbation for the 4 lowest order are in the panel of Figure
further out in the stellar atmOsphere (see Below). we rEstRiCted $\Log \tAu_{\rm c} < -2$ since for lARger Values of $\tau_{\rm c}$ the outer BoundArY Is plACeD in a rEgion whERe THE teMpErAtuRe GRaDient StaRts to beCome signifIcaNt And the periodS Of All modes beComE dependent of $\Tau_{\Rm c}$. It iS cLeaR That tHe pEriodS of the NOrmal mOdes are inDePEndent OF $\log \tau_{\RM C}$ i.E. theY are independent of THe PLacement of the sTellar SuRFaCE. for The Strange modEs, The peRIods varY MaRKEDly WIth $\log \tau_{\rm c}$. THis is consisTEnt With thE fAct THat the StranGe MOdeS are predomiNantLy surface Modes lARgely coNFined to The regIon BetWeen THe HyDroGeN IonIZaTioN ZonE and the sTeLlAr surFace [@BUC97; @SAi98]. ThE reAson That tHe periods of thE stRangE ModEs incRease As thE oUter bOundarY is plAcEd at larger radii Is thAt the ratiO $z_0$ oF tHe rAdIus of THe hydrOgeN ioNizatioN zone to THe sTeLLAR sUrface decreases. As sHoWN In The toy moDels of @BUc97 (SeE Their FigUrE 13), deCreaSINg $z_0$ caUses THe Strange mOde perIOd To IncreasE rElativE tO thE noRmal mODe peRiods. NOte also tHat it IS the decrease in $Z_0$ With luminositY In THE sEQuenCe oF 1.6M$_{\odot}$ modelS shoWN in FIgurE \[NoNad\_P+Gr-logL\] that CaUSeS The strange mode perioDs To incrEase fAster with lumiNosity than THE Normal moDe peRIoDS. The strange modEs in FIgure \[nonad\_P+Gr-tau\] all Have lOwer growTh rates (laRGEr dampinG raTes) ThaN thE NOrMal modes. They aLSO croSs The periOds Of the noRmaL moDes And In Each case iT can be seEn ThAt MoDe iNteraCTion inflUeNceS tHe gRowth RAte as tHe modEs coMe In ANd oUt of resONaNCE. The ToP pAnel Of FIgUre \[eiGfn\] sHOws The outeR structurE of A TypiCaL lUminous Red giant from tHe Centre to thE sUrfAce for TWO placemeNts of the outer boundary, coRRespondIng To $\log \Tau_{\rM c} = -3$ and $\log \tAu_{\rM c} = -8$. The mOdeL With $\loG \tau_{\rm C} = -8$ has a CoNsiDERably LARgEr sUrFace radius THAn tHe modEl With $\Log \tau_{\rM c} = -3$. The physical strucTUre Of the two modelS is EsseNTIaLly INdIStiNgUIshABLe at common radii. the eigenfuNcTIoNs of the radIUs pErTurbatiOn for thE 4 loweST order mOdes are shOwn in the bOtTom pANEl oF Figure \[eigFn\]
further out in the stella r atmosphe re (s eebel ow ). W e re stricted $\log \tau _{\rm c} < -2$ since f or la rg e r va l ue s of$\tau_{ \ rm c }$th eout er bo undar y i s place d in a reg ion w here the tem p er ature grad ien t starts tobec ome si gn ifi c ant a ndthe p eriods of all modes be co m e depe n dent of $ \t au_{ \rm c}$. It is c l ea r that the peri ods of t h en o rma l m odes are i nd epend e nt of $ \ lo g \ tau _ {\rm c}$ i.e. they are i n dep endent o f t h e plac ement o f th e stellar s urfa ce. For t he str a nge mod e s, theperiod s v ary mar k ed ly wi th $\l o g\ta u _{\ rm c}$.Th is is c onsi s t e n t wi ththefactthat the stra nge mod e s a re pr edomi nant ly surf ace mo des l ar gely confined t o th e regionbet we enth e hyd r ogen i oni zat ion zon e and t h e s te l l a rsurface [@buc97; @ sa i 9 8] . The re ason t h at t h e period softhes t range mod e sincrease as th e o ut er boun da ry ispl ace d a t lar g er r adii i s that t he ra t io $z_0$ of th e radius of th e h y d ro g en i oni zation zone tot he s tell a rsur f ace d ecrea se s .A s shown in the toymo dels o f @bu c97 (see thei r Figure 1 3 ) , decreas ing$ z_ 0 $ causes the s trang e mode per i od to in creas e relati ve to the n ormal mo deper iod s.N o te also that it i s th edecreas e i n $z_0$ wi thlum ino si ty in the sequenc eof 1 .6 M$_ {\odo t }$ model ssho wn in Figu r e \[no nad\_ p+gr -l og l \]that ca u se s thest ra ngemod eperio ds t o in creasefaster wi thl umin os it y thanthe normal mo de periods. T hestrang e modes in Figure \[nonad\_p+gr-t a u\] all ha ve lo wergrowth ra tes (larg erd amping rates ) tha nthe n ormal m od es. T hey also c r o ssthe p er iods of the normal modes andi n e ach case it c anbe s e e ntha t m o dein t era c t ion influencesthe growth r a te as the mo d esco me in a nd outof re s onance. The top panel of F igur e \[e igfn\] sho ws the o uter stru c tureo fa typ ica l lumi no usred g iant f r omthe c entreto the s urfac efor twoplacements of the outer bound ary,cor respondin g t o $\ log \tau_ {\rm c} = -3$and $\ log \ tau _ {\rmc} = -8 $.T he mo delw ith $\log \t au_ { \ rm c} = -8$ h a s a c onsid era b ly lar gersurface radius th a n the model wi th $ \ l og\ta u _{\r mc} = -3$. Thephy si c a l struct ur e of the tw o models i s esse ntiall y indi stingui s h ab l e at c ommo n r adii. Th e e ig e nfuncti on so f theradi us pertu rbatio n for t he 4 lowest orde r mod e s ares how n inth e botto m pan el of Figu re \[eigfn\ ]
further_out in_the stellar atmosphere (see_below). We_restricted_$\log \tau_{\rm_c}_< -2$ since_for larger values_of $\tau_{\rm c}$ the_outer boundary is_placed_in a region where the temperature gradient starts to become significant and the periods_of_all modes_become_dependent_of $\tau_{\rm c}$. It is clear_that the periods of the_normal modes_are independent of $\log \tau_{\rm c}$ i.e. they_are_independent of the_placement of the stellar surface. For the strange modes,_the periods vary markedly with $\log_\tau_{\rm c}$. This_is_consistent_with the fact that_the strange modes are predominantly surface_modes largely confined to the region_between the hydrogen ionization zone and the_stellar surface [@buc97; @sai98]. The reason_that the periods of the_strange modes_increase as the outer boundary_is placed at_larger radii_is that the_ratio $z_0$ of the radius of_the hydrogen ionization_zone to the stellar surface decreases._As_shown in the_toy_models_of @buc97_(see their Figure_13),_decreasing $z_0$_causes_the strange mode period to increase_relative_to the normal mode periods. Note also_that it is the_decrease_in $z_0$ with luminosity_in the sequence of 1.6M$_{\odot}$_models shown in Figure \[nonad\_p+gr-logl\] that causes_the strange_mode periods_to increase faster with luminosity than the normal mode periods. The strange_modes in Figure \[nonad\_p+gr-tau\] all have lower_growth rates (larger damping_rates) than_the_normal modes. They_also_cross the_periods of the normal modes and in_each case_it can be seen that mode_interaction influences the growth_rate_as the modes come in and_out of resonance. The top panel of_Figure \[eigfn\] shows the outer structure_of_a_typical luminous red giant from_the centre to the surface for_two placements of_the outer boundary, corresponding to $\log \tau_{\rm_c}_= -3$ and $\log \tau_{\rm c}_=_-8$. The model with $\log \tau_{\rm_c}_=_-8$ has a considerably larger_surface radius than the model with_$\log \tau_{\rm c} = -3$. The physical structure of the_two models is_essentially indistinguishable at common radii. The_eigenfunctions_of_the radius perturbation for the 4 lowest order modes are_shown in_the bottom panel_of Figure \[eigfn\]
can be written as $$\label{eq: 5.4} \delta \mbox{\boldmath$r$} \left( t +\delta t \right) \equiv \mbox{\boldmath$a$}(\mbox{\boldmath$x$}_0 \left( t \right),t) \delta \mbox{\boldmath$r$} \left( t \right) = \left[ \mbox{\boldmath$I$} + \delta \mbox{\boldmath$a$} \left(\mbox{\boldmath$x$}_0\left( t \right),t \right) \right] \delta \mbox{\boldmath$r$} \left( t \right)$$ where $$\label{eq: 5.5} \left[ \delta \mbox{\boldmath$a$} \left(\mbox{\boldmath$x$}_0\left( t \right),t \right) \right]_{ij} = \left(\frac{\partial u_i \left(\mbox{\boldmath$x$}_{0}(t), t\right)}{\partial x_j} \right) \sqrt{\delta t}.$$ The elements of the matrix $ \delta {\bm a}$ are random variables constructed from the second derivatives of the velocity field potentials evaluated at time $t$ and position ${\bm x}_0(t)$ : $$\label{eq: 5.6} \delta {\bf a}=\left(\begin{array}{cc} \psi_{xy}+\beta \phi_{xx} & -\psi_{yy}+\beta \phi_{xy}\cr \psi_{xx}+\beta \phi_{xy} & -\psi_{xy}+\beta \phi_{yy} \end{array}\right)\sqrt{\delta t}.$$ All the derivatives have mean value zero, and are normalized so that their non-zero covariances are: $$\label{eq: 5.7a} \langle \psi_{xx}^2\rangle = \langle\psi_{yy}^2\rangle =3 \ \, \ \ \ \ \langle \psi_{xx}\psi_{yy}\rangle = \langle \psi_{xy}^2\rangle =1\,$$ and similarly for derivatives of $\phi$. Also, since the model is rotationally invariant, and since $\delta \tilde{{\bf a}}_{n}$, defined by (\[eq: 2.1x\]) is a
can be written as $ $ \label{eq: 5.4 } \delta \mbox{\boldmath$r$ } \left (t + \delta t \right) \equiv \mbox{\boldmath$a$}(\mbox{\boldmath$x$}_0 \left (t \right),t) \delta \mbox{\boldmath$r$ } \left (t \right) = \left [ \mbox{\boldmath$I$ } + \delta \mbox{\boldmath$a$ } \left(\mbox{\boldmath$x$}_0\left (t \right),t \right) \right ] \delta \mbox{\boldmath$r$ } \left (t \right)$$ where $ $ \label{eq: 5.5 } \left [ \delta \mbox{\boldmath$a$ } \left(\mbox{\boldmath$x$}_0\left (t \right),t \right) \right]_{ij } = \left(\frac{\partial u_i \left(\mbox{\boldmath$x$}_{0}(t), t\right)}{\partial x_j } \right) \sqrt{\delta t}.$$ The elements of the matrix $ \delta { \bm a}$ are random variable star construct from the second derivatives of the velocity field electric potential evaluated at time $ t$ and put $ { \bm x}_0(t)$: $ $ \label{eq: 5.6 } \delta { \bf a}=\left(\begin{array}{cc } \psi_{xy}+\beta \phi_{xx } & -\psi_{yy}+\beta \phi_{xy}\cr \psi_{xx}+\beta \phi_{xy } & -\psi_{xy}+\beta \phi_{yy } \end{array}\right)\sqrt{\delta t}.$$ All the derivatives have mean value zero, and are normalize so that their non - zero covariances are: $ $ \label{eq: 5.7a } \langle \psi_{xx}^2\rangle = \langle\psi_{yy}^2\rangle =3 \ \, \ \ \ \ \langle \psi_{xx}\psi_{yy}\rangle = \langle \psi_{xy}^2\rangle = 1\,$$ and similarly for derivatives of $ \phi$. besides, since the model is rotationally invariant, and since $ \delta \tilde{{\bf a}}_{n}$, defined by (\[eq: 2.1x\ ]) is a
caj be written as $$\label{eq: 5.4} \delta \mbox{\boldmath$c$} \left( f +\delta g \right) \equiv \mbox{\bolvmaty$a$}(\mboz{\boldmath$x$}_0 \left( t \right),t) \delta \mvox{\biodmath$r$} \lehf( t \rigmc) = \peft[ \mbox{\boldmath$I$} + \delta \mbmx{\boldmath$a$} \ueyt(\mbox{\boldmath$x$}_0\left( t \right),t \right) \wight] \felta \mbox{\boldiath$g$} \jeft( n \vight)$$ where $$\label{eq: 5.5} \left[ \delta \mbox{\bondmath$a$} \left(\mnox{\boldmath$x$}_0\left( t \right),t \gighh) \right]_{ij} = \left(\fgac{\partial o_j \lqdt(\mbox{\boldmagh$x$}_{0}(t), t\righu)}{\pcrtial x_j} \rjght) \sqrt{\delta t}.$$ The elements ow the matrix $ \dglfa {\tm a}$ are raidom vwriables constructed from tne second deriyativxs od the velocity field 'otentials evaluated at time $d$ cnd position ${\bm x}_0(t)$ : $$\lqbwl{eq: 5.6} \geltd {\bf q}=\lewt(\btgii{arday}{cc} \pdi_{xb}+\beta \phi_{xx} & -\psi_{yy}+\beta \phi_{xy}\cr \psi_{xx}+\beta \khi_{vj} & -\psi_{xy}+\beta \lhi_{yy} \egd{wrray}\right)\sqrt{\delta t}.$$ All the derivativts habe mean value zero, and qre normalized so thaj their nog-zero covariances are: $$\label{eq: 5.7a} \langle \psi_{xx}^2\rangle = \lanjld\psn_{nn}^2\raneoe =3 \ \, \ \ \ \ \langle \psi_{xx}\psi_{yy}\rangle = \langle \psi_{xy}^2\wznblv =1\,$$ and similarly nor derivatives of $\pji$. Wlso, since thg model is rotationally invagiant, agd sibce $\delta \tilce{{\bf a}}_{n}$, defined by (\[eq: 2.1x\]) is q
can be written as $$\label{eq: 5.4} \delta t t \right) \mbox{\boldmath$a$}(\mbox{\boldmath$x$}_0 \left( t \right) \left[ \mbox{\boldmath$I$} + \mbox{\boldmath$a$} \left(\mbox{\boldmath$x$}_0\left( t \right) \right] \delta \mbox{\boldmath$r$} \left( t where $$\label{eq: 5.5} \left[ \delta \mbox{\boldmath$a$} \left(\mbox{\boldmath$x$}_0\left( t \right),t \right) \right]_{ij} = \left(\frac{\partial \left(\mbox{\boldmath$x$}_{0}(t), t\right)}{\partial x_j} \right) \sqrt{\delta t}.$$ The elements of the matrix $ \delta a}$ random constructed the second derivatives of the velocity field potentials evaluated at time $t$ and position ${\bm x}_0(t)$ $$\label{eq: 5.6} \delta {\bf a}=\left(\begin{array}{cc} \psi_{xy}+\beta \phi_{xx} & \phi_{xy}\cr \psi_{xx}+\beta \phi_{xy} & \phi_{yy} \end{array}\right)\sqrt{\delta t}.$$ All the have value zero, are so their non-zero covariances $$\label{eq: 5.7a} \langle \psi_{xx}^2\rangle = \langle\psi_{yy}^2\rangle =3 \ \, \ \ \ \ \langle \psi_{xx}\psi_{yy}\rangle = \langle =1\,$$ and derivatives of Also, the is rotationally invariant, $\delta \tilde{{\bf a}}_{n}$, defined by (\[eq:
can be written as $$\label{eq: 5.4} \deltA \mbox{\boldmAth$r$} \lEft( T +\deLtA t \riGht) \eQuiv \mbox{\boldmaTH$a$}(\mbOx{\boldmath$x$}_0 \left( t \right),t) \Delta \MbOX{\bolDMaTh$r$} \leFt( t \righT) = \LeFT[ \MboX{\bOlDmaTh$i$} + \DeLta \mbOx{\bOldmath$A$} \left(\mbox{\bOldMaTh$x$}_0\left( t \righT),T \rIght) \right] \dEltA \mbox{\boldmatH$r$} \lEft( t \riGhT)$$ whERe $$\labEl{eQ: 5.5} \left[ \Delta \mBOx{\boldMath$a$} \left(\MbOX{\boldmATh$x$}_0\left( T \RIgHt),t \rIght) \right]_{ij} = \left(\frAC{\pARtial u_i \left(\mboX{\boldmAtH$X$}_{0}(t), T\RIghT)}{\paRtial x_j} \rigHt) \Sqrt{\dELta t}.$$ The ELeMENTs oF The matrix $ \deltA {\bm a}$ are randOM vaRiableS cOnsTRucted From tHe SEcoNd derivativEs of The velociTy fielD PotentiALs evaluAted at TimE $t$ aNd poSItIoN ${\bm X}_0(t)$ : $$\LAbeL{Eq: 5.6} \DelTA {\bf A}=\left(\begIn{ArRay}{cc} \Psi_{xY}+\BETA \phi_{Xx} & -\pSi_{yy}+\Beta \pHi_{xy}\cr \psi_{xx}+\beTa \pHi_{xy} & -\PSi_{xY}+\beta \Phi_{yy} \End{aRrAy}\rigHt)\sqrt{\Delta T}.$$ ALl the derivativeS havE mean valuE zeRo, And ArE normALized sO thAt tHeir non-Zero covARiaNcES ARe: $$\Label{eq: 5.7a} \langle \psi_{xX}^2\rANGlE = \langle\pSi_{yy}^2\raNGlE =3 \ \, \ \ \ \ \lANgle \psi_{xX}\pSi_{yY}\ranGLE = \langLe \psI_{Xy}^2\Rangle =1\,$$ anD similARlY fOr derivAtIves of $\PhI$. AlSo, sInce tHE modEl is roTationalLy invARiant, and since $\dELta \tilde{{\bf a}}_{n}$, dEFiNED bY (\[Eq: 2.1x\]) iS a
can be written as $$\labe l{eq: 5.4} \de lta \ mbox {\bo ldmath$r$} \le f t( t +\delta t \right) \eq uiv \ mb ox{\b oldmath $ a$ } ( \mb ox {\ bol dm a th $x$}_ 0 \left( t \ rig ht ),t) \delta\ mb ox{\boldma th$ r$} \left( t \r ight)= \left [ \mbox {\bold m ath$I$ } + \delt a\ mbox{\ b oldmath $ a $} \left(\mbox{\bo l dm a th$x$}_0\left( t \ri gh t ), t \ri ght ) \right] \d e lta \mb o x{ \ b o ldm a th$r$} \left( t \right)$ $ wh ere $$ \l abe l {eq: 5 .5} \ l eft [ \delta \m box{ \boldmath $a$} \ l eft(\mb o x{\bold math$x $}_ 0\l eft( t\r igh t) , t \ r ig ht) \ri ght]_{ij }= \l eft( \ f r a c{\p art ialu_i \ left(\mbox{\b old math $ x$} _{0}( t), t \rig ht )}{\p artial x_j} \ right) \sqrt{\d elta t}.$$ Th e e le men ts of t h e matr ix$ \ delta { \bm a}$ are r a n d om variables constru ct e d f rom thesecond de ri v atives o fthe vel o c ity f ield po tentials evalu a te dat time $ t$ and p osi tio n ${\ b m x} _0(t)$ : $$\la bel{e q : 5.6} \delta{ \bf a}=\left( \ be g i n{ a rray }{c c} \psi_{xy }+\b e ta \ phi_ { xx } & -\psi _{yy} +\ b et a \phi_{xy}\cr \psi _{ xx}+\b eta \ phi_{xy} & -\ psi_{xy}+\ b e t a \phi_{ yy}\ en d {array}\right) \sqrt {\delta t} . $$ All t he de rivative s have me a n value z ero , a ndare n or malized so th a t the ir non-ze rocovaria nce s a re: $$ \l abel{eq:5.7a} \l an gl e\p si_ {xx}^ 2 \rangle=\la ng le\ psi_{ y y}^2\r angle =3 \ \ , \\ \ \ \ la n g le \ ps i_ {xx} \ps i_ {yy}\ rang l e = \langl e \psi_{x y}^ 2 \ran gl e=1\,$$and similarly f or derivat iv esof $\p h i $. Also, since the model is rot a tionall y i nvari ant, and sinc e $ \delta \t i lde{{\ bf a}} _{n}$ ,def i n ed by ( \[ eq: 2 .1x\]) isa
can_be written_as $$\label{eq: 5.4} _ _\delta_\mbox{\boldmath$r$} \left(_t_+\delta t \right)_\equiv _ _ \mbox{\boldmath$a$}(\mbox{\boldmath$x$}_0 __ \left( t \right),t) \delta \mbox{\boldmath$r$} \left( t \right) = __ _\left[__\mbox{\boldmath$I$} + \delta \mbox{\boldmath$a$} _ \left(\mbox{\boldmath$x$}_0\left( t_\right),t \right)_ \right] \delta \mbox{\boldmath$r$} \left(_t_\right)$$ where $$\label{eq:_5.5} \left[ \delta \mbox{\boldmath$a$} \left(\mbox{\boldmath$x$}_0\left( t \right),t \right)_\right]_{ij} = \left(\frac{\partial u_i_\left(\mbox{\boldmath$x$}_{0}(t), t\right)}{\partial x_j}_\right)_\sqrt{\delta_t}.$$ The elements of_the matrix $ \delta {\bm a}$_are random variables constructed from the_second derivatives of the velocity field potentials_evaluated at time $t$ and position_${\bm x}_0(t)$ : $$\label{eq: 5.6} \delta_{\bf a}=\left(\begin{array}{cc} \psi_{xy}+\beta_\phi_{xx} & -\psi_{yy}+\beta \phi_{xy}\cr \psi_{xx}+\beta_\phi_{xy} & -\psi_{xy}+\beta_\phi_{yy} \end{array}\right)\sqrt{\delta t}.$$_All the derivatives_have mean value zero, and are_normalized so that_their non-zero covariances are: $$\label{eq: 5.7a} \langle_\psi_{xx}^2\rangle_= \langle\psi_{yy}^2\rangle =3_ \_\,_\ \_\ \ \langle_\psi_{xx}\psi_{yy}\rangle_= \langle_\psi_{xy}^2\rangle_=1\,$$ and similarly for derivatives of_$\phi$._Also, since the model is rotationally invariant,_and since $\delta \tilde{{\bf_a}}_{n}$,_defined by (\[eq: 2.1x\])_is a
}=c^{3}a$. The group von Neumann algebra associated with $G$ is a type II$_{1}$ factor, and the preceding corollary easily implies that $\mathfrak{K}_{2}(L(G))=0$ and $\delta _{0}(L(G))\leq1.$ The next two corollaries follows directly from Corollary 3. Suppose $\mathcal M$ is a nonprime II$_1$ factor, i.e. $\mathcal M\simeq \mathcal N_1\otimes \mathcal N_2$ for some II$_1$ subfactors $ \mathcal N_1, \mathcal N_2$. Then $\mathfrak{K}_{2}( \mathcal{M})=0$, $\delta_{0}( \mathcal{M})\leq1$. Thus $\mathcal{M}$ is not \*-isomorphic to any $L(F(n))$ for $n\geq2$. If $\mathcal{M}=L(SL(\mathbb{Z},2m+1))$ is the group von Neumann algebra associated with $SL(\mathbb{Z},2m+1)$ (the special linear group with integer entries) for $m\geq1$, then $\mathfrak{K}_{2}( \mathcal{M})=0$, $\delta_{0}( \mathcal{M})\leq1$. Thus $\mathcal{M}$ is not \*-isomorphic to any $L(F(n))$ for $n\geq2$. In [@GePopa] L. Ge and S. Popa defined a type II$_{1}$ factor to be *weakly* $n$*-thin*, if it contains hyperfinite subalgebras $\mathcal{R}_{0},\mathcal{R}_{1}$ and $n$ vectors $\xi_{1},\ldots,\xi_{n}$ in $L^{2}\left( \mathcal{M},\tau\right) $ such that $\mathcal{M}=\overline {span}^{\left\Vert { \cdot }\right\Vert _{2}}\left( \mathcal{R}_{0}\{\xi_{1},\ldots,\xi_{n}\}\mathcal{R}_{1}\right) .$ They showed that $L(F_{m})$ is not weakly $n$-thin for $m> 2+2n$. Motivated by these facts, we have the following definition. A type II$_{1}$ factor $\mathcal{M}$ with the tracial state
} = c^{3}a$. The group von Neumann algebra associated with $ G$ is a type II$_{1}$ factor, and the preceding corollary well entail that $ \mathfrak{K}_{2}(L(G))=0 $ and $ \delta _ { 0}(L(G))\leq1.$ The next two corollaries follow immediately from Corollary 3. Suppose $ \mathcal M$ is a nonprime II$_1 $ factor, i.e. $ \mathcal M\simeq \mathcal N_1\otimes \mathcal N_2 $ for some II$_1 $ subfactors $ \mathcal N_1, \mathcal N_2$. Then $ \mathfrak{K}_{2 } (\mathcal{M})=0 $, $ \delta_{0 } ( \mathcal{M})\leq1$. Thus $ \mathcal{M}$ is not \*-isomorphic to any $ L(F(n))$ for $ n\geq2$. If $ \mathcal{M}=L(SL(\mathbb{Z},2m+1))$ is the group von Neumann algebra associate with $ SL(\mathbb{Z},2m+1)$ (the special linear group with integer entries) for $ m\geq1 $, then $ \mathfrak{K}_{2 } (\mathcal{M})=0 $, $ \delta_{0 } ( \mathcal{M})\leq1$. therefore $ \mathcal{M}$ is not \*-isomorphic to any $ L(F(n))$ for $ n\geq2$. In [ @GePopa ] L. Ge and S. Popa define a type II$_{1}$ gene to be * weakly * $ n$*-thin *, if it contains hyperfinite subalgebras $ \mathcal{R}_{0},\mathcal{R}_{1}$ and $ n$ vector $ \xi_{1},\ldots,\xi_{n}$ in $ L^{2}\left ( \mathcal{M},\tau\right) $ such that $ \mathcal{M}=\overline { span}^{\left\Vert { \cdot } \right\Vert _ { 2}}\left ( \mathcal{R}_{0}\{\xi_{1},\ldots,\xi_{n}\}\mathcal{R}_{1}\right) .$ They indicate that $ L(F_{m})$ is not weakly $ n$-thin for $ m > 2 + 2n$. Motivated by these facts, we have the following definition. A type II$_{1}$ component $ \mathcal{M}$ with the tracial state
}=c^{3}a$. Hhe group von Neumann alnebra associated with $G$ is a fype II$_{1}$ wactor, and the preceding corlloary tcsily implies that $\mxthfrak{K}_{2}(L(H))=0$ and $\deota _{0}(L(J))\leq1.$ The next two corollavnes fkplowv directly from Corollary 3. Suppose $\mathcdl M$ is a nonprime II$_1$ factor, i.e. $\mathcal I\simeq \kahhcal N_1\otimes \iathbaj N_2$ ror some II$_1$ subfactors $ \mathcal N_1, \jathcal N_2$. Then $\mathftak{K}_{2}( \mathcal{M})=0$, $\delta_{0}( \mathcal{L})\leq1$. Thus $\mathcal{M}$ is jot \*-isomorpyic ei any $L(F(n))$ fof $n\geq2$. If $\mathcal{M}=L(SL(\majhbb{Z},2m+1))$ is the group von Neumann xlgebxa associatgb wihv $SL(\mathbb{Z},2n+1)$ (the special lincsr grogp with integer entrics) foc $m\gwq1$, then $\mathfrak{K}_{2}( \matical{M})=0$, $\delta_{0}( \mathcal{M})\lgq1$. Thus $\madheal{M}$ is not \*-isomorphix ro ani $L(F(n))$ for $b\ged2$. In [@GxPola] L. Gf aid S. Popa dsfined a type II$_{1}$ factor to be *eewjly* $n$*-thin*, if jt coneayns hyperfinite subalgebras $\mathcal{R}_{0},\matvcam{R}_{1}$ and $n$ vectors $\xi_{1},\ldotw,\xi_{n}$ in $L^{2}\left( \mathcap{M},\tau\righe) $ such that $\mathcal{M}=\overline {span}^{\left\Vert { \cdot }\sight\Teft _{2}}\oent( \oqtjcal{R}_{0}\{\xi_{1},\ldots,\xi_{n}\}\mathcal{R}_{1}\right) .$ They showed thae $L(G_{m})$ is not weakly $n$-thin for $k> 2+2j$. Kjtivated by tfese fcdta, we have the folllwing dgfinituon. A type II$_{1}$ gactor $\mathcal{M}$ with the trqcial state
}=c^{3}a$. The group von Neumann algebra associated is type II$_{1}$ and the preceding and _{0}(L(G))\leq1.$ The next corollaries follows directly Corollary 3. Suppose $\mathcal M$ is nonprime II$_1$ factor, i.e. $\mathcal M\simeq \mathcal N_1\otimes \mathcal N_2$ for some II$_1$ $ \mathcal N_1, \mathcal N_2$. Then $\mathfrak{K}_{2}( \mathcal{M})=0$, $\delta_{0}( \mathcal{M})\leq1$. Thus $\mathcal{M}$ is \*-isomorphic any for If $\mathcal{M}=L(SL(\mathbb{Z},2m+1))$ is the group von Neumann algebra associated with $SL(\mathbb{Z},2m+1)$ (the special linear group with entries) for $m\geq1$, then $\mathfrak{K}_{2}( \mathcal{M})=0$, $\delta_{0}( \mathcal{M})\leq1$. $\mathcal{M}$ is not \*-isomorphic any $L(F(n))$ for $n\geq2$. In L. and S. defined type factor to be $n$*-thin*, if it contains hyperfinite subalgebras $\mathcal{R}_{0},\mathcal{R}_{1}$ and $n$ vectors $\xi_{1},\ldots,\xi_{n}$ in $L^{2}\left( \mathcal{M},\tau\right) $ such that {span}^{\left\Vert { _{2}}\left( \mathcal{R}_{0}\{\xi_{1},\ldots,\xi_{n}\}\mathcal{R}_{1}\right) They that is not weakly $m> 2+2n$. Motivated by these facts, following definition. A type II$_{1}$ factor $\mathcal{M}$ with tracial state
}=c^{3}a$. The group von Neumann algebRa associatEd witH $G$ iS a tYpE II$_{1}$ fActoR, and the precediNG corOllary easily implies thaT $\mathFrAK{K}_{2}(L(G))=0$ ANd $\Delta _{0}(l(G))\leq1.$ ThE NeXT Two CoRoLlaRiES fOllowS diRectly fRom CorollaRy 3. SUpPose $\mathcal M$ IS a Nonprime II$_1$ FacTor, i.e. $\mathcal m\siMeq \matHcAl N_1\OTimes \MatHcal N_2$ For somE iI$_1$ subfActors $ \matHcAL N_1, \mathCAl N_2$. Then $\MAThFrak{k}_{2}( \mathcal{M})=0$, $\delta_{0}( \matHCaL{m})\leq1$. Thus $\mathcaL{M}$ is noT \*-iSOmORPhiC to Any $L(F(n))$ for $n\GeQ2$. If $\maTHcal{M}=L(Sl(\MaTHBB{Z},2m+1))$ IS the group von NEumann algebRA asSociatEd WitH $sL(\mathBb{Z},2m+1)$ (tHe SPecIal linear grOup wIth integeR entriES) for $m\geQ1$, Then $\matHfrak{K}_{2}( \MatHcaL{M})=0$, $\deLTa_{0}( \MaThcAl{m})\Leq1$. tHuS $\maTHcaL{M}$ is not \*-iSoMoRphic To anY $l(f(N))$ For $n\Geq2$. in [@Gepopa] L. ge and S. Popa defIneD a tyPE II$_{1}$ FactoR to be *WeakLy* $N$*-thin*, If it coNtainS hYperfinite subalGebrAs $\mathcal{r}_{0},\maThCal{r}_{1}$ aNd $n$ veCTors $\xi_{1},\LdoTs,\xI_{n}$ in $L^{2}\leFt( \mathcAL{M},\tAu\RIGHt) $ Such that $\mathcal{M}=\ovErLINe {Span}^{\left\vert { \cdOT }\rIgHT\Vert _{2}}\lefT( \mAthCal{R}_{0}\{\XI_{1},\Ldots,\Xi_{n}\}\mAThCal{R}_{1}\righT) .$ They sHOwEd That $L(F_{m})$ Is Not weaKlY $n$-tHin For $m> 2+2n$. mOtivAted by These facTs, we hAVe the following DEfinition. A typE iI$_{1}$ FACtOR $\matHcaL{M}$ with the trAciaL StatE
}=c^{3}a$. The group von N eumann alg ebraass oci at ed w ith$G$ is a typeI I$_{ 1}$ factor, and the pr ecedi ng coro l la ry ea sily im p li e s th at $ \ma th f ra k{K}_ {2} (L(G))= 0$ and $\d elt a_{0}(L(G))\l e q1 .$ The ne xttwo corollar ies follo ws di r ectly fr om Co rollar y 3. S uppose $\ ma t hcal M $ is a n o n pr imeII$_1$ factor, i. e .$ \mathcal M\sim eq \ma th c al N _1\ oti mes \mathc al N_2$ for som e I I $ _ 1$s ubfactors $ \ mathcal N_1 , \m athcal N _2$ . Then$\mat hf r ak{ K}_{2}( \ma thca l{M})=0$, $\del t a_{0}(\ mathcal {M})\l eq1 $.Thus $\ ma thc al { M}$ is no t \* -isomorp hi cto an y $L ( F ( n ))$for $n\ geq2$ . If $\mathc al{ M}=L ( SL( \math bb{Z} ,2m+ 1) )$ is the g roupvo n Neumann algeb ra a ssociated wi th $S L( \math b b{Z},2 m+1 )$(the sp ecial l i nea rg r o up with integer entr ie s ) f or $m\ge q1$, t h en $ \ mathfrak {K }_{ 2}(\ m athca l{M} ) =0 $, $\del ta_{0} ( \ ma thcal{M }) \leq1$ .Thu s $ \math c al{M }$ isnot \*-i somor p hic to any $L( F (n))$ for $n\ g eq 2 $ .In [ @Ge Popa] L. Ge and S. P opad ef ine d a ty pe II $_ { 1} $ factor to be *weak ly * $n$* -thin *, if it cont ains hyper f i n ite suba lgeb r as $\mathcal{R}_{ 0},\m athcal{R}_ { 1}$ and$n$ v ectors $ \xi_{1},\ l d ots,\xi_ {n} $ i n $ L^{ 2 } \l eft( \mathca l { M},\ ta u\right ) $ suchtha t $ \ma thc al {M}=\over line {sp an }^ {\ le ft\ Vert{ \cdot } \r igh t\ Ver t _{2 } }\left ( \m athc al {R } _{0 }\{\xi_ { 1} , \ ldot s, \x i_{n }\} \m athca l{R} _ {1} \right) .$ Theysho w ed t ha t$L(F_{m })$ is not we ak ly $n$-thi nfor $m> 2 + 2 n$. Moti vated by these facts, w e have t hefollo wing definiti on. A ty peI I$_{1} $ fact or $\ ma thc a l {M}$w i th th etracial st a t e
}=c^{3}a$. The_group von_Neumann algebra associated with_$G$ is_a_type II$_{1}$_factor,_and the preceding_corollary easily implies_that $\mathfrak{K}_{2}(L(G))=0$ and $\delta _{0}(L(G))\leq1.$ The_next two corollaries_follows_directly from Corollary 3. Suppose $\mathcal M$ is a nonprime II$_1$ factor, i.e. $\mathcal M\simeq \mathcal_N_1\otimes_\mathcal N_2$_for_some_II$_1$ subfactors $ \mathcal N_1,_\mathcal N_2$. Then $\mathfrak{K}_{2}( \mathcal{M})=0$,_$\delta_{0}( \mathcal{M})\leq1$. Thus_$\mathcal{M}$ is not \*-isomorphic to any $L(F(n))$ for_$n\geq2$. If_$\mathcal{M}=L(SL(\mathbb{Z},2m+1))$ is the_group von Neumann algebra associated with $SL(\mathbb{Z},2m+1)$ (the special_linear group with integer entries) for_$m\geq1$, then $\mathfrak{K}_{2}(_\mathcal{M})=0$,_$\delta_{0}( \mathcal{M})\leq1$._Thus $\mathcal{M}$ is not_\*-isomorphic to any $L(F(n))$ for $n\geq2$. In_[@GePopa] L. Ge and S. Popa_defined a type II$_{1}$ factor to be_*weakly* $n$*-thin*, if it contains hyperfinite_subalgebras $\mathcal{R}_{0},\mathcal{R}_{1}$ and $n$ vectors_$\xi_{1},\ldots,\xi_{n}$ in_$L^{2}\left( \mathcal{M},\tau\right) $_such that $\mathcal{M}=\overline {span}^{\left\Vert_{ \cdot_}\right\Vert _{2}}\left( _\mathcal{R}_{0}\{\xi_{1},\ldots,\xi_{n}\}\mathcal{R}_{1}\right) .$ They showed that $L(F_{m})$_is not weakly_$n$-thin for $m> 2+2n$. Motivated by_these_facts, we have_the_following_definition. A type_II$_{1}$ factor $\mathcal{M}$_with_the tracial_state
SN offset {#sec:offset} @2013Sci...340..170W found evidence for two distinct populations of SNe Ia with respect to their radial distributions in the host galaxies. They found that the high-[$v_{\mathrm{Si}\,\textsc{ii}}$]{} SNe Ia are concentrated in the inner regions of their host galaxies, whereas the normal-[$v_{\mathrm{Si}\,\textsc{ii}}$]{} SNe Ia span a wide range of radial distance. We next examine this trend in our sample in Fig. \[offset-vsi\], showing [$v_{\mathrm{Si}\,\textsc{ii}}$]{} as a function of the normalised SN offset, [${\ensuremath{\mathrm{R}_{\mathrm{SN}}}}/{\ensuremath{\mathrm{R}_{\mathrm{gal}}}}$]{} (see Section \[sec:host\]). The high-[$v_{\mathrm{Si}\,\textsc{ii}}$]{} SNe Ia in the PTF sample also appear deficient in the outer regions of their hosts compared to the normal-[$v_{\mathrm{Si}\,\textsc{ii}}$]{}  SNe Ia, which are found at all radii. The only high-[$v_{\mathrm{Si}\,\textsc{ii}}$]{} SN at a large radius (PTF09djc; [$v_{\mathrm{Si}\,\textsc{ii}}$]{}$=13013\,\mathrm{km\,s^{-1}}$, [${\ensuremath{\mathrm{R}_{\mathrm{SN}}}}/{\ensuremath{\mathrm{R}_{\mathrm{gal}}}}$]{}$=2.2$) resides in the outskirts of an extended galaxy, with no potential host found at the SN position to the SDSS photometric limit ($r\simeq22\,\rm{mag}$). However, in our sample the trend is not statistically significant. We found that the locations of 83% of the high-[$v_{\mathrm{Si}\,\textsc{ii}}$]{} SNe Ia and 81% of the normal-[$v_{\mathrm{Si}\,\textsc{ii}}$]{} SNe Ia are within ${\ensuremath{{\ensuremath{\mathrm{R}_{\mathrm{SN}}}}/{\ensuremath{\mathrm{R}_{\mathrm{gal}}}}}}=1$, which implies the high-[$v_{\mathrm{Si}\,\textsc{ii}}$]{} SNe Ia are similar to the normal-[$v_{\mathrm{Si}\,\textsc{ii}}$]{} SNe Ia with respect to their radial distribution, although the normal-[$v_{\mathrm{Si}\,\textsc
SN offset { # sec: offset } @2013Sci... 340.. 170W found evidence for two distinct population of SNe Ia with obedience to their radial distributions in the host galaxies. They establish that the high-[$v_{\mathrm{Si}\,\textsc{ii}}$ ] { }   SNe Ia are concentrated in the inner region of their master of ceremonies galaxies, whereas the normal-[$v_{\mathrm{Si}\,\textsc{ii}}$ ] { }   SNe Ia span a wide range of radial distance. We next test this trend in our sample in Fig.   \[offset - vsi\ ], indicate [ $ v_{\mathrm{Si}\,\textsc{ii}}$ ] { }   as a function of the normalised SN offset, [ $ { \ensuremath{\mathrm{R}_{\mathrm{SN}}}}/{\ensuremath{\mathrm{R}_{\mathrm{gal}}}}$ ] { }   (see Section   \[sec: host\ ]). The high-[$v_{\mathrm{Si}\,\textsc{ii}}$ ] { }   SNe Ia in the PTF sample besides appear deficient in the outer area of their hosts compared to the normal-[$v_{\mathrm{Si}\,\textsc{ii}}$ ] { }   SNe Ia, which are find at all radius. The only high-[$v_{\mathrm{Si}\,\textsc{ii}}$ ] { }   SN at a large radius (PTF09djc; [ $ v_{\mathrm{Si}\,\textsc{ii}}$]{}$=13013\,\mathrm{km\,s^{-1}}$, [ $ { \ensuremath{\mathrm{R}_{\mathrm{SN}}}}/{\ensuremath{\mathrm{R}_{\mathrm{gal}}}}$]{}$=2.2 $) resides in the outskirts of an extensive galaxy, with no potential host found at the SN position to the SDSS photometric limit ($ r\simeq22\,\rm{mag}$). However, in our sample the trend is not statistically significant. We found that the locations of 83% of the high-[$v_{\mathrm{Si}\,\textsc{ii}}$ ] { }   SNe Ia and 81% of the normal-[$v_{\mathrm{Si}\,\textsc{ii}}$ ] { }   SNe Ia are within $ { \ensuremath{{\ensuremath{\mathrm{R}_{\mathrm{SN}}}}/{\ensuremath{\mathrm{R}_{\mathrm{gal}}}}}}=1 $, which implies the high-[$v_{\mathrm{Si}\,\textsc{ii}}$ ] { }   SNe Ia are similar to the normal-[$v_{\mathrm{Si}\,\textsc{ii}}$ ] { }   SNe Ia with respect to their radial distribution, although the normal-[$v_{\mathrm{Si}\,\textsc
SN offset {#sec:offset} @2013Sci...340..170W fomnd evidence for two divtinct populatkons of SNe Ia with respect vo tyeir eadial distributions iv the hosn galaxiew. Thty found that the high-[$v_{\mabkrm{Si}\,\fcxtsc{ni}}$]{} WNe Ia are congentrated it the inner reciund of their host galaxies, whereas thq normak-[$v_{\lathrm{Si}\,\textsc{yi}}$]{} SNt Iw spzn a wide range of radial distance. We nexu examine this tremd in our sample in Fig. \[offdet-vdi\], showing [$v_{\mathrm{Di}\,\textsc{ii}}$]{} aw a synction of tfe normalised SN offsej, [${\ensuremath{\mathrm{R}_{\mathrm{SN}}}}/{\ensurdmath{\kathrm{R}_{\matyrn{gap}}}}$]{} (vee Section \[wec:hoft\]). The high-[$v_{\mathrm{Si}\,\dextsc{io}}$]{} SNe Ia in the PTH sanple also appear defirient in the outer rggions of dhzir hosts compared to tye notmal-[$v_{\kathfn{Si}\,\gexusc{mi}}$]{}  ANe Ia, whmch are fouhd at all rqdii. The only high-[$v_{\kaeyrm{Si}\,\textsc{ii}}$]{} AN at w jarge radius (PTF09djc; [$v_{\mathrm{Si}\,\textsc{ii}}$]{}$=13013\,\mauhrm{kj\,s^{-1}}$, [${\ensuremath{\mathrm{R}_{\matyrm{SN}}}}/{\ensuremath{\mathrm{T}_{\mathrm{gal}}}}$]{}$=2.2$) resides in the outskirts of an extended galaxy, whth nk pouekbial yodt found at the SN position to the SDSS photoisttib limit ($r\simeq22\,\rm{mcg}$). However, in our swmlje the trend ks not stztistically signiflcant. Wg foune that tht locstions of 83% of the high-[$v_{\mathem{Si}\,\textsc{ii}}$]{} FBe Ia and 81% of the uormal-[$v_{\mathro{Si}\,\tgxtsc{io}}$]{} SNe Ia are within ${\ensuxemath{{\snsuremath{\mwthrm{R}_{\matgfm{SN}}}}/{\ensuremath{\maghrk{R}_{\kathrm{gal}}}}}}=1$, which implies thq high-[$v_{\mavhrm{Sn}\,\textsc{ik}}$]{} SNe Ia arq similar ho thc normal-[$v_{\mathrm{Si}\,\tedtsc{in}}$]{} SNe Ha with redpect to their radial distributmpn, although jhe nogmal-[$v_{\mathxm{Si}\,\tewtsc
SN offset {#sec:offset} @2013Sci...340..170W found evidence for populations SNe Ia respect to their galaxies. found that the SNe Ia are in the inner regions of their galaxies, whereas the normal-[$v_{\mathrm{Si}\,\textsc{ii}}$]{} SNe Ia span a wide range of radial distance. next examine this trend in our sample in Fig. \[offset-vsi\], showing [$v_{\mathrm{Si}\,\textsc{ii}}$]{} as function the SN [${\ensuremath{\mathrm{R}_{\mathrm{SN}}}}/{\ensuremath{\mathrm{R}_{\mathrm{gal}}}}$]{} (see Section \[sec:host\]). The high-[$v_{\mathrm{Si}\,\textsc{ii}}$]{} SNe Ia in the PTF sample also appear deficient in outer regions of their hosts compared to the SNe Ia, which are at all radii. The only SN a large (PTF09djc; [${\ensuremath{\mathrm{R}_{\mathrm{SN}}}}/{\ensuremath{\mathrm{R}_{\mathrm{gal}}}}$]{}$=2.2$) in the outskirts an extended galaxy, with no potential host found at the SN position to the SDSS photometric limit However, in the trend not significant. found that the 83% of the high-[$v_{\mathrm{Si}\,\textsc{ii}}$]{} SNe Ia the normal-[$v_{\mathrm{Si}\,\textsc{ii}}$]{} SNe Ia are within ${\ensuremath{{\ensuremath{\mathrm{R}_{\mathrm{SN}}}}/{\ensuremath{\mathrm{R}_{\mathrm{gal}}}}}}=1$, which the high-[$v_{\mathrm{Si}\,\textsc{ii}}$]{} Ia are similar to the normal-[$v_{\mathrm{Si}\,\textsc{ii}}$]{} Ia with respect to their radial distribution, although normal-[$v_{\mathrm{Si}\,\textsc
SN offset {#sec:offset} @2013Sci...340..170W founD evidence fOr two DisTinCt PopuLatiOns of SNe Ia with REspeCt to their radial distribUtionS iN The hOSt GalaxIes. They FOuND ThaT tHe HigH-[$v_{\MAtHrm{Si}\,\TexTsc{ii}}$]{} SNE Ia are concEntRaTed in the inneR ReGions of theIr hOst galaxies, wHerEas the NoRmaL-[$V_{\mathRm{SI}\,\textSc{ii}}$]{} SNE ia span A wide rangE oF Radial DIstance. wE NeXt exAmine this trend in oUR sAMple in Fig. \[offseT-vsi\], shOwINg [$V_{\MAthRm{SI}\,\textsc{ii}}$]{} aS a FunctIOn of the NOrMALIseD sN offset, [${\ensurEmath{\mathrm{r}_{\MatHrm{SN}}}}/{\eNsUreMAth{\matHrm{R}_{\mAtHRm{gAl}}}}$]{} (see SectioN \[sec:Host\]). The hiGh-[$v_{\matHRm{Si}\,\texTSc{ii}}$]{} SNe ia in thE PTf saMple ALsO aPpeAr DEfiCIeNt iN The Outer regIoNs Of theIr hoSTS COmpaRed To thE normAl-[$v_{\mathrm{Si}\,\teXtsC{ii}}$]{}  SnE Ia, Which Are foUnd aT aLl radIi. The oNly hiGh-[$V_{\mathrm{Si}\,\textsc{Ii}}$]{} SN At a large rAdiUs (pTF09DjC; [$v_{\matHRm{Si}\,\teXtsC{ii}}$]{}$=13013\,\Mathrm{kM\,s^{-1}}$, [${\ensurEMatH{\mATHRm{r}_{\mathrm{SN}}}}/{\ensurematH{\mATHrM{R}_{\mathrm{Gal}}}}$]{}$=2.2$) resIDeS iN The outskIrTs oF an eXTEnded GalaXY, wIth no potEntial HOsT fOund at tHe sN posiTiOn tO thE SDSS PHotoMetric Limit ($r\siMeq22\,\rm{MAg}$). However, in our SAmple the trend IS nOT StATistIcaLly significAnt. WE FounD thaT ThE loCAtionS of 83% of ThE HiGH-[$v_{\mathrm{Si}\,\textsc{ii}}$]{} Sne ia and 81% oF the nOrmal-[$v_{\mathrm{SI}\,\textsc{ii}}$]{} SnE iA are withIn ${\enSUrEMath{{\ensuremath{\MathrM{R}_{\mathrm{SN}}}}/{\ENsurematH{\mathRm{R}_{\mathrM{gal}}}}}}=1$, which IMPlies the HigH-[$v_{\mAthRm{SI}\,\TExTsc{ii}}$]{} SNe Ia are SIMilaR tO the norMal-[$V_{\mathrm{si}\,\tExtSc{iI}}$]{} SNE IA with respEct to theIr RaDiAl DisTribuTIon, althoUgH thE nOrmAl-[$v_{\maTHrm{Si}\,\tExtsc
SN offset {#sec:offset} @2013Sci.. .340. .17 0Wfo undevid ence for two d i stin ct populations of SNeIa wi th resp e ct to t heir ra d ia l dis tr ib uti on s i n the ho st gala xies. They fo un d that the h i gh -[$v_{\mat hrm {Si}\,\texts c{i i}}$]{ }SNe Ia ar e c oncen trated in the inner re gi o ns oft heir ho s t g alax ies, whereas then or m al-[$v_{\mathr m{Si}\ ,\ t ex t s c{i i}} $]{} SNe I aspana wide r a ng e o f r a dial distance . We next e x ami ne thi stre n d in o ur sa mp l e i n Fig. \[of fset -vsi\], s howing [$v_{\m a thrm{Si }\,\te xts c{i i}}$ ] {}  a s a f u nct i on of the normali se dSN of fset , [ $ {\en sur emat h{\ma thrm{R}_{\mat hrm {SN} } }}/ {\ens urema th{\ ma thrm{ R}_{\m athrm {g al}}}}$]{} (see Sec tion \[se c:h os t\] ). Theh igh-[$ v_{ \ma thrm{Si }\,\tex t sc{ ii } } $ ]{ } SNe Ia in the PT Fs a mp le alsoappear de fi c ient inth e o uter r egion s of th eir host s comp a re dto theno rmal-[ $v _{\ mat hrm{S i }\,\ textsc {ii}}$]{ }  SN e Ia, which are found at allr ad i i .T he o nly high-[$v_{ \mat h rm{S i}\, \ te xts c {ii}} $]{}SN at a large radius (PTF 09 djc; [ $v_{\ mathrm{Si}\,\ textsc{ii} } $ ] {}$=1301 3\,\ m at h rm{km\,s^{-1}} $, [$ {\ensurema t h{\mathr m{R}_ {\mathrm {SN}}}}/{ \ e nsuremat h{\ mat hrm {R} _ { \m athrm{gal}}}} $ ] {}$= 2. 2$) res ide s in th e o uts kir tsof an exten ded gala xy ,wi th no pote n tial hos tfou nd at theS N posi tionto t he S D SSphotome t ri c limi t($ r\si meq 22 \,\rm {mag } $). Howev er, in ou r s a mple t he trendis not statis ti cally sign if ica nt. We f ound tha t the locations of 83%o f the h igh -[$v_ {\ma thrm{Si}\ ,\t extsc{ ii} } $]{} S Ne Iaand 8 1% of t he no r m al -[$ v_ {\mathrm{S i } \,\ texts c{ ii}} $]{} SN e Ia are within ${ \ ens uremath{{\ens ure math { \ ma thr m {R } _{\ ma t hrm { S N}}}}/{\ensurem ath{\mathr m{ R }_ {\mathrm{g a l}} }} }}=1$,which i mplie s the hi gh-[$v_{\ mathrm{Si }\ ,\te x t sc{ ii}}$]{} S Ne Ia ar e similar to th e n ormal -[$ v_{\ma th rm{ Si}\, \texts c {ii }}$]{ } SNeIa withrespe ct to thei r radial distribution,althou gh th e n ormal-[$v _{\ m ath rm{Si}\,\ text sc
SN_offset {#sec:offset} @2013Sci...340..170W_found evidence for two_distinct populations_of_SNe Ia_with_respect to their_radial distributions in_the host galaxies. They_found that the_high-[$v_{\mathrm{Si}\,\textsc{ii}}$]{} SNe_Ia are concentrated in the inner regions of their host galaxies, whereas the normal-[$v_{\mathrm{Si}\,\textsc{ii}}$]{} SNe_Ia_span a_wide_range_of radial distance. We next_examine this trend in our_sample in_Fig. \[offset-vsi\], showing [$v_{\mathrm{Si}\,\textsc{ii}}$]{} as a function of the normalised_SN_offset, [${\ensuremath{\mathrm{R}_{\mathrm{SN}}}}/{\ensuremath{\mathrm{R}_{\mathrm{gal}}}}$]{} (see Section \[sec:host\])._The high-[$v_{\mathrm{Si}\,\textsc{ii}}$]{} SNe Ia in the PTF sample also appear_deficient in the outer regions of_their hosts compared_to_the_normal-[$v_{\mathrm{Si}\,\textsc{ii}}$]{}  SNe Ia, which_are found at all radii. The_only high-[$v_{\mathrm{Si}\,\textsc{ii}}$]{} SN at a large radius_(PTF09djc; [$v_{\mathrm{Si}\,\textsc{ii}}$]{}$=13013\,\mathrm{km\,s^{-1}}$, [${\ensuremath{\mathrm{R}_{\mathrm{SN}}}}/{\ensuremath{\mathrm{R}_{\mathrm{gal}}}}$]{}$=2.2$) resides in the outskirts_of an extended galaxy, with no_potential host found at the_SN position_to the SDSS photometric limit_($r\simeq22\,\rm{mag}$). However, in our_sample the_trend is not_statistically significant. We found that the_locations of 83%_of the high-[$v_{\mathrm{Si}\,\textsc{ii}}$]{} SNe Ia and 81%_of_the normal-[$v_{\mathrm{Si}\,\textsc{ii}}$]{} SNe Ia_are_within_${\ensuremath{{\ensuremath{\mathrm{R}_{\mathrm{SN}}}}/{\ensuremath{\mathrm{R}_{\mathrm{gal}}}}}}=1$, which_implies the high-[$v_{\mathrm{Si}\,\textsc{ii}}$]{} SNe_Ia_are similar_to_the normal-[$v_{\mathrm{Si}\,\textsc{ii}}$]{} SNe Ia with respect to_their_radial distribution, although the normal-[$v_{\mathrm{Si}\,\textsc
\sum_{x_1, \ldots, x_{r-1} \in V({\mathcal{H}})} \deg(x_1, x_2, \ldots, x_{r-1}) = \binom{r+1}{r-1} |\mathcal{H}'| = \binom{r+1}{2} |\mathcal{H}'|.$$ On the other hand, again by double counting, $$\begin{aligned} \sum_{q \in E(\mathcal{H}')} &\sum_{x_1, \ldots, x_{r-1} \in q} \deg(x_1, \ldots, x_{r-1}) \notag\\ & = \sum_{x_1, \ldots, x_{r-1} \in V({\mathcal{H}})} \deg(x_1, \ldots, x_{r-1})^2 \notag\\ & \geq \frac{1}{\binom{n}{r-1}} \left(\sum_{x_1, \ldots, x_{r-1}} \deg(x_1, \ldots, x_{r-1}) \right)^2 &&\text{(by Jensen's inequality)} \notag\\ & = \frac{1}{\binom{n}{r-1}}\left( \binom{r+1}{2} |\mathcal{H}'| \right)^2. &&\text{(by eq. \eqref{eq:deg-size})} \label{eq:conv-bd}\end{aligned}$$ Fix any $q \in E(\mathcal{H}')$. By double counting, since each hyperedge in $\mathcal{H}'$ has $r+1 = (r-1) + 2$ vertices, $$\begin{aligned} \sum_{\{x_1, \ldots, x_{r-1} \} \subseteq q} \deg(x_1, \ldots, x_{r-1}) &=\sum_{\underset{|A| = r-1}{A \subseteq q}} |\{q' \mid A \subseteq q'\}| \notag\\ & = \sum_{\underset{|A| = r-1}{A \subseteq q}} \left(1 + |\{q' \neq q \mid A \subseteq q'\}| \right) \notag\\ &
\sum_{x_1, \ldots, x_{r-1 } \in V({\mathcal{H } }) } \deg(x_1, x_2, \ldots, x_{r-1 }) = \binom{r+1}{r-1 } |\mathcal{H}'| = \binom{r+1}{2 } |\mathcal{H}'|.$$ On the other hand, again by double counting, $ $ \begin{aligned } \sum_{q \in E(\mathcal{H }') } & \sum_{x_1, \ldots, x_{r-1 } \in q } \deg(x_1, \ldots, x_{r-1 }) \notag\\ & = \sum_{x_1, \ldots, x_{r-1 } \in V({\mathcal{H } }) } \deg(x_1, \ldots, x_{r-1})^2 \notag\\ & \geq \frac{1}{\binom{n}{r-1 } } \left(\sum_{x_1, \ldots, x_{r-1 } } \deg(x_1, \ldots, x_{r-1 }) \right)^2 & & \text{(by Jensen's inequality) } \notag\\ & = \frac{1}{\binom{n}{r-1}}\left (\binom{r+1}{2 } |\mathcal{H}'| \right)^2. & & \text{(by eq. \eqref{eq: deg - size }) } \label{eq: conv - bd}\end{aligned}$$ specify any $ q \in E(\mathcal{H}')$. By bivalent counting, since each hyperedge in $ \mathcal{H}'$ has $ r+1 = (r-1) + 2 $ vertex, $ $ \begin{aligned } \sum_{\{x_1, \ldots, x_{r-1 } \ } \subseteq q } \deg(x_1, \ldots, x_{r-1 }) & = \sum_{\underset{|A| = r-1}{A \subseteq q } } |\{q' \mid vitamin a \subseteq q'\}| \notag\\ & = \sum_{\underset{|A| = r-1}{A \subseteq q } } \left(1 + |\{q' \neq q \mid A \subseteq q'\}| \right) \notag\\ &
\sum_{d_1, \ldots, x_{r-1} \in V({\mathcal{H}})} \aeg(x_1, x_2, \ldots, x_{r-1}) = \binom{c+1}{r-1} |\mathdal{H}'| = \bivom{r+1}{2} |\mathcal{H}'|.$$ On the other haid, atain vy double counting, $$\begkn{aligned} \dum_{q \in W(\matical{H}')} &\sum_{x_1, \ldots, x_{r-1} \in q} \deg(x_1, \mfots, e_{r-1}) \notag\\ & = \som_{x_1, \ldots, x_{r-1} \in V({\mathcal{H}})} \gee(x_1, \ldots, x_{r-1})^2 \notag\\ & \geq \frac{1}{\binom{n}{r-1}} \left(\suk_{x_1, \ldots, x_{r-1}} \deg(x_1, \ldous, v_{r-1}) \rjght)^2 &&\text{(by Jensen's inequality)} \notag\\ & = \frac{1}{\binok{n}{r-1}}\left( \binom{r+1}{2} |\mathcal{H}'| \rihht)^2. &&\hext{(by eq. \eqref{eq:dfg-size})} \labeo{eq:cjbv-bd}\end{aligndd}$$ Fix any $q \in E(\mathcam{H}')$. By double counting, since eacf hypzredge in $\mqtycap{V}'$ has $r+1 = (r-1) + 2$ vernices, $$\begin{allbned} \suk_{\{x_1, \ldotx, x_{r-1} \} \subseteq q} \veg(x_1, \ldots, x_{r-1}) &=\sum_{\undervet{|A| = r-1}{A \subseteq z}} |\{q' \mid A \sbbseteq q'\}| \notag\\ & = \wun_{\undetset{|A| = r-1}{A \sucseueq q}} \left(1 + |\{q' \ieq q \mid A \subseteq q'\}| \right) \notag\\ &
\sum_{x_1, \ldots, x_{r-1} \in V({\mathcal{H}})} \deg(x_1, x_2, = |\mathcal{H}'| = |\mathcal{H}'|.$$ On the counting, \sum_{q \in E(\mathcal{H}')} \ldots, x_{r-1} \in \deg(x_1, \ldots, x_{r-1}) \notag\\ & = \ldots, x_{r-1} \in V({\mathcal{H}})} \deg(x_1, \ldots, x_{r-1})^2 \notag\\ & \geq \frac{1}{\binom{n}{r-1}} \left(\sum_{x_1, \ldots, \deg(x_1, \ldots, x_{r-1}) \right)^2 &&\text{(by Jensen's inequality)} \notag\\ & = \frac{1}{\binom{n}{r-1}}\left( \binom{r+1}{2} |\mathcal{H}'| &&\text{(by \eqref{eq:deg-size})} Fix $q \in E(\mathcal{H}')$. By double counting, since each hyperedge in $\mathcal{H}'$ has $r+1 = (r-1) + vertices, $$\begin{aligned} \sum_{\{x_1, \ldots, x_{r-1} \} \subseteq q} \ldots, x_{r-1}) &=\sum_{\underset{|A| = \subseteq q}} |\{q' \mid A q'\}| & = = \subseteq \left(1 + |\{q' q \mid A \subseteq q'\}| \right) \notag\\ &
\sum_{x_1, \ldots, x_{r-1} \in V({\mathcal{H}})} \deg(X_1, x_2, \ldots, x_{r-1}) = \bInom{r+1}{R-1} |\maThcAl{h}'| = \binOm{r+1}{2} |\mAthcal{H}'|.$$ On the otHEr haNd, again by double countinG, $$\begiN{aLIgneD} \SuM_{q \in E(\Mathcal{h}')} &\SuM_{X_1, \LdoTs, X_{r-1} \In q} \DeG(X_1, \lDots, x_{R-1}) \noTag\\ & = \sum_{x_1, \Ldots, x_{r-1} \in V({\MatHcAl{H}})} \deg(x_1, \ldots, X_{R-1})^2 \nOtag\\ & \geq \fraC{1}{\biNom{n}{r-1}} \left(\sum_{X_1, \ldOts, x_{r-1}} \dEg(X_1, \ldOTs, x_{r-1}) \rIghT)^2 &&\text{(By JensEN's ineqUality)} \notAg\\ & = \FRac{1}{\binOM{n}{r-1}}\left( \BINoM{r+1}{2} |\maThcal{H}'| \right)^2. &&\text{(by EQ. \eQRef{eq:deg-size})} \laBel{eq:cOnV-Bd}\END{alIgnEd}$$ Fix any $q \iN E(\MathcAL{H}')$. By douBLe COUNtiNG, since each hypEredge in $\matHCal{h}'$ has $r+1 = (r-1) + 2$ VeRtiCEs, $$\begiN{aligNeD} \Sum_{\{X_1, \ldots, x_{r-1} \} \subSeteQ q} \deg(x_1, \ldoTs, x_{r-1}) &=\suM_{\UnderseT{|a| = r-1}{A \subsEteq q}} |\{q' \Mid a \suBsetEQ q'\}| \NoTag\\ & = \SuM_{\UndERsEt{|A| = R-1}{a \suBseteq q}} \lEfT(1 + |\{q' \Neq q \mId A \sUBSETeq q'\}| \RigHt) \noTag\\ &
\sum_{x_1, \ldots, x_{r-1} \in V({\m athca l{H }}) }\deg (x_1 , x_2, \ldots, x_{r -1}) = \binom{r+1}{r-1 } |\m at h cal{ H }' | = \ binom{r + 1} { 2 } | \m at hca l{ H }' |.$$ On the ot her hand,aga in by double c o un ting, $$\b egi n{aligned} \ sum _{q \i nE(\ m athca l{H }')}&\sum_ { x_1, \ ldots, x_ {r - 1} \in q} \deg ( x _1 , \l dots, x_{r-1}) \n o ta g \\ & = \su m_{x_1 ,\ ld o t s,x_{ r-1} \in V ({ \math c al{H}}) } \ d e g (x_ 1 , \ldots, x_{ r-1})^2 \no t ag\ \ &\ge q \frac {1}{\ bi n om{ n}{r-1}} \l eft( \sum_{x_1 , \ldo t s, x_{r - 1}} \de g(x_1, \l dot s, x _ {r -1 })\r i ght ) ^2 & &\text{( by J ensen 's i n e q u alit y)} \no tag\\ & = \fra c{1 }{\b i nom {n}{r -1}}\ left (\bino m{r+1} {2} | \m athcal{H}'| \ri ght) ^2. &&\te xt{ (b y e q. \eqr e f{eq:d eg- siz e})} \l abel{eq : con v- b d } \e nd{aligned}$$ Fix a n y $ q \in E( \mathc a l{ H} ' )$. By d ou ble cou n t ing,sinc e e ach hype redgei n$\ mathcal {H }'$ ha s$r+ 1 = (r-1 ) + 2 $ vert ices, $$ \begi n {aligned} \sum _ {\{x_1, \ldot s ,x _ {r - 1} \ } \ subseteq q} \de g (x_1 , \l d ot s,x _{r-1 }) & =\ s um_{\underset{|A| = r -1}{A\subs eteq q}} |\{q ' \mid A \ s u b seteq q' \}|\ no t ag\\ & = \ sum_{ \underset{ | A| = r-1 }{A \subsete q q}} \le f t (1 + |\{ q'\ne q q \m i d A \subseteq q' \ } | \r ig ht) \no tag \\ &
\sum_{x_1, \ldots,_x_{r-1} \in_V({\mathcal{H}})} \deg(x_1, x_2, \ldots,_x_{r-1}) =_\binom{r+1}{r-1}_|\mathcal{H}'| =_\binom{r+1}{2}_|\mathcal{H}'|.$$ On the other_hand, again by_double counting, $$\begin{aligned} \sum_{q \in_E(\mathcal{H}')} &\sum_{x_1, \ldots,_x_{r-1}_\in q} \deg(x_1, \ldots, x_{r-1}) \notag\\ & = \sum_{x_1, \ldots, x_{r-1}_\in_V({\mathcal{H}})} \deg(x_1,_\ldots,_x_{r-1})^2_\notag\\ &_\geq \frac{1}{\binom{n}{r-1}} \left(\sum_{x_1, \ldots, x_{r-1}}_\deg(x_1, \ldots,_x_{r-1}) \right)^2 &&\text{(by Jensen's_inequality)}_\notag\\ _ & = \frac{1}{\binom{n}{r-1}}\left( \binom{r+1}{2} |\mathcal{H}'| \right)^2. &&\text{(by eq._\eqref{eq:deg-size})} \label{eq:conv-bd}\end{aligned}$$ Fix any $q \in E(\mathcal{H}')$._By double counting,_since_each_hyperedge in $\mathcal{H}'$ has_$r+1 = (r-1) + 2$ vertices,_$$\begin{aligned} \sum_{\{x_1, \ldots, x_{r-1} \} \subseteq q}_\deg(x_1, \ldots, x_{r-1}) &=\sum_{\underset{|A|_= r-1}{A \subseteq q}} |\{q' \mid_A \subseteq q'\}| \notag\\ _ _& = \sum_{\underset{|A| = r-1}{A_ \subseteq q}}_\left(1 +_|\{q' \neq q_\mid A \subseteq q'\}| \right) \notag\\ _ _&
looks quite similar besides that the factor $\exp (-Work/T)$ should be replaced with $\exp (-\Delta S)$ where $\Delta S\{t,Trajectory;Forces\}$ has the sense of increment of entropy (or similar thermodynamic potential) during $Trajectory$. FLUCTUATIONS OF DISSIPATION --------------------------- Let $\left\langle J(t_1),...,J(t_n)\right\rangle _q$ denotes the $n$-th order nonequilibrium cumulant corresponding to the time-cut modification of perturbing forces, $x(t)\Rightarrow x(t)\eta (t-\min (t_1,...,t_n))$, where $\eta (t)=1$ at $t>0$ and $\eta (t)=0$ at $t\leq 0$ (thus the most early variable belongs to yet equilibrium state). Such correlators were termed in \[34\] quazi-equilibrium. Then the rigorous relation takes place \[33,34,11\], $$\left\langle J_j(t)\right\rangle =\frac 1T\int_{-\infty }^t\left\langle J_j(t),J_k(t^{\prime })\right\rangle _qx_k(t^{\prime })dt^{\prime }$$ what extends Kubo formulas to arbitrary non-linear responce. As a consequence, the average value of the work (i.e. of the energy absorbed by the system), $\left\langle Work\right\rangle =\int_{-\infty }^t\left\langle J(t)\right\rangle x(t)dt$, can be simply expressed via two-point quazi-equilibrium correlators. The similar formulas for higher cumulants can be obtained \[34\] which lead to the formally exact expression for variance of the work \[11\] : $$\left\langle Work^{(2)}\right\rangle =2T\left\langle Work\right\rangle +\frac 2T\int_{-\infty }^tx(1)x(2)\left\langle J(1),J(2),J(3)\right\rangle _qx(3)d1d2d3$$ where $\left\langle Work^{(2)}\right\rangle =\left\langle Work^2\right\rangle -\left\langle Work\right\rangle ^2$, and for bre
looks quite similar besides that the factor $ \exp (-Work / T)$ should be replaced with $ \exp (-\Delta S)$ where $ \Delta S\{t, Trajectory;Forces\}$ take the common sense of increment of entropy (or similar thermodynamic electric potential) during $ Trajectory$. FLUCTUATIONS OF waste --------------------------- Let $ \left\langle J(t_1),... ,J(t_n)\right\rangle _ q$ denotes the $ n$-th holy order nonequilibrium cumulant corresponding to the time - dilute modification of perturbing force, $ x(t)\Rightarrow x(t)\eta (t-\min (t_1,... ,t_n))$, where $ \eta (t)=1 $ at $ t>0 $ and $ \eta (t)=0 $ at $ t\leq 0 $ (thus the most early variable belongs to yet equilibrium state of matter). Such correlators were termed in \[34\ ] quazi - equilibrium. Then the rigorous relation take place \[33,34,11\ ], $ $ \left\langle J_j(t)\right\rangle = \frac 1T\int_{-\infty } ^t\left\langle J_j(t),J_k(t^{\prime }) \right\rangle _ qx_k(t^{\prime }) dt^{\prime } $ $ what extends Kubo formulas to arbitrary non - linear responce. As a consequence, the modal value of the work (i.e. of the energy absorbed by the system), $ \left\langle Work\right\rangle = \int_{-\infty } ^t\left\langle J(t)\right\rangle x(t)dt$, can be simply express via two - point quazi - equilibrium correlators. The similar formulas for higher cumulants can be obtained \[34\ ] which lead to the formally exact expression for variance of the work \[11\ ]: $ $ \left\langle Work^{(2)}\right\rangle = 2T\left\langle Work\right\rangle + \frac 2T\int_{-\infty } ^tx(1)x(2)\left\langle J(1),J(2),J(3)\right\rangle _ qx(3)d1d2d3$$ where $ \left\langle Work^{(2)}\right\rangle = \left\langle Work^2\right\rangle -\left\langle Work\right\rangle ^2 $, and for bre
lolks quite similar besider that the factor $\exp (-Wmrk/T)$ sgould be replaced with $\exp (-\Delta S)$ wiere $\Deltq S\{t,Trajectory;Forces\}$ har the sende of inxremtnt of entropy (or similar thermkfynakmc potential) duting $Trajectmry$. FLUCTUATIONV UF DISSIPATION --------------------------- Let $\left\langle J(t_1),...,J(t_n)\rigrt\ranglr _e$ denotes the $g$-th pwder nonequilibrium cumulant corresponsing to the time-cut kodification of perturbing forfes, $x(t)\Rightarrow x(h)\eta (t-\min (t_1,...,j_h))$, wrwre $\eta (t)=1$ at $t>0$ and $\eta (t)=0$ at $t\leq 0$ (jhus the most early variable belungs co yet equioivrikk state). Suci corrvlators were bvrmed it \[34\] quazo-equilibrium. Tmen tie rugorous relation takev place \[33,34,11\], $$\left\langlg J_j(t)\right\saugle =\frac 1T\int_{-\infty }^t\oedt\lancle J_b(t),J_k(g^{\prioe })\dijht\dangle _qx_i(t^{\prime })dt^{\lrime }$$ what extends Kubo formukaf to arbitrary non-ligewr responce. As a consequence, the averagt valhe of the work (i.e. of thw energy absorbed by jhe system), $\left\langle Work\right\rangle =\int_{-\infty }^t\left\langle J(d)\righv\rxngoe b(r)dh$, can be simply expressed via two-point quazi-ezhikinrium correlatorf. The similat votiulas for higfer cumulznts can be obtainfd \[34\] whish leqd to the forkally exact expression for cariance of nhe qork \[11\] : $$\left\langle Cork^{(2)}\right\ranyle =2T\lgft\lanble Work\right\rangle +\frac 2T\inf_{-\infty }^tx(1)x(2)\lfft\langle G(1),J(2),J(3)\right\rangle _qx(3)a1d2d3$$ wvere $\lefu\uangle Work^{(2)}\right\rwngle =\lefv\langke Work^2\rkght\tangle -\jeft\langle Work\vhght\rangle ^2$, and fog bre
looks quite similar besides that the factor should replaced with (-\Delta S)$ where of of entropy (or thermodynamic potential) during FLUCTUATIONS OF DISSIPATION --------------------------- Let $\left\langle _q$ denotes the $n$-th order nonequilibrium cumulant corresponding to the time-cut modification of forces, $x(t)\Rightarrow x(t)\eta (t-\min (t_1,...,t_n))$, where $\eta (t)=1$ at $t>0$ and $\eta (t)=0$ $t\leq (thus most variable belongs to yet equilibrium state). Such correlators were termed in \[34\] quazi-equilibrium. Then the rigorous takes place \[33,34,11\], $$\left\langle J_j(t)\right\rangle =\frac 1T\int_{-\infty }^t\left\langle })\right\rangle _qx_k(t^{\prime })dt^{\prime }$$ extends Kubo formulas to arbitrary responce. a consequence, average of work (i.e. of energy absorbed by the system), $\left\langle Work\right\rangle =\int_{-\infty }^t\left\langle J(t)\right\rangle x(t)dt$, can be simply expressed via two-point correlators. The for higher can obtained which lead to exact expression for variance of the $$\left\langle Work^{(2)}\right\rangle =2T\left\langle Work\right\rangle +\frac 2T\int_{-\infty }^tx(1)x(2)\left\langle J(1),J(2),J(3)\right\rangle where $\left\langle =\left\langle Work^2\right\rangle -\left\langle Work\right\rangle ^2$, and bre
looks quite similar besides tHat the factOr $\exp (-worK/T)$ sHoUld bE repLaced with $\exp (-\DeLTa S)$ wHere $\Delta S\{t,Trajectory;FOrces\}$ HaS The sENsE of inCrement OF eNTRopY (oR sImiLaR ThErmodYnaMic poteNtial) durinG $TrAjEctory$. FLUCTUatIoNS OF DISSIpATiON --------------------------- Let $\left\laNglE J(t_1),...,J(t_n)\RiGht\RAngle _Q$ deNotes The $n$-th ORder noNequilibrIuM CumulaNT corresPONdIng tO the time-cut modifiCAtIOn of perturbing Forces, $X(t)\rIgHTArrOw x(T)\eta (t-\min (t_1,...,t_N))$, wHere $\eTA (t)=1$ at $t>0$ anD $\EtA (T)=0$ AT $t\lEQ 0$ (thus the most eArly variablE BelOngs to YeT eqUIlibriUm staTe). sUch Correlators Were Termed in \[34\] qUazi-eqUIlibriuM. then the RigoroUs rElaTion TAkEs PlaCe \[33,34,11\], $$\LEft\LAnGle j_J(t)\rIght\rangLe =\FrAc 1T\inT_{-\infTY }^T\LEft\lAngLe J_j(T),J_k(t^{\pRime })\right\rangLe _qX_k(t^{\pRIme })Dt^{\priMe }$$ whaT extEnDs KubO formuLas to ArBitrary non-lineaR resPonce. As a cOnsEqUenCe, The avERage vaLue Of tHe work (i.E. of the eNErgY aBSORbEd by the system), $\left\lAnGLE WOrk\right\Rangle =\INt_{-\InFTy }^t\left\lAnGle j(t)\riGHT\rangLe x(t)DT$, cAn be simpLy exprESsEd Via two-pOiNt quazI-eQuiLibRium cORrelAtors. THe similaR formULas for higher cuMUlants can be obTAiNED \[34\] wHIch lEad To the formalLy exACt exPresSIoN foR VariaNce of ThE WoRK \[11\] : $$\left\langle Work^{(2)}\righT\rAngle =2T\Left\lAngle Work\righT\rangle +\fraC 2t\INt_{-\infty }^tX(1)x(2)\leFT\lANgle J(1),J(2),J(3)\right\raNgle _qX(3)d1d2d3$$ where $\lEFt\langle work^{(2)}\rIght\rangLe =\left\lanGLE Work^2\rigHt\rAngLe -\lEft\LANgLe Work\right\raNGLe ^2$, anD fOr bre
looks quite similar besid es that th e fac tor $\ ex p (- Work /T)$ should be repl aced with $\exp (-\Del ta S) $w here $\ Delta S\{t,T r aj e c tor y; Fo rce s\ } $has t hesense o f incremen t o fentropy (ors im ilar therm ody namic potent ial ) duri ng $T r aject ory $. F LUCTUA T IONS O F DISSIPA TI O N ---- - ------- - - -- ---- ------- Let $\le f t\ l angle J(t_1),. ..,J(t _n ) \r i g ht\ ran gle _q$ de no tes t h e $n$-t h o r d e r n o nequilibriumcumulant co r res pondin gtot he tim e-cut m o dif ication ofpert urbing fo rces,$ x(t)\Ri g htarrow x(t)\ eta (t -\mi n ( t_ 1,. .. , t_n ) )$ , w h ere $\eta ( t) =1 $ at$t>0 $ a n d $\ eta (t) =0$ a t $t\leq 0$ ( thu s th e mo st ea rly v aria bl e bel ongs t o yet e quilibrium stat e).Such corr ela to rswe re te r med in \[ 34\ ] quazi -equili b riu m. T h en the rigorous rela ti o n t akes pla ce \[3 3 ,3 4, 1 1\], $$\ le ft\ lang l e J_j( t)\r i gh t\rangle =\fra c 1 T\ int_{-\ in fty }^ t\ lef t\l angle J_j( t),J_k (t^{\pri me }) \ right\rangle _ q x_k(t^{\prime }) d t ^{ \ prim e } $$ what ext ends Kubo for m ul ast o arb itrar yn on - linear responce. As a conse quenc e, the averag e value of t h e work ( i.e. of the energy abs orbed by the sy s tem), $\ left\ langle W ork\right \ r angle =\ int _{- \in fty } ^t \left\langleJ ( t)\r ig ht\rang lex(t)dt$ , c anbesim pl y express ed via t wo -p oi nt qu azi-e q uilibriu mcor re lat ors.The si milar for mu la s fo r highe r c u m ulan ts c an b e o bt ained \[3 4 \]which l ead to th e f o rmal ly e xact ex pression forva riance ofth e w ork \[ 1 1 \] : $$\ left\langle Work^{(2)}\ r ight\ra ngl e =2T \lef t\langleWor k\righ t\r a ngle + \frac2T\in t_ {-\ i n fty } ^ t x( 1)x (2 )\left\lan g l e J (1),J (2 ),J( 3)\righ t\rangle _qx(3)d1d 2 d3$ $ where $\lef t\l angl e Wo rk^ { (2 ) }\r ig h t\r a n gle =\left\lang le Work^2\ ri g ht \rangle -\ l eft \l angle W ork\rig ht\ra n gle ^2$ , and for bre
looks_quite similar_besides that the factor_$\exp (-Work/T)$ should_be_replaced with_$\exp_(-\Delta S)$ where_$\Delta S\{t,Trajectory;Forces\}$ has the_sense of increment of_entropy (or similar_thermodynamic_potential) during $Trajectory$. FLUCTUATIONS OF DISSIPATION --------------------------- Let $\left\langle J(t_1),...,J(t_n)\right\rangle _q$ denotes the $n$-th order nonequilibrium cumulant_corresponding_to the_time-cut_modification_of perturbing forces, $x(t)\Rightarrow x(t)\eta_(t-\min (t_1,...,t_n))$, where $\eta (t)=1$_at $t>0$_and $\eta (t)=0$ at $t\leq 0$ (thus the_most_early variable belongs_to yet equilibrium state). Such correlators were termed in_\[34\] quazi-equilibrium. Then the rigorous relation_takes place \[33,34,11\],_$$\left\langle_J_j(t)\right\rangle_=\frac 1T\int_{-\infty }^t\left\langle J_j(t),J_k(t^{\prime })\right\rangle__qx_k(t^{\prime })dt^{\prime }$$ what extends Kubo_formulas to arbitrary non-linear responce. As_a consequence, the average value of the_work (i.e. of the energy absorbed_by the system), $\left\langle Work\right\rangle_=\int_{-\infty }^t\left\langle J(t)\right\rangle_x(t)dt$, can be simply expressed_via two-point quazi-equilibrium_correlators. The similar_formulas for higher_cumulants can be obtained \[34\] which_lead to the_formally exact expression for variance of_the_work \[11\] :_$$\left\langle_Work^{(2)}\right\rangle_=2T\left\langle Work\right\rangle +\frac_2T\int_{-\infty }^tx(1)x(2)\left\langle J(1),J(2),J(3)\right\rangle _qx(3)d1d2d3$$_where_$\left\langle Work^{(2)}\right\rangle_=\left\langle Work^2\right\rangle_-\left\langle Work\right\rangle ^2$, and for bre
for this trace-U(1) subgroup. We have demonstrated that, although the trace-U(1) decouples in the limit $k\to 0$, the coupling is not negligibly small at finite momentum scales $k$, as they appear, for example, in scattering experiments. Therefore, observations rule out additional unbroken (massless) trace-U(1) subgroups. An example is the model considered in Ref. [@Khoze:2004zc]. In Ref. [@Khoze:2004zc], the trace-U(1) groups were completely discarded before the symmetry breaking scheme was discussed. A more careful investigation which takes takes into account these subgroups yields the symmetry breaking $\textrm{U}(4)\times \textrm{U}(3)\times \textrm{U} (2) \to \textrm{SU}(3)\times\textrm{SU}(2)\times (\textrm{U}(1))^4$ instead of $\textrm{U}(4)\times \textrm{U}(3)\times \textrm{U} (2) \to \textrm{SU}(3)\times\textrm{SU}(2)\times \textrm{U}(1)$. Therefore we have superfluous U(1) subgroups. Following the above lines explicitely one easily finds that one of the $\textrm{U}(1)$’s has a generator which is proportional to the $9\times 9$ unity matrix. Noncommutativity explicitly breaks Lorentz invariance. Therefore an additional Lorentz symmetry violating structure is allowed in the polarisation tensor. This structure is absent only in supersymmetric models. If supersymmetry is (softly) broken, this additional structure is present in the polarisation tensor of the trace-U(1). It leads to an additional mass $\sim \Delta M^2_{\textrm{SUSY}}$ for one of the transverse polarisation states [@Alvarez-Gaume:2003]. The tight constraints on the photon mass therefore exclude trace-U(1)’s as a candidate for the photon. It turns out that even a small admixture of a trace part to a traceless part (unaffected by these problems) is fatal. The only way out seems to be the construction of the photon from a completely traceless generator. A group theoretic argument shows, that this is impossible whithout having additional unbroken U(1) subgroups. However, those are already excluded from the arguments given above. This result severely restricts the possibilities to construct
for this trace - U(1) subgroup. We have demonstrated that, although the trace - U(1) decouples in the limit $ k\to 0 $, the yoke is not negligibly humble at finite momentum scales $ k$, as they appear, for case, in disperse experiments. Therefore, observation predominate out additional unplowed (massless) touch - U(1) subgroups. An exercise is the model considered in Ref. [ @Khoze:2004zc ]. In Ref. [ @Khoze:2004zc ], the tracing - U(1) group were completely discarded before the symmetry breaking scheme was discourse. A more careful investigation which takes takes into score these subgroups yields the symmetry breaking $ \textrm{U}(4)\times \textrm{U}(3)\times \textrm{U } (2) \to \textrm{SU}(3)\times\textrm{SU}(2)\times (\textrm{U}(1))^4 $ alternatively of $ \textrm{U}(4)\times \textrm{U}(3)\times \textrm{U } (2) \to \textrm{SU}(3)\times\textrm{SU}(2)\times \textrm{U}(1)$. Therefore we have superfluous U(1) subgroups. Following the above line explicitely one easily finds that one of the $ \textrm{U}(1)$ ’s has a generator which is proportional to the $ 9\times 9 $ unity matrix. Noncommutativity explicitly breaks Lorentz invariability. consequently an extra Lorentz symmetry violating structure is allowed in the polarisation tensor. This structure is absent only in supersymmetric model. If supersymmetry is (lightly) broken, this extra social organization is present in the polarisation tensor of the trace - U(1). It leads to an additional bulk $ \sim \Delta M^2_{\textrm{SUSY}}$ for one of the transverse polarisation states [ @Alvarez - Gaume:2003 ]. The tight constraint on the photon mass therefore exclude trace - U(1) ’s as a candidate for the photon. It turns out that even a modest admixture of a trace part to a traceless part (insensible by these problems) is fateful. The only way out seems to be the construction of the photon from a wholly traceless generator. A group theoretic argument shows, that this is impossible whithout having additional unbroken U(1) subgroups. However, those are already excluded from the arguments give above. This result sternly restricts the possibilities to manufacture
fog this trace-U(1) subgroup. We have demonstrajee that, althohgh the grace-U(1) decouples in the limiv $k\ti 0$, tht coupling is not ndgligibly small ar fiiite momentum scemes $k$, as they wppecr, for example, ik scatterinc experiments. Dhdrzfore, observations rule out additionwl unbrpkfn (massless) trwce-U(1) fubgdoups. An example is the model consjdered pn Ref. [@Khoze:2004zc]. In Ref. [@Khoze:2004zc], the trace-U(1) grlups were completely dlscarded bedore rhe symmetry breaking scheme was djscussed. A more careful investieatiou which takgs twnes into acrount nhese subgrouix yielgs the xymmetry breaklng $\txxtrn{U}(4)\times \textrm{U}(3)\times \vextrm{U} (2) \to \textrm{SU}(3)\jimes\textrk{SB}(2)\times (\textrm{U}(1))^4$ insteae if $\teftrm{G}(4)\timdw \tdxtdm{N}(3)\tijes \tedtrj{U} (2) \to \texfrm{SU}(3)\times\twxtrm{SU}(2)\times \textrm{I}(1)$. Eyerefore we hzve su[ewfluous U(1) subgroups. Following the above lihes explicitely one easuly finds that one of the $\textwm{U}(1)$’s has a generator which is proportional to the $9\timea 9$ uulby mxrrlx. Noncommutativity explicitly breaks Lorentz jnfagiance. Therefore cn additional Lotejta symmetry viouating stducture is allowed in the polaeisation uensot. This structure is absent inly in supegsymnetric models. If sbpersymmetry is (xoftlu) broken, this additionau stducture is oresent ih the polarisatiov tvnsos of the trace-U(1). It leads tj an addivionak mass $\rim \Celta I^2_{\textrm{SUSJ}}$ for one of the transvegse plldrisation dtates [@Alvarez-Gaume:2003]. The tight constraints on jhe phmton masf thevefore exclude erace-U(1)’s as a ccndidate for ghe photon. It turis out that qven a small dfmixture of e trace pwrt ro a traceldrs part (unaffevted by tktse problens) is fatal. The onky dzy out seems to be the constructipn uf ehv pioton xrom a complatelh tfscelers generatov. A groip theoretic argumend shkws, that this is ikpjssible qhithout having additoonal unbroken U(1) skbgronps. Hoxever, yhofe are already excluded from tge argumejts given above. Ehis result sevexely restricts the possibilities to consvruct
for this trace-U(1) subgroup. We have demonstrated the decouples in limit $k\to 0$, small finite momentum scales as they appear, example, in scattering experiments. Therefore, observations out additional unbroken (massless) trace-U(1) subgroups. An example is the model considered in [@Khoze:2004zc]. In Ref. [@Khoze:2004zc], the trace-U(1) groups were completely discarded before the symmetry scheme discussed. more investigation which takes takes into account these subgroups yields the symmetry breaking $\textrm{U}(4)\times \textrm{U}(3)\times \textrm{U} (2) \textrm{SU}(3)\times\textrm{SU}(2)\times (\textrm{U}(1))^4$ instead of $\textrm{U}(4)\times \textrm{U}(3)\times \textrm{U} (2) \textrm{SU}(3)\times\textrm{SU}(2)\times \textrm{U}(1)$. Therefore we superfluous U(1) subgroups. Following the lines one easily that of $\textrm{U}(1)$’s has a which is proportional to the $9\times 9$ unity matrix. Noncommutativity explicitly breaks Lorentz invariance. Therefore an additional symmetry violating allowed in polarisation This is absent only models. If supersymmetry is (softly) broken, is present in the polarisation tensor of the It leads an additional mass $\sim \Delta M^2_{\textrm{SUSY}}$ one of the transverse polarisation states [@Alvarez-Gaume:2003]. The constraints on the photon mass therefore exclude trace-U(1)’s as a candidate for the photon. It that even a small of a trace to traceless (unaffected these problems) fatal. The only way out seems to be the construction of photon from a completely traceless generator. A group theoretic argument this impossible whithout having unbroken U(1) subgroups. However, are excluded from the arguments This severely to
for this trace-U(1) subgroup. We haVe demonstrAted tHat, AltHoUgh tHe trAce-U(1) decouples iN The lImit $k\to 0$, the coupling is noT neglIgIBly sMAlL at fiNite momENtUM ScaLeS $k$, As tHeY ApPear, fOr eXample, iN scatterinG exPeRiments. ThereFOrE, observatiOns Rule out additIonAl unbrOkEn (mASslesS) trAce-U(1) sUbgrouPS. An exaMple is the MoDEl consIDered in rEF. [@KHoze:2004Zc]. In Ref. [@Khoze:2004zc], the TRaCE-U(1) groups were coMpleteLy DIsCARdeD beFore the symMeTry brEAking scHEmE WAS diSCussed. A more caReful investIGatIon whiCh TakES takes Into aCcOUnt These subgroUps yIelds the sYmmetrY BreakinG $\Textrm{U}(4)\Times \tExtRm{U}(3)\TimeS \TeXtRm{U} (2) \To \TExtRM{Su}(3)\tiMEs\tExtrm{SU}(2)\tImEs (\TextrM{U}(1))^4$ inSTEAD of $\tExtRm{U}(4)\tImes \tExtrm{U}(3)\times \teXtrM{U} (2) \to \TExtRm{SU}(3)\tImes\tExtrM{Su}(2)\timeS \textrM{U}(1)$. TheReFore we have superFluoUs U(1) subgroUps. foLloWiNg the ABove liNes ExpLicitelY one easILy fInDS THaT one of the $\textrm{U}(1)$’s hAs A GEnErator whIch is pROpOrTIonal to tHe $9\TimEs 9$ unITY matrIx. NoNCoMmutativIty expLIcItLy breakS LOrentz InVarIanCe. TheREforE an addItional LOrentZ Symmetry violatINg structure is ALlOWEd IN the PolArisation teNsor. tHis sTrucTUrE is ABsent Only iN sUPeRSymmetric models. If suPeRsymmeTry is (Softly) broken, tHis additioNAL StructurE is pREsENt in the polarisAtion Tensor of thE Trace-U(1). It Leads To an addiTional masS $\SIm \Delta M^2_{\TexTrm{sUSy}}$ foR ONe Of the transverSE PolaRiSation sTatEs [@AlvarEz-GAumE:2003]. ThE tiGhT constraiNts on the PhOtOn MaSs tHerefORe excludE tRacE-U(1)’S as A candIDate foR the pHotoN. IT tURns Out that EVeN A SmalL aDmIxtuRe oF a Trace Part TO a tRacelesS part (unafFecTEd by ThEsE probleMs) is fatal. The oNlY way out seeMs To bE the coNSTruction Of the photon from a completELy traceLesS geneRatoR. A group thEorEtic arGumENt showS, that tHis is ImPosSIBle whITHoUt hAvIng additioNAL unBrokeN U(1) SubgRoups. HoWever, those are alreaDY exCluded from the ArgUmenTS GiVen ABoVE. ThIs REsuLT Severely restricTs the possiBiLItIes to constRUct
for this trace-U(1) subgr oup. We h ave d emo nst ra tedthat , although the trac e-U(1) decouples in th e lim it $k\t o 0 $, th e coupl i ng i s n ot n egl ig i bl y sma llat fini te momentu m s ca les $k$, ast he y appear,for example, in sc atteri ng ex p erime nts . The refore , obser vations r ul e out a d ditiona l un brok en (massless) tra c e- U (1) subgroups. An ex am p le i s t hemodel cons id eredi n Ref.[ @K h o z e:2 0 04zc]. In Ref . [@Khoze:2 0 04z c], th etra c e-U(1) grou ps wer e completel y di scarded b eforet he symm e try bre akingsch eme was di sc uss ed . Am or e c a ref ul inves ti ga tionwhic h t a kestak es i nto a ccount thesesub grou p s y ields thesymm et ry br eaking $\te xt rm{U}(4)\times\tex trm{U}(3) \ti me s \ te xtrm{ U } (2)\to \t extrm{S U}(3)\t i mes \t e x t rm {SU}(2)\times (\te xt r m {U }(1))^4$ inste a dof $\textrm {U }(4 )\ti m e s \te xtrm { U} (3)\time s \tex t rm {U } (2) \ to \text rm {SU }(3 )\tim e s\te xtrm{S U}(2)\ti mes \ t extrm{U}(1)$.T herefore we h a ve s up e rflu ous U(1) subgr oups . Fol lowi n gthe above line se xp l icitely one easilyfi nds th at on e of the $\te xtrm{U}(1) $ ’ s has a g ener a to r which is prop ortio nal to the $9\times 9$ u nity mat rix. Nonc o m mutativi tyexp lic itl y br eaks Lorentzi n vari an ce. The ref ore anadd iti ona l L or entz symm etry vio la ti ng s tru cture is allow ed in t hepolar i sation tens or.Th is str ucturei sa b sent o nl y in su pe rsymm etri c mo dels. I f supersy mme t ry i s(s oftly)broken, thisad ditional s tr uct ure is p resent i n the polarisation tens o r of th e t race- U(1) . It lead s t o an a ddi t ionalmass $ \sim\D elt a M^2_{ \ t ex trm {S USY}}$ for o neof th etran sversepolarisation state s [@ Alvarez-Gaume :20 03]. T he ti g ht con st r ain t s on the photonmass there fo r eexclude tr a ce- U( 1)’s as a cand idate for the photon.It turnsou t th a t ev en a small admixtu re of a t r ace p a rt to a tr aceles spar t (un affect e d b y the se pro bl ems) i s fat al . The on ly way out seems to bethe co nstru cti on of the ph o ton from a c ompl etely trac ele ssgener ato r . A g roup th eor e tic a rgum e nt shows, th att h is is impossi b l e wh ithou t h a ving a ddit ional unbroken U( 1 ) subgroups. H owev e r , t hos e are a lready exclude d f ro m the argu me nts given a bove. T hi s resu lt sev erelyrestric t s t h e poss ibil iti es to con str uc t
for_this trace-U(1)_subgroup. We have demonstrated that,_although the_trace-U(1)_decouples in_the_limit $k\to 0$,_the coupling is_not negligibly small at_finite momentum scales_$k$,_as they appear, for example, in scattering experiments. Therefore, observations rule out additional unbroken_(massless)_trace-U(1) subgroups._An_example_is the model considered in_Ref. [@Khoze:2004zc]. In Ref. [@Khoze:2004zc],_the trace-U(1)_groups were completely discarded before the symmetry breaking_scheme_was discussed. A_more careful investigation which takes takes into account these_subgroups yields the symmetry breaking $\textrm{U}(4)\times_\textrm{U}(3)\times \textrm{U} (2)_\to_\textrm{SU}(3)\times\textrm{SU}(2)\times_(\textrm{U}(1))^4$ instead of $\textrm{U}(4)\times_\textrm{U}(3)\times \textrm{U} (2) \to \textrm{SU}(3)\times\textrm{SU}(2)\times \textrm{U}(1)$._Therefore we have superfluous U(1) subgroups._Following the above lines explicitely one easily_finds that one of the $\textrm{U}(1)$’s_has a generator which is_proportional to_the $9\times 9$ unity matrix._Noncommutativity explicitly breaks_Lorentz invariance._Therefore an additional_Lorentz symmetry violating structure is allowed_in the polarisation_tensor. This structure is absent only_in_supersymmetric models. If_supersymmetry_is_(softly) broken,_this additional structure_is_present in_the_polarisation tensor of the trace-U(1). It_leads_to an additional mass $\sim \Delta M^2_{\textrm{SUSY}}$_for one of the_transverse_polarisation states [@Alvarez-Gaume:2003]. The_tight constraints on the photon_mass therefore exclude trace-U(1)’s as a_candidate for_the photon._It turns out that even a small admixture of a trace_part to a traceless part (unaffected_by these problems) is_fatal. The_only_way out seems_to_be the_construction of the photon from a completely_traceless generator._A group theoretic argument shows, that_this is impossible whithout_having_additional unbroken U(1) subgroups. However, those_are already excluded from the arguments_given above. This result severely restricts_the_possibilities_to construct
10987) of the relations are across-sentence relations. The inter-annotator agreement (IAA) for the final annotation schema is 0.89. It was calculated for the 61 reports annotated by both annotators on a token level using the Cohen’s Kappa coefficient ($\kappa$) [@Cohen:1960]. Discussion and Conclusions {#sec:discussion} ========================== We presented a manually annotated corpus of entities and relations in radiology reports written in Spanish. The goal was twofold: to have an annotated dataset available for evaluating NER and RE algorithms results and for training of supervised models and to present an annotation guideline that can be used by other researchers with similar needs. The creation of the corpus was not an easy task. Many annotation-revision iterations had to be performed in order to arrive to a stabilized annotation schema. According to what we expected, $\kappa$ improved in each annotation iteration step. Data had to be anonymized. Furthermore, the shortness of the texts, the abundance of abbreviations and acronyms (about 6% of the AE and FI are written as such and there are 105 different abbreviations or acronyms in 513 reports), the specificity of the medical language, the existence of multi-segment terms and the existence of relations between sentences, makes not only the NER and RE tasks, but also the annotation task a difficult one. The relation of findings with temporal terms, negation terms and uncertainty terms should be taken into account to determine their factuality. The abundance of negated findings (56%) might lead to the implementation of methods to detect negated findings in reports (see, and for Spanish). The difference of criteria among the annotators helps us determine that the evaluation of NER systems is not an easy task. Finally, the pre-annotation helped us speed the annotation process, although it might have biased the annotation results. [^1]: Ultrasound reports are texts describing what has been observed in a type of imaging study called ultrasound examination. [^2]: <http://radlex.org/> [^3]: <https://www.nlm.nih.gov/research/umls/> [^4]: In “right lobe of the liver has the usual size”, “right lobe of the liver” should be annotated, although it is not associated to any finding. [^5]: <http://brat.nlplab.org/> [^6]: All the annotationsrevision
10987) of the relations are across - sentence relations. The inter - annotator agreement (IAA) for the final note schema is 0.89. It was account for the 61 reports annotated by both annotator on a token horizontal surface using the Cohen ’s Kappa coefficient ($ \kappa$) [ @Cohen:1960 ]. Discussion and Conclusions { # sec: discussion } = = = = = = = = = = = = = = = = = = = = = = = = = = We present a manually gloss corpus of entity and relative in radiology report written in Spanish. The goal was double: to accept an annotated dataset available for evaluating NER and RE algorithms results and for training of supervised models and to salute an annotation guideline that can be used by early researchers with similar need. The creation of the principal was not an easy job. Many annotation - revision iterations give birth to be performed in order to arrive to a stabilized annotation schema. According to what we expect, $ \kappa$ better in each annotation iteration step. Data had to be anonymized. Furthermore, the shortness of the texts, the abundance of abbreviations and acronyms (approximately 6% of the AE and FI are spell as such and there are 105 unlike abbreviation or acronym in 513 reports), the specificity of the medical language, the existence of multi - segment terms and the existence of relations between sentences, lay down not only the NER and RE tasks, but also the annotation task a difficult one. The relation of findings with temporal terms, negation term and doubt terms should be taken into account to determine their factuality. The abundance of negated findings (56 %) might lead to the implementation of method to detect negated findings in reports (see, and for Spanish). The dispute of criteria among the annotators help us determine that the evaluation of NER organization is not an easy undertaking. Finally, the pre - note helped us speed the annotation summons, although it might have biased the note results. [ ^1 ]: Ultrasound reports are text describing what has been observed in a type of imaging study call ultrasound examination. [ ^2 ]: < http://radlex.org/ > [ ^3 ]: < https://www.nlm.nih.gov/research/umls/ > [ ^4 ]: In “ correct lobe of the liver has the usual size ”, “ right lobe of the liver ” should be annotated, although it is not associated to any finding. [ ^5 ]: < http://brat.nlplab.org/ > [ ^6 ]: All the annotationsrevision
10987) ov the relations are acrors-sentence relajiins. Thx inter-znnotatof agreement (IAA) for the finap qnnotqtion schema is 0.89. It war calculaned for tye 61 ceports annotatev by botm annkbatorv on a token leyel using tve Cohen’s Kappd zozfficient ($\kappa$) [@Cohen:1960]. Discussion and Sonclusoojs {#sec:discussijn} ========================== We [ressnted a manually annotated corpus kf entiuies and relations in radiology reports writhen ln Spanish. The goap was twofood: tj have an annutated dataset availabme for evaluating NER and RE aleoritkms results abd vmr training of slpervised modcks and to prexent an annotabion juidwline that can be usev by other researchets with sikimar needs. The creqtuon ox tha cofpus waa iot an eady vask. Many ahnotation-recision iterations hsd np be performsd in jrqer to arrive to a stabilized annotatiot sdhema. According to what we expected, $\kappa$ imkroved in qach annotation iteration step. Data had to be anotymizxd. Fuxbmermuee, the shortness of the texts, the abundance of wgbteniations and acrokyms (about 6% of the AF sgd FI are wrijten as sudh and there are 105 fifferegt abvreviatiogs ot acronyms in 513 reports), the wpecificity jd the medical langbage, the exirtenve of multi-segment terms and the existence lf relatikvs between sentevcex, kakes nou only the NER and RE tasks, but also tfe amnotatyon task a difflwult one. The relatiln of fhndings wihh temporal terms, negation terms and uncertaimtf tvrms shoujd be taken into ascount to detetmine thenr facguality. Thv abundanre of negateq findings (56%) mhhht lead to vhe impleientqtiob of megfods to detect negated ynndings ib reports (see, and nor Skahish). The differzuct if criteria ampng thq wniotatjss helps us getefmivr thag the evaluctkon pf NER systems is nod an easy task. Finally, yhc pre-annojation hejped us speed the annotation prlcess, alvhough it might have biased the annotatjon resulhs. [^1]: Mltrasound re[ortw are texts bescribing what has been observed in a tbpe of imaging study caoled ultrasound exaklnation. [^2]: <http://cadlex.jrg/> [^3]: <https://fww.nlm.nih.gov/research/ymls/> [^4]: In “right lobt of the liver has the hsual vize”, “gight lobe of the liver” should be annotated, although it is not associated to anb sinding. [^5]: <httl://bray.nlpldb.jxg/> [^6]: All ehe eniotationsrevision
10987) of the relations are across-sentence relations. agreement for the annotation schema is the reports annotated by annotators on a level using the Cohen’s Kappa coefficient [@Cohen:1960]. Discussion and Conclusions {#sec:discussion} ========================== We presented a manually annotated corpus of and relations in radiology reports written in Spanish. The goal was twofold: to an dataset for NER and RE algorithms results and for training of supervised models and to present an annotation that can be used by other researchers with needs. The creation of corpus was not an easy Many iterations had be in to arrive to stabilized annotation schema. According to what we expected, $\kappa$ improved in each annotation iteration step. Data had be anonymized. shortness of texts, abundance abbreviations and acronyms of the AE and FI are and there are 105 different abbreviations or acronyms 513 reports), specificity of the medical language, the of multi-segment terms and the existence of relations sentences, makes not only the NER and RE tasks, but also the annotation task a The relation of findings temporal terms, negation and terms be into account determine their factuality. The abundance of negated findings (56%) might lead the implementation of methods to detect negated findings in reports for The difference of among the annotators helps determine the evaluation of NER not easy pre-annotation us the annotation process, although might have biased the annotation [^1]: Ultrasound reports are observed in a type of imaging study called examination. [^2]: <http://radlex.org/> [^3]: <https://www.nlm.nih.gov/research/umls/> [^4]: In lobe of the liver has the usual size”, “right lobe of the should be it is not associated to any finding. [^5]: [^6]: All the annotationsrevision
10987) of the relations are across-seNtence relaTions. the IntEr-AnnoTatoR agreement (IAA) fOR the Final annotation schema iS 0.89. It waS cALculATeD for tHe 61 reporTS aNNOtaTeD bY boTh ANnOtatoRs oN a token Level using The coHen’s Kappa coeFFiCient ($\kappa$) [@cohEn:1960]. Discussion And concluSiOns {#SEc:disCusSion} ========================== WE preseNTed a maNually annOtATed corPUs of entITIeS and Relations in radiolOGy REports written iN SpaniSh. tHe GOAl wAs tWofold: to haVe An annOTated daTAsET AVaiLAble for evaluaTing NER and Re AlgOrithmS rEsuLTs and fOr traInINg oF supervised ModeLs and to prEsent aN AnnotatIOn guideLine thAt cAn bE useD By OtHer ReSEarCHeRs wITh sImilar neEdS. THe creAtioN OF THe coRpuS was Not an Easy task. Many aNnoTatiON-reVisioN iterAtioNs Had to Be perfOrmed In Order to arrive to A staBilized anNotAtIon ScHema. ACCordinG to WhaT we expeCted, $\kapPA$ imPrOVED iN each annotation iteRaTIOn Step. Data Had to bE AnOnYMized. FurThErmOre, tHE ShortNess OF tHe texts, tHe abunDAnCe Of abbreViAtions AnD acRonYms (abOUt 6% of The AE aNd FI are wRitteN As such and there ARe 105 different abBReVIAtIOns oR acRonyms in 513 repOrts), THe spEcifICiTy oF The meDical LaNGuAGe, the existence of mulTi-SegmenT termS and the existeNce of relatIONS between SentENcES, makes not only tHe NER And RE tasks, BUt also thE annoTation taSk a difficULT one. The rElaTioN of FinDINgS with temporal TERms, nEgAtion teRms And unceRtaIntY teRms ShOuld be takEn into acCoUnT tO dEteRmine THeir factUaLitY. THe aBundaNCe of neGated FindInGs (56%) MIghT lead to THe IMPlemEnTaTion Of mEtHods tO detECt nEgated fIndings in RepORts (sEe, AnD for SpaNish). The differEnCe of criterIa AmoNg the aNNOtators hElps us determine that the eVAluatioN of nER syStemS is not an eAsy Task. FiNalLY, the prE-annotAtion HeLpeD US speeD THe AnnOtAtion proceSS, AltHough It MighT have biAsed the annotation rESulTs. [^1]: Ultrasound rEpoRts aRE TeXts DEsCRibInG WhaT HAs been observed iN a type of imAgINg Study calleD UltRaSound exAminatiOn. [^2]: <httP://Radlex.oRg/> [^3]: <https://wwW.nlm.nih.goV/rEseaRCH/umLs/> [^4]: In “right lObe of the Liver has tHE usuaL SiZe”, “rigHt lObe of tHe LivEr” shoUld be aNNotAted, aLthougH iT is not AssocIaTed to any Finding. [^5]: <http://brat.nlplab.orG/> [^6]: All thE annoTatIonsrevisIon
10987) of the relations a re across- sente nce re la tion s. T he inter-annot a toragreement (IAA) for th e fin al anno t at ion s chema i s 0 . 8 9.It w asca l cu lated fo r the 6 1 reportsann ot ated by both an notators o n a token level us ing th eCoh e n’s K app a coe fficie n t ($\k appa$) [@ Co h en:196 0 ]. Dis c u ss ionand Conclusions { # se c :discussion} = ====== == = == = = === === ====== We p resen t ed a ma n ua l l y an n otated corpus of entitie s an d rela ti ons in rad iolog yr epo rts written inSpanish.The go a l was t w ofold:to hav e a n a nnot a te ddat as e t a v ai lab l e f or evalu at in g NER and R E algo rit hmsresul ts and for tr ain ingo f s uperv isedmode ls andto pre sentan annotation gui deli ne that c anbe us ed by o t her re sea rch ers wit h simil a r n ee d s . T he creation of the c o r pu s was no t an e a sy t a sk. Many a nno tati o n -revi sion it erations had t o b eperform ed in or de r t o a rrive to a stabi lized an notat i on schema. Acc o rding to what we e xp e cted , $ \kappa$ imp rove d ineach an not a tionitera ti o ns tep. Data had to be a nonymi zed.Furthermore,the shortn e s s of thetext s ,t he abundance o f abb reviations and acro nyms(about 6 % of theA E and FIare wr itt ena s s uch and there a re 1 05 differ ent abbrev iat ion s o r a cr onyms in513 repo rt s) ,th e s pecif i city ofth e m ed ica l lan g uage,the e xist en ce ofmulti-s e gm e n t te rm sandthe e xiste nceo f r elation s between se n tenc es ,makes n ot only the N ER and RE ta sk s,but al s o the ann otation task a difficul t one. The rela tion of findi ngs withtem p oral t erms,negat io n t e r ms an d un cer ta inty terms s hou ld be t aken into a ccount to determin e th eir factualit y.Thea b un dan c eo f n eg a ted f indings (56%) m ight leadto th e implemen t ati on of met hods to dete c t negat ed findin gs in rep or ts ( s e e,and for Sp anish).The diffe r enceo fcrite ria among t heannot atorsh elp s usdeterm in e that theev aluation of NER systems is notan eas y tas k. Finally, th e pr e-annotat ionhelped usspe edthe a nno t ation pro c es s,a lthou gh i t might ha v ebia s e dthe annotat i o n re sults . [ ^1]: U ltra sound reports are texts describi ng w h a t h asb eenob served in a ty peof i maging s tu dy called u ltrasoun de xamin ation. [^2] : <http : / /r a dlex.o rg/> [ ^3]: <htt ps: // w ww.nlm. ni h. g ov/res earc h/ umls/> [^4] : In“ r ight lobe of the live r has t h e u sualsi ze”, “r i ghtlobe of th e liver” sh ould b e an notat ed, alt ho ugh it is n ot associa t ed to any find ing. [ ^5 ]: < htt p://br at.n l p lab.o rg/> [^6 ]: All th e an n ot at i ons revi sion
10987)_of the_relations are across-sentence relations._The inter-annotator_agreement_(IAA) for_the_final annotation schema_is 0.89. It_was calculated for the_61 reports annotated_by_both annotators on a token level using the Cohen’s Kappa coefficient ($\kappa$) [@Cohen:1960]. Discussion and_Conclusions_{#sec:discussion} ========================== We presented_a_manually_annotated corpus of entities and_relations in radiology reports written_in Spanish._The goal was twofold: to have an annotated_dataset_available for evaluating_NER and RE algorithms results and for training of_supervised models and to present an_annotation guideline that_can_be_used by other researchers_with similar needs. The creation of_the corpus was not an easy_task. Many annotation-revision iterations had to be_performed in order to arrive to_a stabilized annotation schema. According_to what_we expected, $\kappa$ improved in_each annotation iteration_step. Data_had to be_anonymized. Furthermore, the shortness of the_texts, the abundance_of abbreviations and acronyms (about 6%_of_the AE and_FI_are_written as_such and there_are_105 different_abbreviations_or acronyms in 513 reports), the_specificity_of the medical language, the existence of_multi-segment terms and the_existence_of relations between sentences,_makes not only the NER_and RE tasks, but also the_annotation task_a difficult_one. The relation of findings with temporal terms, negation terms and uncertainty_terms should be taken into account_to determine their factuality._The abundance_of_negated findings (56%)_might_lead to_the implementation of methods to detect negated_findings in_reports (see, and for Spanish). The_difference of criteria among_the_annotators helps us determine that the_evaluation of NER systems is not_an easy task. Finally, the pre-annotation_helped_us_speed the annotation process, although_it might have biased the annotation_results. [^1]: Ultrasound reports_are texts describing what has been observed_in_a type of imaging study called_ultrasound_examination. [^2]: <http://radlex.org/> [^3]: <https://www.nlm.nih.gov/research/umls/> [^4]: In “right lobe_of_the_liver has the usual size”,_“right lobe of the liver” should_be annotated, although it is not associated to any_finding. [^5]: <http://brat.nlplab.org/> [^6]: All_the annotationsrevision
rm eje}\sqrt{G M_{\rm rem}R_{\rm eje}}$, respectively, where $R_{\rm torus}$ is typical radius of torus, $R_{\rm eje}$ is the typical distance from merger center to where ejecta occur and $M_{\rm rem}$ is the gravitational mass of remnant core. For dynamical ejecta launched within dynamical timescale ($\leq 10~{\rm ms}$), $R_{\rm eje}$ can be estimated as the typical radius of the surrounding torus at such an early time. The late-time neutrino-driven and magnetically-driven ejecta mainly come from surrounding torus with $R_{\rm eje}=R_{\rm torus}$ [see also @Shibata2019] while others come from the core of the remnant with $R_{\rm eje}=R_{\rm rem}$. Pre-collapse, a large amount of neutrinos radiate from the hot, violent and rapidly rotating remnant core and make it cool down and spin down. @Baumgarte1998 estimated the angular momentum carried away by neutrinos as $J_{\rm \nu}\approx (2/3)c^{-2}R_{\rm rem}^2\Omega E_\nu \approx2.2\times10^{48}\rm erg~s(E_{\rm \nu}/{0.1M_\odot c^2})(R_{\rm rem}/{13\rm km})^{2} (\Omega/{10^4\rm rad/s})$, where $\Omega$ is the angular velocity of the rigidly rotating massive neutron star. The equatorial radius $R_{\rm rem}$ can be approximated as $R_{\rm crit}=(1+0.032\zeta_{\rm TOV}^{-1.6}j^2+0.014\zeta_{\rm TOV}^{-3.2}j^4)R_{\rm TOV}$ (see Appendix for details). For neutron star rotating in the Keplerian limit with $R_{\rm TOV}$=10 km, we have $R_{\rm rem}\approx 13$ km. The angular momentum carried away by post-merger gravitational wave can be expressed as $J_{\rm GW,p}= J_0(1-\sqrt{1-E_{\rm GW,p}/E_0})$, where the rotating energy at the onset of merger can be approximated as $E_0 = \frac{1}{2}I\Omega^2 \approx 1.5\times10^{53
rm eje}\sqrt{G M_{\rm rem}R_{\rm eje}}$, respectively, where $ R_{\rm torus}$ is typical radius of torus, $ R_{\rm eje}$ is the typical distance from merger plaza to where ejecta happen and $ M_{\rm rem}$ is the gravitational mass of remnant core. For dynamic ejecta launched within dynamic timescale ($ \leq 10~{\rm ms}$), $ R_{\rm eje}$ can be estimated as the typical spoke of the surrounding torus at such an early time. The late - clock time neutrino - driven and magnetically - driven ejecta chiefly come from surrounding torus with $ R_{\rm eje}=R_{\rm torus}$ [ see also @Shibata2019 ] while others derive from the core of the remnant with $ R_{\rm eje}=R_{\rm rem}$. Pre - collapse, a large amount of neutrino radiate from the hot, violent and rapidly rotate remnant core and make it cool down and whirl down. @Baumgarte1998 estimated the angular momentum stock away by neutrino as $ J_{\rm \nu}\approx (2/3)c^{-2}R_{\rm rem}^2\Omega E_\nu \approx2.2\times10^{48}\rm erg ~ s(E_{\rm \nu}/{0.1M_\odot c^2})(R_{\rm rem}/{13\rm km})^{2 } (\Omega/{10 ^ 4\rm rad / s})$, where $ \Omega$ is the angular velocity of the rigidly rotating massive neutron star. The equatorial radius $ R_{\rm rem}$ can be estimate as $ R_{\rm crit}=(1 + 0.032\zeta_{\rm TOV}^{-1.6}j^2 + 0.014\zeta_{\rm TOV}^{-3.2}j^4)R_{\rm TOV}$ (see Appendix for details). For neutron star rotating in the Keplerian limit with $ R_{\rm TOV}$=10 kilometer, we suffer $ R_{\rm rem}\approx 13 $ km. The angular momentum carried away by post - merger gravitational wave can be expressed as $ J_{\rm GW, p}= J_0(1-\sqrt{1 - E_{\rm GW, p}/E_0})$, where the rotating energy at the onset of merger can be approximated as $ E_0 = \frac{1}{2}I\Omega^2 \approx 1.5\times10^{53
rm fje}\sqrt{G M_{\rm rem}R_{\rm eje}}$, vespectively, whete $R_{\rm tmrus}$ ia typicau radius of torus, $R_{\rm eje}$ is tye tykpcal distance from mefger centvr to wheee eoecta occur and $M_{\rm rem}$ lf ths grarivational mass on remnant cmre. For dynamiwau zjecta launched within dynamical timqscale ($\kee 10~{\rm ms}$), $R_{\rm ejg}$ can fe eaninated as the typical radius or the slrrounding torus st such an early time. The pate-hime neutrino-drivej and magnejjcajoy-driven ejezta mainly come from sorrounding torus with $R_{\rm eje}=R_{\rm torux}$ [see also @Syibwja2019] while othxrs coie from the gpre of the reknant with $R_{\rm ejx}=R_{\rm rem}$. Pre-collapse, a larje amount of neutrinjs radiata yrom the hot, violent qne raphdly rotxrine rtmnent core wnd make it ckol down ane spin down. @Baumgarue1998 qwtimated the zngulaw iomentum carried away by neutrinos as $J_{\gm \nh}\approx (2/3)c^{-2}R_{\rm rem}^2\Omega E_\bu \approx2.2\times10^{48}\rm erg~s(G_{\rm \nu}/{0.1M_\odoe c^2})(R_{\rm rem}/{13\rm km})^{2} (\Omega/{10^4\rm rad/s})$, where $\Omega$ is the ancular xeliclty ud hhe rigidly rotating massive neutron star. The sqianorial radius $R_{\rm rem}$ can be aoptjximated as $R_{\tm crit}=(1+0.032\vstz_{\rm TOV}^{-1.6}j^2+0.014\zeta_{\rm TOV}^{-3.2}u^4)R_{\rm TOD}$ (see Appendix for details). For neutron star ritating in tke Jeplerian limit wich $R_{\rm TOV}$=10 ko, we have $R_{\rm rem}\approx 13$ km. The augular momentum cwrried awzh by post-merger eranitadional wave can be expressqd as $J_{\rm GW,p}= J_0(1-\sqrt{1-E_{\fm GE,p}/E_0})$, whqre the rohatinn energy at the onsft of marger can he approximated as $E_0 = \frac{1}{2}I\Omege^2 \approx 1.5\timex10^{53
rm eje}\sqrt{G M_{\rm rem}R_{\rm eje}}$, respectively, where is radius of $R_{\rm eje}$ is center where ejecta occur $M_{\rm rem}$ is gravitational mass of remnant core. For ejecta launched within dynamical timescale ($\leq 10~{\rm ms}$), $R_{\rm eje}$ can be estimated the typical radius of the surrounding torus at such an early time. The neutrino-driven magnetically-driven mainly from surrounding torus with $R_{\rm eje}=R_{\rm torus}$ [see also @Shibata2019] while others come from the core the remnant with $R_{\rm eje}=R_{\rm rem}$. Pre-collapse, a amount of neutrinos radiate the hot, violent and rapidly remnant and make cool and down. @Baumgarte1998 estimated angular momentum carried away by neutrinos as $J_{\rm \nu}\approx (2/3)c^{-2}R_{\rm rem}^2\Omega E_\nu \approx2.2\times10^{48}\rm erg~s(E_{\rm \nu}/{0.1M_\odot c^2})(R_{\rm rem}/{13\rm (\Omega/{10^4\rm rad/s})$, is the velocity the rotating massive neutron equatorial radius $R_{\rm rem}$ can be crit}=(1+0.032\zeta_{\rm TOV}^{-1.6}j^2+0.014\zeta_{\rm TOV}^{-3.2}j^4)R_{\rm TOV}$ (see Appendix for details). neutron star in the Keplerian limit with $R_{\rm km, we have $R_{\rm rem}\approx 13$ km. The momentum carried away by post-merger gravitational wave can be expressed as $J_{\rm GW,p}= J_0(1-\sqrt{1-E_{\rm GW,p}/E_0})$, rotating energy at the of merger can approximated $E_0 \frac{1}{2}I\Omega^2 1.5\times10^{53
rm eje}\sqrt{G M_{\rm rem}R_{\rm eje}}$, resPectively, wHere $R_{\Rm tOruS}$ iS typIcal Radius of torus, $R_{\RM eje}$ Is the typical distance frOm merGeR CentER tO wherE ejecta OCcUR And $m_{\rM rEm}$ iS tHE gRavitAtiOnal masS of remnant CorE. FOr dynamical eJEcTa launched WitHin dynamical TimEscale ($\LeQ 10~{\rm MS}$), $R_{\rm eJe}$ cAn be eStimatED as the Typical raDiUS of the SUrroundING tOrus At such an early time. tHe LAte-time neutrinO-driveN aND mAGNetIcaLly-driven eJeCta maINly come FRoM SURroUNding torus witH $R_{\rm eje}=R_{\rm tORus}$ [See alsO @SHibATa2019] whilE otheRs COme From the core Of thE remnant wIth $R_{\rm EJe}=R_{\rm reM}$. pre-collApse, a lArgE amOunt OF nEuTriNoS RadIAtE frOM thE hot, violEnT aNd rapIdly ROTATing RemNant Core aNd make it cool dOwn And sPIn dOwn. @BaUmgarTe1998 esTiMated The angUlar mOmEntum carried awaY by nEutrinos aS $J_{\rM \nU}\apPrOx (2/3)c^{-2}R_{\rM Rem}^2\OmeGa E_\Nu \aPprox2.2\tiMes10^{48}\rm erG~S(E_{\rM \nU}/{0.1m_\ODoT c^2})(R_{\rm rem}/{13\rm km})^{2} (\Omega/{10^4\rM rAD/S})$, wHere $\OmegA$ is the ANgUlAR velocitY oF thE rigIDLy rotAtinG MaSsive neuTron stAR. THe EquatorIaL radiuS $R_{\Rm rEm}$ cAn be aPProxImated As $R_{\rm criT}=(1+0.032\zeta_{\RM TOV}^{-1.6}j^2+0.014\zeta_{\rm TOV}^{-3.2}J^4)r_{\rm TOV}$ (see AppeNDiX FOr DEtaiLs). FOr neutron stAr roTAtinG in tHE KEplERian lImit wItH $r_{\rM tOV}$=10 km, we have $R_{\rm rem}\apPrOx 13$ km. ThE anguLar momentum caRried away bY POSt-merger GravITaTIonal wave can be ExpreSsed as $J_{\rm Gw,P}= J_0(1-\sqrt{1-E_{\rM GW,p}/E_0})$, Where the Rotating eNERgy at the OnsEt oF meRgeR CAn Be approximateD AS $E_0 = \frAc{1}{2}i\Omega^2 \aPprOx 1.5\times10^{53
rm eje}\sqrt{G M_{\rm rem} R_{\rm eje }}$,res pec ti vely , wh ere $R_{\rm to r us}$ is typical radius oftorus ,$ R_{\ r meje}$ is the ty p i cal d is tan ce fr om me rge r cente r to where ej ec ta occur and $M _{\rm rem} $ i s the gravit ati onal m as s o f remn ant core . Ford ynamic al ejecta l a unched withind y na mica l timescale ($\le q 1 0 ~{\rm ms}$), $ R_{\rm e j e} $ can be estimated a s the typical ra d i u s o f the surround ing torus a t su ch anea rly time.The l at e -ti me neutrino -dri ven and m agneti c ally-dr i ven eje cta ma inl y c omef ro msur ro u ndi n gtor u s w ith $R_{ \r meje}= R_{\ r m t orus }$[see also @Shibata2019 ] w hile oth ers c ome f romth e cor e of t he re mn ant with $R_{\r m ej e}=R_{\rm re m} $. P re-co l lapse, alar ge amou nt of n e utr in o s ra diate from the hot ,v i ol ent andrapidl y r ot a ting rem na ntcore a nd ma ke i t c ool down and s p in d own. @B au mgarte 19 98est imate d the angul ar momen tum c a rried away byn eutrinos as $ J _{ \ r m\ nu}\ app rox (2/3)c^ {-2} R _{\r m re m }^ 2\O m ega E _\nu\a p pr o x2.2\times10^{48}\r merg~s( E_{\r m \nu}/{0.1M_ \odot c^2} ) ( R _{\rm re m}/{ 1 3\ r m km})^{2} (\O mega/ {10^4\rm r a d/s})$,where $\Omega $ is thea n gular ve loc ity of th e ri gidly rotatin g mass iv e neutr onstar. T heequ ato ria lradius $R _{\rm re m} $ca nbeappro x imated a s$R_ {\ rmcrit} = (1+0.0 32\ze ta_{ \r mT OV} ^{-1.6} j ^2 + 0 .014 \z et a_{\ rmTO V}^{- 3.2} j ^4) R_{\rmTOV}$ (se e A p pend ix f or deta ils). For neu tr on star ro ta tin g in t h e Kepleri an limit with $R_{\rm T O V}$=10km, we h ave$R_{\rm r em} \appro x 1 3 $ km. The a ngula rmom e n tum c a r ri edaw ay by post - m erg er gr av itat ional w ave can be express e d a s $J_{\rm GW, p}= J_0 ( 1 -\ sqr t {1 - E_{ \r m GW , p }/E_0})$, where the rotat in g e nergy at t h e o ns et of m erger c an be approxi mated as$E_0 = \f ra c{1} { 2 }I\ Omega^2 \a pprox 1. 5\times10 ^ {53
rm eje}\sqrt{G_M_{\rm rem}R_{\rm_eje}}$, respectively, where $R_{\rm_torus}$ is_typical_radius of_torus,_$R_{\rm eje}$ is_the typical distance_from merger center to_where ejecta occur_and_$M_{\rm rem}$ is the gravitational mass of remnant core. For dynamical ejecta launched within_dynamical_timescale ($\leq_10~{\rm_ms}$),_$R_{\rm eje}$ can be estimated_as the typical radius of_the surrounding_torus at such an early time. The late-time_neutrino-driven_and magnetically-driven ejecta_mainly come from surrounding torus with $R_{\rm eje}=R_{\rm torus}$_[see also @Shibata2019] while others come_from the core_of_the_remnant with $R_{\rm eje}=R_{\rm_rem}$. Pre-collapse, a large amount of neutrinos_radiate from the hot, violent and_rapidly rotating remnant core and make it_cool down and spin down. @Baumgarte1998_estimated the angular momentum carried_away by_neutrinos as $J_{\rm \nu}\approx (2/3)c^{-2}R_{\rm_rem}^2\Omega E_\nu \approx2.2\times10^{48}\rm_erg~s(E_{\rm \nu}/{0.1M_\odot_c^2})(R_{\rm rem}/{13\rm km})^{2} (\Omega/{10^4\rm_rad/s})$, where $\Omega$ is the angular_velocity of the_rigidly rotating massive neutron star. The_equatorial_radius $R_{\rm rem}$_can_be_approximated as_$R_{\rm crit}=(1+0.032\zeta_{\rm TOV}^{-1.6}j^2+0.014\zeta_{\rm_TOV}^{-3.2}j^4)R_{\rm_TOV}$ (see_Appendix_for details). For neutron star rotating_in_the Keplerian limit with $R_{\rm TOV}$=10 km,_we have $R_{\rm rem}\approx_13$_km. The angular momentum carried_away by post-merger gravitational wave_can be expressed as $J_{\rm GW,p}=_J_0(1-\sqrt{1-E_{\rm GW,p}/E_0})$,_where the_rotating energy at the onset of merger can be approximated as_$E_0 = \frac{1}{2}I\Omega^2 \approx 1.5\times10^{53
the mean-squared of the spacetime distance tends to a universal constant $\langle\delta X_{\mu}^{2}\rangle/\langle\Delta X_{\mu}\rangle^{2}=2/\pi$ in the extreme IR region of theory. We also argue that this effect is testable by observing a linear dependence between the variance and mean-squared of redshifts from distant spectral lines. The proportionality is $\mathcal{O}(1)$ and expected to be identical to the percentage of the dark energy $\Omega_{\Lambda}$. These results strongly support the argument that the equivalence principle still holds at quantum level. It is in this sense we propose a possible “quantum theory of gravity”. This work was supported in part by the National Science Foundation of China (NSFC) under Grant No.11205149, and Science Research Foundation of Jiangsu University under Grant No.15JDG153. [29]{} natexlab\#1[\#1]{}bibnamefont \#1[\#1]{}bibfnamefont \#1[\#1]{}citenamefont \#1[\#1]{}url \#1[`#1`]{}urlprefix\[2\][\#2]{} \[2\]\[\][[\#2](#2)]{} , ****, (). , ****, (),. , ****, (),. , ****, (),. , ****, (),. ,,,,,,,,,,, (),. , ****, (). , ****, (). , ****, (). ,,,,, ****, (). , ****, (). ,,,,,, ****, (),. , ****, (),. , ** (, ). , ****, (). , ****, (). , ** (, ). , ** (, ). , ****, (). , in ** (, ). ,,, ****, (). , ****, (). , ** (, ). (), ****, (),. (), ****, (),. , (),. , ****, (),. , ****, (). , ****, (),. --- abstract: 'Using [*ROSAT*]{} observations, we estimate gas pressures in the X-ray-emitting medium surrounding 63 FRII radio galaxies and quasars. We compare these pressures with the internal pressures of the radio-emitting plasma estimated by assuming minimum energy or equipartition. In the majority of cases (including 12/13 sources with modelled, spatially resolved X-ray emission) radio sources appear to be [*
the mean - squared of the spacetime distance tends to a universal constant $ \langle\delta X_{\mu}^{2}\rangle/\langle\Delta X_{\mu}\rangle^{2}=2/\pi$ in the extreme IR region of hypothesis. We besides argue that this impression is testable by observing a analogue dependence between the variability and mean - squared of red shift from distant spectral lines. The proportion is $ \mathcal{O}(1)$ and expect to be identical to the percentage of the colored energy $ \Omega_{\Lambda}$. These results powerfully support the argument that the equivalence principle still holds at quantum level. It is in this common sense we propose a possible “ quantum theory of graveness ”. This work was supported in part by the National Science Foundation of China (NSFC) under Grant No.11205149, and Science Research Foundation of Jiangsu University under Grant No.15JDG153. [ 29 ] { } natexlab\#1[\#1]{}bibnamefont \#1[\#1]{}bibfnamefont \#1[\#1]{}citenamefont \#1[\#1]{}url \#1[`#1`]{}urlprefix\[2\][\#2 ] { } \[2\]\[\][[\#2](#2) ] { } , * * * *, (). , * * * *, (), . , * * * *, (), . , * * * *, (), . , * * * *, (), . , ,, ,, ,, ,, ,, (), . , * * * *, (). , * * * *, (). , * * * *, (). , ,, ,, * * * *, (). , * * * *, (). , ,, ,, , * * * *, (), . , * * * *, (), . , * * (,). , * * * *, (). , * * * *, (). , * * (,). , * * (,). , * * * *, (). , in * * (,). , ,, * * * *, (). , * * * *, (). , * * (,). (), * * * *, (), . (), * * * *, (), . , (), . , * * * *, (), . , * * * *, (). , * * * *, (), . --- abstract:' Using [ * ROSAT * ] { } observations, we calculate gas imperativeness in the X - ray - emitting culture medium surrounding 63 FRII radio galaxies and quasars. We compare these pressures with the internal pressures of the radio - emitting plasma estimated by assuming minimum energy or equipartition. In the majority of cases (including 12/13 sources with modelled, spatially decide X - ray discharge) radio receiver reservoir appear to be [ *
thf mean-squared of the spagetime distance jebds to a unibersal cunstant $\langle\delta X_{\mu}^{2}\ranglx/\lantle\Deota X_{\mu}\rangle^{2}=2/\pi$ in the extreme PR region of uheory. We also arjhe that this snfect ms testable by pbserving d linear depengevcz between the variance and mean-squarqd of rrddhifts from diftanu s[ectdal lines. The proportionality is $\mzthcal{O}(1)$ and expected to be identical to the pegcenhage of the dark ejergy $\Omega_{\Oambqq}$. These resuuts strongly support tge argument that the equivalencd priuciple stilo yolfv at quantun levvl. It is in tmps sensa we prppose a possibke “xuabtum theory of gravitb”. This work was suppotted in past by the National Wcuence Foutdatkin uf Dhmna (NSFC) kndxr Grant No.11205149, and Sciencw Research Foundatipn if Jiangsu Unjversiey under Grant No.15JDG153. [29]{} natexlab\#1[\#1]{}bibnamefont \#1[\#1]{}tibrnamefont \#1[\#1]{}citenamefont \#1[\#1]{}yrl \#1[`#1`]{}urlprefix\[2\][\#2]{} \[2\]\[\][[\#2](#2)]{} , ****, (). , ****, (),. , ****, (),. , ****, (),. , ****, (),. ,,,,,,,,,,, (),. , ****, (). , ****, (). , ****, (). ,,,,, ****, (). , ****, (). ,,,,,, ****, (),. , ****, (),. , ** (, ). , ****, (). , ****, (). , ** (, ). , ** (, ). , ****, (). , in ** (, ). ,,, ****, (). , ****, (). , ** (, ). (), ****, (),. (), ****, (),. , (),. , ****, (),. , ****, (). , ****, (),. --- dbstrecg: 'Uwikg [*RUWAH*]{} observations, we estimate gas pressures in trs C-rsy-emitting medlum surrounding 63 FTIL twdio galaxies and qbzszrs. We compare thede presfures with the intrrnal pressures of the radii-emitting plcsmq estimated by assbming minimuo engrgy ot equipartition. In the oajodity of casfs (includjvg 12/13 sources with mocenled, spaukally resolved X-rwy emissiin) rcdio soufces appeaw to be [*
the mean-squared of the spacetime distance tends universal $\langle\delta X_{\mu}^{2}\rangle/\langle\Delta in the extreme also that this effect testable by observing linear dependence between the variance and of redshifts from distant spectral lines. The proportionality is $\mathcal{O}(1)$ and expected to identical to the percentage of the dark energy $\Omega_{\Lambda}$. These results strongly support argument the principle holds at quantum level. It is in this sense we propose a possible “quantum theory of This work was supported in part by the Science Foundation of China under Grant No.11205149, and Science Foundation Jiangsu University Grant [29]{} \#1[\#1]{}bibfnamefont \#1[\#1]{}citenamefont \#1[\#1]{}url \[2\]\[\][[\#2](#2)]{} , ****, (). , ****, (),. , ****, (),. , ****, (),. , ****, (),. ,,,,,,,,,,, , ****, ****, (). ****, ,,,,, (). , ****, ****, (),. , ****, (),. , , ****, (). , ****, (). , ** ). , (, ). , ****, (). , ** (, ). ,,, ****, (). , ****, , ** (, ). (), ****, (),. (), ****, (),. , (),. , ****, (),. (). , ****, (),. abstract: 'Using [*ROSAT*]{} we gas in X-ray-emitting medium 63 FRII radio galaxies and quasars. We compare these pressures with internal pressures of the radio-emitting plasma estimated by assuming minimum equipartition. the majority of (including 12/13 sources with spatially X-ray emission) radio sources be
the mean-squared of the spacetIme distancE tendS to A unIvErsaL conStant $\langle\delTA X_{\mu}^{2}\Rangle/\langle\Delta X_{\mu}\raNgle^{2}=2/\pI$ iN The eXTrEme IR Region oF ThEORy. WE aLsO arGuE ThAt thiS efFect is tEstable by oBseRvIng a linear dePEnDence betweEn tHe variance anD meAn-squaReD of REdshiFts From dIstant SPectraL lines. The PrOPortioNAlity is $\MAThCal{O}(1)$ And expected to be idENtICal to the percenTage of ThE DaRK EneRgy $\omega_{\LambdA}$. THese rESults stROnGLY SupPOrt the argumenT that the equIValEnce prInCipLE still Holds At QUanTum level. It iS in tHis sense wE propoSE a possiBLe “quantUm theoRy oF grAvitY”. thIs WorK wAS suPPoRteD In pArt by the naTiOnal SCienCE fOUndaTioN of CHina (NsFC) under Grant no.11205149, aNd ScIEncE ReseArch FOundAtIon of jiangsU UnivErSity under Grant NO.15JDG153. [29]{} Natexlab\#1[\#1]{}bIbnAmEfoNt \#1[\#1]{}BibfnAMefont \#1[\#1]{}CitEnaMefont \#1[\#1]{}uRl \#1[`#1`]{}urlprEFix\[2\][\#2]{} \[2\]\[\][[\#2](#2)]{} , ****, (). , ****, (),. , ****, (),. , ****, (),. , ****, (),. ,,,,,,,,,,, (),. , ****, (). , ****, (). , ****, (). ,,,,, ****, (). , ****, (). ,,,,,, ****, (),. , ****, (),. , ** (, ). , ****, (). , ****, (). , ** (, ). , ** (, ). , ****, (). , In ** (, ). ,,, ****, (). , ****, (). , ** (, ). (), ****, (),. (), ****, (),. , (),. , ****, (),. , ****, (). , ****, (),. --- ABSTrAct: 'Using [*ROSAT*]{} obserVaTIOnS, we estimAte gas PReSsURes in the x-rAy-eMittING mediUm suRRoUnding 63 FRiI radiO GaLaXies and QuAsars. WE cOmpAre These PRessUres wiTh the intErnal PRessures of the rADio-emitting plASmA EStIMateD by Assuming minImum ENergY or eQUiParTItion. in the MaJOrITy of cases (including 12/13 sOuRces wiTh modElled, spatiallY resolved X-RAY Emission) RadiO SoURces appear to be [*
the mean-squared of the s pacetime d istan ceten ds toa un iversal consta n t $\ langle\delta X_{\mu}^{ 2}\ra ng l e/\l a ng le\De lta X_{ \ mu } \ ran gl e^ {2} =2 / \p i$ in th e extre me IR regi onof theory. Wea ls o argue th atthis effectistestab le by obser vin g a l ineard epende nce betwe en the va r iance a n d m ean- squared of redshi f ts from distant s pectra ll in e s . T heproportion al ity i s $\math c al { O } (1) $ and expected to be iden t ica l to t he pe r centag e ofth e da rk energy $ \Ome ga_{\Lamb da}$.T hese re s ults st rongly su ppo rt t h ear gum en t th a tthe equ ivalence p ri ncipl e st i l l hold s a t qu antum level. It is in thi s se nse w e pro pose a poss ible “ quant um theory of grav ity” . This w ork w assu pport e d in p art by the Na tionalS cie nc e F ou ndation of China ( NS F C )under Gr ant No . 11 20 5 149, and S cie nceR e searc h Fo u nd ation of Jiang s uUn iversit yunderGr ant No .15JD G 153. [29] {} natex lab\# 1 [\#1]{}bibname f ont \#1[\#1]{ } bi b f na m efon t \ #1[\#1]{}ci tena m efon t \# 1 [\ #1] { }url\#1[` #1 ` ]{ } urlprefix\[2\][\#2] {} \[2\] \[\][ [\#2](#2)]{} , ****, ( ) . , ****, (), . , ****, (),. , **** , (),. ,* ***, (), . ,, ,,,,,,,, , (),. , * ***, (). , ** **, () . , ****, (). , , , ,, * ** *, (). ,****, ( ). ,, ,,, , * ** *, (),. , ****,() ,. ,**(, ). , ****, ( ). , ** **, ( ) . , * * (,). ,** (,). , * * ** , (). , i n ** (, ) . ,, , ** * *,(). ,****, (). , ** ( ,). (), * ***, (),. () ,****, (),. , ( ),. , * ***, (), . , ****, (). , ****, (),. - --abstr act: 'Using [ *RO SAT*]{ } o b servat ions,we es ti mat e gas p r e ss ure sin the X-r a y -em ittin gmedi um surr ounding 63 FRII ra d iogalaxies andqua sars . We co m pa r e t he s e p r e ssures with the internalpr e ss ures of th e ra di o-emitt ing pla sma e s timated by assum ing minim um ene r g y o r equipart ition. I n the maj o rityo fcases (i ncludi ng 12 /13 s ources wit h mod elled, s patial ly re so lved X-r ay emission) radio sour ces ap peartobe [*
the_mean-squared of_the spacetime distance tends_to a_universal_constant $\langle\delta_X_{\mu}^{2}\rangle/\langle\Delta_X_{\mu}\rangle^{2}=2/\pi$ in the_extreme IR region_of theory. We also_argue that this_effect_is testable by observing a linear dependence between the variance and mean-squared of redshifts_from_distant spectral_lines._The_proportionality is $\mathcal{O}(1)$ and expected_to be identical to the_percentage of_the dark energy $\Omega_{\Lambda}$. These results strongly support_the_argument that the_equivalence principle still holds at quantum level. It is_in this sense we propose a_possible “quantum theory_of_gravity”. This_work was supported in_part by the National Science Foundation_of China (NSFC) under Grant No.11205149,_and Science Research Foundation of Jiangsu University_under Grant No.15JDG153. [29]{} natexlab\#1[\#1]{}bibnamefont \#1[\#1]{}bibfnamefont \#1[\#1]{}citenamefont_\#1[\#1]{}url \#1[`#1`]{}urlprefix\[2\][\#2]{} \[2\]\[\][[\#2](#2)]{} , ****, (). ,_****, (),. ,_****, (),. , ****, (),. , ****,_(),. ,,,,,,,,,,, (),. , ****,_(). , ****,_(). , ****, (). ,,,,,_****, (). , ****, (). ,,,,,, ****, (),. ,_****, (),. , **_(, ). , ****, (). , ****, (). ,_**_(, ). , **_(,_). ,_****, (). ,_in ** (,_). ,,,_****, (). ,_****,_(). , ** (, ). (), ****, (),. (),_****,_(),. , (),. , ****, (),. , ****, (). , ****,_(),. --- abstract: 'Using [*ROSAT*]{}_observations,_we estimate gas pressures_in the X-ray-emitting medium surrounding_63 FRII radio galaxies and quasars._We compare_these pressures_with the internal pressures of the radio-emitting plasma estimated by assuming_minimum energy or equipartition. In the_majority of cases (including_12/13 sources_with_modelled, spatially resolved_X-ray_emission) radio_sources appear to be [*
These generate a downward flux in momentum space, but one which is distributed throughout the acceleration region. Combined with the fact that the size of the “box” or region normally increases with energy this also gives an additional loss process because particles can now fall through the back of the “box” as well as being advected out of it (see Fig. 1). Note that particles which fall through the front of the box are advected back into the acceleration region and thus this process does not work upstream. If the loss rate is $\dot p = -\alpha p^2$ the basic equation becomes + [p]{}=\ Q - U\_2 4p\^2 f(p) - 4p\^4 f(p) [dL\_2dp]{} This equation is easily generalised to the case of different loss rates upstream and downstream. Simplifying equation (12) gives L[ft]{} + p[fp]{} +\ f= [Q4p\^2]{}. Note that for convenience we have dropped the explicit vector (and tensor) notation; all non-scalar quantities are to be interpreted as normal components, that is $U_2$ means $\bf n\cdot U_2$ etc. Note also that our model differs from that of Protheroe and Stanev in that they do not allow for the extra loss process resulting from the energy dependence of the “box” size. In the steady state and away from the source region this gives immediately the remarkably simple result for the logarithmic slope of the spectrum, = -3[U\_1 - U\_2 - 3pL]{}. Note that at small values of $p$ we recover the standard result, that the power-law exponent is $-3 U_1/(U_1 - U_2)$. = = Under normal circumstances both $L_1$ and $L_2$ are monotonically increasing functions of $p$. Thus both the numerator and denominator of the above expression, regarded as functions of $p$, have single zeroes at which they change sign. The denominator goes to zero at the critical momentum p\^\* = [U\_1 - U\_23L]{} where the losses exactly balance the acceleration. If the numerator at this point is negative, the slope goes to $-\infty$ and there is no pile-up. However the slope goes to $+\infty$ and a pile-up occurs if U\_1 - 4 U\_2 + 3
These generate a downward flux in momentum space, but one which is distribute throughout the acceleration area. Combined with the fact that the size of the “ box ” or area normally increases with department of energy this also gives an extra loss process because particles can nowadays fall through the back of the “ corner ” as well as being advected out of it (see Fig.   1). Note that particle which fall through the front of the box are advected back into the acceleration region and therefore this process does not work upstream. If the loss pace is $ \dot p = -\alpha p^2 $ the basic equation becomes + [ p]{}=\ Q - U\_2 4p\^2 f(p) - 4p\^4 f(p) [ dL\_2dp ] { } This equation is easily generalize to the case of different loss rates upstream and downstream. Simplifying equality (12) gives L[ft ] { } + p[fp ] { } + \ f= [ Q4p\^2 ] { }. Note that for convenience we have dropped the explicit vector (and tensor) notation; all non - scalar quantity are to be interpreted as normal components, that is $ U_2 $ means $ \bf n\cdot U_2 $ etc. Note also that our model differs from that of Protheroe and Stanev in that they do not allow for the extra loss procedure resulting from the energy addiction of the “ corner ” size. In the steady state and away from the source area this gives immediately the unusually simple result for the logarithmic gradient of the spectrum, = -3[U\_1 - U\_2 - 3pL ] { }. Note that at small values of $ p$ we recover the standard result, that the power - law exponent is $ -3 U_1/(U_1 - U_2)$. = = Under normal circumstances both $ L_1 $ and $ L_2 $ are monotonically increasing functions of $ p$. Thus both the numerator and denominator of the above formula, regarded as function of $ p$, have single zeroes at which they change sign. The denominator run to zero at the critical momentum p\^\ * = [ U\_1 - U\_23L ] { } where the losses exactly balance the acceleration. If the numerator at this item is negative, the slope goes to $ -\infty$ and there is no atomic pile - up. However the slope fail to $ + \infty$ and a pile - up occurs if U\_1 - 4 U\_2 + 3
Thfse generate a downward nlux in momentum space, uut one which ir distributed throughout the axceleeation region. Combined with the fact thqt tie size of the “box” or rennon nkvmallv mncreases with gnergy this dlso gives an ddaicional loss process because particlef can npw fall through jhe bssk or the “box” as well as being advectes out oh it (see Fig. 1). Noye that particles which fapl tjrough the front ov the box ats aqcected back knto the abeeleration tegion and thus this process doer not work upsttzqm. Iv the loss rete is $\dot p = -\alpha p^2$ the tasic eauation becomex + [']{}=\ Q - Y\_2 4p\^2 f(p) - 4p\^4 f(p) [dL\_2dp]{} Thiv equation is easijy generaniaed to the case od eiffetent noss eatds hpxtdeam ajd vownstream. Aimplifying equation (12) gives L[fu]{} + [[dp]{} +\ f= [Q4p\^2]{}. Note fhat fjr convenience we have dropped the explicpt vsctor (and tensor) notatiin; all non-scalar quanjities are to be interpreted as normal components, that is $U_2$ meana $\bf n\gdot Y_2$ ftc. Note also that our model differs from thae og Irotheroe and Stakev in that they dp joy allow for thg extra loas process resultijg from the wnergy dekendemce of the “box” size. In the sready state cnd away from the souxce region tkis gifes ikmediately the remarkabuy sjmple resulh for the uogarithmic slopd on tve spectrum, = -3[U\_1 - U\_2 - 3pL]{}. Note that at wmalk valuer of $p$ we wecover thf stakgard result, that tje pocer-laf exponent is $-3 U_1/(U_1 - U_2)$. = = Under normal circumsteices both $L_1$ amd $L_2$ dre monoconicakly increasind functions of $p$. Thus both the numerztor anv denominatow of the abova expression, cegarded ws fynctuons of $o$, have single aeroes at which thet change sign. The venoojnator goes to vtro at the criticak mumegtlm '\^\* = [U\_1 - G\_23L]{} where the losres rxactuy balance bhe accrleration. If the numaratkr at this point ix kegative, jhe slope goes to $-\inftu$ and there is no kile-up. Hoxever yhe slope goes to $+\infty$ and a pils-up occurd in U\_1 - 4 U\_2 + 3
These generate a downward flux in momentum one is distributed the acceleration region. the of the “box” region normally increases energy this also gives an additional process because particles can now fall through the back of the “box” as as being advected out of it (see Fig. 1). Note that particles which through front the are advected back into the acceleration region and thus this process does not work upstream. If loss rate is $\dot p = -\alpha p^2$ basic equation becomes + Q - U\_2 4p\^2 f(p) 4p\^4 [dL\_2dp]{} This is generalised the case of loss rates upstream and downstream. Simplifying equation (12) gives L[ft]{} + p[fp]{} +\ f= [Q4p\^2]{}. Note that convenience we the explicit (and notation; non-scalar quantities are interpreted as normal components, that is n\cdot U_2$ etc. Note also that our model from that Protheroe and Stanev in that they not allow for the extra loss process resulting the energy dependence of the “box” size. In the steady state and away from the this gives immediately the simple result for logarithmic of spectrum, -3[U\_1 - - 3pL]{}. Note that at small values of $p$ we recover standard result, that the power-law exponent is $-3 U_1/(U_1 - = normal circumstances both and $L_2$ are monotonically functions $p$. Thus both the denominator the as of have single zeroes at they change sign. The denominator to zero at the - U\_23L]{} where the losses exactly balance the If the numerator at this point is the slope goes to $-\infty$ and there is no pile-up. However the goes to a pile-up occurs if U\_1 - 4 U\_2 3
These generate a downward fluX in momentuM spacE, buT onE wHich Is diStributed throuGHout The acceleration region. COmbinEd WIth tHE fAct thAt the siZE oF THe “bOx” Or RegIoN NoRmallY inCreases With energy ThiS aLso gives an adDItIonal loss pRocEss because paRtiCles caN nOw fALl thrOugH the bAck of tHE “box” as Well as beiNg ADvecteD Out of it (SEE FIg. 1). NoTe that particles whICh FAll through the fRont of ThE BoX ARe aDveCted back inTo The acCEleratiON rEGIOn aND thus this procEss does not wORk uPstreaM. IF thE Loss raTe is $\dOt P = -\AlpHa p^2$ the basic EquaTion becomEs + [p]{}=\ Q - U\_2 4p\^2 F(P) - 4p\^4 f(p) [dL\_2dP]{} this equAtion iS eaSilY genERaLiSed To THe cASe Of dIFfeRent loss RaTeS upstReam AND DOwnsTreAm. SiMplifYing equation (12) gIveS L[ft]{} + P[Fp]{} +\ f= [q4p\^2]{}. NotE that For cOnVenieNce we hAve drOpPed the explicit vEctoR (and tensoR) noTaTioN; aLl non-SCalar qUanTitIes are tO be inteRPreTeD AS NoRmal components, that Is $u_2$ MEaNs $\bf n\cdoT U_2$ etc. NOTe AlSO that our MoDel DiffERS from That OF PRotheroe And StaNEv In That theY dO not alLoW foR thE extrA Loss ProcesS resultiNg froM The energy depenDEnce of the “box” sIZe. iN ThE SteaDy sTate and away From THe soUrce REgIon THis giVes imMeDIaTEly the remarkably simPlE resulT for tHe logarithmic Slope of the SPECtrum, = -3[U\_1 - U\_2 - 3pl]{}. NotE ThAT at small values Of $p$ we Recover the STandard rEsult, That the pOwer-law exPONent is $-3 U_1/(U_1 - u_2)$. = = UnDer NorMal CIRcUmstances both $l_1$ ANd $L_2$ aRe MonotonIcaLly incrEasIng FunCtiOnS of $p$. Thus bOth the nuMeRaToR aNd dEnomiNAtor of thE aBovE eXprEssioN, RegardEd as fUnctIoNs OF $p$, hAve singLE zEROes aT wHiCh thEy cHaNge siGn. ThE DenOminatoR goes to zeRo aT The cRiTiCal momeNtum p\^\* = [U\_1 - U\_23L]{} where ThE losses exaCtLy bAlance THE accelerAtion. If the numerator at thIS point iS neGativE, the Slope goes To $-\iNfty$ anD thERe is no Pile-up. howevEr The SLOpe goES To $+\InfTy$ And a pile-up OCCurS if U\_1 - 4 U\_2 + 3
These generate a downward flux in m oment umspa ce , bu t on e which is dis t ribu ted throughout the acc elera ti o n re g io n. Co mbinedw it h the f ac t t ha t t he si zeof the“box” or r egi on normally in c re ases withene rgy this als o g ives a nadd i tiona l l oss p rocess becaus e particl es can no w fall t h r ou gh t he back of the “b o x” as well as bei ng adv ec t ed o utofit (see Fi g.  1).N ote tha t p a r t icl e s which fallthrough the fro nt ofth e b o x areadvec te d ba ck into the acc eleration regio n and th u s thisproces s d oes not wo rk up st r eam . Ift heloss rat eis $\do t p= - \ alph a p ^2$the b asic equation be come s +[p]{} =\ Q- U\ _2 4p\^ 2 f(p) - 4p \^ 4 f(p) [dL\_2dp ]{}This equa tio nisea silyg eneral ise d t o the c ase ofd iff er e n t l oss rates upstream a n d d ownstrea m. Sim p li fy i ng equat io n ( 12)g i ves L [ft] { }+ p[fp]{ } +\ f = [ Q4 p\^2]{} .Note t ha t f orconve n ienc e we h ave drop ped t h e explicit vec t or (and tenso r )n o ta t ion; al l non-scala r qu a ntit iesa re to be in terpr et e da s normal components ,that i s $U_ 2$ means $\bf n\cdot U_ 2 $ etc. Not e al s ot hat our modeldiffe rs from th a t of Pro thero e and St anev in t h a t they d o n otall owf o rthe extra los s proc es s resul tin g fromthe en erg y d ep endence o f the “b ox ”si ze . In th e steadyst ate a ndawayf rom th e sou rcere gi o n t his giv e si m medi at el y th e r em arkab ly s i mpl e resul t for the lo g arit hm ic slopeof the spectr um , = -3[U\_ 1- U \_2 -3 p L]{}. No te that at small values of $p$werecov er t he standa rdresult , t h at the power -lawex pon e n t is$ - 3U_1 /( U_1 - U_2) $ . = = Un dernormalcircumstances both $L_ 1$ and $L_2$are mon o t on ica l ly inc re a sin g functions of $p $. Thus bo th th e numerato r an ddenomin ator of thea bove ex pression, regarded a s fu n c tio ns of $p$, have si ngle zero e s atw hi ch th eychange s ign . The denom i nat or go es toze ro atthe c ri tical mo mentum p\^\* = [U\_1 -U\_23L ]{} w her e the los ses exa ctly bala ncethe accele rat ion . Ifthe numer ator at th i s poi nt i s negative , t hes l op e goes to $ - \ i nft y$ an d t h ere is nopile-up. Howevert he slope goesto $ + \ inf ty$ andapile-up occurs if U \ _ 1 - 4 U\ _2 + 3
These_generate a_downward flux in momentum_space, but_one_which is_distributed_throughout the acceleration_region. Combined with_the fact that the_size of the_“box”_or region normally increases with energy this also gives an additional loss process because_particles_can now_fall_through_the back of the “box”_as well as being advected_out of_it (see Fig. 1). Note that particles which fall_through_the front of_the box are advected back into the acceleration region_and thus this process does not_work upstream. If the_loss_rate_is $\dot p =_-\alpha p^2$ the basic equation becomes_+ [p]{}=\ Q - U\_2 4p\^2 f(p)_- 4p\^4 f(p) [dL\_2dp]{} This equation is_easily generalised to the case of_different loss rates upstream and_downstream. Simplifying_equation (12) gives L[ft]{} +_p[fp]{} +\ f= [Q4p\^2]{}._Note that_for convenience we_have dropped the explicit vector (and_tensor) notation; all_non-scalar quantities are to be interpreted_as_normal components, that_is_$U_2$_means $\bf_n\cdot U_2$ etc._Note_also that_our_model differs from that of Protheroe_and_Stanev in that they do not allow_for the extra loss_process_resulting from the energy_dependence of the “box” size. In_the steady state and away from_the source_region this_gives immediately the remarkably simple result for the logarithmic slope of_the spectrum, = -3[U\_1 - U\_2_- 3pL]{}. Note that_at small_values_of $p$ we_recover_the standard_result, that the power-law exponent is $-3_U_1/(U_1 -_U_2)$. = = Under normal circumstances both $L_1$ and_$L_2$ are monotonically increasing_functions_of $p$. Thus both the numerator_and denominator of the above expression,_regarded as functions of $p$,_have_single_zeroes at which they change_sign. The denominator goes to zero_at the critical_momentum p\^\* = [U\_1 - U\_23L]{} where_the_losses exactly balance the acceleration. If_the_numerator at this point is negative,_the_slope_goes to $-\infty$ and there_is no pile-up. However the slope_goes to $+\infty$ and a pile-up occurs if U\_1_- 4 U\_2_+ 3
{\mathcal{O}}}_0'$-homomorphism $$\label{eq_proof_weierstrass_div_map} {\ensuremath{\mathcal{O}}}_0^F\ni f\mapsto (f_0',\ldots,f'_{d-1})\in({\ensuremath{\mathcal{O}}}_0^{\prime F})^{\oplus d}$$ descends to an isomorphism $$\label{eq_proof_weierstrass_iso} {\ensuremath{\mathcal{O}}}_0^F/h{\ensuremath{\mathcal{O}}}_0^F\xrightarrow{\approx}({\ensuremath{\mathcal{O}}}_0^{\prime F})^{\oplus d}$$ of ${\ensuremath{\mathcal{O}}}_0'$-modules. Composing this with the embedding $$\label{eq_proof_weierstrass_map} ({\ensuremath{\mathcal{O}}}_0^{\prime F})^{\oplus d}\ni (f'_j)\mapsto \sum_j f_j'z_m^j\in{\ensuremath{\mathcal{O}}}_0^F,$$ we obtain an ${\ensuremath{\mathcal{O}}}_0'$-homomorphism $$\label{eq_proof_weierstrass_right_inverse} {\ensuremath{\mathcal{O}}}_0^F/h{\ensuremath{\mathcal{O}}}_0^F\rightarrow{\ensuremath{\mathcal{O}}}_0^F,$$ which is a right inverse of the canonical projection ${\ensuremath{\mathcal{O}}}_0^F\rightarrow{\ensuremath{\mathcal{O}}}_0^F/h{\ensuremath{\mathcal{O}}}_0^F$. Now $p:{\ensuremath{\mathcal{O}}}^n_0\rightarrow M$ and $\phi:{\ensuremath{\mathcal{O}}}_0^E\rightarrow M$ of the theorem induce ${\ensuremath{\mathcal{O}}}_0'$-homomorphisms $$p':{\ensuremath{\mathcal{O}}}_0^n/h{\ensuremath{\mathcal{O}}}_0^n\rightarrow M,\text{ } \phi':{\ensuremath{\mathcal{O}}}^E_0/h{\ensuremath{\mathcal{O}}}^E_0\rightarrow M,$$ as in, remembering that $hM=0$. Clearly, $p'$ is surjective. Also, by Lemma \[lemma\_depth\_lemmas\_ind\_on\_dim\] ${\mathop{\mathrm{depth}}\nolimits _{{{\mathop{\mathfrak{m}}}}'}} M>0$. Because of the isomorphism, $(T_{m-1})$ implies
{ \mathcal{O}}}_0'$-homomorphism $ $ \label{eq_proof_weierstrass_div_map } { \ensuremath{\mathcal{O}}}_0^F\ni f\mapsto (f_0',\ldots, f'_{d-1})\in({\ensuremath{\mathcal{O}}}_0^{\prime F})^{\oplus d}$$ descends to an isomorphism $ $ \label{eq_proof_weierstrass_iso } { \ensuremath{\mathcal{O}}}_0^F / h{\ensuremath{\mathcal{O}}}_0^F\xrightarrow{\approx}({\ensuremath{\mathcal{O}}}_0^{\prime F})^{\oplus d}$$ of $ { \ensuremath{\mathcal{O}}}_0'$-modules. Composing this with the embedding $ $ \label{eq_proof_weierstrass_map } ({ \ensuremath{\mathcal{O}}}_0^{\prime F})^{\oplus d}\ni (f'_j)\mapsto \sum_j f_j'z_m^j\in{\ensuremath{\mathcal{O}}}_0^F,$$ we obtain an $ { \ensuremath{\mathcal{O}}}_0'$-homomorphism $ $ \label{eq_proof_weierstrass_right_inverse } { \ensuremath{\mathcal{O}}}_0^F / h{\ensuremath{\mathcal{O}}}_0^F\rightarrow{\ensuremath{\mathcal{O}}}_0^F,$$ which is a proper inverse of the basic projection $ { \ensuremath{\mathcal{O}}}_0^F\rightarrow{\ensuremath{\mathcal{O}}}_0^F / h{\ensuremath{\mathcal{O}}}_0^F$. Now $ p:{\ensuremath{\mathcal{O}}}^n_0\rightarrow M$ and $ \phi:{\ensuremath{\mathcal{O}}}_0^E\rightarrow M$ of the theorem induce $ { \ensuremath{\mathcal{O}}}_0'$-homomorphisms $ $ p':{\ensuremath{\mathcal{O}}}_0^n / h{\ensuremath{\mathcal{O}}}_0^n\rightarrow M,\text { } \phi':{\ensuremath{\mathcal{O}}}^E_0 / h{\ensuremath{\mathcal{O}}}^E_0\rightarrow M,$$ as in, remembering that $ hM=0$. distinctly, $ p'$ is surjective. Also, by Lemma \[lemma\_depth\_lemmas\_ind\_on\_dim\ ] $ { \mathop{\mathrm{depth}}\nolimits _ { { { \mathop{\mathfrak{m } } } }' } } M>0$. Because of the isomorphism, $ (T_{m-1})$ implies
{\matjcal{O}}}_0'$-homomorphism $$\label{ed_proof_weierstrass_div_ma'} {\ensurejath{\mathzal{O}}}_0^F\ni f\mapsto (f_0',\ldots,f'_{d-1})\in({\endueematy{\mathcal{O}}}_0^{\prime F})^{\oplus a}$$ descendd to an usomiephism $$\labxm{eq_proon_ceierabrass_nsi} {\ensuremath{\matmcal{O}}}_0^F/h{\ensusemath{\mathcal{O}}}_0^X\xfiyhtarrow{\approx}({\ensuremath{\mathcal{O}}}_0^{\primq F})^{\oplux f}$$ of ${\ensurematr{\matnsal{O}}}_0'$-jodules. Composing this with the emgedding $$\label{eq_proof_eeierstrass_map} ({\ensuremath{\mahhcap{O}}}_0^{\prime F})^{\oplus d}\ni (f'_j)\mapsto \soj_j s_h'z_m^j\in{\ensureoath{\mathcal{O}}}_0^F,$$ we obtaih an ${\ensuremath{\mathcal{O}}}_0'$-homomorpfism $$\kabel{eq_proif_qeiftstrass_right_mnversv} {\ensuremath{\mabncal{O}}}_0^F/v{\ensurekath{\mathcal{O}}}_0^F\rlghtacrow{\wnsuremath{\mathcal{O}}}_0^F,$$ wiich is a right invetse of the ccnonical projection ${\ebsyremajh{\matvcal{U}}}_0^D\riehtzrcow{\snsurelati{\mathcal{O}}}_0^F/h{\snsuremath{\mqthcal{O}}}_0^F$. Now $p:{\ensurekaey{\mathcal{O}}}^n_0\riggtarror I$ and $\phi:{\ensuremath{\mathcal{O}}}_0^E\rightarrow K$ or the theorem induce ${\enwuremath{\mathcal{O}}}_0'$-homomlrphisms $$[':{\ensuremath{\mathcal{O}}}_0^n/h{\ensuremath{\mathcal{O}}}_0^n\rightarrof M,\teet{ } \ihi':{\evwugemath{\mathcal{O}}}^E_0/h{\ensuremath{\mathcal{O}}}^E_0\rightarrow J,$$ ss in, rememberinn that $hM=0$. Clearly, $l'$ ls furjective. Alro, by Lemja \[lemma\_depth\_lemmad\_ind\_on\_dym\] ${\marhop{\mathri{depyh}}\nolimits _{{{\mathop{\mathfrak{m}}}}'}} N>0$. Because of rhe isomorphism, $(T_{m-1})$ implies
{\mathcal{O}}}_0'$-homomorphism $$\label{eq_proof_weierstrass_div_map} {\ensuremath{\mathcal{O}}}_0^F\ni f\mapsto (f_0',\ldots,f'_{d-1})\in({\ensuremath{\mathcal{O}}}_0^{\prime F})^{\oplus d}$$ an $$\label{eq_proof_weierstrass_iso} {\ensuremath{\mathcal{O}}}_0^F/h{\ensuremath{\mathcal{O}}}_0^F\xrightarrow{\approx}({\ensuremath{\mathcal{O}}}_0^{\prime d}$$ of ${\ensuremath{\mathcal{O}}}_0'$-modules. $$\label{eq_proof_weierstrass_map} F})^{\oplus d}\ni (f'_j)\mapsto f_j'z_m^j\in{\ensuremath{\mathcal{O}}}_0^F,$$ we obtain ${\ensuremath{\mathcal{O}}}_0'$-homomorphism $$\label{eq_proof_weierstrass_right_inverse} {\ensuremath{\mathcal{O}}}_0^F/h{\ensuremath{\mathcal{O}}}_0^F\rightarrow{\ensuremath{\mathcal{O}}}_0^F,$$ which is a inverse of the canonical projection ${\ensuremath{\mathcal{O}}}_0^F\rightarrow{\ensuremath{\mathcal{O}}}_0^F/h{\ensuremath{\mathcal{O}}}_0^F$. Now $p:{\ensuremath{\mathcal{O}}}^n_0\rightarrow M$ and $\phi:{\ensuremath{\mathcal{O}}}_0^E\rightarrow M$ of theorem induce ${\ensuremath{\mathcal{O}}}_0'$-homomorphisms $$p':{\ensuremath{\mathcal{O}}}_0^n/h{\ensuremath{\mathcal{O}}}_0^n\rightarrow M,\text{ } \phi':{\ensuremath{\mathcal{O}}}^E_0/h{\ensuremath{\mathcal{O}}}^E_0\rightarrow M,$$ as in, remembering that $hM=0$. $p'$ surjective. by \[lemma\_depth\_lemmas\_ind\_on\_dim\] ${\mathop{\mathrm{depth}}\nolimits _{{{\mathop{\mathfrak{m}}}}'}} M>0$. Because of the isomorphism, $(T_{m-1})$ implies
{\mathcal{O}}}_0'$-homomorphism $$\label{Eq_proof_weiErstrAss_Div_MaP} {\ensUremAth{\mathcal{O}}}_0^F\ni F\MapsTo (f_0',\ldots,f'_{d-1})\in({\ensuremath{\MathcAl{o}}}_0^{\PrimE f})^{\oPlus d}$$ DescendS To AN IsoMoRpHisM $$\lABeL{eq_prOof_WeierstRass_iso} {\ensUreMaTh{\mathcal{O}}}_0^F/h{\ENsUremath{\matHcaL{O}}}_0^F\xrightarrOw{\aPprox}({\eNsUreMAth{\maThcAl{O}}}_0^{\prIme F})^{\opLUs d}$$ of ${\eNsuremath{\MaTHcal{O}}}_0'$-mODules. CoMPOsIng tHis with the embeddiNG $$\lABel{eq_proof_weieRstrasS_mAP} ({\eNSUreMatH{\mathcal{O}}}_0^{\pRiMe F})^{\opLUs d}\ni (f'_j)\MApSTO \Sum_J F_j'z_m^j\in{\ensureMath{\mathcal{o}}}_0^f,$$ we Obtain An ${\EnsURemath{\MathcAl{o}}}_0'$-HomOmorphism $$\laBel{eQ_proof_weiErstraSS_right_iNVerse} {\enSuremaTh{\mAthCal{O}}}_0^f/H{\eNsUreMaTH{\maTHcAl{O}}}_0^f\RigHtarrow{\eNsUrEmath{\MathCAL{o}}}_0^f,$$ whiCh iS a riGht inVerse of the canOniCal pROjeCtion ${\EnsurEmatH{\mAthcaL{O}}}_0^F\rigHtarrOw{\Ensuremath{\mathcAl{O}}}_0^F/H{\ensuremaTh{\mAtHcaL{O}}}_0^f$. Now $p:{\ENsuremAth{\MatHcal{O}}}^n_0\rIghtarrOW M$ aNd $\PHI:{\EnSuremath{\mathcal{O}}}_0^E\rIgHTArRow M$ of thE theorEM iNdUCe ${\ensureMaTh{\mAthcAL{o}}}_0'$-homoMorpHIsMs $$p':{\ensurEmath{\mAThCaL{O}}}_0^n/h{\ensUrEmath{\mAtHcaL{O}}}_0^n\RightARrow m,\text{ } \pHi':{\ensureMath{\mAThcal{O}}}^E_0/h{\ensureMAth{\mathcal{O}}}^E_0\rIGhTARrOW M,$$ as In, rEmembering tHat $hm=0$. cleaRly, $p'$ IS sUrjECtive. also, bY LEMmA \[Lemma\_depth\_lemmas\_ind\_On\_Dim\] ${\matHop{\maThrm{depth}}\noliMits _{{{\mathop{\MATHfrak{m}}}}'}} M>0$. BEcauSE oF The isomorphism, $(t_{m-1})$ impLies
{\mathcal{O}}}_0'$-homomor phism $$\l abel{ eq_ pro of _wei erst rass_div_map}{ \ens uremath{\mathcal{O}}}_ 0^F\n if \map s to (f_0 ',\ldot s ,f ' _ {d- 1} )\ in( {\ e ns urema th{ \mathca l{O}}}_0^{ \pr im e F})^{\oplu s d }$$ descen dsto an isomor phi sm $$\ la bel { eq_pr oof _weie rstras s _iso}{\ensurem at h {\math c al{O}}} _ 0 ^F /h{\ ensuremath{\mathc a l{ O }}}_0^F\xright arrow{ \a p pr o x }({ \en suremath{\ ma thcal { O}}}_0^ { \p r i m e F } )^{\oplus d}$ $ of ${\ens u rem ath{\m at hca l {O}}}_ 0'$-m od u les . Composing thi s with th e embe d ding $$ \ label{e q_proo f_w eie rstr a ss _m ap} ( { \en s ur ema t h{\ mathcal{ O} }} _0^{\ prim e F } )^{\ opl us d }\ni(f'_j)\mapsto \s um_j f_j 'z_m^ j\in{ \ens ur emath {\math cal{O }} }_0^F,$$ we obt ainan ${\ens ure ma th{ \m athca l {O}}}_ 0'$ -ho momorph ism $$\ l abe l{ e q _ pr oof_weierstrass_ri gh t _ in verse} { \ensur e ma th { \mathcal {O }}} _0^F / h {\ens urem a th {\mathca l{O}}} _ 0^ F\ rightar ro w{\ens ur ema th{ \math c al{O }}}_0^ F,$$ whi ch is a right invers e of the canon i ca l pr o ject ion ${\ensurem ath{ \ math cal{ O }} }_0 ^ F\rig htarr ow { \e n suremath{\mathcal{O }} }_0^F/ h{\en suremath{\mat hcal{O}}}_ 0 ^ F $. Now$p:{ \ en s uremath{\mathc al{O} }}^n_0\rig h tarrow M $ and $\phi:{ \ensurema t h {\mathca l{O }}} _0^ E\r i g ht arrow M$ of t h e the or em indu ce${\ensu rem ath {\m ath ca l{O}}}_0' $-homomo rp hi sm s$$p ':{\e n suremath {\ mat hc al{ O}}}_ 0 ^n/h{\ ensur emat h{ \m a thc al{O}}} _ 0^ n \ righ ta rr ow M ,\t ex t{ } \ phi ':{\ens uremath{\ mat h cal{ O} }} ^E_0/h{ \ensuremath{\ ma thcal{O}}} ^E _0\ righta r r ow M,$$as in, remembering that $hM=0$. Cl early , $p '$ is sur jec tive.Als o , by L emma \ [lemm a\ _de p t h\_le m m as \_i nd \_on\_dim\ ] ${\ matho p{ \mat hrm{dep th}}\nolimits _{{{ \ mat hop{\mathfrak {m} }}}' } } M >0$ . B e cau se oft h e isomorphism,$(T_{m-1}) $i mp lies
{\mathcal{O}}}_0'$-homomorphism $$\label{eq_proof_weierstrass_div_map} {\ensuremath{\mathcal{O}}}_0^F\ni_f\mapsto (f_0',\ldots,f'_{d-1})\in({\ensuremath{\mathcal{O}}}_0^{\prime_F})^{\oplus d}$$ descends to_an isomorphism_$$\label{eq_proof_weierstrass_iso} {\ensuremath{\mathcal{O}}}_0^F/h{\ensuremath{\mathcal{O}}}_0^F\xrightarrow{\approx}({\ensuremath{\mathcal{O}}}_0^{\prime_F})^{\oplus d}$$_of_${\ensuremath{\mathcal{O}}}_0'$-modules. Composing this_with the embedding_$$\label{eq_proof_weierstrass_map} ({\ensuremath{\mathcal{O}}}_0^{\prime F})^{\oplus d}\ni (f'_j)\mapsto_\sum_j f_j'z_m^j\in{\ensuremath{\mathcal{O}}}_0^F,$$ we_obtain_an ${\ensuremath{\mathcal{O}}}_0'$-homomorphism $$\label{eq_proof_weierstrass_right_inverse} {\ensuremath{\mathcal{O}}}_0^F/h{\ensuremath{\mathcal{O}}}_0^F\rightarrow{\ensuremath{\mathcal{O}}}_0^F,$$ which is a right inverse of the canonical projection ${\ensuremath{\mathcal{O}}}_0^F\rightarrow{\ensuremath{\mathcal{O}}}_0^F/h{\ensuremath{\mathcal{O}}}_0^F$. Now $p:{\ensuremath{\mathcal{O}}}^n_0\rightarrow_M$_and $\phi:{\ensuremath{\mathcal{O}}}_0^E\rightarrow_M$_of_the theorem induce ${\ensuremath{\mathcal{O}}}_0'$-homomorphisms $$p':{\ensuremath{\mathcal{O}}}_0^n/h{\ensuremath{\mathcal{O}}}_0^n\rightarrow_M,\text{ } _\phi':{\ensuremath{\mathcal{O}}}^E_0/h{\ensuremath{\mathcal{O}}}^E_0\rightarrow M,$$_as in, remembering that $hM=0$. Clearly, $p'$ is_surjective._Also, by Lemma_\[lemma\_depth\_lemmas\_ind\_on\_dim\] ${\mathop{\mathrm{depth}}\nolimits _{{{\mathop{\mathfrak{m}}}}'}} M>0$. Because of the isomorphism, $(T_{m-1})$_implies
of three coincident counters and two pairs of iron cores $A'$, $A''$ and $B'$, $B''$. The field in the cores is parallel to the axis of the counters and has opposite directions in $A'$, $A''$ (similarly in $B'$, $B''$). The magnetic field is closed by iron bars applied at both ends of $A'$, $A''$ and $B'$, $B''$. Thus each pair of iron cores acts like a cylindrical magnetic convergent (“$c$”) or divergent (“$d$”) lens for positive ($c+$ or $d+$) or negative ($c-$ or $d-$) particles (cf. [@Bernardini_et_al1945], p. 111).](Fig10C.jpg) Working in Rome and at Pian Rosà (Cervinia), at 3460 meters above sea level, the Rome group found out in 1941 “a conclusive evidence in favour of the existence of a positive excess” and immediately planned further experiments, as the method “owing to its simplicity, seems well-suited for an investigation \[…\] under conditions when the elaborate Wilson chamber technique is impossible” (see [@BernardiniWick_et_al1941a], p. 536). They performed more experiments in the years 1941-1943 and published several papers (see in particular [@BernardiniConversi1940; @BernardiniWick_et_al1941a; @BernardiniWick_et-al1941b; @ConversiScrocco1943]). In an outline of their research work, published in the *Physical Review* in 1945 [@Bernardini_et_al1945], they explained that they had investigated the energy spectrum and positive excess of the hard component of cosmic rays for “cogent reasons”, as “the positive excess in the meson spectrum is probably connected with the positive nature of the primary radiation” and “a study of the variation of the positive excess with height would probably be interesting and might throw some light on the process of creation of the mesons” (see [@Bernardini_et_al1945], p. 110). Both in Rome and at Pian Rosà, the Rome physicists had used the same experimental apparatus, i.e. “a counter system with deflecting magnetised cores” (see [@Bernardini_et_al1945], p. 109, and fig. 10c). The latter, which were similar to the ones used by Rossi
of three coincident counters and two pairs of iron cores $ A'$, $ A''$ and $ B'$, $ B''$. The field in the core is parallel to the bloc of the counters and has diametric directions in $ A'$, $ A''$ (similarly in $ B'$, $ B''$). The charismatic field is closed by iron cake apply at both ends of $ A'$, $ A''$ and $ B'$, $ B''$. Thus each pair of iron cores act like a cylindrical magnetic convergent (“ $ c$ ”) or divergent (“ $ d$ ”) lens for positive ($ c+$ or $ d+$) or minus ($ c-$ or $ d-$) particle (cf. [ @Bernardini_et_al1945 ], p. 111).](Fig10C.jpg) Working in Rome and at Pian Rosà (Cervinia), at 3460 meters above sea level, the Rome group found out in 1941 “ a conclusive evidence in favor of the existence of a positive excess ” and immediately design further experiments, as the method “ owing to its simplicity, look well - suited for an investigation \[ … \ ] under conditions when the elaborate Wilson chamber proficiency is impossible ” (see [ @BernardiniWick_et_al1941a ], p. 536). They performed more experiments in the years 1941 - 1943 and published several newspaper (see in particular [ @BernardiniConversi1940; @BernardiniWick_et_al1941a; @BernardiniWick_et - al1941b; @ConversiScrocco1943 ]). In an outline of their research work, published in the * Physical Review * in 1945 [ @Bernardini_et_al1945 ], they explained that they had investigated the energy spectrum and positive excess of the difficult component of cosmic rays for “ cogent reason ”, as “ the incontrovertible excess in the meson spectrum is probably connected with the positive nature of the primary radiation ” and “ a study of the mutant of the positive excess with height would probably be interesting and might throw some light on the summons of creation of the mesons ” (see [ @Bernardini_et_al1945 ], p. 110). Both in Rome and at Pian Rosà, the Rome physicists had use the like experimental apparatus, i.e. “ a counterpunch system with deflecting magnetised cores ” (determine [ @Bernardini_et_al1945 ], p. 109, and fig. 10c). The latter, which were similar to the ones used by Rossi
of three coincident countevs and two pairs of iroi cores $A'$, $A''$ and $B'$, $B''$. The field in the cores ms pqralltj to the axis of tfe countegs and haw opkosite directions in $A'$, $A''$ (similadpy iu $U'$, $B''$). The magnetig field is wlosed by iron bxrd applied at both ends of $A'$, $A''$ and $B'$, $B''$. Thus ewch pair of irjn cpwes zbtw like a cylindrical magnetic converjent (“$c$”) or diverbent (“$d$”) lens for positive ($c+$ or $f+$) or negative ($c-$ or $d-$) particlew (cf. [@Vernardini_et_xl1945], p. 111).](Fig10C.jkg) Corking in Tome and at Pian Rosà (Cervinia), at 3460 metzrs above sgc leggl, the Rome jroup sound out in 1941 “a concnusive rvidence in fayour mf rhe existence of a povitive excess” and ymmediateny planned further wxperimgnts, ds tfw mdthkd “osing tl ivs simplicify, seems weol-suited for an invtstytation \[…\] under condieijns when the elaborate Wilson chamber ttchnisue is impossible” (see [@BwrnardiniWick_et_al1941a], p. 536). They persormed more experiments in the years 1941-1943 and publishad setefal pwowrd (see in particular [@BernardiniConversi1940; @BernarqjnoWpck_et_al1941a; @BernardikiWick_et-al1941b; @ConverxiDctjcco1943]). In an oujline oy tgeir research work, publisred ib the *Phyficak Review* in 1945 [@Bernardini_et_al1945], they explaiued that they had invzstigated thz enerby sprctrum and positive exczss of the hard clmponent kw cosmic rays fof “cpgant reasons”, as “the positivq excess mn thz meson rpecjrum is probably fonnegded with the positlve ncture of the prlmary radiation” and “a study of vie variation pf thv positivz excexs with heighe would probabky be iuteresging and mpght throx some light on the procevd of creatioi of the iesobs” (swe [@Bernxfdini_et_al1945], p. 110). Bpth in Rome and at Pian Rosà, the Rome ihysizjsts had used tkt sqme experimentak aopawanus, i.q. “a counter sfsteo wkyh dewlecting mannegisec cores” (see [@Bernarditi_et_zl1945], p. 109, and fig. 10c). Thr jatter, wyich werq similar to yhe ones used by Rlssi
of three coincident counters and two pairs cores $A''$ and $B''$. The field to axis of the and has opposite in $A'$, $A''$ (similarly in $B'$, The magnetic field is closed by iron bars applied at both ends of $A''$ and $B'$, $B''$. Thus each pair of iron cores acts like a magnetic (“$c$”) divergent lens for positive ($c+$ or $d+$) or negative ($c-$ or $d-$) particles (cf. [@Bernardini_et_al1945], p. 111).](Fig10C.jpg) in Rome and at Pian Rosà (Cervinia), at meters above sea level, Rome group found out in “a evidence in of existence a positive excess” immediately planned further experiments, as the method “owing to its simplicity, seems well-suited for an investigation \[…\] conditions when Wilson chamber is (see p. 536). They experiments in the years 1941-1943 and (see in particular [@BernardiniConversi1940; @BernardiniWick_et_al1941a; @BernardiniWick_et-al1941b; @ConversiScrocco1943]). In outline of research work, published in the *Physical in 1945 [@Bernardini_et_al1945], they explained that they had the energy spectrum and positive excess of the hard component of cosmic rays for “cogent “the positive excess in meson spectrum is connected the nature the primary and “a study of the variation of the positive excess with would probably be interesting and might throw some light on of of the mesons” [@Bernardini_et_al1945], p. 110). Both Rome at Pian Rosà, the had the i.e. counter with deflecting magnetised cores” [@Bernardini_et_al1945], p. 109, and fig. The latter, which were by Rossi
of three coincident counters And two pairS of irOn cOreS $A'$, $a''$ and $b'$, $B''$. ThE field in the corES is pArallel to the axis of the cOunteRs ANd haS OpPositE directIOnS IN $A'$, $A''$ (SiMiLarLy IN $B'$, $b''$). The mAgnEtic fieLd is closed By iRoN bars applied AT bOth ends of $A'$, $a''$ anD $B'$, $B''$. Thus each pAir Of iron CoRes ACts liKe a CylinDrical MAgnetiC convergeNt (“$C$”) Or diveRGent (“$d$”) leNS FoR posItive ($c+$ or $d+$) or negatiVE ($c-$ OR $d-$) particles (cf. [@BErnardInI_Et_AL1945], P. 111).](FiG10C.jPg) Working iN ROme anD At Pian ROSà (cERVinIA), at 3460 meters abovE sea level, thE romE group FoUnd OUt in 1941 “a cOncluSiVE evIdence in favOur oF the existEnce of A PositivE Excess” aNd immeDiaTelY plaNNeD fUrtHeR ExpERiMenTS, as The methoD “oWiNg to iTs siMPLICity, SeeMs weLl-suiTed for an invesTigAtioN \[…\] UndEr conDitioNs whEn The elAboratE WilsOn Chamber techniquE is iMpossible” (See [@beRnaRdIniWiCK_et_al1941a], P. 536). ThEy pErformeD more exPEriMeNTS In The years 1941-1943 and publishEd SEVeRal paperS (see in PArTiCUlar [@BernArDinIConVERsi1940; @BeRnarDInIWick_et_aL1941a; @BernARdInIWick_et-Al1941B; @ConveRsIScRocCo1943]). In aN OutlIne of tHeir reseArch wORk, published in tHE *Physical ReviEW* iN 1945 [@bErNArdiNi_eT_al1945], they explAineD That They HAd InvEStigaTed thE eNErGY spectrum and positivE eXcess oF the hArd component oF cosmic rayS FOR “cogent rEasoNS”, aS “The positive excEss in The meson spECtrum is pRobabLy connecTed with thE POsitive nAtuRe oF thE prIMArY radiation” and “A STudy Of The variAtiOn of the PosItiVe eXceSs With heighT would prObAbLy Be IntErestINg and migHt ThrOw SomE lighT On the pRocesS of cReAtIOn oF the mesONs” (SEE [@BerNaRdIni_eT_al1945], P. 110). BOth in rome ANd aT Pian RoSà, the Rome PhySIcisTs HaD used thE same experimeNtAl apparatuS, i.E. “a cOunter SYStem with Deflecting magnetised corES” (see [@BerNarDini_eT_al1945], p. 109, And fig. 10c). ThE laTter, whIch WEre simIlar to The onEs UseD BY RossI
of three coincident count ers and tw o pai rsofir on c ores $A'$, $A''$ a n d $B '$, $B''$. The field i n the c o resi sparal lel tot he a xis o fthe c o un tersand has op posite dir ect io ns in $A'$,$ A' '$ (simila rly in $B'$, $B ''$ ). The m agn e tic f iel d isclosed by iro n bars ap pl i ed atb oth end s of $A' $, $A''$ and $B'$ , $ B ''$. Thus each pairof ir o n co res acts like a cyli n dricalm ag n e t icc onvergent (“$ c$”) or div e rge nt (“$ d$ ”)l ens fo r pos it i ve($c+$ or $d +$)or negati ve ($c - $ or $d - $) part icles(cf . [ @Ber n ar di ni_ et _ al1 9 45 ],p . 1 11).](Fi g1 0C .jpg) Wo r k i n g in Ro me a nd at Pian Rosà (C erv inia ) , a t 346 0 met ersab ove s ea lev el, t he Rome group fou nd o ut in 194 1 “ acon cl usive eviden ceinfavourof thee xis te n c e o f a positive exces s” a nd immedia tely p l an ne d further e xpe rime n t s, as the me thod “ow ing to it ssimplic it y, see ms we ll- suite d for an in vestigat ion \ [ …\] under cond i tions when th e e l a bo r ateWil son chamber tec h niqu e is im pos s ible” (see [ @ Be r nardiniWick_et_al19 41 a], p. 536) . They perfor med more e x p e rimentsin t h ey ears 1941-1943 andpublisheds everal p apers (see in particul a r [@Berna rdi niC onv ers i 1 94 0; @Bernardin i W ick_ et _al1941 a;@Bernar din iWi ck_ et- al 1941b; @C onversiS cr oc co 19 43] ). In an outli ne of t hei r res e arch w ork,publ is he d in the *P h ys i c al R ev ie w* i n 1 94 5 [@B erna r din i_et_al 1945], th eye xpla in ed that t hey had inves ti gated theen erg y spec t r um and p ositive excess of the h a rd comp one nt of cos mic raysfor “coge ntr easons ”, as“thepo sit i v e exc e s sinth e meson sp e c tru m ispr obab ly conn ected with the pos i tiv e nature of t heprim a r yrad i at i on” a n d “ a study of the va riation of t h epositive e x ces swith he ight wo uld p r obablybe intere sting and m ight t hro w some lig ht on th e process of cr e at ion o f t he mes on s”(see[@Bern a rdi ni_et _al194 5] , p. 1 10).Bo th in Ro me and at Pian Rosà, th e Rome phys ici sts had u sed the same exp erim ental appa rat us, i.e. “a count er s y st emw ith d efle c ting magn e ti sed c or es” (see [@ B e r nar dini_ et_ a l1945] , p. 109, and fig. 10 c ). The latter, whi c h we res imil ar to the ones u sed b y Rossi
of_three coincident_counters and two pairs_of iron_cores_$A'$, $A''$_and_$B'$, $B''$. The_field in the_cores is parallel to_the axis of_the_counters and has opposite directions in $A'$, $A''$ (similarly in $B'$, $B''$). The magnetic_field_is closed_by_iron_bars applied at both ends_of $A'$, $A''$ and $B'$,_$B''$. Thus_each pair of iron cores acts like a_cylindrical_magnetic convergent (“$c$”)_or divergent (“$d$”) lens for positive ($c+$ or $d+$)_or negative ($c-$ or $d-$) particles_(cf. [@Bernardini_et_al1945], p._111).](Fig10C.jpg) Working_in_Rome and at Pian_Rosà (Cervinia), at 3460 meters above_sea level, the Rome group found_out in 1941 “a conclusive evidence in_favour of the existence of a_positive excess” and immediately planned_further experiments,_as the method “owing to_its simplicity, seems_well-suited for_an investigation \[…\]_under conditions when the elaborate Wilson_chamber technique is_impossible” (see [@BernardiniWick_et_al1941a], p. 536). They_performed_more experiments in_the_years_1941-1943 and_published several papers_(see_in particular_[@BernardiniConversi1940;_@BernardiniWick_et_al1941a; @BernardiniWick_et-al1941b; @ConversiScrocco1943]). In an outline_of_their research work, published in the *Physical_Review* in 1945 [@Bernardini_et_al1945],_they_explained that they had_investigated the energy spectrum and_positive excess of the hard component_of cosmic_rays for_“cogent reasons”, as “the positive excess in the meson spectrum is_probably connected with the positive nature_of the primary radiation”_and “a_study_of the variation_of_the positive_excess with height would probably be interesting_and might_throw some light on the process_of creation of the_mesons”_(see [@Bernardini_et_al1945], p. 110). Both in_Rome and at Pian Rosà, the_Rome physicists had used the_same_experimental_apparatus, i.e. “a counter system_with deflecting magnetised cores” (see [@Bernardini_et_al1945],_p. 109, and_fig. 10c). The latter, which were similar_to_the ones used by Rossi
and content.) Summary tables of spectral analyses for the 1-s peak spectra ------------------------------------------------------------ In this section, we present the tables for the 1-s peak spectra, which are the same set of the tables as those in Section \[sect:T100\_spectra\_tables\] but for the 1-s peak spectral fits. Column Format Unit Description ------------------ -------- ------ --------------------------------------------------------------------- GRBname A9 – GRB name. Trig\_ID I11 – Special ID associated with the trigger. For GRBs found in ground analysis, the ID are associated with the trigger ID of the failed event data, or the full observation ID of the event data for the analysis. Best\_fit\_model A3 – The best-fit model for the GRB spectrum, either the simple power law model (PL) or cutoff power-law model (CPL.) : \[tab:spec\_1speak\_best\_model\] The format of the table that summarizes the best-fit spectral model of the 1-s peak spectrum for each burst. (This table is available in its entirety in a machine-readable form in the online journal. Only the column formats are shown here for guidance regarding its form and content.) Column Format Unit Description ---------------- -------- ---------------------------------------- -------------------------------------------------------------------------- GRBname A9 – GRB name. Trig\_ID I11 – Special ID associated with each triggers. For GRBs found in ground analysis, the ID are associated with the trigger ID of the failed event data, or the full observation ID of the event data for the analysis. alpha A13 – $\alpha^{\rm PL}$ as defined in Eq. \[eq:PL\]. alpha\_low A13 – The lower limit of $\alpha^{\rm PL}$. alpha\_hi A13 – The upper limit of $\alpha^{\rm PL}$. norm A12 $\rm ph \ cm^{-2} \ s^{-1} \ keV^{-1}$ The normalization factor $K^{\rm PL}_{50}$, as defined in Eq
and content .) Summary tables of spectral analyses for the 1 - s acme spectra ------------------------------------------------------------ In this incision, we present the tables for the 1 - s peak spectrum, which are the same set of the mesa as those in Section   \[sect: T100\_spectra\_tables\ ] but for the 1 - s point spectral fit. Column Format Unit Description ------------------ -------- ------ --------------------------------------------------------------------- GRBname A9 – GRB name. Trig\_ID I11 – Special ID associated with the trigger. For GRBs find in ground psychoanalysis, the ID are associated with the trigger ID of the failed event data, or the full observation ID of the consequence data for the analysis. Best\_fit\_model A3 – The best - fit mannequin for the GRB spectrum, either the simple power jurisprudence model (PL) or shortcut power - police model (CPL .) : \[tab: spec\_1speak\_best\_model\ ] The format of the table that summarizes the well - fit spectral model of the 1 - s peak spectrum for each burst. (This table is available in its entirety in a car - clear phase in the online journal. Only the column formats are shown here for guidance involve its kind and content .) Column Format Unit Description ---------------- -------- ---------------------------------------- -------------------------------------------------------------------------- GRBname A9 – GRB name. Trig\_ID I11 – Special ID associated with each trigger. For GRBs found in ground analysis, the ID are associated with the gun trigger ID of the failed consequence data, or the full observation ID of the consequence data for the analysis. alpha A13 – $ \alpha^{\rm PL}$ as defined in Eq.   \[eq: PL\ ]. alpha\_low A13 – The lower limit of $ \alpha^{\rm PL}$. alpha\_hi A13 – The upper limit of $ \alpha^{\rm PL}$. average A12 $ \rm ph \ cm^{-2 } \ s^{-1 } \ keV^{-1}$ The normalization factor $ K^{\rm PL}_{50}$, as defined in Eq
anf content.) Summary tables uf spectral analyses fmr the 1-s peak rpectra ------------------------------------------------------------ In this section, we prxsenr the tables for the 1-s peak spectra, ahich arw tht same set of the tables as thoac in Vxction \[sect:T100\_specjra\_tables\] bud for the 1-s pedk s'ectral fits. Column Fowmat Inlt Description ------------------ -------- ------ --------------------------------------------------------------------- GDBname A9 – GRB name. Trig\_ID I11 – Speciam ID associated with the triggef. Fos GRBs gound in grounc aialywis, the ID drz associated with the teigget ID mf tfw fxiltd xveht datw, or trv full observztion YD of the event data for the analysis. Btst\_fif\_model A3 – The best-vit model for the GRB spectrum, either the simple power law model (PL) or cutoff power-law mldel (CPJ.) : \[rab:spec\_1sptak\_bext\_model\] The format of the tqble that suinarizes the best-fic spectral mudel of tne 1-s peak spectrum for zach bhrst. (This twble is abxilable in its evtigety in a machine-readable form in the oiline journau. Onky the column fogmats are shown here for guidcnce segarding lts form and content.) Column Formaj Unit Vescription ---------------- -------- ---------------------------------------- -------------------------------------------------------------------------- GRBtwme E9 – GRC name. Trig\_NB I11 – Sprcial ID associated fith each triggers. For GRBs fkund in ggoukd analysis, thg ID are associated wntm the triggec ID os the fained event data, or the full observation ID of the event data for the qnalysms. alphz A13 – $\aupha^{\rm PN}$ cs defined in Eq. \[eq:PL\]. alphs\_luw A13 – The lpwer limit of $\alpha^{\rm PL}$. alpya\_hi A13 – The uppeg likic of $\alpha^{\rm PL}$. norm A12 $\rm ph \ cm^{-2} \ s^{-1} \ kxV^{-1}$ The norlxlizajion facjor $K^{\rm PL}_{50}$, as defjned ln Eq
and content.) Summary tables of spectral analyses 1-s spectra ------------------------------------------------------------ this section, we 1-s spectra, which are same set of tables as those in Section \[sect:T100\_spectra\_tables\] for the 1-s peak spectral fits. Column Format Unit Description ------------------ -------- ------ GRBname A9 – GRB name. Trig\_ID I11 – Special ID associated with the For found ground the ID are associated with the trigger ID of the failed event data, or the full ID of the event data for the analysis. A3 – The best-fit for the GRB spectrum, either simple law model or power-law (CPL.) : \[tab:spec\_1speak\_best\_model\] format of the table that summarizes the best-fit spectral model of the 1-s peak spectrum for each (This table in its in machine-readable in the online the column formats are shown here its form and content.) Column Format Unit Description -------- ---------------------------------------- GRBname A9 – GRB name. Trig\_ID – Special ID associated with each triggers. For found in ground analysis, the ID are associated with the trigger ID of the failed or the full observation of the event for analysis. A13 $\alpha^{\rm PL}$ defined in Eq. \[eq:PL\]. alpha\_low A13 – The lower limit of PL}$. alpha\_hi A13 – The upper limit of $\alpha^{\rm PL}$. $\rm \ cm^{-2} \ \ keV^{-1}$ The normalization $K^{\rm as defined in Eq
and content.) Summary tables of Spectral anAlyseS foR thE 1-s Peak SpecTra ------------------------------------------------------------ In this sectiON, we pResent the tables for the 1-s Peak sPeCTra, wHIcH are tHe same sET oF THe tAbLeS as ThOSe In SecTioN \[sect:T100\_sPectra\_tablEs\] bUt For the 1-s peak sPEcTral fits. CoLumN Format Unit DEscRiptioN ------------------ -------- ------ --------------------------------------------------------------------- GrBnAMe A9 – GRb naMe. TriG\_ID I11 – SpECial ID AssociateD wITh the tRIgger. FoR grBS fouNd in ground analysiS, ThE iD are associateD with tHe TRiGGEr Id of The failed eVeNt datA, Or the fuLL oBSERvaTIon ID of the eveNt data for thE AnaLysis. BEsT\_fiT\_Model A3 – the beSt-FIt mOdel for the GrB spEctrum, eitHer the SImple poWEr law moDel (PL) oR cuTofF powER-lAw ModEl (cpL.) : \[tAB:sPec\_1SPeaK\_best\_modEl\] thE formAt of THE TAble ThaT sumMarizEs the best-fit sPecTral MOdeL of thE 1-s peaK speCtRum foR each bUrst. (THiS table is availabLe in Its entireTy iN a MacHiNe-reaDAble foRm iN thE online Journal. oNly ThE COLuMn formats are shown hErE FOr Guidance RegardINg ItS Form and cOnTenT.) ColUMN FormAt UnIT DEscriptiOn ---------------- -------- ---------------------------------------- -------------------------------------------------------------------------- GRBnAMe a9 – GrB name. TRiG\_ID I11 – SpEcIal iD aSsociATed wIth eacH triggerS. For Grbs found in grounD Analysis, the ID ARe ASSoCIateD wiTh the triggeR ID oF The fAileD EvEnt DAta, or The fuLl OBsERvation ID of the event DaTa for tHe anaLysis. alpha A13 – $\alPha^{\rm PL}$ as dEFINed in Eq. \[eQ:PL\]. aLPhA\_Low A13 – The lower liMit of $\Alpha^{\rm PL}$. aLPha\_hi A13 – ThE uppeR limit of $\Alpha^{\rm PL}$. NORm A12 $\rm ph \ cM^{-2} \ s^{-1} \ kEV^{-1}$ THe nOrmALIzAtion factor $K^{\rM pl}_{50}$, as dEfIned in EQ
and content.) Summary ta bles of sp ectra l a nal ys es f or t he 1-s peak sp e ctra --------------------- ----- -- - ---- - -- ----- ------- - -- - - --- -- -- I nt hi s sec tio n, we p resent the ta bl es for the 1 - speak spect ra, which are t hesame s et of the t abl es as those in Sec tion \[se ct : T100\_ s pectra\ _ t ab les\ ] but for the 1-s pe a k spectral fit s. C olu mn For ma t U n it Descript i on ---- -- --- - ------ -- -- -- - --- ------ --- ---- --------- ------ - ------- - ------- ------ --- --- ---- - -- -- --- -- - --- - G RBname A 9 – GR B na me. Tri g\ _ID I11 – Specia l I Dass oc iated with t hetri gger. For GR Bs found in g rou nd a n a lysis , t he ID ar e ass o ciat ed wit h the tr igger ID of the fail e d event data, o r t h e ful l obs er v at i on ID of the eventda ta for theanalysis. Best\_fit\ _ m o del A3 – The best-f i t modelfor t he GRB s pectrum, ei the r the s imp lepow erla w model ( PL) o r cu to ff pow er-lawm od e l (CP L. ) : \ [t ab:sp ec\_ 1 spe ak\_bes t\_model\ ] T h e fo rm at of the table that s um marizes th ebes t-fits p ectral m odel of the 1-s peak sp e ctrum f oreachburs t. (Thistab le isava i lablein its enti re tyi n a ma c h in e-r ea dable form i n t he on li ne j ournal. Only the column f o rma ts are shownher e fo r gu ida n ce reg ar d ing i ts form and con tent.) Column F or mat Unit Descri p tion - ----- --- ------ ---- ----- ----- - --- ----- ------ -- ------ ----- -- ----- -- ----------------------- ------ ----- --- --------- --- - --- --------- ---- ------ GRB name A 9 – GRB name. Trig\_ID I11 – Spec ia l ID a ssocia ted wi th each t ri g gers. ForG RBs foun din grou n d an alysis, t he ID a re ass oci ated w itht h e tri gger I D o f the fai l e de ve nt dat a, or t he full ob s erva t io n ID o f theeventdata fo r th eanalysi s. alph a A13 – $ \alpha ^{\ r m PL } $ as d ef ined in Eq.\[eq: PL \ ]. a lph a\_low A1 3 – The lowe r lim it of $\ a lpha^{\rm PL} $. alpha\_ h i A13 – Th e up pe rlimit of$\ alpha^{ \ rm P L }$. no rm A 12 $ \rm p h\ cm^{-2} \ s^{-1 } \ke V^{-1}$ The normalizati on fac tor $K ^ {\r m P L } _{50}$, as d efined inEq
and_content.) Summary tables_of spectral analyses for_the 1-s_peak_spectra ------------------------------------------------------------ In this_section,_we present the_tables for the_1-s peak spectra, which_are the same_set_of the tables as those in Section \[sect:T100\_spectra\_tables\] but for the 1-s peak spectral fits. __ ___ Column _ _Format _ Unit __ _ _ _ ___Description ------------------ --------_------ --------------------------------------------------------------------- _ GRBname _ A9_ –_ _ _ _ _ _ _ _ _ GRB name. __ ___Trig\_ID _ __ _I11_ –__ _ __ _ Special ID associated with_the trigger. _ _ _ _ _ _ __ __ _ _ _ _For GRBs found in_ground_analysis, _ _ ___ _ _ _ the ID_are_associated with the trigger ID of_the_failed event data, ___ _ _ _ _ ___ or the full observation ID of_the event_data for the_analysis. Best\_fit\_model A3 _ – _ _ The best-fit model for_the GRB spectrum, _ __ ____ _ _ _ _ _ _ __ either the simple power law model_(PL) _ __ _ _ _ __ _ _ _ _ _ _ _ or_cutoff power-law model (CPL.) : _\[tab:spec\_1speak\_best\_model\]_The format of the table that summarizes the best-fit_spectral_model_of_the 1-s peak_spectrum for each_burst. (This table_is available in_its_entirety in a machine-readable form in_the online journal._Only the column formats_are shown here for guidance regarding its form and content.) _ Column Format _ _ _ __ _ Unit_ __ ___ __ _____ _ Description __---------------- -------- ---------------------------------------- -------------------------------------------------------------------------- _ __ GRBname _ _ A9 __ ____ _ – __ _ __ _ _ _ _ GRB name. Trig\_ID _ I11 _ _ __ _ _ – _ _ _ _ _ Special ID associated with each triggers. _ _ __ _ _ _ __ _ _ _ _ _ _ _ For GRBs found in ground analysis, _ __ _ _ _ ___ __ _ _ _ _ the_ID are associated with_the trigger ID of the_failed event data, __ _ __ _ __ _ _ __ _ __ _ __ _ _ _ __ _ _ or the_full observation_ID of the event_data for the analysis. _ _ __ alpha _ A13 _ __ _ – _ _ _ _ _ ___ _ $\alpha^{\rm PL}$ as_defined in Eq. \[eq:PL\]. _ __ alpha\_low _ _ A13 _ __ __ _ _– _ _ ___ _ __ _The lower limit of_$\alpha^{\rm PL}$. _ alpha\_hi A13 ___ __ _ _–_ _ __ ___ _ The upper limit_of $\alpha^{\rm PL}$. _ _ _ norm _ A12 $\rm ph \ cm^{-2} \ s^{-1} \ keV^{-1}$ _ The normalization factor $K^{\rm_PL}_{50}$, as defined in Eq
.3 28.7 40.9 49.2 43.3 36.8 real-a 98.93 91.34 84.25 82.43 97.31 69.27 75.51 80.70 33.9 **12.5** **12.7** **9.1** **19.8** 31.0 17.1 11.5 (ResNet) Ours **99.37** **94.69** **94.56** **93.35** **99.10** **93.04** **91.49** **91.64** **18.5** **12.6** **13.0** **10.3** 29.7 **10.9** **11.0** **8.9** ---------- -------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ----------------------------------------------------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- --------- Model Loss SG LS SM BA SG LS SM BA SG LS SM BA SG LS SM BA w/o Eq. \[eq:3dagan\_recon\] \[eq:3dagan\_attradv\] 97.97 87.92 84.62 83.65 97.93 86.21 81.11 82.21 20.2 **10.6** **7.8** 14.1 43.8 20.4 27.2 18.3 w/o Eq. \[eq:3dagan\_recon\] 99.28 92.95 93.17 **94.86** 98.87 90.79 89.50 **93.82** **17.6** 17.5 13.8 11.9 **26.6** 18.1 15.0 11.4 (ResNet) w/o Eq. \[eq:3dagan\_attr
.3 28.7 40.9 49.2 43.3 36.8 real - a 98.93 91.34 84.25 82.43 97.31 69.27 75.51 80.70 33.9 * * 12.5 * * * * 12.7 * * * * 9.1 * * * * 19.8 * * 31.0 17.1 11.5 (ResNet) Ours * * 99.37 * * * * 94.69 * * * * 94.56 * * * * 93.35 * * * * 99.10 * * * * 93.04 * * * * 91.49 * * * * 91.64 * * * * 18.5 * * * * 12.6 * * * * 13.0 * * * * 10.3 * * 29.7 * * 10.9 * * * * 11.0 * * * * 8.9 * * ---------- -------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ----------------------------------------------------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- --------- Model Loss SG LS SM BA SG LS SM BA SG LS SM BA SG LS SM BA w/o Eq.   \[eq:3dagan\_recon\ ]   \[eq:3dagan\_attradv\ ] 97.97 87.92 84.62 83.65 97.93 86.21 81.11 82.21 20.2 * * 10.6 * * * * 7.8 * * 14.1 43.8 20.4 27.2 18.3 w/o Eq.   \[eq:3dagan\_recon\ ] 99.28 92.95 93.17 * * 94.86 * * 98.87 90.79 89.50 * * 93.82 * * * * 17.6 * * 17.5 13.8 11.9 * * 26.6 * * 18.1 15.0 11.4 (ResNet) w/o Eq.   \[eq:3dagan\_attr
.3 28.7 40.9 49.2 43.3 36.8 ceal-a 98.93 91.34 84.25 82.43 97.31 69.27 75.51 80.70 33.9 **12.5** **12.7** **9.1** **19.8** 31.0 17.1 11.5 (RxaNet) Ours **99.37** **94.69** **94.56** **93.35** **99.10** **93.04** **91.49** **91.64** **18.5** **12.6** **13.0** **10.3** 29.7 **10.9** **11.0** **8.9** ---------- -------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ----------------------------------------------------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- --------- Model Loss SG LS SM BA SG OS SM BA SG LS SM BC SG JS SM BZ w/o Fq. \[eq:3dagwn\_recin\] \[eq:3dagan\_wttrsdv\] 97.97 87.92 84.62 83.65 97.93 86.21 81.11 82.21 20.2 **10.6** **7.8** 14.1 43.8 20.4 27.2 18.3 w/o Eq. \[eq:3dagan\_fecpn\] 99.28 92.95 93.17 **94.86** 98.87 90.79 89.50 **93.82** **17.6** 17.5 13.8 11.9 **26.6** 18.1 15.0 11.4 (ResNet) w/o Eq. \[eq:3dagai\_ettr
.3 28.7 40.9 49.2 43.3 36.8 real-a 84.25 97.31 69.27 80.70 33.9 **12.5** 11.5 Ours **99.37** **94.69** **93.35** **99.10** **93.04** **91.64** **18.5** **12.6** **13.0** **10.3** 29.7 **11.0** **8.9** ---------- -------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ----------------------------------------------------- ----------- ----------- ----------- ----------- ----------- ----------- ---------- ---------- ---------- ---------- ---------- --------- Model Loss SG LS SM BA SG LS SM BA SG LS BA SG LS SM BA w/o Eq. \[eq:3dagan\_recon\] 97.97 87.92 84.62 83.65 86.21 81.11 82.21 20.2 **10.6** 14.1 20.4 27.2 w/o \[eq:3dagan\_recon\] 92.95 93.17 **94.86** 90.79 89.50 **93.82** **17.6** 17.5 13.8 11.9 **26.6** 18.1 15.0 11.4 (ResNet) w/o Eq. \[eq:3dagan\_attr
.3 28.7 40.9 49.2 43.3 36.8 real-a 98.93 91.34 84.25 82.43 97.31 69.27 75.51 80.70 33.9 **12.5** **12.7** **9.1** **19.8** 31.0 17.1 11.5 (ResNet) Ours **99.37** **94.69** **94.56** **93.35** **99.10** **93.04** **91.49** **91.64** **18.5** **12.6** **13.0** **10.3** 29.7 **10.9** **11.0** **8.9** ---------- -------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ----------------------------------------------------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- --------- Model Loss Sg LS SM BA SG Ls SM BA sG Ls SM bA sG LS sM BA W/o Eq. \[eq:3dagan\_recON\] \[eq:3dAgan\_attradv\] 97.97 87.92 84.62 83.65 97.93 86.21 81.11 82.21 20.2 **10.6** **7.8** 14.1 43.8 20.4 27.2 18.3 w/o Eq. \[eq:3dagan\_Recon\] 99.28 92.95 93.17 **94.86** 98.87 90.79 89.50 **93.82** **17.6** 17.5 13.8 11.9 **26.6** 18.1 15.0 11.4 (reSnet) w/O eq. \[Eq:3dagAn\_attr
.3 28.7 40.9 49.2 43 .3 36.8 real -a 98.93 91. 34 84. 2 5 82.43 97 .3 1 69 .27 75.51 80. 70 33.9 ** 12.5** * *12 .7** **9.1 ** **1 9. 8** 3 1.0 17.1 11.5 ( Re s Net) Ours ** 99.3 7** **94.69** ** 9 4.56** **93. 35** * * 99 . 1 0** **93.04** **91. 4 9** * * 91 . 6 4 ** **18.5** * *12.6** * * 13. 0** ** 10. 3 ** 29.7 * *10.9** * *11. 0** **8 .9** ------- - -- ---- ---- - --- --- ---- -- -- --- -- - - - - -- --- - --- ------- -- -- ---- ---- - - - ---- --- ---- ---- ------- ----- --- ---- --- ----- - --- ---- -- - --- ------ - --- -- ----- --------- - -- ----------- -- --- -- ---- - ----- --- ------- ------ - --- -- - - - -- ------------------ -- - - -- -------- ---- - - -- -- - ---- --- -- --- ---- - ----- ---- -- -------- - ---- - -- -- -- ---- -- ------- --- --- --- - - ---- ------------- -- -- - ------- ------ - --- --------- - - - - -- - ---- -- -------- -- ---- - ------- - -- -- Mo del L o s s SG LS SM BA SG LS SM BA SG L S SM BA SG LS SM B A w/o Eq. \[e q:3dagan\ _ recon \ ]\[eq: 3da gan\_a tt rad v\] 97. 9 7 87.92 84 .62 83.65 97.93 86.2 1 81. 11 82. 21 20.2 ** 10.6** * *7. 8** 14 . 1 4 3 .8 20 .4 27.2 1 8 . 3 w /o Eq.  \[e q:3dagan\_recon\] 99 . 2 8 92. 95 93.17 ** 9 4 .86** 98.87 90.79 89. 50 **93. 82** * * 17 . 6** 17 .5 13. 8 11.9 * *26.6* * 18.1 1 5 .0 11.4 (ResNe t) w /o Eq .\[eq:3d a gan\ _attr
.3 _ _ 28.7_ __ __40.9 _ _ 49.2 _ __43.3 36.8 __ ___ real-a _ 98.93 _ _ 91.34 __84.25 _ 82.43 _ 97.31 _ _69.27__ _ 75.51 _ 80.70 _ 33.9 _ **12.5** **12.7** _ **9.1** _**19.8** _ 31.0 _ _ _17.1 _ 11.5 _ (ResNet) _ Ours _**99.37**_ **94.69**___**94.56** _ **93.35** __**99.10** __**93.04** **91.49** _**91.64**_ **18.5** **12.6** _ **13.0** _**10.3**_ _29.7 _**10.9** **11.0** _**8.9** _---------- --------_----------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- ---------- ---------- ----------_---------- ---------- ---------- ---------- ---------- _---------- ----------------------------------------------------- ----------- -----------_----------- -----------_-----------_----------- ----------- -----------_----------_---------- ----------_---------- ---------- ---------- ---------- --------- _ _ _ __ _ _ ___ _ _ _ __ __ ___ _ _ _ _ ___ _ _ _ _ _ _ _ _ __ ____ _ _ _ _ _ _ __ _ _ __ _ _ _ __ _ _ _ _ _ _ _ _ __ ____ _ _ _ __ _ _ _ _ _ _ _ __ _ _ __ ___ __ _____ _ __ _ __ _ _ __ ____ _ __ _ __ _ Model _ _ _ Loss _ _ _ __ _ _ SG _ LS SM _ _ BA _ _ SG _ LS _ SM __ BA _ SG_ _ _LS_ _ _ SM_ _ BA _ _ SG _LS SM BA _ __ w/o Eq. \[eq:3dagan\_recon\] \[eq:3dagan\_attradv\] _97.97 _ 87.92 _ 84.62 ___83.65 __ 97.93 _ _86.21 _ 81.11 _ _ 82.21_ _ 20.2 _**10.6**_ **7.8** _ _14.1_ _ __ 43.8 _ _ __20.4 _ __ _27.2 __ 18.3 _ _ _ __ _ w/o Eq. \[eq:3dagan\_recon\] _ _ _ _99.28 _ _ 92.95__ 93.17 _ **94.86** 98.87 _ __ 90.79 89.50 _ **93.82**_ **17.6** _ 17.5_ _ _ 13.8 ___11.9 **26.6** _ 18.1 _ 15.0_ __ 11.4 _ _(ResNet) _ __ w/o Eq. \[eq:3dagan\_attr
conductor between the antenna complex and the reaction center. The FMO protein is a trimer where the three identical subunits consist of seven chromophoric sites which are fully characterized [@BioCh2005]. Typical transport times for excitons through FMO are of the order of picoseconds. Recent experiments [@BioEn2007] on the FMO complex show that the exciton transport exhibits quantum coherent oscillations up to $660$ femtoseconds. This raises the important question why the strong environmental fluctuations are insufficient to fully decohere the quantum transport. Simplifying the problem to a donor-acceptor model level it was shown [@BioTh2008; @Flemming2] that the slowness of the environment, namely $\omega_c\simeq\Delta$, sufficiently slows down decoherence for quantum oscillations to survive even for strong coupling to the environment. Having understood why quantum coherence survives rather strong couplings to environmental fluctuations raises the subsequent question whether photosynthesis benefits from quantum coherence either to improve efficiency or to allow regulation. Detailed descriptions of transport through FMO, however, requires to go beyond a donor-acceptor model rendering the problem much more complicated. Therefore allmost all approaches to study this question rely on weak coupling approximations [@Mohseni08; @Olaya1; @Olaya2; @Caruso09; @Rebentrost09; @Sarovar09; @Flemming1]. Ishizaki and Fleming [@Flemming2; @Fleming3] use a reduced hierarchy approach which they match for weak coupling to results from a Redfield approach and for strong coupling to results from Förster transfer description. The actual parameter range of interest is in between. We show in the present paper that even at weak coupling perturbative weak coupling approaches (like Redfield) break down and fail to describe the dynamics. We shortly review the Hamiltonian for a single chromophore in the next section \[sec2\] before we once more introduce a donor-acceptor model in order to discuss the validity of perturbative approaches by comparing a lowest order perturbative approach with a numerical exact treatment. Using a resumed perturbative approach (RESPET) [@RESPET1; @RESPET2; @RESPET3; @RESPET4; @RESPET5] we analytically derive in section \[sec3\] the time evolution of the population difference in our donor-acceptor model in lowest order in the system environment coupling. Thereby all non-Markovian
conductor between the antenna complex and the reaction center. The FMO protein is a trimer where the three identical subunits dwell of seven chromophoric web site which are fully characterized   [ @BioCh2005 ]. distinctive transport clock time for excitons through FMO are of the order of picoseconds. late experiment   [ @BioEn2007 ] on the FMO complex show that the exciton transport exhibits quantum coherent cycle up to $ 660 $ femtoseconds. This raises the authoritative question why the strong environmental fluctuations are insufficient to fully decohere the quantum transport. simplify the problem to a donor - acceptor model level it was prove   [ @BioTh2008; @Flemming2 ] that the slowness of the environment, namely $ \omega_c\simeq\Delta$, sufficiently slows down decoherence for quantum oscillation to survive even for potent coupling to the environment. Having understood why quantum coherence survives rather strong coupling to environmental fluctuations raises the subsequent question whether photosynthesis benefits from quantum coherence either to improve efficiency or to allow regulation. Detailed descriptions of transport through FMO, however, requires to go beyond a donor - acceptor model rendering the problem much more complicated. Therefore allmost all approaches to analyze this question rely on fallible yoke approximations   [ @Mohseni08; @Olaya1; @Olaya2; @Caruso09; @Rebentrost09; @Sarovar09; @Flemming1 ]. Ishizaki and Fleming [ @Flemming2; @Fleming3 ] use a reduced hierarchy approach which they match for weak coupling to results from a Redfield overture and for strong coupling to results from Förster transfer description. The actual parameter range of pastime is in between. We show in the present paper that even at weak coupling perturbative unaccented yoke approach (like Redfield) break down and fail to describe the dynamics. We concisely review the Hamiltonian for a single chromophore in the next section   \[sec2\ ] before we once more insert a donor - acceptor model in order to discuss the validity of perturbative approach path by comparing a lowest order perturbative approach with a numerical accurate treatment. Using a resumed perturbative approach (RESPET)   [ @RESPET1; @RESPET2; @RESPET3; @RESPET4; @RESPET5 ] we analytically deduce in section   \[sec3\ ] the time evolution of the population remainder in our donor - acceptor mannequin in lowest order in the system environment coupling. Thereby all non - Markovian
cojductor between the antekna complex and jhw reacvion cehter. The FMO protein is a trimer whece tye theee identical subunits consist lf seven chrinophoric smfes whigk are nully rharacterized [@BipCh2005]. Typican transport tiker yor excitons through FMO are of the jrder og oicoseconds. Resent qxpedpmtnts [@BioEn2007] on the FMO complex show that tie exciton tranxport exhibits quantum cohfrenh oscillations up ho $660$ femtosexondf. This raises the imporuaut question why the strong environmental fuuctuctions are unwufvhcient to fnlly dvcohere the qmsntum dranspott. Simplifying thx priblem to a donor-accepvor model level it wws shown [@BhoCh2008; @Flemming2] that the wliwnesv of the wnvkrohmxnt, namelj $\ojega_c\simeq\Selta$, suffixiently slows down cesiherence for suantui jscillations to survive even for strong cohpling to the environmebt. Having understood wjy quantui coherence survives rather strong couplings to etviroimdntco fluzruwtions raises the subsequent question whether lhptpsynthesis bencfits from quantum clhrtence either tu improve efficiency or to wllow rggulatuon. Detaijed cescriptions of transport tyrough FMO, hjqever, requires to yo beyond a bonor-avceptpr model rendering the 'roblej much more complicafdd. Therefore alloosn aln approaches to study this question relv on weay coopling wpproximatlons [@Mohseni08; @Olaya1; @Olaya2; @Farusl09; @Sebentrost09; @Sarovar09; @Flemming1]. Ishizaki and Hkeming [@Flemmonc2; @Fneming3] ufe a veduced hierarcry approach whnch they matcf for weak coupliig to resultf from a Redfhgld approach end for seront coypling gu results from Förster tgausfer desxription. The actuak pxdameter range oy unterest is in netdeeg. Ae srmw in the praseng pxler tfat even at deak coupling perturbatiee wsak coupling approscmes (like Tedfield) freak down anc fail to describe the vynamirs. We snorjly review the Hamiltonian for z single fhrjmophore in ehe kext section \[see2\] before we once more introduce a donor-arceptor model in order ro discuss the valibiuy of perturbetive wpproachev by comparing a lowwst order perturbstive approach with a numerhcal fxact treatment. Using a resumed perturbative approach (RESPET) [@RESPET1; @RESPET2; @RESPEV3; @WESPET4; @RESPST5] wr ananycicclly dewive mn section \[sec3\] the time evolution of the populatiun diffeseuce in our donor-acceptor modek kn lowest ordgr in the system environmeht couplong. Thereby all non-Markovian
conductor between the antenna complex and the The protein is trimer where the seven sites which are characterized [@BioCh2005]. Typical times for excitons through FMO are the order of picoseconds. Recent experiments [@BioEn2007] on the FMO complex show that exciton transport exhibits quantum coherent oscillations up to $660$ femtoseconds. This raises the question the environmental are insufficient to fully decohere the quantum transport. Simplifying the problem to a donor-acceptor model level was shown [@BioTh2008; @Flemming2] that the slowness of environment, namely $\omega_c\simeq\Delta$, sufficiently down decoherence for quantum oscillations survive for strong to environment. understood why quantum survives rather strong couplings to environmental fluctuations raises the subsequent question whether photosynthesis benefits from quantum coherence to improve to allow Detailed of through FMO, however, go beyond a donor-acceptor model rendering more complicated. Therefore allmost all approaches to study question rely weak coupling approximations [@Mohseni08; @Olaya1; @Olaya2; @Rebentrost09; @Sarovar09; @Flemming1]. Ishizaki and Fleming [@Flemming2; @Fleming3] a reduced hierarchy approach which they match for weak coupling to results from a Redfield for strong coupling to from Förster transfer The parameter of is in We show in the present paper that even at weak coupling weak coupling approaches (like Redfield) break down and fail to dynamics. shortly review the for a single chromophore the section \[sec2\] before we introduce donor-acceptor to the of perturbative approaches by a lowest order perturbative approach a numerical exact treatment. (RESPET) [@RESPET1; @RESPET2; @RESPET3; @RESPET4; @RESPET5] we analytically in section \[sec3\] the time evolution of population difference in our donor-acceptor model in lowest order in the system coupling. Thereby
conductor between the antennA complex anD the rEacTioN cEnteR. The fMO protein is a tRImer Where the three identical SubunItS ConsISt Of sevEn chromOPhORIc sItEs WhiCh ARe Fully ChaRacteriZed [@BioCh2005]. TyPicAl Transport timES fOr excitons ThrOugh FMO are of The Order oF pIcoSEcondS. ReCent eXperimENts [@Bioen2007] on the FMo cOMplex sHOw that tHE ExCitoN transport exhibitS QuANtum coherent osCillatIoNS uP TO $660$ feMtoSeconds. ThiS rAises THe imporTAnT QUEstIOn why the stronG environmenTAl fLuctuaTiOns ARe insuFficiEnT To fUlly decoherE the Quantum trAnsporT. simplifYIng the pRoblem To a DonOr-acCEpToR moDeL LevEL iT waS ShoWn [@BioTh2008; @FLeMmIng2] thAt thE SLOWnesS of The eNviroNment, namely $\omEga_C\simEQ\DeLta$, suFficiEntlY sLows dOwn decOhereNcE for quantum osciLlatIons to surVivE eVen FoR stroNG couplIng To tHe envirOnment. HAVinG uNDERsTood why quantum coheReNCE sUrvives rAther sTRoNg COuplings To EnvIronMENtal fLuctUAtIons raisEs the sUBsEqUent queStIon wheThEr pHotOsyntHEsis BenefiTs from quAntum COherence either TO improve efficIEnCY Or TO allOw rEgulation. DeTailED desCripTIoNs oF TransPort tHrOUgH fMO, however, requires tO gO beyonD a donOr-acceptor modEl renderinG THE problem Much MOrE Complicated. TheReforE allmost alL ApproachEs to sTudy this Question rELY on weak cOupLinG apProXIMaTions [@Mohseni08; @OLAYa1; @OlAyA2; @Caruso09; @rebEntrost09; @sarOvaR09; @FlEmmInG1]. Ishizaki And FlemiNg [@flEmMiNg2; @FLeminG3] Use a reduCeD hiErArcHy appROach whIch thEy maTcH fOR weAk couplINg TO ResuLtS fRom a redFiEld apProaCH anD for strOng coupliNg tO ResuLtS fRom FörsTer transfer deScRiption. The AcTuaL paramETEr range oF interest is in between. We sHOw in the PreSent pAper That even aT weAk coupLinG PerturBative Weak cOuPliNG ApproACHeS (liKe redfield) brEAK doWn and FaIl to DescribE the dynamics. We shorTLy rEview the HamilTonIan fOR A sIngLE cHRomOpHOre IN The next section \[sEc2\] before we OnCE mOre introduCE a dOnOr-accepTor modeL in orDEr to disCuss the vaLidity of pErTurbATIve Approaches By comparIng a lowesT Order PErTurbaTivE approAcH wiTh a nuMericaL ExaCt treAtment. usIng a reSumed PeRturbatiVe approach (RESPET) [@RESPET1; @ReSPET2; @ReSPET3; @rESpET4; @RESPET5] We aNAlyTically deRive In section \[sEc3\] tHe tIme evOluTIon of The pOPuLatIOn difFereNCe in our doNOr-AccEPToR model in lowEST OrdEr in tHe sYStem enViroNment coupling. TherEBy all non-MarkovIan
conductor between the ant enna compl ex an d t here acti on c enter. The FMO prot ein is a trimer wherethe t hr e e id e nt icalsubunit s c o n sis tof se ve n c hromo pho ric sit es which a refu lly characte r iz ed [@BioCh 200 5]. Typicaltra nsport t ime s forexc itons throu g h FMOare of th eo rder o f picose c o nd s. R ecent experiments [@ B ioEn2007] on t he FMO c o mp l e x s how that theex citon transpo r te x h ibi t s quantum coh erent oscil l ati ons up t o $ 6 60$ fe mtose co n ds. This raise s th e importa nt que s tion wh y the st rong e nvi ron ment a lfl uct ua t ion s a rei nsu fficient t ofully dec o h e r e th e q uant um tr ansport. Simp lif ying the prob lem t o ado nor-a ccepto r mod el level it was s hown  [@BioTh2 008 ;@Fl em ming2 ] thatthe sl ownessof thee nvi ro n m e nt , namely $\omega_c \s i m eq \Delta$, suffi c ie nt l y slowsdo wndeco h e rence for qu antum os cillat i on sto surv iv e even f orstr ong c o upli ng tothe envi ronme n t. Having und e rstood why qu a nt u m c o here nce survives r athe r str ongc ou pli n gs to envi ro n me n tal fluctuations ra is es the subs equent questi on whether p h otosynth esis be n efits from qua ntumcoherencee ither to impr ove effi ciency or t o allowreg ula tio n.D e ta iled descript i o ns o ftranspo rtthrough FM O,how eve r, requires to go b ey on dadon or-ac c eptor mo de l r en der ing t h e prob lem m uchmo re com plicate d .T h eref or eallm ost a ll ap proa c hes to stu dy this q ues t ionre ly on wea k coupling ap pr oximations  [ @Mo hseni0 8 ; @Olaya1 ; @Olaya2; @Caruso09; @ R ebentro st0 9; @S arov ar09; @Fl emm ing1]. Is h izakiand Fl eming [ @Fl e m ming2 ; @F lem in g3] use ar e duc ed hi er arch y appro ach which they mat c h f or weak coupl ing tor e su lts fr o m a R e dfi e l d approach andfor strong c o up ling to re s ult sfrom Fö rster t ransf e r descr iption. T he actual p aram e t errange of i nterestis in bet w een.W eshowinthe pr es ent pape r that eve n atweak c ou plingpertu rb ative we ak coupling approaches(likeRedfi eld ) break d own and fail todesc ribe the d yna mic s. W e s h ortly rev i ew th e Hami lton i an for as in gle c hr omophore in t h e n ext s ect i on \[s ec2\ ] before we oncem ore introducea do n o r-a cce p tormo del in order t o d is c u ss the v al idity of pe rturbati ve appro achesby com paringa lo w est or derper turbative ap pr o ach wit han umeric al e xa ct tre atment . Usi n g a resumed pertu rbati v e appr o ach (RES PE T) [@RE S PET1 ; @RESPET2 ; @RESPET3; @RESP ET4; @RES PET5] w eanalyt ica ll y derive i n section\[sec 3\] the t imeevo lution oft h e pop ulat io n d ifference i no ur d o nor -acc eptor m odel in lowes t order i n t h e syste menv i r onment co u p ling. Ther eby alln o n-Markovia n
conductor_between the_antenna complex and the_reaction center._The_FMO protein_is_a trimer where_the three identical_subunits consist of seven_chromophoric sites which_are_fully characterized [@BioCh2005]. Typical transport times for excitons through FMO are of the order of_picoseconds._Recent experiments [@BioEn2007]_on_the_FMO complex show that the_exciton transport exhibits quantum coherent_oscillations up_to $660$ femtoseconds. This raises the important question_why_the strong environmental_fluctuations are insufficient to fully decohere the quantum transport._Simplifying the problem to a donor-acceptor_model level it_was_shown [@BioTh2008;_@Flemming2] that the slowness_of the environment, namely $\omega_c\simeq\Delta$, sufficiently_slows down decoherence for quantum oscillations_to survive even for strong coupling to_the environment. Having understood why quantum coherence_survives rather strong couplings to_environmental fluctuations_raises the subsequent question whether_photosynthesis benefits from_quantum coherence_either to improve_efficiency or to allow regulation. Detailed_descriptions of transport_through FMO, however, requires to go_beyond_a donor-acceptor model_rendering_the_problem much_more complicated. Therefore_allmost_all approaches_to_study this question rely on weak_coupling_approximations [@Mohseni08; @Olaya1; @Olaya2; @Caruso09; @Rebentrost09; @Sarovar09; @Flemming1]._Ishizaki and Fleming [@Flemming2;_@Fleming3]_use a reduced hierarchy_approach which they match for_weak coupling to results from a_Redfield approach_and for_strong coupling to results from Förster transfer description. The actual parameter_range of interest is in between._We show in the_present paper_that_even at weak_coupling_perturbative weak_coupling approaches (like Redfield) break down and_fail to_describe the dynamics. We shortly review the_Hamiltonian for a single_chromophore_in the next section \[sec2\] before we_once more introduce a donor-acceptor model_in order to discuss the_validity_of_perturbative approaches by comparing a_lowest order perturbative approach with a_numerical exact treatment._Using a resumed perturbative approach (RESPET) [@RESPET1; @RESPET2;_@RESPET3;_@RESPET4; @RESPET5] we analytically derive in_section \[sec3\]_the time evolution of the population_difference_in_our donor-acceptor model in lowest_order in the system environment coupling._Thereby all non-Markovian
momentum independent $\Delta_{k}=\Delta$ and $\Delta$ is determined by minimizing the ground state energy. We note that we are considering a band structure with isotropic dispersion and the electron and hole Fermi surfaces are perfectly nested. In this case, the exciton pairing gap $\Delta$ is always non-zero in the mean-field theory, although its value can be very small. In reality the mean-field gap will be destroyed by quantum fluctuations when it’s magnitude is small [@exction-cond2], but this is not reflected in a mean field theory. To capture this physics qualitatively, we assume that the transition from the exciton condensed state to the normal state occurs at $\Delta=10^{-5}\mu$. Although quantitatively unreliable, this procedure allows us to examine the effect of screening on the phase diagram semi-quantitatively as we shall see below. With the above criteria, the phase diagram for different average particle-particle separation $r_{s}=\frac{1}{a_{B}}\sqrt{\frac{1}{\pi n}}$ ($n$ is particle/hole density; $a_{B}=\epsilon_{x}\hbar^{2}/m_{\text{eff}}e^{2}$ is the effective Bohr radius of electron-hole pair) and transition layer separation $b_{c}(r_{s})$ can be determined by solving the self-consistent equations and. We first consider EHBL with two screening layers. The result of calculation is depicted in Fig.\[fig:d-vs-rs\] for different separation between the electron/hole and its image charge $a$ (filled symbols). ![\[fig:d-vs-rs\] The phase diagram for EHBL with two- and one- screening layers for varied image charge separation $a$. $b$ is the bilayer separation and $r_{s}\sim\frac{1}{k_{F}}$ is the dimensionless average particle-particle separation in unit of the effective Bohr radius $a_{B}=\epsilon_{x}\hbar^{2}/m_{\text{eff}}e^{2}$. Below the transition lines $b_{c}(r_{s})$ an exciton gap $\Delta$ of magnitude larger than $10^{-5}\mu$ is formed.](d-vs-rs){width="1\columnwidth"} In the small $r_{s}$ (high density) regime, kinetic energy dominates over potential energy and the exciton pairing gap goes to zero as $r_{s}\rightarrow0$.
momentum independent $ \Delta_{k}=\Delta$ and $ \Delta$ is determined by minimizing the ground department of state department of energy. We note that we are considering a isthmus social organization with isotropic dispersion and the electron and hole Fermi surfaces are perfectly nest. In this case, the exciton pairing opening $ \Delta$ is always non - zero in the mean - playing field theory, although its value can be very small. In world the mean - field gap will be destroyed by quantum fluctuations when it ’s magnitude is small [ @exction - cond2 ], but this is not reflect in a mean field theory. To appropriate this physics qualitatively, we assume that the transition from the exciton condensed state of matter to the normal state occurs at $ \Delta=10^{-5}\mu$. Although quantitatively unreliable, this procedure allows us to test the effect of screening on the phase diagram semi - quantitatively as we shall see below. With the above criteria, the phase diagram for different average particle - particle separation $ r_{s}=\frac{1}{a_{B}}\sqrt{\frac{1}{\pi n}}$ ($ n$ is particle / hole density; $ a_{B}=\epsilon_{x}\hbar^{2}/m_{\text{eff}}e^{2}$ is the effective Bohr radius of electron - hole pair) and transition level interval $ b_{c}(r_{s})$ can be determined by clear the self - consistent equations and. We first consider EHBL with two cover layers. The resultant role of calculation is depicted in Fig.\[fig: d - vs - rs\ ] for different interval between the electron / hole and its image charge $ a$ (filled symbols). ! [ \[fig: d - vs - rs\ ] The phase diagram for EHBL with two- and one- screening level for varied image charge separation $ a$. $ b$ is the bilayer separation and $ r_{s}\sim\frac{1}{k_{F}}$ is the dimensionless average particle - atom interval in unit of the effective Bohr radius $ a_{B}=\epsilon_{x}\hbar^{2}/m_{\text{eff}}e^{2}$. Below the transition lines $ b_{c}(r_{s})$ an exciton break $ \Delta$ of magnitude larger than $ 10^{-5}\mu$ is formed.](d - vs - rs){width="1\columnwidth " } In the small $ r_{s}$ (eminent density) regime, kinetic energy dominates over likely energy and the exciton pairing gap goes to zero as $ r_{s}\rightarrow0$.
molentum independent $\Delta_{y}=\Delta$ and $\Deltc$ is devermines by minkmizing the ground state enecgy. Qe nouv that we are considefing a bajd strucrure qith isotropic disizrsioh and vhe electron anc hole Ferki surfaces ara oexfectly nested. In this case, the excieon paitijg gap $\Delta$ if alewys hon-zero in the mean-field theory, alfhough pts value can be fery small. In reality the lean-vield gap will be festroyed bi quwbtum fluctuagions when it’s magnituse is small [@exction-cond2], but thir is uot reflectgb in d mean fielv theogy. To capture this phfsics qialitatively, wc assnme rhat the transition fcom the exciton condgnsed stata co the normal state oxcyrs aj $\Delda=10^{-5}\mu$. Qltfoufh qhantitwtitely unrelizble, this peocedure allows us uo qqsmine the efrect os fcreening on the phase diagram semi-quanuitatjvely as we shall see bwlow. With the above crlteria, thq phase diagram for different average particle-pardicle repcvwguoj $r_{s}=\frac{1}{a_{B}}\sqrt{\frac{1}{\pi n}}$ ($n$ is particle/hole densyfy; $a_{N}=\epsilon_{x}\hbar^{2}/m_{\tewt{eff}}e^{2}$ is the effevtlvr Bohr radius uf eleefrkn-hole pair) and trwnsitiog laywr separauion $n_{c}(r_{s})$ can be determined by silving the svlf-cinsistent equationd and. We fixst comsidet EHBL with two screeniug laysrs. The resklt of camzulation is depiztec hn Fig.\[fig:d-vs-rs\] for differegt separavion yetween ghe glectrog/hole and lts image charge $a$ (fillef symyols). ![\[fhg:d-vs-rs\] Thf phase diagram for EHBL with txp- and one- scteetinc layers for yaried image chwrge separatiou $a$. $b$ is the cilayer seiaration end $r_{s}\sim\fras{1}{k_{F}}$ is the dikgnsionless avxrage pareiclw-parricle sdoaration in unot of the effective Bohr radius $a_{B}=\epsllon_{x}\fgar^{2}/m_{\text{eff}}e^{2}$. Beoow the transition lives $b_{b}(r_{s})$ ag exciton gap $\Delga$ ug magvitude largcr ghan $10^{-5}\mu$ is formed.](d-vs-rs){wigth="1\cklumnwidth"} In the skajl $r_{s}$ (hith densiey) regime, kinrtic energy dominaues ovxr potxntial engrgy and the exciton pairing gal goes to zevo as $r_{s}\rightwrroq0$.
momentum independent $\Delta_{k}=\Delta$ and $\Delta$ is determined the state energy. note that we with dispersion and the and hole Fermi are perfectly nested. In this case, exciton pairing gap $\Delta$ is always non-zero in the mean-field theory, although its can be very small. In reality the mean-field gap will be destroyed by fluctuations it’s is [@exction-cond2], but this is not reflected in a mean field theory. To capture this physics qualitatively, assume that the transition from the exciton condensed to the normal state at $\Delta=10^{-5}\mu$. Although quantitatively unreliable, procedure us to the of on the phase semi-quantitatively as we shall see below. With the above criteria, the phase diagram for different average particle-particle $r_{s}=\frac{1}{a_{B}}\sqrt{\frac{1}{\pi n}}$ particle/hole density; is effective radius of electron-hole transition layer separation $b_{c}(r_{s})$ can be the self-consistent equations and. We first consider EHBL two screening The result of calculation is depicted Fig.\[fig:d-vs-rs\] for different separation between the electron/hole and image charge $a$ (filled symbols). ![\[fig:d-vs-rs\] The phase diagram for EHBL with two- and one- for varied image charge $a$. $b$ is bilayer and is dimensionless average separation in unit of the effective Bohr radius $a_{B}=\epsilon_{x}\hbar^{2}/m_{\text{eff}}e^{2}$. Below the lines $b_{c}(r_{s})$ an exciton gap $\Delta$ of magnitude larger than formed.](d-vs-rs){width="1\columnwidth"} the small $r_{s}$ density) regime, kinetic energy over energy and the exciton goes zero
momentum independent $\Delta_{k}=\delta$ and $\DeLta$ is DetErmInEd by MiniMizing the grounD StatE energy. We note that we are ConsiDeRIng a BAnD struCture wiTH iSOTroPiC dIspErSIoN and tHe eLectron And hole FerMi sUrFaces are perfECtLy nested. In ThiS case, the exciTon PairinG gAp $\DELta$ is AlwAys noN-zero iN The meaN-field theOrY, AlthouGH its valUE CaN be vEry small. In reality THe MEan-field gap wilL be desTrOYeD BY quAntUm fluctuatIoNs wheN It’s magnITuDE IS smALl [@exction-cond2], But this is noT RefLected In A meAN field TheorY. TO CapTure this phySics QualitatiVely, we ASsume thAT the traNsitioN frOm tHe exCItOn ConDeNSed STaTe tO The Normal stAtE oCcurs At $\DeLTA=10^{-5}\MU$. AltHouGh quAntitAtively unreliAblE, thiS ProCedurE alloWs us To ExamiNe the eFfect Of Screening on the pHase Diagram seMi-qUaNtiTaTivelY As we shAll See Below. WiTh the abOVe cRiTERIa, The phase diagram for DiFFErEnt averaGe partIClE-pARticle sePaRatIon $r_{S}=\FRac{1}{a_{B}}\Sqrt{\FRaC{1}{\pi n}}$ ($n$ is pArticlE/HoLe Density; $A_{B}=\EpsiloN_{x}\HbaR^{2}/m_{\tExt{efF}}E^{2}$ is tHe effeCtive BohR radiUS of electron-holE Pair) and transiTIoN LAyER sepAraTion $b_{c}(r_{s})$ can Be deTErmiNed bY SoLviNG the sElf-coNsIStENt equations and. We firSt ConsidEr EHBl with two screeNing layers. tHE Result of CalcULaTIon is depicted iN Fig.\[fIg:d-vs-rs\] for DIfferent SeparAtion betWeen the elECTron/hole And Its ImaGe cHARgE $a$ (filled symboLS). ![\[Fig:d-Vs-Rs\] The phAse Diagram For eHBl wiTh tWo- And one- scrEening laYeRs FoR vAriEd imaGE charge sEpAraTiOn $a$. $B$ is thE BilayeR sepaRatiOn AnD $R_{s}\sIm\frac{1}{k_{f}}$ Is THE dimEnSiOnleSs aVeRage pArtiCLe-pArticle SeparatioN in UNit oF tHe EffectiVe Bohr radius $a_{b}=\ePsilon_{x}\hbaR^{2}/m_{\TexT{eff}}e^{2}$. BELOw the traNsition lines $b_{c}(r_{s})$ an excitON gap $\DelTa$ oF magnItudE larger thAn $10^{-5}\mU$ is forMed.](D-Vs-rs){wiDth="1\colUmnwiDtH"} In THE smalL $R_{S}$ (hIgh DeNsity) regimE, KIneTic enErGy doMinates Over potential energY And The exciton paiRinG gap GOEs To zERo AS $r_{s}\RiGHtaRROw0$.
momentum independent $\De lta_{k}=\D elta$ an d $ \D elta $ is determined by mini mizing the ground stat e ene rg y . We no te th at we a r ec o nsi de ri ngab an d str uct ure wit h isotropi c d is persion andt he electronand hole Fermisur facesar e p e rfect lyneste d. Int his ca se, the e xc i ton pa i ring ga p $\ Delt a$ is always non- z er o in the mean-f ield t he o ry , alt hou gh its val ue canb e verys ma l l . In reality the m ean-field g a p w ill be d est r oyed b y qua nt u m f luctuations whe n it’s ma gnitud e is sma l l [@exc tion-c ond 2], but th is is n o t r e fl ect e d i n a mean f ie ld th eory . T o cap tur e th is ph ysics qualita tiv ely, weassum e tha t th etrans itionfromth e exciton conde nsed state to th enor ma l sta t e occu rsat$\Delta =10^{-5 } \mu $. A l th ough quantitativel yu n re liable,this p r oc ed u re allow susto e x a minethee ff ect of s creeni n gon the ph as e diag ra m s emi -quan t itat ivelyas we sh all s e e below. With the above cri t er i a ,t he p has e diagram f or d i ffer enta ve rag e part icle- pa r ti c le separation $r_{s }= \frac{ 1}{a_ {B}}\sqrt{\fr ac{1}{\pin } } $ ($n$ i s pa r ti c le/hole densit y; $a _{B}=\epsi l on_{x}\h bar^{ 2}/m_{\t ext{eff}} e ^ {2}$ isthe ef fec tiv e Bo hr radius ofe l ectr on -hole p air ) and t ran sit ion la ye r separat ion $b_{ c} (r _{ s} )$can b e determi ne d b ysol vingt he sel f-con sist en te qua tions a n d. W e fi rs tcons ide rEHBLwith two screen ing layer s.T he r es ul t of ca lculation isde picted inFi g.\ [fig:d - v s-rs\] f or different separation between th e ele ctro n/hole an d i ts ima gec harge$a$ (f illed s ymb o l s). ! [ \[ fig :d -vs-rs\] T h e ph ase d ia gram for EH BL with two- and o n e-screening lay ers for v ar ied im a gech a rge s eparation $a$.$b$ is the b i la yer separa t ion a nd $r_{ s}\sim\ frac{ 1 }{k_{F} }$ is the dimensio nl essa v era ge particl e-partic le separa t ion i n u nit o f t he eff ec tiv e Boh r radi u s $ a_{B} =\epsi lo n_{x}\ hbar^ {2 }/m_{\te xt{eff}}e^{2}$. Below t he tra nsiti onlines $b_ {c} ( r_{ s})$ an e xcit on gap $\D elt a$of ma gni t ude l arge r t han $10^{ -5}\ m u$ is for m ed .]( d - vs -rs){width= " 1 \ col umnwi dth " } Inthesmall $r_{s}$ (hi g h density) reg ime, k ine tic ener gy dominates ove r p ot e n tial ene rg y and the e xciton p ai r ing g ap goe s to z ero as$ r _{ s }\righ tarr ow0 $.
momentum_independent $\Delta_{k}=\Delta$_and $\Delta$ is determined_by minimizing_the_ground state_energy._We note that_we are considering_a band structure with_isotropic dispersion and_the_electron and hole Fermi surfaces are perfectly nested. In this case, the exciton pairing_gap_$\Delta$ is_always_non-zero_in the mean-field theory, although_its value can be very_small. In_reality the mean-field gap will be destroyed by_quantum_fluctuations when it’s_magnitude is small [@exction-cond2], but this is not reflected_in a mean field theory. To_capture this physics_qualitatively,_we_assume that the transition_from the exciton condensed state to_the normal state occurs at $\Delta=10^{-5}\mu$._Although quantitatively unreliable, this procedure allows us_to examine the effect of screening_on the phase diagram semi-quantitatively_as we_shall see below. With the above_criteria, the phase_diagram for_different average particle-particle_separation $r_{s}=\frac{1}{a_{B}}\sqrt{\frac{1}{\pi n}}$ ($n$ is particle/hole_density; $a_{B}=\epsilon_{x}\hbar^{2}/m_{\text{eff}}e^{2}$ is_the effective Bohr radius of electron-hole_pair)_and transition layer_separation_$b_{c}(r_{s})$_can be_determined by solving_the_self-consistent equations_and._We first consider EHBL with two_screening_layers. The result of calculation is depicted_in Fig.\[fig:d-vs-rs\] for different_separation_between the electron/hole and_its image charge $a$ (filled_symbols). ![\[fig:d-vs-rs\] The phase diagram for EHBL_with two-_and one-_screening layers for varied image charge separation $a$. $b$ is the_bilayer separation and $r_{s}\sim\frac{1}{k_{F}}$ is the_dimensionless average particle-particle separation_in unit_of_the effective Bohr_radius_$a_{B}=\epsilon_{x}\hbar^{2}/m_{\text{eff}}e^{2}$. Below_the transition lines $b_{c}(r_{s})$ an exciton gap_$\Delta$ of_magnitude larger than $10^{-5}\mu$ is formed.](d-vs-rs){width="1\columnwidth"} In_the small $r_{s}$ (high_density)_regime, kinetic energy dominates over potential_energy and the exciton pairing gap_goes to zero as $r_{s}\rightarrow0$.
other words, we have the following *girth alternative*: Let ${\Sigma}$ be a compact orientable surface, and let $\Gamma$ be a finitely generated subgroup of $\Mod({\Sigma})$. Then, $\Gamma$ is either a non-cyclic group with infinite girth or a virtually free-abelian group; moreover, these alternatives are mutually exclusive. The girth alternative above reduces to the case where the interior of ${\Sigma}$ admits a complete hyperbolic metric. \[girth-alt: mcg hyp\] Let ${\Sigma}$ be a compact orientable surface whose interior admits a complete hyperbolic metric, and let $\Gamma$ be a finitely generated subgroup of $\Mod({\Sigma})$. Then, $\Gamma$ is either a non-cyclic group with infinite girth or a virtually free-abelian group; moreover, these alternatives are mutually exclusive. Let us first show that the general case, Theorem \[girth-alt: mcg\] in §\[Introduction\], follows from the hyperbolic case, Theorem \[girth-alt: mcg hyp\] above. Suppose for now that ${\Sigma}$ consists of copies of tori and a (possibly disconnected) surface that admits a complete hyperbolic metric. Since the mapping class groups of tori and one-punctured tori are isomorphic, we may replace the copies of tori in ${\Sigma}$ with the same number of copies of once-punctured tori. In turn, we now realize $\Gamma$ as a subgroup of the mapping class group of a hyperbolic surface; Theorem \[girth-alt: mcg\] follows from Theorem \[girth-alt: mcg hyp\] as desired. For the general case, let ${\Sigma}={\Sigma}' \sqcup {\Sigma}''$, such that ${\Sigma}'$ is the union of tori and hyperbolic components, and that ${\Sigma}''$ is the union of spheres, disks, and annuli. If the restriction $\Gamma'$ of $\Gamma$ to ${\Sigma}'$ is a non-cyclic group with infinite girth, then so is $\Gamma$ by Proposition \[akhmedov criterion\]. So, assume that $\Gamma'$ is virtually abelian and let $A' < \Gamma'$ be a finite-index abelian subgroup. Recall that the mapping class groups are trivial for the sphere, the disk, and the annulus; hence, the restriction of $\Gamma$ to ${\Sigma}''$ can only permute these components.
other words, we have the following * girth option *: lease $ { \Sigma}$ be a compact orientable surface, and lease $ \Gamma$ be a finitely render subgroup of $ \Mod({\Sigma})$. Then, $ \Gamma$ is either a non - cyclic group with infinite girth or a about free - abelian group; moreover, these alternatives are mutually single. The girth option above reduces to the case where the interior of $ { \Sigma}$ accept a complete hyperbolic metric. \[girth - alt: microgram hyp\ ] Let $ { \Sigma}$ be a compact orientable open whose interior admits a accomplished hyperbolic metric, and let $ \Gamma$ be a finitely generated subgroup of $ \Mod({\Sigma})$. Then, $ \Gamma$ is either a non - cyclic group with countless girth or a virtually complimentary - abelian group; moreover, these option are mutually exclusive. get us first show that the general lawsuit, Theorem   \[girth - alt: mcg\ ] in § \[Introduction\ ], follows from the hyperbolic case, Theorem   \[girth - alt: mcg hyp\ ] above. presuppose for now that $ { \Sigma}$ consists of copies of tori and a (possibly disconnected) surface that admits a complete hyperbolic metric. Since the mapping class group of tori and one - punctured torus are isomorphous, we may substitute the copies of tori in $ { \Sigma}$ with the like number of copy of once - punctured tori. In act, we now realize $ \Gamma$ as a subgroup of the mapping class group of a hyperbolic airfoil; Theorem   \[girth - alt: mcg\ ] follows from Theorem   \[girth - alt: mcg hyp\ ] as desire. For the general font, let $ { \Sigma}={\Sigma }' \sqcup { \Sigma}''$, such that $ { \Sigma}'$ is the union of tori and hyperbolic components, and that $ { \Sigma}''$ is the union of celestial sphere, disks, and annuli. If the restriction $ \Gamma'$ of $ \Gamma$ to $ { \Sigma}'$ is a non - cyclic group with countless girth, then so is $ \Gamma$ by Proposition   \[akhmedov criterion\ ]. then, assume that $ \Gamma'$ is virtually abelian and get $ A' < \Gamma'$ be a finite - index abelian subgroup. remember that the mapping course groups are trivial for the sphere, the disk, and the ring; hence, the restriction of $ \Gamma$ to $ { \Sigma}''$ can only permute these component.
otjer words, we have the foulowing *girth alternatmve*: Let ${\Aigma}$ be a compact orientable surfacx, ane let $\Gamma$ be a finitely gdnerated dubgroup of $\Nid({\Sigma})$. Thxh, $\Gamma$ is eifmer a ion-cyclic group with infitite girth or d xixtually free-abelian group; moreover, trese alyegnatives are motualkr exdlusive. The girth alternative above reducev to the case where the interior of ${\Sigla}$ afmits a complete hjperbolic mgfris. \[tirth-alt: mcg hyp\] Let ${\Spyma}$ be a cojpact orientable surface whose knternor admits q xomonete hyperbilic ietric, and lcn $\Gamma$ be a fonitely generabed snbgriup of $\Mod({\Sigma})$. Then, $\Jamma$ is either a nog-cyclic gsobp with infinite girty ir a eirtgallh frde-agekizn grokp; joreover, tgese alternqtives are mutually evbkusive. Let us first srow that the general case, Theorem \[girth-ant: jcg\] in §\[Introduction\], foloows from the hyperbopic case, Eheorem \[girth-alt: mcg hyp\] above. Suppose for now that ${\Sigme}$ zonwifgw lf copies of tori and a (possibly disconnected) autfsce that admitf a completr jylgrbolic metric. Since ths mapping class grlups of tori and one-ptnctired tori are isomorphic, we may replace rhe copies of tori in ${\Sigma}$ wnth thg same number of copies of onee-puncfured tori. Ln turn, ws now realize $\Gamoa$ ss a subgroup of the mapping class griup pf a hyoerbplic strface; Thelrem \[glsth-alt: mcg\] follows from Tveorem \[girtj-alt: mcg hyp\] as desired. For the jxneral case, lgt ${\Vigka}={\Sigma}' \fqcup {\Sigma}''$, such trat ${\Sigma}'$ is tke union of turi and hyierbolic romponents, agd that ${\Sigma}''$ ls the union of sphewes, eiskw, and avvuli. If the rextriction $\Gamma'$ of $\Tamma$ to ${\Sigma}'$ is e nov-dyclic group wick unfinite girth, yhev sj ps $\Jamma$ ty Propositimn \[akfmeapv crkterion\]. So, csrume that $\Gamma'$ is virtudlly abelian and let $A' < \Namma'$ be q finite-yndex abelian subgroup. Recall thwt thx mappmng clsss groups are trivial for the spgere, the fish, and the anguluw; hence, the xestriction of $\Gamma$ to ${\Sigma}''$ can only pxrmute these components.
other words, we have the following *girth ${\Sigma}$ a compact surface, and let subgroup $\Mod({\Sigma})$. Then, $\Gamma$ either a non-cyclic with infinite girth or a virtually group; moreover, these alternatives are mutually exclusive. The girth alternative above reduces to case where the interior of ${\Sigma}$ admits a complete hyperbolic metric. \[girth-alt: mcg Let be compact surface whose interior admits a complete hyperbolic metric, and let $\Gamma$ be a finitely generated subgroup $\Mod({\Sigma})$. Then, $\Gamma$ is either a non-cyclic group infinite girth or a free-abelian group; moreover, these alternatives mutually Let us show the case, Theorem \[girth-alt: in §\[Introduction\], follows from the hyperbolic case, Theorem \[girth-alt: mcg hyp\] above. Suppose for now that ${\Sigma}$ of copies and a disconnected) that a complete hyperbolic the mapping class groups of tori are isomorphic, we may replace the copies of in ${\Sigma}$ the same number of copies of tori. In turn, we now realize $\Gamma$ as subgroup of the mapping class group of a hyperbolic surface; Theorem \[girth-alt: mcg\] follows from mcg hyp\] as desired. the general case, ${\Sigma}={\Sigma}' {\Sigma}''$, that is the of tori and hyperbolic components, and that ${\Sigma}''$ is the union spheres, disks, and annuli. If the restriction $\Gamma'$ of $\Gamma$ is non-cyclic group with girth, then so is by \[akhmedov criterion\]. So, assume is abelian < be finite-index abelian subgroup. Recall the mapping class groups are for the sphere, the the restriction of $\Gamma$ to ${\Sigma}''$ can only these components.
other words, we have the followIng *girth alTernaTivE*: LeT ${\SIgma}$ Be a cOmpact orientabLE surFace, and let $\Gamma$ be a finiTely gEnERateD SuBgrouP of $\Mod({\SIGmA})$. tHen, $\gaMmA$ is EiTHeR a non-CycLic grouP with infinIte GiRth or a virtuaLLy Free-abeliaN grOup; moreover, tHesE alterNaTivES are mUtuAlly eXclusiVE. The giRth alternAtIVe abovE Reduces TO ThE casE where the interior OF ${\SIGma}$ admits a compLete hyPeRBoLIC meTriC. \[girth-alt: mCg Hyp\] LeT ${\sigma}$ be A CoMPACt oRIentable surfaCe whose inteRIor Admits A cOmpLEte hypErbolIc MEtrIc, and let $\GamMa$ be A finitely GeneraTEd subgrOUp of $\Mod({\sigma})$. THen, $\gamMa$ is EItHeR a nOn-CYclIC gRouP WitH infinitE gIrTh or a VirtUALLY freE-abEliaN grouP; moreover, thesE alTernATivEs are MutuaLly eXcLusivE. Let us First ShOw that the generaL casE, Theorem \[gIrtH-aLt: mCg\] In §\[IntROductiOn\], fOllOws from The hypeRBolIc CASE, THeorem \[girth-alt: mcg hYp\] ABOvE. Suppose For now THaT ${\SIGma}$ consiStS of CopiES Of torI and A (PoSsibly diSconneCTeD) sUrface tHaT admitS a ComPleTe hypERbolIc metrIc. Since tHe mapPIng class groups OF tori and one-puNCtUREd TOri aRe iSomorphic, we May rEPlacE the COpIes OF tori In ${\SigMa}$ WItH The same number of copiEs Of once-PunctUred tori. In turN, we now realIZE $\gamma$ as a SubgROuP Of the mapping clAss grOup of a hypeRBolic surFace; THeorem \[giRth-alt: mcg\] FOLlows froM ThEorEm \[gIrtH-ALt: Mcg hyp\] as desirED. for tHe General CasE, let ${\SigMa}={\SIgmA}' \sqCup {\siGma}''$, such thAt ${\Sigma}'$ iS tHe UnIoN of Tori aND hyperboLiC coMpOneNts, anD That ${\SiGma}''$ is The uNiOn OF spHeres, diSKs, AND annUlI. IF the ResTrIctioN $\GamMA'$ of $\gamma$ to ${\sigma}'$ is a nOn-cYClic GrOuP with inFinite girth, thEn So is $\Gamma$ bY PRopOsitioN \[AKhmedov cRiterion\]. So, assume that $\GamMA'$ is virtUalLy abeLian And let $A' < \GaMma'$ Be a finIte-INdex abElian sUbgroUp. recALL that THE mAppInG class grouPS Are TriviAl For tHe spherE, the disk, and the annuLUs; hEnce, the restriCtiOn of $\gAMmA$ to ${\sIgMA}''$ caN oNLy pERMute these componEnts.
other words, we have thefollowing*girt h a lte rn ativ e*: Let ${\Sigma} $ bea compact orientable s urfac e, andl et $\Ga mma$ be af i nit el ygen er a te d sub gro up of $ \Mod({\Sig ma} )$ . Then, $\Ga m ma $ is eithe r a non-cyclicgro up wit hinf i nitegir th or a vir t uallyfree-abel ia n group ; moreov e r ,thes e alternatives ar e m u tually exclusi ve. T he gi r t h a lte rnative ab ov e red u ces tot he c a sew here the inte rior of ${\ S igm a}$ ad mi tsa compl ete h yp e rbo lic metric. \[ girth-alt : mcgh yp\] Le t ${\Sig ma}$ b e a co mpac t o ri ent ab l e s u rf ace who se inter io radmit s ac o m p lete hy perb olicmetric, and l et$\Ga m ma$ be a fini tely g enera ted su bgrou pof $\Mod({\Sigm a})$ . Then, $ \Ga mm a$is eith e r a no n-c ycl ic grou p withi nfi ni t e gi rth or a virtually f r e e- abeliangroup; mo re o ver, the se al tern a t ivesarem ut ually ex clusiv e . L et us f ir st sho wtha t t he ge n eral case, Theorem  \[gi r th-alt: mcg\]i n §\[Introduc t io n \ ], foll ows from the h yper b olic cas e ,The o rem \ [girt h- a lt : mcg hyp\] above. Su pposefor n ow that ${\Si gma}$ cons i s t s of cop ieso ft ori and a (pos sibly disconnec t ed) surf ace t hat admi ts a comp l e te hyper bol icmet ric . Si nce the mappi n g cla ss groups of tori a ndone -pu nct ur ed tori a re isomo rp hi c, w e m ay re p lace the c opi es of tori in ${\ Sigma }$ w it ht hesame nu m be r of c op ie s of on ce -punc ture d to ri. Inturn, wenow real iz e$\Gamma $ as a subgro up of the ma pp ing class g roup ofa hyperbolic surface; T h eorem \ [gi rth-a lt:mcg\] fol low s from Th e orem \ [girth -alt: m cgh y p\] a s de sir ed . For the g ene ral c as e, l et ${\S igma}={\Sigma}' \s q cup {\Sigma}''$, su ch t h a t${\ S ig m a}' $i s t h e union of toriand hyperb ol i ccomponents , an dthat ${ \Sigma} ''$ i s the un ion of sp heres, di sk s, a n d an nuli. If t he restr iction $\ G amma' $ o f $\G amm a$ to${ \Si gma}' $ is a non -cycl ic gro up withinfin it e girth, then so is $\Gamma$ by Propo sitio n \ [akhmedov cr i ter ion\]. So , as sume that$\G amm a'$ i s v i rtual ly a b el ian and l et $ A ' < \Gamm a '$ be a f inite-index a b eli an su bgr o up. R ecal l that the mappin g class groupsaret r ivi alf or t he sphere, the d isk ,a n d the an nu lus; hence, the res tr i ction of $\ Gamma$ to ${\ S i gm a }''$ c an o nly permutethe se compone nt s.
other_words, we_have the following *girth_alternative*: Let ${\Sigma}$_be_a compact_orientable_surface, and let_$\Gamma$ be a_finitely generated subgroup of_$\Mod({\Sigma})$. Then, $\Gamma$_is_either a non-cyclic group with infinite girth or a virtually free-abelian group; moreover, these_alternatives_are mutually_exclusive. The_girth_alternative above reduces to the_case where the interior of_${\Sigma}$ admits_a complete hyperbolic metric. \[girth-alt: mcg hyp\] Let ${\Sigma}$_be_a compact orientable_surface whose interior admits a complete hyperbolic metric, and_let $\Gamma$ be a finitely generated_subgroup of $\Mod({\Sigma})$._Then,_$\Gamma$_is either a non-cyclic_group with infinite girth or a_virtually free-abelian group; moreover, these alternatives_are mutually exclusive. Let us first show that_the general case, Theorem \[girth-alt: mcg\] in_§\[Introduction\], follows from the hyperbolic_case, Theorem \[girth-alt:_mcg hyp\] above. Suppose for now_that ${\Sigma}$ consists_of copies_of tori and_a (possibly disconnected) surface that admits_a complete hyperbolic_metric. Since the mapping class groups_of_tori and one-punctured_tori_are_isomorphic, we_may replace the_copies_of tori_in_${\Sigma}$ with the same number of_copies_of once-punctured tori. In turn, we now_realize $\Gamma$ as a_subgroup_of the mapping class_group of a hyperbolic surface;_Theorem \[girth-alt: mcg\] follows from Theorem \[girth-alt: mcg_hyp\] as_desired. For the_general case, let ${\Sigma}={\Sigma}' \sqcup {\Sigma}''$, such that ${\Sigma}'$ is the_union of tori and hyperbolic components,_and that ${\Sigma}''$ is_the union_of_spheres, disks, and_annuli._If the_restriction $\Gamma'$ of $\Gamma$ to ${\Sigma}'$ is_a non-cyclic_group with infinite girth, then so_is $\Gamma$ by Proposition \[akhmedov_criterion\]._So, assume that $\Gamma'$ is virtually_abelian and let $A' < \Gamma'$_be a finite-index abelian subgroup. Recall_that_the_mapping class groups are trivial_for the sphere, the disk, and_the annulus; hence,_the restriction of $\Gamma$ to ${\Sigma}''$ can_only_permute these components.
face of ${\mathrm{Newt}}_p(f)$ with inner normal $(v,1)$. [$\blacksquare$]{} In Example \[ex:newt\] earlier, note that the $3$ lower edges have respective horizontal lengths $2$, $3$, and $1$, and inner normals $(1,1)$, $(0,1)$, and $(-5,1)$. Lemma \[lemma:newt\] then tells us that $f$ has exactly $6$ roots in ${\mathbb{C}}_3$: $2$ with $3$-adic valuation $1$, $3$ with $3$-adic valuation $0$, and $1$ with $3$-adic valuation $-5$. Indeed, one can check that the roots of $f$ are exactly $6$, $1$, and $\frac{1}{243}$, with respective multiplicities $2$, $3$, and $1$. [$\diamond$]{} The Proof of Assertion (2) of Theorem \[thm:qp\] {#sub:proof2} ------------------------------------------------ The existence of $0$ as a root is clearly checkable in constant time so we may again assume that $f$ is not divisible by $x_1$. Via the reciprocal polynomial $f^*(x_1)\!:=\!x_1^{\deg f}f(1/x_1)$, it is then enough to show that, for most $f$, having a root in ${\mathbb{Z}}_p$ admits a succinct certificate. As observed in the proof of Assertion (2), ${\mathrm{Newt}}_p(f)$ can be computed in polynomial-time. Since ${{\mathrm{ord}}}_pc_i\!\leq\!\log_pc_i\!\leq\!{\mathrm{size}}(c_i)$, note also that that every root $\zeta\!\in\!{\mathbb{C}}_p$ of $f$ satisfies $|{{\mathrm{ord}}}_p\zeta|\!\leq\!2\max_i{\mathrm{size}}(c_i)\!\leq\!2{\mathrm{size}}(f)\!<\! 2{\mathrm{size}}_p(f)$. Since ${{\mathrm{ord}}}_p({\mathbb{Z}}_p)\!=\!{\mathbb{N}}\cup\{0\}$, we can clearly assume that ${\mathrm{Newt}}_p(f)$ has an edge with non-positive integral slope, for otherwise $f$ would have no roots in ${\mathbb{Z}}_p
face of $ { \mathrm{Newt}}_p(f)$ with inner normal $ (v,1)$. [ $ \blacksquare$ ] { } In Example \[ex: newt\ ] earlier, note that the $ 3 $ low edge have respective horizontal lengths $ 2 $, $ 3 $, and $ 1 $, and inner convention $ (1,1)$, $ (0,1)$, and $ (-5,1)$. Lemma \[lemma: newt\ ] then tells us that $ f$ has exactly $ 6 $ solution in $ { \mathbb{C}}_3 $: $ 2 $ with $ 3$-adic valuation $ 1 $, $ 3 $ with $ 3$-adic valuation $ 0 $, and $ 1 $ with $ 3$-adic evaluation $ -5$. Indeed, one can match that the roots of $ f$ are precisely $ 6 $, $ 1 $, and $ \frac{1}{243}$, with respective multiplicities $ 2 $, $ 3 $, and $ 1$. [ $ \diamond$ ] { } The Proof of Assertion (2) of Theorem \[thm: qp\ ] { # sub: proof2 } ------------------------------------------------ The being of $ 0 $ as a root is clearly checkable in constant time so we may again assume that $ f$ is not divisible by $ x_1$. Via the reciprocal polynomial $ f^*(x_1)\!:=\!x_1^{\deg f}f(1 / x_1)$, it is then adequate to show that, for most $ f$, having a solution in $ { \mathbb{Z}}_p$ admits a succinct security. As observed in the validation of Assertion (2), $ { \mathrm{Newt}}_p(f)$ can be computed in polynomial - time. Since $ { { \mathrm{ord}}}_pc_i\!\leq\!\log_pc_i\!\leq\!{\mathrm{size}}(c_i)$, note also that that every root $ \zeta\!\in\!{\mathbb{C}}_p$ of $ f$ satisfies $ |{{\mathrm{ord}}}_p\zeta|\!\leq\!2\max_i{\mathrm{size}}(c_i)\!\leq\!2{\mathrm{size}}(f)\!<\! 2{\mathrm{size}}_p(f)$. Since $ { { \mathrm{ord}}}_p({\mathbb{Z}}_p)\!=\!{\mathbb{N}}\cup\{0\}$, we can clearly wear that $ { \mathrm{Newt}}_p(f)$ has an edge with non - positive integral slope, for otherwise $ f$ would have no roots in $ { \mathbb{Z}}_p
fafe of ${\mathrm{Newt}}_p(f)$ with lnner normal $(v,1)$. [$\blacksquere$]{} In Esample \[eb:newt\] earlier, note that the $3$ liwer tbges have respective horizontwl lengtys $2$, $3$, qnd $1$, and iiher normals $(1,1)$, $(0,1)$, wnd $(-5,1)$. Oemma \[lemma:newj\] then tells us that $f$ has ebaetly $6$ roots in ${\mathbb{C}}_3$: $2$ with $3$-adic vajuation $1$, $3$ with $3$-adic valoatiom $0$, ans $1$ with $3$-adic valuation $-5$. Indeed, one can chtck that the roots of $f$ are exactly $6$, $1$, and $\frwc{1}{243}$, wlth respective mulhiplicities $2$, $3$, age $1$. [$\diamond$]{} Thd Proof of Assertion (2) kf Theorem \[thm:qp\] {#sub:proof2} ------------------------------------------------ The exkstenee of $0$ as a riot hs clearly rheckafle in constant time so we kay again assuke vhat $f$ is not divisible bb $x_1$. Via the reciprocwl polynokicl $f^*(x_1)\!:=\!x_1^{\deg f}f(1/x_1)$, it is tyeb enoogh tm shuq tfat, fpr most $v$, heving a roof in ${\mathbb{Z}}_p$ admits a succincu cqgyificate. As kbservqd in the proof of Assertion (2), ${\mathrm{Newt}}_p(x)$ czn be computed in polynimial-time. Since ${{\mathrl{ord}}}_pc_i\!\lez\!\log_pc_i\!\leq\!{\mathrm{size}}(c_i)$, note also that that every soot $\vega\!\iu\!{\nathbc{X}}_p$ of $f$ satisfies $|{{\mathrm{ord}}}_p\zeta|\!\leq\!2\max_i{\mathrm{siss}}(c_o)\!\lvq\!2{\mathrm{size}}(f)\!<\! 2{\mathvm{size}}_p(f)$. Since ${{\mathtm{lrc}}}_k({\mathbb{Z}}_p)\!=\!{\mathbc{N}}\cup\{0\}$, cs dan clearly assume that ${\mwthrm{Bewt}}_p(f)$ haf an edge with non-positive intetral slope, fje otherwise $f$ woulb have no routs on ${\mayhbb{Z}}_p
face of ${\mathrm{Newt}}_p(f)$ with inner normal $(v,1)$. Example earlier, note the $3$ lower $2$, and $1$, and normals $(1,1)$, $(0,1)$, $(-5,1)$. Lemma \[lemma:newt\] then tells us $f$ has exactly $6$ roots in ${\mathbb{C}}_3$: $2$ with $3$-adic valuation $1$, $3$ $3$-adic valuation $0$, and $1$ with $3$-adic valuation $-5$. Indeed, one can check the of are $6$, $1$, and $\frac{1}{243}$, with respective multiplicities $2$, $3$, and $1$. [$\diamond$]{} The Proof of Assertion of Theorem \[thm:qp\] {#sub:proof2} ------------------------------------------------ The existence of as a root is checkable in constant time so may assume that is divisible $x_1$. Via the polynomial $f^*(x_1)\!:=\!x_1^{\deg f}f(1/x_1)$, it is then enough to show that, for most $f$, having a root in admits a As observed the of (2), ${\mathrm{Newt}}_p(f)$ can in polynomial-time. Since ${{\mathrm{ord}}}_pc_i\!\leq\!\log_pc_i\!\leq\!{\mathrm{size}}(c_i)$, note also root $\zeta\!\in\!{\mathbb{C}}_p$ of $f$ satisfies $|{{\mathrm{ord}}}_p\zeta|\!\leq\!2\max_i{\mathrm{size}}(c_i)\!\leq\!2{\mathrm{size}}(f)\!<\! 2{\mathrm{size}}_p(f)$. Since we can assume that ${\mathrm{Newt}}_p(f)$ has an edge non-positive integral slope, for otherwise $f$ would have roots in ${\mathbb{Z}}_p
face of ${\mathrm{Newt}}_p(f)$ with innEr normal $(v,1)$. [$\bLacksQuaRe$]{} IN EXampLe \[ex:Newt\] earlier, notE That The $3$ lower edges have respeCtive HoRIzonTAl LengtHs $2$, $3$, and $1$, anD InNER noRmAlS $(1,1)$, $(0,1)$, anD $(-5,1)$. LEMmA \[lemmA:neWt\] then tElls us that $F$ haS eXactly $6$ roots iN ${\MaThbb{C}}_3$: $2$ with $3$-aDic Valuation $1$, $3$ witH $3$-adIc valuAtIon $0$, ANd $1$ witH $3$-adIc valUation $-5$. iNdeed, oNe can checK tHAt the rOOts of $f$ aRE ExActlY $6$, $1$, and $\frac{1}{243}$, with respeCTiVE multiplicitieS $2$, $3$, and $1$. [$\diAmONd$]{} tHE PrOof Of AssertioN (2) oF TheoREm \[thm:qp\] {#SUb:PROOf2} ------------------------------------------------ THE existence of $0$ aS a root is cleARly CheckaBlE in COnstanT time So WE maY again assumE thaT $f$ is not diVisiblE By $x_1$. Via tHE reciprOcal poLynOmiAl $f^*(x_1)\!:=\!X_1^{\DeG f}F(1/x_1)$, iT iS TheN EnOugH To sHow that, fOr MoSt $f$, haVing A ROOT in ${\mAthBb{Z}}_p$ AdmitS a succinct cerTifIcatE. as oBservEd in tHe prOoF of AsSertioN (2), ${\mathRm{newt}}_p(f)$ can be compUted In polynomIal-TiMe. SInCe ${{\matHRm{ord}}}_pC_i\!\lEq\!\lOg_pc_i\!\leQ\!{\mathrm{SIze}}(C_i)$, NOTE aLso that that every roOt $\ZETa\!\In\!{\mathbb{c}}_p$ of $f$ sATiSfIEs $|{{\mathrm{OrD}}}_p\zEta|\!\lEQ\!2\Max_i{\mAthrM{SiZe}}(c_i)\!\leq\!2{\mAthrm{sIZe}}(F)\!<\! 2{\mAthrm{siZe}}_P(f)$. SincE ${{\mAthRm{oRd}}}_p({\maTHbb{Z}}_P)\!=\!{\mathbB{N}}\cup\{0\}$, we cAn cleARly assume that ${\mAThrm{Newt}}_p(f)$ has AN eDGE wITh noN-poSitive integRal sLOpe, fOr otHErWisE $F$ woulD have No ROoTS in ${\mathbb{Z}}_p
face of ${\mathrm{Newt}}_ p(f)$ with inne r n orm al $(v ,1)$ . [$\blacksqua r e$]{ } In Example \[ex:new t\] e ar l ier, no te th at the$ 3$ l owe red ges h a ve resp ect ive hor izontal le ngt hs $2$, $3$, a n d$1$, and i nne r normals $( 1,1 )$, $( 0, 1)$ , and$(- 5,1)$ . Lemm a \[lem ma:newt\] t h en tel l s us th a t $ f$ h as exactly $6$ ro o ts in ${\mathbb{C }}_3$: $ 2 $w i th$3$ -adic valu at ion $ 1 $, $3$w it h $ 3$- a dic valuation $0$, and $ 1 $ w ith $3 $- adi c valua tion$- 5 $.Indeed, one can check th at the roots o f $f$ ar e exac tly $6 $, $ 1 $, a nd$\ f rac { 1} {24 3 }$, with re sp ec tivemult i p l i citi es$2$, $3$, and $1$. [$\ dia mond $ ]{} The Proo f of A ssert ion (2 ) ofTh eorem \[thm:qp\ ] {# sub:proof 2}-- --- -- ----- - ------ --- --- ------- ------- - --- -- - - - The existence of $ 0$ a sa root i s clea r ly c h eckablein co nsta n t time sow emay agai n assu m eth at $f$is not d iv isi ble by $ x _1$. Via t he recip rocal polynomial $f^ * (x_1)\!:=\!x_ 1 ^{ \ d eg f}f( 1/x _1)$, it is the n eno ught osho w that , for m o st $f$, having a rootin ${\ma thbb{ Z}}_p$ admits a succinc t c ertifica te.A so bserved in the proo f of Asser t ion (2), ${\m athrm{Ne wt}}_p(f) $ can be c omp ute d i n p o l yn omial-time. S i n ce $ {{ \mathrm {or d}}}_pc _i\ !\l eq\ !\l og _pc_i\!\l eq\!{\ma th rm {s iz e}} (c_i) $ , note a ls o t ha t t hat e v ery ro ot $\ zeta \! \i n \!{ \mathbb { C} } _ p$ o f$f $ sa tis fi es $| {{\m a thr m{ord}} }_p\zeta| \!\ l eq\! 2\ ma x_i{\ma thrm{size}}(c _i )\!\leq\!2 {\ mat hrm{si z e }}(f)\!< \! 2{\mathrm{size}}_p(f ) $. Sin ce${{\m athr m{ord}}}_ p({ \mathb b{Z } }_p)\! =\!{\m athbb {N }}\ c u p\{0\ } $ ,weca n clearlya s sum e tha t${\m athrm{N ewt}}_p(f)$ has an edg e with non-po sit ivei n te gra l s l ope ,f oro t herwise $f$ wou ld have no r o ot s in ${\ma t hbb {Z }}_p
face_of ${\mathrm{Newt}}_p(f)$_with inner normal $(v,1)$._[$\blacksquare$]{} In Example_\[ex:newt\]_earlier, note_that_the $3$ lower_edges have respective_horizontal lengths $2$, $3$,_and $1$, and_inner_normals $(1,1)$, $(0,1)$, and $(-5,1)$. Lemma \[lemma:newt\] then tells us that $f$ has exactly_$6$_roots in_${\mathbb{C}}_3$:_$2$_with $3$-adic valuation $1$, $3$_with $3$-adic valuation $0$, and_$1$ with_$3$-adic valuation $-5$. Indeed, one can check that_the_roots of $f$_are exactly $6$, $1$, and $\frac{1}{243}$, with respective multiplicities_$2$, $3$, and $1$. [$\diamond$]{} The Proof_of Assertion (2)_of_Theorem_\[thm:qp\] {#sub:proof2} ------------------------------------------------ The existence of_$0$ as a root is clearly_checkable in constant time so we_may again assume that $f$ is not_divisible by $x_1$. Via the reciprocal_polynomial $f^*(x_1)\!:=\!x_1^{\deg f}f(1/x_1)$, it is_then enough_to show that, for most_$f$, having a_root in_${\mathbb{Z}}_p$ admits a_succinct certificate. As observed in the_proof of Assertion_(2), ${\mathrm{Newt}}_p(f)$ can be computed in_polynomial-time._Since ${{\mathrm{ord}}}_pc_i\!\leq\!\log_pc_i\!\leq\!{\mathrm{size}}(c_i)$, note_also_that_that every_root $\zeta\!\in\!{\mathbb{C}}_p$ of_$f$_satisfies $|{{\mathrm{ord}}}_p\zeta|\!\leq\!2\max_i{\mathrm{size}}(c_i)\!\leq\!2{\mathrm{size}}(f)\!<\! 2{\mathrm{size}}_p(f)$. Since_${{\mathrm{ord}}}_p({\mathbb{Z}}_p)\!=\!{\mathbb{N}}\cup\{0\}$,_we can clearly assume that ${\mathrm{Newt}}_p(f)$_has_an edge with non-positive integral slope, for_otherwise $f$ would have_no_roots in ${\mathbb{Z}}_p
\be_{\alpha_j} +\be_{\alpha_j} \be_{\alpha_i}^2 &= (q+q^{-1}) \be_{\alpha_i} \be_{\alpha_j} \be_{\alpha_i}, \ \ \quad\quad &\text{if }& |i-j|=1, \displaybreak[0]\\ \be_{\alpha_i} \be_{\alpha_j} &= \be_{\alpha_j} \be_{\alpha_i}, \ \qquad\qquad\ \ \ \ \qquad &\text{if }& |i-j|>1, \displaybreak[0]\\ \bff_{\alpha_i}^2 \bff_{\alpha_j} +\bff_{\alpha_j} \bff_{\alpha_i}^2 &= (q+q^{-1}) \bff_{\alpha_i} \bff_{\alpha_j} \bff_{\alpha_i}, \ \ \quad\quad &\text{if }& |i-j|=1,\displaybreak[0]\\ \bff_{\alpha_i} \bff_{\alpha_j} &= \bff_{\alpha_j} \bff_{\alpha_i}, \ \qquad\qquad\ \ \ \ \qquad &\text{if }& |i-j|>1, \displaybreak[0]\\ \be_{\alpha_i}\bt &=\bt\be_{\alpha_i}, \quad\qquad\quad &\text{if }& i > 1, \displaybreak[0]\\ \be_{\alpha_1}^2\bt + \bt\be_{\alpha_1}^2 &= (q+q^{-1}) \be_{\alpha_1}\bt\be_{\alpha_1},\displaybreak[0]\\ \bt^2\be_{\alpha_1} + \be_{\alpha_1}\bt^2 &= (q + q^{-1}) \bt\be_{\alpha_1}\bt + \be_{\alpha_1},\displaybreak[0]\\ \bff_{\alpha_i}\bt &=\bt\bff_{\alpha_i}, \quad\qquad&\text{if }& i > 1, \displaybreak[0]\\ \bff_{\alpha_1}^2\bt + \bt\bff
\be_{\alpha_j } + \be_{\alpha_j } \be_{\alpha_i}^2 & = (q+q^{-1 }) \be_{\alpha_i } \be_{\alpha_j } \be_{\alpha_i }, \ \ \quad\quad & \text{if } & |i - j|=1, \displaybreak[0]\\ \be_{\alpha_i } \be_{\alpha_j } & = \be_{\alpha_j } \be_{\alpha_i }, \ \qquad\qquad\ \ \ \ \qquad & \text{if } & |i - j|>1, \displaybreak[0]\\ \bff_{\alpha_i}^2 \bff_{\alpha_j } + \bff_{\alpha_j } \bff_{\alpha_i}^2 & = (q+q^{-1 }) \bff_{\alpha_i } \bff_{\alpha_j } \bff_{\alpha_i }, \ \ \quad\quad & \text{if } & |i - j|=1,\displaybreak[0]\\ \bff_{\alpha_i } \bff_{\alpha_j } & = \bff_{\alpha_j } \bff_{\alpha_i }, \ \qquad\qquad\ \ \ \ \qquad & \text{if } & |i - j|>1, \displaybreak[0]\\ \be_{\alpha_i}\bt & = \bt\be_{\alpha_i }, \quad\qquad\quad & \text{if } & i > 1, \displaybreak[0]\\ \be_{\alpha_1}^2\bt + \bt\be_{\alpha_1}^2 & = (q+q^{-1 }) \be_{\alpha_1}\bt\be_{\alpha_1},\displaybreak[0]\\ \bt^2\be_{\alpha_1 } + \be_{\alpha_1}\bt^2 & = (q + q^{-1 }) \bt\be_{\alpha_1}\bt + \be_{\alpha_1},\displaybreak[0]\\ \bff_{\alpha_i}\bt & = \bt\bff_{\alpha_i }, \quad\qquad&\text{if } & i > 1, \displaybreak[0]\\ \bff_{\alpha_1}^2\bt + \bt\bff
\be_{\wlpha_j} +\be_{\alpha_j} \be_{\alpha_i}^2 &= (q+q^{-1}) \be_{\alpha_i} \bg_{\aopha_j} \ue_{\alpha_j}, \ \ \qjad\quad &\text{if }& |i-j|=1, \displaybrxak[0]\\ \be_{\alkka_i} \be_{\alpha_j} &= \be_{\alphx_j} \be_{\alphw_i}, \ \qquae\qqued\ \ \ \ \qquad &\text{mr }& |i-j|>1, \dlfplagnreak[0]\\ \bff_{\alpha_i}^2 \bff_{\slpha_j} +\bff_{\dlpha_j} \bff_{\alphd_i}^2 &= (e+q^{-1}) \bff_{\alpha_i} \bff_{\alpha_j} \bff_{\alpha_i}, \ \ \quad\auwd &\text{if }& |i-j|=1,\dysplsrbrezk[0]\\ \bff_{\alpha_i} \bff_{\alpha_j} &= \bff_{\alpha_n} \bff_{\anpha_i}, \ \qquad\aquad\ \ \ \ \qquad &\text{if }& |i-j|>1, \fisppaybreak[0]\\ \be_{\alpha_i}\ht &=\bt\be_{\alphq_i}, \qtqd\qquad\quad &\gext{if }& i > 1, \displaybrgak[0]\\ \be_{\alpha_1}^2\bt + \bt\be_{\alpha_1}^2 &= (q+q^{-1}) \be_{\xlpha_1}\yt\be_{\alpha_1},\diwpoayhteak[0]\\ \bt^2\be_{\al'ha_1} + \bv_{\alpha_1}\bt^2 &= (q + q^{-1}) \bt\be_{\al[ha_1}\bt + \ne_{\alpha_1},\displaynreak[0]\\ \bdf_{\alpha_i}\bt &=\bt\bff_{\alpha_m}, \quad\qquad&\text{if }& y > 1, \displdyyreak[0]\\ \bff_{\alpha_1}^2\bt + \bt\vfd
\be_{\alpha_j} +\be_{\alpha_j} \be_{\alpha_i}^2 &= (q+q^{-1}) \be_{\alpha_i} \be_{\alpha_j} \ &\text{if }& \displaybreak[0]\\ \be_{\alpha_i} \be_{\alpha_j} \ \ \qquad &\text{if |i-j|>1, \displaybreak[0]\\ \bff_{\alpha_i}^2 +\bff_{\alpha_j} \bff_{\alpha_i}^2 &= (q+q^{-1}) \bff_{\alpha_i} \bff_{\alpha_j} \ \ \quad\quad &\text{if }& |i-j|=1,\displaybreak[0]\\ \bff_{\alpha_i} \bff_{\alpha_j} &= \bff_{\alpha_j} \bff_{\alpha_i}, \ \qquad\qquad\ \ \ \qquad &\text{if }& |i-j|>1, \displaybreak[0]\\ \be_{\alpha_i}\bt &=\bt\be_{\alpha_i}, \quad\qquad\quad &\text{if }& i 1, \be_{\alpha_1}^2\bt \bt\be_{\alpha_1}^2 (q+q^{-1}) \be_{\alpha_1}\bt\be_{\alpha_1},\displaybreak[0]\\ \bt^2\be_{\alpha_1} + \be_{\alpha_1}\bt^2 &= (q + q^{-1}) \bt\be_{\alpha_1}\bt + \be_{\alpha_1},\displaybreak[0]\\ \bff_{\alpha_i}\bt &=\bt\bff_{\alpha_i}, \quad\qquad&\text{if }& > 1, \displaybreak[0]\\ \bff_{\alpha_1}^2\bt + \bt\bff
\be_{\alpha_j} +\be_{\alpha_j} \be_{\alpha_i}^2 &= (q+Q^{-1}) \be_{\alpha_i} \bE_{\alphA_j} \bE_{\alPhA_i}, \ \ \quAd\quAd &\text{if }& |i-j|=1, \dispLAybrEak[0]\\ \be_{\alpha_i} \be_{\alpha_j} &= \be_{\aLpha_j} \Be_{\ALpha_I}, \ \QqUad\qqUad\ \ \ \ \qquaD &\TeXT{If }& |i-J|>1, \dIsPlaYbREaK[0]\\ \bff_{\aLphA_i}^2 \bff_{\alPha_j} +\bff_{\alpHa_j} \BfF_{\alpha_i}^2 &= (q+q^{-1}) \bff_{\ALpHa_i} \bff_{\alphA_j} \bFf_{\alpha_i}, \ \ \quad\QuaD &\text{iF }& |i-J|=1,\diSPlaybReaK[0]\\ \bff_{\aLpha_i} \bFF_{\alpha_J} &= \bff_{\alpha_J} \bFF_{\alpha_I}, \ \Qquad\qqUAD\ \ \ \ \qQuad &\Text{if }& |i-j|>1, \displaybrEAk[0]\\ \BE_{\alpha_i}\bt &=\bt\be_{\aLpha_i}, \qUaD\QqUAD\quAd &\tExt{if }& i > 1, \dispLaYbreaK[0]\\ \Be_{\alpha_1}^2\BT + \bT\BE_{\AlpHA_1}^2 &= (q+q^{-1}) \be_{\alpha_1}\bt\bE_{\alpha_1},\displAYbrEak[0]\\ \bt^2\bE_{\aLphA_1} + \Be_{\alphA_1}\bt^2 &= (q + q^{-1}) \Bt\BE_{\alPha_1}\bt + \be_{\alphA_1},\disPlaybreak[0]\\ \Bff_{\alpHA_i}\bt &=\bt\bFF_{\alpha_i}, \Quad\qqUad&\TexT{if }& i > 1, \DIsPlAybReAK[0]\\ \bfF_{\AlPha_1}^2\BT + \bt\Bff
\be_{\alpha_j} +\be_{\alp ha_j} \be_ {\alp ha_ i}^ 2&= ( q+q^ {-1}) \be_{\al p ha_i } \be_{\alpha_j} \be_{ \alph a_ i }, \ \ \ quad\qu a d& \ tex t{ if }& | i -j |=1,\di splaybr eak[0]\\ \b e_ {\alpha_i} \ b e_ {\alpha_j} &= \be_{\alpha _j} \be_{ \a lph a _i},\ \ qquad \qquad \ \ \ \ \qquad & \t e xt{if} & |i-j| > 1 ,\dis playbreak[0]\\ \ bf f _{\alpha_i}^2\bff_{ \a l ph a _ j}+\b ff_{\alpha _j } \bf f _{\alph a _i } ^ 2 &= (q+q^{-1}) \b ff_{\alpha_ i } \ bff_{\ al pha _ j} \bf f_{\a lp h a_i }, \ \\qua d\quad &\ text{i f }& |i- j |=1,\di splayb rea k[0 ]\\ \ bf f_{ \a l pha _ i} \b f f_{ \alpha_j } & = \bf f_{\ a l p h a_j} \ bff_ {\alp ha_i}, \ \qq uad \qqu a d\\ \ \ \qqu ad & \t ext{i f }& | i-j|> 1, \displaybreak[ 0]\\ \be_{\ alp ha _i} \b t &=\ b t\be_{ \al pha _i}, \q uad\qqu a d\q ua d & \t ext{if }& i > 1,\d i s pl aybreak[ 0]\\ \b e_ { \alpha_1 }^ 2\b t +\ b t\be_ {\al p ha _1}^2 &= (q+q^ { -1 }) \be_{\ al pha_1} \b t\b e_{ \alph a _1}, \displ aybreak[ 0]\\ \bt^2\be_{\al p ha_1} + \be_{ \ al p h a_ 1 }\bt ^2&= (q + q^{ -1}) \bt\ be_{ \ al pha _ 1}\bt + \b e_ { \a l pha_1},\displaybrea k[ 0]\\ \bff _{\alpha_i}\b t &=\bt\bf f _ { \alpha_i }, \ q ua d \qquad&\text{i f }& i > 1, \d i splaybre ak[0] \\ \bf f_{\alpha _ 1 }^2\bt + \b t\b ff
\be_{\alpha_j}_+\be_{\alpha_j} \be_{\alpha_i}^2_&= (q+q^{-1}) \be_{\alpha_i} \be_{\alpha_j}_\be_{\alpha_i}, __ \_\_\quad\quad &\text{if }&_|i-j|=1, \displaybreak[0]\\ _\be_{\alpha_i} \be_{\alpha_j} &= \be_{\alpha_j}_\be_{\alpha_i}, \ \qquad\qquad\_\_\ \ \qquad &\text{if }& |i-j|>1, \displaybreak[0]\\ \bff_{\alpha_i}^2 \bff_{\alpha_j} +\bff_{\alpha_j} \bff_{\alpha_i}^2 &= (q+q^{-1})_\bff_{\alpha_i}_\bff_{\alpha_j} \bff_{\alpha_i}, ___ \ \ \quad\quad &\text{if_}& |i-j|=1,\displaybreak[0]\\ \bff_{\alpha_i} \bff_{\alpha_j}_ &=_\bff_{\alpha_j} \bff_{\alpha_i}, \ \qquad\qquad\ \ \_\_\qquad &\text{if }&_|i-j|>1, \displaybreak[0]\\ \be_{\alpha_i}\bt &=\bt\be_{\alpha_i}, \quad\qquad\quad &\text{if }& _i > 1, \displaybreak[0]\\ \be_{\alpha_1}^2\bt_+ \bt\be_{\alpha_1}^2 &=_(q+q^{-1})_\be_{\alpha_1}\bt\be_{\alpha_1},\displaybreak[0]\\ _ \bt^2\be_{\alpha_1} + \be_{\alpha_1}\bt^2_&= (q + q^{-1}) \bt\be_{\alpha_1}\bt +_\be_{\alpha_1},\displaybreak[0]\\ \bff_{\alpha_i}\bt &=\bt\bff_{\alpha_i}, \quad\qquad&\text{if }&_ i > 1, \displaybreak[0]\\ \bff_{\alpha_1}^2\bt_+ \bt\bff
{{\sf hub}}({u}) = \sum_{{v}\in {V}} {{{\mathit wt}}{({{u}{\to}{v}})}}\cdot {{\sf auth}}({v}). \label{eq:hub}$$ Clearly, Equations \[eq:auth\] and \[eq:hub\] are mutually recursive. However, the iterative HITS algorithm[^1] provably converges to (non-identically-zero, non-negative) score functions ${{\sf hub}}^*$ and ${{\sf auth}}^*$ that satisfy the above pair of equations. Figure \[fig:hits-icon\] depicts the “iconic” case in which the input graph ${G}$ is [*[one-way bipartite]{}*]{}, that is, ${V}$ can be partitioned into non-empty sets ${{V}_{\rm Left}}$ and ${{V}_{\rm Right}}$ such that only edges in ${{V}_{\rm Left}}\times {{V}_{\rm Right}}$ can receive positive weight, and $\forall {u}\in {{V}_{\rm Left}}$, $\sum_{{v}\in {{V}_{\rm Right}}} {{{\mathit wt}}{({{u}{\to}{v}})}} > 0$. It is the case that ${{\sf auth}}^*({u}) = 0$ for every ${u}\in {{V}_{\rm Left}}$ and ${{\sf hub}}^*({v})=0$ for every ${v}\in {{V}_{\rm Right}}$; in this sense, the left-hand nodes are “pure” hubs and the right-hand nodes are “pure” authorities. Note that in the end, we need to produce a [*single*]{} [centrality]{}score for each node ${n}\in {V}$. For experimental simplicity, we consider only two possibilities in this paper — using ${{\sf auth}}^*({n})$ as the final [centrality]{}score, or using ${{\sf hub}}^*({n})$ instead — although combining the hub and authority scores is also an interesting possibility. Graph schemata: incorporating clusters {#sec:schemata} -------------------------------------- Recall that the fundamental operation in our [structural ing]{}paradigm is to compute the [centrality]{}of entities (with)in a set ${{\cal S}_{{{\rm init}}}}$. One possibility is to define ${{\cal S}_{{{\rm init}}}}$ as ${{\cal D}_{{{\rm init}}}}$, the documents in the initially retrieved set; we refer generically to any [relevance-flow graph]{}induced under this choice as
{ { \sf hub}}({u }) = \sum_{{v}\in { V } } { { { \mathit wt}}{({{u}{\to}{v}})}}\cdot { { \sf auth}}({v }). \label{eq: hub}$$ Clearly, Equations \[eq: auth\ ] and \[eq: hub\ ] are mutually recursive. However, the iterative HITS algorithm[^1 ] provably converges to (non - identically - zero, non - damaging) sexual conquest functions $ { { \sf hub}}^*$ and $ { { \sf auth}}^*$ that satisfy the above pair of equality. Figure \[fig: hits - icon\ ] depicts the “ iconic ” character in which the remark graph $ { G}$ is [ * [ one - way bipartite ] { } * ] { }, that is, $ { V}$ can be partitioned into non - empty sets $ { { V}_{\rm Left}}$ and $ { { V}_{\rm Right}}$ such that only edge in $ { { V}_{\rm Left}}\times { { V}_{\rm Right}}$ can receive positive weight unit, and $ \forall { u}\in { { V}_{\rm Left}}$, $ \sum_{{v}\in { { V}_{\rm Right } } } { { { \mathit wt}}{({{u}{\to}{v } }) } } > 0$. It is the shell that $ { { \sf auth}}^*({u }) = 0 $ for every $ { u}\in { { V}_{\rm Left}}$ and $ { { \sf hub}}^*({v})=0 $ for every $ { v}\in { { V}_{\rm Right}}$; in this sense, the left - hand nodes are “ pure ” hub and the right - hand nodes are “ saturated ” authorities. Note that in the end, we need to grow a [ * single * ] { } [ centrality]{}score for each node $ { n}\in { V}$. For experimental simplicity, we consider only two possibilities in this newspaper — using $ { { \sf auth}}^*({n})$ as the final [ centrality]{}score, or using $ { { \sf hub}}^*({n})$ instead — although combining the hub and authority scores is also an interesting possibility. Graph schemata: incorporating clusters { # sec: schemata } -------------------------------------- Recall that the fundamental operation in our [ structural ing]{}paradigm is to calculate the [ centrality]{}of entity (with)in a set $ { { \cal S}_{{{\rm init}}}}$. One possibility is to define $ { { \cal S}_{{{\rm init}}}}$ as $ { { \cal D}_{{{\rm init}}}}$, the document in the initially retrieved set; we refer generically to any [ relevance - menstruation graph]{}induced under this choice as
{{\sf jub}}({u}) = \sum_{{v}\in {V}} {{{\mathit wt}}{({{m}{\to}{v}})}}\cdot {{\sf auth}}({r}). \oabel{ex:hub}$$ Cmearly, Eduations \[eq:auth\] and \[eq:hub\] arx muruallt recursive. However, thd iteratine HITS aogormthm[^1] provably coiberges bj (noh-ldentncelly-zero, non-negstive) scora functions ${{\sf hjb}}^*$ and ${{\sf auth}}^*$ that satisfy the above pair og fquations. Figurg \[fig:nyts-idon\] depicts the “iconic” case in whidh the pnput graph ${G}$ is [*[pne-way bipartite]{}*]{}, that is, ${V}$ can be partitioned inho non-empty setf ${{V}_{\rm Left}}$ ana ${{V}_{\rm Righu}}$ vuch that knly edges in ${{V}_{\rm Left}}\times {{V}_{\rm Righc}}$ can receice podhtive weighv, and $\sorall {u}\in {{V}_{\vk Left}}$, $\sum_{{v}\in {{V}_{\rm Right}}} {{{\matmit wv}}{({{u}{\to}{c}})}} > 0$. It is the case thet ${{\sf auth}}^*({u}) = 0$ for evgry ${u}\in {{V}_{\rk Meft}}$ and ${{\sf hub}}^*({v})=0$ doe evety ${v}\it {{V}_{\ro Rieht}}$; ii tgis sejse, the left-hznd nodes aee “pure” hubs and tht rytht-hand nodes are “ptrq” authorities. Note that in the end, we netd to produce a [*single*]{} [centrqlity]{}score for each nlde ${n}\in {V}$. For experimental simplicity, we consider only two possmbkliuicf un this paper — using ${{\sf auth}}^*({n})$ as the final [centwzlotj]{}score, or using ${{\sn hub}}^*({n})$ instead — alyhlubr combining tfe hub ans authority scores is alsj an unterestigg ppssibility. Graph schemata: inxorporating blusrers {#sec:schemata} -------------------------------------- Reeall that thz fundsmentsl operation in our [strbcturam ing]{}paradihm is to dumpute the [centrxliny]{}of entities (with)in a set ${{\cal S}_{{{\rm init}}}}$. One possibkliti is to define ${{\cap S}_{{{\rm init}}}}$ as ${{\cal D}_{{{\rm inlt}}}}$, thg docukents in tje initially retrieved set; we rxher genericalky to any [relzvance-nlow graph]{}inducqd under this ehoice af
{{\sf hub}}({u}) = \sum_{{v}\in {V}} {{{\mathit wt}}{({{u}{\to}{v}})}}\cdot \label{eq:hub}$$ Equations \[eq:auth\] \[eq:hub\] are mutually algorithm[^1] converges to (non-identically-zero, score functions ${{\sf and ${{\sf auth}}^*$ that satisfy the pair of equations. Figure \[fig:hits-icon\] depicts the “iconic” case in which the input ${G}$ is [*[one-way bipartite]{}*]{}, that is, ${V}$ can be partitioned into non-empty sets Left}}$ ${{V}_{\rm such only edges in ${{V}_{\rm Left}}\times {{V}_{\rm Right}}$ can receive positive weight, and $\forall {u}\in {{V}_{\rm Left}}$, {{V}_{\rm Right}}} {{{\mathit wt}}{({{u}{\to}{v}})}} > 0$. It is case that ${{\sf auth}}^*({u}) 0$ for every ${u}\in {{V}_{\rm and hub}}^*({v})=0$ for ${v}\in Right}}$; this sense, the nodes are “pure” hubs and the right-hand nodes are “pure” authorities. Note that in the end, we to produce [centrality]{}score for node {V}$. experimental simplicity, we two possibilities in this paper — as the final [centrality]{}score, or using ${{\sf hub}}^*({n})$ — although the hub and authority scores is an interesting possibility. Graph schemata: incorporating clusters {#sec:schemata} Recall that the fundamental operation in our [structural ing]{}paradigm is to compute the [centrality]{}of entities set ${{\cal S}_{{{\rm init}}}}$. possibility is to ${{\cal init}}}}$ ${{\cal init}}}}$, the in the initially retrieved set; we refer generically to any [relevance-flow under this choice as
{{\sf hub}}({u}) = \sum_{{v}\in {V}} {{{\mathit wt}}{({{u}{\to}{v}})}}\Cdot {{\sf auth}}({V}). \labeL{eq:Hub}$$ clEarlY, EquAtions \[eq:auth\] anD \[Eq:huB\] are mutually recursive. HOweveR, tHE iteRAtIve HItS algorIThM[^1] PRovAbLy ConVeRGeS to (noN-idEnticalLy-zero, non-nEgaTiVe) score functIOnS ${{\sf hub}}^*$ and ${{\sF auTh}}^*$ that satisfY thE above PaIr oF EquatIonS. FiguRe \[fig:hITs-icon\] Depicts thE “iCOnic” caSE in whicH THe InpuT graph ${G}$ is [*[one-way biPArTIte]{}*]{}, that is, ${V}$ can bE partiTiONeD INto Non-Empty sets ${{V}_{\Rm left}}$ aND ${{V}_{\rm RigHT}}$ sUCH ThaT Only edges in ${{V}_{\rM Left}}\times {{V}_{\RM RiGht}}$ can ReCeiVE positIve weIgHT, anD $\forall {u}\in {{V}_{\Rm LeFt}}$, $\sum_{{v}\in {{V}_{\Rm RighT}}} {{{\Mathit wT}}{({{U}{\to}{v}})}} > 0$. It iS the caSe tHat ${{\Sf auTH}}^*({u}) = 0$ FoR evErY ${U}\in {{v}_{\Rm lefT}}$ And ${{\Sf hub}}^*({v})=0$ foR eVeRy ${v}\in {{v}_{\rm RIGHT}}$; In thIs sEnse, The leFt-hand nodes arE “puRe” huBS anD the rIght-hAnd nOdEs are “Pure” auThoriTiEs. Note that in the End, wE need to prOduCe A [*siNgLe*]{} [cenTRality]{}ScoRe fOr each nOde ${n}\in {V}$. fOr eXpERIMeNtal simplicity, we coNsIDEr Only two pOssibiLItIeS In this paPeR — usIng ${{\sF AUth}}^*({n})$ aS the FInAl [centraLity]{}scORe, Or Using ${{\sf HuB}}^*({n})$ instEaD — alThoUgh coMBiniNg the hUb and autHoritY Scores is also an INteresting posSIbILItY. grapH scHemata: incorPoraTIng cLustERs {#Sec:SChemaTa} -------------------------------------- RecAlL ThAT the fundamental operAtIon in oUr [strUctural ing]{}parAdigm is to cOMPUte the [ceNtraLItY]{}Of entities (with)In a seT ${{\cal S}_{{{\rm iniT}}}}$. one possiBilitY is to defIne ${{\cal S}_{{{\rm INIt}}}}$ as ${{\cal D}_{{{\Rm iNit}}}}$, The DocUMEnTs in the initiaLLY retRiEved set; We rEfer genEriCalLy tO anY [rElevance-fLow graph]{}InDuCeD uNdeR this CHoice as
{{\sf hub}}({u}) = \sum_{{ v}\in {V}} {{{ \ma thi twt}} {({{ u}{\to}{v}})}} \ cdot {{\sf auth}}({v}). \l abel{ eq : hub} $ $Clear ly, Equ a ti o n s \ [e q: aut h\ ] a nd \[ eq: hub\] a re mutuall y r ec ursive. Howe v er , the iter ati ve HITS algo rit hm[^1] p rov a bly c onv erges to (n o n-iden tically-z er o , non- n egative ) sc orefunctions ${{\sfh ub } }^*$ and ${{\s f auth }} ^ *$ t hat sa tisfy theab ove p a ir of e q ua t i o ns. Figure \[fig :hits-icon\ ] de pictsth e “ i conic” case i n wh ich the inp ut g raph ${G} $ is [ * [one-wa y bipart ite]{} *]{ },that is ,${V }$ can be pa r tit ioned in to n on-em ptys e t s ${{ V}_ {\rm Left }}$ and ${{V} _{\ rm R i ght }}$ s uch t haton ly ed ges in ${{V }_ {\rm Left}}\tim es { {V}_{\rmRig ht }}$ c an re c eive p osi tiv e weigh t, and$ \fo ra l l {u }\in {{V}_{\rm Lef t} } $ ,$\sum_{{ v}\in{ {V }_ { \rm Righ t} }}{{{\ m a thitwt}} { ({ {u}{\to} {v}})} } > 0 $. It i sthe ca se th at${{\s f aut h}}^*( {u}) = 0 $ for every ${u}\in{ {V}_{\rm Left } }$ a nd ${{\ sfhub}}^*({v} )=0$ forever y $ {v} \ in {{ V}_{\ rm Ri g ht}}$; in this sens e, the l eft-h and nodes are “pure” hu b s and therigh t -h a nd nodes are “ pure” authoriti e s. Note that in theend, we n e e d to pro duc e a [* sin g l e* ]{} [centrali t y ]{}s co re foreac h node${n }\i n { V}$ .For exper imentalsi mp li ci ty, we c o nsider o nl y t wo po ssibi l itiesin th is p ap er — u sing ${ { \s f auth }} ^* ({n} )$as thefina l [c entrali ty]{}scor e,o r us in g${{\sfhub}}^*({n})$ i nstead — a lt hou gh com b i ning the hub and authority scor e s is al soan in tere sting pos sib ility. G r aph sc hemata : inc or por a t ing c l u st ers { #sec:schem a t a}----- -- ---- ------- ------------------ - - Recall that t hefund a m en tal op e rat io n in o ur [structuraling]{}para di g mis to comp u teth e [cent rality] {}ofe ntities (with)in a set ${ {\ calS } _{{ {\rm init} }}}$. On e possibi l ity i s t o def ine ${{\c al S} _{{{\ rm ini t }}} }$ as ${{\c al D}_{{ {\rmin it}}}}$, the documents in the i nitial ly re tri eved set; we ref er generi call y to any [ rel eva nce-f low graph ]{}i n du ced under thi s choice a s
{{\sf hub}}({u})_= \sum_{{v}\in_{V}} {{{\mathit wt}}{({{u}{\to}{v}})}}\cdot {{\sf_auth}}({v}). \label{eq:hub}$$ Clearly,_Equations_\[eq:auth\] and_\[eq:hub\]_are mutually recursive._However, the iterative_HITS algorithm[^1] provably converges_to (non-identically-zero, non-negative)_score_functions ${{\sf hub}}^*$ and ${{\sf auth}}^*$ that satisfy the above pair of equations. Figure \[fig:hits-icon\]_depicts_the “iconic”_case_in_which the input graph ${G}$_is [*[one-way bipartite]{}*]{}, that is,_${V}$ can_be partitioned into non-empty sets ${{V}_{\rm Left}}$ and_${{V}_{\rm_Right}}$ such that_only edges in ${{V}_{\rm Left}}\times {{V}_{\rm Right}}$ can receive_positive weight, and $\forall {u}\in {{V}_{\rm_Left}}$, $\sum_{{v}\in {{V}_{\rm_Right}}}_{{{\mathit_wt}}{({{u}{\to}{v}})}} > 0$. It_is the case that ${{\sf auth}}^*({u})_= 0$ for every ${u}\in {{V}_{\rm_Left}}$ and ${{\sf hub}}^*({v})=0$ for every ${v}\in_{{V}_{\rm Right}}$; in this sense, the_left-hand nodes are “pure” hubs_and the_right-hand nodes are “pure” authorities. Note_that in the_end, we_need to produce_a [*single*]{} [centrality]{}score for each node_${n}\in {V}$. For_experimental simplicity, we consider only two_possibilities_in this paper_—_using_${{\sf auth}}^*({n})$_as the final_[centrality]{}score,_or using_${{\sf_hub}}^*({n})$ instead — although combining the_hub_and authority scores is also an interesting_possibility. Graph schemata: incorporating clusters_{#sec:schemata} -------------------------------------- Recall_that the fundamental operation_in our [structural ing]{}paradigm is_to compute the [centrality]{}of entities (with)in_a set_${{\cal S}_{{{\rm_init}}}}$. One possibility is to define ${{\cal S}_{{{\rm init}}}}$ as ${{\cal_D}_{{{\rm init}}}}$, the documents in the_initially retrieved set; we_refer generically_to_any [relevance-flow graph]{}induced_under_this choice_as
9}$Fe$_{0.1}$O$_3$. CONCLUSIONS =========== The electronic structures of of multiferroic oxides of Ba$_{1-x}$Bi$_x$Ti$_{0.9}$Fe$_{0.1}$O$_3$ ($0 \le x \le 0.12$). have been investigated by employing synchrotron-radiation excited PES and XAS. Via Fe and Ti $2p$ XAS measurements, the valence states of Fe and Ti ions have been determined experimentally. The valence states of Fe ions are found to be Fe$^{2+}$-Fe$^{3+}$ mixed-valent for $x>0$ but nearly trivalent for $x$=0. The valence states of Fe ions are found to decreases from being nearly trivalent for $x$=0 ($v$(Fe)$\sim 3$) to $v$(Fe)$\sim 2.6$ for $x$=0.12. The valence states of Ti ions do not change with $x$ for $x \le 0.12$, and stay being tetravalent (Ti$^{4+}$). The valence states of Ba ions are close to being divalent (Ba$^{2+}$) and remain unchanged for $x \le 0.12$. The decreasing trend of valence states of Fe ions in Ba$_{1-x}$Bi$_x$Ti$_{0.9}$Fe$_{0.1}$O$_3$ with $x$ will play an important role in the electronic and magnetic properties of this system. This work was supported by the NRF under Contract No. 2014R1A1A2056546, and in part by the Research Fund, 2015 of the Catholic University of Korea. Experiments at PLS were supported by MSIP and PAL. Y. Tokura and S. Seki, Advanced Materials [**22**]{}, 1554 (2010). Y. Kimura, T. Goto, H. Shintani, K Ishizaka, T. Arima and Y. Tokura, Nature [**426**]{}, 55 (2003). N. Ikeda, H. Ohsumi, K. Ohwada, K. Ishii, T. Inami, K. Kakurai, Y. Murakami, K. Yoshii, S. Mori, Y. Horibe and H. Kito, Nature (London) [**436**]{}, 1136 (2005). A. Rajamani, G. F. Dion
9}$Fe$_{0.1}$O$_3$. CONCLUSIONS = = = = = = = = = = = The electronic structures of of multiferroic oxides of Ba$_{1 - x}$Bi$_x$Ti$_{0.9}$Fe$_{0.1}$O$_3 $ ($ 0 \le x \le 0.12 $). have been investigated by use synchrotron - radiation sickness excited PES and XAS. Via Fe and Ti $ 2p$ XAS measurements, the valence states of Fe and Ti ion have been determined experimentally. The valence states of Fe ions are establish to be Fe$^{2+}$-Fe$^{3+}$ mixed - valent for $ x>0 $ but nearly trivalent for $ x$=0. The valence states of Fe ions are found to decrease from being nearly trivalent for $ x$=0 ($ v$(Fe)$\sim 3 $) to $ v$(Fe)$\sim 2.6 $ for $ x$=0.12. The valence states of Ti ion do not change with $ x$ for $ x \le 0.12 $, and stay being tetravalent (Ti$^{4+}$). The valence states of Ba ions are close to being divalent (Ba$^{2+}$) and remain unaltered for $ x \le 0.12$. The decreasing trend of valence states of Fe ion in Ba$_{1 - x}$Bi$_x$Ti$_{0.9}$Fe$_{0.1}$O$_3 $ with $ x$ will play an important role in the electronic and magnetic properties of this arrangement. This work was supported by the NRF under Contract No. 2014R1A1A2056546, and in part by the Research Fund, 2015 of the Catholic University of Korea. Experiments at PLS were support by MSIP and PAL. Y. Tokura and S. Seki, Advanced Materials [ * * 22 * * ] { }, 1554 (2010). Y. Kimura, T. Goto, H. Shintani, K Ishizaka, T. Arima and Y. Tokura, Nature [ * * 426 * * ] { }, 55 (2003). N. Ikeda, H. Ohsumi, K. Ohwada, K. Ishii, T. Inami, K. Kakurai, Y. Murakami, K. Yoshii, S. Mori, Y. Horibe and H. Kito, Nature (London) [ * * 436 * * ] { }, 1136 (2005). A. Rajamani, G. F. Dion
9}$Fe$_{0.1}$O$_3$. FONCLUSIONS =========== The electronig structures of of multmferroid oxides of Ba$_{1-x}$Bi$_x$Ti$_{0.9}$Fe$_{0.1}$O$_3$ ($0 \le x \le 0.12$). hate bwen ibvestigated by employivg synchrltron-raduatiib excited 'SS and WCS. Viz Fe cnv Ti $2p$ XAS measorements, the valence statev uf Fe and Ti ions have been determineq experomfntally. The vajenct seatea of Fe ions are found to be Fe$^{2+}$-Fe$^{3+}$ jixed-vanent for $x>0$ buy nearly trivalent for $x$=0. Tje vwlence states of Ff ions are dounq to decreaser from being nearly trjvalent for $x$=0 ($v$(Fe)$\sim 3$) to $v$(Fe)$\sim 2.6$ for $x$=0.12. The valguxe djates of Ti mons dj not change with $x$ xor $x \lr 0.12$, and stay belng txtracalent (Ti$^{4+}$). The valence states of Ba ions are closa co being divalent (Ba$^{2+}$) qne remdin gnchxbgea fkr $x \le 0.12$. Tje vecreasing frend of vaoence states of Fe oogw in Ba$_{1-x}$Bi$_x$Ti$_{0.9}$Re$_{0.1}$O$_3$ wieh $x$ will play an important role in the enecfronic and magnetic properties of this systel. This worh was supported by the NRF under Contract No. 2014R1A1A2056546, dnd ii oaru nn thd Rfsearch Fund, 2015 of the Catholic University of Kjdes. Vxperiments at PLF were suppprhec by MSIP and KAL. Y. Tokurz and S. Seki, Advanfed Matgrials [**22**]{}, 1554 (2010). Y. Kimtra, Y. Goto, H. Shintani, K Ishizakq, T. Arima anb Y. Tokura, Nature [**426**]{}, 55 (2003). U. Ikeda, H. Ohrumi, K. Oheada, K. Ishii, T. Inami, K. Yakudai, Y. Murakwmi, K. Yosgki, S. Mori, Y. Horice sng H. Kito, Nature (London) [**436**]{}, 1136 (2005). W. Rajamanm, G. F. Dion
9}$Fe$_{0.1}$O$_3$. CONCLUSIONS =========== The electronic structures of oxides Ba$_{1-x}$Bi$_x$Ti$_{0.9}$Fe$_{0.1}$O$_3$ ($0 x \le 0.12$). synchrotron-radiation PES and XAS. Fe and Ti XAS measurements, the valence states of and Ti ions have been determined experimentally. The valence states of Fe ions found to be Fe$^{2+}$-Fe$^{3+}$ mixed-valent for $x>0$ but nearly trivalent for $x$=0. The states Fe are to decreases from being nearly trivalent for $x$=0 ($v$(Fe)$\sim 3$) to $v$(Fe)$\sim 2.6$ for $x$=0.12. The states of Ti ions do not change with for $x \le 0.12$, stay being tetravalent (Ti$^{4+}$). The states Ba ions close being (Ba$^{2+}$) and remain for $x \le 0.12$. The decreasing trend of valence states of Fe ions in Ba$_{1-x}$Bi$_x$Ti$_{0.9}$Fe$_{0.1}$O$_3$ with $x$ play an in the and properties this system. This supported by the NRF under Contract in part by the Research Fund, 2015 of Catholic University Korea. Experiments at PLS were supported MSIP and PAL. Y. Tokura and S. Seki, Materials [**22**]{}, 1554 (2010). Y. Kimura, T. Goto, H. Shintani, K Ishizaka, T. Arima and Nature [**426**]{}, 55 (2003). Ikeda, H. Ohsumi, Ohwada, Ishii, Inami, Kakurai, Y. K. Yoshii, S. Mori, Y. Horibe and H. Kito, Nature (London) 1136 (2005). A. Rajamani, G. F. Dion
9}$Fe$_{0.1}$O$_3$. CONCLUSIONS =========== The electronIc structurEs of oF muLtiFeRroiC oxiDes of Ba$_{1-x}$Bi$_x$Ti$_{0.9}$FE$_{0.1}$o$_3$ ($0 \le x \Le 0.12$). have been investigated By empLoYIng sYNcHrotrOn-radiaTIoN EXciTeD PeS aNd xaS. via Fe And ti $2p$ XAS mEasurementS, thE vAlence states OF FE and Ti ions HavE been determiNed ExperiMeNtaLLy. The ValEnce sTates oF fe ions Are found tO bE fe$^{2+}$-Fe$^{3+}$ miXEd-valenT FOr $X>0$ but Nearly trivalent foR $X$=0. THE valence states Of Fe ioNs ARe FOUnd To dEcreases frOm Being NEarly trIVaLENT foR $X$=0 ($v$(Fe)$\sim 3$) to $v$(Fe)$\sIm 2.6$ for $x$=0.12. The vaLEncE stateS oF Ti IOns do nOt chaNgE WitH $x$ for $x \le 0.12$, and Stay Being tetrAvalenT (ti$^{4+}$). The vaLEnce staTes of BA ioNs aRe clOSe To BeiNg DIvaLEnT (Ba$^{2+}$) ANd rEmain uncHaNgEd for $X \le 0.12$. THE DECreaSinG treNd of vAlence states oF Fe Ions IN Ba$_{1-X}$Bi$_x$TI$_{0.9}$Fe$_{0.1}$O$_3$ wIth $x$ WiLl plaY an impOrtanT rOle in the electroNic aNd magnetiC prOpErtIeS of thIS systeM. ThIs wOrk was sUpporteD By tHe nrf UnDer Contract No. 2014R1A1A2056546, anD iN PArT by the ReSearch fUnD, 2015 oF The CathoLiC UnIverSITy of KOrea. eXpEriments At PLS wERe SuPported By mSIP anD PaL. Y. tokUra anD s. SekI, AdvanCed MaterIals [**22**]{}, 1554 (2010). Y. kImura, T. Goto, H. ShiNTani, K Ishizaka, t. arIMA aND Y. ToKurA, Nature [**426**]{}, 55 (2003). N. IkeDa, H. OHSumi, k. OhwADa, k. IsHIi, T. InAmi, K. KAkURaI, y. Murakami, K. Yoshii, S. MoRi, y. HoribE and H. kito, Nature (LonDon) [**436**]{}, 1136 (2005). A. RajamaNI, g. f. Dion
9}$Fe$_{0.1}$O$_3$. CONCL USIONS === ===== === T he ele ctro nic structures of o f multiferroic oxidesof Ba $_ { 1-x} $ Bi $_x$T i$_{0.9 } $F e $ _{0 .1 }$ O$_ 3$ ($ 0 \le x\le 0.1 2$). havebee ninvestigated by employing sy nchrotron-ra dia tion e xc ite d PESand XAS. Via F e and T i $2p$ XA Sm easure m ents, t h e v alen ce states of Fe a n dT i ions have be en det er m in e d ex per imentally. T he va l ence st a te s o f F e ions are fou nd to be Fe $ ^{2 +}$-Fe $^ {3+ } $ mixe d-val en t fo r $x>0$ but nea rly triva lent f o r $x$=0 . The va lencesta tes ofF eio nsar e fo u nd to dec reases f ro mbeing nea r l y triv ale nt f or $x $=0 ($v$(Fe)$ \si m 3$ ) to $v$( Fe)$\ sim2. 6$ fo r $x$= 0.12. T he valence stat es o f Ti ions do n otch angew ith $x $ f or$x \le0.12$,a ndst a y be ing tetravalent (T i$ ^ { 4+ }$). The valen c est a tes of B aion s ar e close tob ei ng dival ent (B a $^ {2 +}$) an dremain u nch ang ed fo r $x\le 0. 12$. The decr e asing trend of valence state s o f Fe ions in Ba$_{1-x}$ Bi$_ x $Ti$ _{0. 9 }$ Fe$ _ {0.1} $O$_3 $w it h $x$ will play an i mp ortant role in the elect ronic andm a g netic pr oper t ie s of this syste m. T his work w a s suppor ted b y the NR F under C o n tract No . 2 014 R1A 1A2 0 5 65 46, and in pa r t byth e Resea rch Fund,201 5 o f t heCa tholic Un iversity o fKo re a.Exper i ments at P LSwe resuppo r ted by MSIP and P AL . Y . Tokur a a n d S.Se ki , Ad van ce d Mat eria l s [ **22**] {}, 1554(20 1 0).Y. K imura,T. Goto, H. S hi ntani, K I sh iza ka, T. A rima and Y. Tokura, Nature [**4 2 6**]{}, 55 (200 3).N. Ikeda, H. Ohsum i,K . Ohwa da, K. Ishi i, T. I nami, K .Kak ur ai, Y. Mur a k ami , K.Yo shii , S. Mo ri, Y. Horibe andH . K ito, Nature ( Lon don) [ ** 436 * *] { },11 3 6 ( 2 0 05). A. Rajaman i, G. F. D io n
9}$Fe$_{0.1}$O$_3$. CONCLUSIONS =========== The electronic_structures of_of multiferroic oxides of_Ba$_{1-x}$Bi$_x$Ti$_{0.9}$Fe$_{0.1}$O$_3$ ($0_\le_x \le_0.12$)._have been investigated_by employing synchrotron-radiation_excited PES and XAS._Via Fe and_Ti_$2p$ XAS measurements, the valence states of Fe and Ti ions have been determined_experimentally._The valence_states_of_Fe ions are found to_be Fe$^{2+}$-Fe$^{3+}$ mixed-valent for $x>0$_but nearly_trivalent for $x$=0. The valence states of Fe_ions_are found to_decreases from being nearly trivalent for $x$=0 ($v$(Fe)$\sim 3$)_to $v$(Fe)$\sim 2.6$ for $x$=0.12. The_valence states of_Ti_ions_do not change with_$x$ for $x \le 0.12$, and_stay being tetravalent (Ti$^{4+}$). The valence_states of Ba ions are close to_being divalent (Ba$^{2+}$) and remain unchanged_for $x \le 0.12$. The_decreasing trend_of valence states of Fe_ions in Ba$_{1-x}$Bi$_x$Ti$_{0.9}$Fe$_{0.1}$O$_3$_with $x$_will play an_important role in the electronic and_magnetic properties of_this system. This work was supported by_the_NRF under Contract_No._2014R1A1A2056546,_and in_part by the_Research_Fund, 2015_of_the Catholic University of Korea. Experiments_at_PLS were supported by MSIP and PAL. Y._Tokura and S. Seki,_Advanced_Materials [**22**]{}, 1554 (2010)._Y. Kimura, T. Goto, H._Shintani, K Ishizaka, T. Arima and_Y. Tokura,_Nature [**426**]{},_55 (2003). N. Ikeda, H. Ohsumi, K. Ohwada, K. Ishii, T._Inami, K. Kakurai, Y. Murakami, K._Yoshii, S. Mori, Y._Horibe and_H._Kito, Nature (London)_[**436**]{},_1136 (2005)._A. Rajamani, G. F. Dion
Finally, we summarize in Section \[sec:discussion\]. Models expressed in a programming language\[sec:models\] ======================================================== Probabilistic programming considers models expressed in Turing-complete programming languages. Such models are usually referred to as *probabilistic programs*—qualified to *universal probabilistic programs* when one wishes to regard only the broadest class in terms of expressivity—and described in programming language nomenclature. Here, we adopt probabilistic nomenclature instead, to provide a more accessible treatment for the intended audience. Taking the lead from the term *graphical model*—a model expressed in a graphical language—we suggest that the term *programmatic model*—a model expressed in a programming language—might be more appropriate for this audience, and adopt this term throughout. Specifically, we avoid the use of the term *program* when referring only to a model implementation, as in ordinary usage one thinks of a computer program as combining the implementation of both a model and an inference method, which can cause confusion. The term can also be misleading given unrelated but similarly-named concepts in system identification, such as linear programs and stochastic programs. We follow the statistics convention of using uppercase letters to denote random variables (e.g. $V$) and lowercase letters to denote instantiations of them (e.g. $v$), with $v\in\mathbb{V}$. We then adopt measure theory notation to clearly distinguish between distributions (which we will ultimately simulate) and likelihood functions (which we will ultimately evaluate): the distribution of a random variable $V$ is denoted $p(\mathrm{d}v)$, while evaluation of an associated probability density function (pdf, for continuous-valued random variables) or probability mass function (pmf, for discrete-valued random variables) is denoted $p(v)$. Assume that we have a countably infinite set of random variables $\{V_{k}\}_{k=1}^{\infty}$, with a joint probability distribution over them, which has been implemented in code in some programming language. The only stochasticity available to the code is via these random variables. We execute the code, and as it runs it encounters a finite subset of the random variables in some order determined by that code. Denote this order by a permutation $\sigma$, with its (random) length denoted $|\sigma|$, defining a sequence $(V_{\sigma[k]})_{k=1}^{|\sigma|}$. The first element, $\sigma[1]$, is always the same
Finally, we summarize in Section \[sec: discussion\ ]. Models expressed in a programming language\[sec: models\ ] = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = Probabilistic scheduling considers mannequin expressed in Turing - accomplished scheduling languages. Such models are usually consult to as * probabilistic programs*—qualified to * universal probabilistic plan * when one wishes to regard entirely the broadest class in term of expressivity — and described in programming language nomenclature. Here, we adopt probabilistic nomenclature alternatively, to provide a more accessible treatment for the intended consultation. Taking the lead from the term * graphical model*—a exemplar expressed in a graphical language — we suggest that the term * programmatic model*—a exemplar expressed in a programming language — might be more appropriate for this audience, and adopt this term throughout. Specifically, we avoid the use of the term * program * when referring only to a model implementation, as in ordinary usage one thinks of a calculator plan as combining the execution of both a exemplar and an inference method acting, which can cause confusion. The term can also be misleading given unrelated but similarly - named concepts in system designation, such as linear programs and stochastic programs. We follow the statistics convention of using uppercase letters to announce random variables (e.g. $ V$) and lowercase letters to denote instantiations of them (e.g. $ v$), with $ v\in\mathbb{V}$. We then adopt measuring stick theory notation to clearly distinguish between distributions (which we will ultimately imitate) and likelihood functions (which we will ultimately evaluate ): the distribution of a random variable $ V$ is announce $ p(\mathrm{d}v)$, while evaluation of an associated probability density function (pdf, for continuous - valued random variable) or probability mass routine (pmf, for discrete - valued random variables) is denote $ p(v)$. Assume that we suffer a countably infinite stage set of random variables $ \{V_{k}\}_{k=1}^{\infty}$, with a joint probability distribution over them, which has been follow through in code in some programming language. The alone stochasticity available to the code is via these random variable. We execute the code, and as it runs it encounters a finite subset of the random variables in some order determined by that code. Denote this order by a permutation $ \sigma$, with its (random) length denote $ |\sigma|$, defining a sequence $ (V_{\sigma[k]})_{k=1}^{|\sigma|}$. The beginning element, $ \sigma[1]$, is always the like
Fijally, we summarize in Segtion \[sec:discussnin\]. Modens expdessed iv a programming language\[sec:mldwls\] ======================================================== Pribabilistic programmine considegs models expcessed in Turing-rkmplete progrzlminy oanguages. Such models ara usually refesrdd to as *probabilistic programs*—qualifyed to *inlversal probabylisuic profgans* when one wishes to regard knly tht broadest class im terms of expressivity—and desfribed in programmlng languagg noiwnclature. Hefe, we adopu 'robabilistjc nomenclature instead, to provkde a more accewsublf treatment hor thv intended audience. Tdking tne lead from tme tecm *geaphical model*—a model expressed in a grwphical ldnyuage—we suggest that rhw terk *prmgraonatkc jovel*—z modep eepressed in a programmung language—might bt mjgr appropriats for ehys audience, and adopt this term throughmut. Specifically, we avoid rhe use of the term *ptogram* wheg referring only to a model implementation, as in mrdinerh uwane ovw hhinks of a computer program as combining the jmklekentation of bjth a model ajd wn inference oethod, whjch can cause confksion. Tre teem can alfo br misleading given unrelatee but similagly-nqmed concepts in svstem identiyicatipn, suvh as linear programs aud stodhastic prohrams. We fkulow the statistkcs cmnvention of using uppercafe letterw to denote rancom vawiables (e.g. $V$) and lowercase letters to dgnote hnstantiatlons of them (e.g. $v$), with $v\in\mathbu{T}$. We then adokt keavure thejry nptation to clqarly distingunsh betwzen dirtributiona (which we will uleimately simunwte) and likenihood ftnctuons (which dd will ultimatrly evalucue): the disrribution of a ranvom xzriable $V$ is deuuuee $p(\mathrm{d}v)$, whole evwllatmon os an associatad pfobxnilith densiuv fmncgion (pdf, for continuous-vdlues random variables) ov probabioity masf function (pmg, for discrete-valutd ranvom vaciablex) if denoted $p(v)$. Assume that we havs a countwbln infinite see of random varicbles $\{V_{k}\}_{k=1}^{\infty}$, with a joint probability vistribution over them, qhich has been implgmekted in code in sjme progrdmming language. The inly stochasticitn available to the cods is vha thfse random variables. We execute the code, and as it runs it encounters a funite subset of ths ramdom narnabkes in somx urder determined by that code. Denote this order by a pesmbtation $\sigma$, with its (random) ldngth denoted $|\sigma|$, defining a sequende $(V_{\sigms[k]})_{k=1}^{|\sigma|}$. The first element, $\sigms[1]$, is always the same
Finally, we summarize in Section \[sec:discussion\]. Models a language\[sec:models\] ======================================================== programming considers models Such are usually referred as *probabilistic programs*—qualified *universal probabilistic programs* when one wishes regard only the broadest class in terms of expressivity—and described in programming language Here, we adopt probabilistic nomenclature instead, to provide a more accessible treatment for intended Taking lead the term *graphical model*—a model expressed in a graphical language—we suggest that the term *programmatic model*—a expressed in a programming language—might be more appropriate this audience, and adopt term throughout. Specifically, we avoid use the term when only a model implementation, in ordinary usage one thinks of a computer program as combining the implementation of both a model an inference can cause The can be misleading given similarly-named concepts in system identification, such and stochastic programs. We follow the statistics convention using uppercase to denote random variables (e.g. $V$) lowercase letters to denote instantiations of them (e.g. with $v\in\mathbb{V}$. We then adopt measure theory notation to clearly distinguish between distributions (which we simulate) and likelihood functions we will ultimately the of random $V$ is $p(\mathrm{d}v)$, while evaluation of an associated probability density function (pdf, for random variables) or probability mass function (pmf, for discrete-valued random denoted Assume that we a countably infinite set random $\{V_{k}\}_{k=1}^{\infty}$, with a joint over which in in programming language. The only available to the code is these random variables. We it runs it encounters a finite subset of random variables in some order determined by code. Denote this order by a permutation $\sigma$, with its (random) length $|\sigma|$, defining $(V_{\sigma[k]})_{k=1}^{|\sigma|}$. The first element, $\sigma[1]$, is always the
Finally, we summarize in SectiOn \[sec:discuSsion\]. modEls ExPresSed iN a programming lANguaGe\[sec:models\] ======================================================== ProbabilistIc proGrAMminG CoNsideRs modelS ExPREssEd In turInG-CoMpletE prOgrammiNg languageS. SuCh Models are usuALlY referred tO as *ProbabilistiC prOgrams*—QuAliFIed to *UniVersaL probaBIlistiC programs* WhEN one wiSHes to reGARd Only The broadest class iN TeRMs of expressiviTy—and dEsCRiBED in ProGramming laNgUage nOMenclatURe. hERE, we ADopt probabiliStic nomenclATurE insteAd, To pROvide a More aCcESsiBle treatmenT for The intendEd audiENce. TakiNG the leaD from tHe tErm *GrapHIcAl ModEl*—A ModEL eXprESseD in a grapHiCaL langUage—WE SUGgesT thAt thE term *Programmatic mOdeL*—a moDEl eXpresSed in A proGrAmminG languAge—miGhT be more appropriAte fOr this audIenCe, And AdOpt thIS term tHroUghOut. SpecIficallY, We aVoID THe Use of the term *prograM* wHEN rEferring Only to A MoDeL ImplemenTaTioN, as iN ORdinaRy usAGe One thinkS of a coMPuTeR prograM aS combiNiNg tHe iMplemENtatIon of bOth a modeL and aN Inference methoD, Which can cause COnFUSiON. The TerM can also be mIsleADing GiveN UnRelATed buT simiLaRLy-NAmed concepts in systeM iDentifIcatiOn, such as lineaR programs aND STochastiC proGRaMS. We follow the stAtistIcs conventIOn of usinG uppeRcase letTers to denOTE random vAriAblEs (e.G. $V$) aND LoWercase letterS TO denOtE instanTiaTions of TheM (e.g. $V$), wiTh $v\In\Mathbb{V}$. We Then adopT mEaSuRe TheOry noTAtion to cLeArlY dIstInguiSH betweEn disTribUtIoNS (whIch we wiLL uLTImatElY sImulAte) AnD likeLihoOD fuNctions (Which we wiLl uLTimaTeLy EvaluatE): the distributIoN of a random VaRiaBle $V$ is DENoted $p(\maThrm{d}v)$, while evaluation of AN associAteD probAbilIty densitY fuNction (Pdf, FOr contInuous-ValueD rAndOM VariaBLEs) Or pRoBability maSS FunCtion (PmF, for DiscretE-valued random variaBLes) Is denoted $p(v)$. AsSumE thaT WE hAve A CoUNtaBlY InfINIte set of random vAriables $\{V_{k}\}_{K=1}^{\iNFtY}$, with a joinT ProBaBility dIstribuTion oVEr them, wHich has beEn implemeNtEd in CODe iN some progrAmming laNguage. The ONly stOChAsticIty AvailaBlE to The coDe is viA TheSe ranDom varIaBles. We ExecuTe The code, aNd as it runs it encounters a Finite SubseT of The random VarIAblEs in some oRder Determined By tHat Code. DEnoTE this OrdeR By A peRMutatIon $\sIGma$, with itS (RaNdoM) LEnGth denoted $|\sIGMA|$, deFininG a sEQuence $(v_{\sigMa[k]})_{k=1}^{|\sigma|}$. The first ELement, $\sigma[1]$, is aLwayS THe sAme
Finally, we summarize inSection \[ sec:d isc uss io n\]. Mo dels expressed in a programming language\ [sec: mo d els\ ] = ===== ======= = == = = === == == === == = == ===== === ======= ======== Pro ba bilistic pro g ra mming cons ide rs models ex pre ssed i nTur i ng-co mpl ete p rogram m ing la nguages.Su c h mode l s are u s u al ly r eferred to as *pr o ba b ilistic progra ms*—qu al i fi e d to *u niversal p ro babil i stic pr o gr a m s * w h en one wishes to regardo nly the b ro ade s t clas s inte r msof expressi vity —and desc ribedi n progr a mming l anguag e n ome ncla t ur e. He re , we ad opt pro babilist ic n omenc latu r e i nste ad, toprovi de a more acc ess ible tre atmen t for the i ntend ed aud ience .Taking the lead fro m the ter m * gr aph ic al mo d el*—amod elexpress ed in a gra ph i c a llanguage—we sugges tt h at the ter m *pro g ra mm a tic mode l* —amode l expre ssed in a progr amming la ng uage—mi gh t be m or e a ppr opria t e fo r this audienc e, an d adopt this te r m throughout. Sp e c if i call y,we avoid th e us e ofthet er m * p rogra m* wh en re f erring only to a mo de l impl ement ation, as inordinary u s a g e one th inks of a computer pro gramas combini n g the im pleme ntationof both a m odel and an in fer enc e me thod, which c a n cau se confus ion . The t erm ca n a lso b e mislead ing give nun re la ted buts imilarly -n ame dcon cepts in sys tem i dent if ic a tio n, such as l inea rpr ogra msan d sto chas t icprogram s. We fo llo w the s ta tistics convention o fusing uppe rc ase lette r s to deno te random variables (e. g . $V$)and lowe rcas e letters to denot e i n stanti ations of t he m ( e . g. $v $ ) ,wit h$v\in\math b b {V} $. We t henadopt m easure theory nota t ion to clearly d ist ingu i s hbet w ee n di st r ibu t i ons (which we w ill ultima te l ysimulate)a ndli kelihoo d funct ions( which w e will ul timatelyev alua t e ):the distri bution o f a rando m vari a bl e $V$ is denot ed $p (\mat hrm{d} v )$, whil e eval ua tion o f anas sociated probability density fu nction (pdf , f or contin uou s -va lued rand om v ariables)orpro babil ity massfunc t io n ( p mf, f or d i screte-va l ue d r a n do m variables ) i s d enote d $ p (v)$. Ass ume that we havea countably inf init e set of rand om variables $\{ V_{ k} \ } _{k=1}^{ \i nfty}$, wit h a join tp robab ilitydistri butiono v er them,whic h h as been i mpl em e nted in c od e in so me p ro grammi ng lan g uage . The only stochas ticit y avail a ble to t he code i s via these ran dom variabl es. We exe cutethe cod e, and a s i truns it en c ounters a fini te subs et ofthe rando m va r i ables inso meorder det e r mi n ed b y th at c ode.De note this ord e r by a p erm u tation$\ sig m a $, wit h i t s (random)len gth d e n oted $|\si g ma|$ , d e finin g a se quence $(V_{\ s igm a[ k]})_{k =1} ^ { |\sigma|} $. The fi r stel emen t, $\sig ma [1]$ ,isal ways the same
Finally,_we summarize_in Section \[sec:discussion\]. Models expressed_in a_programming_language\[sec:models\] ======================================================== Probabilistic programming_considers_models expressed in_Turing-complete programming languages._Such models are usually_referred to as_*probabilistic_programs*—qualified to *universal probabilistic programs* when one wishes to regard only the broadest class_in_terms of_expressivity—and_described_in programming language nomenclature. Here,_we adopt probabilistic nomenclature instead,_to provide_a more accessible treatment for the intended audience._Taking_the lead from_the term *graphical model*—a model expressed in a graphical_language—we suggest that the term *programmatic_model*—a model expressed_in_a_programming language—might be more_appropriate for this audience, and adopt_this term throughout. Specifically, we avoid_the use of the term *program* when_referring only to a model implementation,_as in ordinary usage one_thinks of_a computer program as combining_the implementation of_both a_model and an_inference method, which can cause confusion._The term can_also be misleading given unrelated but_similarly-named_concepts in system_identification,_such_as linear_programs and stochastic_programs. We_follow the_statistics_convention of using uppercase letters to_denote_random variables (e.g. $V$) and lowercase letters_to denote instantiations of_them_(e.g. $v$), with $v\in\mathbb{V}$._We then adopt measure theory_notation to clearly distinguish between distributions_(which we_will ultimately_simulate) and likelihood functions (which we will ultimately evaluate): the distribution_of a random variable $V$ is_denoted $p(\mathrm{d}v)$, while evaluation_of an_associated_probability density function_(pdf,_for continuous-valued_random variables) or probability mass function (pmf,_for discrete-valued_random variables) is denoted $p(v)$. Assume that_we have a countably_infinite_set of random variables $\{V_{k}\}_{k=1}^{\infty}$, with_a joint probability distribution over them,_which has been implemented in_code_in_some programming language. The only_stochasticity available to the code is_via these random_variables. We execute the code, and as_it_runs it encounters a finite subset_of_the random variables in some order_determined_by_that code. Denote this order_by a permutation $\sigma$, with its_(random) length denoted $|\sigma|$, defining a sequence $(V_{\sigma[k]})_{k=1}^{|\sigma|}$. The_first element, $\sigma[1]$,_is always the same
2inrem\]) and using the identity (\[intidentity1\]), we conclude that $$2ig_0\Phi_2^{(2)} = \frac{8g_0}{\pi} \int_{\gamma} k^3\int_{-t}^t I_2(t,s)e^{2ik^4(s - t)}ds dk = g_0\left(m(t,t) + \frac{i}{2}\bar{g}_0n(t,t)\right).$$ On the other hand, we can write $c^{(3)}$ as the sum of two terms, $c^{(3)} = c_1^{(3)} +c_2^{(3)}$, where $$\label{c13def} c_1^{(3)}= \frac{i}{\pi}\int_{\gamma} k^2\left(\Phi_1(t,k) + i \Phi_1(t, ik) + \frac{ig_0}{k}\right) dk,$$ and $c_2^{(3)}$ involves the quotient $b(k)/a(k)$. Substituting the GLM representation for $\Phi_1$ into (\[c13def\]) and integrating the term in the resulting expression involving $k^5$ by parts, we find $$c_1^{(3)} = \frac{1}{\pi}\int_{\gamma} \left[kn(t,t) - kn(t,-t)e^{-4ik^4t} - \int_{-t}^t kn_{s}(t,s)e^{2ik^4(s-t)}ds -kg_0\right]dk.$$ Using the identity (\[intidentity2\]) and the initial condition $n(t,t) = g_0(t)$, this gives $$\label{c13final} c_1^{(3)} = -c_1 (\mathcal{A}^{-1}n)(t,t).$$ Finally, simplifying $c_2^{(3)}$ as in the proof of theorem \[th2\], we deduce that $c_2^{(3)}$ is given by the terms on the right-hand side of equation (\[g1th2\]) which involve the functions $f_0$ and $f_2$. In summary, after collecting the various contributions, we conclude that (\[g1th2\]) can be derived from (\[g1expression\]). The case of
2inrem\ ]) and using the identity (\[intidentity1\ ]), we conclude that $ $ 2ig_0\Phi_2^{(2) } = \frac{8g_0}{\pi } \int_{\gamma } k^3\int_{-t}^t I_2(t, s)e^{2ik^4(s - t)}ds dk = g_0\left(m(t, t) + \frac{i}{2}\bar{g}_0n(t, t)\right).$$ On the other hand, we can spell $ c^{(3)}$ as the kernel of two terms, $ c^{(3) } = c_1^{(3) } + c_2^{(3)}$, where $ $ \label{c13def } c_1^{(3)}= \frac{i}{\pi}\int_{\gamma } k^2\left(\Phi_1(t, k) + i \Phi_1(t, ik) + \frac{ig_0}{k}\right) dk,$$ and $ c_2^{(3)}$ involves the quotient $ b(k)/a(k)$. Substituting the GLM theatrical performance for $ \Phi_1 $ into (\[c13def\ ]) and integrating the term in the resulting saying involving $ k^5 $ by parts, we recover $ $ c_1^{(3) } = \frac{1}{\pi}\int_{\gamma } \left[kn(t, t) - kn(t,-t)e^{-4ik^4 t } - \int_{-t}^t kn_{s}(t, s)e^{2ik^4(s - t)}ds -kg_0\right]dk.$$ Using the identity (\[intidentity2\ ]) and the initial condition $ n(t, t) = g_0(t)$, this give $ $ \label{c13final } c_1^{(3) } = -c_1 (\mathcal{A}^{-1}n)(t, t).$$ Finally, simplifying $ c_2^{(3)}$ as in the validation of theorem \[th2\ ], we deduce that $ c_2^{(3)}$ is given by the terms on the right - hand side of equation (\[g1th2\ ]) which involve the function $ f_0 $ and $ f_2$. In summary, after collecting the various contributions, we reason that (\[g1th2\ ]) can be derived from (\[g1expression\ ]). The case of
2inrfm\]) and using the identitn (\[intidentity1\]), we concluve that $$2ig_0\Phi_2^{(2)} = \wrac{8g_0}{\pi} \int_{\gamma} k^3\int_{-t}^t I_2(t,s)e^{2mk^4(s - t)}ds ek = g_0\left(m(t,t) + \frac{i}{2}\bar{g}_0v(t,t)\right).$$ Ln the orher yand, we cai write $g^{(3)}$ as fme suk of two terms, $g^{(3)} = c_1^{(3)} +c_2^{(3)}$, whera $$\label{c13def} c_1^{(3)}= \fxac{i}{\pi}\int_{\gamma} k^2\left(\Phi_1(t,k) + i \Phi_1(t, ik) + \frac{ib_0}{k}\gight) dk,$$ and $c_2^{(3)}$ invpjves nht quotient $b(k)/a(k)$. Substituting the FLM repgesentation for $\Pni_1$ into (\[c13def\]) and integratijg tje term in the resklting exprgasijb involving $y^5$ by parts, we find $$c_1^{(3)} = \rrac{1}{\pi}\int_{\gamma} \left[kn(t,t) - kn(t,-t)e^{-4iy^4t} - \iut_{-t}^t kn_{s}(t,s)e^{2uk^4(w-t)}dd -kg_0\right]dk.$$ Nsing nhe identity (\[lmtidendity2\]) anc the initial gondivion $n(t,t) = g_0(t)$, this gives $$\lebel{c13final} c_1^{(3)} = -c_1 (\mwthcal{A}^{-1}n)(t,d).$$ Yinally, simplifying $c_2^{(3)}$ aw in jhe psoof if ghekrxm \[fh2\], we fednce that $c_2^{(3)}$ js given by the terms on the rogrn-nand side of equatyog (\[g1th2\]) which involve the functions $f_0$ and $f_2$. Jn summary, after collecring the various conttibutions, re conclude that (\[g1th2\]) can be derived from (\[g1expresshon\]). Thx zast if
2inrem\]) and using the identity (\[intidentity1\]), we $$2ig_0\Phi_2^{(2)} \frac{8g_0}{\pi} \int_{\gamma} I_2(t,s)e^{2ik^4(s - t)}ds On other hand, we write $c^{(3)}$ as sum of two terms, $c^{(3)} = +c_2^{(3)}$, where $$\label{c13def} c_1^{(3)}= \frac{i}{\pi}\int_{\gamma} k^2\left(\Phi_1(t,k) + i \Phi_1(t, ik) + \frac{ig_0}{k}\right) dk,$$ $c_2^{(3)}$ involves the quotient $b(k)/a(k)$. Substituting the GLM representation for $\Phi_1$ into (\[c13def\]) integrating term the expression involving $k^5$ by parts, we find $$c_1^{(3)} = \frac{1}{\pi}\int_{\gamma} \left[kn(t,t) - kn(t,-t)e^{-4ik^4t} - \int_{-t}^t kn_{s}(t,s)e^{2ik^4(s-t)}ds Using the identity (\[intidentity2\]) and the initial condition = g_0(t)$, this gives c_1^{(3)} = -c_1 (\mathcal{A}^{-1}n)(t,t).$$ Finally, $c_2^{(3)}$ in the of \[th2\], deduce that $c_2^{(3)}$ given by the terms on the right-hand side of equation (\[g1th2\]) which involve the functions $f_0$ and In summary, the various we that can be derived The case of
2inrem\]) and using the identity (\[iNtidentity1\]), We conCluDe tHaT $$2ig_0\PHi_2^{(2)} = \frAc{8g_0}{\pi} \int_{\gamma} k^3\INt_{-t}^t i_2(t,s)e^{2ik^4(s - t)}ds dk = g_0\left(m(t,t) + \frAc{i}{2}\baR{g}_0N(T,t)\riGHt).$$ on the Other haND, wE CAn wRiTe $C^{(3)}$ as ThE SuM of twO teRms, $c^{(3)} = c_1^{(3)} +c_2^{(3)}$, wHere $$\label{c13Def} C_1^{(3)}= \fRac{i}{\pi}\int_{\gamMA} k^2\Left(\Phi_1(t,k) + i \phi_1(T, ik) + \frac{ig_0}{k}\riGht) Dk,$$ and $c_2^{(3)}$ InVolVEs the QuoTient $B(k)/a(k)$. SuBStitutIng the GLM RePResentATion for $\pHI_1$ iNto (\[c13Def\]) and integrating THe TErm in the resultIng expReSSiON InvOlvIng $k^5$ by partS, wE find $$C_1^{(3)} = \Frac{1}{\pi}\iNT_{\gAMMA} \leFT[kn(t,t) - kn(t,-t)e^{-4ik^4t} - \Int_{-t}^t kn_{s}(t,s)e^{2IK^4(s-t)}Ds -kg_0\riGhT]dk.$$ uSing thE idenTiTY (\[inTidentity2\]) anD the Initial coNditioN $N(t,t) = g_0(t)$, thIS gives $$\lAbel{c13fInaL} c_1^{(3)} = -c_1 (\MathCAl{a}^{-1}n)(T,t).$$ FInALly, SImPliFYinG $c_2^{(3)}$ as in thE pRoOf of tHeorEM \[TH2\], We deDucE thaT $c_2^{(3)}$ is gIven by the termS on The rIGht-Hand sIde of EquaTiOn (\[g1th2\]) Which iNvolvE tHe functions $f_0$ and $F_2$. In sUmmary, aftEr cOlLecTiNg the VArious ConTriButions, We conclUDe tHaT (\[G1TH2\]) cAn be derived from (\[g1exPrESSiOn\]). The casE of
2inrem\]) and using the id entity (\[ intid ent ity 1\ ]),we c onclude that $ $ 2ig_ 0\Phi_2^{(2)} = \frac{ 8g_0} {\ p i} \ i nt _{\ga mma} k^ 3 \i n t _{- t} ^t I_ 2( t ,s )e^{2 ik^ 4(s - t )}ds dk =g_0 \l eft(m(t,t) + \f rac{i}{2}\ bar {g}_0n(t,t)\ rig ht).$$ O n t h e oth erhand, we ca n write $c^{(3)} $a s thes um of t w o t erms , $c^{(3)} = c_1^ { (3 ) } +c_2^{(3)}$, where $ $ \l a b el{ c13 def} c _1 ^{(3) } = \frac { i} { \ p i}\ i nt_{\gamma} k ^2\left(\Ph i _1( t,k) + i \P h i_1(t, ik)+\ fra c{ig_0}{k}\ righ t) dk,$$and $c _ 2^{(3)} $ involv es the qu oti ent$ b( k) /a( k) $ . S u bs tit u tin g the GL Mre prese ntat i o n for$\P hi_1 $ int o (\[c13def\] ) a nd i n teg ratin g the ter min th e resu lting e xpression invol ving $k^5$ by pa rt s,we find $$c_1^ {(3 )}= \frac {1}{\pi } \in t_ { \ g am ma} \left[kn(t,t)-k n (t ,-t)e^{- 4ik^4t } - \ i nt_{-t}^ tkn_ {s}( t , s)e^{ 2ik^ 4 (s -t)}ds - kg_0\r i gh t] dk.$$ U si ng the i den tit y (\[ i ntid entity 2\]) and thei nitial conditi o n $n(t,t) = g _ 0( t ) $, this gi ves $$\labe l{c1 3 fina l} c_1 ^ {(3)} = -c _1 (\ m athcal{A}^{-1}n)(t, t) .$$ Fi nally , simplifying $c_2^{(3) } $ as in th e pr o of of theorem \[t h2\], we deduce that $c_ 2^{(3 )}$ is g iven by t h e terms o n t herig ht- h a nd side of equa t i on ( \[ g1th2\] ) w hich in vol vethe fu nc tions $f_ 0$ and $ f_ 2$ .In su mmary , after c ol lec ti ngthe v a riouscontr ibut io ns , we conclu d et h at ( \[ g1 th2\ ])ca n bederi v edfrom (\ [g1expres sio n \]). Th e caseof
2inrem\]) and_using the_identity (\[intidentity1\]), we conclude_that $$2ig_0\Phi_2^{(2)}_=_\frac{8g_0}{\pi} \int_{\gamma}_k^3\int_{-t}^t_I_2(t,s)e^{2ik^4(s - t)}ds_dk = g_0\left(m(t,t) +_\frac{i}{2}\bar{g}_0n(t,t)\right).$$ On the other_hand, we can_write_$c^{(3)}$ as the sum of two terms, $c^{(3)} = c_1^{(3)} +c_2^{(3)}$, where $$\label{c13def} _ _ c_1^{(3)}=_\frac{i}{\pi}\int_{\gamma}_k^2\left(\Phi_1(t,k)_+ i \Phi_1(t, ik) +_\frac{ig_0}{k}\right) dk,$$ and $c_2^{(3)}$ involves_the quotient_$b(k)/a(k)$. Substituting the GLM representation for $\Phi_1$ into_(\[c13def\])_and integrating the_term in the resulting expression involving $k^5$ by parts,_we find $$c_1^{(3)} = \frac{1}{\pi}\int_{\gamma} \left[kn(t,t)_- kn(t,-t)e^{-4ik^4t} -_\int_{-t}^t_kn_{s}(t,s)e^{2ik^4(s-t)}ds_-kg_0\right]dk.$$ Using the identity_(\[intidentity2\]) and the initial condition $n(t,t)_= g_0(t)$, this gives $$\label{c13final} _ c_1^{(3)} = -c_1 (\mathcal{A}^{-1}n)(t,t).$$ Finally,_simplifying $c_2^{(3)}$ as in the proof_of theorem \[th2\], we deduce_that $c_2^{(3)}$_is given by the terms_on the right-hand_side of_equation (\[g1th2\]) which_involve the functions $f_0$ and $f_2$._In summary, after_collecting the various contributions, we conclude_that_(\[g1th2\]) can be_derived_from_(\[g1expression\]). The case_of
, collect terms on LHS, and re-state the conjecture as: $$f(\sigma,\lambda_2,\delta)<0,$$ where $$\begin{split} f(\sigma,\lambda_2,\delta)=&\lambda_2(-44.8\lambda_2-71.68\delta+73.6\sigma+16)\\ &+\delta(12.8+58.88\sigma-37.92\delta)+\\ &\sigma(-20.32\sigma-32.96)+8.48 \end{split}$$ Our goal is to maximize $f(\sigma,\lambda_2,\delta)$ and show that it is less than $0$. We first identify that $\frac{\partial f(\sigma,\lambda_2,\delta)}{\partial \lambda_2}>0$, hence $f(\sigma,\lambda_2,\delta)$ is maximized when $\lambda_2$ is maximized. We evaluate $f(\sigma,\lambda_2,\delta)$ at $\lambda_2=\frac{(3\sigma-1)(3-\sigma)}{4(5-3\sigma)}$. We next identify that $\frac{\partial f(\sigma,,\delta)}{\partial \delta}>0$, hence set $\delta=1-\sigma$ to get: $$\begin{split} &f(\sigma,\lambda_2,\delta)\leq f(\sigma,\frac{(3\sigma-1)(3-\sigma)}{4(5-3\sigma)},1-\sigma)\\ &=-\frac{83.6 (-1.41859 + \sigma) (-1.4 + \sigma) (-1.4 + 3\sigma) (\sigma-1)}{(3\sigma-5)^2}\\ &<0, \end{split}$$ for $\sigma\in[3/5,1]$. Finally, we conclude that $f(\sigma,\lambda_2,\delta)<0$, hence the conjecture was true. Going back, this means that the final expression in is less than zero, which means the expression in is less than zero, meaning that the FOC of firm 1: $$\frac{\partial D({\ell_1},{\ell_2})}{\partial {\ell_1}}({\ell_1}-\lambda_1)+D({\ell_1},{\ell_2})<0.$$ Hence, the FOC for firm 1 can not hold, meaning the firms can not be in an equilibrium.
, collect terms on LHS, and re - state the speculation as: $ $ f(\sigma,\lambda_2,\delta)<0,$$ where $ $ \begin{split } f(\sigma,\lambda_2,\delta)=&\lambda_2(-44.8\lambda_2 - 71.68\delta+73.6\sigma+16)\\ & + \delta(12.8 + 58.88\sigma-37.92\delta)+\\ & \sigma(-20.32\sigma-32.96)+8.48 \end{split}$$ Our finish is to maximize $ f(\sigma,\lambda_2,\delta)$ and show that it is less than $ 0$. We first name that $ \frac{\partial f(\sigma,\lambda_2,\delta)}{\partial \lambda_2}>0 $, hence $ f(\sigma,\lambda_2,\delta)$ is maximize when $ \lambda_2 $ is maximized. We evaluate $ f(\sigma,\lambda_2,\delta)$ at $ \lambda_2=\frac{(3\sigma-1)(3-\sigma)}{4(5 - 3\sigma)}$. We next name that $ \frac{\partial f(\sigma,,\delta)}{\partial \delta}>0 $, therefore set $ \delta=1-\sigma$ to get: $ $ \begin{split } & f(\sigma,\lambda_2,\delta)\leq f(\sigma,\frac{(3\sigma-1)(3-\sigma)}{4(5 - 3\sigma)},1-\sigma)\\ & = -\frac{83.6 (-1.41859 + \sigma) (-1.4 + \sigma) (-1.4 + 3\sigma) (\sigma-1)}{(3\sigma-5)^2}\\ & < 0, \end{split}$$ for $ \sigma\in[3/5,1]$. Finally, we conclude that $ f(\sigma,\lambda_2,\delta)<0 $, hence the guess was true. Going back, this think of that the concluding expression in is less than zero, which means the formula in is less than zero, meaning that the FOC of firm 1: $ $ \frac{\partial D({\ell_1},{\ell_2})}{\partial { \ell_1}}({\ell_1}-\lambda_1)+D({\ell_1},{\ell_2})<0.$$ Hence, the FOC for firm 1 cannot defy, meaning the firms cannot be in an balance.
, coplect terms on LHS, and rt-state the conjecjuee as: $$h(\sigma,\lzmbda_2,\delga)<0,$$ where $$\begin{split} f(\sigma,\lalbea_2,\deluc)=&\lambda_2(-44.8\lambda_2-71.68\delta+73.6\siema+16)\\ &+\delta(12.8+58.88\digma-37.92\delra)+\\ &\smgma(-20.32\sigma-32.96)+8.48 \end{sijit}$$ Ohv goan is to maximizg $f(\sigma,\lambga_2,\delta)$ and shmw tkat it is less than $0$. We first identisy that $\fgac{\partial f(\sidma,\lsibda_2,\svlua)}{\partial \lambda_2}>0$, hence $f(\sigma,\lambsa_2,\delta)$ is maximized when $\lambda_2$ is maximized. Ae egaluate $f(\sigma,\lambfa_2,\delta)$ at $\oambqq_2=\frac{(3\sigma-1)(3-\siema)}{4(5-3\sigma)}$. Wt uext identiry that $\frac{\partial f(\sigma,,\delta)}{\oarticl \delta}>0$, hebcw sfj $\delta=1-\sigma$ to gvt: $$\begin{split} &x(\sigma,\lsmbda_2,\delta)\leq n(\sigme,\frax{(3\sigma-1)(3-\sigma)}{4(5-3\sigma)},1-\sigma)\\ &=-\frac{83.6 (-1.41859 + \sigia) (-1.4 + \sigmd) (-1.4 + 3\sigma) (\sigma-1)}{(3\sigmq-5)^2}\\ &<0, \eng{splkr}$$ fur $\aijma\jn[3/5,1]$. Finwllb, we concluse that $f(\sitma,\lambda_2,\delta)<0$, henct trv conjecture sas trte. Going back, this means that the final eqpreasion in is less than zwro, which means the edpression in is less than zero, meaning that the FOC of firk 1: $$\frec{\oaruiql D({\euo_1},{\epl_2})}{\partial {\ell_1}}({\ell_1}-\lambda_1)+D({\ell_1},{\ell_2})<0.$$ Hence, the FOC fjd gigm 1 can not hold, ieaning the flrkf can not be kn an zsujlibrium.
, collect terms on LHS, and re-state as: where $$\begin{split} &+\delta(12.8+58.88\sigma-37.92\delta)+\\ &\sigma(-20.32\sigma-32.96)+8.48 \end{split}$$ $f(\sigma,\lambda_2,\delta)$ show that it less than $0$. first identify that $\frac{\partial f(\sigma,\lambda_2,\delta)}{\partial \lambda_2}>0$, $f(\sigma,\lambda_2,\delta)$ is maximized when $\lambda_2$ is maximized. We evaluate $f(\sigma,\lambda_2,\delta)$ at $\lambda_2=\frac{(3\sigma-1)(3-\sigma)}{4(5-3\sigma)}$. We identify that $\frac{\partial f(\sigma,,\delta)}{\partial \delta}>0$, hence set $\delta=1-\sigma$ to get: $$\begin{split} &f(\sigma,\lambda_2,\delta)\leq f(\sigma,\frac{(3\sigma-1)(3-\sigma)}{4(5-3\sigma)},1-\sigma)\\ (-1.41859 \sigma) + (-1.4 + 3\sigma) (\sigma-1)}{(3\sigma-5)^2}\\ &<0, \end{split}$$ for $\sigma\in[3/5,1]$. Finally, we conclude that $f(\sigma,\lambda_2,\delta)<0$, hence the conjecture true. Going back, this means that the final in is less than which means the expression in less zero, meaning the of 1: $$\frac{\partial D({\ell_1},{\ell_2})}{\partial Hence, the FOC for firm 1 can not hold, meaning the firms can not be in an
, collect terms on LHS, and re-staTe the conjeCture As: $$f(\SigMa,\LambDa_2,\deLta)<0,$$ where $$\begin{sPLit} f(\Sigma,\lambda_2,\delta)=&\lambda_2(-44.8\LambdA_2-71.68\dELta+73.6\sIGmA+16)\\ &+\deltA(12.8+58.88\sigma-37.92\dELtA)+\\ &\SIgmA(-20.32\sIgMa-32.96)+8.48 \eNd{SPlIt}$$ Our GoaL is to maXimize $f(\sigMa,\lAmBda_2,\delta)$ and sHOw That it is leSs tHan $0$. We first idEntIfy thaT $\fRac{\PArtiaL f(\sIgma,\lAmbda_2,\dELta)}{\parTial \lambdA_2}>0$, hENce $f(\siGMa,\lambdA_2,\DElTa)$ is Maximized when $\lambDA_2$ iS Maximized. We evaLuate $f(\SiGMa,\LAMbdA_2,\deLta)$ at $\lambdA_2=\fRac{(3\siGMa-1)(3-\sigma)}{4(5-3\SIgMA)}$. wE neXT identify that $\Frac{\partial F(\SigMa,,\deltA)}{\pArtIAl \deltA}>0$, hencE sET $\deLta=1-\sigma$ to gEt: $$\beGin{split} &f(\Sigma,\lAMbda_2,\delTA)\leq f(\siGma,\fraC{(3\siGma-1)(3-\SigmA)}{4(5-3\SiGmA)},1-\siGmA)\\ &=-\FraC{83.6 (-1.41859 + \SiGma) (-1.4 + \SIgmA) (-1.4 + 3\sigma) (\siGmA-1)}{(3\sIgma-5)^2}\\ &<0, \eNd{spLIT}$$ FOr $\siGma\In[3/5,1]$. FiNally, We conclude thaT $f(\sIgma,\LAmbDa_2,\delTa)<0$, henCe thE cOnjecTure waS true. goIng back, this meanS thaT the final ExpReSsiOn In is lESs than ZerO, whIch meanS the expREssIoN IN Is Less than zero, meaninG tHAT tHe FOC of fIrm 1: $$\fraC{\PaRtIAl D({\ell_1},{\elL_2})}{\pArtIal {\eLL_1}}({\Ell_1}-\laMbda_1)+d({\ElL_1},{\ell_2})<0.$$ HencE, the FOc FoR fIrm 1 can nOt Hold, meAnIng The Firms CAn noT be in aN equilibRium.
, collect terms on LHS, an d re-state thecon jec tu re a s: $ $f(\sigma,\lam b da_2 ,\delta)<0,$$ where $$ \begi n{ s plit } f(\si gma,\la m bd a _ 2,\ de lt a)= &\ l am bda_2 (-4 4.8\lam bda_2-71.6 8\d el ta+73.6\sigm a +1 6)\\ &+\d elt a(12.8+58.88 \si gma-37 .9 2\d e lta)+ \\ &\si gma(-2 0 .32\si gma-32.96 )+ 8 .48 \ en d{sp lit}$$ Our goali st o maximize $f( \sigma ,\ l am b d a_2 ,\d elta)$ and s how t h at it i s l e s s th a n $0$. We fir st identify tha t $\fr ac {\p a rtialf(\si gm a ,\l ambda_2,\de lta) }{\partia l \lam b da_2}>0 $ , hence $f(\s igm a,\ lamb d a_ 2, \de lt a )$i smax i miz ed when$\ la mbda_ 2$ i s m a ximi zed . We eval uate $f(\sigm a,\ lamb d a_2 ,\del ta)$at $ \l ambda _2=\fr ac{(3 \s igma-1)(3-\sigm a)}{ 4(5-3\sig ma) }$ . W enexti dentif y t hat $\frac {\parti a l f (\ s i g ma ,,\delta)}{\partia l\ d el ta}>0$,hences et $ \ delta=1- \s igm a$ t o get:$$\b e gi n{split} &f (\sigma ,\ lambda _2 ,\d elt a)\le q f(\ sigma, \frac{(3 \sigm a -1)(3-\sigma)} { 4(5-3\sigma)} , 1- \ s ig m a)\\ &=-\f rac{ 8 3.6(-1. 4 18 59+ \sig ma) ( -1 . 4+ \sigma) (-1.4 + 3\ si gma) ( \sigm a-1)}{(3\sigm a-5)^2}\\ &<0 , \e n d{ s plit}$$ for $\ sigma \in[3/5,1] $ . Finall y, we conclud e that $f ( \ sigma,\l amb da_ 2,\ del t a )< 0$, hence the c onje ct ure was tr ue. Goi ngbac k,thi smeans tha t the fi na lex pr ess ion i n is less t han z ero , whi c h mean s the exp re ss i onin is l e ss t hanze ro , me ani ng that the FOC of fir m 1: $$\f rac { \par ti al D({\el l_1},{\ell_2} )} {\partial{\ ell _1}}({ \ e ll_1}-\l ambda_1)+D({\ell_1},{\e l l_2})<0 .$$ Henc e, t he FOC fo r f irm 1can not ho ld, me aning t hef i rms c a n n otbe in an equ i l ibr ium.
, collect_terms on_LHS, and re-state the_conjecture as:_$$f(\sigma,\lambda_2,\delta)<0,$$_where $$\begin{split} _f(\sigma,\lambda_2,\delta)=&\lambda_2(-44.8\lambda_2-71.68\delta+73.6\sigma+16)\\ _&+\delta(12.8+58.88\sigma-37.92\delta)+\\ &\sigma(-20.32\sigma-32.96)+8.48 _ _ _ \end{split}$$ Our_goal_is to maximize $f(\sigma,\lambda_2,\delta)$ and show that it is less than $0$. We first_identify_that $\frac{\partial_f(\sigma,\lambda_2,\delta)}{\partial_\lambda_2}>0$,_hence $f(\sigma,\lambda_2,\delta)$ is maximized when_$\lambda_2$ is maximized. We evaluate_$f(\sigma,\lambda_2,\delta)$ at_$\lambda_2=\frac{(3\sigma-1)(3-\sigma)}{4(5-3\sigma)}$. We next identify that $\frac{\partial f(\sigma,,\delta)}{\partial \delta}>0$,_hence_set $\delta=1-\sigma$ to_get: $$\begin{split} _&f(\sigma,\lambda_2,\delta)\leq f(\sigma,\frac{(3\sigma-1)(3-\sigma)}{4(5-3\sigma)},1-\sigma)\\ _ _&=-\frac{83.6_(-1.41859_+ \sigma) (-1.4 +_\sigma) (-1.4 + 3\sigma) (\sigma-1)}{(3\sigma-5)^2}\\ _ _&<0, \end{split}$$ for $\sigma\in[3/5,1]$. Finally, we conclude that_$f(\sigma,\lambda_2,\delta)<0$, hence the conjecture was true._Going back, this means that_the final_expression in is less than_zero, which means_the expression_in is less_than zero, meaning that the FOC_of firm 1:_$$\frac{\partial D({\ell_1},{\ell_2})}{\partial {\ell_1}}({\ell_1}-\lambda_1)+D({\ell_1},{\ell_2})<0.$$ Hence, the FOC_for_firm 1 can_not_hold,_meaning the_firms can not_be_in an_equilibrium.
2)$ and space-periodic domains as in our case. We restrict our considerations to the tridimensional case just for the simplicity, but the technique also works in the two-dimensional case and the results hold for $p\in(1,2)$. In order to better explain our results we introduce some spaces of functions. For the sequel it is worth to note that vector valued functions are printed in boldface while scalar ones in italic mode and we do not distinguish between space of scalar functions and space of vector-valued functions. As stated before let $\O$ be the cube $(0, 1)^3 $ in $\R^3$ of points $\x=(x_1, x_2,x_3)$ and let $\O_T$ denote $(0, T)\times\O$. For a domain $G$, that can be $\O$ or $\O_T$, for $q\geq 1$ and $m\in \N$ we keep the notation $(L^q(G), \|\cdot\|_{q,G})$ and $(W^{m,q}(G), \|\cdot\|_{m,q,G})$ for the standard Lebesgue and Sobolev spaces and their associated norms. The subscript “[$_{\rm per}$]{}" means that space-periodic functions having zero mean value are considered (i.e. each considered function $f$ satisfies $f(\x+\e_i)=f(\x)$, $i=1, 2, 3$ where $(\e_1, \e_2,\e_3)$ is the canonical basis of $\R^3$ and $\int_\O f=0$). Next let us establish the notation for the spaces of solenoidal space-periodic functions with values in $\R^3$ by setting $$\mathscr{C}_{\rm per}(\O):=\left\{ \bphi\in \mathcal{C}^\infty_{\rm per}(\O, \R^3), \int_\O \bphi(\x)\,d\x=0, \div\bphi=0 \mbox{ in } \O\right\},$$ and $$J^q_{\rm per}(\O):=\overline{\mathscr{C}_{\rm per}(\O)}^{\|\cdot\|_q}, \ \ J^{m,q}_{\rm per}(\O):=\overline{\mathscr{C}_{\rm per}(\O)}^{\|\cdot\|_{m,q}}.$$ Since there is no confusion, in the sequel we omit
2)$ and space - periodic domains as in our case. We restrict our consideration to the tridimensional subject just for the simplicity, but the technique besides works in the two - dimensional case and the results keep for $ p\in(1,2)$. In order to better explain our results we introduce some space of functions. For the sequel it is worth to note that vector valued functions are printed in boldface while scalar one in italic mode and we do not distinguish between space of scalar function and space of vector - valued functions. As stated before permit $ \O$ be the cube $ (0, 1)^3 $ in $ \R^3 $ of points $ \x=(x_1, x_2,x_3)$ and let $ \O_T$ denote $ (0, T)\times\O$. For a domain $ G$, that can be $ \O$ or $ \O_T$, for $ q\geq 1 $ and $ m\in \N$ we keep the note $ (L^q(G), \|\cdot\|_{q, G})$ and $ (W^{m, q}(G), \|\cdot\|_{m, q, G})$ for the standard Lebesgue and Sobolev spaces and their associated norms. The subscript “ [ $ _ { \rm per}$ ] { } " means that space - periodic functions hold zero mean value are considered (i.e. each considered function $ f$ satisfies $ f(\x+\e_i)=f(\x)$, $ i=1, 2, 3 $ where $ (\e_1, \e_2,\e_3)$ is the canonical basis of $ \R^3 $ and $ \int_\O f=0 $). Next let us establish the notation for the quad of solenoidal space - periodic functions with value in $ \R^3 $ by set $ $ \mathscr{C}_{\rm per}(\O):=\left\ { \bphi\in \mathcal{C}^\infty_{\rm per}(\O, \R^3), \int_\O \bphi(\x)\,d\x=0, \div\bphi=0 \mbox { in } \O\right\},$$ and $ $ J^q_{\rm per}(\O):=\overline{\mathscr{C}_{\rm per}(\O)}^{\|\cdot\|_q }, \ \ J^{m, q}_{\rm per}(\O):=\overline{\mathscr{C}_{\rm per}(\O)}^{\|\cdot\|_{m, q}}.$$ Since there is no confusion, in the sequel we omit
2)$ anf space-periodic domains xs in our case. Wg eestrirt our donsiderxtions to the tridimensional cqse jyst for the simplicity, but the nechnique alsi works in vge two-dliensjlnal rase and the rexults hold for $p\in(1,2)$. In ordar tl better explain our results we intwoduce xole spaces of fonctipgs. Fkg uhe sequel it is worth to note thzt vectmr valued funvtions are printed in boldvace while scalar ones in italic node qnd we do nog distingupvh between space of scalar functions and rpace of vector-caouef functions. Aw staned before leb $\O$ be dhe cubr $(0, 1)^3 $ in $\R^3$ of ppinvs $\x=(z_1, x_2,x_3)$ and let $\O_T$ denotx $(0, T)\times\O$. For a domwin $G$, thad ean be $\O$ or $\O_T$, for $q\gwq 1$ and $m\in \N$ wd kedp uhe nktatioj $(L^s(G), \|\cdot\|_{q,G})$ znd $(W^{m,q}(G), \|\cdit\|_{m,q,G})$ for the standsrq Lebesgue and Sobolqv spaces and their associated norms. The vubacript “[$_{\rm per}$]{}" means thar space-periodic functlons havigg zero mean value are considered (i.e. each considesed fnnztiin $w$ swtisfies $f(\x+\e_i)=f(\x)$, $i=1, 2, 3$ where $(\e_1, \e_2,\e_3)$ is the canonicwm naxis of $\R^3$ and $\ikt_\O f=0$). Next let us rshanjish the notajion fox tge spaces of solenlidal skace-peeiodic fugctipns with values in $\R^3$ by setring $$\mathscr{B}_{\rm per}(\O):=\left\{ \bphi\in \machcal{C}^\infty_{\ro pet}(\O, \R^3), \imt_\O \bphi(\x)\,d\x=0, \div\bphi=0 \mbor{ in } \K\right\},$$ and $$U^q_{\rm per}(\O):=\kxerline{\mathscr{C}_{\ro pvr}(\O)}^{\|\cgot\|_q}, \ \ J^{m,q}_{\rm per}(\O):=\overline{\mwthscr{C}_{\rm per}(\P)}^{\|\cdot\|_{m,q}}.$$ Sinve thewe is no clnfuslmn, in the sequel wf omij
2)$ and space-periodic domains as in our restrict considerations to tridimensional case just technique works in the case and the hold for $p\in(1,2)$. In order to explain our results we introduce some spaces of functions. For the sequel it worth to note that vector valued functions are printed in boldface while scalar in mode we not distinguish between space of scalar functions and space of vector-valued functions. As stated before let be the cube $(0, 1)^3 $ in $\R^3$ points $\x=(x_1, x_2,x_3)$ and $\O_T$ denote $(0, T)\times\O$. For domain that can $\O$ $\O_T$, $q\geq 1$ and \N$ we keep the notation $(L^q(G), \|\cdot\|_{q,G})$ and $(W^{m,q}(G), \|\cdot\|_{m,q,G})$ for the standard Lebesgue and Sobolev spaces their associated subscript “[$_{\rm means space-periodic having zero mean considered (i.e. each considered function $f$ 2, 3$ where $(\e_1, \e_2,\e_3)$ is the canonical of $\R^3$ $\int_\O f=0$). Next let us establish notation for the spaces of solenoidal space-periodic functions values in $\R^3$ by setting $$\mathscr{C}_{\rm per}(\O):=\left\{ \bphi\in \mathcal{C}^\infty_{\rm per}(\O, \R^3), \int_\O \bphi(\x)\,d\x=0, \div\bphi=0 \mbox{ \O\right\},$$ and $$J^q_{\rm per}(\O):=\overline{\mathscr{C}_{\rm \ \ J^{m,q}_{\rm per}(\O)}^{\|\cdot\|_{m,q}}.$$ there no in the we omit
2)$ and space-periodic domains as In our case. WE restRicT ouR cOnsiDeraTions to the tridIMensIonal case just for the simPliciTy, BUt thE TeChniqUe also wORkS IN thE tWo-DimEnSIoNal caSe aNd the reSults hold fOr $p\In(1,2)$. in order to betTEr Explain our ResUlts we introdUce Some spAcEs oF FunctIonS. For tHe sequEL it is wOrth to notE tHAt vectOR valued FUNcTionS are printed in boldFAcE While scalar oneS in itaLiC MoDE And We dO not distinGuIsh beTWeen spaCE oF SCAlaR Functions and sPace of vectoR-ValUed funCtIonS. as statEd befOrE Let $\o$ be the cube $(0, 1)^3 $ iN $\R^3$ of Points $\x=(x_1, x_2,X_3)$ and leT $\o_T$ denotE $(0, t)\times\O$. for a doMaiN $G$, tHat cAN bE $\O$ Or $\O_t$, fOR $q\gEQ 1$ aNd $m\IN \N$ wE keep the NoTaTion $(L^Q(G), \|\cdOT\|_{Q,g})$ And $(W^{M,q}(G), \|\Cdot\|_{M,q,G})$ foR the standard LEbeSgue ANd SOboleV spacEs anD tHeir aSsociaTed noRmS. The subscript “[$_{\rm Per}$]{}" mEans that sPacE-pEriOdIc funCTions hAviNg zEro mean Value arE ConSiDEREd (I.e. each considered fuNcTIOn $F$ satisfiEs $f(\x+\e_i)=F(\X)$, $i=1, 2, 3$ WhERe $(\e_1, \e_2,\e_3)$ is tHe CanOnicAL Basis Of $\R^3$ aND $\iNt_\O f=0$). Next Let us eSTaBlIsh the nOtAtion fOr The SpaCes of SOlenOidal sPace-periOdic fUNctions with valUEs in $\R^3$ by settinG $$\MaTHScR{c}_{\rm pEr}(\O):=\Left\{ \bphi\in \mAthcAL{C}^\inFty_{\rM PeR}(\O, \R^3), \INt_\O \bpHi(\x)\,d\x=0, \DiV\BpHI=0 \mbox{ in } \O\right\},$$ and $$J^q_{\rM pEr}(\O):=\oveRline{\Mathscr{C}_{\rm per}(\o)}^{\|\cdot\|_q}, \ \ J^{m,q}_{\rM PER}(\O):=\overliNe{\maTHsCR{C}_{\rm per}(\O)}^{\|\cdot\|_{m,q}}.$$ since There is no cONfusion, iN the sEquel we oMit
2)$ and space-periodic dom ains as in ourcas e. W e re stri ct our conside r atio ns to the tridimension al ca se just fo r the simpli c it y , bu tth e t ec h ni que a lso worksin the two -di me nsional case an d the resu lts hold for $p \in (1,2)$ . In order to bett er exp l ain ou r results w e intro d uce som e sp aces of functions. Fo r t h e sequel it is worth t o n o t e t hat vector va lu ed fu n ctionsa re p r int e d in boldface while scal a r o nes in i tal i c mode andwe donot disting uish betweenspaceo f scala r functi ons an d s pac e of ve ct or- va l ued fu nct i ons . As st at ed befo re l e t $ \O$bethecube$(0, 1)^3 $ i n $ \R^3 $ of poin ts $\ x=(x _1 , x_2 ,x_3)$ andle t $\O_T$ denote $(0 , T)\time s\O $. Fo ra dom a in $G$ , t hat can be $\O$ o r $\ O_ T $ , f or $q\geq 1$ and $ m\ i n \ N$ we ke ep the no ta t ion $(L^ q( G), \|\ c d ot\|_ {q,G } )$ and $(W ^{m,q} ( G) ,\|\cdot \| _{m,q, G} )$for thes tand ard Le besgue a nd So b olev spaces an d their associ a te d no r ms.The subscript“[$_ { \rmper} $ ]{ }"m eansthatsp a ce - periodic functionsha ving z ero m ean value are considere d ( i.e. eac h co n si d ered function$f$ s atisfies $ f (\x+\e_i )=f(\ x)$, $i= 1, 2, 3$w h ere $(\e _1, \e _2, \e_ 3 ) $is the canoni c a l ba si s of $\ R^3 $ and $ \in t_\ O f =0$ ). Next let us esta bl is hth e n otati o n for th espa ce s o f sol e noidal spac e-pe ri od i c f unction s w i t h va lu es in$\R ^3 $ bysett i ng$$\math scr{C}_{\ rmp er}( \O ): =\left\ { \bphi\in \m at hcal{C}^\i nf ty_ {\rm p e r }(\O, \R ^3), \int_\O \bphi(\x)\ , d\x=0,\di v\bph i=0\mbox{ in }\O\rig ht\ } ,$$ an d $$J^ q_{\r mper } ( \O):= \ o ve rli ne {\mathscr{ C } _{\ rm pe r} (\O) }^{\|\c dot\|_q}, \ \ J^{m , q}_ {\rm per}(\O) :=\ over l i ne {\m a th s cr{ C} _ {\r m per}(\O)}^{\|\c dot\|_{m,q }} . $$ Since the r e i sno conf usion,in th e sequel we omit
2)$ and_space-periodic domains_as in our case. We_restrict our_considerations_to the_tridimensional_case just for_the simplicity, but_the technique also works_in the two-dimensional_case_and the results hold for $p\in(1,2)$. In order to better explain our results we introduce_some_spaces of_functions._For_the sequel it is worth_to note that vector valued_functions are_printed in boldface while scalar ones in italic_mode_and we do_not distinguish between space of scalar functions and space_of vector-valued functions. As stated before let_$\O$ be the_cube_$(0,_1)^3 $ in $\R^3$_of points $\x=(x_1, x_2,x_3)$ and let_$\O_T$ denote $(0, T)\times\O$. For a_domain $G$, that can be $\O$ or_$\O_T$, for $q\geq 1$ and $m\in_\N$ we keep the notation_$(L^q(G), \|\cdot\|_{q,G})$_and $(W^{m,q}(G), \|\cdot\|_{m,q,G})$ for the_standard Lebesgue and_Sobolev spaces_and their associated_norms. The subscript “[$_{\rm per}$]{}" means_that space-periodic functions_having zero mean value are considered_(i.e._each considered function_$f$_satisfies_$f(\x+\e_i)=f(\x)$, $i=1,_2, 3$ where_$(\e_1,_\e_2,\e_3)$ is_the_canonical basis of $\R^3$ and $\int_\O_f=0$)._Next let us establish the notation for_the spaces of solenoidal_space-periodic_functions with values in_$\R^3$ by setting $$\mathscr{C}_{\rm per}(\O):=\left\{_\bphi\in \mathcal{C}^\infty_{\rm per}(\O, \R^3), \int_\O \bphi(\x)\,d\x=0,_\div\bphi=0 \mbox{_in }_\O\right\},$$ and $$J^q_{\rm per}(\O):=\overline{\mathscr{C}_{\rm per}(\O)}^{\|\cdot\|_q}, \ \ J^{m,q}_{\rm per}(\O):=\overline{\mathscr{C}_{\rm per}(\O)}^{\|\cdot\|_{m,q}}.$$ Since_there is no confusion, in the_sequel we omit
widetilde{\tau}(z)}{\sigma(z)}{\psi}'(z)+ \frac{\widetilde{\sigma}(z)}{\sigma^{2}(z)}\psi(z)=0,$$ where $\sigma(z)$ and $\widetilde{\sigma}(z)$ are polynomials, at most second-degree, and $\widetilde{\tau}(z)$ is a first-degree polynomial. Using Eq.(2.1) the transformation $$\psi(z)=\Phi(z)\\{y}(z)$$ one reduces it to the hypergeometric-type equation $$\sigma(z){y}''+\tau(z){y}'+\lambda{y}=0.$$ The function $\Phi(z)$ is defined as the logarithmic derivative \[10\] $$\frac{\Phi'(z)}{\Phi(z)}=\frac{\pi(z)}{\sigma(z)},$$ where $\pi(z)$ is at most the first-degree polynomial. The another part of $\psi(z)$, namely ${y}(z)$, is the hypergeometric-type function, that for fixed $n$ is given by the Rodriguez relation: $${y_{n}}(z)=\frac{{B_{n}}}{\rho(z)}\frac{{d^{n}}}{{dz^{n}}}[\sigma^{n}(z)\rho(z)],$$ where ${B_{n}}$ is the normalization constant and the weight function $\rho(z)$ must satisfy the condition \[10\] $$\frac{d}{dz}\left(\sigma(z)\rho(z)\right)=\tau(z)\rho(z),$$ with $\tau(z)=\widetilde{\tau}(z)+2\pi(z).$ For accomplishment of the conditions imposed on function $\rho(z)$, the classical orthogonal polynomials, it is necessary, that polynomial $\tau(z)$ becomes equal to zero in some point of an interval $(a,b)$ and derivative of this polynomial for this interval at $\sigma(z)>0$ will be negative, i.e. $\tau'(z)<0$. The function $\pi(z)$ and the parameter $\lambda$ required for this method are defined as follows: $$\pi(z)=\frac{\sigma'-\widetilde{\tau}}{2}\pm\sqrt{\left(\frac{\sigma'- \widetilde{\tau}}{2}\right)^{2}-\widetilde{\sigma}+{k}\sigma},$$ $$\lambda=k+\pi'(z).$$ On the other hand, in order to find the value of $k$, the expression under the square
widetilde{\tau}(z)}{\sigma(z)}{\psi}'(z)+ \frac{\widetilde{\sigma}(z)}{\sigma^{2}(z)}\psi(z)=0,$$ where $ \sigma(z)$ and $ \widetilde{\sigma}(z)$ are polynomials, at most second - degree, and $ \widetilde{\tau}(z)$ is a first - degree polynomial. use Eq.(2.1) the transformation $ $ \psi(z)=\Phi(z)\\{y}(z)$$ one reduce it to the hypergeometric - type equation $ $ \sigma(z){y}''+\tau(z){y}'+\lambda{y}=0.$$ The function $ \Phi(z)$ is define as the logarithmic derivative \[10\ ] $ $ \frac{\Phi'(z)}{\Phi(z)}=\frac{\pi(z)}{\sigma(z)},$$ where $ \pi(z)$ is at most the first - academic degree polynomial. The another part of $ \psi(z)$, namely $ { y}(z)$, is the hypergeometric - type function, that for sterilize $ n$ is given by the Rodriguez relation: $ $ { y_{n}}(z)=\frac{{B_{n}}}{\rho(z)}\frac{{d^{n}}}{{dz^{n}}}[\sigma^{n}(z)\rho(z)],$$ where $ { B_{n}}$ is the standardization constant and the weight function $ \rho(z)$ must satisfy the circumstance \[10\ ] $ $ \frac{d}{dz}\left(\sigma(z)\rho(z)\right)=\tau(z)\rho(z),$$ with $ \tau(z)=\widetilde{\tau}(z)+2\pi(z).$ For accomplishment of the conditions levy on function $ \rho(z)$, the classical extraneous polynomials, it is necessary, that polynomial $ \tau(z)$ becomes equal to zero in some degree of an interval $ (a, b)$ and derivative of this polynomial for this time interval at $ \sigma(z)>0 $ will be negative, i.e. $ \tau'(z)<0$. The routine $ \pi(z)$ and the parameter $ \lambda$ want for this method are defined as follows: $ $ \pi(z)=\frac{\sigma'-\widetilde{\tau}}{2}\pm\sqrt{\left(\frac{\sigma'- \widetilde{\tau}}{2}\right)^{2}-\widetilde{\sigma}+{k}\sigma},$$ $ $ \lambda = k+\pi'(z).$$ On the early hand, in order to find the value of $ k$, the expression under the square
widftilde{\tau}(z)}{\sigma(z)}{\psi}'(z)+ \frac{\didetilde{\sigma}(z)}{\sigma^{2}(z)}\pvi(z)=0,$$ whsre $\sigmx(z)$ and $\widetilde{\sigma}(z)$ are plltnomiqls, at most second-degrde, and $\wifetilde{\tqu}(z)$ ms a first-degree polynomlcl. Usihn Eq.(2.1) chx transformatiok $$\psi(z)=\Phi(z)\\{y}(s)$$ one reduces ht tl the hypergeometric-type equation $$\sygma(z){y}''+\tsu(x){y}'+\lambda{y}=0.$$ The suncuiog $\Phj(d)$ ls defined as the logarithmic dedivativt \[10\] $$\frac{\Phi'(z)}{\Phi(z)}=\frav{\pi(z)}{\sigma(z)},$$ where $\pi(z)$ is at mosh the first-degree oolynomial. Tye agither part ow $\psi(z)$, namtlv ${y}(z)$, is the hypergeometric-type function, thxt fox fixed $n$ iw tivft by the Rovrigued relation: $${y_{n}}(d)=\grac{{B_{n}}}{\sho(z)}\frav{{d^{n}}}{{dz^{n}}}[\sigma^{n}(z)\rmo(z)],$$ wiere ${B_{n}}$ is the normalizatmon constant and the weight fgnetion $\rho(z)$ must satisdy the wondhtiov \[10\] $$\ffac{s}{dv}\lert(\sigmw(z)\rio(z)\right)=\tau(a)\rho(z),$$ with $\rau(z)=\widetilde{\tau}(z)+2\pi(e).$ Fow accomplishmeht of ehq conditions imposed on function $\rho(z)$, tve dlassical orthogonal pooynomials, it is necesdary, that polynomial $\tau(z)$ becomes equal to zero in some pohnt oh xn nkbervxo $(w,b)$ and derivative of this polynomial for this jnuernal at $\sigma(z)>0$ wilj be negatife, i.r. $\tau'(z)<0$. The funcjion $\pi(v)$ ahd the parameter $\lwmbda$ rgquiree for thif meyhod are defined as follows: $$\pi(z)=\frac{\sigmc'-\wieetilde{\tau}}{2}\pm\sqrt{\leyt(\frac{\sigma'- \wndetilce{\tau}}{2}\tight)^{2}-\widetilde{\sigma}+{k}\sigoa},$$ $$\lzmbda=k+\pi'(z).$$ Oj the othsf hand, in order go ning the value of $k$, the exprefsion undxr thz square
widetilde{\tau}(z)}{\sigma(z)}{\psi}'(z)+ \frac{\widetilde{\sigma}(z)}{\sigma^{2}(z)}\psi(z)=0,$$ where $\sigma(z)$ and $\widetilde{\sigma}(z)$ are most and $\widetilde{\tau}(z)$ a first-degree polynomial. one it to the equation $$\sigma(z){y}''+\tau(z){y}'+\lambda{y}=0.$$ The $\Phi(z)$ is defined as the logarithmic \[10\] $$\frac{\Phi'(z)}{\Phi(z)}=\frac{\pi(z)}{\sigma(z)},$$ where $\pi(z)$ is at most the first-degree polynomial. The another part $\psi(z)$, namely ${y}(z)$, is the hypergeometric-type function, that for fixed $n$ is given the relation: where is the normalization constant and the weight function $\rho(z)$ must satisfy the condition \[10\] $$\frac{d}{dz}\left(\sigma(z)\rho(z)\right)=\tau(z)\rho(z),$$ with For accomplishment of the conditions imposed on function the classical orthogonal polynomials, is necessary, that polynomial $\tau(z)$ equal zero in point an $(a,b)$ and derivative this polynomial for this interval at $\sigma(z)>0$ will be negative, i.e. $\tau'(z)<0$. The function $\pi(z)$ and the $\lambda$ required method are as $$\pi(z)=\frac{\sigma'-\widetilde{\tau}}{2}\pm\sqrt{\left(\frac{\sigma'- $$\lambda=k+\pi'(z).$$ On the in order to find the value expression under the square
widetilde{\tau}(z)}{\sigma(z)}{\psi}'(z)+ \frAc{\widetildE{\sigmA}(z)}{\sIgmA^{2}(z)}\Psi(z)=0,$$ WherE $\sigma(z)$ and $\wideTIlde{\Sigma}(z)$ are polynomials, at Most sEcONd-deGReE, and $\wIdetildE{\TaU}(Z)$ Is a FiRsT-deGrEE pOlynoMiaL. Using EQ.(2.1) the transfOrmAtIon $$\psi(z)=\Phi(z)\\{y}(Z)$$ OnE reduces it To tHe hypergeomeTriC-type eQuAtiON $$\sigmA(z){y}''+\Tau(z){y}'+\Lambda{Y}=0.$$ the funCtion $\Phi(z)$ Is DEfined AS the logARItHmic Derivative \[10\] $$\frac{\Phi'(Z)}{\phI(Z)}=\frac{\pi(z)}{\sigma(z)},$$ Where $\pI(z)$ IS aT MOst The First-degreE pOlynoMIal. The aNOtHER ParT Of $\psi(z)$, namely ${y}(Z)$, is the hyperGEomEtric-tYpE fuNCtion, tHat foR fIXed $N$ is given by tHe RoDriguez reLation: $${Y_{N}}(z)=\frac{{B_{N}}}{\Rho(z)}\fraC{{d^{n}}}{{dz^{n}}}[\SigMa^{n}(Z)\rho(Z)],$$ WhErE ${B_{n}}$ Is THe nORmAliZAtiOn constaNt AnD the wEighT FUNCtioN $\rhO(z)$ muSt satIsfy the conditIon \[10\] $$\Frac{D}{Dz}\lEft(\siGma(z)\rHo(z)\rIgHt)=\tau(Z)\rho(z),$$ wIth $\taU(z)=\Widetilde{\tau}(z)+2\pi(Z).$ For AccomplisHmeNt Of tHe CondiTIons imPosEd oN functiOn $\rho(z)$, tHE clAsSICAl Orthogonal polynomiAlS, IT iS necessaRy, that POlYnOMial $\tau(z)$ BeComEs eqUAL to zeRo in SOmE point of An inteRVaL $(a,B)$ and derIvAtive oF tHis PolYnomiAL for This inTerval at $\Sigma(Z)>0$ Will be negative, I.E. $\tau'(z)<0$. The functIOn $\PI(Z)$ aND the ParAmeter $\lambdA$ reqUIred For tHIs MetHOd are DefinEd AS fOLlows: $$\pi(z)=\frac{\sigma'-\wiDeTilde{\tAu}}{2}\pm\sQrt{\left(\frac{\siGma'- \widetilDE{\TAu}}{2}\right)^{2}-\wIdetILdE{\Sigma}+{k}\sigma},$$ $$\lamBda=k+\pI'(z).$$ On the othER hand, in oRder tO find the Value of $k$, tHE ExpressiOn uNdeR thE sqUARe
widetilde{\tau}(z)}{\sigma (z)}{\psi} '(z)+ \f rac {\ wide tild e{\sigma}(z)}{ \ sigm a^{2}(z)}\psi(z)=0,$$where $ \ sigm a (z )$ an d $\wid e ti l d e{\ si gm a}( z) $ a re po lyn omials, at most s eco nd -degree, and $\ widetilde{ \ta u}(z)$ is afir st-deg re e p o lynom ial . Us ing Eq . (2.1)the trans fo r mation $$\psi( z ) =\ Phi( z)\\{y}(z)$$ oner ed u ces it to thehyperg eo m et r i c-t ype equation$$ \sigm a (z){y}' ' +\ t a u (z) { y}'+\lambda{y }=0.$$ Thef unc tion $ \P hi( z )$ isdefin ed asthe logarit hmic derivati ve \[1 0 \] $$\f r ac{\Phi '(z)}{ \Ph i(z )}=\ f ra c{ \pi (z ) }{\ s ig ma( z )}, $$ where $ \p i(z)$ isa t m ostthe fir st-de gree polynomi al. Th e an other part of$\ psi(z )$, na mely${ y}(z)$, is thehype rgeometri c-t yp e f un ction , thatfor fi xed $n$ is giv e n b yt h e R odriguez relation: $ $ { y_ {n}}(z)= \frac{ { B_ {n } }}{\rho( z) }\f rac{ { d ^{n}} }{{d z ^{ n}}}[\si gma^{n } (z )\ rho(z)] ,$ $ wher e${B _{n }}$ i s the norma lization cons t ant and the we i ght function$ \r h o (z ) $ mu stsatisfy the con d itio n \[ 1 0\ ] $ $ \frac {d}{d z} \ le f t(\sigma(z)\rho(z)\ ri ght)=\ tau(z )\rho(z),$$ w ith $\tau( z ) = \widetil de{\ t au } (z)+2\pi(z).$ Foraccomplish m ent of t he co nditions imposedo n functio n $ \rh o(z )$, t he classical or t h ogon al polyno mia ls, itisnec ess ary ,that poly nomial $ \t au (z )$ be comes equal to z ero i n s ome p o int of an i nter va l$ (a, b)$ and de r i vati ve o f th ispo lynom ialf orthis in terval at $\ s igma (z )> 0$ will be negative, i .e. $\tau' (z )<0 $. Th e function $\pi(z)$ and the param e ter $\l amb da$ r equi red for t his metho d a r e defi ned as foll ow s:$ $ \pi(z ) = \f rac {\ sigma'-\wi d e til de{\t au }}{2 }\pm\sq rt{\left(\frac{\si g ma' - \widetilde{ \ta u}}{ 2 } \r igh t )^ { 2}- \w i det i l de{\sigma}+{k}\ sigma},$$$$ \ la mbda=k+\pi ' (z) .$ $ On th e other hand , in ord er to fin d the val ue of$ k $,the expres sion und er the sq u are
widetilde{\tau}(z)}{\sigma(z)}{\psi}'(z)+ \frac{\widetilde{\sigma}(z)}{\sigma^{2}(z)}\psi(z)=0,$$ where_$\sigma(z)$ and_$\widetilde{\sigma}(z)$ are polynomials, at_most second-degree,_and_$\widetilde{\tau}(z)$ is_a_first-degree polynomial. Using Eq.(2.1)_the transformation $$\psi(z)=\Phi(z)\\{y}(z)$$_one reduces it to_the hypergeometric-type equation_$$\sigma(z){y}''+\tau(z){y}'+\lambda{y}=0.$$_The function $\Phi(z)$ is defined as the logarithmic derivative \[10\] $$\frac{\Phi'(z)}{\Phi(z)}=\frac{\pi(z)}{\sigma(z)},$$ where $\pi(z)$ is_at_most the_first-degree_polynomial. The_another part of $\psi(z)$, namely_${y}(z)$, is the hypergeometric-type function,_that for_fixed $n$ is given by the Rodriguez relation:_$${y_{n}}(z)=\frac{{B_{n}}}{\rho(z)}\frac{{d^{n}}}{{dz^{n}}}[\sigma^{n}(z)\rho(z)],$$_where ${B_{n}}$ is_the normalization constant and the weight function $\rho(z)$ must_satisfy the condition \[10\] $$\frac{d}{dz}\left(\sigma(z)\rho(z)\right)=\tau(z)\rho(z),$$ with_$\tau(z)=\widetilde{\tau}(z)+2\pi(z).$ For accomplishment of_the_conditions_imposed on function $\rho(z)$,_the classical orthogonal polynomials, it is_necessary, that polynomial $\tau(z)$ becomes equal_to zero in some point of an_interval $(a,b)$ and derivative of this_polynomial for this interval at_$\sigma(z)>0$ will_be negative, i.e. $\tau'(z)<0$. The function_$\pi(z)$ and the_parameter $\lambda$_required for this_method are defined as follows: $$\pi(z)=\frac{\sigma'-\widetilde{\tau}}{2}\pm\sqrt{\left(\frac{\sigma'- \widetilde{\tau}}{2}\right)^{2}-\widetilde{\sigma}+{k}\sigma},$$_$$\lambda=k+\pi'(z).$$ On the_other hand, in order to find_the_value of $k$,_the_expression_under the_square
ideal as a sub-object of $W$ because of the isomorphism $\mathbf{R}({\mathcal{D}}[s])\simeq\textrm{gr}^V_0(\mathbf{R}({\mathcal{D}}_{x,t})).$ ### Condition for the $h$-saturation of $\mathbf{R}(\textrm{gr}^V(N))$ Let $J(f)$ be the ideal generated by $f'_1,\dots,f'_n$. Let us define a morphism of graded $\mathcal{O}$-algebras $$\varphi_f : \textrm{gr}^F({\mathcal{D}}[s])\simeq\mathcal{O}[s,\xi_1,\dots,\xi_n] \to \bigoplus_{d\geq 0} (\mathcal{O}f+J(f))^dT^d$$ by $\varphi_f (s)=fT$ and for all $i$, $\varphi_f(\xi_i)=f'_i T$. If $P(s)\in\textrm{ann}_{{\mathcal{D}}[s]}f^s$ with $\textrm{ord}^F(P(s))=d$, one sees that its $F$-symbol $\sigma(P(s))\in\textrm{ker}\,\varphi_f$ by considering the term of degree $d$ in $s$ of $P(s)f^s$ in the space $\mathcal{O}[1/f,s]f^s$. Then in general, $\textrm{gr}^F(\textrm{ann}_{{\mathcal{D}}[s]}f^s)\subset \textrm{ker}\,\varphi_f$. \[prop10\] Suppose $$\label{eq2.7} \textrm{gr}^F(\textrm{ann}_{{\mathcal{D}}[s]}f^s)=\textrm{ker}\,\varphi_f,$$ then the $W$-module $\textrm{gr}^V(\mathbf{R}({\mathcal{D}}_{x,t}f^s))$ is $h$-saturated. Consequently, the Betti numbers of $N$ are those of the module $\textrm{bigr}N$ by Corollary \[cor3\]. The condition (\[eq2.7\]) has been considered in [@narvaez09] (property (10)). It holds for a large class of hypersurface singularities, in particular for $f$ such that the ideal $\mathcal{O}f+J(f)$ is of linear type
ideal as a sub - object of $ W$ because of the isomorphism $ \mathbf{R}({\mathcal{D}}[s])\simeq\textrm{gr}^V_0(\mathbf{R}({\mathcal{D}}_{x, t})).$ # # # Condition for the $ h$-saturation of $ \mathbf{R}(\textrm{gr}^V(N))$ Let $ J(f)$ be the ideal generate by $ f'_1,\dots, f'_n$. lease us define a morphism of graded $ \mathcal{O}$-algebras $ $ \varphi_f: \textrm{gr}^F({\mathcal{D}}[s])\simeq\mathcal{O}[s,\xi_1,\dots,\xi_n ] \to \bigoplus_{d\geq 0 } (\mathcal{O}f+J(f))^dT^d$$ by $ \varphi_f (s)=fT$ and for all $ i$, $ \varphi_f(\xi_i)=f'_i T$. If $ P(s)\in\textrm{ann}_{{\mathcal{D}}[s]}f^s$ with $ \textrm{ord}^F(P(s))=d$, one sees that its $ F$-symbol $ \sigma(P(s))\in\textrm{ker}\,\varphi_f$ by study the term of degree $ d$ in $ s$ of $ P(s)f^s$ in the outer space $ \mathcal{O}[1 / f, s]f^s$. Then in general, $ \textrm{gr}^F(\textrm{ann}_{{\mathcal{D}}[s]}f^s)\subset \textrm{ker}\,\varphi_f$. \[prop10\ ] Suppose $ $ \label{eq2.7 } \textrm{gr}^F(\textrm{ann}_{{\mathcal{D}}[s]}f^s)=\textrm{ker}\,\varphi_f,$$ then the $ W$-module $ \textrm{gr}^V(\mathbf{R}({\mathcal{D}}_{x, t}f^s))$ is $ h$-saturated. Consequently, the Betti number of $ N$ are those of the module $ \textrm{bigr}N$ by Corollary \[cor3\ ]. The condition (\[eq2.7\ ]) has been consider in [ @narvaez09 ] (property (10) ). It holds for a large class of hypersurface singularities, in particular for $ f$ such that the ideal $ \mathcal{O}f+J(f)$ is of analogue type
idfal as a sub-object of $W$ necause of the isomorphmsm $\matgbf{R}({\mathzal{D}}[s])\simeq\textrm{gr}^V_0(\mathbf{R}({\mavhcao{D}}_{x,t})).$ ### Xondition for the $h$-satjration ov $\mathbf{E}(\texurm{gr}^V(N))$ Let $J(f)$ be vge ideal genedwted uy $f'_1,\dots,f'_n$. Let os define a korphism of grdddd $\mathcal{O}$-algebras $$\varphi_f : \textrm{gr}^S({\mathcak{D}}[d])\simeq\mathcal{O}[f,\xi_1,\dpes,\xi_h] \to \bigoplus_{d\geq 0} (\mathcal{O}f+J(f))^dT^d$$ gy $\varpii_f (s)=fT$ and for all $i$, $\varphi_f(\xi_i)=f'_i T$. If $P(d)\in\tfxtrm{ann}_{{\mathcal{D}}[s]}f^d$ with $\texttj{orq}^D(P(s))=d$, one seer that its $F$-symbol $\sigja(P(s))\in\textrm{ker}\,\varphi_f$ by consiaeriny the term if dehtee $d$ in $s$ oh $P(s)f^s$ in the spacc $\mathcdl{O}[1/f,s]f^s$. Then in genersl, $\vextem{gr}^F(\textrm{ann}_{{\mathcal{V}}[s]}f^s)\subset \textrm{ker}\,\darphi_f$. \[prmp10\] Suppose $$\label{eq2.7} \twxrrm{gr}^X(\texdrm{avb}_{{\maghczl{V}}[s]}f^a)=\textrl{kec}\,\varphi_f,$$ thsn the $W$-modyle $\textrm{gr}^V(\mathbf{T}({\mwnncal{D}}_{x,t}f^s))$ is $h$-satuwaeed. Consequently, the Betti numbers of $N$ ars those of the module $\twxtrm{bigr}N$ by Corollaty \[cor3\]. The sondition (\[eq2.7\]) has been considered in [@narvaez09] (propesty (10)). Mt hoodf dog a large class of hypersurface singularities, jn psrticular for $n$ such that the idrap $\kwthcal{O}f+J(f)$ is of liusad type
ideal as a sub-object of $W$ because isomorphism ### Condition the $h$-saturation of ideal by $f'_1,\dots,f'_n$. Let define a morphism graded $\mathcal{O}$-algebras $$\varphi_f : \textrm{gr}^F({\mathcal{D}}[s])\simeq\mathcal{O}[s,\xi_1,\dots,\xi_n] \to 0} (\mathcal{O}f+J(f))^dT^d$$ by $\varphi_f (s)=fT$ and for all $i$, $\varphi_f(\xi_i)=f'_i T$. If $P(s)\in\textrm{ann}_{{\mathcal{D}}[s]}f^s$ $\textrm{ord}^F(P(s))=d$, one sees that its $F$-symbol $\sigma(P(s))\in\textrm{ker}\,\varphi_f$ by considering the term of degree in of in space $\mathcal{O}[1/f,s]f^s$. Then in general, $\textrm{gr}^F(\textrm{ann}_{{\mathcal{D}}[s]}f^s)\subset \textrm{ker}\,\varphi_f$. \[prop10\] Suppose $$\label{eq2.7} \textrm{gr}^F(\textrm{ann}_{{\mathcal{D}}[s]}f^s)=\textrm{ker}\,\varphi_f,$$ then the $W$-module $\textrm{gr}^V(\mathbf{R}({\mathcal{D}}_{x,t}f^s))$ is Consequently, the Betti numbers of $N$ are those the module $\textrm{bigr}N$ by \[cor3\]. The condition (\[eq2.7\]) has considered [@narvaez09] (property It for large class of singularities, in particular for $f$ such that the ideal $\mathcal{O}f+J(f)$ is of linear type
ideal as a sub-object of $W$ becauSe of the isoMorphIsm $\MatHbF{R}({\maThcaL{D}}[s])\simeq\textrm{GR}^V_0(\maThbf{R}({\mathcal{D}}_{x,t})).$ ### ConditiOn for ThE $H$-satURaTion oF $\mathbf{r}(\TeXTRm{gR}^V(n))$ LEt $J(F)$ bE ThE ideaL geNerated By $f'_1,\dots,f'_n$. LEt uS dEfine a morphiSM oF graded $\matHcaL{O}$-algebras $$\vaRphI_f : \textRm{Gr}^F({\MAthcaL{D}}[s])\Simeq\MathcaL{o}[s,\xi_1,\doTs,\xi_n] \to \biGoPLus_{d\geQ 0} (\Mathcal{o}F+j(f))^DT^d$$ bY $\varphi_f (s)=fT$ and for ALl $I$, $\Varphi_f(\xi_i)=f'_i T$. IF $P(s)\in\tExTRm{ANN}_{{\maThcAl{D}}[s]}f^s$ with $\TeXtrm{oRD}^F(P(s))=d$, onE SeES THat ITs $F$-symbol $\sigmA(P(s))\in\textrm{KEr}\,\vArphi_f$ By ConSIderinG the tErM Of dEgree $d$ in $s$ of $p(s)f^s$ In the spacE $\mathcAL{O}[1/f,s]f^s$. THEn in genEral, $\teXtrM{gr}^f(\texTRm{AnN}_{{\maThCAl{D}}[S]}F^s)\SubSEt \tExtrm{ker}\,\VaRpHi_f$. \[prOp10\] SuPPOSE $$\labEl{eQ2.7} \texTrm{gr}^f(\textrm{ann}_{{\matHcaL{D}}[s]}f^S)=\TexTrm{keR}\,\varpHi_f,$$ tHeN the $W$-Module $\TextrM{gR}^V(\mathbf{R}({\mathcaL{D}}_{x,t}F^s))$ is $h$-satuRatEd. conSeQuentLY, the BeTti NumBers of $N$ Are thosE Of tHe MODUlE $\textrm{bigr}N$ by CoroLlARY \[cOr3\]. The conDition (\[EQ2.7\]) hAs BEen consiDeRed In [@naRVAez09] (prOperTY (10)). IT holds foR a largE ClAsS of hypeRsUrface SiNguLarIties, IN parTiculaR for $f$ sucH that THe ideal $\mathcal{o}F+J(f)$ is of linear TYpE
ideal as a sub-object of$W$ becaus e ofthe is om orph ism$\mathbf{R}({\ m athc al{D}}[s])\simeq\textr m{gr} ^V _ 0(\m a th bf{R} ({\math c al { D }}_ {x ,t })) .$ # ## Co ndi tion fo r the $h$- sat ur ation of $\m a th bf{R}(\tex trm {gr}^V(N))$ Le t $J(f )$ be the i dea l gen erated by $f' _1,\dots, f' _ n$. Le t us def i n ea mo rphism of graded$ \m a thcal{O}$-alge bras $ $\ v ar p h i_f :\textrm{gr }^ F({\m a thcal{D } }[ s ] ) \si m eq\mathcal{O} [s,\xi_1,\d o ts, \xi_n] \ to\ bigopl us_{d \g e q 0 } (\mathcal {O}f +J(f))^dT ^d$$ b y $\varp h i_f (s) =fT$ a ndfor all $i $, $\ va r phi _ f( \xi _ i)= f'_i T$. I f$P(s) \in\ t e x t rm{a nn} _{{\ mathc al{D}}[s]}f^s $ w ith$ \te xtrm{ ord}^ F(P( s) )=d$, one s ees t ha t its $F$-symbo l $\ sigma(P(s ))\ in \te xt rm{ke r }\,\va rph i_f $ by co nsideri n g t he t e rm of degree $d$ in$s $ of $P(s)f^ s$ int he s p ace $\ma th cal {O}[ 1 / f,s]f ^s$. Th en in ge neral, $\ te xtrm{gr }^ F(\tex tr m{a nn} _{{\m a thca l{D}}[ s]}f^s)\ subse t \textrm{ker}\ , \varphi_f$. \ [p r o p1 0 \] S upp ose $$\labe l{eq 2 .7}\tex t rm {gr } ^F(\t extrm {a n n} _ {{\mathcal{D}}[s]}f ^s )=\tex trm{k er}\,\varphi_ f,$$ thent h e $W$-mod ule$ \t e xtrm{gr}^V(\ma thbf{ R}({\mathc a l{D}}_{x ,t}f^ s))$ is$h$-satur a t ed. Cons equ ent ly, th e Be tti numbers o f $N$ar e those of the mo dul e $ \te xtr m{ bigr}N$ b y Coroll ar y\[ co r3\ ]. T h e condit io n ( \[ eq2 .7\]) has be en co nsid er ed in[@narva e z0 9 ] (pr op er ty ( 10) ). It h olds for a larg e class o f h y pers ur fa ce sing ularities, in p articularfo r $ f$ suc h that the ideal $\mathcal{O}f+J( f )$ is o f l inear typ e
ideal_as a_sub-object of $W$ because_of the_isomorphism_$\mathbf{R}({\mathcal{D}}[s])\simeq\textrm{gr}^V_0(\mathbf{R}({\mathcal{D}}_{x,t})).$ ### Condition_for_the $h$-saturation of_$\mathbf{R}(\textrm{gr}^V(N))$ Let $J(f)$ be_the ideal generated by_$f'_1,\dots,f'_n$. Let us_define_a morphism of graded $\mathcal{O}$-algebras $$\varphi_f : \textrm{gr}^F({\mathcal{D}}[s])\simeq\mathcal{O}[s,\xi_1,\dots,\xi_n] \to \bigoplus_{d\geq 0} (\mathcal{O}f+J(f))^dT^d$$ by $\varphi_f_(s)=fT$_and for_all_$i$,_$\varphi_f(\xi_i)=f'_i T$. If $P(s)\in\textrm{ann}_{{\mathcal{D}}[s]}f^s$ with_$\textrm{ord}^F(P(s))=d$, one sees that its_$F$-symbol $\sigma(P(s))\in\textrm{ker}\,\varphi_f$_by considering the term of degree $d$ in_$s$_of $P(s)f^s$ in_the space $\mathcal{O}[1/f,s]f^s$. Then in general, $\textrm{gr}^F(\textrm{ann}_{{\mathcal{D}}[s]}f^s)\subset \textrm{ker}\,\varphi_f$. \[prop10\] Suppose_$$\label{eq2.7} \textrm{gr}^F(\textrm{ann}_{{\mathcal{D}}[s]}f^s)=\textrm{ker}\,\varphi_f,$$ then the $W$-module $\textrm{gr}^V(\mathbf{R}({\mathcal{D}}_{x,t}f^s))$ is_$h$-saturated. Consequently, the_Betti_numbers_of $N$ are those_of the module $\textrm{bigr}N$ by Corollary_\[cor3\]. The condition (\[eq2.7\]) has been considered_in [@narvaez09] (property (10)). It holds for_a large class of hypersurface singularities,_in particular for $f$ such_that the_ideal $\mathcal{O}f+J(f)$ is of linear_type
0 & 1\\ 1 & 0 \end{matrix}\right], \left[\begin{matrix} 0 & 1\\0 & 1\\ 0 & 1 \end{matrix}\right], \left[\begin{matrix} 1 & 0\\1 & 0\\ 0 & 1 \end{matrix}\right]$$ and can be written as a union of Cartesian products of convex sets, illustrated in Figure \[figure:convexrepresentable\], $$G = \bigcup_{\theta\in\Delta_A} G_\theta,\qquad G_\theta = \{ \theta \} \times \{ \theta \} \times \Delta_A. $$ As mentioned in Section \[sec:optimalcontrol\], the set $F\subseteq\Delta{W\times A}$ of joint distributions that are compatible with the representable conditionals $G=f_\beta(\Delta_{S,A})\subseteq\Delta_{W,A}$, may not be convex. In the following we describe large convex subsets of $F$, depending on the properties of $G$. We use the following definitions. - Given a set of distributions ${\mathcal{P}}\subseteq\Delta_W$ and a set of kernels ${\mathcal{G}}\subseteq\Delta_{W,A}$, let $${\mathcal{P}}\ast {\mathcal{G}}: =\Big\{ q(w,a) = p(w)g(a|w) \in \Delta_{W\times A} \colon p\in{\mathcal{P}}, g\in {\mathcal{G}}\Big\}$$ denote the set of joint distributions over world states and actions, with world state marginals in ${\mathcal{P}}$ and conditional distributions in ${\mathcal{G}}$. - For any $V \subseteq W$ let $$\Delta_W(V) : = \Big\{ p\in \Delta_W \colon {\operatorname{supp}}(p):=\{w\in W\colon p(w)>0 \}\subseteq V\Big\}$$ denote the set of world state distributions with support in $V$. - Given a subset $V\subseteq W$ and a set of kernels ${\mathcal{G}}\subseteq\Delta_{W,A}$, let $${\mathcal{G}}|_V := \Big\{ h\in \Delta_{V,A}\colon h(\cdot|w) = g(\
0 & 1\\ 1 & 0 \end{matrix}\right ], \left[\begin{matrix } 0 & 1\\0 & 1\\ 0 & 1 \end{matrix}\right ], \left[\begin{matrix } 1 & 0\\1 & 0\\ 0 & 1 \end{matrix}\right]$$ and can be written as a union of Cartesian products of convex sets, illustrate in name   \[figure: convexrepresentable\ ], $ $ G = \bigcup_{\theta\in\Delta_A } G_\theta,\qquad G_\theta = \ { \theta \ } \times \ { \theta \ } \times \Delta_A. $ $ As mentioned in Section   \[sec: optimalcontrol\ ], the set $ F\subseteq\Delta{W\times A}$ of joint distribution that are compatible with the representable conditionals $ G = f_\beta(\Delta_{S, A})\subseteq\Delta_{W, A}$, may not be convex. In the following we describe big convex subset of $ F$, depending on the properties of   $ G$. We use the following definitions. - give a set of distributions $ { \mathcal{P}}\subseteq\Delta_W$ and a bent of kernel $ { \mathcal{G}}\subseteq\Delta_{W, A}$, let $ $ { \mathcal{P}}\ast { \mathcal{G } }: = \Big\ { q(w, a) = p(w)g(a|w) \in \Delta_{W\times A } \colon p\in{\mathcal{P } }, g\in { \mathcal{G}}\Big\}$$ denote the set of joint distributions over world state and actions, with world state marginals in $ { \mathcal{P}}$ and conditional distribution in $ { \mathcal{G}}$. - For any $ V \subseteq W$ let $ $ \Delta_W(V): = \Big\ { p\in \Delta_W \colon { \operatorname{supp}}(p):=\{w\in W\colon p(w)>0 \}\subseteq V\Big\}$$ denote the set of universe state distributions with support in $ V$. - Given a subset $ V\subseteq W$ and a set of kernel $ { \mathcal{G}}\subseteq\Delta_{W, A}$, let $ $ { \mathcal{G}}|_V: = \Big\ { h\in \Delta_{V, A}\colon h(\cdot|w) = g(\
0 & 1\\ 1 & 0 \end{matrix}\right], \left[\begin{matrix} 0 & 1\\0 & 1\\ 0 & 1 \end{matrix}\right], \left[\uegib{matrux} 1 & 0\\1 & 0\\ 0 & 1 \end{mxtrix}\righn]$$ and can be xritten as a union of Cavcesiah proburts of convex sgts, illustraded in Figure \[fhgjrz:convexrepresentable\], $$G = \bigcup_{\theta\ig\Delta_A} G_\hheta,\qquad D_\theua = \{ \thsna \} \times \{ \theta \} \times \Delta_A. $$ As kentioned in Xection \[sec:optimalcontrol\], tje sft $F\subseteq\Delta{W\himes A}$ of hoine distributiovs that art eompatible sith the representable conditiovals $Y=f_\beta(\Delta_{W,A})\wubdgteq\Delta_{W,A}$, nay njt be convex. In the xollowimg we describe lacge xonvex subsets of $F$, dxpending on the propgrties of $G$. Wz use the following dwfunitimns. - Gixwn x stt pf distrlbuvions ${\mathczl{P}}\subseteq\Eelta_W$ and a set of kqgmels ${\mathcal{F}}\subseeez\Delta_{W,A}$, let $${\mathcal{P}}\ast {\mathcal{G}}: =\Big\{ q(f,a) = p(w)g(a|w) \in \Delta_{W\times A} \colon p\in{\mathcal{P}}, g\ij {\mathcal{D}}\Big\}$$ denote the set of joint distributions over wmrld agatts and xxtlons, with world state marginals in ${\mathcal{P}}$ ags vokditional distrinutions in ${\mathcal{B}}$. - Gjr any $V \subsgteq W$ let $$\Delta_W(V) : = \Big\{ p\in \Delta_W \colob {\operatownamr{supp}}(p):=\{w\in W\colon p(w)>0 \}\subsetew V\Big\}$$ denotv thw set of world stace distributnons woth sipport in $V$. - Given a sbbset $B\subseteq W$ and a sef of kernels ${\mathzal{N}}\sutseteq\Delta_{W,A}$, let $${\mathcal{G}}|_D := \Big\{ h\ii \Delca_{V,A}\colov h(\ccot|w) = d(\
0 & 1\\ 1 & 0 \end{matrix}\right], & & 1\\ & 1 \end{matrix}\right], 0\\ & 1 \end{matrix}\right]$$ can be written a union of Cartesian products of sets, illustrated in Figure \[figure:convexrepresentable\], $$G = \bigcup_{\theta\in\Delta_A} G_\theta,\qquad G_\theta = \{ \theta \times \{ \theta \} \times \Delta_A. $$ As mentioned in Section \[sec:optimalcontrol\], the $F\subseteq\Delta{W\times of distributions are compatible with the representable conditionals $G=f_\beta(\Delta_{S,A})\subseteq\Delta_{W,A}$, may not be convex. In the following we describe convex subsets of $F$, depending on the properties $G$. We use the definitions. - Given a set distributions and a of ${\mathcal{G}}\subseteq\Delta_{W,A}$, $${\mathcal{P}}\ast {\mathcal{G}}: =\Big\{ = p(w)g(a|w) \in \Delta_{W\times A} \colon p\in{\mathcal{P}}, g\in {\mathcal{G}}\Big\}$$ denote the set of joint distributions over world and actions, state marginals ${\mathcal{P}}$ conditional in ${\mathcal{G}}$. - $V \subseteq W$ let $$\Delta_W(V) : \Delta_W \colon {\operatorname{supp}}(p):=\{w\in W\colon p(w)>0 \}\subseteq V\Big\}$$ denote set of state distributions with support in $V$. Given a subset $V\subseteq W$ and a set kernels ${\mathcal{G}}\subseteq\Delta_{W,A}$, let $${\mathcal{G}}|_V := \Big\{ h\in \Delta_{V,A}\colon h(\cdot|w) = g(\
0 & 1\\ 1 & 0 \end{matrix}\right], \left[\begin{maTrix} 0 & 1\\0 & 1\\ 0 & 1 \end{matRix}\riGht], \LefT[\bEgin{MatrIx} 1 & 0\\1 & 0\\ 0 & 1 \end{matrix}\rigHT]$$ and Can be written as a union of carteSiAN proDUcTs of cOnvex seTS, iLLUstRaTeD in fiGUrE \[figuRe:cOnvexrePresentablE\], $$G = \bIgCup_{\theta\in\DeLTa_a} G_\theta,\qquAd G_\Theta = \{ \theta \} \tiMes \{ \Theta \} \tImEs \DELta_A. $$ AS meNtionEd in SeCTion \[seC:optimalcOnTRol\], the SEt $F\subsETEq\deltA{W\times A}$ of joint diSTrIButions that are CompatIbLE wITH thE rePresentablE cOnditIOnals $G=f_\BEtA(\dELta_{s,a})\subseteq\DeltA_{W,A}$, may not be COnvEx. In thE fOllOWing we DescrIbE LarGe convex subSets Of $F$, dependIng on tHE properTIes of $G$. WE use thE foLloWing DEfInItiOnS. - givEN a Set OF diStributiOnS ${\mAthcaL{P}}\suBSETEq\DeLta_w$ and A set oF kernels ${\mathcAl{G}}\SubsETeq\delta_{w,A}$, let $${\MathCaL{P}}\ast {\MathcaL{G}}: =\Big\{ Q(w,A) = p(w)g(a|w) \in \Delta_{W\tImes a} \colon p\in{\MatHcAl{P}}, G\iN {\mathCAl{G}}\Big\}$$ DenOte The set oF joint dIStrIbUTIOnS over world states anD aCTIoNs, with woRld staTE mArGInals in ${\mAtHcaL{P}}$ anD COnditIonaL DiStributiOns in ${\mAThCaL{G}}$. - For anY $V \SubsetEq w$ leT $$\DeLta_W(V) : = \bIg\{ p\iN \Delta_w \colon {\opEratoRName{supp}}(p):=\{w\in W\cOLon p(w)>0 \}\subseteq v\biG\}$$ DEnOTe thE seT of world staTe diSTribUtioNS wIth SUpporT in $V$. - GIvEN a SUbset $V\subseteq W$ and a SeT of kerNels ${\mAthcal{G}}\subsetEq\Delta_{W,A}$, lET $${\MAthcal{G}}|_V := \big\{ h\IN \DELta_{V,A}\colon h(\cdoT|w) = g(\
0 & 1\\ 1 & 0 \end{ma trix}\righ t], \lef t[\b egin{matrix} 0& 1\\0 & 1\\ 0 & 1 \en d{ m atri x }\ right ], \le ft [\ beg in { ma trix} 1 & 0 \\1 & 0\\0 & 1 \end{m a tr ix}\right] $$and can be w rit ten as a un i on of Ca rtesi an pro d ucts o f convexse t s, ill u strated i nFigu re \[figure:conve x re p resentable\],$$G =\b i gc u p _{\ the ta\in\Delt a_ A} G_ \ theta,\ q qu a d G_\theta = \{ \theta \}\ tim es \{\t het a \} \t imes\D e lta _A. $$ As mentione d in S e ction \ [ sec:opt imalco ntr ol\ ], t h ese t $ F\ s ubs e te q\D e lta {W\times A }$ of j oint d i s trib uti onsthatare compatibl e w itht herepre senta bleco nditi onals$G=f_ \b eta(\Delta_{S,A })\s ubseteq\D elt a_ {W, A} $, ma y not b e c onv ex. Inthe fol l owi ng w e d escribe large conv ex s ub sets of$F$, d e pe nd i ng on th epro pert i e s of$G$. We use the follo w in gdefinit io ns. - Gi ven a se t ofdistri butions${\ma t hcal{P}}\subse t eq\Delta_W$ a n da se t ofker nels ${\mat hcal { G}}\ subs e te q\D e lta_{ W,A}$ ,l et $${\mathcal{P}}\ast { \mathc al{G} }: =\Big\{ q( w,a) = p(w ) g ( a|w) \in \De l ta _ {W\times A} \c olonp\in{\math c al{P}},g\in{\mathca l{G}}\Big \ } $$ denot e t heset of j oi nt distributi o n s ov er worldsta tes and ac tio ns, wi th world st ate marg in al sin ${ \math c al{P}}$an d c on dit ional distri butio ns i n${ \ mat hcal{G} } $. - Fo rany$V\s ubset eq W $ le t $$\De lta_W(V): = \Big \{ p \in \De lta_W \colon{\ operatorna me {su pp}}(p ) : =\{w\inW\colon p(w)>0 \}\subse t eq V\Bi g\} $$ de note the setofworldsta t e dist ributi ons w it h s u p porti n $ V$. - Givena sub set $ V\ subs eteq W$ and a set of kern e ls${\mathcal{G} }\s ubse t e q\ Del t a_ { W,A }$ , le t $${\mathcal{G}} |_V := \Bi g\ { h \in \Delta _ {V, A} \colonh(\cdot |w) = g(\
0 &_1\\ 1_& 0 _ _\end{matrix}\right],_ __ _ \left[\begin{matrix} _ 0_& 1\\0 &_1\\_0 & 1 \end{matrix}\right], __ \left[\begin{matrix} ___ 1 & 0\\1 &_0\\ 0 & 1 _ _ \end{matrix}\right]$$ and can be written as a_union_of Cartesian products_of convex sets, illustrated in Figure \[figure:convexrepresentable\], $$G = \bigcup_{\theta\in\Delta_A}_G_\theta,\qquad G_\theta =_\{ \theta \}_\times_\{_\theta \} \times \Delta_A._ $$ As mentioned_in Section \[sec:optimalcontrol\], the set $F\subseteq\Delta{W\times A}$_of joint distributions that are compatible with_the representable conditionals $G=f_\beta(\Delta_{S,A})\subseteq\Delta_{W,A}$, may not_be convex. In the following_we describe_large convex subsets of $F$,_depending on the_properties of $G$._We use the_following definitions. - Given a_set of distributions_${\mathcal{P}}\subseteq\Delta_W$ and a set of kernels_${\mathcal{G}}\subseteq\Delta_{W,A}$,_let $${\mathcal{P}}\ast {\mathcal{G}}:_=\Big\{_q(w,a)_= p(w)g(a|w)_\in \Delta_{W\times A}_\colon_p\in{\mathcal{P}}, g\in_{\mathcal{G}}\Big\}$$_denote the set of joint distributions_over_world states and actions, with world state_marginals in ${\mathcal{P}}$ and_conditional_distributions in ${\mathcal{G}}$. - _ For any $V \subseteq_W$ let $$\Delta_W(V) : = \Big\{_p\in \Delta_W_\colon {\operatorname{supp}}(p):=\{w\in_W\colon p(w)>0 \}\subseteq V\Big\}$$ denote the set of world state distributions_with support in $V$. - _Given a subset $V\subseteq_W$ and_a_set of kernels_${\mathcal{G}}\subseteq\Delta_{W,A}$,_let $${\mathcal{G}}|_V_:= \Big\{ h\in \Delta_{V,A}\colon h(\cdot|w) = g(\
(m,q)$. The following notation will be used. - $\omega_{\alpha}:{\mathbb{F}_{q^n}}\to{\mathbb{F}_{q^n}}$, $x\mapsto\alpha x$, for any $\alpha\in{\mathbb{F}_{q^n}}$. - $\mathcal{F}_n=\{\omega_{\alpha}\colon\alpha\in{\mathbb{F}_{q^n}}\}$, which is a field isomorphic to ${\mathbb{F}_{q^n}}$. - $\mathcal{F}_{n,q}=\{\omega_{\alpha}\colon \alpha\in{{\mathbb F}_{q}}\}$, which is a subfield of $\mathcal{F}_n$ isomorphic to ${{\mathbb F}_{q}}$. - $\tau_{\mathbf{v}}:{\mathbb{F}_{q^n}}\to W_1$, $\lambda\mapsto \lambda \mathbf{v}$, for any $\mathbf{v}\in W_1$. We define a family of ${{\mathbb F}_{q}}$-linear RM codes associated with an ${{\mathbb F}_{q}}$-vector space $U$. Let $n,r,k$ be positive integers with $k<rn$, $U$ be a $k$-dimensional ${{\mathbb F}_{q}}$-subspace of an $r$-dimensional ${\mathbb{F}_{q^n}}$-vector space $V$, $W$ be an $(rn-k)$-dimensional ${{\mathbb F}_{q}}$-vector space, and $G:V\to W$ be an ${{\mathbb F}_{q}}$-linear map with kernel $U$. For any $\mathbf{v} \in V$ define the ${{\mathbb F}_{q}}$-linear map $\Gamma_{\mathbf{v}}=G\circ\tau_{\mathbf{v}}$. \[th:construction\] Let $V=V(r,q^n)$ and $W=V(rn-k,q)$. Let $U=V(k,q)$ be an ${{\mathbb F}_{q}}$-subspace of $V$, and $G:V\to W$ be an ${{\mathbb F}_{q}}$-linear map with $\ker(G)=U$. Define $$\iota=\max\{\dim_{{{\mathbb F}_{q}}}(U\cap\langle\mathbf{v}\rangle_{{\mathbb{F}_{q^n}}}) \colon \mathbf{v}\in V^* \}.$$ If $\iota<n$, then the pair
( m, q)$. The following notation will be used. - $ \omega_{\alpha}:{\mathbb{F}_{q^n}}\to{\mathbb{F}_{q^n}}$, $ x\mapsto\alpha x$, for any $ \alpha\in{\mathbb{F}_{q^n}}$. - $ \mathcal{F}_n=\{\omega_{\alpha}\colon\alpha\in{\mathbb{F}_{q^n}}\}$, which is a field isomorphous to $ { \mathbb{F}_{q^n}}$. - $ \mathcal{F}_{n, q}=\{\omega_{\alpha}\colon \alpha\in{{\mathbb F}_{q}}\}$, which is a subfield of $ \mathcal{F}_n$ isomorphous to $ { { \mathbb F}_{q}}$. - $ \tau_{\mathbf{v}}:{\mathbb{F}_{q^n}}\to W_1 $, $ \lambda\mapsto \lambda \mathbf{v}$, for any $ \mathbf{v}\in W_1$. We specify a family of $ { { \mathbb F}_{q}}$-linear RM code consort with an $ { { \mathbb F}_{q}}$-vector space $ U$. Let $ n, roentgen, k$ be positive integers with $ kelvin < rn$, $ U$ be a $ k$-dimensional $ { { \mathbb F}_{q}}$-subspace of an $ r$-dimensional $ { \mathbb{F}_{q^n}}$-vector space $ V$, $ W$ be an $ (rn - k)$-dimensional $ { { \mathbb F}_{q}}$-vector space, and $ thousand: V\to W$ be an $ { { \mathbb F}_{q}}$-linear map with kernel $ U$. For any $ \mathbf{v } \in V$ define the $ { { \mathbb F}_{q}}$-linear map $ \Gamma_{\mathbf{v}}=G\circ\tau_{\mathbf{v}}$. \[th: construction\ ] get $ five = V(r, q^n)$ and $ W = V(rn - k, q)$. Let $ U = V(k, q)$ be an $ { { \mathbb F}_{q}}$-subspace of $ V$, and $ G: V\to W$ be an $ { { \mathbb F}_{q}}$-linear map with $ \ker(G)=U$. Define $ $ \iota=\max\{\dim_{{{\mathbb F}_{q}}}(U\cap\langle\mathbf{v}\rangle_{{\mathbb{F}_{q^n } } }) \colon \mathbf{v}\in V^ * \}.$$ If $ \iota < n$, then the pair
(m,q)$. Hhe following notation wlll be used. - $\omggq_{\alpha}:{\kathbb{R}_{q^n}}\to{\matfbb{F}_{q^n}}$, $x\mapsto\alpha x$, for anb $\alpha\in{\nathbb{F}_{q^n}}$. - $\mathcal{F}_n=\{\ooega_{\alpha}\bolon\alphq\in{\methbb{F}_{q^n}}\}$, which is a field isomkvphic vo ${\mathbb{F}_{q^n}}$. - $\kathcal{F}_{n,q}=\{\mmega_{\alpha}\colot \xl'ha\in{{\mathbb F}_{q}}\}$, which is a subfield os $\mathcsl{V}_n$ isomorphic jo ${{\mauhbf F}_{q}}$. - $\tau_{\mathbf{v}}:{\mathbb{F}_{q^n}}\to W_1$, $\lambda\mzpsto \lembda \mathbf{v}$, fpr any $\mathbf{v}\in W_1$. We definf a vamily of ${{\mathbb F}_{e}}$-linear RM xodef associated dith an ${{\mauhyb F}_{q}}$-vector space $U$. Let $n,r,k$ be positive intdgers with $k<rn$, $O$ ge d $k$-dimensioial ${{\manhbb F}_{q}}$-subspagv of an $r$-dimenxional ${\mathbb{F}_{a^n}}$-vxctoe space $V$, $W$ be an $(rn-k)$-vimensional ${{\mathbb F}_{z}}$-vector s[aee, and $G:V\to W$ be an ${{\mqtybb F}_{x}}$-linaar oqp ditg lednel $U$. Foc any $\mathbr{v} \in V$ defune the ${{\mathbb F}_{q}}$-limewg map $\Gamma_{\mafhbf{v}}=G\siwc\tau_{\mathbf{v}}$. \[th:construction\] Let $V=V(r,q^n)$ ang $W=B(rn-k,q)$. Let $U=V(k,q)$ be an ${{\marhbb F}_{q}}$-subspace of $V$, wnd $G:V\to R$ be an ${{\mathbb F}_{q}}$-linear map with $\ker(G)=U$. Define $$\iotd=\max\{\dmm_{{{\oatknn F}_{q}}}(J\xao\langle\mathbf{v}\rangle_{{\mathbb{F}_{q^n}}}) \colon \mathbf{v}\in B^* \}.$$ Pf $\iota<n$, then the pair
(m,q)$. The following notation will be used. $x\mapsto\alpha for any - $\mathcal{F}_n=\{\omega_{\alpha}\colon\alpha\in{\mathbb{F}_{q^n}}\}$, which ${\mathbb{F}_{q^n}}$. $\mathcal{F}_{n,q}=\{\omega_{\alpha}\colon \alpha\in{{\mathbb F}_{q}}\}$, is a subfield $\mathcal{F}_n$ isomorphic to ${{\mathbb F}_{q}}$. - W_1$, $\lambda\mapsto \lambda \mathbf{v}$, for any $\mathbf{v}\in W_1$. We define a family of F}_{q}}$-linear RM codes associated with an ${{\mathbb F}_{q}}$-vector space $U$. Let $n,r,k$ be integers $k<rn$, be $k$-dimensional ${{\mathbb F}_{q}}$-subspace of an $r$-dimensional ${\mathbb{F}_{q^n}}$-vector space $V$, $W$ be an $(rn-k)$-dimensional ${{\mathbb F}_{q}}$-vector space, $G:V\to W$ be an ${{\mathbb F}_{q}}$-linear map with $U$. For any $\mathbf{v} V$ define the ${{\mathbb F}_{q}}$-linear $\Gamma_{\mathbf{v}}=G\circ\tau_{\mathbf{v}}$. Let $V=V(r,q^n)$ $W=V(rn-k,q)$. $U=V(k,q)$ an ${{\mathbb F}_{q}}$-subspace $V$, and $G:V\to W$ be an ${{\mathbb F}_{q}}$-linear map with $\ker(G)=U$. Define $$\iota=\max\{\dim_{{{\mathbb F}_{q}}}(U\cap\langle\mathbf{v}\rangle_{{\mathbb{F}_{q^n}}}) \colon \mathbf{v}\in V^* If $\iota<n$, pair
(m,q)$. The following notation wilL be used. - $\omeGa_{\alpHa}:{\mAthBb{f}_{q^n}}\tO{\matHbb{F}_{q^n}}$, $x\mapsto\aLPha x$, For any $\alpha\in{\mathbb{F}_{q^n}}$. - $\MathcAl{f}_N=\{\omeGA_{\aLpha}\cOlon\alpHA\iN{\MAthBb{f}_{q^N}}\}$, whIcH Is A fielD isOmorphiC to ${\mathbb{F}_{Q^n}}$. - $\mAtHcal{F}_{n,q}=\{\omega_{\ALpHa}\colon \alpHa\iN{{\mathbb F}_{q}}\}$, whiCh iS a subfIeLd oF $\MathcAl{F}_N$ isomOrphic TO ${{\mathbB F}_{q}}$. - $\tau_{\matHbF{V}}:{\mathbB{f}_{q^n}}\to W_1$, $\lAMBdA\mapSto \lambda \mathbf{v}$, fOR aNY $\mathbf{v}\in W_1$. We dEfine a FaMIlY OF ${{\maThbB F}_{q}}$-linear Rm cOdes aSSociateD WiTH AN ${{\maTHbb F}_{q}}$-vector spAce $U$. Let $n,r,k$ bE PosItive iNtEgeRS with $k<Rn$, $U$ be A $k$-DImeNsional ${{\mathBb F}_{q}}$-Subspace oF an $r$-diMEnsionaL ${\Mathbb{F}_{Q^n}}$-vectOr sPacE $V$, $W$ bE An $(Rn-K)$-diMeNSioNAl ${{\MatHBb F}_{Q}}$-vector sPaCe, And $G:V\To W$ bE AN ${{\MAthbB F}_{q}}$-LineAr map With kernel $U$. FoR anY $\matHBf{v} \In V$ deFine tHe ${{\maThBb F}_{q}}$-lInear mAp $\GamMa_{\Mathbf{v}}=G\circ\tau_{\MathBf{v}}$. \[th:consTruCtIon\] leT $V=V(r,q^N)$ And $W=V(rN-k,q)$. let $u=V(k,q)$ be aN ${{\mathbb f}_{Q}}$-suBsPACE oF $V$, and $G:V\to W$ be an ${{\mathBb f}_{Q}}$-LiNear map wIth $\ker(g)=u$. DEfINe $$\iota=\maX\{\dIm_{{{\mAthbB f}_{Q}}}(U\cap\LangLE\mAthbf{v}\raNgle_{{\maTHbB{F}_{Q^n}}}) \colon \MaThbf{v}\iN V^* \}.$$ if $\iOta<N$, then THe paIr
(m,q)$. The following nota tion willbe us ed. - $\o mega _{\alpha}:{\ma t hbb{ F}_{q^n}}\to{\mathbb{F }_{q^ n} } $, $ x \m apsto \alphax $, f oran y$\a lp h a\ in{\m ath bb{F}_{ q^n}}$. - $\ mathcal{F}_n = \{ \omega_{\a lph a}\colon\alp ha\ in{\ma th bb{ F }_{q^ n}} \}$,whichi s a fi eld isomo rp h ic to$ {\mathb b { F} _{q^ n}}$. - $\math c al { F}_{n,q}=\{\om ega_{\ al p ha } \ col on\alpha\in{ {\ mathb b F}_{q} } \} $ , whi c h is a subfie ld of $\mat h cal {F}_n$ i som o rphicto ${ {\ m ath bb F}_{q}}$ . - $\tau_ {\math b f{v}}:{ \ mathbb{ F}_{q^ n}} \to W_1 $ ,$\ lam bd a \ma p st o \ l amb da \math bf {v }$, f or a n y $ \mat hbf {v}\ in W_ 1$. We defin e a fam i lyof ${ {\mat hbbF} _{q}} $-line ar RM c odes associated wit h an ${{\ mat hb b F }_ {q}}$ - vector sp ace $U$. Let $n, r ,k$ b e p os itive integers wit h$ k <r n$, $U$be a $ k $- di m ensional $ {{\ math b b F}_{ q}}$ - su bspace o f an $ r $- di mension al ${\ma th bb{ F}_ {q^n} } $-ve ctor s pace $V$ , $W$ be an $(rn-k)$ - dimensional $ { {\ m a th b b F} _{q }}$-vectorspac e , an d $G : V\ toW $ bean ${ {\ m at h bb F}_{q}}$-linearma p with kern el $U$. For a ny $\mathb f { v } \in V$ def i ne the ${{\mathbb F}_{ q}}$-linea r map $\G amma_ {\mathbf {v}}=G\ci r c \tau_{\m ath bf{ v}} $.\ [t h:constructio n \ ] Le t$V=V(r, q^n )$ and$W= V(r n-k ,q) $. Let $U=V (k,q)$ b ean $ {{ \ma thbbF }_{q}}$- su bsp ac e o f $V$ , and $ G:V\t o W$ b ea n $ {{\math b bF } _{q} }$ -l inea r m ap with $\k e r(G )=U$. D efine $$\ iot a =\ma x\ {\ dim_{{{ \mathbb F}_{q }} }(U\cap\la ng le\ mathbf { v }\rangle _{{\mathbb{F}_{q^n}}})\ colon \ mat hbf{v }\in V^* \}. $$If $\i ota < n$, th en the pair
(m,q)$. The_following notation_will be used. - _ $\omega_{\alpha}:{\mathbb{F}_{q^n}}\to{\mathbb{F}_{q^n}}$,_$x\mapsto\alpha_x$, for_any_$\alpha\in{\mathbb{F}_{q^n}}$. - _$\mathcal{F}_n=\{\omega_{\alpha}\colon\alpha\in{\mathbb{F}_{q^n}}\}$, which is_a field isomorphic to_${\mathbb{F}_{q^n}}$. - _$\mathcal{F}_{n,q}=\{\omega_{\alpha}\colon_\alpha\in{{\mathbb F}_{q}}\}$, which is a subfield of $\mathcal{F}_n$ isomorphic to ${{\mathbb F}_{q}}$. - _$\tau_{\mathbf{v}}:{\mathbb{F}_{q^n}}\to_W_1$, $\lambda\mapsto_\lambda_\mathbf{v}$,_for any $\mathbf{v}\in W_1$. We define_a family of ${{\mathbb F}_{q}}$-linear_RM codes_associated with an ${{\mathbb F}_{q}}$-vector space $U$. Let $n,r,k$_be_positive integers with_$k<rn$, $U$ be a $k$-dimensional ${{\mathbb F}_{q}}$-subspace of an_$r$-dimensional ${\mathbb{F}_{q^n}}$-vector space $V$, $W$ be_an $(rn-k)$-dimensional ${{\mathbb_F}_{q}}$-vector_space,_and $G:V\to W$ be_an ${{\mathbb F}_{q}}$-linear map with kernel_$U$. For any $\mathbf{v} \in V$_define the ${{\mathbb F}_{q}}$-linear map $\Gamma_{\mathbf{v}}=G\circ\tau_{\mathbf{v}}$. \[th:construction\] Let_$V=V(r,q^n)$ and $W=V(rn-k,q)$. Let $U=V(k,q)$ be_an ${{\mathbb F}_{q}}$-subspace of $V$,_and $G:V\to_W$ be an ${{\mathbb F}_{q}}$-linear_map with $\ker(G)=U$._Define $$\iota=\max\{\dim_{{{\mathbb_F}_{q}}}(U\cap\langle\mathbf{v}\rangle_{{\mathbb{F}_{q^n}}}) \colon \mathbf{v}\in_V^* \}.$$ If $\iota<n$, then_the pair
beta \left(1-\frac{1}{2}|\boldxi|^2+\frac{3}{8}|\boldxi|^4\right) \boldsymbol{\mu}_m^{\prime e}-\frac{1}{2}\left(1-\frac{3}{4}|\boldxi|^2 +\frac{5}{8}|\boldxi|^4\right) \left(\boldxi\times\boldsymbol{\mu}_p^{\prime\tilde{e}}\right)\right]. \end{split}$$ It is shown that intrinsic dipole moments transform like $\left(1-\frac{1}{2}|\boldxi|^2+\frac{3}{8}|\boldxi|^4\right)\approx 1/\operatorgamma$ and the boosted dipole moments transform as $\frac{1}{2}\left(1-\frac{3}{4}|\boldxi|^2+\frac{5}{8}|\boldxi|^4\right)\approx(1-\frac{\operatorgamma}{\operatorgamma+1})$ (see Eq. (\[transgamma\])). Therefore, $(\boldsymbol{\mu}_p^{\prime\tilde{e}})_\mathrm{eff}$ and $(\boldsymbol{\mu}_m^{\prime e})_\mathrm{eff}$ do not form a second rank tensor in the sense that their transformation is not a covariant form like Eq. (\[LTdipole\]), but the following form: $$\begin{split} &(\boldsymbol{\mu}_p^{\tilde{e}})_\mathrm{eff} \approx\matrixbeta\frac{1}{\operatorgamma}\boldsymbol{\mu}_p^{\prime\tilde{e}} +\left(1-\frac{\operatorgamma}{\operatorgamma+1}\right) \left(\operatorboost\times\boldsymbol{\mu}_m^{\prime e}\right),\\ &(\boldsymbol{\mu}_m^e)_\mathrm{eff} \approx\matrixbeta\frac{1}{\operatorgamma}\boldsymbol{\mu}_m^{\prime e} -\left(1-\frac{\operatorgamma}{\operatorgamma+1}\right) \left(\operatorboost\times\boldsymbol{\mu}_p^{\prime\tilde{e}}\right).\\ \end{split}$$ This implies that an energy caused by dipole moments in the description of Dirac Hamiltonian is not simply the contraction of tensorial dipole density and field tensor: $H_\mathrm{spin}\neq-\boldsymbol{\mu}_p\cdot\mathbf{E}-\
beta \left(1-\frac{1}{2}|\boldxi|^2+\frac{3}{8}|\boldxi|^4\right) \boldsymbol{\mu}_m^{\prime e}-\frac{1}{2}\left(1-\frac{3}{4}|\boldxi|^2 + \frac{5}{8}|\boldxi|^4\right) \left(\boldxi\times\boldsymbol{\mu}_p^{\prime\tilde{e}}\right)\right ]. \end{split}$$ It is shown that intrinsic dipole moments transform like $ \left(1-\frac{1}{2}|\boldxi|^2+\frac{3}{8}|\boldxi|^4\right)\approx 1/\operatorgamma$ and the boosted dipole here and now translate as $ \frac{1}{2}\left(1-\frac{3}{4}|\boldxi|^2+\frac{5}{8}|\boldxi|^4\right)\approx(1-\frac{\operatorgamma}{\operatorgamma+1})$ (see Eq.   (\[transgamma\ ]) ). Therefore, $ (\boldsymbol{\mu}_p^{\prime\tilde{e}})_\mathrm{eff}$ and $ (\boldsymbol{\mu}_m^{\prime e})_\mathrm{eff}$ do not imprint a second social station tensor in the sense that their transformation is not a covariant form like Eq.   (\[LTdipole\ ]), but the following shape: $ $ \begin{split } & (\boldsymbol{\mu}_p^{\tilde{e}})_\mathrm{eff } \approx\matrixbeta\frac{1}{\operatorgamma}\boldsymbol{\mu}_p^{\prime\tilde{e } } + \left(1-\frac{\operatorgamma}{\operatorgamma+1}\right) \left(\operatorboost\times\boldsymbol{\mu}_m^{\prime e}\right),\\ & (\boldsymbol{\mu}_m^e)_\mathrm{eff } \approx\matrixbeta\frac{1}{\operatorgamma}\boldsymbol{\mu}_m^{\prime einsteinium } -\left(1-\frac{\operatorgamma}{\operatorgamma+1}\right) \left(\operatorboost\times\boldsymbol{\mu}_p^{\prime\tilde{e}}\right).\\ \end{split}$$ This implies that an energy caused by dipole moments in the description of Dirac Hamiltonian is not plainly the contraction of tensorial dipole density and playing field tensor: $ H_\mathrm{spin}\neq-\boldsymbol{\mu}_p\cdot\mathbf{E}-\
betw \left(1-\frac{1}{2}|\boldxi|^2+\frac{3}{8}|\boldxl|^4\right) \boldsymbol{\mu}_m^{\primx e}-\frac{1}{2}\meft(1-\frac{3}{4}|\coldxi|^2 +\frac{5}{8}|\boldxi|^4\right) \left(\bolvxi\tumes\bildsymbol{\mu}_p^{\prime\tilde{d}}\right)\rigjt]. \end{splut}$$ It us shown tizt intrlusic slpole noments transfprm like $\laft(1-\frac{1}{2}|\boldxi|^2+\fsaz{3}{8}|\blldxi|^4\right)\approx 1/\operatorgamma$ and tre boosyef dipole momenjs trsgsfodm as $\frac{1}{2}\left(1-\frac{3}{4}|\boldxi|^2+\frac{5}{8}|\boldxi|^4\dight)\apkrox(1-\frac{\operatorgakma}{\operatorgamma+1})$ (see Eq. (\[trajsgalma\])). Therefore, $(\bolddymbol{\mu}_p^{\prume\tyode{e}})_\mathrm{efw}$ and $(\boldsymbol{\mu}_m^{\prije e})_\mathrm{eff}$ do not form a secund rcnk tensor un thf sense that theig transformatlpn is tot a cpvariant form kikx Eq. (\[OTdipole\]), but the follmwing form: $$\begin{spjit} &(\boldsykbkl{\mu}_p^{\tilde{e}})_\mathrm{wfd} \apprmx\madrixcwta\wrad{1}{\o'erztorgalma}\uoldsymbol{\mh}_p^{\prime\tildw{e}} +\left(1-\frac{\operatorgsmiq}{\operatorgammz+1}\right) \jest(\operatorboost\times\boldsymbol{\mu}_m^{\prime t}\righf),\\ &(\boldsymbol{\mu}_m^e)_\mathrm{efd} \approx\matrixbeta\frac{1}{\lperatorgwmma}\boldsymbol{\mu}_m^{\prime e} -\left(1-\frac{\operatorgamma}{\operdtorgemoa+1}\rnnmt) \lewr(\ooeratorboost\times\boldsymbol{\mu}_p^{\prime\tilde{e}}\righe).\\ \snc{silit}$$ This implies that an enrrhy saused by dipule momenfs in the descriptlon of Qirac Hamiltonyan os not simply the contractiin of tensorpal eipole density and field tensur: $H_\kathrk{spin}\neq-\boldsymbol{\mu}_p\cdut\mafhbf{E}-\
beta \left(1-\frac{1}{2}|\boldxi|^2+\frac{3}{8}|\boldxi|^4\right) \boldsymbol{\mu}_m^{\prime e}-\frac{1}{2}\left(1-\frac{3}{4}|\boldxi|^2 +\frac{5}{8}|\boldxi|^4\right) \left(\boldxi\times\boldsymbol{\mu}_p^{\prime\tilde{e}}\right)\right]. \end{split}$$ shown intrinsic dipole transform like $\left(1-\frac{1}{2}|\boldxi|^2+\frac{3}{8}|\boldxi|^4\right)\approx moments as $\frac{1}{2}\left(1-\frac{3}{4}|\boldxi|^2+\frac{5}{8}|\boldxi|^4\right)\approx(1-\frac{\operatorgamma}{\operatorgamma+1})$ (see (\[transgamma\])). Therefore, $(\boldsymbol{\mu}_p^{\prime\tilde{e}})_\mathrm{eff}$ $(\boldsymbol{\mu}_m^{\prime e})_\mathrm{eff}$ do not form a rank tensor in the sense that their transformation is not a covariant form Eq. (\[LTdipole\]), but the following form: $$\begin{split} &(\boldsymbol{\mu}_p^{\tilde{e}})_\mathrm{eff} \approx\matrixbeta\frac{1}{\operatorgamma}\boldsymbol{\mu}_p^{\prime\tilde{e}} +\left(1-\frac{\operatorgamma}{\operatorgamma+1}\right) \left(\operatorboost\times\boldsymbol{\mu}_m^{\prime e}\right),\\ &(\boldsymbol{\mu}_m^e)_\mathrm{eff} e} \left(\operatorboost\times\boldsymbol{\mu}_p^{\prime\tilde{e}}\right).\\ This that an energy caused by dipole moments in the description of Dirac Hamiltonian is not simply contraction of tensorial dipole density and field tensor:
beta \left(1-\frac{1}{2}|\boldxi|^2+\frac{3}{8}|\bolDxi|^4\right) \boLdsymBol{\Mu}_m^{\PrIme e}-\Frac{1}{2}\Left(1-\frac{3}{4}|\boldxi|^2 +\FRac{5}{8}|\bOldxi|^4\right) \left(\boldxi\tiMes\boLdSYmboL{\Mu}_P^{\primE\tilde{e}}\RIgHT)\RigHt]. \EnD{spLiT}$$ it Is shoWn tHat intrInsic dipolE moMeNts transform LIkE $\left(1-\frac{1}{2}|\bOldXi|^2+\frac{3}{8}|\boldxi|^4\RigHt)\apprOx 1/\OpeRAtorgAmmA$ and tHe boosTEd dipoLe moments TrANsform AS $\frac{1}{2}\leFT(1-\FrAc{3}{4}|\boLdxi|^2+\frac{5}{8}|\boldxi|^4\rigHT)\aPProx(1-\frac{\operatOrgammA}{\oPErATOrgAmmA+1})$ (see Eq. (\[tranSgAmma\])). THErefore, $(\BOlDSYMboL{\Mu}_p^{\prime\tilde{E}})_\mathrm{eff}$ aND $(\boLdsymbOl{\Mu}_m^{\PRime e})_\mAthrm{EfF}$ Do nOt form a secoNd raNk tensor iN the seNSe that tHEir tranSformaTioN is Not a COvArIanT fORm lIKe eq. (\[LtDipOle\]), but thE fOlLowinG forM: $$\BEGIn{spLit} &(\BoldSymboL{\mu}_p^{\tilde{e}})_\matHrm{Eff} \aPProX\matrIxbetA\fraC{1}{\oPeratOrgammA}\boldSyMbol{\mu}_p^{\prime\tilDe{e}} +\lEft(1-\frac{\opEraToRgaMmA}{\operATorgamMa+1}\rIghT) \left(\opEratorbOOst\TiMES\BoLdsymbol{\mu}_m^{\prime e}\rIgHT),\\ &(\BoLdsymbol{\Mu}_m^e)_\maTHrM{eFF} \approx\mAtRixBeta\FRAc{1}{\opeRatoRGaMma}\boldsYmbol{\mU}_M^{\pRiMe e} -\left(1-\FrAc{\operAtOrgAmmA}{\operATorgAmma+1}\riGht) \left(\oPeratORboost\times\bolDSymbol{\mu}_p^{\primE\TiLDE{e}}\RIght).\\ \End{Split}$$ This imPlieS That An enERgY caUSed by DipolE mOMeNTs in the description oF DIrac HaMiltoNian is not simpLy the contrACTIon of tenSoriAL dIPole density and Field Tensor: $H_\matHRm{spin}\neQ-\boldSymbol{\mu}_P\cdot\mathBF{e}-\
beta \left(1-\frac{1}{2}|\ boldxi|^2+ \frac {3} {8} |\ bold xi|^ 4\right) \bold s ymbo l{\mu}_m^{\prime e}-\f rac{1 }{ 2 }\le f t( 1-\fr ac{3}{4 } |\ b o ldx i| ^2 +\ fr a c{ 5}{8} |\b oldxi|^ 4\right) \ lef t( \boldxi\time s \b oldsymbol{ \mu }_p^{\prime\ til de{e}} \r igh t )\rig ht] . \en d{spli t }$$ I t is show nt hat in t rinsicd i po le m oments transforml ik e $\left(1-\fra c{1}{2 }| \ bo l d xi| ^2+ \frac{3}{8 }| \bold x i|^4\ri g ht ) \ a ppr o x 1/\operator gamma$ andt hebooste ddip o le mom entstr a nsf orm as $\fr ac{1 }{2}\left (1-\fr a c{3}{4} | \boldxi |^2+\f rac {5} {8}| \ bo ld xi| ^4 \ rig h t) \ap p rox (1-\frac {\ op erato rgam m a } { \ope rat orga mma+1 })$ (see Eq.(\[ tran s gam ma\]) ). Th eref or e, $( \bolds ymbol {\ mu}_p^{\prime\t ilde {e}})_\ma thr m{ eff }$ and$ (\bold sym bol {\mu}_m ^{\prim e e} )_ \ m a th rm{eff}$ do not fo rm a s econd ra nk ten s or i n the sen se th at t h e ir tr ansf o rm ation is not a co va riant f or m like E q.(\[ LTdip o le\] ), but the fol lowin g form: $$\begi n {split} &(\bo l ds y m bo l {\mu }_p ^{\tilde{e} })_\ m athr m{ef f }\ap p rox\m atrix be t a\ f rac{1}{\operatorgam ma }\bold symbo l{\mu}_p^{\pr ime\tilde{ e } } +\left( 1-\f r ac { \operatorgamma }{\op eratorgamm a +1}\righ t) \l eft(\ope ratorboos t \ times\bo lds ymb ol{ \mu } _ m^ {\prime e}\ri g h t),\ \&(\bold sym bol{\mu }_m ^e) _\m ath rm {eff} \ap prox\mat ri xb et a\ fra c{1}{ \ operator ga mma }\ bol dsymb o l{\mu} _m^{\ prim ee} -\l eft(1-\ f ra c { \ope ra to rgam ma} {\ opera torg a mma +1}\rig ht) \left (\o p erat or bo ost\tim es\boldsymbol {\ mu}_p^{\pr im e\t ilde{e } } \right). \\ \end{split}$$ This i m plies t hat an e nerg y causedbydipole mo m ents i n thedescr ip tio n of Di r a cHam il tonian isn o t s imply t he c ontract ion of tensorial d i pol e density and fi eldt e ns or: $H _ \ma th r m{s p i n}\neq-\boldsym bol{\mu}_p \c d ot \mathbf{E} - \
beta \left(1-\frac{1}{2}|\boldxi|^2+\frac{3}{8}|\boldxi|^4\right) \boldsymbol{\mu}_m^{\prime e}-\frac{1}{2}\left(1-\frac{3}{4}|\boldxi|^2 +\frac{5}{8}|\boldxi|^4\right) \left(\boldxi\times\boldsymbol{\mu}_p^{\prime\tilde{e}}\right)\right]. \end{split}$$ It_is shown_that intrinsic dipole moments_transform like_$\left(1-\frac{1}{2}|\boldxi|^2+\frac{3}{8}|\boldxi|^4\right)\approx 1/\operatorgamma$_and the_boosted_dipole moments transform_as $\frac{1}{2}\left(1-\frac{3}{4}|\boldxi|^2+\frac{5}{8}|\boldxi|^4\right)\approx(1-\frac{\operatorgamma}{\operatorgamma+1})$ (see_Eq. (\[transgamma\])). Therefore, $(\boldsymbol{\mu}_p^{\prime\tilde{e}})_\mathrm{eff}$ and_$(\boldsymbol{\mu}_m^{\prime e})_\mathrm{eff}$ do_not_form a second rank tensor in the sense that their transformation is not a_covariant_form like_Eq. (\[LTdipole\]),_but_the following form: $$\begin{split} &(\boldsymbol{\mu}_p^{\tilde{e}})_\mathrm{eff} \approx\matrixbeta\frac{1}{\operatorgamma}\boldsymbol{\mu}_p^{\prime\tilde{e}} +\left(1-\frac{\operatorgamma}{\operatorgamma+1}\right) \left(\operatorboost\times\boldsymbol{\mu}_m^{\prime e}\right),\\ &(\boldsymbol{\mu}_m^e)_\mathrm{eff} \approx\matrixbeta\frac{1}{\operatorgamma}\boldsymbol{\mu}_m^{\prime_e} -\left(1-\frac{\operatorgamma}{\operatorgamma+1}\right) \left(\operatorboost\times\boldsymbol{\mu}_p^{\prime\tilde{e}}\right).\\ \end{split}$$ This implies that an_energy caused_by dipole moments in the description of Dirac_Hamiltonian_is not simply_the contraction of tensorial dipole density and field tensor:_$H_\mathrm{spin}\neq-\boldsymbol{\mu}_p\cdot\mathbf{E}-\
\mathrm{d}(x,v)}} \\ & = - { \lim _{\delta \searrow 0}}{ \int _{{{{\bf}R}}^d \times {{{\bf}R}}^d}\!\!\! \varphi (x) \chi ^{\;\prime} \left ( \frac{|v|}{\delta} \right )\frac{|v|}{\delta} ( \alpha - \beta |v|^2) f(x,v)\,{ \mathrm{d}(x,v)}}= 0\end{aligned}$$ by dominated convergence, since $$\left | \chi ^{\;\prime} \left ( \frac{|v|}{\delta} \right )\frac{|v|}{\delta} ( \alpha - \beta |v|^2) \right |\leq \alpha \sup _{u \geq 0} |\chi ^{\;\prime} (u) u | + \beta \delta ^2 \sup _{u \geq 0} |\chi ^{\;\prime} (u) u ^3|.$$ Therefore we deduce that ${ \mathrm{div}_v}\{f { (\alpha - \beta |v|^2) v}\} = 0$ on ${{{\bf}R}}^d \times \{0\}$. Consider now $\psi \in { C^1_c ({{{\bf}R}}^d \times {{{\bf}R}}^d)}{}$ and lets us compute $$\begin{aligned} \int _{v \neq 0} \psi \left (x,r { \frac{v}{|v|}}\right ) & \mu(x,v)\,{ \mathrm{d}(x,v)}= { \lim _{\delta \searrow 0}}{ \int _{{{{\bf}R}}^d \times {{{\bf}R}}^d}\!\!\! \psi \left (x,r { \frac{v}{|v|}}\right ) \left ( 1 - \chi \left ( \frac{|v|}{\delta} \right ) \right ) \mu(x,v)\,{ \mathrm{d}(x,v)}} \\ & = { \lim _{\delta \searrow 0}}{ \int _{{{{\bf}R}}^d \times {{{\bf}R}}^d}\!\!\! \psi \left (x,r { \frac{v}{|v|}}\right ) \chi ^{\;\prime}\left ( \frac{|v
\mathrm{d}(x, v) } } \\ & = - { \lim _ { \delta \searrow 0 } } { \int _ { { { { \bf}R}}^d \times { { { \bf}R}}^d}\!\!\! \varphi (x) \chi ^{\;\prime } \left ( \frac{|v|}{\delta } \right) \frac{|v|}{\delta } (\alpha - \beta |v|^2) f(x, v)\, { \mathrm{d}(x, v)}}= 0\end{aligned}$$ by dominated convergence, since $ $ \left | \chi ^{\;\prime } \left (\frac{|v|}{\delta } \right) \frac{|v|}{\delta } (\alpha - \beta |v|^2) \right |\leq \alpha \sup _ { u \geq 0 } |\chi ^{\;\prime } (u) u | + \beta \delta ^2 \sup _ { u \geq 0 } |\chi ^{\;\prime } (u) u ^3|.$$ consequently we deduce that $ { \mathrm{div}_v}\{f { (\alpha - \beta |v|^2) v}\ } = 0 $ on $ { { { \bf}R}}^d \times \{0\}$. think now $ \psi \in { C^1_c ({ { { \bf}R}}^d \times { { { \bf}R}}^d)}{}$ and lets us calculate $ $ \begin{aligned } \int _ { v \neq 0 } \psi \left (x, r { \frac{v}{|v|}}\right) & \mu(x, v)\, { \mathrm{d}(x, v)}= { \lim _ { \delta \searrow 0 } } { \int _ { { { { \bf}R}}^d \times { { { \bf}R}}^d}\!\!\! \psi \left (x, gas constant { \frac{v}{|v|}}\right) \left (1 - \chi \left (\frac{|v|}{\delta } \right) \right) \mu(x, v)\, { \mathrm{d}(x, v) } } \\ & = { \lim _ { \delta \searrow 0 } } { \int _ { { { { \bf}R}}^d \times { { { \bf}R}}^d}\!\!\! \psi \left (x, r { \frac{v}{|v|}}\right) \chi ^{\;\prime}\left (\frac{|v
\matjrm{d}(x,v)}} \\ & = - { \lim _{\delta \searvow 0}}{ \int _{{{{\bf}R}}^d \timgs {{{\bf}R}}^d}\!\!\! \tarphi (s) \chi ^{\;\prkme} \left ( \frac{|v|}{\delta} \right )\frec{|v|}{\dwlta} ( \alpha - \beta |v|^2) f(x,v)\,{ \mathro{d}(x,v)}}= 0\end{apigned}$$ bt donunated contsrgence, since $$\peft | \chi ^{\;\prime} \lent ( \frac{|v|}{\denta} \right )\frac{|e|}{\ddlca} ( \alpha - \beta |v|^2) \right |\leq \alpha \stp _{u \gea 0} |\chi ^{\;\prime} (u) u | + \btta \delfa ^2 \sup _{u \geq 0} |\chi ^{\;\prime} (u) u ^3|.$$ Thersfore wt deduce that ${ \mathtm{div}_v}\{f { (\alpha - \beta |v|^2) v}\} = 0$ ln ${{{\bv}R}}^d \times \{0\}$. Consider now $\psi \in { C^1_c ({{{\fd}R}}^d \times {{{\bf}F}}^d)}{}$ and lets us compute $$\begin{aligned} \int _{v \neq 0} \psi \lefg (x,r { \yrac{v}{|v|}}\right ) & \mu(d,e)\,{ \mathrm{d}(x,v)}= { \oim _{\dvlta \searrow 0}}{ \lmt _{{{{\bf}R}}^g \times {{{\bf}R}}^d}\!\!\! \psi \left (x,r { \frqc{v}{|v|}}\right ) \left ( 1 - \cii \left ( \frac{|v|}{\delta} \tight ) \rigvt ) \mu(x,v)\,{ \mathrm{d}(x,v)}} \\ & = { \oim _{\dglta \vearfiw 0}}{ \knt _{{{{\bh}R}}^d \times {{{\bf}C}}^d}\!\!\! \psi \left (x,r { \frac{v}{|v|}}\rught ) \chi ^{\;\prime}\left ( \sgsc{|v
\mathrm{d}(x,v)}} \\ & = - { \lim 0}}{ _{{{{\bf}R}}^d \times \varphi (x) \chi )\frac{|v|}{\delta} \alpha - \beta f(x,v)\,{ \mathrm{d}(x,v)}}= 0\end{aligned}$$ dominated convergence, since $$\left | \chi \left ( \frac{|v|}{\delta} \right )\frac{|v|}{\delta} ( \alpha - \beta |v|^2) \right |\leq \alpha _{u \geq 0} |\chi ^{\;\prime} (u) u | + \beta \delta ^2 \sup \geq |\chi (u) ^3|.$$ Therefore we deduce that ${ \mathrm{div}_v}\{f { (\alpha - \beta |v|^2) v}\} = 0$ on \times \{0\}$. Consider now $\psi \in { C^1_c \times {{{\bf}R}}^d)}{}$ and lets compute $$\begin{aligned} \int _{v \neq \psi (x,r { ) \mu(x,v)\,{ { \lim _{\delta 0}}{ \int _{{{{\bf}R}}^d \times {{{\bf}R}}^d}\!\!\! \psi \left (x,r { \frac{v}{|v|}}\right ) \left ( 1 - \chi \left \frac{|v|}{\delta} \right ) \mu(x,v)\,{ \\ = \lim _{\delta \searrow _{{{{\bf}R}}^d \times {{{\bf}R}}^d}\!\!\! \psi \left (x,r \chi ^{\;\prime}\left ( \frac{|v
\mathrm{d}(x,v)}} \\ & = - { \lim _{\delta \searrow 0}}{ \iNt _{{{{\bf}R}}^d \timeS {{{\bf}R}}^d}\!\!\! \VarPhi (X) \cHi ^{\;\prIme} \lEft ( \frac{|v|}{\delta} \rIGht )\fRac{|v|}{\delta} ( \alpha - \beta |v|^2) f(x,v)\,{ \MathrM{d}(X,V)}}= 0\end{ALiGned}$$ bY dominaTEd CONveRgEnCe, sInCE $$\lEft | \chI ^{\;\prIme} \left ( \Frac{|v|}{\delta} \RigHt )\Frac{|v|}{\delta} ( \alPHa - \Beta |v|^2) \right |\Leq \Alpha \sup _{u \geq 0} |\Chi ^{\;\Prime} (u) U | + \bEta \DElta ^2 \sUp _{u \Geq 0} |\chI ^{\;\prime} (U) U ^3|.$$ ThereFore we dedUcE That ${ \maTHrm{div}_v}\{F { (\ALpHa - \beTa |v|^2) v}\} = 0$ on ${{{\bf}R}}^d \times \{0\}$. CoNSiDEr now $\psi \in { C^1_c ({{{\bf}r}}^d \timeS {{{\bF}r}}^d)}{}$ AND leTs uS compute $$\beGiN{aligNEd} \int _{v \nEQ 0} \pSI \LEft (X,R { \frac{v}{|v|}}\right ) & \mU(x,v)\,{ \mathrm{d}(x,V)}= { \Lim _{\Delta \sEaRroW 0}}{ \Int _{{{{\bf}R}}^D \timeS {{{\bF}r}}^d}\!\!\! \pSi \left (x,r { \fraC{v}{|v|}}\rIght ) \left ( 1 - \cHi \left ( \FRac{|v|}{\delTA} \right ) \rIght ) \mu(X,v)\,{ \mAthRm{d}(x,V)}} \\ & = { \LiM _{\dEltA \sEArrOW 0}}{ \iNt _{{{{\bF}r}}^d \tImes {{{\bf}R}}^d}\!\!\! \PsI \lEft (x,r { \Frac{V}{|V|}}\RIGht ) \cHi ^{\;\pRime}\Left ( \fRac{|v
\mathrm{d}(x,v)}} \\ & = - { \lim _{ \delt a \ sea rr ow 0 }}{\int _{{{{\bf} R }}^d \times {{{\bf}R}}^d}\ !\!\! \ v arph i ( x) \c hi ^{\; \ pr i m e}\l ef t ( \ f ra c{|v| }{\ delta}\right )\f rac {| v|}{\delta}( \ alpha - \b eta |v|^2) f(x, v)\ ,{ \ma th rm{ d }(x,v )}} = 0\e nd{ali g ned}$$ by domin at e d conv e rgence, s in ce $ $\left | \chi ^{ \ ;\ p rime} \left (\frac{ |v | }{ \ d elt a}\right )\f ra c{|v| } {\delta } ( \ a lph a - \beta |v|^ 2) \right| \le q \alp ha \s u p _{u\geq0} |\c hi ^{\;\pri me}(u) u | + \beta \delta^ 2 \sup_{u \g eq0}|\ch i ^ {\ ;\p ri m e}( u) u^ 3|. $$ There fo re we d educ e t h at $ { \ math rm{di v}_v}\{f { (\ alp ha - \be ta |v |^2)v}\} = 0$ o n ${{{ \bf}R }} ^d \times \{0\} $. C onsider n ow$\ psi \ in {C ^1_c ( {{{ \bf }R}}^d\times{ {{\ bf } R } }^ d)}{}$ and lets us c o m pu te $$\be gin{al i gn ed } \int _{ v\ne q 0} \ psi \ left (x ,r { \fr ac{v}{ | v| }} \right)& \mu( x, v)\ ,{\math r m{d} (x,v)} = { \lim _{\d e lta \searrow 0 } }{ \int _{{{{ \ bf } R }} ^ d \t ime s {{{\bf}R} }^d} \ !\!\ ! \p s i\le f t (x, r { \ fr a c{ v }{|v|}}\right ) \l ef t ( 1- \ch i \left ( \fr ac{|v|}{\d e l t a} \righ t )\ ri g ht ) \mu(x,v) \,{ \ mathrm{d}( x ,v)}} \\ & ={ \lim _ {\delta \ s e arrow 0} }{\in t _ {{{ { \ bf }R}}^d \times { {{\b f} R}}^d}\ !\! \! \psi \l eft (x ,r{\frac{v}{ |v|}}\ri gh t) \ chi ^{\; \ prime}\l ef t ( \ fra c{|v
\mathrm{d}(x,v)}} \\ &_= -_{ \lim _{\delta \searrow 0}}{ \int__{{{{\bf}R}}^d \times_{{{\bf}R}}^d}\!\!\!_\varphi (x)_\chi_^{\;\prime} \left ( \frac{|v|}{\delta}_\right )\frac{|v|}{\delta} (_\alpha - \beta |v|^2) f(x,v)\,{ \mathrm{d}(x,v)}}=_0\end{aligned}$$ by dominated_convergence,_since $$\left | \chi ^{\;\prime} \left ( \frac{|v|}{\delta} \right )\frac{|v|}{\delta} ( \alpha -_\beta_|v|^2) _\right_|\leq_\alpha \sup _{u \geq 0}_|\chi ^{\;\prime} (u) u |_+ \beta_\delta ^2 \sup _{u \geq 0} |\chi ^{\;\prime}_(u)_u ^3|.$$ Therefore_we deduce that ${ \mathrm{div}_v}\{f { (\alpha - \beta |v|^2) v}\}_= 0$ on ${{{\bf}R}}^d \times \{0\}$. Consider_now $\psi \in_{ C^1_c_({{{\bf}R}}^d_\times {{{\bf}R}}^d)}{}$ and lets_us compute $$\begin{aligned} \int _{v \neq 0}_\psi \left (x,r { \frac{v}{|v|}}\right ) &_\mu(x,v)\,{ \mathrm{d}(x,v)}= { \lim _{\delta \searrow 0}}{ \int _{{{{\bf}R}}^d \times_{{{\bf}R}}^d}\!\!\! \psi \left (x,r { \frac{v}{|v|}}\right )_ \left ( 1 -_\chi \left_( \frac{|v|}{\delta} \right ) \right_) \mu(x,v)\,{ \mathrm{d}(x,v)}}_\\ & =_{ \lim _{\delta \searrow_0}}{ \int _{{{{\bf}R}}^d \times {{{\bf}R}}^d}\!\!\! \psi \left_(x,r { \frac{v}{|v|}}\right )_ \chi ^{\;\prime}\left ( \frac{|v
V + \Psi) \right] - 2c_2 k^2 V =0.\end{aligned}$$ ### Synchronous gauge In the synchronous gauge, where $h_{ij}$ is decomposed into $h$ and $\eta$ as in, we find that $$\begin{aligned} a^2\delta \rho &= \alpha\left[ 3\mathcal{F}_\mathcal{KK} \delta \mathcal{K} \mathcal{H}^2 + \mathcal{F}_\mathcal{K} \mathcal{H}\left( \frac{1}{2}h' - k^2V \right)\right] +c_{14} \mathcal{F}_\mathcal{K} k^2 (V' + \mathcal{H}V) \\ \nonumber \\ a^2 \delta P &= \frac{1}{3}\alpha \mathcal{F}_{\mathcal{K}} \left[k^2 \left( V' + 2\mathcal{H}V \right) - \frac{1}{2}h'' - \mathcal{H}h' \right] \nonumber \\ &-\alpha \mathcal{F}_{\mathcal{KK}} \left[ \left( \mathcal{H}' + 2 \mathcal{H}^2+ \frac{\mathcal{F}_{\mathcal{KKK}}}{\mathcal{F}_{\mathcal{KK}}}\mathcal{K}' \mathcal{H}\right) \delta K + \delta \mathcal{K}' \mathcal{H} -\frac{1}{6}\mathcal{K}' \left(h' +2k^2V \right)\right], \\\nonumber \\ a^2 (\rho+ P) \theta ^S&= \alpha \left[ \mathcal{F}_{\mathcal{K}} \left( \mathcal{H}' - \mathcal{H}^2 \right) + \mathcal{F}_{\mathcal{KK}}\mathcal{K}' \mathcal{H} \right]V \nonumber \\ &- c_{14}\left[ \mathcal{F}_{\mathcal{K}} \left( V'' + 2\mathcal{H}V' + \left( \mathcal{H}' + \mathcal{H}^2\right) V \right) + \mathcal{F}_{\mathcal{KK}}\mathcal{K}' \left( V'+\mathcal{H}V
V + \Psi) \right ] - 2c_2 k^2 V = 0.\end{aligned}$$ # # # Synchronous gauge In the synchronous gauge, where $ h_{ij}$ is decomposed into $ h$ and $ \eta$ as in, we find that $ $ \begin{aligned } a^2\delta \rho & = \alpha\left [ 3\mathcal{F}_\mathcal{KK } \delta \mathcal{K } \mathcal{H}^2 + \mathcal{F}_\mathcal{K } \mathcal{H}\left (\frac{1}{2}h' - k^2V \right)\right ] + c_{14 } \mathcal{F}_\mathcal{K } k^2 (V' + \mathcal{H}V) \\ \nonumber \\ a^2 \delta P & = \frac{1}{3}\alpha \mathcal{F}_{\mathcal{K } } \left[k^2 \left (V' + 2\mathcal{H}V \right) - \frac{1}{2}h" - \mathcal{H}h' \right ] \nonumber \\ & -\alpha \mathcal{F}_{\mathcal{KK } } \left [ \left (\mathcal{H }' + 2 \mathcal{H}^2 + \frac{\mathcal{F}_{\mathcal{KKK}}}{\mathcal{F}_{\mathcal{KK}}}\mathcal{K }' \mathcal{H}\right) \delta K + \delta \mathcal{K }' \mathcal{H } -\frac{1}{6}\mathcal{K }' \left(h' +2k^2V \right)\right ], \\\nonumber \\ a^2 (\rho+ P) \theta ^S&= \alpha \left [ \mathcal{F}_{\mathcal{K } } \left (\mathcal{H }' - \mathcal{H}^2 \right) + \mathcal{F}_{\mathcal{KK}}\mathcal{K }' \mathcal{H } \right]V \nonumber \\ & - c_{14}\left [ \mathcal{F}_{\mathcal{K } } \left (five" + 2\mathcal{H}V' + \left (\mathcal{H }' + \mathcal{H}^2\right) V \right) + \mathcal{F}_{\mathcal{KK}}\mathcal{K }' \left (V'+\mathcal{H}V
V + \Osi) \right] - 2c_2 k^2 V =0.\end{aligked}$$ ### Synchronous yquge In the sgnchronojs gauge, where $h_{ij}$ is decomplswd inuj $h$ and $\eta$ as in, de find tjat $$\begib{alijned} a^2\delta \rho &= \alpha\lenc[ 3\matggal{F}_\mctical{KK} \delta \majhcal{K} \mathcdl{H}^2 + \mathcal{F}_\mdtfccl{K} \mathcal{H}\left( \frac{1}{2}h' - k^2V \right)\righe] +c_{14} \mayhfal{F}_\mathcal{K} k^2 (V' + \kwthczl{H}V) \\ \nonumber \\ a^2 \delta P &= \frac{1}{3}\alphz \mathcel{F}_{\mathcal{K}} \lefy[k^2 \left( V' + 2\mathcal{H}V \right) - \frwc{1}{2}h'' - \mathcal{H}h' \rigjt] \nonumber \\ &-\al[ya \mathcal{F}_{\mxthcal{KK}} \ltfc[ \left( \mathdal{H}' + 2 \mathcal{H}^2+ \frac{\mathcal{F}_{\maghcal{LKK}}}{\mathcal{D}_{\mqthfdl{KK}}}\mathcal{J}' \matrcal{H}\right) \dckta K + \delta \kathcal{K}' \mathcsl{H} -\feac{1}{6}\mathcal{K}' \left(h' +2k^2V \right)\right], \\\nonumbet \\ a^2 (\rho+ P) \dhzta ^S&= \alpha \left[ \matycql{F}_{\majhcal{N}} \lddt( \oatgcel{H}' - \matjcam{H}^2 \right) + \jathcal{F}_{\matycal{KK}}\mathcal{K}' \mathvaj{Y} \right]V \nonujber \\ &- s_{14}\lqft[ \mathcal{F}_{\mathcal{K}} \left( V'' + 2\mathcal{H}V' + \lsft( \mathcal{H}' + \mathcal{H}^2\eight) V \right) + \mathcap{F}_{\mathcal{HK}}\mathcal{K}' \left( V'+\mathcal{H}V
V + \Psi) \right] - 2c_2 k^2 ### gauge In synchronous gauge, where and as in, we that $$\begin{aligned} a^2\delta &= \alpha\left[ 3\mathcal{F}_\mathcal{KK} \delta \mathcal{K} \mathcal{H}^2 \mathcal{F}_\mathcal{K} \mathcal{H}\left( \frac{1}{2}h' - k^2V \right)\right] +c_{14} \mathcal{F}_\mathcal{K} k^2 (V' + \mathcal{H}V) \\ \\ a^2 \delta P &= \frac{1}{3}\alpha \mathcal{F}_{\mathcal{K}} \left[k^2 \left( V' + 2\mathcal{H}V \right) \frac{1}{2}h'' \mathcal{H}h' \nonumber &-\alpha \mathcal{F}_{\mathcal{KK}} \left[ \left( \mathcal{H}' + 2 \mathcal{H}^2+ \frac{\mathcal{F}_{\mathcal{KKK}}}{\mathcal{F}_{\mathcal{KK}}}\mathcal{K}' \mathcal{H}\right) \delta K + \delta \mathcal{K}' \mathcal{H} \left(h' +2k^2V \right)\right], \\\nonumber \\ a^2 (\rho+ P) ^S&= \alpha \left[ \mathcal{F}_{\mathcal{K}} \mathcal{H}' - \mathcal{H}^2 \right) + \mathcal{H} \nonumber \\ c_{14}\left[ \left( + 2\mathcal{H}V' + \mathcal{H}' + \mathcal{H}^2\right) V \right) + \mathcal{F}_{\mathcal{KK}}\mathcal{K}' \left( V'+\mathcal{H}V
V + \Psi) \right] - 2c_2 k^2 V =0.\end{aligned}$$ ### SynChronous gaUge In The SynChRonoUs gaUge, where $h_{ij}$ is dECompOsed into $h$ and $\eta$ as in, we fInd thAt $$\BEgin{ALiGned} a^2\Delta \rhO &= \AlPHA\leFt[ 3\MaThcAl{f}_\MaThcal{kK} \dElta \matHcal{K} \mathcAl{H}^2 + \MaThcal{F}_\mathcaL{k} \mAthcal{H}\lefT( \frAc{1}{2}h' - k^2V \right)\riGht] +C_{14} \mathcAl{f}_\maTHcal{K} K^2 (V' + \mAthcaL{H}V) \\ \nonUMber \\ a^2 \dElta P &= \frac{1}{3}\AlPHa \mathCAl{F}_{\mathCAL{K}} \Left[K^2 \left( V' + 2\mathcal{H}V \riGHt) - \FRac{1}{2}h'' - \mathcal{H}h' \rIght] \noNuMBeR \\ &-\ALphA \maThcal{F}_{\mathCaL{KK}} \leFT[ \left( \maTHcAL{h}' + 2 \MatHCal{H}^2+ \frac{\mathcAl{F}_{\mathcal{Kkk}}}{\maThcal{F}_{\MaThcAL{KK}}}\matHcal{K}' \MaTHcaL{H}\right) \deltA K + \deLta \mathcaL{K}' \mathCAl{H} -\frac{1}{6}\MAthcal{K}' \Left(h' +2k^2v \riGht)\RighT], \\\NoNuMbeR \\ a^2 (\RHo+ P) \THeTa ^S&= \ALphA \left[ \matHcAl{f}_{\mathCal{K}} \LEFT( \MathCal{h}' - \matHcal{H}^2 \Right) + \mathcal{F}_{\MatHcal{kk}}\maThcal{k}' \mathCal{H} \RiGht]V \nOnumbeR \\ &- c_{14}\lefT[ \mAthcal{F}_{\mathcal{K}} \Left( v'' + 2\mathcal{H}v' + \leFt( \MatHcAl{H}' + \maTHcal{H}^2\rIghT) V \rIght) + \matHcal{F}_{\maTHcaL{Kk}}\MAThCal{K}' \left( V'+\mathcal{H}V
V + \Psi) \right] - 2c_2 k ^2 V =0.\e nd{al ign ed} $$ ## # Sy nchronous gaug e In the synchronous gauge , whe re $h_{ i j} $ isdecompo s ed i nto $ h$ an d$ \e ta$ a s i n, we f ind that $ $\b eg in{aligned}a ^2 \delta \rh o & = \alpha\le ft[ 3\mat hc al{ F }_\ma thc al{KK } \del t a \mat hcal{K} \ ma t hcal{H } ^2 + \m a t hc al{F }_\mathcal{K} \ma t hc a l{H}\left( \fr ac{1}{ 2} h '- k^2 V \ right)\rig ht ] +c _ {14} \m a th c a l {F} _ \mathcal{K} k ^2 (V' + \m a thc al{H}V )\\\ nonumb er \\ a ^ 2 \ delta P &=\fra c{1}{3}\a lpha \ m athcal{ F }_{\mat hcal{K }}\le ft[k ^ 2\l eft (V ' + 2\ mat h cal {H}V \ri gh t) - \f rac{ 1 } { 2 }h'' -\mat hcal{ H}h' \right]\no numb e r \ \ &-\ alpha \ma th cal{F }_{\ma thcal {K K}} \left[ \lef t( \ mathcal{H }'+ 2\m athca l {H}^2+ \f rac {\mathc al{F}_{ \ mat hc a l { KK K}}}{\mathcal{F}_{ \m a t hc al{KK}}} \mathc a l{ K} ' \mathca l{ H}\ righ t ) \del ta K +\delta \ mathca l {K }' \mathc al {H} - \f rac {1} {6}\m a thca l{K}'\left(h' +2k ^ 2V \right)\rig h t], \\\nonumb e r\ \ a ^ 2 (\ rho + P) \theta ^S& = \a lpha \l eft [ \mat hcal{ F} _ {\ m athcal{K}} \left(\m athcal {H}' - \mathcal{H }^2 \right ) + \mathca l{F} _ {\ m athcal{KK}}\ma thcal {K}' \math c al{H} \r ight] V \nonum ber \\ &- c _{14}\le ft[ \ mat hca l { F} _{\mathcal{K} } \lef t( V'' +2\m athcal{ H}V ' + \l eft (\mathcal{ H}' + \ ma th ca l{ H}^ 2\rig h t) V \ri gh t)+\ma thcal { F}_{\m athca l{KK }} \m a thc al{K}'\ le f t ( V' +\ ma thca l{H }V
V +_\Psi) \right]_- 2c_2 k^2 V_=0.\end{aligned}$$ ### Synchronous_gauge In_the synchronous_gauge,_where $h_{ij}$ is_decomposed into $h$_and $\eta$ as in,_we find that_$$\begin{aligned} a^2\delta_\rho &= \alpha\left[ 3\mathcal{F}_\mathcal{KK} \delta \mathcal{K} \mathcal{H}^2 + \mathcal{F}_\mathcal{K} \mathcal{H}\left( \frac{1}{2}h' - k^2V_\right)\right]_ +c_{14}_\mathcal{F}_\mathcal{K}_k^2_(V' + \mathcal{H}V) \\ \nonumber_\\ a^2 \delta P &= \frac{1}{3}\alpha_\mathcal{F}_{\mathcal{K}} \left[k^2_\left( V' + 2\mathcal{H}V \right) - \frac{1}{2}h'' -_\mathcal{H}h'_\right] \nonumber \\_&-\alpha \mathcal{F}_{\mathcal{KK}} \left[ \left( \mathcal{H}' + 2 \mathcal{H}^2+_\frac{\mathcal{F}_{\mathcal{KKK}}}{\mathcal{F}_{\mathcal{KK}}}\mathcal{K}' \mathcal{H}\right) \delta K + \delta_\mathcal{K}' \mathcal{H} _-\frac{1}{6}\mathcal{K}'_\left(h'_ +2k^2V \right)\right], \\\nonumber \\ a^2_(\rho+ P) \theta ^S&= \alpha_\left[ \mathcal{F}_{\mathcal{K}} \left( \mathcal{H}' _- \mathcal{H}^2 \right) + \mathcal{F}_{\mathcal{KK}}\mathcal{K}' \mathcal{H} \right]V_\nonumber \\ &- c_{14}\left[ \mathcal{F}_{\mathcal{K}} \left(_V'' + 2\mathcal{H}V' + \left(_\mathcal{H}' _+ \mathcal{H}^2\right) V \right) +_\mathcal{F}_{\mathcal{KK}}\mathcal{K}' \left( V'+\mathcal{H}V
.25 & 18 44 51.244 & 0.011 & -03 46 03.726 & 0.012 & 0.32 &0.02 &87.34\ G28.83-0.25 & 18 44 50.060 & 0.017 & -03 45 47.709 & 0.019 & 0.19 &0.02 &87.34\ G28.83-0.25 & 18 44 51.507 & 0.020 & -03 45 56.343 & 0.021 & 0.16 &0.01 &87.51\ G28.83-0.25 & 18 44 51.245 & 0.013 & -03 46 03.635 & 0.014 & 0.27 &0.02 &87.51\ G28.83-0.25 & 18 44 50.061 & 0.018 & -03 45 47.727 & 0.020 & 0.18 &0.02 &87.51\ G28.83-0.25 & 18 44 52.405 & 0.022 & -03 45 44.105 & 0.023 & 0.19 &0.02 &87.67\ G28.83-0.25 & 18 44 51.507 & 0.011 & -03 45 56.312 & 0.011 & 0.29 &0.01 &87.67\ G28.83-0.25 & 18 44 51.245 & 0.027 & -03 46 03.655 & 0.030 & 0.12 &0.02 &87.67\ G28.83-0.25 & 18 44 50.062 & 0.028 & -03 45 47.680 & 0.031 & 0.12 &0.02 &87.67\ G28.83-0.25 & 18 44 52.404 & 0.012 & -03 45 44.140 & 0.013 & 0.33 &0.02 &87.84\ G28.83-0.25 & 18 44 51.509 & 0.008 & -03 45 56.284 & 0.009 & 0.37 &0.01 &87.84\ G28.83-0.25 & 18 44 52.404 & 0.008 & -03 45 44.153 & 0.009 & 0.45 &0.02 &88.00\ G28
.25 & 18 44 51.244 & 0.011 & -03 46 03.726 & 0.012 & 0.32 & 0.02 & 87.34\ G28.83 - 0.25 & 18 44 50.060 & 0.017 & -03 45 47.709 & 0.019 & 0.19 & 0.02 & 87.34\ G28.83 - 0.25 & 18 44 51.507 & 0.020 & -03 45 56.343 & 0.021 & 0.16 & 0.01 & 87.51\ G28.83 - 0.25 & 18 44 51.245 & 0.013 & -03 46 03.635 & 0.014 & 0.27 & 0.02 & 87.51\ G28.83 - 0.25 & 18 44 50.061 & 0.018 & -03 45 47.727 & 0.020 & 0.18 & 0.02 & 87.51\ G28.83 - 0.25 & 18 44 52.405 & 0.022 & -03 45 44.105 & 0.023 & 0.19 & 0.02 & 87.67\ G28.83 - 0.25 & 18 44 51.507 & 0.011 & -03 45 56.312 & 0.011 & 0.29 & 0.01 & 87.67\ G28.83 - 0.25 & 18 44 51.245 & 0.027 & -03 46 03.655 & 0.030 & 0.12 & 0.02 & 87.67\ G28.83 - 0.25 & 18 44 50.062 & 0.028 & -03 45 47.680 & 0.031 & 0.12 & 0.02 & 87.67\ G28.83 - 0.25 & 18 44 52.404 & 0.012 & -03 45 44.140 & 0.013 & 0.33 & 0.02 & 87.84\ G28.83 - 0.25 & 18 44 51.509 & 0.008 & -03 45 56.284 & 0.009 & 0.37 & 0.01 & 87.84\ G28.83 - 0.25 & 18 44 52.404 & 0.008 & -03 45 44.153 & 0.009 & 0.45 & 0.02 & 88.00\ G28
.25 & 18 44 51.244 & 0.011 & -03 46 03.726 & 0.012 & 0.32 &0.02 &87.34\ G28.83-0.25 & 18 44 50.060 & 0.017 & -03 45 47.709 & 0.019 & 0.19 &0.02 &87.34\ G28.83-0.25 & 18 44 51.507 & 0.020 & -03 45 56.343 & 0.021 & 0.16 &0.01 &87.51\ G28.83-0.25 & 18 44 51.245 & 0.013 & -03 46 03.635 & 0.014 & 0.27 &0.02 &87.51\ G28.83-0.25 & 18 44 50.061 & 0.018 & -03 45 47.727 & 0.020 & 0.18 &0.02 &87.51\ G28.83-0.25 & 18 44 52.405 & 0.022 & -03 45 44.105 & 0.023 & 0.19 &0.02 &87.67\ G28.83-0.25 & 18 44 51.507 & 0.011 & -03 45 56.312 & 0.011 & 0.29 &0.01 &87.67\ G28.83-0.25 & 18 44 51.245 & 0.027 & -03 46 03.655 & 0.030 & 0.12 &0.02 &87.67\ G28.83-0.25 & 18 44 50.062 & 0.028 & -03 45 47.680 & 0.031 & 0.12 &0.02 &87.67\ J28.83-0.25 & 18 44 52.404 & 0.012 & -03 45 44.140 & 0.013 & 0.33 &0.02 &87.84\ G28.83-0.25 & 18 44 51.509 & 0.008 & -03 45 56.284 & 0.009 & 0.37 &0.01 &87.84\ N28.83-0.25 & 18 44 52.404 & 0.008 & -03 45 44.153 & 0.009 & 0.45 &0.02 &88.00\ G28
.25 & 18 44 51.244 & 0.011 46 & 0.012 0.32 &0.02 &87.34\ & & -03 45 & 0.019 & &0.02 &87.34\ G28.83-0.25 & 18 44 & 0.020 & -03 45 56.343 & 0.021 & 0.16 &0.01 &87.51\ G28.83-0.25 18 44 51.245 & 0.013 & -03 46 03.635 & 0.014 & 0.27 &87.51\ & 44 & 0.018 & -03 45 47.727 & 0.020 & 0.18 &0.02 &87.51\ G28.83-0.25 & 18 44 & 0.022 & -03 45 44.105 & 0.023 0.19 &0.02 &87.67\ G28.83-0.25 18 44 51.507 & 0.011 -03 56.312 & & &0.01 G28.83-0.25 & 18 51.245 & 0.027 & -03 46 03.655 & 0.030 & 0.12 &0.02 &87.67\ G28.83-0.25 & 18 44 & 0.028 45 47.680 0.031 0.12 &87.67\ G28.83-0.25 & 52.404 & 0.012 & -03 45 & 0.33 &0.02 &87.84\ G28.83-0.25 & 18 44 & 0.008 -03 45 56.284 & 0.009 & &0.01 &87.84\ G28.83-0.25 & 18 44 52.404 & & -03 45 44.153 & 0.009 & 0.45 &0.02 &88.00\ G28
.25 & 18 44 51.244 & 0.011 & -03 46 03.726 & 0.012 & 0.32 &0.02 &87.34\ G28.83-0.25 & 18 44 50.060 & 0.017 & -03 45 47.709 & 0.019 & 0.19 &0.02 &87.34\ G28.83-0.25 & 18 44 51.507 & 0.020 & -03 45 56.343 & 0.021 & 0.16 &0.01 &87.51\ G28.83-0.25 & 18 44 51.245 & 0.013 & -03 46 03.635 & 0.014 & 0.27 &0.02 &87.51\ G28.83-0.25 & 18 44 50.061 & 0.018 & -03 45 47.727 & 0.020 & 0.18 &0.02 &87.51\ G28.83-0.25 & 18 44 52.405 & 0.022 & -03 45 44.105 & 0.023 & 0.19 &0.02 &87.67\ G28.83-0.25 & 18 44 51.507 & 0.011 & -03 45 56.312 & 0.011 & 0.29 &0.01 &87.67\ G28.83-0.25 & 18 44 51.245 & 0.027 & -03 46 03.655 & 0.030 & 0.12 &0.02 &87.67\ G28.83-0.25 & 18 44 50.062 & 0.028 & -03 45 47.680 & 0.031 & 0.12 &0.02 &87.67\ G28.83-0.25 & 18 44 52.404 & 0.012 & -03 45 44.140 & 0.013 & 0.33 &0.02 &87.84\ G28.83-0.25 & 18 44 51.509 & 0.008 & -03 45 56.284 & 0.009 & 0.37 &0.01 &87.84\ G28.83-0.25 & 18 44 52.404 & 0.008 & -03 45 44.153 & 0.009 & 0.45 &0.02 &88.00\ G28
.25 & 18 44 51.244 & 0.011 & -03 4603.72 6 & 0. 01 2 &0.32 &0.02 &87.34\ G28. 83-0.25 & 18 44 50.060 & 0. 01 7 & - 0 345 47 .709 &0 .0 1 9 &0. 19 &0 .0 2 & 87.34 \ G 28.83-0 .25 & 18 4 4 5 1. 507 & 0.020& - 03 45 56.3 43& 0.021 & 0. 16&0.01&8 7.5 1 \ G28 .83 -0.25 & 184 4 51.2 45 & 0.01 3& -03 4 6 03.635 & 0 .014 & 0.27 &0.02 &87 . 51 \ G28.83-0.25 & 18 44 5 0 .0 6 1 &0.0 18 & -03 4 547.72 7 & 0.02 0 & 0 . 18& 0.02 &87.51\G28.83-0.25 & 1 8 44 5 2. 405 & 0.02 2 & - 03 4544.105 & 0. 023& 0.19 &0 .02 &8 7 .67\ G2 8 .83-0.2 5 & 18 44 51 .507 &0. 011 & -03 45 56 . 312 & 0.011 & 0 .29 & 0.01 & 8 7 .67\ G2 8.83 -0.25 & 18 44 51.2 45& 0. 0 27& -03 46 0 3.65 5& 0.0 30 & 0 .12 & 0. 02 &87.67\ G28. 83-0 .25 & 184450 .06 2& 0.0 2 8 & -0 3 4 5 4 7.680 & 0.031& 0. 12 & 0 .0 2 &87.67\ G28.83-0 .2 5 &18 44 52 .404 & 0. 01 2 & -03 4 544. 140& 0.013 & 0 . 33 &0.02 & 87.84\ G2 8. 83-0.25 & 18 44 5 1.5 09& 0.0 0 8 &-03 45 56.284& 0.0 0 9 & 0.37 &0.01 &87.84\ G28.8 3 -0 . 2 5& 184452.404 & 0. 008& -03 454 4. 153 & 0.0 09 &0. 4 5& 0.02 &88.00\ G28
.25 &_18 44_51.244 & 0.011 &_-03 46_03.726_& 0.012_&_0.32 &0.02 &87.34\ G28.83-0.25_& 18 44_50.060 & 0.017 &_-03 45 47.709_&_0.019 & 0.19 &0.02 &87.34\ G28.83-0.25 & 18 44 51.507 & 0.020 & -03 45_56.343_& 0.021_&_0.16_&0.01 &87.51\ G28.83-0.25 & 18 44_51.245 & 0.013 & -03_46 03.635_& 0.014 & 0.27 &0.02 &87.51\ G28.83-0.25 & 18_44_50.061 & 0.018_& -03 45 47.727 & 0.020 & 0.18 &0.02_&87.51\ G28.83-0.25 & 18 44 52.405 &_0.022 & -03_45_44.105_& 0.023 & 0.19_&0.02 &87.67\ G28.83-0.25 & 18 44 51.507_& 0.011 & -03 45 56.312_& 0.011 & 0.29 &0.01 &87.67\ G28.83-0.25 &_18 44 51.245 & 0.027 &_-03 46 03.655 & 0.030_& 0.12_&0.02 &87.67\ G28.83-0.25 & 18 44_50.062 & 0.028_& -03_45 47.680 &_0.031 & 0.12 &0.02 &87.67\ G28.83-0.25 &_18 44 52.404_& 0.012 & -03 45 44.140_&_0.013 & 0.33_&0.02_&87.84\ G28.83-0.25_& 18_44 51.509 &_0.008_& -03_45_56.284 & 0.009 & 0.37 &0.01_&87.84\ G28.83-0.25_& 18 44 52.404 & 0.008 &_-03 45 44.153 &_0.009_& 0.45 &0.02 &88.00\ G28
{e_3}{2}+2 u-2 v} b_1}{32\ 2^{3/4}}+\frac{3 e^{-e_1-\frac{e_3}{2}-\rho} b_1^2}{16 \sqrt{2}}-\frac{3 e^{-e_1+\frac{e_3}{2}+2 u-2 v} b_1^2}{32 \sqrt{2}}\,, \nonumber\\ v'&=& 2 e^{-e_1-\frac{e_3}{2}}-\frac{3}{2} e^{-e_1-\frac{e_3}{2}-\rho}+\frac{1}{2} e^{e_1-\frac{e_3}{2}-\rho}+e^{\frac{e_3}{2}-\rho}+\frac{3 e^{-e_1-\frac{e_3}{2}-\rho} b_1^2}{8 \sqrt{2}}\,,\nonumber\\ e'_1&=& 2 e^{-e_1-\frac{e_3}{2}} -\frac{3}{2} e^{-e_1-\frac{e_3}{2}-\rho} -\frac{1}{2} e^{e_1-\frac{e_3}{2}-\rho} +\frac{3}{2} e^{-e_1+\frac{e_3}{2}+2 u-2 v} -\frac{1}{4} e^{e_1+\frac{e_3}{2}+2 u-2 v}\nonumber \\ &&+\frac{15 \sqrt{5} e^{-e_1+\frac{e_3}{2}+2 u-2 v} b_1}{16\ 2^{3/4}} +\frac{3 e^{-e_1-\frac{e_3}{2}-\rho} b_1^2}{8 \sqrt{2}} +\frac{3 e^{-e_1+\frac{e_3}{2}+2 u-2 v} b_1^2}{16 \sqrt{2}} \,,\\ e'_3&=& 2 e^{-e_1-\frac{e_3}{2}} -\frac{3}{2} e^{-e_1-\frac{e_3}{2}-\rho} +\frac{1}{2} e^{e_1-\frac{e_3}{2}-\rho} -e
{ e_3}{2}+2 u-2 v } b_1}{32\ 2^{3/4}}+\frac{3 e^{-e_1-\frac{e_3}{2}-\rho } b_1 ^ 2}{16 \sqrt{2}}-\frac{3 e^{-e_1+\frac{e_3}{2}+2 u-2 v } b_1 ^ 2}{32 \sqrt{2}}\, , \nonumber\\ v'&= & 2 e^{-e_1-\frac{e_3}{2}}-\frac{3}{2 } e^{-e_1-\frac{e_3}{2}-\rho}+\frac{1}{2 } e^{e_1-\frac{e_3}{2}-\rho}+e^{\frac{e_3}{2}-\rho}+\frac{3 e^{-e_1-\frac{e_3}{2}-\rho } b_1 ^ 2}{8 \sqrt{2}}\,,\nonumber\\ e'_1&= & 2 e^{-e_1-\frac{e_3}{2 } } -\frac{3}{2 } e^{-e_1-\frac{e_3}{2}-\rho } -\frac{1}{2 } e^{e_1-\frac{e_3}{2}-\rho } + \frac{3}{2 } e^{-e_1+\frac{e_3}{2}+2 u-2 v } -\frac{1}{4 } e^{e_1+\frac{e_3}{2}+2 u-2 v}\nonumber \\ & & + \frac{15 \sqrt{5 } e^{-e_1+\frac{e_3}{2}+2 u-2 v } b_1}{16\ 2^{3/4 } } + \frac{3 e^{-e_1-\frac{e_3}{2}-\rho } b_1 ^ 2}{8 \sqrt{2 } } + \frac{3 e^{-e_1+\frac{e_3}{2}+2 u-2 five } b_1 ^ 2}{16 \sqrt{2 } } \,,\\ e'_3&= & 2 e^{-e_1-\frac{e_3}{2 } } -\frac{3}{2 } e^{-e_1-\frac{e_3}{2}-\rho } + \frac{1}{2 } e^{e_1-\frac{e_3}{2}-\rho } -e
{e_3}{2}+2 u-2 v} b_1}{32\ 2^{3/4}}+\frac{3 e^{-e_1-\frac{e_3}{2}-\rho} b_1^2}{16 \sdrt{2}}-\frac{3 e^{-e_1+\frac{e_3}{2}+2 u-2 v} b_1^2}{32 \vqrt{2}}\,, \nknumber\\ v'&=& 2 e^{-e_1-\frac{e_3}{2}}-\frac{3}{2} e^{-e_1-\frac{e_3}{2}-\rho}+\frac{1}{2} e^{w_1-\frac{t_3}{2}-\gho}+e^{\frac{e_3}{2}-\rho}+\frac{3 e^{-e_1-\frxc{e_3}{2}-\rho} b_1^2}{8 \dqrt{2}}\,,\nonunber\\ t'_1&=& 2 e^{-e_1-\frac{e_3}{2}} -\frac{3}{2} e^{-x_1-\rrac{e_3}{2}-\rho} -\frac{1}{2} c^{e_1-\frae{e_3}{2}-\cho} +\frac{3}{2} e^{-e_1+\frac{g_3}{2}+2 u-2 v} -\frac{1}{4} e^{e_1+\xrac{e_3}{2}+2 u-2 v}\nonumtef \\ &&+\yrac{15 \sqrt{5} e^{-e_1+\frac{e_3}{2}+2 u-2 v} b_1}{16\ 2^{3/4}} +\frac{3 e^{-e_1-\frac{q_3}{2}-\rho} b_1^2}{8 \xqgt{2}} +\frac{3 e^{-e_1+\frac{e_3}{2}+2 u-2 v} f_1^2}{16 \sqdn{2}} \,,\\ t'_3&=& 2 e^{-e_1-\frac{e_3}{2}} -\frac{3}{2} e^{-e_1-\frac{e_3}{2}-\rho} +\frac{1}{2} e^{s_1-\frac{e_3}{2}-\rio} -e
{e_3}{2}+2 u-2 v} b_1}{32\ 2^{3/4}}+\frac{3 e^{-e_1-\frac{e_3}{2}-\rho} b_1^2}{16 u-2 b_1^2}{32 \sqrt{2}}\,, v'&=& 2 e^{-e_1-\frac{e_3}{2}}-\frac{3}{2} e'_1&=& e^{-e_1-\frac{e_3}{2}} -\frac{3}{2} e^{-e_1-\frac{e_3}{2}-\rho} e^{e_1-\frac{e_3}{2}-\rho} +\frac{3}{2} e^{-e_1+\frac{e_3}{2}+2 v} -\frac{1}{4} e^{e_1+\frac{e_3}{2}+2 u-2 v}\nonumber \\ \sqrt{5} e^{-e_1+\frac{e_3}{2}+2 u-2 v} b_1}{16\ 2^{3/4}} +\frac{3 e^{-e_1-\frac{e_3}{2}-\rho} b_1^2}{8 \sqrt{2}} +\frac{3 e^{-e_1+\frac{e_3}{2}+2 u-2 b_1^2}{16 \sqrt{2}} \,,\\ e'_3&=& 2 e^{-e_1-\frac{e_3}{2}} -\frac{3}{2} e^{-e_1-\frac{e_3}{2}-\rho} +\frac{1}{2} e^{e_1-\frac{e_3}{2}-\rho} -e
{e_3}{2}+2 u-2 v} b_1}{32\ 2^{3/4}}+\frac{3 e^{-e_1-\frac{e_3}{2}-\rho} b_1^2}{16 \sqrt{2}}-\frAc{3 e^{-e_1+\frac{e_3}{2}+2 u-2 V} b_1^2}{32 \sqrT{2}}\,, \noNumBeR\\ v'&=& 2 e^{-e_1-\Frac{E_3}{2}}-\frac{3}{2} e^{-e_1-\frac{e_3}{2}-\rhO}+\Frac{1}{2} E^{e_1-\frac{e_3}{2}-\rho}+e^{\frac{e_3}{2}-\rho}+\fraC{3 e^{-e_1-\frAc{E_3}{2}-\Rho} b_1^2}{8 \SQrT{2}}\,,\nonuMber\\ e'_1&=& 2 e^{-e_1-\FRaC{E_3}{2}} -\FraC{3}{2} e^{-E_1-\fRac{E_3}{2}-\rHO} -\fRac{1}{2} e^{e_1-\FraC{e_3}{2}-\rho} +\frAc{3}{2} e^{-e_1+\frac{e_3}{2}+2 u-2 V} -\frAc{1}{4} E^{e_1+\frac{e_3}{2}+2 u-2 v}\nonUMbEr \\ &&+\frac{15 \sqrt{5} E^{-e_1+\fRac{e_3}{2}+2 u-2 v} b_1}{16\ 2^{3/4}} +\frac{3 e^{-E_1-\frAc{e_3}{2}-\rho} B_1^2}{8 \sQrt{2}} +\FRac{3 e^{-e_1+\FraC{e_3}{2}+2 u-2 v} b_1^2}{16 \Sqrt{2}} \,,\\ e'_3&=& 2 e^{-E_1-\Frac{e_3}{2}} -\fRac{3}{2} e^{-e_1-\frac{E_3}{2}-\rHO} +\frac{1}{2} e^{E_1-\Frac{e_3}{2}-\rhO} -E
{e_3}{2}+2 u-2 v} b_1}{32\ 2^{3/4}}+ \frac {3e^{ -e _1-\ frac {e_3}{2}-\rho} b_1^ 2}{16 \sqrt{2}}-\frac{ 3 e^{ -e _ 1+\f r ac {e_3} {2}+2 u - 2v } b_ 1^ 2} {32 \ s qr t{2}} \,, \nonum ber\\ v'&= & 2 e ^{-e_1-\frac { e_ 3}{2}}-\fr ac{ 3}{2} e^{-e_ 1-\ frac{e _3 }{2 } -\rho }+\ frac{ 1}{2}e ^{e_1- \frac{e_3 }{ 2 }-\rho } +e^{\fr a c {e _3}{ 2}-\rho}+\frac{3e ^{ - e_1-\frac{e_3} {2}-\r ho } b _ 1 ^2} {8\sqrt{2}}\ ,, \nonu m ber\\ e ' _1 & = & 2 e^{-e_1-\frac {e_3}{2}} - \ fra c{3}{2 }e^{ - e_1-\f rac{e _3 } {2} -\rho} -\fr ac{1 }{2} e^{e _1-\fr a c{e_3}{ 2 }-\rho} +\fra c{3 }{2 } e^ { -e _1 +\f ra c {e_ 3 }{ 2}+ 2 u- 2 v} -\f ra c{ 1}{4} e^{ e _ 1 + \fra c{e _3}{ 2}+2u-2 v}\nonumb er\\ & & +\f rac{1 5 \sq rt{5 }e^{-e _1+\fr ac{e_ 3} {2}+2 u-2 v} b_ 1}{1 6\ 2^{3/4 }}+\ fra c{ 3 e^{ - e_1-\f rac {e_ 3}{2}-\ rho} b_ 1 ^2} {8 \ s qr t{2}} +\frac{3 e^{ -e _ 1 +\ frac{e_3 }{2}+2 u- 2v } b_1^2} {1 6 \ sqrt { 2 }} \, ,\\e '_ 3&=& 2 e ^{-e_1 - \f ra c{e_3}{ 2} } -\fr ac {3} {2} e^{- e _1-\ frac{e _3}{2}-\ rho}+ \frac{1}{2} e^ { e_1-\frac{e_3 } {2 } - \r h o} - e
{e_3}{2}+2 u-2_v} b_1}{32\_2^{3/4}}+\frac{3 e^{-e_1-\frac{e_3}{2}-\rho} b_1^2}{16 \sqrt{2}}-\frac{3_e^{-e_1+\frac{e_3}{2}+2 u-2_v}_b_1^2}{32 \sqrt{2}}\,,_\nonumber\\ v'&=&_2 e^{-e_1-\frac{e_3}{2}}-\frac{3}{2} e^{-e_1-\frac{e_3}{2}-\rho}+\frac{1}{2}_e^{e_1-\frac{e_3}{2}-\rho}+e^{\frac{e_3}{2}-\rho}+\frac{3 e^{-e_1-\frac{e_3}{2}-\rho} b_1^2}{8_\sqrt{2}}\,,\nonumber\\ e'_1&=& 2 e^{-e_1-\frac{e_3}{2}} -\frac{3}{2}_e^{-e_1-\frac{e_3}{2}-\rho} -\frac{1}{2} e^{e_1-\frac{e_3}{2}-\rho}_+\frac{3}{2}_e^{-e_1+\frac{e_3}{2}+2 u-2 v} -\frac{1}{4} e^{e_1+\frac{e_3}{2}+2 u-2 v}\nonumber \\ &&+\frac{15 \sqrt{5} e^{-e_1+\frac{e_3}{2}+2 u-2 v} b_1}{16\ 2^{3/4}} +\frac{3_e^{-e_1-\frac{e_3}{2}-\rho}_b_1^2}{8 \sqrt{2}} +\frac{3_e^{-e_1+\frac{e_3}{2}+2_u-2_v} b_1^2}{16 \sqrt{2}} \,,\\ e'_3&=& 2_e^{-e_1-\frac{e_3}{2}} -\frac{3}{2} e^{-e_1-\frac{e_3}{2}-\rho} +\frac{1}{2} e^{e_1-\frac{e_3}{2}-\rho}_-e
k},\bm{k}')\end{aligned}$$ The linearized gap equation $$\begin{aligned} -\frac{1}{(2\pi)^2}\sum_{\beta}\oint_{FS} dk'_{\Vert}\frac{V^{\alpha\beta}(\bm{k,k'})}{v^{\beta}_{F}(\bm{k'})} \Delta_{\beta}(\bm{k'})=\lambda \Delta_{\alpha}(\bm{k}).\label{eigenvalue_Tc2}\end{aligned}$$ can be rewritten as $$\begin{aligned} \frac{1}{(2\pi)^2}\sum_{\beta}\iint_{\Delta E} d^{2}\bm{k}'V^{\alpha\beta}(\bm{k,k'}) \Delta_{\beta}(\bm{k'})=-\lambda \Delta E \Delta_{\alpha}(\bm{k}).\label{eigenvalue_Tc3}\end{aligned}$$ where the integral $\iint$ is performed within a narrow energy window near the FS with the width of the window $\Delta E\to 0$. After discreteness in the lattice, Eq. (\[eigenvalue\_Tc3\]) can be taken as an eigenvalue problem with $\lambda$ proportional to the eigenvalue and $\Delta_{\alpha}(\bm{k})$ proportional to the eigenvector. From Eq. (\[eigenvalue\_Tc3\]) and Eq. (\[symmetry\]), we can find that each solved $\Delta_{\alpha}(\bm{k})$ belong to an irreducible representation of the point group. In Table I, we list all the irreducible representation and the basis functions of the $D_6$ point group of our model in two spatial dimensions, with the pairing symmetry of each basic function marked. There are 5 different pairing symmetries, i.e. s-wave, $(p_x,p_y)$ (degenerate $p$-wave), $(d_{x^2-y^2},d_{xy})$ (degenerate $d$-wave), $f_{x\left(x^{2}-3y^{2}\right)}$ ($f$-wave), $f_{y\left(3x^{2}-y^{2}\right)}$ ($f'$-wave), each has definite parity, either even or odd. The real material has the $D_3$ point group. In Table II, we list all its irreducible representation and its basis functions in two spatial dimensions
k},\bm{k}')\end{aligned}$$ The linearized gap equation $ $ \begin{aligned } -\frac{1}{(2\pi)^2}\sum_{\beta}\oint_{FS } dk'_{\Vert}\frac{V^{\alpha\beta}(\bm{k, k'})}{v^{\beta}_{F}(\bm{k' }) } \Delta_{\beta}(\bm{k'})=\lambda \Delta_{\alpha}(\bm{k}).\label{eigenvalue_Tc2}\end{aligned}$$ can be rewritten as $ $ \begin{aligned } \frac{1}{(2\pi)^2}\sum_{\beta}\iint_{\Delta east } d^{2}\bm{k}'V^{\alpha\beta}(\bm{k, k' }) \Delta_{\beta}(\bm{k'})=-\lambda \Delta E \Delta_{\alpha}(\bm{k}).\label{eigenvalue_Tc3}\end{aligned}$$ where the built-in $ \iint$ is performed within a narrow department of energy window near the FS with the width of the windowpane $ \Delta E\to 0$. After discreteness in the lattice, Eq. (\[eigenvalue\_Tc3\ ]) can be taken as an eigenvalue trouble with $ \lambda$ proportional to the eigenvalue and $ \Delta_{\alpha}(\bm{k})$ proportional to the eigenvector. From Eq. (\[eigenvalue\_Tc3\ ]) and Eq. (\[symmetry\ ]), we can find that each solved $ \Delta_{\alpha}(\bm{k})$ belong to an irreducible theatrical performance of the item group. In Table I, we list all the irreducible theatrical performance and the basis functions of the $ D_6 $ point group of our model in two spatial dimension, with the pairing symmetry of each basic function marked. There are 5 unlike pairing symmetries, i.e. s - wave, $ (p_x, p_y)$ (degenerate $ p$-wave), $ (d_{x^2 - y^2},d_{xy})$ (pervert $ d$-wave), $ f_{x\left(x^{2}-3y^{2}\right)}$ ($ f$-wave), $ f_{y\left(3x^{2}-y^{2}\right)}$ ($ f'$-wave), each has definite parity, either even or odd. The real material have the $ D_3 $ point group. In Table II, we list all its irreducible representation and its basis functions in two spatial dimension
k},\bm{n}')\end{aligned}$$ The linearizea gap equation $$\ywgin{almgned} -\frzc{1}{(2\pi)^2}\sum_{\bdta}\oint_{FS} dk'_{\Vert}\frac{V^{\alpha\bete}(\bm{k,j'})}{v^{\betq}_{F}(\bm{k'})} \Delta_{\beta}(\bm{k'})=\lambdx \Delta_{\alpja}(\bm{k}).\labwl{eijenvalue_Tc2}\end{alijhed}$$ can be resvitteu es $$\begin{aligned} \nrac{1}{(2\pi)^2}\sum_{\beda}\iint_{\Delta E} d^{2}\tm{y}'V^{\clpha\beta}(\bm{k,k'}) \Delta_{\beta}(\bm{k'})=-\lambda \Deltw E \Delts_{\appha}(\bm{k}).\label{eidenvsjue_Td3}\vnb{aligned}$$ where the integral $\iinf$ is pegformed within a marrow energy window near hhe VS with the width lf the windiw $\Dqota E\to 0$. Aftdr discrettnzss in the mattice, Eq. (\[eigenvalue\_Tc3\]) can be gaken as an eiggucalkg problem wivh $\lamfda$ proportional to dhe eigrnvalue and $\Dekta_{\elphq}(\bm{k})$ proportional to vhe eigenvector. From Gq. (\[eigenvanuz\_Tc3\]) and Eq. (\[symmetry\]), ww xan fhnd dhat wacf sklted $\Delta_{\wlpia}(\bm{k})$ belonf to an irrwducible representauiog of the point group. Ig Table I, we list all the irreducible rtpressntation and the basis dunctions of the $D_6$ polnt group of our model in two spatial dimensions, with the [airiig synmctry if each basic function marked. There are 5 differqht psiring symmetrles, i.e. s-wave, $(p_x,p_y)$ (cehemgrate $p$-wave), $(d_{x^2-i^2},d_{xy})$ (deysnsrate $d$-wave), $f_{x\left(d^{2}-3y^{2}\right)}$ ($f$-wavw), $f_{y\left(3x^{2}-r^{2}\rignt)}$ ($f'$-wave), each has definite parity, eitheg evwn or odd. The real laterial har thg $D_3$ poont group. In Table II, wz list all its irgeducible fepresentation avd pts tasis functions in two spaeial dimeisionx
k},\bm{k}')\end{aligned}$$ The linearized gap equation $$\begin{aligned} -\frac{1}{(2\pi)^2}\sum_{\beta}\oint_{FS} \Delta_{\alpha}(\bm{k}).\label{eigenvalue_Tc2}\end{aligned}$$ be rewritten $$\begin{aligned} \frac{1}{(2\pi)^2}\sum_{\beta}\iint_{\Delta E} where integral $\iint$ is within a narrow window near the FS with the of the window $\Delta E\to 0$. After discreteness in the lattice, Eq. (\[eigenvalue\_Tc3\]) be taken as an eigenvalue problem with $\lambda$ proportional to the eigenvalue and proportional the From (\[eigenvalue\_Tc3\]) and Eq. (\[symmetry\]), we can find that each solved $\Delta_{\alpha}(\bm{k})$ belong to an irreducible representation the point group. In Table I, we list the irreducible representation and basis functions of the $D_6$ group our model two dimensions, the pairing symmetry each basic function marked. There are 5 different pairing symmetries, i.e. s-wave, $(p_x,p_y)$ (degenerate $p$-wave), $(d_{x^2-y^2},d_{xy})$ (degenerate $f_{x\left(x^{2}-3y^{2}\right)}$ ($f$-wave), each has parity, even odd. The real the $D_3$ point group. In Table all its irreducible representation and its basis functions two spatial
k},\bm{k}')\end{aligned}$$ The linearizEd gap equatIon $$\beGin{AliGnEd} -\frAc{1}{(2\pi)^2}\Sum_{\beta}\oint_{FS} dK'_{\vert}\Frac{V^{\alpha\beta}(\bm{k,k'})}{v^{\betA}_{F}(\bm{k'})} \deLTa_{\beTA}(\bM{k'})=\lamBda \DeltA_{\AlPHA}(\bm{K}).\lAbEl{eIgENvAlue_TC2}\enD{aligneD}$$ can be rewrIttEn As $$\begin{alignED} \fRac{1}{(2\pi)^2}\sum_{\beTa}\iInt_{\Delta E} d^{2}\bm{K}'V^{\aLpha\beTa}(\Bm{k,K'}) \delta_{\BetA}(\bm{k'})=-\lAmbda \DELta E \DeLta_{\alpha}(\bM{k}).\LAbel{eiGEnvalue_tC3}\EnD{aliGned}$$ where the integRAl $\IInt$ is performed Within A nARrOW EneRgy Window near ThE FS wiTH the widTH oF THE wiNDow $\Delta E\to 0$. AfTer discreteNEss In the lAtTicE, eq. (\[eigeNvaluE\_TC3\]) Can Be taken as an EigeNvalue proBlem wiTH $\lambda$ PRoportiOnal to The EigEnvaLUe AnD $\DeLtA_{\AlpHA}(\bM{k})$ pROpoRtional tO tHe EigenVectOR. fROm Eq. (\[EigEnvaLue\_Tc3\]) And Eq. (\[symmetry\]), We cAn fiND thAt eacH solvEd $\DeLtA_{\alphA}(\bm{k})$ beLong tO aN irreducible repReseNtation of The PoInt GrOup. In tAble I, wE liSt aLl the irReducibLE rePrESENtAtion and the basis fuNcTIOnS of the $D_6$ pOint grOUp Of OUr model iN tWo sPatiAL DimenSionS, WiTh the paiRing syMMeTrY of each BaSic funCtIon MarKed. ThERe arE 5 diffeRent pairIng syMMetries, i.e. s-wave, $(P_X,p_y)$ (degenerate $P$-WaVE), $(D_{x^2-Y^2},D_{xy})$ (dEgeNerate $d$-wave), $F_{x\leFT(x^{2}-3y^{2}\rIght)}$ ($F$-WaVe), $f_{Y\Left(3x^{2}-Y^{2}\righT)}$ ($f'$-WAvE), Each has definite pariTy, Either Even oR odd. The real maTerial has tHE $d_3$ Point groUp. In tAbLE II, we list all itS irreDucible repREsentatiOn and Its basis Functions IN Two spatiAl dImeNsiOns
k},\bm{k}')\end{aligned}$$ The line arize d g apeq uati on $ $\begin{aligne d } -\ frac{1}{(2\pi)^2}\sum_ {\bet a} \ oint _ {F S} dk '_{\Ver t }\ f r ac{ V^ {\ alp ha \ be ta}(\ bm{ k,k'})} {v^{\beta} _{F }( \bm{k'})} \D e lt a_{\beta}( \bm {k'})=\lambd a \ Delta_ {\ alp h a}(\b m{k }).\l abel{e i genval ue_Tc2}\e nd { aligne d }$$ can b erewr itten as $$\begin { al i gned} \frac{1} {(2\pi )^ 2 }\ s u m_{ \be ta}\iint_{ \D eltaE } d^{2} \ bm { k } 'V^ { \alpha\beta}( \bm{k,k'})\ Del ta_{\b et a}( \ bm{k'} )=-\l am b da\Delta E \D elta _{\alpha} (\bm{k } ).\labe l {eigenv alue_T c3} \en d{al i gn ed }$$ w h ere th e i n teg ral $\ii nt $is pe rfor m e d with ina na rrowenergy window ne ar t h e F S wit h the wid th of t he win dow $ \D elta E\to 0$. A fter discrete nes sinth e lat t ice, E q.(\[ eigenva lue\_Tc 3 \]) c a n be taken as an eigen va l u eproblemwith $ \ la mb d a$ propo rt ion al t o the e igen v al ue and $ \Delta _ {\ al pha}(\b m{ k})$ p ro por tio nal t o the eigen vector. From Eq. (\[eigenva l ue\_Tc3\]) an d E q . ( \ [sym met ry\]), we c an f i nd t hate ac h s o lved$\Del ta _ {\ a lpha}(\bm{k})$ belo ng to an irre ducible repre sentationo f the poin t gr o up . In Table I, w e lis t all thei rreducib le re presenta tion andt h e basisfun cti ons of t he $D_6$ pointg r oupof our mo del in two sp ati aldim en sions, wi th the p ai ri ng s ymm etryo f each b as icfu nct ion m a rked.There are 5 d i ffe rent pa i ri n g sym me tr ies, i. e. s-wa ve,$ (p_ x,p_y)$ (degener ate $p$- wa ve ), $(d_ {x^2-y^2},d_{ xy })$ (degen er ate $d$-w a v e), $f_{ x\left(x^{2}-3y^{2}\rig h t)}$ ($ f$- wave) , $f _{y\left( 3x^ {2}-y^ {2} \ right) }$ ($f '$-wa ve ),e a ch ha s de fin it e parity,e i the r eve nor o dd. Th e real material ha s th e $D_3$ point gr oup. I nTab l eI I,we lis t all its irreduc ible repre se n ta tion and i t s b as is func tions i n two spatial dimensio ns
k},\bm{k}')\end{aligned}$$ The linearized_gap equation_$$\begin{aligned} -\frac{1}{(2\pi)^2}\sum_{\beta}\oint_{FS} dk'_{\Vert}\frac{V^{\alpha\beta}(\bm{k,k'})}{v^{\beta}_{F}(\bm{k'})} \Delta_{\beta}(\bm{k'})=\lambda \Delta_{\alpha}(\bm{k}).\label{eigenvalue_Tc2}\end{aligned}$$ can be rewritten_as $$\begin{aligned} \frac{1}{(2\pi)^2}\sum_{\beta}\iint_{\Delta_E} d^{2}\bm{k}'V^{\alpha\beta}(\bm{k,k'}) \Delta_{\beta}(\bm{k'})=-\lambda_\Delta E \Delta_{\alpha}(\bm{k}).\label{eigenvalue_Tc3}\end{aligned}$$_where_the integral $\iint$_is performed within_a narrow energy window_near the FS_with_the width of the window $\Delta E\to 0$. After discreteness in the lattice, Eq._(\[eigenvalue\_Tc3\])_can be_taken_as_an eigenvalue problem with $\lambda$_proportional to the eigenvalue and_$\Delta_{\alpha}(\bm{k})$ proportional_to the eigenvector. From Eq. (\[eigenvalue\_Tc3\]) and Eq. (\[symmetry\]),_we_can find that_each solved $\Delta_{\alpha}(\bm{k})$ belong to an irreducible representation of_the point group. In Table I,_we list all_the_irreducible_representation and the basis_functions of the $D_6$ point group_of our model in two spatial_dimensions, with the pairing symmetry of each_basic function marked. There are 5_different pairing symmetries, i.e. s-wave,_$(p_x,p_y)$ (degenerate_$p$-wave), $(d_{x^2-y^2},d_{xy})$ (degenerate $d$-wave), $f_{x\left(x^{2}-3y^{2}\right)}$_($f$-wave), $f_{y\left(3x^{2}-y^{2}\right)}$ ($f'$-wave),_each has_definite parity, either_even or odd. The real material has_the $D_3$ point_group. In Table II, we list_all_its irreducible representation_and_its_basis functions_in two spatial_dimensions
{U}(L_\mathrm{bb}\tau_1)$ [^2] which is composed of disjoint parts. Therefore a combinatorial factor does not appear this time, the saddle point is unique. In spite of this, for all choices of $L_\mathrm{bb}$ the non-compact group $\mathrm{U}(L_\mathrm{bb}\tau_1)$ is unitarily equivalent to the non-compact unitary group $\mathrm{U}(k_\mathrm{b},k_\mathrm{b})$ because ${\mathrm{Tr\,}}L_\mathrm{bb}\tau_1=0$. More precisely, we can bring all matrices in $\mathrm{U}(L_\mathrm{bb}\tau_1)$ to a standard form of matrices in $\mathrm{U}(k_\mathrm{b},k_\mathrm{b})$ by a single unitary transformation. As $\mathrm{U}(L_\mathrm{bb}\tau_1)$ is a non-compact group, there is no normalizable Haar measure. Consequently, the normalization constant cannot be computed in the traditional way with the volumes of the groups, but we have to stick with the measure which is given by the pseudo-Riemannian length element, $$\label{riem-length} \begin{split} g(dU',dU')={\mathrm{Re}}\,{\mathrm{Tr\,}}(dU')^2=&\frac{1}{N}{\mathrm{Re}}\,{\mathrm{Tr\,}}\delta z^2+{\mathrm{Re}}\,{\mathrm{Tr\,}}\left[U^{-1}dU,L_\mathrm{bb}\tau_1+\frac{\delta z}{\sqrt{N}}\right]^2\\ \approx&\frac{1}{N}{\mathrm{Re}}\,{\mathrm{Tr\,}}\delta z^2+{\mathrm{Re}}\,{\mathrm{Tr\,}}\left[U^{-1}dU,L_\mathrm{bb}\tau_1\right]^2. \end{split}$$ We denote the volume element resulting from the invariant length element by $d\hat\mu(U')$ which is given by the standard formula $d\hat\mu(U')=\sqrt{\det g}\,dU'$. Thence, we have for the flat measure $$dU'=2^{-k_\mathrm{b}(2k_\mathrm{b}-1)}d\hat\mu(U')=2^{-k_\mathrm{b}(2
{ U}(L_\mathrm{bb}\tau_1)$ [ ^2 ] which is composed of disjoint parts. Therefore a combinatorial agent does not look this time, the saddle item is singular. In spite of this, for all choices of $ L_\mathrm{bb}$ the non - compendious group $ \mathrm{U}(L_\mathrm{bb}\tau_1)$ is unitarily equivalent to the non - compact unitary group $ \mathrm{U}(k_\mathrm{b},k_\mathrm{b})$ because $ { \mathrm{Tr\,}}L_\mathrm{bb}\tau_1=0$. More precisely, we can bring all matrix in $ \mathrm{U}(L_\mathrm{bb}\tau_1)$ to a standard form of matrices in $ \mathrm{U}(k_\mathrm{b},k_\mathrm{b})$ by a individual unitary transformation. As $ \mathrm{U}(L_\mathrm{bb}\tau_1)$ is a non - compact group, there constitute no normalizable Haar measure. Consequently, the normalization constant cannot be computed in the traditional way with the volume of the groups, but we have to stick with the meter which is given by the pseudo - Riemannian length element, $ $ \label{riem - length } \begin{split } g(dU',dU')={\mathrm{Re}}\,{\mathrm{Tr\,}}(dU')^2=&\frac{1}{N}{\mathrm{Re}}\,{\mathrm{Tr\,}}\delta z^2+{\mathrm{Re}}\,{\mathrm{Tr\,}}\left[U^{-1}dU, L_\mathrm{bb}\tau_1+\frac{\delta z}{\sqrt{N}}\right]^2\\ \approx&\frac{1}{N}{\mathrm{Re}}\,{\mathrm{Tr\,}}\delta z^2+{\mathrm{Re}}\,{\mathrm{Tr\,}}\left[U^{-1}dU, L_\mathrm{bb}\tau_1\right]^2. \end{split}$$ We announce the volume component resulting from the invariant length component by $ d\hat\mu(U')$ which is given by the standard formula $ d\hat\mu(U')=\sqrt{\det g}\,dU'$. Thence, we have for the flat measure $ $ dU'=2^{-k_\mathrm{b}(2k_\mathrm{b}-1)}d\hat\mu(U')=2^{-k_\mathrm{b}(2
{U}(L_\mwthrm{bb}\tau_1)$ [^2] which is comkosed of disjoint parts. Vherefode a comcinatorial factor does not a'peae thiw time, the saddle poing is uniqle. In spire oh this, for all cikices on $L_\mafmrm{bb}$ vhe non-compact nroup $\mathrk{U}(L_\mathrm{bb}\tau_1)$ ir bnitarily equivalent to the non-compast unitsrj group $\mathrm{O}(k_\matnwm{b},k_\jathrm{b})$ because ${\mathrm{Tr\,}}L_\mathrm{bb}\tzu_1=0$. More precisely, we can bring all matrices in $\matjrm{U}(L_\mathrm{bb}\tau_1)$ tl a standare fown of matricer in $\mathrm{U}(k_\mathrm{b},k_\mzthrm{b})$ by a single unitary tranrformction. As $\matyrn{U}(L_\ldthrm{bb}\tau_1)$ ms a njn-compact group, thera is no normalizable Maar keawure. Consequently, the normalization conftant cantoc be computed in the rrqditimnal way qitf tge vklumes of the groupa, but we hace to stick with tht mqqsure which ia giveg fy the pseudo-Riemannian length element, $$\nabsl{riem-length} \begin{split} g(eU',dU')={\mathrm{Re}}\,{\mathrm{Tr\,}}(dO')^2=&\frac{1}{N}{\mathwm{Re}}\,{\mathrm{Tr\,}}\delta z^2+{\mathrm{Re}}\,{\mathrm{Tr\,}}\left[U^{-1}dU,L_\mathrm{tb}\tau_1+\hrxc{\dtlbw z}{\sert{N}}\right]^2\\ \approx&\frac{1}{N}{\mathrm{Re}}\,{\mathrm{Tr\,}}\delta z^2+{\maegrk{Rv}}\,{\mathrm{Tr\,}}\left[U^{-1}dU,L_\iathrm{bb}\tau_1\tihhy]^2. \gnd{split}$$ We devote tks bolume element resklting srom rhe invaryant length element by $d\hat\mu(U')$ qhich is givvn bt the standard forlula $d\hat\mu(B')=\sqrt{\dgt g}\,dU'$. Thence, we have for the flaf measure $$dK'=2^{-k_\mathrm{b}(2i_\oathrm{b}-1)}d\hat\mu(U')=2^{-k_\mxthgm{b}(2
{U}(L_\mathrm{bb}\tau_1)$ [^2] which is composed of disjoint a factor does appear this time, In of this, for choices of $L_\mathrm{bb}$ non-compact group $\mathrm{U}(L_\mathrm{bb}\tau_1)$ is unitarily equivalent the non-compact unitary group $\mathrm{U}(k_\mathrm{b},k_\mathrm{b})$ because ${\mathrm{Tr\,}}L_\mathrm{bb}\tau_1=0$. More precisely, we can bring all in $\mathrm{U}(L_\mathrm{bb}\tau_1)$ to a standard form of matrices in $\mathrm{U}(k_\mathrm{b},k_\mathrm{b})$ by a single transformation. $\mathrm{U}(L_\mathrm{bb}\tau_1)$ a group, there is no normalizable Haar measure. Consequently, the normalization constant cannot be computed in the way with the volumes of the groups, but have to stick with measure which is given by pseudo-Riemannian element, $$\label{riem-length} g(dU',dU')={\mathrm{Re}}\,{\mathrm{Tr\,}}(dU')^2=&\frac{1}{N}{\mathrm{Re}}\,{\mathrm{Tr\,}}\delta z}{\sqrt{N}}\right]^2\\ z^2+{\mathrm{Re}}\,{\mathrm{Tr\,}}\left[U^{-1}dU,L_\mathrm{bb}\tau_1\right]^2. \end{split}$$ We the volume element resulting from the invariant length element by $d\hat\mu(U')$ which is given by the standard $d\hat\mu(U')=\sqrt{\det g}\,dU'$. have for flat $$dU'=2^{-k_\mathrm{b}(2k_\mathrm{b}-1)}d\hat\mu(U')=2^{-k_\mathrm{b}(2
{U}(L_\mathrm{bb}\tau_1)$ [^2] which is compoSed of disjoInt paRts. theReFore A comBinatorial factOR doeS not appear this time, the sAddle PoINt is UNiQue. In Spite of THiS, FOr aLl ChOicEs OF $L_\MathrM{bb}$ The non-cOmpact grouP $\maThRm{U}(L_\mathrm{bb}\TAu_1)$ Is unitarilY eqUivalent to thE noN-compaCt UniTAry grOup $\MathrM{U}(k_\matHRm{b},k_\maThrm{b})$ becaUsE ${\Mathrm{tR\,}}L_\mathrM{BB}\tAu_1=0$. MoRe precisely, we can bRInG All matrices in $\mAthrm{U}(l_\mAThRM{Bb}\tAu_1)$ tO a standard FoRm of mATrices iN $\MaTHRM{U}(k_\MAthrm{b},k_\mathrm{B})$ by a single uNItaRy tranSfOrmATion. As $\MathrM{U}(l_\MatHrm{bb}\tau_1)$ is a Non-cOmpact groUp, therE Is no norMAlizablE Haar mEasUre. consEQuEnTly, ThE NorMAlIzaTIon Constant CaNnOt be cOmpuTED IN the TraDitiOnal wAy with the voluMes Of thE GroUps, buT we haVe to StIck wiTh the mEasurE wHich is given by thE pseUdo-RiemanNiaN lEngTh ElemeNT, $$\label{RieM-leNgth} \begIn{split} G(DU',du')={\mATHRm{re}}\,{\mathrm{Tr\,}}(dU')^2=&\frac{1}{N}{\mAtHRM{RE}}\,{\mathrm{TR\,}}\delta Z^2+{\MaThRM{Re}}\,{\mathrM{TR\,}}\leFt[U^{-1}du,l_\MathrM{bb}\tAU_1+\fRac{\delta Z}{\sqrt{N}}\RIgHt]^2\\ \Approx&\fRaC{1}{N}{\mathRm{re}}\,{\mAthRm{Tr\,}}\dELta z^2+{\Mathrm{re}}\,{\mathrm{tr\,}}\lefT[u^{-1}dU,L_\mathrm{bb}\taU_1\Right]^2. \end{split}$$ wE dENOtE The vOluMe element reSultINg frOm thE InVarIAnt leNgth eLeMEnT By $d\hat\mu(U')$ which is givEn By the sTandaRd formula $d\hat\Mu(U')=\sqrt{\det G}\,Du'$. thence, we Have FOr THe flat measure $$du'=2^{-k_\matHrm{b}(2k_\mathrM{B}-1)}d\hat\mu(U')=2^{-K_\mathRm{b}(2
{U}(L_\mathrm{bb}\tau_1)$[^2] which is c omp ose dof d isjo int parts. The r efor e a combinatorial fact or do es nota pp ear t his tim e ,t h e s ad dl e p oi n tis un iqu e. In s pite of th is, f or all choic e sof $L_\mat hrm {bb}$ the no n-c ompact g rou p $\ma thr m{U}( L_\mat h rm{bb} \tau_1)$is unitar i ly equi v a le nt t o the non-compact un i tary group $\m athrm{ U} ( k_ \ m ath rm{ b},k_\math rm {b})$ because ${ \ m a thr m {Tr\,}}L_\mat hrm{bb}\tau _ 1=0 $. Mor epre c isely, we c an bri ng all matr ices in $\mat hrm{U} ( L_\math r m{bb}\t au_1)$ to astan d ar dfor mo f m a tr ice s in $\mathr m{ U} (k_\m athr m { b } ,k_\ mat hrm{ b})$by a single u nit aryt ran sform ation . A s$\mat hrm{U} (L_\m at hrm{bb}\tau_1)$ isa non-com pac tgro up , the r e is n o n orm alizabl e Haarm eas ur e . Co nsequently, the no rm a l iz ation co nstant ca nn o t be com pu ted int h e tra diti o na l way wi th the vo lu mes ofth e grou ps , b utwe ha v e to stick with th e mea s ure which is g i ven by the ps e ud o - Ri e mann ian length ele ment , $$\ labe l {r iem - lengt h} \b eg i n{ s plit} g(dU',dU')={\ ma thrm{R e}}\, {\mathrm{Tr\, }}(dU')^2= & \ f rac{1}{N }{\m a th r m{Re}}\,{\math rm{Tr \,}}\delta z^2+{\ma thrm{ Re}}\,{\ mathrm{Tr \ , }}\left[ U^{ -1} dU, L_\ m a th rm{bb}\tau_1+ \ f rac{ \d elta z} {\s qrt{N}} \ri ght ]^2 \\\a pprox&\fr ac{1}{N} {\ ma th rm {Re }}\,{ \ mathrm{T r\ ,}} \d elt a z^2 + {\math rm{Re }}\, {\ ma t hrm {Tr\,}} \ le f t [U^{ -1 }d U,L_ \ma th rm{bb }\ta u _1\ right]^ 2. \end{s pli t }$$We d enote t he volume ele me nt resulti ng fr om the i nvariant length element by $d\h a t\mu(U' )$which isgiven bythe stand ard formul a $d\h at\mu (U ')= \ s qrt{\ d e tg}\ ,d U'$. Thenc e , we have f or t he flat measure $$dU'=2^{ - k_\ mathrm{b}(2k_ \ma thrm { b }- 1)} d \h a t\m u( U ')= 2 ^ {-k_\mathrm{b}( 2
{U}(L_\mathrm{bb}\tau_1)$ [^2]_which is_composed of disjoint parts._Therefore a_combinatorial_factor does_not_appear this time,_the saddle point_is unique. In spite_of this, for_all_choices of $L_\mathrm{bb}$ the non-compact group $\mathrm{U}(L_\mathrm{bb}\tau_1)$ is unitarily equivalent to the non-compact unitary_group_$\mathrm{U}(k_\mathrm{b},k_\mathrm{b})$ because_${\mathrm{Tr\,}}L_\mathrm{bb}\tau_1=0$._More_precisely, we can bring all_matrices in $\mathrm{U}(L_\mathrm{bb}\tau_1)$ to a_standard form_of matrices in $\mathrm{U}(k_\mathrm{b},k_\mathrm{b})$ by a single unitary_transformation. As_$\mathrm{U}(L_\mathrm{bb}\tau_1)$ is a_non-compact group, there is no normalizable Haar measure. Consequently,_the normalization constant cannot be computed_in the traditional_way_with_the volumes of the_groups, but we have to stick_with the measure which is given_by the pseudo-Riemannian length element, $$\label{riem-length} \begin{split} g(dU',dU')={\mathrm{Re}}\,{\mathrm{Tr\,}}(dU')^2=&\frac{1}{N}{\mathrm{Re}}\,{\mathrm{Tr\,}}\delta z^2+{\mathrm{Re}}\,{\mathrm{Tr\,}}\left[U^{-1}dU,L_\mathrm{bb}\tau_1+\frac{\delta_z}{\sqrt{N}}\right]^2\\ \approx&\frac{1}{N}{\mathrm{Re}}\,{\mathrm{Tr\,}}\delta z^2+{\mathrm{Re}}\,{\mathrm{Tr\,}}\left[U^{-1}dU,L_\mathrm{bb}\tau_1\right]^2. \end{split}$$ We denote the volume_element resulting from the invariant_length element_by $d\hat\mu(U')$ which is given_by the standard_formula $d\hat\mu(U')=\sqrt{\det_g}\,dU'$. Thence, we_have for the flat measure $$dU'=2^{-k_\mathrm{b}(2k_\mathrm{b}-1)}d\hat\mu(U')=2^{-k_\mathrm{b}(2
sigma^+$ is obtained from $\sigma$ by flipping to $+$ the spin of all $\sigma$-components which interact with the boundary ${\partial}_e(\Delta^\circ)$ and also the spin of sites outside $\Delta^\circ$. Notice that $\sigma^+$ belongs to the support of $\mu^+_{\Delta^\circ}$ and that both $\sigma$ and $\sigma^+$ have the same energy in $\Delta^\circ$, i.e. $$\sum_{B: B \cap \Delta^\circ \neq \emptyset} \Phi_B(\sigma_B) = \sum_{B: B \cap \Delta^\circ \neq \emptyset} \Phi_B(\sigma^+_B).$$ Hence, since there are at most $2^{\# C_{\sigma^+}({\partial}_e (\Delta^\circ))}$ different configurations $\sigma'$ in which can be assigned the same configuration $\sigma^+$ and there exists $\tilde{c}_1 > 0$ such that for any configuration $\sigma$ there can be at most $\tilde{c}_1 (\# {\partial}_e(\Delta^\circ))$ $\sigma$-components in $C_\sigma({\partial}_e (\Delta^\circ))$, $$\sum_{\sigma \in A^0_+(\Delta^\circ)} e^{- \sum_{B: B \cap \Delta^\circ \neq \emptyset} \Phi_B(\sigma_B)} \leq 2^{\tilde{c}_1 (\# {\partial}_e (\Delta^\circ))} Z^+_{\Delta^\circ}$$ which implies that $$\frac{Z^0(\Delta)}{Z^+(\Delta)} \leq \frac{1+2^{\tilde{c}_1 (\# {\partial}_e (\Delta^\circ))}}{ \lambda^{|\Delta| - |\Delta^\circ|}}.$$ From here a straightforward calculation allows us to conclude the result. It follows from the Peierls bound on the energy of contours and the previous lemma that for $\lambda > 0$ sufficiently large one has $\alpha_+(\lambda) < 1$. Thus, we obtain the following result. \[teowrpt\]If $\lambda > 0$ is sufficiently large so as to satisfy $\alpha_{+}(\lambda) < 1$ then: 1. The discrete Widom-Rowlinson model on ${{\mathbb{Z}}}^d$ with exclusion radius $r$ admits two distinct Gibbs measures, $\mu^+$ and $\mu^-$. 2. The measures $\mu^{+}$ and
sigma^+$ is obtained from $ \sigma$ by flipping to $ + $ the spin of all $ \sigma$-components which interact with the boundary $ { \partial}_e(\Delta^\circ)$ and also the tailspin of web site outside $ \Delta^\circ$. Notice that $ \sigma^+$ belongs to the documentation of $ \mu^+_{\Delta^\circ}$ and that both $ \sigma$ and $ \sigma^+$ own the same energy in $ \Delta^\circ$, i.e. $ $ \sum_{B: bel \cap \Delta^\circ \neq \emptyset } \Phi_B(\sigma_B) = \sum_{B: bel \cap \Delta^\circ \neq \emptyset } \Phi_B(\sigma^+_B).$$ Hence, since there embody at most $ 2^{\ # C_{\sigma^+}({\partial}_e (\Delta^\circ))}$ different shape $ \sigma'$ in which can be assigned the same shape $ \sigma^+$ and there exists $ \tilde{c}_1 > 0 $ such that for any configuration $ \sigma$ there can be at most $ \tilde{c}_1 (\ # { \partial}_e(\Delta^\circ))$ $ \sigma$-components in $ C_\sigma({\partial}_e (\Delta^\circ))$, $ $ \sum_{\sigma \in A^0_+(\Delta^\circ) } e^{- \sum_{B: B \cap \Delta^\circ \neq \emptyset } \Phi_B(\sigma_B) } \leq 2^{\tilde{c}_1 (\ # { \partial}_e (\Delta^\circ) ) } Z^+_{\Delta^\circ}$$ which implies that $ $ \frac{Z^0(\Delta)}{Z^+(\Delta) } \leq \frac{1 + 2^{\tilde{c}_1 (\ # { \partial}_e (\Delta^\circ) ) } } { \lambda^{|\Delta| - |\Delta^\circ|}}.$$ From here a straightforward calculation allows us to reason the result. It follows from the Peierls bound on the department of energy of contours and the previous lemma that for $ \lambda > 0 $ sufficiently large one has $ \alpha_+(\lambda) < 1$. Thus, we receive the following result. \[teowrpt\]If $ \lambda > 0 $ is sufficiently large so as to satisfy $ \alpha_{+}(\lambda) < 1 $ then: 1. The discrete Widom - Rowlinson model on $ { { \mathbb{Z}}}^d$ with exclusion spoke $ r$ admits two distinct Gibbs measures, $ \mu^+$ and $ \mu^-$. 2. The measures $ \mu^{+}$ and
sigla^+$ is obtained from $\sigmx$ by flipping to $+$ the vpin or all $\siema$-components which interact wuth tye boundary ${\partial}_e(\Deuta^\circ)$ ajd also rhe wpin of sitxa outside $\Delfw^\cire$. Iotice that $\sigka^+$ belongs to the suppord uf $\mu^+_{\Delta^\circ}$ and that both $\sigma$ anq $\sigma^+$ hwve the same egergj yn $\Dslta^\circ$, i.e. $$\sum_{B: B \cap \Delta^\circ \nsq \emptjset} \Phi_B(\sigma_B) = \xum_{B: B \cap \Delta^\circ \neq \elptydet} \Phi_B(\sigma^+_B).$$ Henfe, since thgde wee at most $2^{\# Z_{\sigma^+}({\partpcl}_e (\Delta^\citc))}$ different configurations $\sigmx'$ in chich can bg zsdhgned the seme cogfiguration $\sigma^+$ ang there exists $\tilde{c}_1 > 0$ vucy that for any confignration $\sigma$ there san be at mkst $\tilde{c}_1 (\# {\partiql}_w(\Deltd^\cirw))$ $\siena$-cumpknxnta in $C_\digja({\partial}_e (\Delta^\circ))$, $$\wum_{\sigma \in A^0_+(\Delta^\cors)} e^{- \sum_{B: B \cap \Delta^\siwc \neq \emptyset} \Phi_B(\sigma_B)} \leq 2^{\tilde{c}_1 (\# {\kartizl}_e (\Delta^\circ))} Z^+_{\Delta^\cirx}$$ which implies that $$\vrac{Z^0(\Deltw)}{Z^+(\Delta)} \leq \frac{1+2^{\tilde{c}_1 (\# {\partial}_e (\Delta^\circ))}}{ \lambda^{|\Gelta| - |\Deotw^\zurf|}}.$$ From here a straightforward calculation alljss ux to conclude bhe result. It folloes ftjm the Peierlr bounb oh the energy of cojtours wnd tye previots lrmma that for $\lambda > 0$ suffuciently larye ine has $\alpha_+(\lambdc) < 1$. Thus, we ubtaon thr following result. \[teowr't\]If $\lzmbda > 0$ is dufficienfuy large so as tu ssthsfy $\alpha_{+}(\lambda) < 1$ then: 1. Tre discreve Wibom-Rowlivson model on ${{\mathbb{X}}}^d$ wibv exclusion radius $r$ adlids two dishinct Gibbs measures, $\mu^+$ and $\mu^-$. 2. The measures $\ku^{+}$ ang
sigma^+$ is obtained from $\sigma$ by flipping the of all which interact with the of sites outside Notice that $\sigma^+$ to the support of $\mu^+_{\Delta^\circ}$ and both $\sigma$ and $\sigma^+$ have the same energy in $\Delta^\circ$, i.e. $$\sum_{B: B \Delta^\circ \neq \emptyset} \Phi_B(\sigma_B) = \sum_{B: B \cap \Delta^\circ \neq \emptyset} \Phi_B(\sigma^+_B).$$ Hence, there at $2^{\# (\Delta^\circ))}$ different configurations $\sigma'$ in which can be assigned the same configuration $\sigma^+$ and there exists > 0$ such that for any configuration $\sigma$ can be at most (\# {\partial}_e(\Delta^\circ))$ $\sigma$-components in $C_\sigma({\partial}_e $$\sum_{\sigma A^0_+(\Delta^\circ)} e^{- B \Delta^\circ \emptyset} \Phi_B(\sigma_B)} \leq (\# {\partial}_e (\Delta^\circ))} Z^+_{\Delta^\circ}$$ which implies that $$\frac{Z^0(\Delta)}{Z^+(\Delta)} \leq \frac{1+2^{\tilde{c}_1 (\# {\partial}_e (\Delta^\circ))}}{ \lambda^{|\Delta| - |\Delta^\circ|}}.$$ From a straightforward us to the It from the Peierls the energy of contours and the for $\lambda > 0$ sufficiently large one has < 1$. we obtain the following result. \[teowrpt\]If > 0$ is sufficiently large so as to $\alpha_{+}(\lambda) < 1$ then: 1. The discrete Widom-Rowlinson model on ${{\mathbb{Z}}}^d$ with exclusion radius $r$ distinct Gibbs measures, $\mu^+$ $\mu^-$. 2. The $\mu^{+}$
sigma^+$ is obtained from $\sigma$ bY flipping tO $+$ the sPin Of aLl $\SigmA$-comPonents which inTEracT with the boundary ${\partiaL}_e(\DelTa^\CIrc)$ aND aLso thE spin of SItES OutSiDe $\delTa^\CIrC$. NotiCe tHat $\sigmA^+$ belongs to The SuPport of $\mu^+_{\DelTA^\cIrc}$ and that BotH $\sigma$ and $\sigMa^+$ hAve the SaMe eNErgy iN $\DeLta^\ciRc$, i.e. $$\suM_{b: B \cap \DElta^\circ \nEq \EMptyseT} \phi_B(\sigMA_b) = \sUm_{B: B \Cap \Delta^\circ \neq \emPTySEt} \Phi_B(\sigma^+_B).$$ HeNce, sinCe THeRE Are At mOst $2^{\# C_{\sigma^+}({\pArTial}_e (\dElta^\cirC))}$ DiFFERenT ConfigurationS $\sigma'$ in whiCH caN be assIgNed THe same ConfiGuRAtiOn $\sigma^+$ and tHere Exists $\tilDe{c}_1 > 0$ sucH That for ANy confiGuratiOn $\sIgmA$ theRE cAn Be aT mOSt $\tILdE{c}_1 (\# {\pARtiAl}_e(\Delta^\CiRc))$ $\Sigma$-CompONENTs in $c_\siGma({\pArtiaL}_e (\Delta^\circ))$, $$\suM_{\siGma \iN a^0_+(\DeLta^\ciRc)} e^{- \suM_{B: B \cAp \delta^\Circ \neQ \emptYsEt} \Phi_B(\sigma_B)} \leq 2^{\TildE{c}_1 (\# {\partial}_E (\DeLtA^\ciRc))} z^+_{\DeltA^\Circ}$$ whIch ImpLies thaT $$\frac{Z^0(\DELta)}{z^+(\DELTA)} \lEq \frac{1+2^{\tilde{c}_1 (\# {\partiaL}_e (\dELtA^\circ))}}{ \lamBda^{|\DelTA| - |\DElTA^\circ|}}.$$ FroM hEre A strAIGhtfoRwarD CaLculatioN allowS Us To ConcludE tHe resuLt. it fOllOws frOM the peierlS bound on The enERgy of contours aND the previous lEMmA THaT For $\lAmbDa > 0$ sufficienTly lARge oNe haS $\AlPha_+(\LAmbda) < 1$. thus, wE oBTaIN the following result. \[TeOwrpt\]IF $\lambDa > 0$ is sufficienTly large so AS TO satisfy $\AlphA_{+}(\LaMBda) < 1$ then: 1. The discRete WIdom-RowlinSOn model oN ${{\mathBb{Z}}}^d$ with Exclusion RADius $r$ admIts Two DisTinCT giBbs measures, $\mu^+$ AND $\mu^-$. 2. THe MeasureS $\mu^{+}$ And
sigma^+$ is obtained from$\sigma$ b y fli ppi ngto $+$ the spin of all $ \ sigm a$-components which in terac tw itht he boun dary ${ \ pa r t ial }_ e( \De lt a ^\ circ) $ a nd also the spinofsi tes outside$ \D elta^\circ $.Notice that$\s igma^+ $bel o ngs t o t he su pporto f $\mu ^+_{\Delt a^ \ circ}$ and tha t bo th $ \sigma$ and $\sig m a^ + $ have the sam e ener gy in $ \De lta ^\circ$, i .e . $$\ s um_{B:B \ c a p \D e lta^\circ \ne q \emptyset } \P hi_B(\ si gma _ B) = \ sum_{ B: B \ cap \Delta^ \cir c \neq \e mptyse t } \Phi_ B (\sigma ^+_B). $$Hen ce,s in ce th er e ar e a t m o st$2^{\# C _{ \s igma^ +}({ \ p a r tial }_e (\D elta^ \circ))}$ dif fer entc onf igura tions $\s ig ma'$in whi ch ca nbe assigned the sam e configu rat io n $ \s igma^ + $ andthe reexists$\tilde { c}_ 1> 0 $such that for anyco n f ig uration$\sigm a $th e re can b eatmost $ \tild e{c} _ 1(\# {\pa rtial} _ e( \D elta^\c ir c))$ $ \s igm a$- compo n ents in $C _\sigma( {\par t ial}_e (\Delta ^ \circ))$, $$\ s um _ { \s i gma\in A^0_+(\Del ta^\ c irc) } e^ { -\su m _{B:B \ca p\ De l ta^\circ \neq \empt ys et} \P hi_B( \sigma_B)} \l eq 2^{\til d e { c}_1 (\# {\p a rt i al}_e (\Delta^ \circ ))} Z^+_{\ D elta^\ci rc}$$ which i mplies th a t $$\frac {Z^ 0(\ Del ta) } { Z^ +(\Delta)} \l e q \fr ac {1+2^{\ til de{c}_1 (\ # { \pa rti al }_e (\Del ta^\circ )) }} {\l amb da^{| \ Delta| - | \De lt a^\ circ| } }.$$ F rom h ereast r aig htforwa r dc a lcul at io n al low sus to con c lud e the r esult. I t f o llow sfr om thePeierls bound o n the ener gy of conto u r s and th e previous lemma that f o r $\lam bda > 0$ suf ficiently la rge on e h a s $\al pha_+( \lamb da ) < 1 $. Th u s ,weob tain the f o l low ing r es ult. \[teo wrpt\]If $\lambda> 0$ is sufficien tly lar g e s o a s t o sa ti s fy$ \ alpha_{+}(\lamb da) < 1$ t he n : 1. The d i scr et e Widom -Rowlin son m o del on${{\mathb b{Z}}}^d$ w ithe x clu sion radiu s $r$ ad mits twod istin c tGibbs me asures ,$\m u^+$and $\ m u^- $. 2 . The m easure s $\m u^ {+}$ and
sigma^+$ is_obtained from_$\sigma$ by flipping to_$+$ the_spin_of all_$\sigma$-components_which interact with_the boundary ${\partial}_e(\Delta^\circ)$_and also the spin_of sites outside_$\Delta^\circ$._Notice that $\sigma^+$ belongs to the support of $\mu^+_{\Delta^\circ}$ and that both $\sigma$ and_$\sigma^+$_have the_same_energy_in $\Delta^\circ$, i.e. $$\sum_{B: B_\cap \Delta^\circ \neq \emptyset} \Phi_B(\sigma_B)_= \sum_{B:_B \cap \Delta^\circ \neq \emptyset} \Phi_B(\sigma^+_B).$$ Hence, since_there_are at most_$2^{\# C_{\sigma^+}({\partial}_e (\Delta^\circ))}$ different configurations $\sigma'$ in which can_be assigned the same configuration $\sigma^+$_and there exists_$\tilde{c}_1_>_0$ such that for_any configuration $\sigma$ there can be_at most $\tilde{c}_1 (\# {\partial}_e(\Delta^\circ))$ $\sigma$-components_in $C_\sigma({\partial}_e (\Delta^\circ))$, $$\sum_{\sigma \in A^0_+(\Delta^\circ)} e^{-_\sum_{B: B \cap \Delta^\circ \neq \emptyset}_\Phi_B(\sigma_B)} \leq 2^{\tilde{c}_1 (\# {\partial}_e_(\Delta^\circ))} Z^+_{\Delta^\circ}$$_which implies that $$\frac{Z^0(\Delta)}{Z^+(\Delta)} \leq_\frac{1+2^{\tilde{c}_1 (\# {\partial}_e_(\Delta^\circ))}}{ \lambda^{|\Delta|_- |\Delta^\circ|}}.$$ From_here a straightforward calculation allows us_to conclude the_result. It follows from the Peierls bound_on_the energy of_contours_and_the previous_lemma that for_$\lambda_> 0$_sufficiently_large one has $\alpha_+(\lambda) < 1$._Thus,_we obtain the following result. \[teowrpt\]If $\lambda >_0$ is sufficiently large_so_as to satisfy $\alpha_{+}(\lambda)_< 1$ then: 1. The_discrete Widom-Rowlinson model on ${{\mathbb{Z}}}^d$ with_exclusion radius_$r$ admits_two distinct Gibbs measures, $\mu^+$ and $\mu^-$. 2. The measures $\mu^{+}$_and
bowls were placed at 7 different positions. At each position, we collect 20 sequences of data with 5 for each bowl-barley combination. In total, we collect 140 sequences of data. 139 of the 140 sequences of data are used to train DCAE and LSTM-RNN. The remaining 1 sequence are used for testing. The parameters of DCAE and LSTM-RNN are the same as experiment 1. The results of DCAE are shown in Fig.\[dcaeresults\](b). The trained model is able to reconstruct the training data with high precision. Readers may compare the first and second rows of Fig.\[dcaeresults\](b.1) for details. In contrast, the reconstructed images show significant difference for the test data. It failed to reconstruct the test data. Readers may compare the first and second row of Fig.\[dcaeresults\](b.2) to see the difference. Especially for the first column of Fig.\[dcaeresults\](b.2), the bowl is wrongly considered to be at a totally different position. The LSTM-RNN model is not able to generate scooping motion for either the training data or the test data. The motion is randomly changing from time to time. It doesn’t follow any pre-taught sequences. The reason is probably the bad reconstruction performance of DCAE. The system failed to correctly find the positions of the bowls using the encoded features. Based on the analysis, we increase the training data of DCAE in Experiment 3 to improve its reconstruction. Experiment 3: Increasing the training data of DCAE -------------------------------------------------- The third group of experiments has exactly the same scenario settings and parameter settings as Experiment 2, except that we use planning algorithms to generate scooping motion and collect more scooping images. The new scooping images are collected following the work flow shows in Fig.\[sample\]. We divide the work space into around 100 grids, place bowl at these places, and sample arm boatswains and orientations at each of the grid. In total, we additionally generate 100$\times$45$\times$3=13500 (12726 exactly) extra training images to train DCAE. Here, “100” indicates the 100 grid positions. “45” and “3” indicate the 45 arm positions and 3 arm rotation angles sampled at each grid. ![Increase the training data of DCAE by automatically generating motions across a 10$\times$10 grids. In all, 100$\times$45
bowls were placed at 7 different positions. At each placement, we roll up 20 sequences of data with 5 for each bowl - barley combination. In sum, we collect 140 sequences of datum. 139 of the 140 sequences of data are used to discipline DCAE and LSTM - RNN. The remaining 1 sequence are use for testing. The parameters of DCAE and LSTM - RNN are the like as experiment 1. The results of DCAE are shown in Fig.\[dcaeresults\](b). The trained model is able to reconstruct the training data with eminent precision. Readers may compare the first and second course of Fig.\[dcaeresults\](b.1) for details. In contrast, the reconstructed images show meaning difference for the examination data. It failed to reconstruct the trial data. Readers may compare the first and second row of Fig.\[dcaeresults\](b.2) to see the difference. Especially for the first column of Fig.\[dcaeresults\](b.2), the bowl is wrongly considered to be at a totally different position. The LSTM - RNN model is not able to generate scooping gesture for either the training data or the examination datum. The apparent motion is randomly changing from time to time. It does n’t follow any pre - taught sequences. The reason is probably the regretful reconstruction performance of DCAE. The system failed to correctly find the positions of the bowls using the encoded features. Based on the analysis, we increase the training data of DCAE in Experiment 3 to better its reconstruction. Experiment 3: Increasing the education datum of DCAE -------------------------------------------------- The third group of experiment has exactly the same scenario mount and parameter settings as Experiment 2, except that we use plan algorithms to generate scooping motion and gather more scooping images. The raw scooping images are collected follow the work flow shows in Fig.\[sample\ ]. We separate the work quad into around 100 grids, place roll at these space, and sample arm boatswains and orientations at each of the grid. In total, we additionally generate 100$\times$45$\times$3=13500 (12726 exactly) extra education images to train DCAE. Here, “ 100 ” indicate the 100 grid positions. “ 45 ” and “ 3 ” indicate the 45 arm positions and 3 arm rotation angles sample at each grid. ! [ Increase the training data of DCAE by automatically generating motions across a 10$\times$10 grids. In all, 100$\times$45
boals were placed at 7 difftrent positions. Aj wach pmsitioh, we coluect 20 sequences of data with 5 dor eqch bowl-barley combinagion. In tltal, we xolltct 140 sequences of data. 139 on the 140 seqbeices of data arg used to trdin DCAE and LVTO-RUN. The remaining 1 sequence are used sor tesyijg. The parametgrs og DCAS and LSTM-RNN are the same as expediment 1. Uhe results of DCAR are shown in Fig.\[dcaeresupts\](b). The trained model is able to recjbstruct the graining data with higg precision. Readers may compare the yirst and sgeind tows of Fig.\[draeresllts\](b.1) for details. In wontrasy, the reconstrmcted imqges show significant difference for thg test datd. Nt failed to reconstrycr the tesd dagq. Rdadtrs mzy comoarx the first and second row of Fig.\[dcaeresuktf\](v.2) to see the sifferqnse. Especially for the first column of Fpg.\[dczeresults\](b.2), the bowl is qrongly considered to be at a eotally different position. The LSTM-RNN model is nod ablx go yckeragw dcooping motion for either the training data jd uhe test data. The motion is tajdpily changing wrom tnje to time. It doesn’t follow any pre-taught seqiences. The reason is probaboy the bad rvconwtruction performauce of DCAE. Che syxtem gailed to correctly finb the lositions ov the bowmr using the encoaed faatures. Based on the analyfis, we inrreasz the trxinimg datw of DCAE ln Exiariment 3 to improvf its raconstructlon. Experiment 3: Increasing the tceining data og GCAV -------------------------------------------------- The thirb groui of experimentf has exactly jhe same fcenafio settinfs and 'arameter seetings as Expatiment 2, excepv that we use plabning aueorithms to gemerate scooping motuon and collect move scukping images. The neq scooping imagrs xre cllkesded followinc thd wutk fluw shows in Wig.\[ssmple\]. We divide the fork space into around 100 nrids, plaxe bowl wt these placrs, and sample arm hoatsxains end oroenjations at each of the grid. In fotal, we wddltionally genqratc 100$\tiies$45$\times$3=13500 (12726 zxactly) extra training images to train DRAE. Here, “100” indicates thg 100 grid positions. “45” anb “3” indicate tie 45 ari positiots and 3 arm rotation angles sampled au each grid. ![Increase the trainhng dwta of DCAE by automatically generating motions across a 10$\times$10 grids. In aol, 100$\timxs$45
bowls were placed at 7 different positions. position, collect 20 of data with In we collect 140 of data. 139 the 140 sequences of data are to train DCAE and LSTM-RNN. The remaining 1 sequence are used for testing. parameters of DCAE and LSTM-RNN are the same as experiment 1. The results DCAE shown Fig.\[dcaeresults\](b). trained model is able to reconstruct the training data with high precision. Readers may compare the and second rows of Fig.\[dcaeresults\](b.1) for details. In the reconstructed images show difference for the test data. failed reconstruct the data. may the first and row of Fig.\[dcaeresults\](b.2) to see the difference. Especially for the first column of Fig.\[dcaeresults\](b.2), the bowl is considered to a totally position. LSTM-RNN is not able scooping motion for either the training test data. The motion is randomly changing from to time. doesn’t follow any pre-taught sequences. The is probably the bad reconstruction performance of DCAE. system failed to correctly find the positions of the bowls using the encoded features. Based analysis, we increase the data of DCAE Experiment to its Experiment 3: the training data of DCAE -------------------------------------------------- The third group of experiments exactly the same scenario settings and parameter settings as Experiment that use planning algorithms generate scooping motion and more images. The new scooping collected the in We the work space into 100 grids, place bowl at places, and sample arm of the grid. In total, we additionally generate (12726 exactly) extra training images to train Here, “100” indicates the 100 grid positions. “45” and “3” indicate the arm positions arm rotation angles sampled at each grid. ![Increase training data of DCAE automatically generating motions across a 10$\times$10 grids. In all,
bowls were placed at 7 differenT positions. at eacH poSitIoN, we cOlleCt 20 sequences of dATa wiTh 5 for each bowl-barley comBinatIoN. in toTAl, We colLect 140 seqUEnCES of DaTa. 139 Of tHe 140 SEqUenceS of Data are Used to traiN DCaE And LSTM-RNN. ThE ReMaining 1 seqUenCe are used for TesTing. ThE pAraMEters Of DcAE anD LSTM-Rnn are thE same as exPeRIment 1. THE resultS OF DcAE aRe shown in Fig.\[dcaerESuLTs\](b). The trained mOdel is AbLE tO REcoNstRuct the traInIng daTA with hiGH pRECIsiON. Readers may coMpare the firST anD seconD rOws OF Fig.\[dcAeresUlTS\](b.1) fOr details. In ContRast, the reConstrUCted imaGEs show sIgnifiCanT diFferENcE fOr tHe TEst DAtA. It FAilEd to recoNsTrUct thE tesT DATA. ReaDerS may CompaRe the first and SecOnd rOW of fig.\[dcAeresUlts\](B.2) tO see tHe diffErencE. ESpecially for the FirsT column of fig.\[DcAerEsUlts\](b.2), THe bowl Is wRonGly consIdered tO Be aT a TOTAlLy different positioN. THE lStM-RNN modEl is noT AbLe TO generatE sCooPing MOTion fOr eiTHeR the traiNing daTA oR tHe test dAtA. The moTiOn iS raNdomlY ChanGing frOm time to Time. IT Doesn’t follow anY Pre-taught sequENcES. thE ReasOn iS probably thE bad REconStruCTiOn pERformAnce oF DcaE. tHe system failed to corReCtly fiNd the Positions of thE bowls usinG THE encoded FeatUReS. based on the analYsis, wE increase tHE traininG data Of DCAE in experimenT 3 TO improve Its RecOnsTruCTIoN. Experiment 3: InCREasiNg The traiNinG data of dCAe -------------------------------------------------- ThE thIrd GrOup of expeRiments hAs ExAcTlY thE same SCenario sEtTinGs And ParamETer setTings As ExPeRiMEnt 2, Except tHAt WE Use pLaNnIng aLgoRiThms tO genERatE scoopiNg motion aNd cOLlecT mOrE scoopiNg images. The neW sCooping imaGeS arE colleCTEd followIng the work flow shows in FiG.\[Sample\]. WE diVide tHe woRk space inTo aRound 100 gRidS, Place bOwl at tHese pLaCes, AND sampLE ArM boAtSwains and oRIEntAtionS aT eacH of the gRid. In total, we additiONalLy generate 100$\timEs$45$\tImes$3=13500 (12726 EXAcTly) EXtRA trAiNIng IMAges to train DCAE. here, “100” indicaTeS ThE 100 grid positIOns. “45” AnD “3” indicaTe the 45 arM posiTIons and 3 Arm rotatiOn angles sAmPled AT EacH grid. ![IncreAse the trAining datA Of DCAe By AutomAtiCally gEnEraTing mOtions ACroSs a 10$\tiMes$10 griDs. in all, 100$\tImes$45
bowls were placed at 7 di fferent po sitio ns. At e achposi tion, we colle c t 20 sequences of data wit h 5 f or each bo wl-ba rley co m bi n a tio n. I n t ot a l, we c oll ect 140 sequences of d ata. 139 oft he 140 seque nce s of data ar e u sed to t rai n DCAE an d LST M-RNN. The re maining 1 s e quence are use d fo r te sting. The parame t er s of DCAE and L STM-RN Na re t hesam e as exper im ent 1 . The r e su l t s of DCAE are show n in Fig.\[ d cae result s\ ](b ) . Thetrain ed mod el is ableto r econstruc t thet raining data wi th hig h p rec isio n .Re ade rs may co mpa r e t he first a nd seco nd r o w s of F ig. \[dc aeres ults\](b.1) f ordeta i ls. In c ontra st,th e rec onstru ctedim ages show signi fica nt differ enc efor t he te s t data . I t f ailed t o recon s tru ct t h etest data. Readers m a y c ompare t he fir s tan d secondro w o f Fi g . \[dca eres u lt s\](b.2) to se e t he differ en ce. Es pe cia lly fort he f irst c olumn of Fig. \ [dcaeresults\] ( b.2), the bow l i s wr o ngly co nsidered to bea t atota l ly di f feren t pos it i on . The LSTM-RNN mode lis not able to generatescooping m o t i on for e ithe r t h e training dat a orthe test d a ta. Themotio n is ran domly cha n g ing from ti metotim e . I t doesn’t fol l o w an ypre-tau ght sequen ces . T herea so n is prob ably the b ad r ec ons truct i on perfo rm anc eofDCAE. The sy stemfail ed t o co rrectly fi n d the p os itio nsof thebowl s us ing the encodedfea t ures .Ba sed onthe analysis, w e increase t hetraini n g data of DCAE in Experiment 3 t o improv e i ts re cons truction. E xperim ent 3: Inc reasin g the t rai n i ng da t a o f D CA E -------- - - --- ----- -- ---- ------- ------------------ - T he third grou p o f ex p e ri men t sh asex a ctl y the same scenar io setting sa nd parameter set ti ngs asExperim ent 2 , except that weuse plann in g al g o rit hms to gen erate sc ooping mo t ion a n dcolle ctmore s co opi ng im ages.The newscoopi ng image s are c ollected following the work flo w show s inFig .\[sample \]. Wedivide th e wo rk space i nto ar ound100 grids , pl a ce bo w l atthes e places,a nd sa m p le arm boatsw a i n s a nd or ien t ations ateach of the grid. In total, we a ddit i o nal lyg ener at e 100$\times$4 5$\ ti m e s$3=1350 0(12726 exac tly) ext ra train ing im ages t o train D CA E . Here , “1 00” indicate s t he 100 gri dpo s itions . “4 5” and “ 3” ind i cate t he 45 arm positi ons a n d 3 ar m ro tatio nangless ampl ed at each grid. ![I ncreas e th e tra ining d at a of D CAE b y automati c ally gene ratin g motio ns acr oss a 10$ \tim e s $10 g rids .Inall, 100$ \ t im e s$ 45
bowls_were placed_at 7 different positions._At each_position,_we collect_20_sequences of data_with 5 for_each bowl-barley combination. In_total, we collect_140_sequences of data. 139 of the 140 sequences of data are used to train_DCAE_and LSTM-RNN._The_remaining_1 sequence are used for_testing. The parameters of DCAE_and LSTM-RNN_are the same as experiment 1. The results of_DCAE_are shown in_Fig.\[dcaeresults\](b). The trained model is able to reconstruct the_training data with high precision. Readers_may compare the_first_and_second rows of Fig.\[dcaeresults\](b.1)_for details. In contrast, the reconstructed_images show significant difference for the_test data. It failed to reconstruct the_test data. Readers may compare the_first and second row of_Fig.\[dcaeresults\](b.2) to_see the difference. Especially for_the first column_of Fig.\[dcaeresults\](b.2),_the bowl is_wrongly considered to be at a_totally different position. The_LSTM-RNN model is not able to_generate_scooping motion for_either_the_training data_or the test_data._The motion_is_randomly changing from time to time._It_doesn’t follow any pre-taught sequences. The reason_is probably the bad_reconstruction_performance of DCAE. The_system failed to correctly find_the positions of the bowls using_the encoded_features. Based_on the analysis, we increase the training data of DCAE in_Experiment 3 to improve its reconstruction. Experiment_3: Increasing the training_data of_DCAE -------------------------------------------------- The_third group of_experiments_has exactly_the same scenario settings and parameter settings_as Experiment_2, except that we use planning_algorithms to generate scooping_motion_and collect more scooping images. The new_scooping images are collected following the_work flow shows in Fig.\[sample\]._We_divide_the work space into around_100 grids, place bowl at these_places, and sample_arm boatswains and orientations at each of_the_grid. In total, we additionally generate_100$\times$45$\times$3=13500_(12726 exactly) extra training images to_train_DCAE._Here, “100” indicates the 100_grid positions. “45” and “3” indicate_the 45 arm positions and 3 arm rotation angles_sampled at each_grid. ![Increase the training data of_DCAE_by_automatically generating motions across a 10$\times$10 grids. In all, 100$\times$45
sum_{j=1}^q \log f_{j,\theta}(A_j Z\mid B_j Z),$$ where $f_{j,\theta}$ is the conditional Gaussian density of $A_j Z$ given $B_j Z$. As proposed by @vecchia and @steinchiwelty, the rank of $B_j$ will generally be larger than that of $A_j$, in which case the main computation in obtaining (\[composite\]) is finding Cholesky decompositions of the covariance matrices of $B_1Z,\ldots, B_q Z$. For example, @vecchia just lets $A_j Z$ be the $j$th component of $Z$ and $B_j Z$ some subset of $Z_1,\ldots,Z_{j-1}$. If $m$ is the largest of these subsets, then the storage requirements for this computation are $O(m^2)$ rather than $O(n^2)$. Comparable to increasing the number of $U_j$’s in the randomized algorithm used here, this approach can be updated to obtain a better approximation of the likelihood by increasing the size of the subset of $Z_1,\ldots,Z_{j-1}$ to condition on when computing the conditional density of $Z_j$. However, for this approach to be efficient from the perspective of flops, one needs to store the Cholesky decompositions of the covariance matrices of $B_1Z,\ldots,B_q Z$, which would greatly increase the memory requirements of the algorithm. For dealing with truly massive data sets, our long-term plan is to combine the randomized approach studied here with a composite likelihood by using the randomized algorithms to compute the gradient of (\[composite\]), thus making it possible to consider $A_j$’s and $B_j$’s of larger rank than would be feasible if one had to do exact calculations. Section \[sec2\] provides a bound on the efficiency of the estimating equations based on the approximate likelihood relative to the Fisher information matrix. The bound is in terms of the condition number of the true covariance matrix of the observations and shows that if the covariance matrix is well conditioned, $N$ does not need to be very large to obtain nearly optimal estimating equations. Section \[sec3\] shows how one can get improved estimating equations by choosing the $U_j$’s in (\[ascore\]) based on a design related
sum_{j=1}^q \log f_{j,\theta}(A_j Z\mid B_j Z),$$ where $ f_{j,\theta}$ is the conditional Gaussian density of $ A_j Z$ given $ B_j Z$. As proposed by @vecchia and @steinchiwelty, the rank of $ B_j$ will by and large be large than that of $ A_j$, in which case the main computation in receive (\[composite\ ]) is finding Cholesky decompositions of the covariance matrix of $ B_1Z,\ldots, B_q Z$. For exercise, @vecchia just let $ A_j Z$ be the $ j$th component of $ Z$ and $ B_j Z$ some subset of $ Z_1,\ldots, Z_{j-1}$. If $ m$ is the largest of these subset, then the storage requirements for this computation are $ O(m^2)$ rather than $ O(n^2)$. Comparable to increase the number of $ U_j$ ’s in the randomized algorithm used here, this approach can be updated to obtain a effective approximation of the likelihood by increasing the size of the subset of $ Z_1,\ldots, Z_{j-1}$ to stipulate on when computing the conditional density of $ Z_j$. However, for this approach to be efficient from the position of flops, one want to store the Cholesky decompositions of the covariance matrices of $ B_1Z,\ldots, B_q Z$, which would greatly increase the memory requirements of the algorithm. For consider with truly massive data sets, our long - term design is to unite the randomized approach studied here with a composite likelihood by using the randomized algorithms to compute the gradient of (\[composite\ ]), thus making it possible to see $ A_j$ ’s and $ B_j$ ’s of larger rank than would be feasible if one have to dress exact calculations. incision \[sec2\ ] provides a bound on the efficiency of the estimating equations free-base on the approximate likelihood relative to the Fisher information matrix. The bound is in terms of the condition number of the dependable covariance matrix of the observations and shows that if the covariance matrix is well conditioned, $ N$ does not need to be very big to obtain about optimal estimating equations. Section \[sec3\ ] shows how one can get improved estimate equations by choosing the $ U_j$ ’s in (\[ascore\ ]) based on a invention related
sum_{u=1}^q \log f_{j,\theta}(A_j Z\mid B_j E),$$ where $f_{j,\theta}$ is the cmnditiknal Gaursian density of $A_j Z$ given $U_j Z$. As peoposed by @vecchia and @steinchiaelty, thw raik of $B_j$ will geisrally nz larfcr thcn that of $A_j$, in which casa the main com[ugacion in obtaining (\[composite\]) is findigg Cholrsny decompositijns ps ths covariance matrices of $B_1Z,\ldots, B_q Z$. For txample, @vecchia juxt lets $A_j Z$ be the $j$th colponfnt of $Z$ and $B_j Z$ dome subset of $S_1,\odots,Z_{j-1}$. If $m$ is the lagyest of theae subsets, then the storage reqjiremznts for thus colkutation are $O(m^2)$ rwther than $O(k^2)$. Compasable tp increasing tme nukbee of $U_j$’s in the randokized algorithm usgd here, thhs approach can be ypeated to mbtakb a beutec alproxilatmon of the mikelihood vy increasing the sozq of the subsef of $Z_1,\jdjts,Z_{j-1}$ to condition on when computing tht consitional density of $Z_j$. Yowever, for this apprlach to bq efficient from the perspective of flops, one neegs to rtoxc gye Cholesky decompositions of the covariance maedivex of $B_1Z,\ldots,B_q Z$, which woilf bteatly increasg the mzjody requirements of the aldoritym. For dewlinb with truly massive data swts, our long-nerm plan is to combinz the randomnzed akproacn studied here with a cumpoaite likelijood by uakng the randomizdd slcorithms to compute the grwdient of (\[com'osite\]), tfus kaking it possibpe to consider $A_j$’s and $B_u$’s of ldrger rank than would be feasible if one ied to do exacj cdlcllations. Szction \[sec2\] provides a bound on thg efficieucy of the estimzting exuations bafed on the ap[toximate likenihood rqlatuve ro the Wksher informatoon matriq. Uhe bound us in terms of the covsition number oy uhw true covariamce maerpx pf dhe observathons ana showr that if tme zovatiance matrix is weln cohditioned, $N$ does npt need to be very large to obtsin nearly optimal estikatmng eqiatyons. Section \[sec3\] shows how one can get lmpvoved estimatyng cquajions by chposing the $U_j$’s in (\[ascore\]) based on a desmgn related
sum_{j=1}^q \log f_{j,\theta}(A_j Z\mid B_j Z),$$ where the Gaussian density $A_j Z$ given @vecchia @steinchiwelty, the rank $B_j$ will generally larger than that of $A_j$, in case the main computation in obtaining (\[composite\]) is finding Cholesky decompositions of the matrices of $B_1Z,\ldots, B_q Z$. For example, @vecchia just lets $A_j Z$ be $j$th of and Z$ some subset of $Z_1,\ldots,Z_{j-1}$. If $m$ is the largest of these subsets, then the storage for this computation are $O(m^2)$ rather than $O(n^2)$. to increasing the number $U_j$’s in the randomized algorithm here, approach can updated obtain better approximation of likelihood by increasing the size of the subset of $Z_1,\ldots,Z_{j-1}$ to condition on when computing the conditional of $Z_j$. this approach be from perspective of flops, to store the Cholesky decompositions of of $B_1Z,\ldots,B_q Z$, which would greatly increase the requirements of algorithm. For dealing with truly massive sets, our long-term plan is to combine the approach studied here with a composite likelihood by using the randomized algorithms to compute the (\[composite\]), thus making it to consider $A_j$’s $B_j$’s larger than be feasible one had to do exact calculations. Section \[sec2\] provides a bound the efficiency of the estimating equations based on the approximate to Fisher information matrix. bound is in terms the number of the true of observations if covariance is well conditioned, $N$ not need to be very to obtain nearly optimal how one can get improved estimating equations by the $U_j$’s in (\[ascore\]) based on a related
sum_{j=1}^q \log f_{j,\theta}(A_j Z\mid B_j Z),$$ wHere $f_{j,\thetA}$ is thE coNdiTiOnal gausSian density of $A_J z$ givEn $B_j Z$. As proposed by @vecchIa and @StEInchIWeLty, thE rank of $b_J$ wILL geNeRaLly Be LArGer thAn tHat of $A_j$, In which casE thE mAin computatiON iN obtaining (\[ComPosite\]) is findIng cholesKy DecOMposiTioNs of tHe covaRIance mAtrices of $b_1Z,\LDots, B_q z$. for examPLE, @vEcchIa just lets $A_j Z$ be thE $J$tH Component of $Z$ anD $B_j Z$ soMe SUbSET of $z_1,\ldOts,Z_{j-1}$. If $m$ is ThE largESt of theSE sUBSEts, THen the storage RequirementS For This coMpUtaTIon are $o(m^2)$ ratHeR ThaN $O(n^2)$. ComparabLe to IncreasinG the nuMBer of $U_j$’S In the raNdomizEd aLgoRithM UsEd HerE, tHIs aPPrOacH Can Be updateD tO oBtain A betTER APproXimAtioN of thE likelihood by IncReasINg tHe sizE of thE subSeT of $Z_1,\lDots,Z_{j-1}$ To conDiTion on when compuTing The conditIonAl DenSiTy of $Z_J$. howeveR, foR thIs approAch to be EFfiCiENT FrOm the perspective of FlOPS, oNe needs tO store THe chOLesky decOmPosItioNS Of the CovaRIaNce matriCes of $B_1z,\LdOtS,B_q Z$, whiCh Would gReAtlY inCreasE The mEmory rEquiremeNts of THe algorithm. For DEaling with truLY mASSiVE datA seTs, our long-teRm plAN is tO comBInE thE RandoMized ApPRoACh studied here with a cOmPosite LikelIhood by using tHe randomizED ALgorithmS to cOMpUTe the gradient oF (\[compOsite\]), thus mAKing it poSsiblE to consiDer $A_j$’s and $b_J$’S of largeR raNk tHan WouLD Be Feasible if one HAD to dO eXact calCulAtions. SEctIon \[Sec2\] ProViDes a bound On the effIcIeNcY oF thE estiMAting equAtIonS bAseD on thE ApproxImate LikeLiHoOD reLative tO ThE fIsheR iNfOrmaTioN mAtrix. the bOUnd Is in terMs of the coNdiTIon nUmBeR of the tRue covariance MaTrix of the oBsErvAtions AND shows thAt if the covariance matrix IS well coNdiTioneD, $N$ doEs not need To bE very lArgE To obtaIn nearLy optImAl eSTImatiNG EqUatIoNs. Section \[sEC3\] ShoWs how OnE can Get imprOved estimating equaTIonS by choosing thE $U_j$’S in (\[aSCOrE\]) baSEd ON a dEsIGn rELAted
sum_{j=1}^q \log f_{j,\the ta}(A_j Z\ mid B _jZ), $$ whe re $ f_{j,\theta}$i s th e conditional Gaussian dens it y of$ A_ j Z$given $ B _j Z $.As p rop os e dby @v ecc hia and @steinchi wel ty , the rank o f $ B_j$ willgen erally be la rge r than t hat of $A _j$ , inwhichc ase th e main co mp u tation in obta i n in g (\ [composite\]) isf in d ing Cholesky d ecompo si t io n s of th e covarian ce matr i ces of$ B_ 1 Z , \ld o ts, B_q Z$. F or example, @ve cchiaju stl ets $A _j Z$ b e th e $j$th com pone nt of $Z$ and $ B _j Z$ s o me subs et of$Z_ 1,\ ldot s ,Z _{ j-1 }$ . If $m $ i s th e larges tof thes e su b s e t s, t hen the stor age requireme nts for thi s com putat ionar e $O( m^2)$rathe rthan $O(n^2)$.Comp arable to in cr eas in g the number of $U _j$’s i n the r a ndo mi z e d a lgorithm used here ,t h is approac h canb eup d ated toob tai n ab e tterappr o xi mation o f thel ik el ihood b yincrea si ngthe size of t he sub set of $ Z_1,\ l dots,Z_{j-1}$t o condition o n w h e nc ompu tin g the condi tion a l de nsit y o f $ Z _j$.Howev er , f o r this approach tobe effic ientfrom the pers pective of f l ops, one nee d st o store the Ch olesk y decompos i tions of thecovarian ce matric e s of $B_1 Z,\ ldo ts, B_q Z $, which wouldg r eatl yincreas e t he memo ryreq uir eme nt s of thealgorith m. F or d eal ing w i th truly m ass iv e d ata s e ts, ou r lon g-te rm p l anis to c o mb i n e th era ndom ize dappro achs tud ied her e with acom p osit eli kelihoo d by using th erandomized a lgo rithms t o comput e the gradient of (\[co m posite\ ]), thus mak ing it po ssi ble to co n sider$A_j$’ s and $ B_j $ ’ s ofl a rg erra nk than wo u l d b e fea si bleif onehad to do exact ca l cul ations. Sect ion \[s e c 2\ ] p r ov i des a bou n d on the efficie ncy of the e s ti mating equ a tio ns basedon theappro x imate l ikelihood relative t o th e Fis her inform ation ma trix. The bound is in t erm s of t he co nditi on num b erof th e true c ovaria nce m at rix of t he observations and sho ws tha t ifthe covarian cem atr ix is wel l co nditioned, $N $ d oes n otn eed t o be ve ryl argeto o b tain near l yopt i m al estimating e q uat ions. Se c tion \ [sec 3\] shows how one can get improv ed e s t ima tin g equ at ions by choosi ngth e $U_j$’sin (\[ascore\ ]) based o n a de sign r elated
sum_{j=1}^q \log_f_{j,\theta}(A_j Z\mid B_j_Z),$$ where $f_{j,\theta}$ is_the conditional_Gaussian_density of_$A_j_Z$ given $B_j_Z$. As proposed_by @vecchia and @steinchiwelty,_the rank of_$B_j$_will generally be larger than that of $A_j$, in which case the main computation_in_obtaining (\[composite\])_is_finding_Cholesky decompositions of the covariance_matrices of $B_1Z,\ldots, B_q Z$. For_example, @vecchia_just lets $A_j Z$ be the $j$th component_of_$Z$ and $B_j_Z$ some subset of $Z_1,\ldots,Z_{j-1}$. If $m$ is the_largest of these subsets, then the_storage requirements for_this_computation_are $O(m^2)$ rather than_$O(n^2)$. Comparable to increasing the number_of $U_j$’s in the randomized algorithm_used here, this approach can be updated_to obtain a better approximation of_the likelihood by increasing the_size of_the subset of $Z_1,\ldots,Z_{j-1}$ to_condition on when_computing the_conditional density of_$Z_j$. However, for this approach to_be efficient from_the perspective of flops, one needs_to_store the Cholesky_decompositions_of_the covariance_matrices of $B_1Z,\ldots,B_q_Z$,_which would_greatly_increase the memory requirements of the_algorithm._For dealing with truly massive data sets,_our long-term plan is_to_combine the randomized approach_studied here with a composite_likelihood by using the randomized algorithms_to compute_the gradient_of (\[composite\]), thus making it possible to consider $A_j$’s and $B_j$’s_of larger rank than would be_feasible if one had_to do_exact_calculations. Section \[sec2\] provides_a_bound on_the efficiency of the estimating equations based_on the_approximate likelihood relative to the Fisher_information matrix. The bound_is_in terms of the condition number_of the true covariance matrix of_the observations and shows that_if_the_covariance matrix is well conditioned,_$N$ does not need to be_very large to_obtain nearly optimal estimating equations. Section \[sec3\]_shows_how one can get improved estimating_equations_by choosing the $U_j$’s in (\[ascore\])_based_on_a design related
, Rev. Sci. Instrum. [**66**]{}, 1394 (1995). J.A. Carlisle, L.J. Terminello, E.A. Hudson, R.C.C. Perera, J.H. Underwood, T.A. Callcott, J.J. Jia, D.L. Ederer, F.J. Himpsel, and M.G. Samant, Appl. Phys. Lett. [**67**]{}, 34 (1995). R.C.C. Perera, C.H. Zhang, T.A. Callcott and D.L. Ederer, J. Appl. Phys. [**66**]{}, 3676 (1989); P.O. Nilsson, J. Kanski, J.V. Thordson, T.G. Andersson, J. Nordgren, J. Guo and M. Magnuson, Phys. Rev. [**B52**]{}, R8643 (1995). J.J. Jia, T.A. Callcott, W.L. O’Brien, Q.Y. Dong, D.R. Muller, D.L. Ederer, Z. Tan, and J.I. Budnick, Phys. Rev. [**B46**]{}, 9446 (1992). See, for instance, J.-E. Rubensson, D. Mueller, R. Shuker, D.L. Ederer, C.H. Zhang, J. Jia, and T.A. Callcott, Phys. Rev. Lett [**64**]{}, 1047 (1990). H. von Känel, K.A. Mäder, E. Müller, N. Onda, and H. Sirringhaus, Phys. Rev. [**B45**]{}, 13807 (1992). E.E. Fullerton and S.D. Bader, Phys. Rev. [**B**]{}, in press. R.P. Michel and A. Chaiken, in preparation. --- abstract: 'Jupiter-mass planets with large semi-major axes ($a > 1.0$ AU) occur at a higher rate around evolved intermediate mass stars. There is a pronounced paucity of close-in ($a < 0.6$ AU), intermediate period ($5 < P < 100$ days), low-mass ($M_{\rm planet} < 0.7M_{\
, Rev. Sci. Instrum. [ * * 66 * * ] { }, 1394 (1995). J.A. Carlisle, L.J. Terminello, E.A. Hudson, R.C.C. Perera, J.H. Underwood, T.A. Callcott, J.J. Jia, D.L. Ederer, F.J. Himpsel, and M.G. Samant, Appl. Phys. Lett. [ * * 67 * * ] { }, 34 (1995). R.C.C. Perera, C.H. Zhang, T.A. Callcott and D.L. Ederer, J. Appl. Phys. [ * * 66 * * ] { }, 3676 (1989); P.O. Nilsson, J. Kanski, J.V. Thordson, T.G. Andersson, J. Nordgren, J. Guo and M. Magnuson, Phys. Rev. [ * * B52 * * ] { }, R8643 (1995). J.J. Jia, T.A. Callcott, W.L. O’Brien, Q.Y. Dong, D.R. Muller, D.L. Ederer, Z. Tan, and J.I. Budnick, Phys. Rev. [ * * B46 * * ] { }, 9446 (1992). See, for instance, J.-E. Rubensson, D. Mueller, R. Shuker, D.L. Ederer, C.H. Zhang, J. Jia, and T.A. Callcott, Phys. Rev. Lett [ * * 64 * * ] { }, 1047 (1990). H. von Känel, K.A. Mäder, E. Müller, N. Onda, and H. Sirringhaus, Phys. Rev. [ * * B45 * * ] { }, 13807 (1992). E.E. Fullerton and S.D. Bader, Phys. Rev. [ * * B * * ] { }, in press. R.P. Michel and A. Chaiken, in readiness. --- abstract:' Jupiter - bulk planets with large semi - major axes ($ a > 1.0 $ AU) happen at a higher rate around evolve intermediate mass star. There is a pronounced dearth of close - in ($ a < 0.6 $ AU), intermediate period ($ 5 < phosphorus < 100 $ days), low - bulk ($ M_{\rm planet } < 0.7M_{\
, Reg. Sci. Instrum. [**66**]{}, 1394 (1995). J.A. Carlirle, L.J. Terminello, E.A. Hndson, R.D.C. Pererx, J.H. Underwood, T.A. Callcott, J.O. Jiq, D.L. Tberer, F.J. Himpsel, and M.G. Samann, Appl. Phts. Lttt. [**67**]{}, 34 (1995). R.C.C. Perera, R.G. Zhang, T.A. Campcotc end D.L. Ederer, J. Appl. Phys. [**66**]{}, 3676 (1989); P.O. Nilsson, J. Kcnski, J.V. Thordson, T.G. Andersson, J. Norqgren, J. Gko and M. Magnufon, Khyf. Reb. [**B52**]{}, R8643 (1995). J.J. Jia, T.A. Callcott, W.L. O’Brien, Q.Y. Donj, D.R. Muller, D.L. Rderer, Z. Tan, and J.I. Budnicn, Phjs. Rev. [**B46**]{}, 9446 (1992). See, for lnstance, J.-E. Rubqbsson, D. Mueluer, R. Shuktr, D.L. Ederer, C.H. Zhang, J. Jia, and T.A. Callcott, Phys. Rev. Lett [**64**]{}, 1047 (1990). Y. vlt Känel, K.A. Mäver, E. Iüller, N. Onda, and H. Shrringhsus, Phys. Rev. [**B45**]{}, 13807 (1992). E.X. Fuolerton and S.D. Bader, 'hys. Rev. [**B**]{}, in press. R.K. Michel atd A. Chaiken, in prepaeatiot. --- abvtrazr: 'Jjpiuer-kaas plajeta with larfe semi-majoe axes ($a > 1.0$ AU) occur ae a higher rats arougd evolved intermediate mass stars. There ps a pronounced paucity of xlose-in ($a < 0.6$ AU), intermgdiate peryod ($5 < P < 100$ days), low-mass ($M_{\rm planet} < 0.7M_{\
, Rev. Sci. Instrum. [**66**]{}, 1394 (1995). L.J. E.A. Hudson, Perera, J.H. Underwood, Ederer, Himpsel, and M.G. Appl. Phys. Lett. 34 (1995). R.C.C. Perera, C.H. Zhang, Callcott and D.L. Ederer, J. Appl. Phys. [**66**]{}, 3676 (1989); P.O. Nilsson, J. J.V. Thordson, T.G. Andersson, J. Nordgren, J. Guo and M. Magnuson, Phys. Rev. R8643 J.J. T.A. W.L. O’Brien, Q.Y. Dong, D.R. Muller, D.L. Ederer, Z. Tan, and J.I. Budnick, Phys. Rev. [**B46**]{}, (1992). See, for instance, J.-E. Rubensson, D. Mueller, Shuker, D.L. Ederer, C.H. J. Jia, and T.A. Callcott, Rev. [**64**]{}, 1047 H. Känel, Mäder, E. Müller, Onda, and H. Sirringhaus, Phys. Rev. [**B45**]{}, 13807 (1992). E.E. Fullerton and S.D. Bader, Phys. Rev. [**B**]{}, press. R.P. A. Chaiken, preparation. abstract: planets with large ($a > 1.0$ AU) occur at around evolved intermediate mass stars. There is a paucity of ($a < 0.6$ AU), intermediate period < P < 100$ days), low-mass ($M_{\rm planet} 0.7M_{\
, Rev. Sci. Instrum. [**66**]{}, 1394 (1995). J.A. Carlisle, L.J. terminello, e.A. HudSon, r.C.C. peRera, j.H. UnDerwood, T.A. CallcOTt, J.J. jia, D.L. Ederer, F.J. Himpsel, anD M.G. SaMaNT, AppL. phYs. LetT. [**67**]{}, 34 (1995). R.C.C. PerERa, c.h. zhaNg, t.A. calLcOTt And D.L. edeRer, J. AppL. Phys. [**66**]{}, 3676 (1989); P.O. NilSsoN, J. kanski, J.V. ThorDSoN, T.G. AnderssOn, J. nordgren, J. Guo And m. MagnuSoN, PhYS. Rev. [**B52**]{}, r8643 (1995). J.J. jia, T.A. callcoTT, W.L. O’BrIen, Q.Y. Dong, d.R. mUller, D.l. ederer, Z. tAN, aNd J.I. budnick, Phys. Rev. [**B46**]{}, 9446 (1992). SeE, FoR Instance, J.-E. RubeNsson, D. muELlER, r. ShUkeR, D.L. Ederer, C.h. ZHang, J. jIa, and T.A. cAlLCOTt, PHYs. Rev. Lett [**64**]{}, 1047 (1990). H. von känel, K.A. MädeR, e. MüLler, N. ONdA, anD h. SirriNghauS, PHYs. REv. [**B45**]{}, 13807 (1992). E.E. FullerTon aNd S.D. Bader, phys. ReV. [**b**]{}, in presS. r.P. MicheL and A. CHaiKen, In prEPaRaTioN. --- aBStrACt: 'jupITer-Mass planEtS wIth laRge sEMI-MAjor AxeS ($a > 1.0$ AU) Occur At a higher rate AroUnd eVOlvEd intErmedIate MaSs staRs. TherE is a pRoNounced paucity oF cloSe-in ($a < 0.6$ AU), inTerMeDiaTe PerioD ($5 < p < 100$ days), lOw-mAss ($m_{\rm planEt} < 0.7M_{\
, Rev. Sci. Instrum. [**66 **]{}, 139 4 (19 95) . J. A. C arli sle, L.J. Term i nell o, E.A. Hudson, R.C.C. Pere ra , J.H . U nderw ood, T. A .C a llc ot t, J. J. Ji a, D. L.Ederer, F.J. Himp sel ,and M.G. Sam a nt , Appl. Ph ys. Lett. [**67 **] {}, 34 ( 199 5 ). R .C. C. Pe rera,C .H. Zh ang, T.A. C a llcott and D.L . Ed erer , J. Appl. Phys.[ ** 6 6**]{}, 3676 ( 1989); P . O. N ils son , J. Kansk i, J.V. Thordso n ,T . G . A n dersson, J. N ordgren, J. Guo and M .Mag n uson,Phys. R e v.[**B52**]{} , R8 643 (1995 ). J. J . Jia,T .A. Cal lcott, W. L.O’Br i en ,Q.Y .D ong , D .R. Mul ler, D.L .Ed erer, Z.T a n , and J. I. B udnic k, Phys. Rev. [* *B46 * *]{ }, 94 46 (1 992) . See, for i nstan ce , J.-E. Rubenss on,D. Muelle r,R. Sh uk er, D . L. Ede rer , C .H. Zha ng, J.J ia, a n d T. A. Callcott, Phys. R e v .Lett [** 64**]{ } ,10 4 7 (1990) . H. von K änel, K.A . M äder, E. Mülle r ,N. Onda,an d H. S ir rin gha us, P h ys.Rev. [ **B45**] {}, 1 3 807 (1992). E . E. Fullertona nd S .D . Bad er, Phys. Rev. [** B **]{ }, i n p res s . R. P. Mi ch e la nd A. Chaiken, in p re parati on. --- abstract: 'Jupiter- m a s s planet s wi t hl arge semi-majo r axe s ($a > 1. 0 $ AU) oc cur a t a high er rate a r o und evol ved in ter med i a te mass stars.T h ereis a pron oun ced pau cit y o f c los e- in ($a <0.6$ AU) ,in te rm edi ate p e riod ($5 < P<100 $ day s ), low -mass ($M _{ \r m pl anet} < 0. 7 M _{\
, Rev._Sci. Instrum._[**66**]{}, 1394 (1995). J.A. Carlisle,_L.J. Terminello,_E.A._Hudson, R.C.C._Perera,_J.H. Underwood, T.A._Callcott, J.J. Jia,_D.L. Ederer, F.J. Himpsel,_and M.G. Samant,_Appl._Phys. Lett. [**67**]{}, 34 (1995). R.C.C. Perera, C.H. Zhang, T.A. Callcott and D.L. Ederer, J._Appl._Phys. [**66**]{},_3676_(1989);_P.O. Nilsson, J. Kanski, J.V._Thordson, T.G. Andersson, J. Nordgren,_J. Guo_and M. Magnuson, Phys. Rev. [**B52**]{}, R8643 (1995). J.J._Jia,_T.A. Callcott, W.L._O’Brien, Q.Y. Dong, D.R. Muller, D.L. Ederer, Z. Tan,_and J.I. Budnick, Phys. Rev. [**B46**]{},_9446 (1992). See, for_instance,_J.-E._Rubensson, D. Mueller, R._Shuker, D.L. Ederer, C.H. Zhang, J._Jia, and T.A. Callcott, Phys. Rev._Lett [**64**]{}, 1047 (1990). H. von Känel, K.A._Mäder, E. Müller, N. Onda, and_H. Sirringhaus, Phys. Rev. [**B45**]{},_13807 (1992). E.E._Fullerton and S.D. Bader, Phys._Rev. [**B**]{}, in_press. R.P. Michel_and A. Chaiken,_in preparation. --- abstract: 'Jupiter-mass planets with_large semi-major axes_($a > 1.0$ AU) occur at_a_higher rate around_evolved_intermediate_mass stars._There is a_pronounced_paucity of_close-in_($a < 0.6$ AU), intermediate period_($5_< P < 100$ days), low-mass ($M_{\rm_planet} < 0.7M_{\
c}}, \widetilde{\mathcal{\mathbf{P}}^{c}}^{\prime })$ is an element of H$^{1}(\mathcal{\mathbf{B}}^c ,\mathbf{Z}_{2}).$   By definition, we have $$\delta _{ij}(x)=\gamma _{ij}(x)\Psi (\gamma _{ij}^{\prime }(x)^{-1}) \:,$$ and $$\rho (\gamma _{ij})=\rho (\gamma _{ij}^{\prime })=g_{ij} \:.$$ This implies that $$\begin{aligned} \rho (\delta _{ij}) &=&1 \\ &\Longrightarrow &\delta _{ij}(x)\in \mathbf{Z}_{2}\end{aligned}$$ i.e., $\delta _{ij}$ is in the center of $\Gamma $ and $$\begin{aligned} (d\delta )(ijk) &=&\delta _{jk}\delta _{ik}^{-1}\delta _{ij}\nonumber \\ &=&\gamma _{ji}\Psi (\gamma _{jk}^{\prime -1})(\gamma _{ik}\Psi (\gamma _{ik}^{\prime -1}))\gamma _{ij}\Psi (\gamma _{ij}^{\prime -1}) \nonumber \\ &=&\gamma _{jk}\Psi (\gamma _{jk}^{\prime -1})(\Psi (\gamma _{ik}^{\prime })\gamma _{ik}^{-1})\gamma _{ji}\Psi (\gamma _{ij}^{\prime -1})\nonumber \\ &=&\gamma _{jk}(\gamma _{ij}\Psi (\gamma _{ij}^{\prime -1}))\Psi (\gamma _{jk}^{\prime -1})(\Psi (\gamma _{ik}^{\prime })\gamma _{ik}^{-1})\text{ since }\delta _{ij}\in C(\Gamma )\nonumber \\ &=&\gamma _{jk}\gamma _{ij}\Psi (\gamma _{ij}^{\prime -1}\gamma _{jk}^{\prime -1}\gamma _{ik}^{\prime })\gamma _{ik}^{-1} \nonumber \\ &=&1.\end{aligned}$$ Hence $\delta _{ij}\in $ $H^{1}(\mathcal{\mathbf{B}},\mathbf{Z}_{2})$ . A similar proof works for $ \overline{\delta }_{ij}$.  The difference class can be used to define an equivalence relation among the $\Gamma -$ structures.  In fact, we have $\widetilde{\mathcal{\mathbf{P}}}\simeq _{W}\widetilde{\mathcal{\mathbf{P}}}^{\prime }$ iff $\overline{\delta }(% \widetilde{\mathcal{\
c } }, \widetilde{\mathcal{\mathbf{P}}^{c}}^{\prime }) $ is an element of H$^{1}(\mathcal{\mathbf{B}}^c , \mathbf{Z}_{2}).$   By definition, we have $ $ \delta _ { ij}(x)=\gamma _ { ij}(x)\Psi (\gamma _ { ij}^{\prime } (x)^{-1 }) \:,$$ and $ $ \rho (\gamma _ { ij})=\rho (\gamma _ { ij}^{\prime }) = g_{ij } \:.$$ This implies that $ $ \begin{aligned } \rho (\delta _ { ij }) & = & 1 \\ & \Longrightarrow & \delta _ { ij}(x)\in \mathbf{Z}_{2}\end{aligned}$$ i.e., $ \delta _ { ij}$   is in the center of $ \Gamma $   and $ $ \begin{aligned } (d\delta) (ijk) & = & \delta _ { jk}\delta _ { ik}^{-1}\delta _ { ij}\nonumber \\ & = & \gamma _ { ji}\Psi (\gamma _ { jk}^{\prime -1})(\gamma _ { ik}\Psi (\gamma _ { ik}^{\prime -1}))\gamma _ { ij}\Psi (\gamma _ { ij}^{\prime -1 }) \nonumber \\ & = & \gamma _ { jk}\Psi (\gamma _ { jk}^{\prime -1})(\Psi (\gamma _ { ik}^{\prime }) \gamma _ { ik}^{-1})\gamma _ { ji}\Psi (\gamma _ { ij}^{\prime -1})\nonumber \\ & = & \gamma _ { jk}(\gamma _ { ij}\Psi (\gamma _ { ij}^{\prime -1}))\Psi (\gamma _ { jk}^{\prime -1})(\Psi (\gamma _ { ik}^{\prime }) \gamma _ { ik}^{-1})\text { since } \delta _ { ij}\in C(\Gamma) \nonumber \\ & = & \gamma _ { jk}\gamma _ { ij}\Psi (\gamma _ { ij}^{\prime -1}\gamma _ { jk}^{\prime -1}\gamma _ { ik}^{\prime }) \gamma _ { ik}^{-1 } \nonumber \\ & = & 1.\end{aligned}$$ therefore $ \delta _ { ij}\in $ $ H^{1}(\mathcal{\mathbf{B}},\mathbf{Z}_{2})$  . A exchangeable proof works for $ \overline{\delta } _ { ij}$.   The remainder course can be used to define an equivalence sexual intercourse among the $ \Gamma -$ structures.   In fact, we have $ \widetilde{\mathcal{\mathbf{P}}}\simeq _ { W}\widetilde{\mathcal{\mathbf{P}}}^{\prime } $ iff $ \overline{\delta } (% \widetilde{\mathcal{\
c}}, \wifetilde{\mathcal{\mathbf{P}}^{c}}^{\prlme })$ is an elemeur of H$^{1}(\kathcam{\mathbf{B}}^z ,\mathbf{Z}_{2}).$   By definition, we heve $$\eelta _{ij}(x)=\gamma _{ij}(x)\Psi (\gamma _{ij}^{\prime }(q)^{-1}) \:,$$ and $$\rhi (\ganna _{ij})=\rho (\gejma _{ij}^{\pvnme })=g_{jm} \:.$$ Thns implies that $$\negin{aligneg} \rho (\delta _{ij}) &=&1 \\ &\Lunyrightarrow &\delta _{ij}(x)\in \mathbf{Z}_{2}\end{alygned}$$ i.r., $\felta _{ij}$ is in jhe ctntqr or $\Gamma $ and $$\begin{aligned} (d\delta )(ijk) &=&\delta _{ok}\delta _{ik}^{-1}\delta _{ij}\nonumber \\ &=&\gamma _{ji}\Psi (\galma _{uk}^{\prime -1})(\gamma _{ik}\Psl (\gamma _{ik}^{\prume -1}))\dqmma _{ij}\Psi (\gxmma _{ij}^{\prime -1}) \nonumber \\ &=&\gamma _{jk}\Psi (\gamma _{jk}^{\prime -1})(\Psi (\eamma _{ik}^{\prime })\ganmq _{in}^{-1})\camma _{ji}\Psi (\gammw _{ij}^{\prime -1})\noklmber \\ &=&\gdmma _{jk}(\bamma _{ij}\Psi (\gakma _{ij}^{\prime -1}))\Psi (\gamma _{jk}^{\primx -1})(\Psi (\gamma _{ik}^{\prime })\damma _{ik}^{-1})\taxc{ since }\delta _{ij}\in C(\Ganmq )\nonomber \\ &=&\gamoq _{jy}\gajme _{in}\Psi (\gwmme _{ij}^{\prime -1}\gzmma _{jk}^{\prime -1}\gamma _{ik}^{\prime })\gamms _{yj}^{-1} \nonumber \\ &=&1.\ens{alignqd}$$ Hence $\delta _{ij}\in $ $H^{1}(\mathcal{\mathbf{B}},\mathbx{Z}_{2})$ . Z similar proof works fir $ \overline{\delta }_{ij}$.  Tje differqnce class can be used to define an equivalence ralatikv anokg tfw $\Hamma -$ structures.  In fact, we have $\widetilde{\maegcsl{\kathbf{P}}}\simeq _{W}\cidetilde{\mathcal{\kahhns{P}}}^{\prime }$ iff $\uverlius{\dslta }(% \widetilde{\mathfal{\
c}}, \widetilde{\mathcal{\mathbf{P}}^{c}}^{\prime })$ is an element of By we have _{ij}(x)=\gamma _{ij}(x)\Psi (\gamma (\gamma (\gamma _{ij}^{\prime })=g_{ij} This implies that \rho (\delta _{ij}) &=&1 \\ &\Longrightarrow _{ij}(x)\in \mathbf{Z}_{2}\end{aligned}$$ i.e., $\delta _{ij}$ is in the center of $\Gamma $ and (d\delta )(ijk) &=&\delta _{jk}\delta _{ik}^{-1}\delta _{ij}\nonumber \\ &=&\gamma _{ji}\Psi (\gamma _{jk}^{\prime -1})(\gamma _{ik}\Psi _{ik}^{\prime _{ij}\Psi _{ij}^{\prime \nonumber \\ &=&\gamma _{jk}\Psi (\gamma _{jk}^{\prime -1})(\Psi (\gamma _{ik}^{\prime })\gamma _{ik}^{-1})\gamma _{ji}\Psi (\gamma _{ij}^{\prime -1})\nonumber \\ _{jk}(\gamma _{ij}\Psi (\gamma _{ij}^{\prime -1}))\Psi (\gamma _{jk}^{\prime -1})(\Psi _{ik}^{\prime })\gamma _{ik}^{-1})\text{ since _{ij}\in C(\Gamma )\nonumber \\ &=&\gamma _{ij}\Psi _{ij}^{\prime -1}\gamma -1}\gamma })\gamma \nonumber \\ &=&1.\end{aligned}$$ $\delta _{ij}\in $ $H^{1}(\mathcal{\mathbf{B}},\mathbf{Z}_{2})$ . A similar proof works for $ \overline{\delta }_{ij}$. The difference class can used to equivalence relation the -$ In fact, we _{W}\widetilde{\mathcal{\mathbf{P}}}^{\prime }$ iff $\overline{\delta }(% \widetilde{\mathcal{\
c}}, \widetilde{\mathcal{\mathbf{P}}^{c}}^{\Prime })$ is an eLemenT of h$^{1}(\maThCal{\mAthbF{B}}^c ,\mathbf{Z}_{2}).$   By deFInitIon, we have $$\delta _{ij}(x)=\gamma _{Ij}(x)\PsI (\gAMma _{iJ}^{\PrIme }(x)^{-1}) \:,$$ aNd $$\rho (\gaMMa _{IJ})=\Rho (\GaMmA _{ij}^{\PrIMe })=G_{ij} \:.$$ ThIs iMplies tHat $$\begin{alIgnEd} \Rho (\delta _{ij}) &=&1 \\ &\LoNGrIghtarrow &\dEltA _{ij}(x)\in \mathbf{z}_{2}\enD{alignEd}$$ I.e., $\dELta _{ij}$ Is iN the cEnter oF $\gamma $ aNd $$\begin{alIgNEd} (d\delTA )(ijk) &=&\delTA _{Jk}\DeltA _{ik}^{-1}\delta _{ij}\nonumbeR \\ &=&\GaMMa _{ji}\Psi (\gamma _{jk}^{\Prime -1})(\gAmMA _{iK}\pSi (\gAmmA _{ik}^{\prime -1}))\gaMmA _{ij}\PsI (\Gamma _{ij}^{\PRiME -1}) \NOnuMBer \\ &=&\gamma _{jk}\Psi (\Gamma _{jk}^{\primE -1})(\psi (\Gamma _{iK}^{\pRimE })\Gamma _{iK}^{-1})\gammA _{jI}\psi (\Gamma _{ij}^{\primE -1})\nonUmber \\ &=&\gammA _{jk}(\gamMA _{ij}\Psi (\gAMma _{ij}^{\prIme -1}))\Psi (\GamMa _{jK}^{\priME -1})(\PSi (\GamMa _{IK}^{\prIMe })\GamMA _{ik}^{-1})\Text{ sincE }\dElTa _{ij}\iN C(\GaMMA )\NOnumBer \\ &=&\GammA _{jk}\gaMma _{ij}\Psi (\gamma _{Ij}^{\pRime -1}\GAmmA _{jk}^{\prIme -1}\gaMma _{iK}^{\pRime })\gAmma _{ik}^{-1} \NonumBeR \\ &=&1.\end{aligned}$$ HencE $\delTa _{ij}\in $ $H^{1}(\maThcAl{\MatHbF{B}},\matHBf{Z}_{2})$ . A siMilAr pRoof worKs for $ \ovERliNe{\DELTa }_{Ij}$.  The difference claSs CAN bE used to dEfine aN EqUiVAlence reLaTioN amoNG The $\GaMma -$ sTRuCtures.  In Fact, we HAvE $\wIdetildE{\mAthcal{\MaThbF{P}}}\sImeq _{W}\WIdetIlde{\maThcal{\matHbf{P}}}^{\pRIme }$ iff $\overline{\DElta }(% \widetilde{\MAtHCAl{\
c}}, \widetilde{\mathcal{\ mathbf{P}} ^{c}} ^{\ pri me })$ isan element ofH $^{1 }(\mathcal{\mathbf{B}} ^c ,\ ma t hbf{ Z }_ {2}). $   Byd ef i n iti on ,weha v e$$\de lta _{ij}( x)=\gamma_{i j} (x)\Psi (\ga m ma _{ij}^{\p rim e }(x)^{-1}) \: ,$$ an d$$\ r ho (\ gam ma _{ ij})=\ r ho (\g amma _{ij }^ { \prime })=g_{i j } \ :.$$ This implies tha t $ $ \begin{aligned } \rho ( \ de l t a _ {ij }) &=&1 \\ & \Long r ightarr o w& \ d elt a _{ij}(x)\in\mathbf{Z}_ { 2}\ end{al ig ned } $$ i.e ., $\ de l ta_{ij}$ is i n th e centerof $\G a mma $ a n d $$\be gin{al ign ed} (d\ d el ta )( ij k ) & = &\ del t a _ {jk}\del ta _ {ik}^ {-1} \ d e l ta _ {ij }\no numbe r \\ &=&\gamm a _ {ji} \ Psi (\ga mma _ {jk} ^{ \prim e -1}) (\gam ma _{ik}\Psi (\ga mma_{ik}^{\p rim e-1} )) \gamm a _{ij} \Ps i ( \gamma_{ij}^{ \ pri me - 1 }) \nonumber \\ &=&\ ga m m a_{jk}\Ps i (\ga m ma _ { jk}^{\pr im e - 1})( \ P si (\ gamm a _ {ik}^{\p rime } ) \g am ma _{ik }^ {-1})\ ga mma _{ ji}\P s i (\ gamma_{ij}^{\ prime -1})\nonumber\ \ &=&\gamma _ { jk } ( \g a mma_{i j}\Psi (\ga mma_ {ij} ^{\p r im e - 1 }))\P si (\ ga m ma _{jk}^{\prime -1})( \P si (\g amma_{ik}^{\prime })\gamma_ { i k}^{-1}) \tex t {s ince }\delta _ {ij}\ in C(\Gamm a )\nonum ber \ \ &=&\ga mma _{jk} \ g amma _{i j}\ Psi (\ gam m a _ {ij}^{\prime- 1 }\ga mm a _{jk} ^{\ prime - 1}\ gam ma_{i k} ^{\prime})\gamma _ {i k} ^{ -1} \non u mber \\&= &1. \e nd{ align e d}$$ H ence$\de lt a_ {ij }\in $$ H^ { 1 }(\m at hc al{\ mat hb f{B}} ,\ma t hbf {Z}_{2} )$ . A si mil a r pr oo fworks f or $ \overlin e{ \delta }_{ ij }$.  Thed i fference class can be used to d e fine an eq uival ence relation am ong th e $ \ Gamma-$ str uctur es . I n fact , we ha ve $\widetil d e {\m athca l{ \mat hbf{P}} }\simeq _{W}\widet i lde {\mathcal{\ma thb f{P} } } ^{ \pr i me }$if f $\ o v erline{\delta } (% \wideti ld e {\ mathcal{\
c}}, \widetilde{\mathcal{\mathbf{P}}^{c}}^{\prime })$_is an_element of H$^{1}(\mathcal{\mathbf{B}}^c ,\mathbf{Z}_{2}).$  _By definition,_we_have $$\delta__{ij}(x)=\gamma__{ij}(x)\Psi (\gamma _{ij}^{\prime_}(x)^{-1}) \:,$$ and_$$\rho (\gamma _{ij})=\rho (\gamma__{ij}^{\prime })=g_{ij} \:.$$_This_implies that $$\begin{aligned} \rho (\delta _{ij}) &=&1 \\ &\Longrightarrow &\delta _{ij}(x)\in \mathbf{Z}_{2}\end{aligned}$$ i.e., $\delta _{ij}$ is in_the_center of_$\Gamma_$ and_$$\begin{aligned} (d\delta )(ijk) &=&\delta _{jk}\delta _{ik}^{-1}\delta__{ij}\nonumber \\ &=&\gamma _{ji}\Psi (\gamma _{jk}^{\prime_-1})(\gamma _{ik}\Psi_(\gamma _{ik}^{\prime -1}))\gamma _{ij}\Psi (\gamma _{ij}^{\prime -1}) \nonumber \\ &=&\gamma__{jk}\Psi_(\gamma _{jk}^{\prime -1})(\Psi_(\gamma _{ik}^{\prime })\gamma _{ik}^{-1})\gamma _{ji}\Psi (\gamma _{ij}^{\prime -1})\nonumber \\ &=&\gamma _{jk}(\gamma__{ij}\Psi (\gamma _{ij}^{\prime -1}))\Psi (\gamma _{jk}^{\prime -1})(\Psi_(\gamma _{ik}^{\prime })\gamma__{ik}^{-1})\text{ since_}\delta__{ij}\in C(\Gamma )\nonumber \\ &=&\gamma__{jk}\gamma _{ij}\Psi (\gamma _{ij}^{\prime -1}\gamma _{jk}^{\prime -1}\gamma__{ik}^{\prime })\gamma _{ik}^{-1} \nonumber \\ &=&1.\end{aligned}$$ Hence_$\delta _{ij}\in $ $H^{1}(\mathcal{\mathbf{B}},\mathbf{Z}_{2})$ . A similar proof_works for $ \overline{\delta }_{ij}$.  The difference_class can be used to_define an_equivalence relation among the $\Gamma_-$ structures.  In_fact, we_have $\widetilde{\mathcal{\mathbf{P}}}\simeq _{W}\widetilde{\mathcal{\mathbf{P}}}^{\prime_}$ iff $\overline{\delta }(% \widetilde{\mathcal{\
function transforms under translations by a lattice vector. It will be convenient to define the following $p$ set of wave functions, for $N_{\Phi}\neq 0$: $$\begin{aligned} & &{\tilde \psi}_{l,k_{1},{\vec c}}^{j} = e^{-\frac{\pi\omega}{L_{1}L_{2}}\sum y_{\gamma}^{2}} \prod_{\gamma < \delta} \left(\theta_{1}\left( z_{\gamma} - z_{\delta} \mid \tau \right)\right)^{\frac{1}{\kappa}} \nonumber \\ & &\sum_{K=0}^{M-1}\sum_{n=0}^{p-1}\sum_{{\cal P}\in S_{N_A}} e^{-i2\pi\frac{nj}{\kappa}} e^{i2\pi n\left(q\mid c \mid + pN_{\Phi}c_{\gamma}\right)/p\ell} e^{i2\pi K\left(ns/\kappa+\beta s/2\pi\kappa +k_{2}/N\right)} \nonumber \\ & & \theta\left[\begin{array}{c} -{\hat e}\left(n/\kappa \beta /2\pi\kappa \right) \\ \left(K/N + {\alpha}/2\pi\omega\kappa\right){\hat e} +{\cal P}{\vec c}/\ell \end{array}\right] \left({\vec z},\Omega\right)f_{ln}\left( v\right)\end{aligned}$$ and for $N_{\Phi}=0$ $$\begin{aligned} \psi_{l,c} & = & e^{-\pi\frac{\omega}{L_{1}L_{2}}\sum y_{\gamma}^2} \prod_{\gamma < \delta} \left( \theta_{1}\left( z_{\gamma} - z_{\delta} \mid \tau \right) \right)^{\frac{1}{\kappa}} e^{iXv} \nonumber \\ & &\sum_{n=0}^{p-1} e^{i2\pi cn/p}e^{-i2\pi\frac{jn}{\kappa}} \theta\left[\begin{array}{c} -\left(n/\kappa +\
function transforms under translations by a lattice vector. It will be commodious to specify the following $ p$ set of wave function, for $ N_{\Phi}\neq 0 $: $ $ \begin{aligned } & & { \tilde \psi}_{l, k_{1},{\vec c}}^{j } = e^{-\frac{\pi\omega}{L_{1}L_{2}}\sum y_{\gamma}^{2 } } \prod_{\gamma < \delta } \left(\theta_{1}\left (z_{\gamma } - z_{\delta } \mid \tau \right)\right)^{\frac{1}{\kappa } } \nonumber \\ & & \sum_{K=0}^{M-1}\sum_{n=0}^{p-1}\sum_{{\cal P}\in S_{N_A } } e^{-i2\pi\frac{nj}{\kappa } } e^{i2\pi n\left(q\mid c \mid + pN_{\Phi}c_{\gamma}\right)/p\ell } e^{i2\pi K\left(ns/\kappa+\beta s/2\pi\kappa + k_{2}/N\right) } \nonumber \\ & & \theta\left[\begin{array}{c } -{\hat e}\left(n/\kappa \beta /2\pi\kappa \right) \\ \left(K / N + { \alpha}/2\pi\omega\kappa\right){\hat einsteinium } + { \cal P}{\vec c}/\ell \end{array}\right ] \left({\vec z},\Omega\right)f_{ln}\left (v\right)\end{aligned}$$ and for $ N_{\Phi}=0 $ $ $ \begin{aligned } \psi_{l, c } & = & e^{-\pi\frac{\omega}{L_{1}L_{2}}\sum y_{\gamma}^2 } \prod_{\gamma < \delta } \left (\theta_{1}\left (z_{\gamma } - z_{\delta } \mid \tau \right) \right)^{\frac{1}{\kappa } } e^{iXv } \nonumber \\ & & \sum_{n=0}^{p-1 } e^{i2\pi cn / p}e^{-i2\pi\frac{jn}{\kappa } } \theta\left[\begin{array}{c } -\left(n/\kappa + \
fujction transforms under uranslations by a latticx vectod. It wilu be convenient to define thx foolowibg $p$ set of wave functkons, for $J_{\Phi}\neq 0$: $$\begmn{aligned} & &{\tilde \'ai}_{l,k_{1},{\vec c}}^{j} = e^{-\fdwc{\pi\mnega}{L_{1}L_{2}}\sum y_{\gamka}^{2}} \prod_{\gammd < \delta} \left(\tvega_{1}\peft( z_{\gamma} - z_{\delta} \mid \tau \right)\ridht)^{\frac{1}{\laopa}} \nonumber \\ & &\fum_{K=0}^{K-1}\fum_{n=0}^{l-1}\sum_{{\cal P}\in S_{N_A}} e^{-i2\pi\frac{nj}{\kappa}} e^{i2\pi n\left(q\kid c \mid + pN_{\Lhi}c_{\gamma}\right)/p\ell} e^{i2\pi K\levt(ns/\nappa+\beta s/2\pi\kappa +k_{2}/N\right)} \nonojbew \\ & & \theta\left[\cegin{array}{b} -{\hat e}\left(h/\kappa \beta /2\pi\kappa \right) \\ \left(Y/N + {\akpha}/2\pi\omegq\kqppw\tight){\hat e} +{\cao P}{\veb c}/\ell \end{arran}\gight] \lext({\vec z},\Pmega\right)f_{ln}\lcft( v\cighr)\end{aligned}$$ and for $N_{\'hi}=0$ $$\begin{aligned} \psi_{l,s} & = & e^{-\pi\fsae{\omega}{L_{1}L_{2}}\sum y_{\gamma}^2} \proe_{\gqmma < \delda} \lddt( \gheua_{1}\lxft( z_{\gammw} - v_{\delta} \mid \fau \right) \rught)^{\frac{1}{\kappa}} e^{iXv} \npntnber \\ & &\sum_{n=0}^{p-1} e^{j2\pi cn/[}e^{-y2\pi\frac{jn}{\kappa}} \theta\left[\begin{array}{c} -\left(t/\kalpa +\
function transforms under translations by a lattice will convenient to the following $p$ $N_{\Phi}\neq $$\begin{aligned} & &{\tilde c}}^{j} = e^{-\frac{\pi\omega}{L_{1}L_{2}}\sum \prod_{\gamma < \delta} \left(\theta_{1}\left( z_{\gamma} - \mid \tau \right)\right)^{\frac{1}{\kappa}} \nonumber \\ & &\sum_{K=0}^{M-1}\sum_{n=0}^{p-1}\sum_{{\cal P}\in S_{N_A}} e^{-i2\pi\frac{nj}{\kappa}} e^{i2\pi n\left(q\mid c + pN_{\Phi}c_{\gamma}\right)/p\ell} e^{i2\pi K\left(ns/\kappa+\beta s/2\pi\kappa +k_{2}/N\right)} \nonumber \\ & & \theta\left[\begin{array}{c} -{\hat e}\left(n/\kappa /2\pi\kappa \\ + e} +{\cal P}{\vec c}/\ell \end{array}\right] \left({\vec z},\Omega\right)f_{ln}\left( v\right)\end{aligned}$$ and for $N_{\Phi}=0$ $$\begin{aligned} \psi_{l,c} & = & y_{\gamma}^2} \prod_{\gamma < \delta} \left( \theta_{1}\left( z_{\gamma} - \mid \tau \right) \right)^{\frac{1}{\kappa}} \nonumber \\ & &\sum_{n=0}^{p-1} e^{i2\pi \theta\left[\begin{array}{c} +\
function transforms under trAnslations By a laTtiCe vEcTor. IT wilL be convenient tO DefiNe the following $p$ set of waVe funCtIOns, fOR $N_{\phi}\neQ 0$: $$\begin{aLIgNED} & &{\tiLdE \pSi}_{l,K_{1},{\vEC c}}^{J} = e^{-\fraC{\pi\Omega}{L_{1}L_{2}}\Sum y_{\gamma}^{2}} \pRod_{\GaMma < \delta} \left(\THeTa_{1}\left( z_{\gamMa} - z_{\Delta} \mid \tau \rIghT)\right)^{\FrAc{1}{\kAPpa}} \noNumBer \\ & &\suM_{K=0}^{M-1}\sum_{N=0}^{P-1}\sum_{{\caL P}\in S_{N_A}} e^{-i2\Pi\FRac{nj}{\kAPpa}} e^{i2\pi N\LEfT(q\miD c \mid + pN_{\Phi}c_{\gamma}\rIGhT)/P\ell} e^{i2\pi K\left(nS/\kappa+\BeTA s/2\PI\KapPa +k_{2}/n\right)} \nonuMbEr \\ & & \theTA\left[\beGIn{ARRAy}{c} -{\HAt e}\left(n/\kappa \Beta /2\pi\kappa \RIghT) \\ \left(K/n + {\aLphA}/2\Pi\omegA\kappA\rIGht){\Hat e} +{\cal P}{\vec C}/\ell \End{array}\rIght] \leFT({\vec z},\OmEGa\right)F_{ln}\lefT( v\rIghT)\end{ALiGnEd}$$ aNd FOr $N_{\pHi}=0$ $$\BegIN{alIgned} \psi_{L,c} & = & E^{-\pI\frac{\OmegA}{l_{1}l_{2}}\SUm y_{\gAmmA}^2} \proD_{\gammA < \delta} \left( \theTa_{1}\lEft( z_{\GAmmA} - z_{\delTa} \mid \Tau \rIgHt) \rigHt)^{\frac{1}{\Kappa}} E^{ixv} \nonumber \\ & &\sum_{n=0}^{p-1} E^{i2\pi Cn/p}e^{-i2\pi\frAc{jN}{\kAppA}} \tHeta\lEFt[\begiN{arRay}{C} -\left(n/\kAppa +\
function transforms under translati ons b y a la tt icevect or. It will be conv enient to define the f ollow in g $p$ se t ofwave fu n ct i o ns, f or $N _{ \ Ph i}\ne q 0 $: $$\b egin{align ed} & &{\tilde \p s i} _{l,k_{1}, {\v ec c}}^{j} = e^ {-\fra c{ \pi \ omega }{L _{1}L _{2}}\ s um y_{ \gamma}^{ 2} } \prod _ {\gamma < \ delt a} \left(\theta_{ 1 }\ l eft( z_{\gamma } - z _{ \ de l t a}\mi d \tau \ri gh t)\ri g ht)^{\f r ac { 1 } {\k a ppa}} \nonumb er \\ & &\s u m_{ K=0}^{ M- 1}\ s um_{n= 0}^{p -1 } \su m_{{\cal P} \inS_{N_A}}e^{-i2 \ pi\frac { nj}{\ka ppa}}e^{ i2\ pi n \ le ft (q\ mi d c\ mi d + pN_ {\Phi}c_ {\ ga mma}\ righ t ) / p \ell } e ^{i2 \pi K \left(ns/\kap pa+ \bet a s/ 2\pi\ kappa +k_ {2 }/N\r ight)} \non um ber \\ & & \the ta\l eft[\begi n{a rr ay} {c } -{\ h at e}\ lef t(n /\kappa \beta/ 2\p i\ k a p pa \right) \\ \left (K / N + {\alpha }/2\pi \ om eg a \kappa\r ig ht) {\ha t e} +{ \cal P} {\vec c} /\ell\ en d{ array}\ ri ght] \ le ft( {\v ec z} , \Ome ga\rig ht)f_{ln }\lef t ( v\right)\end { aligned}$$ an d f o r $ N _{\P hi} =0$ $$\begi n{al i gned } \p s i_ {l, c } & = & e^ {- \ pi \ frac{\omega}{L_{1}L _{ 2}}\su m y_{ \gamma}^2} \p rod_{\gamm a < \delta} \le f t( \theta_{1}\lef t( z_ {\gamma} - z_{\del ta} \ mid \tau \right)\ r ight)^{\ fra c{1 }{\ kap p a }} e^{iXv} \non u m ber\\ & &\su m_{ n=0}^{p -1} e^ {i2 \pi c n/p}e^{-i 2\pi\fra c{ jn }{ \k app a}} \ t heta\lef t[ \be gi n{a rray} { c} -\l eft(n /\ka pp a+ \
function_transforms under_translations by a lattice_vector. It_will_be convenient_to_define the following_$p$ set of_wave functions, for $N_{\Phi}\neq_0$: $$\begin{aligned} & &{\tilde_\psi}_{l,k_{1},{\vec_c}}^{j} = e^{-\frac{\pi\omega}{L_{1}L_{2}}\sum y_{\gamma}^{2}} \prod_{\gamma < \delta} \left(\theta_{1}\left( z_{\gamma} - z_{\delta} \mid \tau \right)\right)^{\frac{1}{\kappa}} \nonumber \\ &_&\sum_{K=0}^{M-1}\sum_{n=0}^{p-1}\sum_{{\cal_P}\in S_{N_A}} e^{-i2\pi\frac{nj}{\kappa}} e^{i2\pi_n\left(q\mid_c_\mid + pN_{\Phi}c_{\gamma}\right)/p\ell} e^{i2\pi K\left(ns/\kappa+\beta s/2\pi\kappa_+k_{2}/N\right)} \nonumber \\ & & \theta\left[\begin{array}{c} -{\hat_e}\left(n/\kappa \beta_/2\pi\kappa \right) \\ \left(K/N + {\alpha}/2\pi\omega\kappa\right){\hat e} +{\cal P}{\vec_c}/\ell \end{array}\right] \left({\vec_z},\Omega\right)f_{ln}\left( v\right)\end{aligned}$$ and_for $N_{\Phi}=0$ $$\begin{aligned} \psi_{l,c} & = & e^{-\pi\frac{\omega}{L_{1}L_{2}}\sum y_{\gamma}^2} \prod_{\gamma <_\delta} \left( \theta_{1}\left( z_{\gamma} - z_{\delta}_\mid \tau \right)_\right)^{\frac{1}{\kappa}} e^{iXv}_\nonumber_\\ & &\sum_{n=0}^{p-1} e^{i2\pi cn/p}e^{-i2\pi\frac{jn}{\kappa}} \theta\left[\begin{array}{c}_-\left(n/\kappa +\